Circadian Factors in Coping with Chronic Stress

by

Daniela Bellicoso

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy

Department of Psychology University of Toronto

© Copyright by Daniela Bellicoso 2017

ii

Circadian Factors in Coping with Chronic Stress

Daniela Bellicoso

Doctor of Philosophy

Department of Psychology University of Toronto

2017

Abstract

Circadian rhythms, cyclic changes that repeat approximately once every 24 hours, regulate daily

temporal processes of physiology and behaviour in all living organisms. Rhythmicity is adaptive,

providing advantages for survival, growth, and reproduction. For human beings, this includes

rhythmic regulation of mental, physical, and emotional responses to stressors. In this dissertation,

we addressed how human beings facing major stressors such as chronic disease (in this case,

cancer), are influenced by circadian timing, specifically chronotype and sleep quality. A broad

epidemiological study was conducted among individuals in different roles (patients, medical

staff, and familial caregivers) facing a common stressor to assesses their ability to cope with the

situation. Breast cancer patients provided real-time reports of their coping across the day.

Patients and familial caregivers completed retrospective average ratings for their coping across

the day across treatment, along with details on use of coping behaviours. Oncology staff

provided ratings of their burnout, and Professional Quality of Life (ProQoL). These data were all

assessed in relation to chronotype and sleep quality, and to an extent, in relation to personality.

Chronotype and sleep quality influenced coping within each group, but their impacts were not

correlated. Nonetheless, working at chronotypically optimal times improved ProQoL. ‘In the

moment’ coping ratings from patients reflected their chronotype, as did their recalled coping.

‘Openness’ was positively linked with ProQoL among staff, and with an engagement coping

iii

style among patients and caregivers. The data reflect an influence of circadian timing on the

expression of coping responses during chronic stress. A better understanding of changes in

coping ability as they relate to one’s innate rhythms will allow for the development of a

cognitive and emotion-based chronotherapy regime intended to maximize proactive coping

among individuals facing chronic stress, such as providing care or undergoing treatment for

cancer.

iv

Acknowledgments

Without the help of the following people, this PhD would not have been possible – I owe each of

you many thanks for the kindness you showed me.

My Primary Supervisor, Dr. Martin Ralph: Thank you for taking a chance on me and allowing

me to begin conducting research in your lab as an undergraduate student, over 10 years ago.

Your guidance, patience, and kindness through the years have allowed me the freedom to explore

my own ideas and grow as a researcher in my own way. Thank you ever so much.

Dr. Marg Fitch: Thank you for lending your expertise in the field of coping and adaptation to

illness. Your invaluable guidance on how to approach the various topics explored in this research

are greatly appreciated. It was a pleasure to work with you, and get to know you over the years.

Dr. Maureen Trudeau: There are no words to express my gratitude for the guidance, kindness,

and support you’ve shown me. Without you, this research would not have been possible. It was a

pleasure to learn from you. Thank you for the innumerable and invaluable opportunities for

research, learning, and personal growth that you provided.

The kind staff at Sunnybrook Hospital – Odette Cancer Centre & Princess Margaret Hospital:

Thank you for taking the time out of your busy schedules to offer suggestions on topics and

questions for research, and to identify participants for my studies. Your participation in my

studies and taking the time to sharing some of your own struggles relating to facing chronic

stress were invaluable. I hope that my research, even if only in a small way, helps to make your

job slightly more manageable.

The Breast Cancer Patients and their Caregivers: Your willingness to partake in research, without

any personal reward during your most difficult days is a testament to the good nature and kind

spirit you each possess. Thank you for taking the time to chat with me, share your stories, and

open-up to a stranger about some of your own daily struggles. Thank you for your selflessness.

My husband Matthew, and my sister Elisa: You’ve both given me courage in so many ways. The

love and encouragement you both provided were invaluable, and greatly appreciated.

v

My grandparents, who came to this wonderful country over 60 years ago, in search of a better

life for their families: The life lessons of hard work and determination that you each taught me

helped me to achieve this degree. Thank you for the sacrifices you made, so I could have so

many opportunities and such a wonderful education.

And finally, my parents: You each instilled a love of learning, and a strong work ethic from early

on in my life. You have been my biggest cheerleaders, never failing to give encouragement, and

let me know how much you believed in me. The experiences and opportunities you provided

from early on opened my eyes to so many things around me, and piqued my curiosity to always

learn more and understand why things are a certain way. While you’ve always told me how

proud I make you, I’m proud to be the daughter of two such wonderful people who allowed me

to explore, to be me, and to make my own path in the world.

vi

Table of Contents

Acknowledgments ........................................................................................................................ iv

Table of Contents ......................................................................................................................... vi

List of Tables ................................................................................................................................ xi

List of Figures ............................................................................................................................. xiv

Chapter 1 ....................................................................................................................................... 1

General Introduction ............................................................................................................ 1

1.1 Context, Hypotheses, and Rationale .........................................................................................1 1.2 Coping and Survival Strategies .................................................................................................3

1.2.1 Coping .....................................................................................................................................3 1.2.2 Brief-COPE Questionnaire ...................................................................................................3 1.2.3 Circadian Rhythms and Coping Strategies .........................................................................6 1.2.4 Sleep and Coping ....................................................................................................................7 1.2.5 Depression, Stress, and Coping .............................................................................................9 1.2.6 Burnout, Stress, and Coping in Caregivers .......................................................................10 1.2.7 Personality and Coping .......................................................................................................11

1.3 Cancer as a Chronic Stress ......................................................................................................13 1.3.1 Emotional Distress in Patients ............................................................................................14

1.3.1.1 Why Choose Breast Cancer Patients .........................................................................................14 1.3.1.2 Specific Distress Involving Genetic Issues ................................................................................15

1.3.2 Emotional Distress in Caregivers .......................................................................................15 1.3.2.1 Specific Distress Involving Partners ..........................................................................................15 1.3.2.2 Specific Distress Involving Oncology Staff ...............................................................................16

1.4 Rhythmicity ...............................................................................................................................16 1.4.1 Biological Clocks in Nature .................................................................................................16 1.4.2 Biological Clocks in Mammals ............................................................................................17 1.4.3 Biological Clocks in Human Beings ....................................................................................18 1.4.4 Cognition, Emotion, and Circadian Rhythms ...................................................................18 1.4.5 Chronotype ...........................................................................................................................18

1.4.5.1 Genetic Basis of Chronotype ......................................................................................................19 1.4.5.2 Questionnaires ............................................................................................................................19

vii

1.4.5.3 Performance Variances ..............................................................................................................20 1.4.5.4 Emotionality ................................................................................................................................21

1.5 Sleep ...........................................................................................................................................21 1.5.1 Two-Process Model of Sleep ................................................................................................22 1.5.2 The Functions of Sleep .........................................................................................................22 1.5.3 The Impact of Sleep and Sleep Loss ...................................................................................23

1.5.3.1 Sleep, Cognition, and Memory ...................................................................................................23 1.5.3.2 Polysomnography and Actigraphy .............................................................................................25 1.5.3.3 Sleep Quality ...............................................................................................................................25

1.5.3.3.1 Cancer ....................................................................................................................................26 1.5.3.3.2 Breast Cancer Patients ...........................................................................................................27 1.5.3.3.3 Caregivers ..............................................................................................................................28 1.5.3.3.4 Oncology Staff ........................................................................................................................29

1.6 Breast Cancer Background .....................................................................................................29 1.6.1 Cancer Biology .....................................................................................................................29 1.6.2 Breast Cancer Biology .........................................................................................................30 1.6.3 Genetics and Mutations .......................................................................................................31 1.6.4 Incidence ...............................................................................................................................31 1.6.5 Diagnosis and Treatment .....................................................................................................32

1.6.5.1 Staging ........................................................................................................................................32 1.6.5.2 Procedures ..................................................................................................................................34 1.6.5.3 Other Tumour Characteristics ...................................................................................................34 1.6.5.4 Treatment ....................................................................................................................................36

1.6.5.4.1 Localized .................................................................................................................................36 1.6.5.4.2 Systemic ..................................................................................................................................37

1.6.5.5 Prognosis ....................................................................................................................................39

Chapter 2 ..................................................................................................................................... 40

General Methods ................................................................................................................. 40

2.1 Procedures .................................................................................................................................40 2.1.1 Nursing Study .......................................................................................................................40 2.1.2 Hospital Staff ........................................................................................................................40 2.1.3 Patients and Caregivers .......................................................................................................41

2.2 Instruments ...............................................................................................................................41 2.2.1 General Questionnaires .......................................................................................................41 2.2.2 Group Specific Questionnaires ...........................................................................................42

2.2.2.1 Nursing Staff ..............................................................................................................................42

viii

2.2.2.2 Oncology Staff ............................................................................................................................42 2.2.2.3 Patients and Caregivers ..............................................................................................................43

2.2.2.3.1 Questionnaire Package ..........................................................................................................44 2.3 Statistics .....................................................................................................................................45

Chapter 3 ..................................................................................................................................... 46

Burnout Among Oncology Nurses: Influence of Chronotype and Sleep Quality ......... 46

3.1 Abstract .....................................................................................................................................46 3.2 Introduction ..............................................................................................................................47 3.3 Materials and Methods ............................................................................................................49

3.3.1 Participants ...........................................................................................................................49 3.3.2 Procedure ..............................................................................................................................50 3.3.3 Measures ...............................................................................................................................50 3.3.4 Statistical Analysis ...............................................................................................................51

3.4 Results ........................................................................................................................................51 3.5 Discussion ..................................................................................................................................61

3.5.1 Limitations ............................................................................................................................64 3.6 Conclusions ...............................................................................................................................65

Chapter 4 ..................................................................................................................................... 66

Chronobiological Factors for Compassion Satisfaction and Fatigue Among

Ambulatory Oncology Caregivers ............................................................................................. 66

4.1 Abstract .....................................................................................................................................66 4.2 Introduction ..............................................................................................................................67 4.3 Materials and Methods ............................................................................................................69

4.3.1 Participants and Procedures ...............................................................................................69 4.3.2 Measures ...............................................................................................................................70 4.3.3 Statistical Analysis ...............................................................................................................70

4.4 Results ........................................................................................................................................71 4.4.1 Descriptive Group Statistics ................................................................................................71 4.4.2 Correlation Analysis ............................................................................................................73

4.5 Discussion ..................................................................................................................................80 4.5.1 Chronotype and ProQoL .....................................................................................................80 4.5.2 Sleep and ProQoL ................................................................................................................81 4.5.3 Personality and ProQoL ......................................................................................................82

4.5.3.1 Agreeableness .............................................................................................................................82

ix

4.5.3.2 Emotional Stability .....................................................................................................................82 4.5.3.3 Openness .....................................................................................................................................83 4.5.3.4 Conscientiousness and Extraversion .........................................................................................83

4.5.4 Job Satisfaction and ProQoL ..............................................................................................83 4.5.5 Limitations ............................................................................................................................86

4.6 Conclusion .................................................................................................................................86

Chapter 5 ..................................................................................................................................... 91

Primary Circadian Impacts on Patients, Caregivers, and Dyads .................................. 91

5.1 Abstract .....................................................................................................................................91 5.2 Introduction ..............................................................................................................................92 5.3 Materials and Methods ............................................................................................................94

5.3.1 Participants and Procedures ...............................................................................................94 5.3.2 Measures ...............................................................................................................................95 5.3.3 Statistical Analysis ...............................................................................................................95

5.4 Results ........................................................................................................................................96 5.4.1 Descriptive Statistics ............................................................................................................96

5.4.1.1 Overall Demographic Comparisons ..........................................................................................96 5.4.1.2 Morningness-Eveningness Distribution ....................................................................................97 5.4.1.3 Sleep Quality Comparison ..........................................................................................................97 5.4.1.4 UTIME Performance Results ....................................................................................................97

5.4.2 Analysis 1: Total Patient Population Analysis ................................................................105 5.4.2.1 Multifactorial Correlation Analysis .........................................................................................105 5.4.2.2 Global Correlations Among Patients’ MEQ, PSQI and UTIME data ...................................105 5.4.2.3 Total Patient Population Mixed Measures ANOVAs .............................................................107

5.4.3 Analysis 2: Total Caregiver Population Analysis ............................................................111 5.4.3.1 Global Correlations Among Caregivers’ MEQ, PSQI and UTIME data ..............................111 5.4.3.2 Caregiver Population Mixed Measures ANOVAs ...................................................................113

5.4.4 Analysis 3: Matched Patient and Caregiver Population Analysis .................................116 5.4.4.1 Global Associations and Differences Among Matched Patient and Caregiver MEQ, PSQI

and UTIME data ..........................................................................................................................................116 5.4.4.2 Matched Patient and Caregiver Group Mixed Measures ANOVAs .......................................118

5.5 Discussion ................................................................................................................................123 5.5.1 On the Independent Impacts of Sleep and Chronotype .................................................123 5.5.2 Patient Cognitive Function and Emotional Regulation ..................................................126

5.5.2.1 Chronotype-UTIME Correlations ...........................................................................................126 5.5.2.2 Multivariate Analysis of UTIME Correlations in the Patient Population .............................127

x

5.5.3 Caregiver Cognitive Function and Emotional Regulation .............................................131 5.5.4 Matched Patient and Caregiver Cognitive Function and Emotional Regulation ........132 5.5.5 Limitations ..........................................................................................................................135

5.6 Conclusion ...............................................................................................................................136

Chapter 6 ................................................................................................................................... 138

Coping Behaviour in Chronic Disease ............................................................................ 138

6.1 Abstract ...................................................................................................................................138 6.2 Introduction ............................................................................................................................139 6.3 Materials and Methods ..........................................................................................................142

6.3.1 Participants and Procedures .............................................................................................142 6.3.2 Measures .............................................................................................................................143 6.3.3 Statistics ..............................................................................................................................144

6.4 Results ......................................................................................................................................144 6.5 Discussion ................................................................................................................................178

6.5.1 Personal Coping Assessment .............................................................................................178 6.5.2 Coping Behaviours: Changes in Coping Scores ..............................................................180 6.5.3 In the Moment Coping vs. Coping Behaviours ...............................................................189 6.5.4 Limitations ..........................................................................................................................190

6.6 Conclusion ...............................................................................................................................192

Chapter 7 ................................................................................................................................... 194

Conclusion ......................................................................................................................... 194

7.1 Chronotype ..............................................................................................................................195 7.2 Sleep Quality ...........................................................................................................................197 7.3 Personality ...............................................................................................................................199 7.4 Future Directions ....................................................................................................................200

Bibliography .............................................................................................................................. 202

xi

List of Tables

Table 3.1 Demographic characteristics and questionnaire response ratings of participants…………………………………………………………..

52

Table 3.2 Comparison of burnout ratings between respondents with good and bad sleep quality and between MEQ types ………………………….

54

Table 3.3 Analysis of bivariate correlations for participants’ questionnaire response ratings ……………………………………………………..

56

Table 3.4a Summary of hierarchical multivariate regression analysis for variables predicting personal burnout among oncology nurses (N = 64) ………………………….………………………………………..

58

Table 3.4b Summary of hierarchical multivariate regression analysis for variables predicting work related burnout among oncology nurses (N = 64) ………………………….…………………………………..

59

Table 3.4c Summary of hierarchical multivariate regression analysis for variables predicting client related burnout among oncology nurses (N = 64) ………………………….………………………………….

60

Table 4.1 Means and frequencies of participant demographic and questionnaire

data ………………………….………………………………………. 72

Table 4.2 Non-parametric correlations between ProQoL domains (columns) and covariates (rows) ………………………….…………………….

75

Table 4.3 Final models in backwards multiple regression with professional quality of life (CS, BO, and STS) as dependent variables, and significant values* (as tested by univariate regression) for MEQ, PSQI, JSS and TIPI as independent variables ……………………….

77

Table 4.4 Models in backwards multiple regression with professional quality of life (CS, BO, and STS) as dependent variables, and significant values* (as tested by univariate regression) for MEQ, PSQI, TIPI and working on multiple vs. single cancer sites as independent variables ………………………….………………………………….

79

Table S4.1 Spearman correlations (rs) between continuous variables, and point biserial correlations for categorical variables (demographics, chronotype and sleep quality) ……………………………………….

88

Table S4.2 Spearman correlations (rs) between continuous variables, and point biserial correlations for categorical variables related to job satisfaction ………………………….……………………………….

89

Table S4.3 Spearman correlations (rs) between continuous variables, and point biserial correlations for categorical variables related to personality and professional quality of life ……………………………………...

90

xii

Table 5.1 Means and frequencies of demographic and questionnaire data for patients and caregivers ………………………………………………

99

Table 5.2 Means and frequencies for chronotype and sleep quality questionnaires for patients and caregivers …………………………..

100

Table 5.3 Spearman correlations for chronotype versus sleep quality and UTIME among patients ……………………………………………..

106

Table 5.4 Mixed ANOVAs on patient sleep quality and UTIME: main effects and interactions ………………………….…………………………..

110

Table 5.5 Spearman correlations for caregiver chronotype versus sleep quality and UTIME ………………………….………………………………

112

Table 5.6 Mixed ANOVAs on caregiver sleep quality and UTIME: main effects and interactions ……………………………………………...

115

Table 5.7 Median and Mann Whitney U significance values comparing MEQ, PSQI, and UTIME response scores between patients and their caregivers ………………………….………………………………...

117

Table 5.8 Mixed ANOVAs on patient AND caregiver sleep quality and UTIME: main effects and interactions ……………………………...

122

Table 6.1 Descriptive data for patient coping logs ……………………………. 146 Table 6.2 Median and Friedman test data for changes in raw patient coping

log scores across treatment, with Wilcoxon signed-rank tests with Bonferroni correction applied for differences in raw patient coping log scores across treatment ………………………………………….

148

Table 6.3 Mean ± standard deviation and paired samples t-test data for pre- and post-chemotherapy coping rating comparisons ………………...

150

Table 6.4 Pearson correlations between MEQ and PSQI and patients’ coping log UTIME scores …………………………………………………...

152

Table 6.5 Descriptive data for Brief COPE are mean ± standard deviation …... 154 Table 6.6a Median and Friedman test data for changes in Brief COPE scores

across treatment among patients and caregivers ……………………. 157

Table 6.6b Wilcoxon signed-rank tests with Bonferroni correction applied for differences in Brief COPE scores across treatment …………………

158

Table 6.7 Summary of Multiple Regression Analyses for Brief-COPE Scores Across Treatment, assessing the predictive value of participant role, chronotype, and sleep quality ……………………………………….

163

Table 6.8 Descriptive data and independent samples t-test for BFAS between men and women. Descriptive data are mean ± standard deviation, for patients and caregivers …………………………………………..

166

xiii

Table 6.9 Summary of Multiple Regression Analyses for Brief-COPE Scores Across Treatment, assessing the predictive value of participant role, chronotype, sleep quality and personality …………………………..

172

xiv

List of Figures

Figure 5.1a UTIME scores across treatment for the total patient group ………… 101 Figure 5.1b UTIME scores across treatment for patients without a caregiver

involved in the study ………………………….…………………….. 102

Figure 5.1c UTIME scores across treatment for patients with a caregiver involved in the study ………………………….……………………..

103

Figure 5.1d UTIME scores across treatment for caregivers …………………….. 104

1

Chapter 1!

! General Introduction

1.1! Context, Hypotheses, and Rationale

Circadian rhythms, cyclic changes that repeat approximately once every 24 hours, regulate daily

temporal processes of biology and physiology in all living organisms. Rhythmicity is expressed

in processes at all levels of biological organization. At a molecular level, cell division and

replication are known to be rhythmic. At a higher level, cognition, physical functioning, and

emotionality have all been linked to a rhythmic preference for morning (M) versus evening (E)

performance, known as chronotype. Our biology, physiology, and behaviour are governed by

clocks. In this same vein, previous studies have explored the impact of rhythmicity on various

domains of health. Different findings in areas such as cardiology (e.g., Portaluppi et al., 2012),

musco-skeletal function (e.g., Riley & Esser, 2017), gynecology/obstetrics (e.g., Dogru et al.,

2016), and mental health (e.g., Hasler, Allen, Sbarra, Bootzin, & Bernert, 2010) support evidence

for health and rhythms.

Coping strategies, specifically, coping with chronic trauma that accompany health issues, has not

been examined in depth in relation to rhythmicity. Very little is known about how human beings

might respond at different times of day to a major stressor such as chronic disease, along with

other environmental factors that will influence an individual’s behaviour.

This dissertation focuses on how human beings facing major stressors (specifically, chronic

disease) are influenced by circadian timing. Based on existing information on rhythmicity of

emotional responses (e.g., Costa-Martins et al., 2016; Lenaert, Barry, Schruers, Vervliet, &

Hermans, 2016), a person’s ability to cope with not only the chronic disease but the other

stressors in their environment might vary according to the timing of their circadian cycle. Given

that research has suggested that how an individual deals with their disease or stressors can

impact outcome (e.g., Demytteraera et al., 1998; Shehmar & Gupta, 2010) it is important to

understand if and how these abilities might vary through the day. Furthermore, stress can

contribute to perpetuating a disease and disrupting circadian rhythms (McEwen & Karatsoreos,

2015). We should recognize that if this is the case, then not only will patients be affected, but

2

also everyone involved with the patient and going through the highly stressful situation will also

be influenced. It is reasonable to predict that the stress that occurs due to the disease itself,

together with the stresses felt by patients and caregivers will have a mix of influence on coping.

We initiated a large scale epidemiological study of a group of individuals facing the same

stressor (cancer), but each in a different role (patient, medical staff, familial caregiver), looking

at their ability to cope with the situation. We examined a number of important variables believed

to influence coping behaviour and memory for coping, by conducting various cross-sectional and

longitudinal studies. Using surveys and data logs, we collected measures of participants’

chronotype, sleep quality, personality, and coping. To examine chronotype and sleep quality, this

requires an understanding of the rhythmic processes, starting with the generation of rhythms by

biological clocks, and the expression of that timing in the regulation of behaviour. It requires

also an understanding of the individual and the disease itself. It also involves an understanding

of sleep stages and implications for good versus poor quality sleep. We narrowed down the focus

to address a group of breast cancer patients where the disease itself is relatively well defined, as

is treatment for the disease. The prognosis is also relatively well defined.

Given that circadian rhythms influence cognitive function and emotional regulation throughout

the day in the general population, (e.g., Blatter & Cajochen, 2006; Ottoni, Antoniolli & Lara,

2012) it is important to understand how these rhythms fluctuate among both cancer patients and

caregivers faced with a chronic daily stressor such as cancer. While it is understood that

disturbed rhythms can perpetuate poor health, less is known in general about circadian rhythms

and emotionality and emotional responding, including behaviour and memory for behaviour. An

understanding of emotionality and rhythmicity is required. It was hypothesized that rhythmicity

would impact individuals’ responses to chronic stress, but the influence of circadian rhythms

would vary depending on one’s role, and whether coping ratings were retrospective or in the

moment. A better understanding and documentation of changes in coping ability as they relate to

one’s innate rhythms will allow for the development of a cognitive and emotion-based

chronotherapy regime intended to maximize proactive coping among cancer patients and their

caregivers, both in the hospital and the home.

3

1.2! Coping and Survival Strategies

1.2.1! Coping

Coping refers to those “constantly changing cognitive and behavioral efforts to manage specific

external and internal demands that are appraised as taxing or exceeding the resources of the

person” (Lazarus & Folkman, 1984, p. 141). These taxing demands that exceed one’s personal

resources are known as stress. Stress is a real or perceived interruption to the homeostasis of

one’s physical state or mental well-being. Stress can result from a range of positive and negative

demands. For example, the stress of organizing a large event, or the physical stress that comes

from exercise both have the potential to exceed one’s mental or physical resources, respectively,

to deal with the situation. However, the body’s response to these self-sought out tasks can mimic

the stress response seen when responding to a negative stressor (National Research Council,

2008). Individuals are each taxed differently by various demands, and will not necessarily

respond the same way to a particular stressor. The individual efforts people perform to cope with

stress affects their physical, psychological, and social well-being (Folkman & Lazarus, 1980). As

such, it is important to understand how a specific stressor can elicit various coping responses

from different individuals, and the effect these efforts will have on well-being.

1.2.2! Brief-COPE Questionnaire

The cognitive and behavioural efforts an individual may use to cope or deal with the stress of a

situation can range from healthy to negative, yet it can be difficult to clearly distinguish which

specific efforts fall into either category. The Ways of Coping Questionnaire, developed by

Folkman and Lazarus (1980), suggests certain efforts can be categorized into problem or emotion

focused coping styles. Problem focused coping strategies are intended for problem solving or

performing some action to change the source of the stress. Emotion focused coping efforts are

intended to manage or reduce the emotional distress being created by or associated with the

stressor. It is important to note that of the two techniques, neither is necessarily more or less

positive or negative than the other. This distinction between problem and emotion focused

coping is very basic, and it is important to note that not all efforts aimed at reducing a stressor

necessarily fall neatly into one category or the other (Carver, Scheier, & Weintraub, 1989).

4

The COPE is a 60-item, 15-scale measure, and contains many items which can be considered as

either emotion or problem focused coping. The COPE was also found to be correlated in varying

degrees to certain personality traits. Findings demonstrated that functional coping strategies are

generally linked to personality traits from various questionnaires that are seen as beneficial,

while less functional coping strategies showed inverse correlations with desirable personality

traits. The COPE was designed to address three key issues the authors believed existed with

previous coping measures:

1.! Provide a more complete and comprehensive assessment of the various coping efforts

people may engage in to deal with a stressor

2.! Reduce ambiguity and produce questions with a direct focus

3.! Develop a scale that is theoretically rather than empirically based, focusing on specific

theoretical arguments that assess functional properties of coping strategies

(Austenfeld & Stanton, 2004; Carver, Scheier, & Weintraub, 1989)

This research assesses coping strategies using the Brief COPE, a 28-item 14-scale measure

which is a condensed version of the original COPE questionnaire. The Brief COPE was created

to:

•! Minimize time demands on participants

•! Revise the questionnaire to exclude two irrelevant scales

•! Slightly refocus three scales

•! Include a self-blame scale evidence that had since been proved was important

The Brief COPE continues to address the three key goals of the original COPE. Like the original

COPE, the Brief COPE continues to include both adaptive and dysfunctional measures of

coping. The items of the Brief COPE can be presented in three formats:

•! Situational retrospective (e.g., “I’ve been doing things to try and take my mind off the

situation”)

•! Dispositional (e.g., “I do things to try and take my mind off the situation”)

•! Situational concurrent (e.g., “I’m doing things to try and take my mind off the situation”).

5

The 14 scales of the Brief COPE refer to coping in the following ways:

1.! Active Coping: taking measures to attempt to remove or avoid the stress, or ameliorate its

effects

2.! Planning: coming up with strategies to deal with the stressor and best handle the situation

3.! Positive Reframing: managing or reframing distressing emotions resulting from the

stressor or situation in positive terms

4.! Acceptance: accepting the reality of and attempting to deal with the situation or stressor;

opposite of denial

5.! Humour: making fun of the stressor or situation in an attempt to make light of the

situation

6.! Religion: using religion to provide a source of comfort, or to clear or organize one’s

thoughts about a stressor

7.! Using Emotional Support: seeking out moral support, sympathy, and or understanding

from others about having to deal with the stressor or situation; this is emotion-focused

coping

8.! Using Instrumental Support: getting advice, assistance, and or information from others on

how to deal with the stressor or situation; this is strictly problem-focused coping

9.! Self-Distraction: focusing away from the stress; intentionally performing activities to take

one’s mind off the stressor

10.!Denial: attempting to push away or ignore the reality of the situation; opposite of

acceptance

11.!Venting: focusing on the stress

12.!Substance Use: using alcohol or drugs to think less about the stressor

13.!Behavioural Disengagement: lessening or giving up one’s attempts to deal with the

stressor and/or achieve goals the stressor interferes with

14.!Self-Blame: criticizing oneself as being responsible for the stressor or situation

Certain scales of the Brief COPE are clearly dysfunctional or adaptive, or carry a distinct

negative or positive tone. For example, behavioural disengagement is clearly negative as it

involves one giving up any attempt of working with the situation or stressor. Conversely,

acceptance carries a strong positive tone as it requires one come to terms with the stressor, which

creates opportunity to move forward and deal with the stresses one is facing. Yet other categories

6

are less clearly defined on whether they are adaptive or dysfunctional coping strategies. For

example, humour can be used to shed light on the situation and possibly make it easier to face a

stressor, however, this can also become dysfunctional if someone uses humour to not have to

face the severity of a stressor or situation. It is important to keep the nature of each scale in mind

when assessing the various coping strategies used by patients. Furthermore, it is important to

assess if certain coping styles as indicated by particular scales tend to co-occur (Carver, 1997).

1.2.3! Circadian Rhythms and Coping Strategies

While the literature focusing specifically on circadian rhythms and coping is sparse, research

does indicate that the cognitive and emotional processes which regulate one’s use of particular

coping strategies are under circadian control. Functioning of cognitive and emotional processes

vary over the course of the 24 h day. An appropriately timed wake and sleep cycle that reflects

one’s internal biological clock facilitates maximal cognitive and emotional performance.

Conversely, a wake sleep schedule that does not mirror one’s biological clock can reduce an

individual’s cognitive and or emotional regulatory abilities (Wright, Lowry, & LeBourgeois,

2012).

Given that the use or disuse of cognitive, emotional, and or inhibitory control in different

combinations plays a role in each of the coping strategies people commonly use (i.e., such as

those assessed in the Brief COPE), these dimensions provide an important link between circadian

rhythms and the coping strategies an individual may use. Circadian oscillators are known to

regulate cognition based functions, such as maintaining alertness, and learning and memory

formation and recall. Performance of these cognitive functions is significantly reduced when

occurring out of synchrony with one’s innate circadian rhythm, such as at one’s off peak time as

indicated by their chronotype (Krishnan & Lyons, 2015). One’s inhibitory control is also

modulated by chronotype. For example, on a task measuring vigilance, M type individuals

maintained high performance when tested in the morning, whereas their performance decreased

with time on task in the evening. Conversely, E type individuals showed worse inhibitory control

with greater time on task in the morning session and greater performance in the evening testing

session (Lara, Madrid & Correa, 2014). This study indicates that for cognitive measures, the

negative effects of time on task can be mediated by testing an individual at his or her

chronotypically optimal time in accordance with their circadian rhythm. Inhibitory control is

7

particularly important to coping as it may allow for blocking certain coping strategies that may

be negative or dysfunctional, and that an individual may be aware are not in their best interest,

but may be a natural response. At one’s chronotypically optimal time, it may be easier to work to

actively avoid particular unwanted strategies, whereas at one’s off peak times, inhibitory control

may be lacking and subsequently facilitate the use of these otherwise blocked out strategies.

In addition to changes in cognitive regulatory abilities across the day in line with one’s circadian

rhythm, it appears that chronotype is also linked with changes in emotionality and mood. When

measuring affect across the day for an entire week in healthy individuals, M type individuals

showed the quickest rise in positive affect in the morning between 9 a.m. and noon, followed by

a dramatic decrease after 9 p.m. Conversely, N (neither type, i.e., intermediate between M and E

types) and E types did not demonstrate the same rapid rise in positive affect in the morning as

was seen among M types (Clark, Watson & Leeka, 1989). Among healthy individuals,

depressive symptomatology is more common among E type individuals than M types, suggesting

an E chronotype may be a predisposing factor for depression (Hidalgo et al., 2009). These results

have been replicated even among individuals with different physical health levels. For example,

between normal versus overweight females, E typology remains associated with a greater

number of depressive symptoms (Pabst, Negriff, Dorn, Susman & Huang, 2009), suggesting E

types report more depressive symptomatology regardless of physical health. Even among healthy

individuals, one’s innate rhythm may predispose an individual towards greater use of negative or

dysfunctional coping strategies, and overall poorer stress management abilities.

1.2.4! Sleep and Coping

Everyone copes differently. The coping strategies one uses will influence their life in various

ways, including impacting their sleep quality. Cognitive arousal among healthy individuals

dealing with stressful life events has been linked to sleep disruptions and or chronic insomnia

(Friedman, Brooks, Bliwise, Yesavage & Wicks, 1995). It is believed that adequate sleep

duration may act as a biobehavioural regulatory and restorative process that regulates one’s daily

emotional experiences and allostatic loads of emotional stress (Vandekerckhove & Cluydts,

2010). Among individuals with major depression, one’s use of avoidance behaviours as a coping

mechanism, along with the intrusion of unwanted thoughts are known to contribute to poor sleep

(Hall et al., 1997). Among physically healthy persons, emotional arousal caused by anxiety is

8

also known to produce sleep disruptions, due in large part to activation of the corticotropin-

releasing hormone system which is recruited for reacting to emotional stress and is believed to

regulate spontaneous waking (Staner, 2003).

Cancer patients and their caregivers – both in the home and hospital – are under cognitive and

emotional strain. Cancer patients’ sleep is known to generally be poor. Among healthy

individuals, sleep plays an important role in mediating coping as they face regular daily events.

Therefore, it is important to assess sleep as a variable associated with changes in coping across

the day in a cancer patient or caregiver’s ability to face the various stresses involved in their role.

Among early stage breast cancer patients, the use of avoidance coping has been linked with

greater time needed to fall asleep across the treatment trajectory. Similar results have been found

in men with prostate cancer, along with decreased sleep onset time (i.e., sleep latency) both at

baseline and across treatment when approach coping strategies are used (Thomas, Bower, Hoyt

& Sepah, 2010). In a study of a varied sample of men with cancer, use of avoidance coping at

baseline was associated with poorer sleep at follow up testing. The authors suggested that using

avoidance coping towards cancer-related stressors or circumstances is likely due to poorer mood

and reduced sleep (Hoyt, Thomas, Epstein & Dirksen, 2009). In this case, it is possible that

poorer mood creates greater emotional arousal, leading to subsequent sleep disruptions. Familial

cancer patient caregivers report similar results. The use of less functional coping strategies (e.g.,

venting, self-distraction, self-blame) have been associated with increased sleep disruptions

(Aslan, Sanisoglu, Akyol & Yetkin, 2009; Carter & Acton, 2006; Northouse, Williams, Given &

McCorkle, 2012). Associations have also been found in some studies suggesting that the use of

positive, proactive coping strategies among caregivers has been associated with reduced numbers

of sleep disturbances (Zhang, Yao, Yang, & Zhou, 2014). Interestingly, in studies among

patients and caregivers, while research seems to consistently point to a positive correlation

between less functional coping styles and increased reports of poor sleep, not all studies seem to

find this association between functional or proactive coping strategies and better sleep quality.

Research on the association between sleep quality and coping strategies required to face the

stresses of being an oncology staff member (e.g., oncologist, oncology nurse, etc.) is more

scarce. Given the cognitive and emotional burden of caring for cancer patients – both treatable

and terminal – it is important to have well developed, functional coping strategies. In a study of

nurses following a shiftwork schedule, on average, sleep quality was found to be poor, as rated

9

by the PSQI. Among this same sample, high emotional disturbance was correlated with poorer

overall sleep quality and greater sleep disturbances (Lee, Chen, Tseng, Lee & Huang, 2015).

While no actual coping measure was used in this study, the high level of emotional disturbance

suggests a lack of coping strategies being used to mediate the emotional demands of the job.

While ambulatory oncology staff do not all follow shiftwork schedules, there are still several

cognitive and emotional demands to be dealt with on a daily basis that require well developed

functional, proactive coping strategies.

While the current literature indicates that sleep and coping are related, it is important to

understand how sleep may be associated with one’s perceived coping abilities as they change

across the day. A better understanding of the influence of sleep on various coping strategies will

guide the development of strategies to teach necessary healthy coping behaviours to patients and

caregivers to better help them in their role. A better understanding of changes in coping across

the day as they relate to one’s sleep quality will allow for necessary assistance to be provided at

times of day when additional help may be required to cope in a proactive fashion.

1.2.5! Depression, Stress, and Coping

In North America, major depressive disorder is the leading cause of disability, and by 2020 is

projected to become the second leading cause of disability worldwide (Muscatell, Slavich,

Monroe & Gotlib, 2009). Depression, which refers to a range of mental health problems, is

characterized by persistent traits such as low mood state, little or no positive affect, and

functional and social impairment. According to Radloff (1977) who designed the Centre for

Epidemiological Studies Depression Scale (CES-D), key major components of depressive

symptomatology include low mood, feelings of guilt and worthlessness, a sense of helplessness

and hopelessness, little or no appetite, psychomotor retardation or delay, and sleep disturbances.

Stress – either chronic or acute – can play a role in the development of depression (Hammen,

Kim, Eberhart & Brennan, 2009; Muscatell, Slavich, Monroe & Gotlib, 2009). One’s reaction to,

or interpretation of a stressor influences the mental impact it has on the individual and affects the

stressors’ contribution to depression onset. When faced with stressors, individuals who exhibit

depression or a high number of depressive symptoms are more likely to use dysfunctional coping

strategies based on avoidance and denial; healthy controls are more likely to use positive,

adaptive coping strategies aimed at accepting the stressor and making plans to move forward

10

towards one’s goals (Orzechowska, Zajaczkowska, Talarowska & Galecki, 2013). Given the

high stress levels typically experienced by cancer patient and caregiver populations, it is

necessary to understand how one’s use of specific coping strategies can reflect depression onset;

such an understanding would allow treatment to be provided before the depression worsens.

1.2.6! Burnout, Stress, and Coping in Caregivers

Burnout is believed to be a consequence of a broader feeling known as compassion fatigue (CF),

that often develops among caregivers of trauma victims and/or patients with a grave illness,

particularly after providing care for an extended period (Stamm, 2010). CF refers to the negative

outcomes of being a caregiver to such individuals, and includes specific feelings such as

exhaustion, frustration, depression, or even fear resulting from working with this population. In

addition to burnout, the other outcome of CF is secondary traumatic stress (STS), resulting due

to prolonged exposure to traumatized individuals, and manifesting as an ongoing combination of

fear, intrusive imagery, and/or sleep disturbance.

While caring for sick or traumatized individuals over an extended period can be mentally and

emotionally draining, there is also the opportunity for personal reward stemming from knowing

one has contributed to ameliorating the quality of life of another individual in a time of need.

These positive feelings, known as compassion satisfaction (CS) are the opposite of CF, and refer

specifically to the pleasure or fulfillment one feels from helping others, in particular those faced

with illness or trauma, and carrying out this role well (Stamm, 2002). Unlike CF, CS does not

break down into further subcategories. Stamm (2010), suggests that together, CS and CF

represent a worker’s overall professional quality of life. The Professional Quality of Life Scale

(ProQoL) produces a rating of these two components.

The coping strategies a caregiver uses to deal with the stresses associated with caring for

traumatized and/or grievously ill patients may contribute to their level of CS or CF. However,

the literature on specific coping strategies that contribute towards increased or decreased CS

and/or CF is sparse. The coping strategies used may alter one’s perception of the stresses faced

on a daily basis, producing higher or lower levels of CS or CF. Understanding which specific

coping strategies are associated with increased or decreased levels of CS and CF is important.

This information can gauge which strategies increase the satisfaction one obtains from their

caregiving role, and help to assess and provide suggestions to alter one’s coping style when it is

11

known to reduce satisfaction and increase fatigue. Given that previous research has shown that a

caregiver’s mental state impacts the quality of care provided to patients (e.g., Beach et al., 2005),

this research will contribute to the field by providing suggestions for appropriate coping styles to

maximize CS and reduce CF.

1.2.7! Personality and Coping

Personality, which is made up of the individual characteristics that shape a person’s behaviour,

feelings, and thoughts, has been linked with circadian rhythms and sleep quality (e.g., Cavallera,

Gatto & Boari, 2014; Duggan, Friedman, McDevitt & Mednick, 2014; Hintsanen et al., 2014;

Hsu, Gau, Shang, Chiu & Lee, 2012). In addition, personality has been linked with coping, such

that certain coping styles occur in increasing or decreasing frequency with certain personality

traits. Some research suggests that even prior to coping, one’s personality predicts frequency of

exposure to stressors, type of stressors experienced, and subsequent appraisals of the stressors.

For example, neuroticism contributes to one’s exposure to interpersonal stress, a tendency to

classify events as highly threatening, and to feel unequipped with the necessary coping

resources, while scoring higher on conscientiousness is reflective of lower stress exposure, likely

due to advanced planning and impulse control (Carver & Connor-Smith, 2010). The idea that

personality predicts stress exposure may be especially true in average daily life scenarios where

one has some control over the roles he or she takes on, the interactions that may be had or

avoided during the day, or the scenarios where a person may find themselves. However, this

theory may not apply in scenarios where a person has less control over the stressor.

Among cancer patients, lifestyle choices may at times contribute to the disease’s development,

but on the whole, people are equally susceptible to a cancer diagnosis in general, regardless of

personality. Among familial or spousal caregiver, one’s relationship with the cancer patient, and

not personality, determines this role. In both these groups, personality will contribute to one’s

adjustment to the role, and their ability to cope with the stressors being presented. In these cases,

it is important to understand how personality predisposes an individual to cope with stressors

when they are presented. An individual’s decision to work in oncology (either as an oncologist,

oncology nurse, pharmacist, etc.), may be more so determined by personality compared to cancer

patients and familial caregivers. This relates back to the previously mentioned point that

personality may predispose people to find themselves in particular scenarios. However, in regard

12

to selecting a career, personality may contribute to one’s reasons for taking on a paid caregiver

role in the field of oncology, in addition to mediating how they will cope with the various

stressors that will be presented in this career. In each of these cases, personality will affect how

one copes with the stresses of their role. While research on coping strategies and personality

traits among cancer patients, and spousal and oncology staff caregivers is sparse, studies

conducted among the general population may serve as a starting point.

Research has suggested various associations between personality traits and choice of coping

strategies. Bartley and Roesch (2011) found that college students high in conscientiousness were

more likely to engaged in problem-focused coping strategies (e.g., problem-solving, cognitive

decision making) which in turn was associated with greater positive affect when faced with a

daily stressor. In a meta-analysis assessing the relationship between coping and personality traits,

high conscientiousness and extraversion were more common among individuals using problem-

solving and cognitive restructuring, which generally reflects problem-focused coping strategies.

Extraversion was also noted to produce support-seeking coping behaviours. This same analysis

noted that neuroticism was more predictive of emotion-focused coping strategies, but also

showed a link to support-seeking, similar to that found among individuals high in extraversion

(Connor-Smith & Flachsbart, 2007). Among intensive care unit (ICU) nurses, while personality

was not associated with workplace stress, it was linked with various coping styles used by the

participants to approach daily stresses. Among nurses reporting greater conscientiousness and or

agreeableness, there was a greater association with active coping to work towards making the

situation better, and planning by using strategies to resolve a stressor. Nurses high in openness

were also more likely to cope by using positive reframing in order to see stressors in a more

positive light. These personalities showed associations with coping strategies indicative of

problem focused coping to resolve the problem. Conversely, high neuroticism was strongly

related to venting as a coping strategy to verbally express negative feelings, reflective of an

emotion-based coping style aimed at reducing one’s negative feelings about the threat as

opposed to altering the source of stress itself (Burgess, Irvine & Wallymahmed, 2010).

In the research discussed in this dissertation, in addition to assessing personality and coping in

this cancer-related group, we addressed the role of personality among patients and caregivers as

a mediating factor on the association between chronotype and sleep quality on coping, and one’s

ability to handle the stresses associated with their role, both across the day and overall.

13

Understanding the association between chronotype, sleep quality and personality and their

influence on coping will allow for more tailored strategies to foster healthy coping strategies

among patients and caregivers with different chronotypes and sleep patterns, as well as

personality types.

1.3! Cancer as a Chronic Stress

Stress results when the demands or outcomes of a positive or negative situation impinge upon or

threaten one’s behavioural, emotional, or physical state. More specifically, stress is deemed as an

unpleasant negative emotional experience resulting in behavioural, biochemical, and

physiological changes intended to adapt to the stressor, either via its manipulation, alteration, or

accommodation (Baum, 1990). Stressors can be acute or chronic. Acute stressors are those that

are specific episodic events with relatively discreet onset and offset points, and are general

occurrences in everyday life (e.g., an interview, narrowly missing a car accident, etc.). Chronic

stressors refer to ongoing events that continue over a prolonged period of time (e.g., poverty,

long term illness). The body’s stress response system is activated when dealing with acute and

chronic stress to prepare the body for the challenges that must be faced. However, chronic long-

term activation of this system – at both a low or high stress level – is detrimental to one’s mental

and physical health, contributing to a range of health problems varying in severity (e.g., high

blood pressure, obesity, anxiety, and depression, etc.). Interestingly, events and factors that cause

stress for one individual may not produce the same stresses for another person. Different people

can react to stress in a variety of ways, with the distinction typically depending on one’s

perception of the stressor (e.g., can it be overcome, how will it affect life, etc.) (Baum, 1990;

Hammen, Kim, Eberhart & Brennan, 2009).

Cancer can be contributed to by the detrimental effects that stress causes on the body, and itself

acts as a long-term chronic stressor to both the patient and his or her caregivers, including

medical professionals, and friends and family members. For patients, the gravity and uncertainty

of cancer overall causes stress, but various forms of cancer and the according treatments may

cause their own specific stresses such as the loss of a body part, or changes in one’s sexual

functioning. For medical caregivers, being tasked with continuously providing care and

assistance to ill individuals, coupled with being charged with their care and not always seeing

14

favourable outcomes produces a chronic stress. As friends and family members providing care,

chronic stress can result for many reasons including constant worry for a loved one, and changes

to one’s own personal life including taking on a new role. In each case, one’s personal coping

strategies will determine how an individual will cope with and be affected by the stresses

associated with cancer.

1.3.1! Emotional Distress in Patients

It is well understood that in addition to the physical symptoms associated with breast cancer and

its treatment, there are increased risks for developing psychological problems and suffering

emotional distress (Barre, Padmaja, Saxena & Rana, 2015). Bultz and Carlson (2006) reviewed

the literature and found that in North America the incidence of distress among cancer patients

across the trajectory of the illness (diagnosis, treatment, survival, and or palliation) is between

35-45%. Emotional distress during these times can stem from a number of reasons such as worry

about one’s health and survival, changes in family life and or job, financial security, marital

problems, etc. Research indicates that when cancer patients at large experience greater emotional

distress, they are more likely to visit emergency facilities and community health services, thus

placing a larger burden on the healthcare system, while often leaving the emotional distress

unresolved (Carlson & Bultz, 2004). In addition to the economic burden that emotional distress

can cause when patients seek additional emergency or community care, the emotional distress

compounds with the physical symptoms patients may already be experiencing thus adding extra

burden to the situation. While a certain level of emotional distress is normal, and a patient can

still continue to function, emotional distress can often become all-consuming and debilitating.

Emotional distress may stem from poor coping, and may also fuel continued poor coping styles.

For these reasons, it is important to understand solutions to treat or alleviate emotional distress,

and maximize positive, proactive, and healthy coping strategies among patients.

1.3.1.1! Why Choose Breast Cancer Patients

Survival rates for patients diagnosed with stages I to III breast cancer have increased

significantly in recent decades, with five-year survival rates between approximately 72% to

almost 100% (Canadian Cancer Society, 2016b). While survival rates from breast cancer are

increasing, incidence of breast cancer is still relatively high; it was estimated that in 2016 alone,

approximately 25,700 Canadian women would be diagnosed with breast cancer (Canadian

15

Cancer Society, 2016a). Therefore, while breast cancer survival itself has increased significantly

in recent decades, it continues to be a commonly diagnosed cancer. Given the relatively strong

prognosis, early stage breast cancer patients are a unique population who can experience worries

and fears that contribute to emotional distress, but can also see an end to their treatment and the

eventual high likelihood of recovery. Conversely, among metastatic breast cancer patients, while

some treatment might be aimed at tumour reduction and slowed growth, patients typically

understand that their disease or complications from it will lead to eventual death. The eventual

“light at the end of the tunnel” so to speak that applies to early stage breast cancer patients allows

for a shorter period of study between diagnosis, treatment, and remission, and the according

changes in emotional distress that come with being diagnosed, seeing effects of treatment, and

beginning the road to recovery.

1.3.1.2! Specific Distress Involving Genetic Issues

Given the increased risk of developing breast cancer when one is a BRCA1 or BRCA2 carrier, it

is likely that these individuals may experience increased distress. While learning that one carries

a BRCA1 or BRCA2 mutation can potentially promote efforts towards breast cancer prevention,

concern over adverse emotional, cognitive, and behavioural consequences have also been

expressed (Lerman & Croyle, 1996, Lerman & Schwartz, 1993). Baum et al. (1997) suggest that

one’s level of emotional distress following genetic testing indicating that one is a carrier is

mediated by questions regarding disease occurrence, timing, potential severity, treatment course,

and potential for preventability. Other factors that may influence one’s distress relating to

carrying a BRCA1/BRCA2 mutation include quality of genetic counselling and medical

surveillance available, and age and gender. Time since testing also influences one’s emotional

distress related to being a carrier, such that with time, distress appears to decrease among

carriers, and return to pre-testing levels. Patients who receive confirmation of being

BRCA1/BRCA2 carriers with a personal cancer history experience less distress post diagnosis

than those without a personal cancer history (Hamilton, Lobel & Moyer, 2009).

1.3.2! Emotional Distress in Caregivers

1.3.2.1! Specific Distress Involving Partners

Patient’s partners are typically deeply affected by the cancer diagnosis and are often the patient’s

primary support or caregiver outside the hospital. Only recently has the emotional distress of

16

caregivers begun to be studied. Merckaert et al. (2013) reviewed the literature and found that

high levels of distress are experienced by between 10-50% of caregivers, and their distress level

is often similar to that of their partners. Partners have reported receiving less social support

compared to patients, and their distress increases with disease progression into palliation (Davis-

Ali, Chesler, & Chesney, 1993; Kurtz, Given, Kurtz & Given, 1994). In a recent study among

cancer patient caregivers, 85.6% reported unmet needs (i.e., the requirement or desire for the

provision of actions or resources to help achieve an optimal state of well-being), while 69.1%

reported positive for emotional distress, 26.5% for depression, and 34.9% for anxiety

(Sklenarova et al., 2015). Given the frequency of the issue, it is important to understand causes

of and changes in emotional distress among caregivers in order to provide the necessary tools to

alleviate these feelings.

1.3.2.2! Specific Distress Involving Oncology Staff

Hospital staff who must provide care to cancer patients are also known to experience emotional

distress, often in the form of burnout, fatigue, and low job satisfaction. Work stress and the

stressful situations associated that are faced by doctors, nurses, radiation therapists, and

pharmacists working in the field of oncology and dealing directly with patients can combine to

produce feelings of burnout and fatigue. Even among those oncology professionals who report

high levels of personal accomplishment, occupational distress levels are still rated as being high.

It is important to understand the causes of emotional distress and patterns of change in stress

management across the day in order to provide the necessary tools within the hospital setting to

reduce these negative feelings. Reducing these negative feelings will both reduce the number of

employees needing to take long-term leave as a result of work related burnout or fatigue. At the

same time, reduced emotional distress, burnout, and fatigue among home medical staff is known

to increase the overall satisfaction of their patients (Ruggieri, Zeppegno, Gramaglia, Gill,

Deantonio & Krengli, 2014; Vahey, Aiken, Sloane, Clarke, & Vargas, 2004).

1.4! Rhythmicity

1.4.1! Biological Clocks in Nature

Biological clocks that moderate physiology and behavior reflect cyclic changes in the

surrounding environment. The systematic changes can be moderated by the earth’s rotation, and

17

the sun and moon, and can range in duration based on daily (circadian), annual (circannual),

lunar (circalunar), or tidal (circatidal) cycle length. Rhythms provide a significant adaptive

advantage by allowing an organism to anticipate patterns such as food availability, mating

seasons, migratory periods, times to avoid predation by other organisms, etc. Circadian clocks

are the best understood sources of these daily rhythms. These regulate specifically the

endogenous cyclic changes exhibited by organisms from cyanobacteria to humans across the 24-

hour day. A defining characteristic of these clocks is that their action persists when the organism

is held in constant conditions (Chaudhury & Colwell, 2002; McClung, 2011; Reppert & Weaver,

2001).

1.4.2! Biological Clocks in Mammals

In mammals (including humans), the circadian system is a hierarchical organization of clocks

with a master oscillator that coordinates rhythmicity throughout.

The suprachiasmatic nuclei (SCN) of the hypothalamus has the primary responsibility of

synchronizing physiology and behavior with daily changes in the environment (Ralph, Foster,

Davis & Menaker, 1990). Environmental light received by specialized cells in the retina transmit

this information to the SCN, thereby entraining its rhythm according to the surroundings (Morin

& Allen, 2005).

It is believed that the two primary adaptive reasons for organisms to have biological clocks are

for organizational and anticipatory purposes. From an organizational perspective, biological

clocks allow for energy conservation by ensuring metabolic or behavioural processes are

sequenced to reduce energy use and increase overall efficiency. For example, in the evening, the

biological clock is organized to begin lowering overall body temperature to maximize the body’s

energy reserves; simultaneously, metabolism is also slowed as an organizational measure to

enhance energy conservation since the body does not require a constant elevated supply of

glucose to support activity. From an anticipatory perspective, biological clocks allow organisms

to predict changes both within the organism (serving again as an energy conservation measure)

and in the surrounding environment (to allow the organism to use energy efficiently, while

simultaneously protecting itself). For example, research in rats has demonstrated anticipatory

behaviours for meal times across the day. In free feeding conditions, rats with SCN lesions

demonstrated anticipated restricted food access at 24-hour intervals, regardless of the absence of

18

a functioning master circadian clock, suggesting circadian oscillators beyond the SCN are

entrainable by restricted feeding schedules (Stephan, Shwann & Sisk, 1979).

1.4.3! Biological Clocks in Human Beings

Virtually all cells and tissues within the human body express circadian rhythms, which influence

several bodily processes. At the molecular level, phase is determined according to a set of

canonical clock genes comprised of a transcription translation feedback (TTF) oscillation, that

may be set primarily by light-dark cycle, and influenced by a variety of nonphotic zeitgebers. At

the hormonal level, several rhythmic patterns of expression for various bodily hormones have

been studied. Most notably, the human melatonin cycle shows rhythmic oscillations across the

24-hour day with melatonin synthesis confined predominately to the subjective night (Arendt,

2006). A peak in hunger times across the day also appears to be tied to circadian rhythms,

allowing for a specific peak in evening hunger (dinner time) before the overnight fasting period

(Scheer, Morris, & Shea, 2013). Human performance on physical activities is also known to

fluctuate daily in accordance with circadian timing. For example, several aspects of sports

performance including flexibility and muscle strength show time-of-day variances tied to

circadian rhythms (Atkinson & Reilly, 1996).

1.4.4! Cognition, Emotion, and Circadian Rhythms Performance fluctuations across the day on higher level cognitive processes are driven by

circadian rhythms (e.g., Natale, Alzani, & Cocogna, 2003; Schmidt, Collette, Cajochen &

Peigneux, 2007). Cognitive function is an umbrella term referring to the general mental

processes by which information is acquired, and includes various categories including but not

limited to attention, memory, and language. Emotional regulation requires a high level of

cognitive functioning and control.

1.4.5! Chronotype

M versus E preference is often called chronotype, which broadly refers to self-perceived

changing performance capabilities across the day. Chronotype variances support the idea that

people have distinct innate personal preferences for carrying out tasks with varying demands

earlier or later in the day. These personal preferences may apply to several tasks, including those

with a cognitive, physical, or emotional basis. Furthermore, chronotype is known to shift across

19

the lifespan; children are typically M types, with tendency to shift towards E types with the onset

of adolescence followed by a cross-cultural shift back towards M type by approximately age 50.

(Schmidt, Collette, Cajochen & Peigneux, 2007). Chronotype refers to circadian phase and is

linked to the circadian clock. For example, preference for sleep timing has been correlated with

melatonin secretion profiles which are moderated by the circadian clock (e.g., Burgess & Fogg,

2008).

In humans, the most clearly visible and easily measurable display of rhythmicity is the daily

sleep-wake cycle. While individuals exposed to the same solar cycles should all be similarly

entrained to the daytime, human circadian behaviour research has demonstrated a significant

interindividual variability in the temporal relationships that exist with the light/dark cycle

(Roenneberg et al., 2007). Significant interindividual variability exists for the timing of sleep and

wake behaviour. People are commonly characterized as larks (early risers) and owls (late risers),

as a categorization for their M versus E preference, respectively. Kleitman (1963) found that

human sleep-wake patterns demonstrated that the existence of M and E type individuals is

correlated with “early” or “late” peaks in body temperature and performance efficiency curves

throughout the day.

1.4.5.1! Genetic Basis of Chronotype

There has been long standing interest in understanding the genetic basis for circadian

rhythmicity, and its contribution to chronotype – a behavioural phenotype. Chronotype is

normally distributed in the population across the age span, and many common genetic variants

with modest effects have been found to influence these phenotypes (Kalmbach et al., 2017).

Research indicates that a series of molecules function together in a feedback loop to form a core

circadian clock in every cell, and the SCN appears to maintain synchrony among these cellular

clocks (Kalmbach et al., 2017; Lowrey & Takahashi, 2011). Twin and family studies suggest that

genetic factors explain up to 50% of the population variability in chronotype (Kalmbach et al.,

2017).

1.4.5.2! Questionnaires

Several questionnaires have been devised to assess chronotype, however the gold standard in the

literature remains the Horne-Östberg Morningness Eveningness Questionnaire (MEQ) (Adan &

20

Natale, 2002; Horne & Östberg, 1976). The MEQ is a 19-item multiple choice measure that asks

respondents to indicate their optimal time of day for performing various activities between

morning and night, and yields a total score in a range of 16-86 (16 = extremely evening types; 86

– extremely morning type). More recently, the Munich Chronotype Questionnaire (MCTQ) was

devised, and suggests that the timing of midsleep is also an indicator of chronotype on work- and

free-days (Roenneberg, Wirz-Justice & Merrow, 2003). However, while a satisfactory

correlation exists between MEQ and the uncorrected-MCTQ score, the correlation diminishes as

corrections for various scenarios are applied to the MCTQ score (Di Milia, Adan, Natale, &

Randler, 2013).

1.4.5.3! Performance Variances

Following early research demonstrating that chronotype is associated with performance curves

across the day, further studies have examined this concept in greater detail. Basic performance

timing among athletes appears tied in some extent to chronotype, such that some physical

activity performance appears better at one’s optimal time of day (Brown, Neft, & LaJambe,

2008). In terms of the psycho-physical factors behind performing a physical activity, completion

of such tasks at one’s chronotypically optimal time is linked to increased enjoyment and

affective response, and general motivation to exercise. Engaging in physical activity during one’s

chronotypically preferred time of day is likely to predict greater long term adherence to an

exercise regime (Vitale, Colagiuri & Weydahl, 2013).

Changing cognitive function throughout the day is related to chronotype. Morning and evening

type individuals show optimal functioning of executive control abilities when completing

cognitive-based attentional tasks at their chronotypically optimal time of day; conversely,

executive control abilities are diminished at non-optimal times of day (Lara, Madrid & Correa,

2014). Performance on some memory-related tasks tends to decrease across the day in M type

individuals, and to increase across the day in E type individuals (Hasher, Zacks & Rahhal, 1999).

The same results have been found between different age groups, on a range of cognitive and

executive function type tasks (e.g., Hahn et al., 2012; May, Hasher & Stoltzfus, 1993). The

existence of optimal times for performance on lab-based tasks depending on one’s chronotype

and age indicates that this phenomenon exists throughout the lifespan. While these tasks can be

tested at controlled times in the day in a laboratory setting, it is important to consider that

21

optimal times of day exist in regular day-to-day life as well, and as such, performance variances

will exist here as well.

1.4.5.4! Emotionality

Chronotype is also related to emotionality. Studies have indicated that individuals with a greater

propensity towards morningness rate more highly in positive affect than E types, while those

with tendency towards eveningness display greater tendency for higher depression scores (Clark,

Watson & Leeka,1989; Hasler, Allen, Sbarra, Bootzin, & Bernert, 2010). When comparing

younger and older adults, M types from both age groups also rated higher on their overall

positive affect as compared to age matched E types (Biss & Hasher, 2012). Among young

teenagers with a tendency for eveningness, significantly more emotional and behavioural

problems were reported compared to age matched participants with a tendency for morningness

(Gau et al., 2007). Emotional dysregulation has also been linked to chronotype. Among patients

who have attempted suicide, eveningness tendencies were most common among those patients

who had attempted violent suicide, and rated highly on tests of impulsivity; similarly, suicidal

ideation is highest among patients with major depressive disorder and a preference for E (Bahk,

Han & Lee, 2014; Selvi et al., 2011). Aggression and antisocial behaviour also appear more

prevalent in individuals with a preference for E (Schlarb, Sopp, Ambiel & Grünwald, 2014). The

literature suggests E types are more likely to experience mental, physical and behavioural

disorders than M types. It is unclear whether this is purely based on conflicts between optimal

sleep-wake times versus societally imposed timing structures in daily work and life, or whether

an E chronotype itself is specifically associated with the onset of these disorders. It is important

to consider that morningness may provide a protective factor, and it is necessary to understand

how to protect E types against the onset of these issues. These findings show a clear link between

chronotype and emotionality. However, it remains to be seen how changes in one’s emotional

state across the day, particularly when considering coping, relate to chronotype.

1.5! Sleep

In today’s world, the traditional timing of human behaviours and activities linked to the sleep-

wake and light-dark cycles has been altered due to the use of artificial lighting sources. Societal

22

demands including work schedules, a global network and a wide range of modes of

telecommunication, active lifestyles, and daily tasks also contribute to these shifts in timing of

human behaviour. While it is understood that a need for sleep exists, many people, particularly in

the modern world, do not place high importance on the body’s need for sleep. In the 24-hour

world in which we live, night time sleep is often replaced with other activities, resulting in sleep

deprivation (Ferrara & De Gennaro, 2001).

1.5.1! Two-Process Model of Sleep

Adequate and good quality sleep is necessary for maintaining optimal cognitive, mental,

physical, and emotional health. There is debate regarding how much sleep is necessary each day,

with a wide range of hours of necessary sleep being reported by various researchers. While little

consensus exists on the general number of hours that should be necessary for proper daytime

functioning and health maintenance, on the whole, it is understood that sleep deprivation

negatively impacts mental and physical wellbeing (Alhola & Polo-Kantola, 2007; Ferrara & De

Gennaro, 2001). The most common theory of sleep regulation is the two-process model; it

suggests that sleep is dependent on both a homeostatic sleep process (which reflects the

increased need for sleep with sustained wakefulness), and a circadian process (which determines

a window during which sleep is most likely to occur). These two processes are believed to

interact to determine an individual’s sleep/wake cycle, and to influence changes in one’s level of

alertness and attention throughout the day (Achermann, 2004; Alhola & Polo-Kantola, 2007). A

simple result of environments (real world or experimental) that interrupt sleep is sleep

deprivation, which can drive a greater propensity for sleep during wake time. This is a natural

consequence of sleep disruption, not to be confused with insomnia, which can cause sleep

deprivation, but likely has a primary cause within the sleep regulation system itself. Whereas

insomnia refers to the inability to obtain adequate and or good quality sleep despite the

opportunity to sleep, sleep deprivation refers to truncated sleep length due to an environmental

circumstance that restricts the opportunity to sleep.

1.5.2! The Functions of Sleep

A key function of sleep – to regulate and restore brain energy – is well documented (Dworak,

McCarley, Kim, Kalinchuk & Basheer, 2010). In rats, sleep brings about a surge in adenosine

triphosphate (ATP), which acts as the energy currency of brain cells (Dworak et al., 2010). The

23

function of the hypothalamic pituitary adrenal (HPA) axis is also altered by partial sleep

deprivation. In a study of partial sleep deprivation (between 04:00 and 08:00 hours), the normal

quiescence of cortisol secretion showed a significant delay, producing elevated evening cortisol

levels on the following evening (Leproult, Copinschi, Buxton & Van Cauter, 1997). These

findings suggest sleep deprivation can have adverse health effects. In the United States, average

sleep duration decreased from between 8 to 9 hours in the 1960s, to approximately 7 hours in the

1990s, paralleling the increased prevalence in the country’s diabetes and obesity cases. A review

of the literature supports the notion that chronic partial sleep loss results in dysregulation of

glucose pathways and neuroendocrine control of appetite, resulting in overeating and reduced

energy expenditure, subsequently leading to an increased risk for obesity and or diabetes

(Knutson, Spiegel, Penev & Van Cauter, 2007). On the whole, this research indicates that sleep

not only affects brain function directly, but other downstream pathways tied into one’s health

and biological well-being.

1.5.3! The Impact of Sleep and Sleep Loss

In addition to a biological function, sleep also influences one’s performance and mental state.

Sleep quality is associated with the onset of, or prevention against a variety of disorders.

Individuals with poor sleep quality are significantly more likely to experience at least one

common mental disorder compared to individuals who report good quality sleep (Rose et al.,

2015). Problems with cognition have also been reported in the literature among individuals with

poor sleep quality and particularly among those with untreated sleep apnea, which is known to

cause reduced sleep quality (Vaessen, Overeem, & Sitskoorn, 2015). As with chronotype, it is

evident that sleep quality is also a significant contributor to both one’s health and ability to

function throughout the day.

1.5.3.1! Sleep, Cognition, and Memory

In response to demanding work, school and social schedules in daily life today, many people

trade sleep for additional time to devote to these activities, resulting in high occurrences of sleep

deprivation in modern society. Insufficient sleep dramatically impacts alertness and attention,

without which it is extremely difficult to carry out complex cognitive processing (Lim & Dinges,

2008; 2010). In a study of sleep deprivation using a psychomotor vigilance test among otherwise

healthy subjects, a general slowing of response times and average reaction time was observed

24

(Lim & Dinges, 2008). Sleep restriction research has demonstrated that compared to individuals

sleeping 8 hours, those on a restricted sleep cycle of 6 hours for 2 weeks demonstrated cognitive

performance deficits equivalent to 2 nights of sleep deprivation. Furthermore, subjects on a 6-

hour restricted sleep cycle were relatively unaware of the cognitive deficits they were displaying,

which may explain why many members of the general public assume the effects of sleep

restriction are relatively benign (Van Dongen, Maislin, Mullington & Dinges, 2003).

Furthermore, it seems age may influence the extent to which psychomotor vigilance on

attentional tasks is affected by sleep deprivation. A review of the literature indicates that among

younger adults, sleep deprivation results in poorer performance and slowed reaction time on

vigilance tests, along with unintentional sleep episodes during sleep deprivation or restriction

periods as compared to older adults. Conversely, older adults may show slower reaction times on

such measures, yet compared to younger adults, their overall performance is less impaired by a

night or sometimes more of lost sleep (Killgore, 2010). In general, cognitive processing appears

reduced across the lifespan when an individual loses sleep, however the extent of disruption may

vary by age group.

Sleep loss is also known to affect one’s emotional functioning and processing abilities, which

subsequently alters associated cognitive processes such as memory for events, judgment and

decision making skills. Sleep deprivation specifically appears to result in greater intolerant and

unforgiving behaviour, along with more self-focused tendencies, compared to when one is fully

rested (Killgore, 2010). Sleep plays a role in one’s emotional evaluations. Individuals on a sleep

deprivation schedule for one night reported greater negative mood and perceived neutral images

significantly more negatively compared to individuals on an undisrupted sleep schedule. The

negative emotional bias for neutral stimuli did not appear related to the negative mood known to

typically follow sleep loss (Tempesta et al., 2010). Furthermore, research has also demonstrated

that sleep deprivation results in a greater tendency towards negative cognitions and intolerance

for frustrating social scenarios or greater emotional reactivity to threatening visual stimuli. In a

two-night sleep deprivation study, participants were asked to respond to a cartoon demonstrating

two characters experiencing a frustrating situation. Sleep deprived participants were more likely

to direct blame or show hostility towards other people or objects in the environment, while

simultaneously showing less willingness to suggest amends that would be mutually satisfying for

both parties. These findings were indicative of a weakened inhibition of aggression and poor

25

willingness to behave in ways that would increase effective social interaction (Kahn-Greene,

Lipizzi, Conrad, Kamimori & Killgore, 2006). In studies of REM sleep deprivation, emotional

responses to visually threatening stimuli were also enhanced relative to baseline after one night

of REM-deprivation. This work highlights the specific behavioural regulatory role played by

REM sleep in order to modulate emotional responsivity (Rosales-Lagarde et al., 2012). In

general, it appears that sleep deprivation plays a key role in regulating various aspects of mood

and emotional processing, particularly in elevating negative-response styles in reaction to

potentially trying situations.

1.5.3.2! Polysomnography and Actigraphy

Objective sleep measures often require participants to use polysomnography, and sleep in a

laboratory setting, many times for more than one night. Among patients receiving medical care,

particularly those being treated for a chronic illness (e.g., cancer) some researchers may use

alternate measures to assess sleep quality without having individuals change their routine and

sleep away from home. While not as detailed as polysomnography, actigraphy can be used as a

noninvasive measure of one’s daytime versus nighttime activity and sleep across the 24 h day.

Actigraphy devices are typically worn on the non-dominant wrist and record motor activity

across the day. Sleep estimates are made based on the amount of movement versus rest. The data

are translated into a histogram, with the x-axis representing a 24 h time period, and the y-axis

representing activity frequency. However, unlike polysomnography, actigraphy does not detail

the various stages of sleep that one passes through, or the amount of time spent in each state.

Nonetheless, actigraphy is still deemed an effective and accurate tool to measure circadian

activity rhythms, and has been especially noted for use among cancer populations due to its

nonintrusive nature. (Fiorentino & Ancoli-Israel, 2007; Roscoe et al., 2007).

1.5.3.3! Sleep Quality

Subjective sleep measures are not based on physiological measures such as REM sleep, or

changes in hormone levels, but rather on one’s self-reported ratings of their sleep. Various

measures exist, and can be as simple as a sleep log – where an individual tracks their daily sleep

and wake times, or can include retrospective questions requiring an individual to reflect on

various aspects of their sleep (i.e., duration, number of awakenings, body temperature sensation,

etc.) over an extended period. Subjective sleep measures reporting on one’s own sleep appear to

26

reflect objective polysomnographic data (Armitage, Trivedi, Hoffmann & Rush, 1997), however,

subjective sleep quality reports on the sleep of another individual (e.g., by parents on their

child’s sleep), were less verified by objective measures (Choi, Yoon, Kim, Chung & Yoo, 2010).

The Pittsburgh Sleep Quality Index (PSQI) (Buysse, Reynolds III, Monk, Berman, Kupfer, 1988)

is a commonly used subjective sleep measure. It calculates a score based on one’s recall for sleep

quality, latency, duration, efficiency, and disturbances, along with one’s use of medications and

daytime dysfunction over the past month. It is a short questionnaire, and has good reliability and

validity. It is intended for use in research and clinical settings, and is intended for easy patient

use.

1.5.3.3.1! Cancer

Sleep quality refers to both objective and subjective aspects of sleep. Objective characteristics of

sleep quality can include factors such as sleep duration, latency, and number of arousals, while

subjective aspects refer to the degree to which sleep is restful or deep (Buysse, Reynolds III,

Monk, Berman, Kupfer, 1988). Sleep quality is associated with sleep disorders. Sleep disorders

refer to issues such as difficulty falling asleep or maintaining sleep, low sleep efficiency, and

insomnia, and can become chronic. Sleep quality among cancer patients as a collective group is

typically reported as poor. There is a high prevalence of problems with achieving and

maintaining sleep, low sleep efficiency (amount of time spent asleep while in bed), along with

high levels of sleepiness during waking hours (Fiorentino & Ancoli-Israel, 2007). In a telephone

survey study of 150 lung and breast cancer patients reporting on their sleep quality over the past

month, 44% of participants reported sleep problems, yet only 16.6% of individuals reported these

sleep complaints to their health care providers (Engstrom, Strohl, Rose, Lewandowski &

Stefanek, 1999). While the importance of sleep is generally well understood, this research

indicates that the prevalence of sleep problems and general poor sleep quality among cancer

patients is often not well understood and subsequently left unaddressed.

It is important to also distinguish between cancer related fatigue and cancer related sleep

disorders. Cancer related sleep disorders are those that may result due to the effects of the

disease itself or its treatment (e.g., chemotherapy, radiation, surgery, etc.). Cancer related fatigue

however refers to the feelings of tiredness, weakness, and or low energy, that remain unrelieved

by good quality sleep or even rest. These feelings can occur both as a side effect of the disease

27

itself and or its treatment, however the exact cause is unknown. Poor sleep quality and sleep

disturbances or problems are known to affect patients at all stages of treatment, and are often

reported even following treatment, indicating that poor sleep is often a chronic problem within

this population (Fiorentino & Ancoli-Israel, 2007; Hofman, Ryan, Figueroa-Moseley, Jean-

Pierre, & Morrow, 2007; Roscoe et al., 2007).

1.5.3.3.2! Breast Cancer Patients

Much of the research on cancer patient sleep quality has focused on breast cancer patients. In a

sample of nonmetastatic breast cancer patients recruited during routine follow-up exams after

having completed treatment, the average sleep score on the PSQI was 7.0, which is higher than

the recommended cutoff for good quality sleep as determined by the questionnaire. This

indicates that on average, breast cancer patients report significant sleep problems and poor sleep

quality (Carpenter & Andrykowski, 1998). In a study of 72 women with breast cancer recruited

at various time points of the diagnosis, treatment, and recovery trajectory, 61% reported

significant sleep problems on the PSQI. In particular, sleep in this population was characterized

by reduced total sleep time, pain, nocturia, hot flashes, and trouble breathing caused by coughing

or snoring (Fortner, Stepanski, Wang, Kasprowicz, & Heith Durrence, 2002). These data confirm

previous research indicating that sleep quality is often poor among breast cancer patients.

Insomnia is often reported as a common contributor to poor sleep quality among breast cancer

patients. Specifically, it is characterized by trouble initiating or maintaining sleep, non-

restorative sleep, must persist for at least one month, and must result in distress or interfere with

one’s ability to function. Insomnia as a contributor to poor sleep quality may be more common

among women with breast cancer for three key predisposing, precipitating and perpetuating

reasons. Female gender is a known predisposing factor in the general population to cause

insomnia. Among breast cancer patients, female gender may contribute to higher than average

rates of insomnia, together with factors such as increasing age. Depression is also known to be

more common among females and has a strong link to insomnia. Given the high number of

breast cancer patients reporting depression (on average approximately 20 – 30%), this may be a

common concomitant health mental health issue that contributes to insomnia and poor sleep

quality in this population. Insomnia may be precipitated by treatment side effects which may

include increased frequency and severity of hot flashes due to the sudden onset of drug induced

28

menopause. Finally, it may be perpetuated by the disruptions to the typical sleep-wake cycle,

such as frequent napping or desynchrony of one’s circadian clock. Given the number of common

factors that can contribute to the onset of chronic insomnia, and its negative effects on sleep

quality, it is especially important that the skills to prevent insomnia onset be taught to cancer

patients. Both cognitive and behavioural therapies separately and in combination have been used

with success to teach patients the necessary skills to avoid and or combat the onset of insomnia

(Bower, 2008; Fiorentino & Ancoli-Israel, 2007).

1.5.3.3.3! Caregivers

Familial cancer patient caregivers play a key role in providing care to cancer patients in both the

home, and additional assistance even in a hospital or hospice setting. The difficulties faced by

caregivers are believed to contribute to the sleep disturbances they face during the cancer care

process. Research has indicated that key worries, fears and emotional burdens of cancer

caregivers relate to the nature and metastasis of the patient’s disease, helping patients cope with

or face the emotional consequences of the illness, and managing life disruptions caused by

cancer (Aslan, Sanisoglu, Akyol, & Yetkin, 2009; Osse, Vernooij-Dassen, Schadé, & Grol,

2006). These worries, fears, and burdens are believed to contribute to sleep disturbances found to

be common among cancer caregivers such as insomnia, nocturnal sleep disturbances, chronic

sleep loss. Furthermore, sleep problems are found to persist among several caregivers following

the death of a loved one, particularly if the patient’s symptoms were unrelieved during their final

three months of life (Berger et al., 2005). These sleep disturbances among caregivers are known

to negatively impact on the individual’s ability to carry out daily tasks and activities, quality of

life, emotional status, and performance as a caregiver. Conversely, improved sleep quality allows

caregivers to improve the quality of care which they provide to patients (Aslan, Sanisoglu,

Akyol, & Yetkin, 2009). Given the positive versus negative effects that varying degrees of sleep

quality can have both directly on the caregiver and on the patient, it is important to understand

specifically how functionality is affected across the day, particularly as it relates to one’s

circadian rhythm. A better understanding of the variances in caregiver performance as they relate

to circadian rhythms will produce a clearer picture of performance variances across the day, at

both the physical and emotional levels of functionality.

29

1.5.3.3.4! Oncology Staff

Far less research has focused specifically on the role of sleep quality among oncology staff (i.e.,

oncologists, and nurses and pharmacists working in oncology). Among the general population of

nurses, independently of the stressful work situations faced on a daily basis, sleep debt is known

to result in feelings of physical and emotional fatigue, reduced cognitive functioning, and a

general sense of burnout (Bellicoso, Ralph, Trudeau, 2014). In general, physicians also obtain

inadequate sleep, with some even working more than 24 consecutive hours. These interrupted

sleep cycles result in disrupted circadian rhythms. Evidence has demonstrated that this loss of

sleep experienced by many physicians can directly affect their personal health as well as the

health and safety of patients. A review of the literature indicates that lack of sleep among

physicians has been associated with several negative outcomes including misinterpretation of

patient records and scans, increased procedure error rates, and lack of empathy towards and poor

communication with patients (Eddy, 2005). Given the negative side effects sleep loss and poor

sleep quality can have on one’s self and the patients one cares for, sleep quality among oncology

caregiving staff in hospitals should be examined relative to changes in their performance ability.

A better understanding of performance fluctuations, and how they are related to sleep quality can

yield solutions to maximizing oncology staff performance and quality of patient care.

1.6! Breast Cancer Background

1.6.1! Cancer Biology

While several forms of cancer exist, in general terms, cancer is a disease that begins in the cells.

The normal cycle of cellular reproduction involves cellular division and subsequent death of

older cells. When this typical cycle is interrupted, uncontrolled cell growth results, with cells

producing tumours, lumps, or spreading through the bloodstream and lymphatic system to other

areas of the body. Tumours that stay in one controlled section of the body are known as benign,

and are typically non-life threatening. However, malignant (spreading) tumours have the

potential to invade nearby tissues and travel to other areas of the body. Once a malignant tumour

has travelled to other areas, these new tumours are known as metastases. Even though tumours

have the ability to metastasize to other body areas, the particular cancer is named after the

30

location in the body where it originated (e.g., breast cancer begins in the breast but can

metastasize outside the breast tissues).

1.6.2! Breast Cancer Biology

Breast cancer involves the growth of a malignant tumour that originates in any of the cells of

breast tissue. The breast consists of both fatty and connective tissue, and contains lobules (milk

production glands) and ducts (pathway for milk transport between lobules and nipple); breast

cancer can originate in any of these tissues. The nipple and areola are also susceptible to the

development of breast cancer, however development of cancer is more rare in this tissue. While

significantly less common, men can also develop breast cancer in the cells of their limited breast

tissue.

During replication, cells of the breast tissue can sometimes become mutated, which can result in

the irregular proliferation of cells. These irregular cells may create benign breast growths (e.g.,

cysts, fibroademonas, or benign tumours), or sometimes result in malignant breast tumours. Most

commonly, irregular cell growth is observed in the milk ducts, but can also occur in any other

breast tissue. These malignant tumours have the potential to grow in size as a solid mass, to

produce several smaller tumours in the breast, to spread into the lymph nodes in the underarm, or

to break off and travel to distant areas of the body, including other organs, the bones, or the

brain. So long as disease spread is confined to the breast and or nearby lymph nodes of the

underarm, the disease is considered to have the possibility of being cured. However, once the

tumour metastasizes into distant areas of the body, care is considered palliative.

Several types of breast cancer exist; however, some forms are better understood than others.

Certain tumour-specific biological factors play a role in determining tumour cell proliferation

and subsequent treatment. Circulating hormones are a key contributor to many types of breast

cancer. Estrogen is known to play a key role in the promotion of both normal and abnormal

breast cell growth, and an expression of estrogen receptors is believed to account for up to

approximately 70% of breast cancers (Russo & Russo, 2006; Shi, 2015). These breast cancers

are known as ER-positive, and can be treated with hormones to prevent recurrence. Progesterone

receptor (PR) positive tumours can also be treated with hormone suppression therapy, and are

known to have a better long-term prognosis than PR-negative breast cancer tumours (Arpino et

al., 2005). The over expression of human epidermal growth factor receptor 2 (HER2) in breast

31

tumour cells results in rapid cell division and causes faster tumour cell growth and replication

when left untreated. Over expression of the HER2 gene is exhibited in approximately 25% of

breast cancers and can be suppressed by the anti-HER2 antibody Trastuzumab (Hicks &

Kulkarni, 2008). Breast tumours that are ER-negative, PR-negative, and HER2-negative are

known as triple negative breast cancers.

1.6.3! Genetics and Mutations

In addition to the many forms of breast cancer, the mutation of certain genes are also known to

influence one’s risk of developing breast cancer. When functioning normally, BRCA1 and

BRCA2 produce suppressor proteins that function to repair DNA damage. When an individual

has a mutation in either of these genes, the person is at an increased risk of DNA not being

properly repaired, resulting in cells being more likely to develop additional genetic alterations

that increase one’s risk of developing breast or ovarian cancer. These harmful BRCA1 or BRCA2

mutations are heritable, and can be passed on from one’s mother or father. In the United States,

approximately 12% of women in the general population will develop breast cancer at some point

in their lives (Howlader et al., 2014). Instead, by the time they are 70 years old, women who

have a BRCA1 mutation are estimated to have a 55-65% chance of developing breast cancer, and

those with a BRCA2 mutations are believed to have a 45% chance of developing breast cancer

(Antoniou et al., 2003; Chen & Parmigiani, 2007). These are only estimates, and further research

may indicate differences in these numbers. Genetic testing is available to detect these mutations.

1.6.4! Incidence

Incidence refers to the total number of new cases of breast cancer being diagnosed and does not

include the number of deaths caused by the disease. The most recent statistics by the Canadian

Cancer Society (2016a) indicate that in 2016, it was estimated that 25,700 new cases of breast

cancer would be diagnosed, making up approximately 26% of the cancers diagnosed in all

women in Canada. As of 2009, it was believed that Canadian women had a one in nine chance of

developing breast cancer (Canadian Cancer Society, 2016a).

With the increasing population in Canada, the number of breast cancer diagnoses has

increased since the 1980s, however the increased incidence rate has paralleled the rate of

population growth (Canadian Breast Cancer Foundation, 2016). Data for 2016 estimated that

32

approximately 51% of new breast cancers diagnosed in Canadian women would be seen between

those aged 50 to 69 making this the most common age to receive such a diagnosis. It was also

estimated that approximately 32% of newly diagnosed breast cancers would be found among

those over 69 years of age, while roughly 17% of diagnoses would be for individuals less than 50

years old. While the incidence of breast cancer among Canadian women under 50 is relatively

lower than that among older women, the disease tends to be more aggressive among this cohort.

For example, only 12% of cancer related deaths in women over 60 are due to breast cancer,

while 22% of cancer related deaths among women 30-59 are due to breast cancer (Canadian

Breast Cancer Foundation, 2016).

The detection of breast cancer at earlier stages and greater prevention of deaths caused by

breast cancer can be attributed to both a better understanding of the biology of various breast

tumours, along with the development of modern screening technologies and treatment.

1.6.5! Diagnosis and Treatment

1.6.5.1! Staging

Breast cancer staging is significant to one’s prognosis and determines treatment plan.

Classification of breast cancer staging, according to the American Joint Committee on Cancer

(AJCC) is based on a combination of three key components (tumour size, regional node

involvement, and metastases status) (Edge et al., 2010, p. 347-376). Additional details regarding

hormone receptor status of the tumour tissue, human epidermal growth factor receptor 2 (HER2)

status, and the patient’s menopausal state and general health will also affect one’s treatment plan,

but are not related to staging. One’s stage is generally characterized based on TNM

classifications, which include:

Tumour (T) Classification:

T0 No evidence of primary tumour

Tis Carcinoma in situ

T1 Invasive tumour ≤ 2cm

T2 Invasive tumour > 2cm but ≤ 5 cm

T3 Invasive tumour > 5cm

33

T4 Invasive tumour of any size, either attached to or invading surrounding tissues outside the breast, axillary and mammary lymph nodes; includes inflammatory carcinoma

Lymph Node (N) Classification:

N0 No detectable disease spread to regional tissue; node negative

N1 Spread to underarm lymph nodes; not fixed to other nodes/tissue; node positive

N2 Spread to underarm lymph nodes; fixed to other nodes/tissues; node positive

N3 Spread to breast bone or distant underarm lymph nodes; fixed to other nodes/tissues; node positive

Metastasis (M) Classification:

M0 No evidence of disease outside the breast and nearby region (clinically or radiographically)

M1 Detectable disease spread > 0.2mm beyond the breast and nearby underarm region

This TNM information is then reduced into categories known as anatomic stages or prognostic

groups. A general breakdown of these stages includes:

Stages:

Stage 0: Tis N0 M0

Stage I: T1 N0 M0

Stage IIA: T0-2 N0-1 M0

Stage IIB: T2-3 N0-1 M0

Stage IIIA: T0-3 N1-2 M0

Stage IIIB: T4 N0-2 M0

Stage IIIC: Any T, N3 M0

Stage IV: Any T, Any N, Any M1

Additional further breakdowns of stage I includes invasion of micrometastases to nearby lymph

nodes.

34

1.6.5.2! Procedures

To determine TNM staging, various procedures are conducted to assess tumour size, node

involvement, and if necessary, check for distant metastases. Various procedures may be used for

staging upon determining the presence of breast cancer.

Axillary Lymph Node Dissection: An axially lymph node dissection involves the removal of

lymph nodes from the underarm to verify whether cancer cells have spread outside the breast to

this region. This is a more aggressive lymph node dissection procedure, compared to a sentinel

node biopsy, which involves removing fewer nodes.

Sentinel Node Biopsy: The sentinel node is the first lymph node in a collection of nodes which

lymph fluid surrounding a tumour passes through; cancer cells typically spread to the sentinel

node first. A sentinel node biopsy involves the injection of blue radioactive dye into the breast

tissue surrounding the tumour area, which passes to the sentinel nodes. The node/nodes stained

blue are removed and assessed for presence of cancer cells. If no cancer is detected, the

remaining nodes are left in place; if cancer is detected, further lymph node dissection is required.

This is less aggressive lymph node removal method.

Ultrasound, x-ray, CT scan, MRI, and Bone Scan: These procedures assess the presence of

distant disease metastases outside the breast and nearby lymph node region to other areas

including bones and other organs or tissues.

Measurement of Tumour Size: This measurement reflects the longest dimension of the tumour.

1.6.5.3! Other Tumour Characteristics

As briefly discussed in section 1.6.2 Breast Cancer Biology, in addition to TNM staging, certain

biological traits concerning tumour make up must be consider as they will affect one’s treatment

plan and subsequent disease-free survival. Understanding the ER and PR status of a tumour will

indicate the degree to which a tumour will likely respond to hormone suppression therapy. ER-

positive and PR-positive tumours typically have a greater association to longer periods of

disease-free survival when compared to any combination of tumours that are negative for either

one or both of these hormone receptors. The benefit of an ER-positive PR-positive tumour is due

to the ability to take hormone suppression drugs that will prevent either the release of hormones

35

or subsequent uptake by neoplastic breast cells, thereby reducing the chance of tumour cell

growth. Hormones can sometimes be an alternate to chemotherapy and sometimes even radiation

if a tumour is small enough, however, they can also be taken in conjunction with treatment plans

involving chemotherapy or radiation. Over-expression of HER2, while typically associated with

more aggressive, fast growing tumours can now also be suppressed using Trastuzumab

(Herceptin ®), thereby preventing the expression of the tumour cell growth factor and reducing

the risk of recurrence. Trastuzumab is given in conjunction with chemotherapy, as it is most

effective when combined. Tumours that do not display an over-expression of ER, PR, or HER2,

known as triple negative breast cancers, cannot be treated with hormone suppression therapy or

growth factor antibodies. Instead, the only treatment options are surgery, radiation and

chemotherapy. Typically, triple negative tumours are known to have a worse disease-free

survival rate in the one to four years following initial diagnosis due to a high rate of recurrence

when compared to tumours with positive ER/PR status and or HER2 status. However, long-term

disease-free survival after eight years among triple negative disease survivors appears to be very

good (Dent et al., 2007).

When the benefit of chemotherapy is uncertain, in some cases the Oncotype DX ® (a gene

expression profiling test) may be conducted to assess whether such treatment will provide a

significant additional survival benefit. In Canada, the Oncotype DX ® test is given only to

patients with breast cancer that is hormone receptor positive, HER2 negative, and node negative.

Such patients have the potential to be successfully treated using some combination of surgery,

radiation, and or hormone treatment. However, the Oncotype DX ® test can assess whether

chemotherapy may provide an additional survival benefit based on the analysis of a sample of

tumour tissue. The test assesses the biology of each tumour tissue sample by assessing the

activity of 21 genes, and using these results to calculate a recurrence score on a scale of 0 (no

likelihood of recurrence) to 100 (certain chance of recurrence) if chemotherapy is or is not given.

Chemotherapy’s benefit is sometimes debated given the immediate and long-term side effects

that a patient may suffer. As such, it is important to ensure that only patients who will likely

benefit from chemotherapy receive the treatment. The Oncotype DX ® test can help to increase

both the doctor’s and patient’s confidence in the decision of whether or not to use chemotherapy.

36

1.6.5.4! Treatment

Patients receive individual medical treatment plans based on the particular type of cancer they

have, and the characteristics and stage of the tumour. The goals of any method of treatment

include controlling the spread of and potentially curing the current cancer, and preventing its

future recurrence. When a cure is not possible, treatment may be provided to shrink and reduce

the spread of the disease, reduce cancer related symptoms, prolong life, and increase quality of

life. Treatment may involve local or systemic therapies.

1.6.5.4.1! Localized

Local treatment is targeted to the specific area where cancer is present. Surgery and radiation

therapy are considered local treatments. Depending on cancer stage, one or both treatments may

be used.

Surgical: Various surgical procedures exist for treating breast cancer. The end goal of surgery is

to remove the tumour and achieve clean surgical margins with no disease tissue remaining.

Breast conserving surgery (lumpectomy) is often recommended when a tumour is small and

enough healthy breast tissue will remain to achieve a desirable cosmetic effect. A lumpectomy is

generally accompanied by a sentinel node biopsy and or an axillary node dissection.

Alternately, a mastectomy may be conducted to remove the entire breast. Typically, a

mastectomy is carried out for a larger tumour or higher stage, or if someone has a small tumour

but wishes to avoid radiation therapy. The 3 options for mastectomy include:

•! Modified Radical Mastectomy: The entire breast and nipple, some of the lymph nodes,

and the chest muscle lining (fascia) are removed. Typically, the nerves and chest muscles

are left intact and not removed. This is the most common form of mastectomy today for

invasive tumour treatment. Sentinel node biopsies have replaced more extensive removal

of lymph nodes.

•! Radical Mastectomy: This is no longer the typical surgical procedure for a mastectomy

today as it does not provide a significant survival benefit over the modified version and

can greatly reduce one’s range of motion and posture, but may be used if there is tumour

extension into the chest wall muscle. The entire breast and nipple, all the lymph nodes,

37

and chest muscle fascia are removed. Much of the chest muscle and nerves are also

removed.

•! Total Mastectomy: The least invasive mastectomy, it involves removal of the entire

breast, nipple, and chest muscle fascia. The chest muscle and nerves along with lymph

nodes are left intact. The skin of the breast may also be spared if the patient is opting for

a breast reconstruction. This procedure is either conducted prophylactically or if the

cancer is early stage or non-invasive.

Radiation: For treatment of breast cancer, radiation may either be used following surgery and or

chemotherapy to ensure that any residual tumour cells at the original tumour site are destroyed,

or to shrink tumours either in the breast or in areas of the body with distant disease spread in

patients with metastatic stage IV disease. While in general radiation for local control and

prevention acts as a lifesaving measure, at times patients do not undergo radiation because the

benefits they receive will be outweighed by secondary effects that may develop as a result of

radiation including damage to the cardiac muscle or increased risk of a secondary cancer.

Radiation may be avoidable in those who presented with a very small early stage tumour, and/or

those who subsequently had a mastectomy for a very small tumour, and/or those who had a high

dose of chemotherapy for a small tumour with favourable characteristics. The typical course of

radiation therapy for early breast cancer treatment in Canada is external beam therapy that

involves daily exposure to targeted doses (between 16 and 25) of radiation aimed specifically to

the tumour site.

1.6.5.4.2!Systemic

Systemic therapy is aimed to care for the whole body, preventing tumour recurrence at both the

original site and distant sites. Chemotherapy and hormone therapy are systemic treatment options

for breast cancer. Depending on the stage and characteristics of one’s tumour, neither, one, or

both of these treatments may be necessary.

Chemotherapy: Chemotherapy is used to kill cancer cells that are actively dividing in order to

damage the cells and reduce their ability to reproduce and spread. Several different

chemotherapy regimes may be used for breast cancer depending on the tumour staging and

characteristics, but most are given in a regimented cycle over a specific number of weeks or

months meant to interrupt the cancer cell division and replication cycle. Chemotherapy can be

38

used adjuvantly to prevent recurrence locally and distantly following surgery and tumour

removal by destroying any remaining cancer cells, or neoadjuvantly to shrink a tumour prior to

surgery and kill any microscopic distant metastases. Neoadjuvant chemotherapy is often used in

breast cancer cases where the tumour is too large to conduct surgery without causing significant

damage to the surrounding tissue, when the tumour has affected too much tissue to properly

close the surgical wound, or at times when a surgery date cannot be scheduled soon enough.

Neoadjuvant therapy also allows for the measurement of tumour response to the chemotherapy

drugs – if there is no response to the treatment, it can be stopped and a patient can proceed to

alternate modes of treatment such as surgery, radiation, or hormone treatment. Neoadjuvant and

adjuvant chemotherapy are equivalent in terms of survival outcome. Chemotherapy may not

always be necessary depending on the particular tumour characteristics and may be avoided in

cases where its immediate side effects or potential future complications (e.g., reduced cardiac

function, organ damage, increased risk for secondary cancers) may outweigh the benefits it

provides.

Hormone Therapy: Hormone therapy is a long term, non-cyclic treatment available as treatment

to prevent breast cancer recurrence. The typical hormone targeted by such therapy is estrogen.

These drugs may be used to prevent or slow the growth of cancer cells that are otherwise fueled

by the presence of estrogen. Depending whether a breast cancer patient is pre- or post-

menopausal, tamoxifen may be given to block hormone binding to estrogen specific receptor

sites, or an aromatase inhibitor may be given to prevent estrogen production.

Targeted: As a greater understanding of the specific cells associated with particular cancers

develops, it is possible to create therapies that target these particular aberrant cells fueling the

growth of particular tumours. These treatments, known as targeted therapies, are meant to

specifically attack cancer cells while generally sparing the normal cells. Targeted therapy can be

used in the treatment of some breast cancers where there is an over expression of HER2

receptors (HER2 positive) on the cancer cell surface that facilitate rapid cell proliferation.

Trastuzumab can be used to block HER2 receptors in patients with HER2 positive breast cancer,

and thereby slow or prevent the proliferation of cancer cells. Trastuzumab is given intravenously,

and is given in conjunction with chemotherapy drugs.

39

1.6.5.5! Prognosis

Disease prognosis can play a role in the way patients and their caregivers cope with the

diagnosis. For breast cancer, several factors affect their prognosis. Stage is a key prognostic

factor such that a diagnosis between stages I to III, even with increasing severity, still allows for

treatment options intended to be curative. A stage IV breast cancer diagnosis means treatment

will be to prolong life, but will ultimately not cure the disease. In general, across all stages, the

higher the stage at time of initial diagnosis, the worse one’s prognosis, but this does not mean

long-term survival is not possible. This indicates that greater tumour size and or greater nodal

involvement is associated with a greater chance of local or distant disease recurrence.

Additional factors outside TNM staging include grade (rate of tumour cell reproduction) and

type (e.g., ER/PR/HER-status) of tumour. Higher grade, poorly differentiated cell structure, or

inflammatory type of tumour are typically faster growing and have a worse prognosis. Greater

invasion of the blood vessels around the tumour results in a generally poorer prognosis.

Hormone receptor status affects prognosis, such that greater estrogen receptor status is associated

with a better prognosis, while breast tumours with an overall negative hormone receptor status

typically have a greater risk of recurrence. If distant metastasis does occur, those with hormone

receptor positive tumours typically exhibit first metastases to the bone, while those with hormone

negative tumours exhibit first metastases more commonly to the lung or liver. A more recently

understood prognostic factor is HER2 status. Until recently when the Trastuzumab drug became

available, HER2 positive tumours had a worse prognosis because they exhibited over expression

of a growth promoting protein produced on the tumour cell surface. Trastuzumab blocks HER2

positive receptors to slow and reduce the division of the corresponding cells. Finally, age at

initial diagnosis affects prognosis, such that those of ≤ 35 years of age typically have a worse

prognosis including greater risk of recurrence given the often more aggressive, higher-grade

nature of the tumour, than that typically seen among breast cancer patients who are older and

postmenopausal at time of diagnosis.

40

Chapter 2!

! General Methods This research collectively assessed how coping ability among patients and caregivers (medical

and familial) facing a common stressor (cancer) was influenced by chronotype, sleep quality, and

personality. Individuals in different positions but facing a common stressor were hypothesized to

react in ways specific to their roles. Both cross-sectional and longitudinal data collection

schedules were used. Oncology staff completed cross-sectional assessments of their coping

(based on measures of burnout, compassion fatigue (CF), and compassion satisfaction (CS)).

Patients and caregivers partook in longitudinal data collection schedules. Among each group

involved, specific surveys were used to collect data. Using these questionnaires, we developed an

assessment of how individuals in each group coped with cancer in relation to their chronotype,

sleep quality, and personality. The following is a description of the individual groups involved in

this study, and the various questionnaires used to assess participants across the study.

2.1! Procedures

2.1.1! Nursing Study

A literature search was conducted to verify a gap in the research on the influence of chronotype

and sleep quality on burnout among oncology nursing staff. A set of surveys were compiled to

assess demographic information and job satisfaction, chronotype, sleep quality, and burnout.

Ethics approval was received from Sunnybrook Health Sciences Centre (SH) and Princess

Margaret Hospital (PMH) to approach and conduct research among ambulatory oncology nurses.

2.1.2! Hospital Staff

A literature search was conducted to verify a gap in the research on the influence of chronotype

and sleep quality on CS and CF among oncology hospital staff providing cancer patient care. A

set of surveys was compiled to assess demographic information, professional quality of life,

chronotype, sleep quality, personality and job satisfaction. The professional quality of life

measure assesses CS/CF. Ethics approval was received from Sunnybrook Health Sciences Centre

41

(SH) to request voluntary participation from oncologists, oncology nurses, radiation therapists,

pharmacists, and pharmacy technicians working at the Sunnybrook Odette Cancer Centre. Upon

receiving ethics approval and prior to reaching out to participants, the questionnaire package was

created in an online survey format using Survey Monkey.

2.1.3! Patients and Caregivers

An extensive literature search was conducted to assess chronotype, sleep quality, and coping

among cancer patients and their familial caregivers. A gap in the literature was found on the

influence of chronotype and sleep quality as factors influencing coping, with personality as a

potentially mediating factor, among breast cancer patients undergoing chemotherapy and their

spousal caregivers. Questionnaire packages were assembled to assess demographic information,

chronotype, sleep quality, recalled coping across the day on various domains, and use of coping

behaviours. These were completed at baseline, midpoint and endpoint. A daily log was created

so participants could track their perceived changes in coping across the day on a Likert scale of 1

(poor) to 5 (excellent). Ethics approval was received from SH to approach early stage (I, II, and

III) breast cancer patients, and their spousal caregivers for participation. While spousal

caregivers were only granted ethics approval to participate if doing so in conjunction with their

partner, patients could participate with out without a spousal caregiver so long as they were

going to receive chemotherapy and were given a breast cancer diagnosis between stages I to III.

2.2! Instruments

2.2.1! General Questionnaires

Horne-Östberg Morningness Eveningness Questionnaire (MEQ): a 19-item assessment tool that

measures M or E chronotype. Each question is scored in the range of 0-6 points, depending on

question. Individual items are tallied to produce a single total score in the range of 16 and 86.

Scores in the range of 16 to 41 indicate E-tendency; 42-58 indicates a chronotype of neither (N)

morning nor evening, but likely more a tendency towards a preference for midday; 59-86

indicates a preference for M.

Pittsburgh Sleep Quality Index (PSQI): a 19-item assessment tool that provides a global sleep

quality index. Each question is scored on a scale of 0-3 (0 indicates no particular sleep

42

disturbance based on that dimension). Following data transformation, a global sleep quality

rating is provided between 0 and 21. In the general population, a global score of ≤ 5 indicates

good sleep quality, while scores > 5 indicate poor sleep quality. Further studies have suggested

that among cancer patients, a global score ≤ 8 indicates good quality sleep, while > 8 indicates

poor sleep quality (Carpenter & Andrykowski, 1998; Vargas, Wohlgemuth, Antoni, Lechner,

Holley & Carver, 2010).

2.2.2! Group Specific Questionnaires

2.2.2.1! Nursing Staff

Copenhagen Burnout Inventory (CBI): a 3-scale, 19-item measure that assesses personal, work-

related, and client-related burnout. A corresponding burnout score is provided for each category.

Each component variable is rated with a score of 0, 25, 50, 75, or 100. The total score for each

burnout category is based on the mean value of the component questions in that section. On each

scale, a higher score is indicative of a greater degree of burnout for that specific dimension

(Kristensen, Borritz, Villadsen & Christensen, 2005).

2.2.2.2! Oncology Staff

Job Satisfaction Survey (JSS): a 9-subscale, 36-item measure of satisfaction with a range of

conceptually distinct topics relating to one’s job. Developed specifically for use with human

service, public, and nonprofit sector organizations, the scale provides 9-component scores, and

one total composite score. Individual response choices are scored on a scale of 1 (disagree very

much) to 6 (agree very much); negatively worded items require reverse scoring. Individual

component scores are summed to produce the value for each subscale between 4 (dissatisfied)

and 24 (satisfied), with those between 12-16 indicating ambivalence. All subscale values are

totaled to provide a total composite score between 36 (dissatisfied) and 216 (satisfied), with

those between 108-144 indicating overall ambivalence (Spector, 1985).

Ten Item Personality Inventory (TIPI): a 10-item scale that produces five outcome scores:

Extraversion, Agreeableness, Conscientiousness, Emotional stability, and Openness to

experience. Each outcome score is based on the mean score of two individual questions

describing the particular trait, scored on a scale of 1 (disagree strongly) to 7 (agree strongly);

reverse scoring is used where indicated by questionnaire scoring rules. The TIPI is used as a

43

short version measure to assess the Big Five Personality Traits. The five subscales range as

follows:

•! Extraversion: reserved, quiet to extraverted, enthusiastic

•! Agreeableness: critical, quarrelsome to sympathetic, warm

•! Conscientiousness: dependable, self-disciplined to disorganized, careless

•! Emotional Stability: anxious, easily upset to calm, emotionally stable

•! Openness to Experience: conventional, uncreative to open to new experiences, complex

(Gosling, Rentfrow, & Swann Jr., 2003).

Professional Quality of Life (ProQoL), Version 5: a 3-subscale, 30-item measure of CS, burnout,

and STS. The 3 scales are distinct and cannot be combined to produce a total score. Burnout and

STS are the 2 subcomponents of CF, the counterpart to CS. Each question is scored on a scale of

1 (never) to 5 (very often). Reverse scoring is applied to applicable questions as indicated by the

scoring guide. Total scores for each subscale are achieved by summing the individual questions,

to achieve raw scores in the range of 10 to 50. Each subscale follows the same raw cutoff score

scale: 22 or less indicates low CS, burnout, or STS; between 23 and 41 indicates average CS,

burnout, or STS; 42 or more indicates high CS, burnout, or STS (Stamm, 2010).

2.2.2.3! Patients and Caregivers

Daily coping logs: these were devised for the purposes of this research for participants to keep

track of their coping in the moment across the day. Coping logs were intended for the use by

patients undergoing chemotherapy and their caregivers. Coping is assessed on a scale of 1 (poor)

to 5 (excellent). The three time slots for coping assessment were: 7:00 am to 11:00 am, 1:00 pm

to 5:00 pm, and 7:00 pm to 11:00 pm. The timing slots were assigned as such to reflect a

morning, afternoon, and evening coping time, and to allow for a reflection of coping across the

day. Participants were asked to give a rating of their coping “in the moment” at any point in the

time slot, but to skip/leave blank any coping times that had been missed, even if just by a few

minutes. Participants were told not to reflect back on their coping, but rather give a rating of how

he or she felt in the moment. A specific definition of coping was not provided, instead

participants were asked to give a coping rating based on their own personal understanding of

coping and to provide a rating in terms of everything being faced and in light either their own

44

diagnosis or that of their partner. No coping definition was given in order to remove any bias

towards searching for positive or negative coping traits.

2.2.2.3.1!Questionnaire Package

University of Toronto Inventory of Morningness and Eveningness (UTIME) Questionnaire: asks

respondents to report their recalled best, average, and worst performance times for various

activities they could be asked to perform across the day. UTIME data can be used comparatively

against scores of other questionnaires. Against MEQ scores, the UTIME can be used to

understand how self-reported preference for M versus E (based on MEQ score) is associated with

one’s own recalled performance across the day on various tasks (UTIME score). Against daily

coping log data, the UTIME can be used to assess how one’s retrospective memory of emotional

stability on various coping-based tasks compares to in the moment ratings.

Big Five Aspects Scales (BFAS): a 100-item measure that produces 10-outcome scores that

represent the 2 distinct but related aspects within each of the Big Five domains of personality.

Each question is scored on a scale of 1 (strongly disagree) to 5 (strongly agree). Reverse scoring

is applied where indicating by the scoring guide. Completed questions within each of the 10

scales are summed and averaged to produce a score. The Big Five scores are computed by

averaging a score for the two aspects within each domain. Higher scores (both the distinct pairs,

and the overall domain) indicate a greater propensity for displaying that personality trait

(DeYoung, Quilty, & Peterson, 2007).

Brief-COPE: a modified version of the original 14-subscale, 60-item COPE scale. The Brief

COPE is a 14-subscale, 28-item measure of coping that assesses multiple responses known to be

relevant to coping. The scale does not provide a breakdown of whether the coping responses fall

into categories such as problem-focused or emotion-focused, but rather presents stressor

response-types that may be relevant to either effective or ineffective coping. Each question is

scored on a scale of 1 (I haven’t been doing this at all) to 4 (I’ve been doing this a lot). To

achieve a score for each subscale, take the mean of the two corresponding questions for that

particular scale. The score for each subscale can range between 1 and 4, with higher scores

indicating a greater tendency to engage in the particular coping technique described by the

particular subscale. This scale does not provide a total or overall score that represents a dominant

coping style (Carver, 1997).

45

2.3! Statistics

SPSS (Statistical Package for the Social Sciences) version 20.0, 23.0, and 24.0 (Armonk, NY)

for Mac were used to perform various descriptive and analytical tests on the data.

46

Chapter 3!

Bellicoso, D., Ralph, M. R., & Trudeau, M. E. (2014). Burnout among oncology nurses: Influence of chronotype and sleep quality. Journal of Nursing Education and Practice, 4, 80-89. doi: https://doi.org/10.5430/jnep.v4n8p80

! Burnout Among Oncology Nurses: Influence of Chronotype and Sleep Quality

3.1! Abstract

The study sought to clarify the impact of chronotype and sleep quality on feelings of personal,

work-related, and client-related burnout among ambulatory care oncology nurses following

regular dayshift work schedules. Ninty-four participants from two Toronto, ON hospitals took

part. The Horne-Östberg Morningness Eveningness Questionnaire, Pittsburgh Sleep Quality

Index, and the Copenhagen Burnout Inventory were used to assess the impact of chronotype and

sleep quality, together with subjective measures of job and place of employment satisfaction and

work stressfulness on burnout. Findings showed that participants reporting greater tendency for

evening-type or neither-type chronotype, and/or poor sleep quality had significantly higher levels

of personal, work-related, and client-related burnout than individuals with either a morning

tendency and/or good sleep quality. Work stressfulness also contributed to elevated burnout.

Working at one’s optimal time and obtaining good quality sleep contributes to decreased

burnout. Future studies should consider the effect of chronotype and sleep quality on mediating

burnout among shift work oncology nurses. When creating nursing work schedules, employees’

chronotype and associated sleep quality should be considered to achieve decreased burnout,

optimal performance, and potentially increased employee retention and patient care quality and

satisfaction.

Keywords: Burnout, Chronotype, Circadian Rhythms, Nursing, Oncology, Sleep Quality

47

3.2! Introduction

The impacts of chronotype and sleep quality on burnout have yet to be studied and require

greater examination. Chronotype reflects one’s performance capability as it changes throughout

the day and the associated preference for morning (M) vs. evening (E) activities (Horne &

Östberg, 1976; Nielsen, 2010). Sleep quality refers to both quantitative (e.g., sleep duration and

latency, and number of arousals) and subjective (e.g., degree of restfulness) aspects of sleep –

however, inter-individual differences exist in the elements of sleep composition and their relative

importance (Buysse, Reynolds III, Monk, Berman, Kupfer, 1988). Healthcare workers are

particularly prone to burnout, and oncology nurses in general are at great risk, as the uncertainty

associated with patient outcome will elevate stress contributing to poor quality sleep (Chen &

McMurray, 2001; Cubrilo-Turek, Urek & Turek, 2006; Potter et al., 2010). Given the negative

feelings associated with burnout, it is important to understand how chronotype and sleep quality

both contribute to, and can be used to alleviate burnout.

Chronotype is associated with a range of outcomes. For example, research has shown systematic

differences in blood oxygenation levels in the brain across the day that relate with one’s

chronotype (Peres et al., 2011). In terms of measurable behaviour, one’s executive functioning

and alerting reactions have also shown daily fluctuations that correspond to chronotype

(Matchock & Mordkoff, 2009). Chronotype is linked with the time of sleep onset (Horne &

Östberg, 1976; Roenneberg, Wirz-Justice & Merrow, 2003). Significant differences in wake and

sleep onset times exist between M and E type individuals, with E types going to bed

approximately 99 minutes after M types and awaking approximately 114 minutes after them,

without significant differences in each group’s sleep duration (Horne & Östberg, 1976).

Adequate and good quality sleep is a basic human need that is restorative both physically and

cognitively (Karagozoglu & Bingöl, 2008; Stepanski, 2002). Like chronotype, sleep quality is

also known to affect a range of measurable outcomes. In relation to health, poor sleep quality has

been shown to increase many negative physical and mental health problems (Araghi et al., 2013;

Fujiwara, 2013). In terms of executive function, good sleep quality has been shown to increase

academic performance and executive functioning, while poor sleep quality, particularly among

elderly individuals, is associated with reduced cognitive performance (Lemma, Berhane, Worku,

Gelaye & Williams, 2014; Miyata et al., 2013). While proper and adequate sleep is important for

48

everyone, many people do not obtain the necessary amount of slow-wave and REM sleep each

night, with approximately 30% of the general population reporting sleep problems (van

Litsenburg et al., 2011). While the concept of sleep quality is difficult to define objectively, it

does involve certain key quantitative and subjective constructs such as sleep duration and

restfulness, respectfully (Buysse, Reynolds III, Monk, Berman, Kupfer, 1988). Shortened sleep

duration impedes physical and cognitive restoration, and decreases the following day’s

wakefulness, while regularly oversleeping can increase tiredness due to the strenuous and energy

consuming nature of REM sleep (Bonnet & Arand, 1995). Sleep restfulness relies both on the

quality of one’s previous waking period and its ability to generate proper homeostatic sleep

drive, and how one perceived their personal level of tiredness following a sleep period

(Bersagliere et al., 2012). Thus, sleep quality – based both on its quantitative and subjective

aspects, is important in allowing a person to go through the sleep stages necessary for their body

to be properly restored, well rested, and ready for the next day. However, while the importance

of adequate and good quality sleep has been studied abundantly, many conventional societal

practices interfere with achieving proper sleep quality when they conflict with the hours

associated with innate chronotype.

Work schedules in continuous disharmony with one’s circadian rhythm induce stress, which

negatively affects sleep quality (Karagozoglu & Bingöl, 2008). Disturbed sleep and its associated

fatigue make it difficult to carry out daily activities and work tasks. Among nurses, sleep debt

induces stress independently of the high stress work situations they face on a daily basis,

oftentimes leading to feelings of extreme emotional and physical fatigue, decreased cognitive

function, weariness, and exhaustion, which together are known as burnout (Brand et al., 2010;

Vela-Bueno et al., 2008).

Burnout is determined from feelings of emotional and physical fatigue or exhaustion, cognitive

weariness, and chronic energy resource depletion due to continued exposure to stress. However,

there is no concrete definition of the term, nor is there a standardized general procedure from

which to obtain a burnout diagnosis (Brand et al., 2010; Korczak, Huber & Kister, 2010). A

common hypothesis is that stress-induced sleep disturbances over time compound to produce

mental and physical exhaustion, which are generally agreed to be key contributors to burnout

(Söderström, Ekstedt, Åkerstedt, Nilsson & Axelsson, 2004). In line with this hypothesis, the

Copenhagen Burnout Inventory (CBI) used in the present study suggests that at the core of

49

burnout are fatigue and exhaustion (Kristensen, Borritz, Villadsen & Christensen, 2005). The

CBI subscribes to the definition of burnout that suggests it is a state of emotional, physical, and

mental exhaustion resulting from prolonged involvement in emotionally demanding work

situations (Schaufeli & Greenglass, 2001). In healthcare organizations, burnout is a significant

phenomenon due to its demonstrated negative effects on patient satisfaction and safety, and on

nurse retention and turnover (Potter et al., 2010). Among clinical care oncology nurses, higher

burnout rates have been found compared to hospice care oncology nurses, suggesting that

working in a hospital setting with the goal to heal sick patients rather than care for those that are

dying, may increase stress levels, and cause subsequent sleep disturbances which may contribute

to the gap in burnout reports (Ostacoli et al., 2010).

While many studies have examined the triggers and protective factors for burnout, the influence

of chronotype and sleep quality on this phenomenon have not been studied. We sought to

demonstrate the potential for chronotype and sleep quality to mediate feelings of burnout among

ambulatory care oncology nurses working in a hospital setting on a fixed Monday to Friday

daytime work schedule by using a set of standardized questionnaires to assess the influence of

chronotype and sleep quality on burnout. Our objectives were to understand the separate links

between good versus bad sleep and one’s ratings of the various types of burnout, and the link

between chronotype and burnout. It was also important to understand the differential

contribution of various predictors (chronotype, sleep quality, job satisfaction, overall place of

employment satisfaction, and work stressfulness) on the various types of burnout. It was

predicted that M types and respondents with better sleep would exhibit less overall burnout

compared to E types and people reporting poor sleep quality. It was also predicted that in

addition to one’s chronotype and sleep quality, stressfulness and satisfaction ratings relating to

one’s job would differentially influence each type of burnout.

3.3! Materials and Methods

3.3.1! Participants

Registered ambulatory care oncology nurses working at one of two primary care hospitals in

Toronto, Ontario, Canada – University Health Network Princess Margaret Cancer Centre (PMH),

and Sunnybrook Health Sciences Centre Odette Cancer Centre (SB) – were recruited to

participate. A total of 64 nurses completed questionnaires with useable results (PMH, n = 42, all

50

female, age 27 – 63, mean age 45.29 ± 10.18; SH, n = 22, 21 females, 1 male, age 28 – 65, mean

age 50.54 ± 11.70). The mean ages for these groups are not significantly different (see Table

3.1). Inclusion criteria were as follows: participants must be ambulatory care oncology nurse

responsible for the care and support of oncology patients, be currently employed in one of the

hospitals, follow a regular dayshift work schedule Monday to Friday, and not be connected with

the present study. The study was granted approval by the ethics committees of both hospitals.

Informed consent was obtained from all participants prior to giving them the questionnaire

package. Study participation was voluntary.

3.3.2! Procedure

A preliminary meeting was held to present the aims of the study, and an anonymous survey

package was distributed to nurses who agreed to partake in the present study. Time was given in

the meeting to complete the survey package. Nurses provided demographic information and job

descriptions (age, gender, job satisfaction, place of employment satisfaction, and stressfulness of

current job) in a background information survey.

3.3.3! Measures

The Horne-Östberg Morningness Eveningness Questionnaire (MEQ) is a 19-item measure that

was used to assess tendency for M or E chronotype. Each component question is scored on a

scale in the range of 0-6 depending on the question, based on guidelines provided by the

questionnaire’s authors, to obtain a global score between 16 and 86. Based on one’s global score,

a rating of definitely E- (16 to 30), moderately E- (31 to 41), neither- (42 to 58), or moderately

M- (59 to 69) or definitely M-type (70-86) is given. The Pittsburgh Sleep Quality Index

(PSQI) is a 19-item measure that yields a global sleep quality index. Each question is scored on

a frequency scale of 0 to 3 (0 indicates no occurrence of a particular sleep disturbance). A global

sleep quality rating is provided on a scale of 0 to 21 (following data transformations), where a

global score of ≤ 5 indicates good sleep quality, while > 5 indicates poor sleep quality. The

primary outcome variables – personal, work-related, and client-related burnout – were measured

using the Copenhagen Burnout Inventory (CBI) which is a 3-scale, 19-item measure used to

assess these 3 types of burnout that provides a corresponding burnout score for each category.

Each component question is rated with a score of 0, 25, 50, 75 or 100. Total scores for each

burnout category are based on the mean of the component questions pertaining to that section.

51

The higher a scale’s associated score, the greater the degree of burnout that is associated with

that specific dimension.

3.3.4! Statistical Analysis

A priori tests indicated that to detect a moderate effect size, with 80% power and α = .05, sample

size calculations for a one-tailed test required 27 subjects. Statistical analyses were performed

using SPSS (Statistical Package for the Social Sciences) version 20.0 for Mac. Analyses were

run as follows: demographic and test variables of the two hospital groups of nurses were

compared using unpaired t -tests. Each group of nurses was analyzed by their MEQ

classification, and again by the PSQI good or poor quality sleep rating using t -tests to look for

burnout differences. Correlations were run between the test variables to help determine where

significant associations lay between different items. Lastly, multivariate regression analyses were

run to clarify the differential contribution of the predictors of interest to the outcome measures. A

p-value < .05 was accepted as statistically significant.

3.4! Results

Of the 94 nurses who volunteered to participate, 64 (68.09%) provided useable data, resulting in

a strong effect size with 98.96% power (α = .05, one-tailed test). At PMH, 42/60 (70.00%)

returned useable data, while 22/34 (64.71%) with useable data were returned from SH. None of

the demographic characteristics and questionnaire response ratings between the two groups are

statistically different for these categories, indicating that the two groups are well matched (see

Table 3.1).

52

Table 3.1 Demographic characteristics and questionnaire response ratings of participants Nurse group PMH (n = 42) SB (n = 22) Frequency Frequency Gender (M:F) 0:42 1:21 Mean (SD) Mean (SD) t-test* Age in years 45.29 (10.18) 50.54 (11.70) NS Oncology nurse experience (yrs) 15.10 (11.64) 14.19 (10.25) NS Job satisfaction 3.46 (1.09) 3.93 (0.76) NS Overall place of employment satisfaction

3.70 (1.01) 4.05 (0.95) NS

Work stressfulness 3.82 (1.03) 3.68 (1.09) NS MEQ 57.02 (8.47) 59.82 (9.52) NS PSQI 6.60 (3.59) 5.50 (2.20) NS CBI

Personal 50.20 (19.63) 44.32 (16.94) NS Work 51.70 (22.16) 44.48 (16.52) NS

Client Related 21.22 (19.13) 21.02 (14.52) NS Note. NS = not significant. *p < .05, 2-tailed.

53

The three separate CBI categories each conceptualize burnout as a continuous variable on a scale

of 0 to 100. Table 3.2 shows burnout ratings among respondents with good (≤ 5) or poor (> 5)

sleep quality as rated by the PSQI, and burnout ratings among MEQ types. Given the low

number of moderately E types (n = 2) and definitely M types (n = 4), these two categories have

been added to the neither type and moderately M type groups, respectively. This break point also

allows for a nearly equal frequency distribution of participants with 48.44% of respondents

qualifying as moderately E type or neither type, and 51.56% qualifying as M types. It is

important to note that while both of these conceptualizations of burnout in Table 3.2 are broken

down into groups of n = 31 and n = 33, the individuals making up these divisions are not

necessarily the same across both sets of data.

54

Table 3.2 Comparison of burnout ratings between respondents with good and bad sleep quality and between MEQ types

Sleep Quality (PSQI) MEQ Good (≤ 5)

Bad (> 5) t-test Moderately

& Definitely M type

N & Moderately E type

t-test

CBI Personal 39.65 (18.81) 56.19 (15.13) -3.88*** 42.17 (21.16) 54.57 (13.58) 2.78**

Work 42.51 (21.77) 55.52 (17.41) -2.65** 42.86 (22.23) 55.99 (16.38) 2.68** Client Related 15.59 (15.51) 26.52 (17.94) -2.60* 15.15 (16.23) 27.69 (16.82) 3.03**

*p < .05, 2-tailed. **p < .01, 2-tailed. ***p < .001, 2-tailed.

55

As shown in Table 3.3, MEQ and PSQI scores are not significantly correlated. Significant

correlations exist between all CBI scores and the associated MEQ and PSQI scores. Stressfulness

associated with place of employment is significantly correlated to all burnout subscales, while

job satisfaction is significantly correlated only with work related burnout. Overall place of

employment satisfaction was not correlated with chronotype, sleep quality, or any of the burnout

ratings.

56

Table 3.3 Analysis of bivariate correlations for participants’ questionnaire response ratings CBI MEQ PSQI Personal Work Client MEQ 1.00 -.19 -.33** -.26* -.36** PSQI -.19 1.00 .52** .43** .39** Job Satisfaction .05 -.12 -.21 -.38** -.04 Overall place of employment satisfaction

-.12 .11 -.07 -.23 .08

Work Stressfulness .03 .26* .62** .66** .32** *p < .05, 2-tailed. **p < .01, 2-tailed.

57

To clarify the differential contribution of the predictors of interest on the various categories of

burnout, multivariate regression analyses were run. Table 3.4a shows the relationship between

personal burnout and each predictor. Statistical model #2 (Step 2) yields a satisfactory proportion

of variances explained, as indicated by the associated R2 value (.61). Table 3.4b shows the

relationship between work related burnout and each predictor. For this relationship, statistical

model #3 (Step 2) yields a satisfactory proportion of variances explained as indicated by the

associated R2 value (0.62). Table 3.4c shows the relationship between client related burnout and

each predictor. For this relationship, statistical model #2 (Step 2) yields a satisfactory proportion

of variances explained as indicated by the associated R2 value (0.31). Note that the R2 value

explaining the proportion of variance for client related burnout due to these predictors is much

lower than that seen for personal and work related burnout. This suggests that other factors not

being considered in this study may influence client related burnout.

58

Table 3.4a Summary of hierarchical multivariate regression analysis for variables predicting personal burnout among oncology nurses (N = 64) Variables b SE b ßa Step 1

Constant 60.91 14.56 MEQ -0.52 0.23 -.24* PSQI 2.77 0.63 .47***

Step 2 Constant 34.52 11.93

MEQ -0.61 0.18 .29*** PSQI 1.90 0.50 .32***

Work Stressfulness 9.88 1.51 .55*** Note. R2 = .33 for step 1; ΔR2 = .28 for step 2. a standardized coefficient *p < .05, 2-tailed. ***p < .001, 2-tailed.

59

Table 3.4b Summary of hierarchical multivariate regression analysis for variables predicting work related burnout among oncology nurses (N = 64) Variables b SE b ßa Step 1

Constant 58.18 17.39 MEQ -0.43 0.27 -.18 PSQI 2.52 0.74 .39***

Step 2 Constant 26.07 13.70

MEQ -0.54 0.20 .23** PSQI 1.46 0.58 .23*

Work Stressfulness 12.04 1.74 .61*** Step 3

Constant 45.41 14.72 MEQ -0.52 0.19 -.22** PSQI 1.37 0.55 .21*

Work Stressfulness 11.20 1.68 .57*** Job Satisfaction -4.69 1.68 -.23**

Note. R2 = .22 for step 1; ΔR2 = .35 for step 2; ΔR2 = .05 for step 3. a standardized coefficient *p < .05, 2-tailed. **p < .01, 2-tailed. ***p < .001; 2-tailed.

60

Table 3.4c Summary of hierarchical multivariate regression analysis for variables predicting client related burnout among oncology nurses (N = 64) Variables b SE b ßa Step 1

Constant 44.00 14.23 MEQ -0.59 0.23 -.30* PSQI 1.84 0.63 .33**

Step 2 Constant 32.03 14.77

MEQ -0.63 0.22 .32** PSQI 1.44 0.62 .26*

Work Stressfulness 4.49 1.88 .27* Note: R2 = .24 for step 1; ΔR2 = .07 for step 2. a standardized coefficient *p < .05, 2-tailed. **p < .01, 2-tailed.

61

According to the experimental hypothesis, chronotype and sleep quality significantly influenced

all the CBI subscale scores. In addition, each respondent’s perception of work stressfulness

significantly influenced each CBI subscale score. Job satisfaction was a significant factor

influencing the work-related CBI subscale. These findings correspond with the analysis of

bivariate correlations indicated in Table 3.3.

3.5! Discussion

Chronotype and sleep quality are known to influence a wide range of important aspects of daily

life such as physical and mental health, blood oxygenation, and cognitive functioning to name a

few (Peres et al., 2011; Fujiwara et al., 2013; Miyata et al., 2013). However, the influence of

one’s chronotype and sleep quality on the degree of burnout has not been greatly explored. Given

the high degree of burnout associated with several careers, and particularly with oncology

nursing, it was important to consider the impacts of chronotype and sleep quality on burnout.

This information will be informative to the design of schedules and approaches that reduce

burnout among members of this group. The present study investigated the contribution of

chronotype and sleep quality, together with situational factors, on symptoms of burnout among

ambulatory care oncology nurses working in outpatient clinics in one of two hospitals in

Toronto, Ontario, Canada. It was found that innate chronotype and sleep quality strongly

predicted burnout, such that individuals with an M type tendency, and respondents with better

sleep quality, experienced less burnout when following a set dayshift work schedule.

The two groups of nurses from each hospital were well matched for demographic characteristics.

A comparison of the two nurse groups found the absence of a significant difference in terms of

job satisfaction, overall place of employment satisfaction, place of employment stressfulness,

age, and years as a nurse, thus ruling out the possibility that burnout was triggered by place of

employment satisfaction differences. The two groups were also well matched for their MEQ and

PSQI scores.

No significant difference was found between the two nurse groups, regarding the presence of

feelings of personal, work-related, and client-related burnout as rated by the CBI. However,

significant differences in tendency for feelings of burnout were found between individuals

62

reporting good versus poor sleep quality based on PSQI cutoff ratings. Respondents with good

sleep quality reported significantly lower levels of burnout across all three subscales, compared

to participants who described themselves as having poor quality sleep. Regarding chronotype,

individuals with a greater propensity for morningness reported lower levels of burnout across all

three subscales compared to participants with a tendency for eveningness or neither M nor E. In

both cases of burnout analysis – by sleep quality and by chronotype – personal and work related

burnout were reported to be in the midrange level of severity, while client-related burnout

appeared to be much lower. It is possible that client related burnout may have been reported as

being much lower than personal and work related burnout scores if nurses are inclined to express

compassion towards patients as opposed to being frustrated with them. However, no data was

collected on this topic therefore no formal conclusions can be drawn. In the future, it will be

important to look at both compassion fatigue and compassion satisfaction to examine changes in

nurses’ feelings on these topics as they relate to chronotype and sleep quality.

In agreement with the idea by Söderström et al. (2004) that stress and sleep quality are

associated, correlational analysis demonstrated that individuals expressing greater work

stressfulness tend to have poorer sleep quality. Interestingly however, while the data were not

shown, chronotype and sleep quality were not associated in this sample. This may be due in part

to the sample size. Had more extreme chronotypes been obtained, this may have led to an

association. Furthermore, this result may change if sleep quality and chronotype were to be

examined in shiftworkers. It is more likely that in such a group, people would be working at

hours that may significantly conflict with their chronotype, or if they are extreme M or E types,

may even compliment them. Furthermore, shiftworkers likely have fragmented sleep which

would affect their sleep quality ratings, and this together with chronotype, might yield different

results on the relationship between chronotype and sleep quality.

To further investigate the evidence supporting the significant contribution made by chronotype

tendency, sleep quality, job satisfaction, place of employment satisfaction, and overall place of

employment stressfulness, in determining feelings of burnout, multivariate regression analyses

were applied. Chronotype tendency and sleep quality were significant factors influencing all of

the CBI subscale scores, however, here again the effect of these two factors on client related

burnout was much less than on personal and work-related burnout. As such, it appears that both

propensity for morningness and good sleep quality lower the risk of burnout among oncology

63

nurses working regular day shifts Monday to Friday. Overall workplace stressfulness was a

contributor to each burnout subscale score, indicating that reduced workplace stressfulness

protects against each type of burnout examined by this study. Job satisfaction also contributes

significantly to workplace related burnout, suggesting that higher job satisfaction leads to

reduced burnout in this domain. It did not however, influence personal or client related burnout.

Overall place of employment satisfaction was not a significant contributing factor, suggesting

that one’s burnout is more so influenced by the specific role or duties they must carry out, rather

than by the institution as a whole. The lack of correlation between overall place of employment

satisfaction and any of the burnout ratings supports the finding that overall place of employment

satisfaction was not a differential predictor for any of the burnout ratings examined.

According to the experimental hypothesis, this study’s findings support the view that chronotype

and sleep quality are key predictors of personal, work-related, and client-related burnout

symptoms, indicating that oncology nurses following a permanent dayshift schedule with a

propensity for morningness present with relatively low levels of burnout, compared to their

colleagues working at the same time but who have a tendency for eveningness or neither type of

chronotype. Furthermore, ambulatory care oncology nurses obtaining better sleep quality also

experience significantly decreased feelings of burnout across all three domains as compared to

their colleagues who report poor sleep quality. Many people – either through necessity or

voluntarily – work at times that conflict with their chronotype, oftentimes foregoing adequate

sleep in order to do so. This results in poorer sleep quality for a number of reasons including but

not limited to shortened sleep, going to bed at times not reflective of one’s chronotypic needs,

and even sleeping at times of day when one should actually be awake and functioning, to name a

few. While other personal and environmental factors likely also underlie these significant

burnout differences, allowing nurses to begin work at slightly modified times that better

accommodate their chronotypic needs and may help modify or enhance their sleep situation and

quality, might significantly help to reduce levels of personal, work related, and client related

burnout. However, further studies need to be completed in order to understand which of these

burnout subscale categories would be more or less affected by such changes.

This study’s results have important implications in the healthcare profession, not only regarding

nurse related burnout, but also relating to patient care. If burnout brings on feelings of cognitive

weariness, one’s decision-making skills may be affected such that their ability to make sound

64

judgments may be hindered by increased feelings of burnout. Physical and emotional fatigue

may also be heightened, potentially reducing one’s ability to react quickly in dire situations to

help a patient in distress. Emotional control or regulation may also be compromised, which could

elevate a patient’s level of distress should they be forced to deal with a nurse who is emotionally

burnt out. Burnout is known to increase the risk of absence from work (Westermann Kozak,

Harling & Nienhaus, 2014). An increasing number of people absent from work due to burnout

will further compound the workload and related stress felt by many nurses working in the same

department who will need to take over the responsibilities of their coworkers. Based on these

negative effects of burnout, it is in a medical institution’s best interest to consider how key

contributors such as chronotype and sleep quality can be better managed and used to alleviate

symptoms of burnout. Given the large number of nurses following shift work schedules both in

and out of the field of oncology, future studies should address burnout among oncology nurses

engaging in shift work in order to understand how chronotype and sleep quality could be used to

structure schedules that allow for decreased reports of burnout.

3.5.1! Limitations

One limitation of this study is related to sample size. Future studies using larger numbers may

likely be able to achieve a greater and more equal distribution of E and M type participants. This

might allow for reanalysis of the current statistics in this study that suggest there is no correlation

between chronotype and sleep quality among oncology nurses, as well as a greater understanding

of how extreme the symptoms of burnout are among E type participants working earlier in the

day. Secondly, in the future it would be beneficial to study client-related burnout, by including

other objective measures to rate one’s degree of compassion, in order to understand if reduced

ratings in this domain are related to feelings of compassion for patients, or whether some other

factor was involved in causing nurses to report such markedly lower burnout level in this

domain. The role of other factors such as extracurricular activities, medical conditions, and

medications were not investigated, as many nurses did not report this information in the personal

survey, and it would be important to understand the role of these variables in future.

65

3.6! Conclusions

Individual chronotype and sleep quality are important determinants of alertness and cognitive

ability variance throughout the day. The findings of this study help to clarify the contributions of

chronotype and sleep quality to burnout, together with workplace stressfulness and job

satisfaction, and as such fully support initiatives for programs that would modify work schedules

to accommodate the needs of one’s chronotype and that would also make suggestions for

modifications that would ameliorate one’s sleep quality. Such initiatives would result in

decreased burnout among ambulatory oncology nurses, and may subsequently increase quality of

patient care and satisfaction. This research also presents further research opportunities for

understanding burnout among oncology nurses following shift work schedules, by considering

the disruption their work brings to their circadian rhythm and sleep quality.

66

Chapter 4!

Bellicoso, D., Trudeau, M., Fitch, M. I., & Ralph, M. R. (2017). Chronobiological factors for compassion satisfaction and fatigue among ambulatory oncology caregivers. Chronobiology International, Epub Ahead of Print. doi: 10.1080/07420528.2017.1314301

! Chronobiological Factors for Compassion Satisfaction and Fatigue Among Ambulatory Oncology Caregivers

4.1! Abstract

Primary caregivers for victims of chronic illness and or trauma experience both positive

and negative emotional consequences. These are broadly classified as compassion satisfaction

(CS) and compassion fatigue (CF). Because one of the components of CF, burnout, varies with

chronotype and sleep quality, we assessed the influence of chronobiological features on the

broader constructs of CS and CF. Responses from primary ambulatory care oncology staff

working dayshifts were assessed for potential relationships of chronotype and sleep quality with

CS and CF using the professional quality of life scale (ProQoL). These were analyzed further in a

multivariate model that included personality and job satisfaction as cofactors. We found that

sleep quality was a key contributor to CS development, and CF reduction. Morningness was

positively linked to CS, but the univariate association was masked in the multivariate model. Job

satisfaction (contingent rewards, nature of work, and operating procedures) heavily influenced CS

and CF development. Agreeableness and openness showed positive correlations with CS and

negative with burnout, while emotional stability was linked to reduced CF. While job satisfaction

and personality predictably played roles in the development of CS and CF, sleep quality and

chronotype contributed significantly to benefits and negative consequences of providing care.

67

4.2! Introduction

Compassion fatigue (CF) describes a set of negative consequences experienced by caregivers to

one or many people who are dealing with trauma. It includes feelings of exhaustion, frustration,

depression, and even fear driven by working with such individuals. Burnout has been recognized

as one of the major results that often develops among caregivers who experience chronic CF

from working with trauma victims and or individuals with long term illness (Stamm, 2010).

Burnout is preceded by a progressive increase in the sense of hopelessness and or difficulty in

dealing with work (Brand et al., 2010). It may be expressed as feelings that one’s efforts are

useless, or that workloads are unmanageable. Common responses are anger, frustration, sadness,

cognitive weariness, and emotional and physical fatigue (Bellicoso et al., 2014; Stamm, 2010).

Along with burnout, CF can also produce secondary traumatic stress (STS) resulting from

exposure to individuals who have experienced recently, or are currently experiencing, traumatic

events. STS may be expressed as a lingering combination of fear, sleep disturbance, and or

intrusive imagery.

In contrast, compassion satisfaction (CS) is a positive outcome of providing care. CS refers to the

pleasure or fulfillment a caregiver feels from doing a job well and from helping others,

particularly those faced with traumas or chronic illness (Stamm, 2002). Therefore, taken together,

CF and CS reflect a worker’s total professional quality of life. These positive and negative

outcomes of caregiving are known to exist among those working in various positions in the

healthcare field, (e.g., nurses, physicians, therapists, and even audiologists) (Drury et al., 2014;

El-Bar et al., 2013; Severn et al., 2012; Sodeke-Gregson et al., 2013). A quantification of the two

components is provided by the Professional Quality of Life (ProQoL) Scale.

Given the prevalence and impact of CS and CF among healthcare workers, it is important to

understand factors that contribute to their development and management. In a previous study,

sleep quality and chronotype were shown to be significant factors in the propensity and

development of burnout among oncology nurses (Bellicoso et al., 2014). This raised the question

of what roles these two factors play in the broader constructs of CS and CF.

Chronotype refers to the relative preference shown by individuals for performing various

activities during the morning (M) hours (M types) versus evening (E) hours (E types) independent

of societally imposed time constraints (Horne & Östberg, 1976; Neilsen, 2010). Basic biological

68

processes such as cellular functioning, hormonal fluctuation, and fluctuations in body temperature

have been linked to chronotype and circadian rhythmicity (Brown et al., 2008; Horne & Östberg,

1976; Lack et al., 2009). Chronotype is also correlated with overt, measurable behaviours

including executive control, and physical performance (Brown et al., 2008; Lara, Madrid, &

Correa, 2014). On several tasks with a high cognitive demand, a synchrony effect has been

reported, showing that performance is often best at optimal (morning for M types, afternoon or

evening for E types) over other times of day (Hasher, Chung, May & Foong, 2002).

A key measurable expression of chronotype and rhythmicity is the sleep wake pattern (Horne &

Östberg, 1976; Roenneberg, Wirz-Justice, & Merrow, 2003). M and E types consistently show

no difference in sleep duration, but demonstrate significantly earlier sleep wake timing (Horne &

Östberg, 1976; Lara et al., 2014). While average sleep duration is consistent across chronotypes,

M types often report better sleep quality (Merikanto et al., 2012; Nielsen, 2010; Vitale et al.,

2015).

Importantly, chronotype and sleep quality have known associations with other factors that

ultimately affect CS and CF. Personality has been found to be associated with both chronotype

and compassion. Poorer sleep quality has been found in individuals reporting low

conscientiousness together with high neuroticism, while those reporting higher levels of

extraversion, agreeableness, and conscientiousness typically experience better sleep quality

(Duggan, Friedman, McDevitt, & Mednick, 2014; Hintsanen et al., 2014). Morningness has been

positively correlated with high conscientiousness and agreeableness; eveningness has been

correlated with greater neuroticism (e.g., Cavallera, Gatto & Boari, 2014; Duggan, Friedman,

McDevitt, & Mednick, 2014; Hsu, Gau, Shang, Chiu, & Lee, 2012). Studies that have examined

specifically the relationship between CS/CF and personality are sparse. Some research has

suggested a link with optimism, which is a broadly representative trait on the Big Five factors of

personality (Sharpe, Martin & Roth, 2011). In a group of genetic councilors, those reporting low

optimism had increased CF (moderate to high burnout), and low to moderate CS (Injeyan,

Shuman, Shugar, Chitayat, Atenafu, & Kaiser 2011). Among volunteer counselors caring for the

terminally ill, burnout is greater among those individuals reporting high neuroticism (Bakker,

Van der Zee, Lewig, & Dollard, 2006). Comparative roles for personality traits, chronotype, and

sleep quality on the development of CS and CF within the same subject pool have not been

assessed.

69

In addition, CS and CF are likely to be influenced by work conditions and job satisfaction. Low

job satisfaction is linked with greater stress and lower job ratings compared to workers reporting

high job satisfaction. Those reporting high job satisfaction tend to also report better health than

their colleagues (Rahman & Sen, 1987). Assessing the effect of job satisfaction when considering

CS and CF will provide a more complete picture of how specific personal and environmental

factors contribute to one’s overall professional quality of life.

All the factors that have been studied or theorized to contribute to CS and CF are likely to

interact. For example, chronic health issues have been linked to sleep disturbance, both as

outcomes as well as causative factors. Many societal conventions such as responsibilities in the

home, social protocols, and daily work schedules, interfere with achieving proper sleep quality

when they are desynchronized with the hours associated with one’s innate chronotype. Therefore,

to provide a more complete perspective on the factors contributing to CS and CF, in a defined

work environment, we have assessed the complex effects and interactions among sleep timing

and quality, chronotype, personality, job satisfaction, with self-ratings of CS and CF in oncology

staff working in a large hospital.

4.3! Materials and Methods

4.3.1! Participants and Procedures

Primary ambulatory care staff at the Odette Cancer Centre of the Sunnybrook Health Sciences

Centre (Toronto, Ontario, Canada) were sent a general email requesting their voluntary

participation in this study. The job categories included medical/radiation/surgical oncologists,

registered ambulatory care oncology nurses, radiation therapists, and pharmacists/pharmacy

technicians. The email explained the purpose of the study, and provided a link to participate. The

inclusion criteria indicated that the study was only intended for oncologists, oncology nurses,

radiation therapists or pharmacy team members working in the Odette Cancer Centre, providing

direct care to patients, and following a dayshift Monday to Friday work schedule. It was

explained that by proceeding to answer the survey questions, participants were granting their

consent to use the data provided. Participants were informed that their responses would not be an

indication of their work performance, nor would their individual scores be revealed to

70

supervisors or made public. While participation was voluntary, study subjects were given the

option of providing an email address to participate in a draw for one of three iPad minis, which

were awarded based on a random draw once the entire study had been completed.

In total, 140 staff members responded to the questionnaire. Twelve of the submissions were

discarded because the respondents either did not fit the job description criteria, worked an

irregular shift schedule, or the questionnaire was submitted incomplete, leaving a total of 128

completed useable questionnaires. The test group comprised 25 oncologists, 44 nurses, 17

pharmacy team members, and 42 radiation therapists.

4.3.2! Measures

The questionnaire package included six survey items: background information, Horne-Östberg

Morningness Eveningness Questionnaire (MEQ; Horne & Östberg, 1976), Pittsburg Sleep

Quality Index (PSQI; Buysse, Reynolds III, Monk, Berman, Kupfer, 1988), Job Satisfaction

Scale (JSS; Spector, 1985), Ten Item Personality Inventory (TIPI; Gosling, Rentfrow, & Swann,

2003), and the Professional Quality of Life Scale (ProQoL; Stamm, 2010). The short TIPI scale

was used instead of the full BFAS (DeYoung, Quilty, & Peterson, 2007) to assess personality to

reduce the number of questions for participants to complete, and the time commitment, increasing

the number of volunteers. The JSS was included because workers will have some degree of

opinion on tasks they perform, and the environment in which they work, which will affect

various domains of ProQoL and CF/CS beyond the direct impact of caregiving per se (see Renzi,

Tabolli, Ianni, Di Pietro, & Puddu, 2005).

4.3.3! Statistical Analysis

Data from oncologists, nurses, radiation therapists, pharmacists, and pharmacy technicians was

analyzed collectively, as a group. Descriptive data and demographic frequencies were used to

describe the sample. Questionnaires (MEQ, PSQI, ProQol, TIPI and JSS) were scored according to

their scoring manuals or published scoring guidelines in the original article. Descriptive analyses

were conducted on MEQ, PSQI, each of the nine components and the total JSS scores, the TIPI

scales, and the three components of ProQoL to assess whether data were normally distributed.

Natural log transformation was performed on burnout, STS and PSQI to produce normally

distributed variables later used in the regression analyses. Univariate linear regression was used

to assess which continuous and categorical variables were significantly associated with each

71

outcome variable (CS, burnout and STS). Factors significantly associated with study outcome

variables were included in a backwards multivariate linear regression analysis to build predictive

models of each ProQoL score. A p value < .05 was accepted as statistically significant. Statistical

analyses were performed using SPSS (Statistical Package for the Social Sciences) version 23.0

for Mac.

4.4! Results

4.4.1! Descriptive Group Statistics

Descriptive and frequency statistics for continuous and categorical demographic characteristics

and questionnaire data from the 128 participants are presented in Table 4.1.

72

Table 4.1 Means and frequencies of participant demographic and questionnaire data

Mean (SD) Frequencies (%) Basic Demographics Age 44.41 (11.83) yrs Gender

Males 27 (21.1%) Females 101 (78.9%)

No. of years in healthcare 20.48 (12.81) yrs No. of years in oncology 15.73 (11.27) yrs No. of Sites Worked

Single 25 (19.5%) Multiple 103 (80.5%)

Relationship Status Single 42 (32.8%)

Spouse 86 (67.2%) Current Position

Oncologist 25 (19.5%) Outpatient Nurse 44 (34.4%) Pharmacy Team 17 (13.3%)

Radiation Therapist 42 (32.8%) MEQ

Total Group of Scores 56.09 (9.907) Definitely Evening 1 (0.8%)

Moderately Evening 12 (9.4%) Neither 59 (46.1%)

Moderately Morning 47 (36.7%) Definitely Morning 9 (7.0%)

PSQI Total Group of Scores 5.30 (3.05)

Good Quality Sleep (≤ 5) 76 (59.4%) Poor Quality Sleep (> 5) 52 (40.6%)

Job Satisfaction Pay 14.13 (4.58)

Promotion 11.91 (4.37) Supervision 18.09 (4.76)

Benefits 14.00 (4.17) Contingent Rewards 14.70 (4.68)

Operating Procedures 12.68 (3.50) Coworkers 17.18 (3.72)

Nature of Work 20.04 (3.69) Communication 15.39 (3.96)

Total 138.13 (26.63) Personality

Extraversion 4.17 (1.45) Agreeableness 5.54 (1.09)

Conscientiousness 6.17 ( .93) Emotional Stability 5.46 (1.30)

Openness to Experience 5.39 (1.09) ProQoL

CS 40.16 (6.04) BO 22.40 (5.63)

STS 20.63 (5.80)

73

Respondents were predominantly female (78.8%), were married (67.2%), and or worked with

multiple cancer sites (80.5%). The mean MEQ score (56.06) indicates that on average,

participants were classified as neither M nor E type individuals; however, the average score is

only two points below the cutoff for being moderately M type. 43.7% scored as M type and 10.2%

were E type.

The mean sleep quality score (5.30) is slightly above the cutoff for good quality sleep. 40.6% of

respondents met the criteria for varying degrees of poor quality sleep, while 21.0% of the

respondents who were rated as having good quality sleep were right on the borderline of poor

quality sleep.

JSS scores were all comparable to those presented in the original article (Spector, 1985).

Cumulative scores ranged from 36 to 216 with an average of 138.13 (the highest possible JSS

score is 216). Therefore, respondents overall were above 50% for this index of job satisfaction.

Individual components of job satisfaction range from a score of 4 (not satisfied) to 24 (satisfied).

Average scores for individual components of job satisfaction averaged between 11.91 (promotion)

and 20.04 (nature of work).

Personality is scored on a scale of 1 (strongly disagree) to 7 (agree strongly). On average, the

strongest trait among respondents was conscientiousness (6.17), and the weakest was

extraversion (4.17), which was still above the halfway mark of the scale.

The mean score for CS (40.16) indicates that respondents fell in the high end of the range for

average satisfaction derived from their job (ProQoL manual, Stamm, 2010). The mean burnout

score fell just outside the cutoff to qualify as low burnout, suggesting participants were close to

the low end of an average degree of burnout. The mean STS score indicates that participants

typically feel a low level of STS related to their job.

4.4.2! Correlation Analysis

We first conducted a comprehensive univariate analysis using all component variables of the five

scales and demographic data. The correlations are presented in three supplementary tables

highlighting relationships with (a) demographic, chronotype, and sleep quality (Supplementary

Table S4.1); (b) job satisfaction (Supplementary Table S4.2); and (c) personality and

professional quality of life (Supplementary Table S4.3).

74

ProQoL components were strongly but differentially correlated with MEQ, PSQI, JSS, and TIPI

(Table 4.2). The majority of correlations for CS were positive, with a range of significant

correlation coefficients between rs = 0.179 to 0.768. Most correlations with burnout were

negative, with a range of significant correlation coefficients between rs = - 0.216 and - 0.589.

Fewer significant correlations existed for STS, but again, the majority of the significant

correlations were negative, in the range of rs = - 0.034 to - 0.512. As expected, the correlations

between the three ProQoL domains and PSQI were the inverse of the rest of the significant

correlations for the group, with a range of rs = - 0.322 to 0.463.

75

Table 4.2 Non-parametric correlations between ProQoL domains (columns) and covariates (rows) Compassion

Satisfaction Burnout Secondary

Traumatic Stress MEQ .185* -.106 .053 PSQI -.322** .463** .270** Job Satisfaction

Pay .316** -.280** -.020 Promotion .272** -.216* .066

Supervision .386** -.398** -.110 Benefits .203* -.204* -.034*

Contingent Rewards .452** -.478** -.157 Operating Procedures .233** -.414** -.257**

Coworkers .470** -.504** -.235** Nature of Work .768** -.589** -.237** Communication .413** -.504** -.177*

Total .543** -.552** -.171

TIPI Extraversion .179* -.133 -.100

Agreeableness .390** -.399** -.269** Conscientiousness .126 -.128 -.247**

Emotional Stability .384** -.543** -.512** Openness to Experience .249** -.295** -.180*

Spearman’s rho is shown * p<0.05; **p<0.01

76

Following univariate regression analyses of the predictive value of MEQ, PSQI, and each

component of the JSS and TIPI scales on each ProQoL domain, we assessed the extent to which

significant univariate predictors combined as a multiple regression model (backwards multiple

regression; data not shown) to predict CS, burnout, and STS (Table 4.3). MEQ was not

significantly related to the ProQoL domains when considered with the other factors, and is

therefore not included in Table 4.3. However, the univariate regression analysis for MEQ alone

indicated that chronotype can account for 2.7% of the explained variability in CS development

(univariate regression data not included in tables). Based on the correlation data (Table 4.2), it

appears that greater M type tendency is associated with higher CS. The adjusted R2 values were

higher for CS and burnout than STS, suggesting that other factors not explored here may

contribute to the development of STS. Overall, components of job satisfaction and certain

personality factors contributed heavily to all ProQoL domains. Interestingly, according to these

models, PSQI only contributes to burnout, while MEQ does not contribute significantly to any of

the separate domains of ProQoL when considered together with other factors.

77

Table 4.3 Final models in backwards multiple regression with professional quality of life (CS, BO, and STS) as dependent variables, and significant values* (as tested by univariate regression) for MEQ, PSQI, JSS and TIPI as independent variables ProQoL domain Included covariates B (SE) ß p value Adjusted R2 CS .642 Constant 6.878 (2.490) .007 JSS Contingent Rewards 0.176 (0.082) .136 .035 Nature of Work 1.036 (0.109) .632 .000 TIPI Agreeableness 0.965 (0.319) .174 .003 Openness to Experience 0.849 (0.307) .153 .007 Burnout .593 Constant 4.193 (.154) .000 PSQI 0.088 (0.032) .172 .007 JSS Contingent Rewards -0.010 (0.004) -.174 .013 Operating Procedures -0.020 (0.004) -.268 .000 Nature of Work -0.018 (0.005) -.255 .000 TIPI Emotional Stability -0.059 (0.013) -.297 .000 Openness to Experience -0.038 (0.015) -.160 .011 STS .281 Constant 3.750 (0.112) .000 JSS Operating Procedures -0.015 (0.006) -.186 .016 TIPI Emotional Stability -0.105 (0.016) -.484 .000

Note. B (SE) represents the unstandardized coefficient (standard error). ß represents standardized coefficients. *Only covariates with p < 0.10 in the univariate linear regression were included in the analyses

78

In these regressions, the explained variance is higher for CS (adjusted R2 = .642), and burnout

(adjusted R2 = .593) as compared to the variance for STS (adjusted R2 = .281).

Furthermore, CS and burnout, and burnout and STS shared similar covariates, yet CS and STS

did not share covariates.

Job satisfaction significantly predicts the development of each component of ProQoL. Given that

job satisfaction may be masking the contribution of the other covariates being examined,

backwards multiple regression was run again to assess the various domains of ProQoL, this time

without including the influence of job satisfaction (Table 4.4). The explained variance in these

multiple regressions is lower than when the models included JSS scores.

79

Table 4.4 Models in backwards multiple regression with professional quality of life (CS, BO, and STS) as dependent variables, and significant values* (as tested by univariate regression) for MEQ, PSQI, TIPI and working on multiple vs. single cancer sites as independent variables ProQoL domain Included covariates B (SE) ß p value Adjusted R2 CS .253 Constant 39.469 (3.568) .000 PSQI -3.290 (0.940) -.274 .001 TIPI Agreeableness 1.991 (0.429) .358 .000 Work on multiple vs. single cancer sites -2.589 (1.182) -.171 .030 Burnout .395 Constant 3.425 (0.131) .000 PSQI 0.166 (0.037) .326 .000 TIPI Agreeableness -0.049 (0.018) -.206 .009 Emotional Stability -0.067 (0.016) -.338 .000 STS .253 Constant 3.590 (0.093) .000 TIPI Emotional Stability -0.110 (0.017) -.509 .000

Note. B (SE) represents the unstandardized coefficient (standard error). ß represents standardized coefficients. *Only covariates with p < 0.10in the univariate linear regression were included in the analyses; MEQ did not remain a significant predictor for any of the dependent variables

80

Of the three ProQoL domains, the explained variance is highest for burnout (adjusted R2 = .395)

when job satisfaction is not included in the multiple regression; the explained variance for CS

and STS is equal in this model (adjusted R2 = .253). PSQI score is a covariate in determining

both CS and burnout when JSS scores are not included. Working with multiple instead of a single

cancer site is also a significant covariate in determining CS when JSS scores are not included,

however it is not a significant contributor to any other models. As with the original models,

burnout shares a particular set of covariates with CS, and burnout shares a different set of

covariates with STS; no similar set is shared between CS and STS.

4.5! Discussion

The primary objective of this study was to understand how the internal factors of chronotype and

sleep quality influence the various components contributing to ProQoL. This study focused on

healthcare professionals working directly with cancer patients in a hospital setting. Given the

known pervasive influences of both circadian rhythms and sleep on human physical, mental, and

emotional performance, it is not a surprise to find that both contribute in some way to aspects of

professional quality of life measured by the ProQoL Scale. A significant finding from this study

is that chronotype and sleep quality have distinctly different effects on CS and CF.

4.5.1! Chronotype and ProQoL

While the three components of ProQoL (CS, burnout, and STS) are consistently and strongly

predicted by aspects of job satisfaction, personality, and sleep quality, chronotype was only

correlated with CS (Table 4.2). We corroborated this using univariate linear regression analysis

(data not shown), which indicated that MEQ score predicted CS, but neither burnout nor STS.

However, in the multiple regression model (which includes personality, sleep quality, and JSS as

covariates), MEQ score did not predict CS. Therefore, these employees are able to earn

satisfaction from their work regardless of their chronotype. Chronotype appears to be a weak factor

compared with the others in producing CS. Unfortunately, the impact of chronotype could be

underestimated due to the limited range of chronotypes represented in the data set. Moreover, the

impact of chronotype depends on the synchrony of performance demand with the subject’s optimal

81

time (estimated by the MEQ score). Because the subjects were day shift workers only, the optimal

performance times would coincide with performance demands for the majority of participants, thereby

reducing the impact of chronotype on the dependent measures of CS and CF. Neither burnout nor STS

were linked to MEQ chronotype. This suggests that the underlying CF can develop independently

of whether or not one works at their optimal time.

4.5.2! Sleep and ProQoL

PSQI scores were negatively correlated with CS, and positively correlated with burnout and STS

(Table 4.2); therefore, better sleep quality is associated with higher CS and lower CF. As part of a

multivariate regression model, sleep quality negatively predicts burnout, regardless of whether or

not job satisfaction is included in the model (Table 4.3 and 4.4). As sleep score increases

(indicating worse sleep quality), one’s burnout rating also increases. Given that poor sleep quality

has been associated with perseveration (e.g., Banks & Dinges, 2007), a possible explanation is

that perseverative thoughts relating particularly to difficult work experiences result in this

increased burnout. Rzeszutek and Schier (2014) documented evidence supporting this conclusion,

indicating that among a sample of therapists, perseveration proved to be a significant predictor of

burnout symptoms.

When job satisfaction (JSS) is not included, PSQI score is an even greater predictor of burnout

(Table 4.4). This suggests that rest and good quality sleep increase one’s ability to achieve

satisfaction from providing compassionate care for patients. However, sleep quality is no longer a

significant predictor of CS if job satisfaction is included, suggesting that the benefits gained by

job satisfaction outweigh those of good sleep quality in this population.

Sleep quality does not appear to be a significant predictor of STS in this population, regardless of

the inclusion of JSS in the model, which reflects the ProQoL literature indicating that sleep

difficulties are the result of STS (Stamm, 2010) (Table 4.3 and 4.4). This implies one’s level of

STS is a predictor for sleep quality or sleep disturbance onset. Interestingly, the ProQoL survey

literature does not indicate that either CS or burnout contribute to an inability to sleep or

diminished sleep quality, and only lists poor sleep as an outcome of STS. While the lack of

predictive value for PSQI on STS in this study may be due to sample size, it seems likely that in

general, as part of a model, sleep quality is not a predictor of STS.

82

4.5.3! Personality and ProQoL

Personality traits were all positively correlated to CS and negatively correlated with burnout and

STS; although not all correlations were significant (Table 4.2). There were no overlapping

personality traits acting as predictors for both CS and STS in multivariate regression models;

however, both CS and STS each have shared personality traits predictors with burnout (Table 4.3

and Table 4.4). The same trend exists both when considering or excluding components of JSS as

predictors. This pattern of shared versus non-shared personality trait predictors for CS, burnout,

and STS occurs regardless of whether JSS components are included in the multiple regression

model.

4.5.3.1! Agreeableness

Agreeableness significantly predicts CS (higher agreeableness predicts greater CS), regardless of

whether or not JSS is considered as a predictor (Table 4.3 and 4.4). In the TIPI, agreeableness is

described on a scale of critical or quarrelsome (low agreeableness) to sympathetic and warm

(high agreeableness) (Gosling, Rentfrow, & Swann, 2003). Those who are high in agreeableness

are likely to achieve high CS from their job as they may be able to get along with their peers and

show understanding for their patients, leading to an overall positive outcome and sense of

satisfaction from their work.

When JSS components are not being considered, agreeableness also predicts burnout (higher

agreeableness is linked to lower burnout rates). Agreeableness may allow a person to avoid

overthinking certain hospital policies that may be burdening them, or make it easier to get along

with patients or family members who may be acting difficult, subsequently reducing feelings of

burnout and allowing a person to obtain greater satisfaction from the tasks they perform.

4.5.3.2! Emotional Stability

Emotional stability significantly predicts burnout and STS, regardless of whether or not JSS is

considered as a predictor. Poor emotional stability predicts higher levels of burnout and or STS

(Table 4.3 and 4.4). The TIPI describes emotional stability on a scale of anxious or easily upset

(low) to calm and emotionally stable (high) (Gosling, Rentfrow, & Swann, 2003). Those high in

emotional stability are likely better able to remain calm in the face of emotionally taxing

83

situations at work, allowing them to avoid feeling overwhelmed and bogged down by their

experiences, and able to separate their work versus private life. Being able to avoid these feelings

reduces one’s level of burnout and STS.

4.5.3.3! Openness

Openness to experience also predicts both CS and burnout (higher openness is linked to greater

CS and lower burnout), but only when JSS components are considered as predictors in the

multivariate regression model (Table 4.3). The TIPI describes openness to experience on a scale

of conventional and uncreative (low) to open to new experiences and complex (high) (Gosling,

Rentfrow, & Swann, 2003). When considering job satisfaction, openness to experience may help

an individual think outside the box, and be more creative in ways to perform their job, allowing

them to gain greater satisfaction from finding new ways to help others, or being innovative in

ways to avoid feelings of fatigue or work around hospital policies, subsequently reducing

burnout. However, future studies should further examine this relationship to understand if it is

specifically linked to JSS components, or if it is a stable strong predictor of CS and burnout.

4.5.3.4! Conscientiousness and Extraversion

Neither of these traits was significantly correlated with all the ProQoL indicators. High

conscientiousness was linked to higher STS but not to CS, nor to the rate of burnout.

Extraversion was only weakly linked to CS, and not to CF.

While these results do not indicate shared predictors between CS and STS, the existence of

overlapping personality traits that simultaneously influence CS and burnout, and burnout and

STS suggest that burnout has specific aspects that directly relate to one’s level of CS and STS.

4.5.4! Job Satisfaction and ProQoL

Average job satisfaction scores were comparable to those obtained from a larger population of

human service workers. This indicates that the current sample of participants is reflective of the

scores seen in the general population. Spearman correlations showed a significant positive

relationship between all aspects of the JSS and CS, and a significant negative association with

burnout. The positive individual correlations between the components of the JSS and CS indicate

that as one’s job satisfaction increases, so does one’s level of CS, and vice versa. The negative

84

correlations between the JSS and burnout indicate that those with higher burnout, report poorer

job satisfaction, and vice versa. However, not all aspects of the JSS are correlated with STS.

Furthermore, while almost all correlations between the JSS and STS are negative, the correlation

between the component promotion and STS is a positive correlation, but it is not significant.

Further research with a larger sample size should reassess this association to verify if a significant

correlation can be found, and whether it will be a positive or negative association.

Spearman correlations (Table 4.2) indicate that the specific components of job satisfaction that

most positively correlate with CS are nature of the work, coworkers, and contingent rewards.

Communication was also showed a significant positive correlation with CS. Nature of the work

refers to an overall sense of pleasure derived from the work tasks themselves, and indicates that

there is a sense of pride in one’s job along with the belief that one’s work serves a benefit.

Contingent rewards refer to the recognition and appreciation one receives for their job well done.

Overall, these four domains of job satisfaction refer to the benefits one sees their work provides,

the personal satisfaction one gains from the job, and the accolades received for one’s efforts.

When considering JSS as part of a multivariate regression model (Table 4.3) as a predictor of CS,

the only two components that remain significant predictors are contingent rewards and nature of

work. While total JSS score is significantly correlated to CS, it did not predict CS as part of

model, suggesting that one’s type of job, and the rewards the job brings (in the form of accolades

and appreciation from others), when considered together with other factors such as personality

and sleep quality, are the strongest predictors.

The Spearman correlations between the JSS and burnout (Table 4.2) are all significant and

negative, indicating that high job satisfaction ratings are associated with lower burnout. The four

components of JSS most significantly correlated to CS are also the most significantly correlated

to burnout. Understandably, when one receives little personal pleasure from the nature of his or

her job, and or little recognition in the form of praise or accolades from peers or patients, it is

more difficult to see that one’s work is appreciated. Accordingly, if one feels their work is

serving little or no benefit, it becomes difficult to obtain satisfaction when one feels their efforts

are undervalued. Operating procedures, which refer to one’s impression of the manageability of

workload and the extent to which one feels efforts to do a good job are either facilitated or

85

blocked by organizational policy, are also significantly negative correlated with burnout. A sense

of struggling with a large and unmanageable workload, or that one’s efforts to care for patients

are being blocked by hospital policy, would likely increase one’s job-related sense of frustration

and sense of fatigue, subsequently bolstering work-related feelings of burnout. Further research is

necessary to understand which specific operating procedures in the hospital are contributing to

increased feelings of burnout, in order to identify appropriate coping strategies. As part of a

multivariate regression model, the three components that remain significant predictors of burnout

are contingent rewards, operating procedures, and nature of work (Table 4.3).

The majority of correlations between the JSS and STS are negative, yet unlike the correlations

with CS and burnout, not all correlations with STS are significant (Table 4.2). The correlations

between the JSS and STS are also smaller than those with CS or burnout. Compared with CS and

burnout, one’s job satisfaction is less correlated to their level of STS. Of the correlations between

select components of the JSS and STS, only one’s satisfaction with work operating procedures

remains a significant negative predictor of STS when considered as part of a multivariate

regression model (Table 4.3). This model suggests low satisfaction with operating procedures is

predictive of higher STS. It is possible that certain hospital rules make staff more vulnerable to

the negative emotional effects of the job. However, as with the impact of operating procedures on

burnout, it is necessary to conduct further studies among caregivers working in a hospital setting

to understand how and why certain operating procedures are increasing one’s vulnerability to

developing the negative domains of ProQoL.

The total score for JSS, does not appear to have any predictive value as part of a model on the

development of CS, burnout or STS. This is interesting to note as it appears that specific

components of job satisfaction play more prominent roles in contributing to the development of

particular domains of ProQoL than does one’s overall level of job satisfaction. This can prove

beneficial in the workplace as it indicates specific areas that can directly affect employee levels

of CS and CF. Furthermore, as with personality, there were no overlapping predictors of job

satisfaction for CS and STS. However, both CS and burnout, and burnout and STS shared certain

JSS predictors, again supporting the idea that burnout shares associations with one’s levels of CS

and STS.

86

4.5.5! Limitations

While as a group, this sample allows for a significant effect size, it was not large enough to allow

independent assessments of the professional categories (oncologists, outpatient nurses,

pharmacists, pharmacy technicians, radiation therapists). Separate assessments will paint a clearer

picture of whether certain careers are differentially influenced by M versus E preference, sleep

quality, job satisfaction components, or different personality traits in the development of CS,

burnout, and STS. The individuals included in this study are a cross section of oncology

caregivers, and are sampled from a single hospital. Subject samples from different hospitals will

be necessary to determine how factors such as job satisfaction might contribute to CS and CF; it

may be important to give this current population further examination given the elevated level of

professional quality of life which they report.

The group included a limited range of chronotypes, with disproportionate makeup of N and M

type subjects as compared to E types. A broader range approaching that of the larger population,

may reveal a greater influence of M versus E preference on one’s CS and CF. Given that the

group was sampled from dayshift workers, these individuals may have a greater propensity

towards morningness in this population. Still, even without a large number of E type dayshift

workers, a univariate regression effect of chronotype on CS score was still present.

4.6! Conclusion

We have found that job satisfaction, personality, and sleep quality all are significant predictors of

CS, burnout, and STS. On the other hand, MEQ score was a significant predictor of CS only at

the univariate regression level, and did not show significance in the multiple regression model for

either CS, burnout or STS. But while chronotype was only a weak predictor in our study, it

cannot be discounted as a factor, and may be more relevant in other groups. Importantly, the M

chronotype was over-represented in our subject group, possibly due to self-selection of work

schedule. A more normal distribution of MEQ scores across the E to M spectrum may reveal a

more substantial influence of chronotype, with a greater proportion of individuals working at their

less-than-optimal times.

87

Job satisfaction is important in promoting CS while reducing burnout and STS. However, as CS

and STS shared no predictor variables, healthcare workers could be obtaining CS from their job

while simultaneously developing STS from job related stresses and emotions. It is particularly

important to monitor for the development of STS among all workers, even those who seem

satisfied with their job.

Sleep quality is highly predictive of burnout. Although our data do not indicate the causal

relationship (poor sleep could lead to burnout, or burnout due to other factors, could produce poor

sleep), it seems likely that the relationship would be reciprocal. The strong relationship suggests at least

that, regardless of all other factors, promoting good sleep quality among ambulatory healthcare

providers working in hospitals is beneficial in many ways.

88

Supplementary Table S4.1 Spearman correlations (rs) between continuous variables, and point biserial correlations for categorical variables (demographics, chronotype and sleep quality)

Demographics MEQ PSQI Age Genderƒ Yrs in

Oncology Yrs in

Healthcare No. of Sites

Worked‡

Age 1.000 .092 .796** .928** .009 .360** .034 Genderƒ .092 1.000 .082 .149 .206* .027 .071 Yrs in Oncology .796** .082 1.000 .861** .061 .274** .022 Yrs in Healthcare .928** .149 .861** 1.000 .061 .361** .063 No. of Sites Worked‡ .009 .206* .061 .061 1.000 .147 .017 MEQ .360** .027 .274** .361** .147 1.000 -.172 PSQI .034 .071 .022 .063 .017 -.172 1.000 Job Satisfaction

Pay .190* -.186* .112 .134 -.064 .104 .035 Promotion .003 -.173 -.055 -.036 -.124 .091 -.099

Supervision .057 .041 -.119 .004 -.063 -.002 -.037 Benefits -.025 -.201* -.043 -.049 -.033 .138 -.103

Contingent Rewards -.033 -.135 -.139 -.103 -.054 .019 -.177* Operating Procedures -.016 .030 -.074 -.050 .086 .019 -.109

Coworkers .073 -.068 -.031 .034 -.004 .115 -.282** Nature of Work .199* -.115 .031 .126 -.084 .200* -.317** Communication -.062 -.002 -.236** -.111 .172 .045 -.213*

TOTAL .042 -.123 -.104 -.028 -.024 .098 -.187* TIPI

Extraversion .147 .121 .078 .127 -.034 .032 -.125 Agreeableness .033 .470 .012 .060 .029 .100 -.073

Conscientiousness .142 -.028 .074 .129 -.045 .151 -.047 Emotional Stability .176* -.031 .185* .196* .049 .105 -.266**

Openness to Experience .005 .130 .040 .034 .002 .007 -.160 ProQoL

CS .145 -.021 .779 .092 -.057 .178* -.324** Burnout .105 .041 .109 .114 -.051 -.075 .406**

STS -.046 .159 -.101 -.014 -.030 .118 .272** * p < .05, ** p < .01 ƒ Coding: 1=male, 2=female; ‡ Coding: 1=single, 2=multiple

89

Supplementary Table S4.2 Spearman correlations (rs) between continuous variables, and point biserial correlations for categorical variables related to job satisfaction

Job Satisfaction Survey Pay Promotion Supervision Benefits Contingent

Rewards Operating Procedures

Coworkers Nature of Work

Commu-nication

TOTAL

Age .190* .003 .057 -.025 -.033 -.016 .073 .199* -.062 .042 Gender ƒ -.186 -.173 .041 -.201* -.135 .030 -.068 -.115 -.002 -.123 Years in Oncology .112 -.055 -.119 -.043 -.139 -.074 -.031 .031 -.236** -.104 Years in Healthcare .134 -.036 .004 -.049 -.103 -.050 .034 .126 -.111 -.028 No. of Sites Worked‡ -.064 -.124 -.063 -.033 -.054 .086 -.004 -.084 .172 -.024 MEQ .104 .091 -.002 .138 .019 .019 .115 .200* .045 .098 PSQI .035 -.099 -.037 -.103 -.177* -.109 -.282** -.317** -.213* -.187* Job Satisfaction

Pay 1.000 .531** .432** .541** .483** .109 .398** .354** .388** .686** Promotion .531** 1.000 .501** .460** .587** .088 .319** .307** .404** .681**

Supervision .432** .501** 1.000 .321** .620** .333** .513** .486** .600** .771** Benefits .541** .460** .321** 1.000 .492** .142 .342** .269** .422** .656**

Contingent Rewards .483** .587** .620** .492** 1.000 .245** .542** .500** .567** .823** Operating Procedures .109 .088 .333** .142 .245** 1.000 .318** .110 .425** .395**

Coworkers .398** .319** .513** .342** .542** .318** 1.000 .513** .593** .709** Nature of Work .354** .307** .486** .269** .500** .110 .513** 1.000 .490** .602** Communication .388** .404** .600** .422** .567** .425** .593** .490** 1.000 .774**

TOTAL .686** .681** .771** .656** .823** .395** .709** .602** .774** 1.000 TIPI

Extraversion .120 -.017 .046 -.036 -.019 -.063 .018 .099 .051 .046 Agreeableness -.120 -.250** .015 -.137 -.086 .134 .103 .314** .025 -.065

Conscientiousness -.134 -.059 -.032 -.081 -.118 -.090 -.172 .158 -.001 -.101 Emotional Stability .059 -.006 .164 -.043 .130 .083 .269** .324** .192* .154

Openness to Experience

-.061 -.204* -.101 .125 -.111 -.119 -.029 .103 -.029 -.097

ProQoL CS .273** .221* .380** .187* .416** .151 .436** .742** .388** .470**

Burnout -.263** -.167 -.342** -.223* .432** -.373** -.483** -.499** -.474** -.495** STS .011 .126 -.108 -.021 -.122 -.214* -.245** -.248** -.143 -.137

* p < .05, ** p < .01; ƒ Coding: 1=male, 2=female; ‡ Coding: 1=single, 2=multiple

90

Supplementary Table S4.3. Spearman correlations (rs) between continuous variables, and point biserial correlations for categorical variables related to personality and professional quality of life.

Ten Item Personality Inventory ProQoL Extraversion Agreeableness Conscientiousness Emotional

Stability Openness to Experience

CS Burnout STS

Age .147 .033 .142 .176* .005 .145 .105 -.046 Gender ƒ .121 .064 -.028 -.031 .130 -.021 .041 .159 Years in Oncology .078 .012 .074 .185* .040 .025 .109 -.101 Years in Healthcare .127 .060 .129 .196* .034 .092 .114 -.014 No. of Sites Worked‡ -.034 .029 -.045 .049 .002 -.057 -.051 -.030 MEQ .032 .100 .151 .105 .007 .178* -.075 .118 PSQI -.125 -.073 -.047 -.266** -.160 -.324** .406** .272 Job Satisfaction

Pay .120 -.120 -.134 .059 -.061 .273** -.263** .011 Promotion -.017 -.250** -.059 -.006 -.204* .221* -.167 .126

Supervision .046 .015 -.032 .164 -.101 .380** -.342** -.108 Benefits -.036 -.137 -.081 -.043 -.125 .187* -.223* -.021

Contingent Rewards -.019 -.086 -.118 .130 -.111 .416** -.432** -.122 Operating Procedures -.063 .134 -.090 .083 -.119 .151 -.373** -.214*

Coworkers .018 .103 -.172 .269** -.029 .436** .483** -.245** Nature of Work .099 .314** .158 .324** .103 .742** -.499** -.248** Communication .051 .025 -.001 .192* -.029 .388** -.474** -.143

TOTAL .046 -.065 -.101 .154 -.097 .470** -.495** -.137 TIPI

Extraversion 1.000 .065 .130 .141 .421** .171 -.146 -.124 Agreeableness .065 1.000 .240** .484** .292** .363** -.375** -.282**

Conscientiousness .130 .240** 1.000 .290** .136 .181* -.143 -.264** Emotional Stability .141 .484** .290** 1.000 .312** .405** -.528** -.504**

Openness to Experience .421** .292** .136 .312** 1.000 .267** -.279** -.169 ProQoL

CS .171 .363** .181* .405** .267** 1.000 -.651** -.189* Burnout -.146 -.375** -.143 -.528** -.279** -.651** 1.000 .502**

STS -.124 -.282** -.264** -.504** -.169 -.189* .502** 1.000 * p < .05, ** p < .01; ƒ Coding: 1=male, 2=female; ‡ Coding: 1=single, 2=multiple

91

Chapter 5!

! Primary Circadian Impacts on Patients, Caregivers, and Dyads

5.1! Abstract

We assessed the influence of chronotype on one’s memory for their sleep quality and coping,

(including their compassion and prosocial behaviour), among cancer patients and spousal

caregivers across chemotherapy treatment. Mixed ANOVAs were used to assess patient and

caregiver responses for potential relationships between chronotype and one’s recalled sleep

quality or coping (assessed by the University of Toronto Inventory of Morningness Eveningness

(UTIME)). Among patients without a caregiver involved in the study, chronotype and sleep were

correlated; caregivers and patients with a caregiver participating in the study, did not show a

correlation between chronotype and sleep quality. In the total patient group, peaks in recalled

coping, compassion and prosocial behaviour reflected chronotype; morning (M) types typically

indicated earlier peak performance compared to neither (N) and evening (E) types. To an extent,

peak performance times were recalled earlier in the day as treatment progressed. Caregivers only

reported a chronotype dependent performance difference at endpoint, relating to their need for

time and space to themselves. Patient chronotype is significantly related to recalled performance

across the day and to an extent to sleep quality. Results among patients with caregivers involved

in the study suggest the possibility that spending too much time going through the cancer

treatment process together may produce poorer sleep quality. Caregivers may not report

chronotype dependent performance fluctuations for several reasons, including being required to

function at their non-optimal time, or believing they were providing consistent care at all times.

92

5.2! Introduction

Current published data suggest that two in five Canadians will develop cancer in their lifetime. It

was estimated that in 2016 alone, there would be 202,400 new cancer diagnoses (Canadian

Cancer Statistics, 2016a). While research on cancer care is dominated by diagnosis, treatment,

and patient care, the disease also has profound effects on caregivers and others who are

associated with the affected individuals. Caregivers in hospitals or hospices, including doctors,

nurses, radiation therapists, and or pharmacists, who provide both physical and emotional

medical attention as part of their professional and voluntary work. Outside of the hospital, non-

vocational caregivers, including typically family members and/or friends also provide physical

and emotional support. In this chapter, we refer to caregivers as the individuals who each patient

identifies as their primary provider of non-vocational care or support, outside of the hospital. We

specifically sought caregivers that were spouses of patients also involved in the study, with the

intention that the same person would be providing consistent care across treatment, and would

identify themselves as being closely involved in the patient’s care. While other family members

and friends can also be caregivers, we sought caregivers that would be closely involved across

treatment throughout the day in their spouse’s care, with the intention of examining the influence

of such a role on patients, the spousal caregivers themselves, and on the dyad relationship.

While cancer research is dominated by understanding, diagnosing and treating the disease per se,

other factors beyond the biology of disease also affect patient and caregiver wellbeing. The

potential impact of temporal biology, specifically, has been given comparatively little attention

in either the understanding of cancer or its treatment (e.g., Hede, 2009), yet it has long been

recognized that both are subject to circadian modulation (e.g., Innominato, Levi, & Bjarnason,

2010; Levi, 2001; Levi, 2006).

Chronotype and sleep timing are expressions of the temporal biology of human beings. When

quantified, they are a means of recognizing and categorizing human behavior and physiology

based on both physiological changes and self-perceptions of optimal timing for physical,

emotional, and mental performance. Chronotype is a stable trait that persist for extended time

periods across one’s lifespan, and demonstrates large shifts coinciding with major changes

during development and aging (Schmidt, Collette, Cajochen & Peigneux, 2007). As chronotype

93

shifts, it affects changes in sleep timing (Koscec, Radosevic-Vidacek & Bakotic, 2014; Simpkin

et al., 2014).

Chronotype has been shown to strongly influence cognitive and emotional regulation and

reactivity in human beings. Cognitive task performance typically changes across the day with

high and low performance correlated with chronotype (for review, see Schmidt, Collette,

Cajochen & Peigneux, 2007). Emotional regulation is related to cognitive function. For example,

cognitive reappraisal, a form of emotional regulation, requires individuals to reinterpret the

significance of emotion-eliciting stimuli or situations. Fluid cognitive ability is necessary to

adjust and reinterpret one’s assessment of stimuli and adjust emotional responding (Opitz, Lee,

Gross & Urry, 2014). In the general population, performance recollection for activities with

different cognitive, physical and emotional demands varies. We have shown previously that

memory for performance throughout the day on tasks with high cognitive demands is

significantly higher than it is for emotionally based tasks (Bellicoso, 2010). Since emotional

regulation, and coping in particular, depend on various cognitive processes (e.g., problem

solving, reframing), it is important to understand how coping is related to chronotype, as this

may shed light on how people form memories for their coping ability.

Sleep quality influences performance ability on a variety of domains. Cognitive function is

enhanced by good sleep quality, and reduced following a poor night of sleep (e.g., Lim &

Dinges, 2008). Cancer patients and caregivers (including medical staff, family and friends) often

report sleep disruptions and poor sleep quality (Ancoli-Israel, Moore & Jones, 2001; Zhang,

Yao, Yang & Zhou, 2014). Sleep disruptions among patients have been attributed to a variety of

predisposing, precipitating, and perpetuating factors. For example, Savard and Morin (2001)

suggest cancer patients may have a heightened predisposition to poor sleep quality due to factors

such as familial history of insomnia, aging, hyper-arousability, and female gender. In the general

population, these factors contribute to poor sleep. Savard and Morin (2001) suggest precipitating

factors that bring about poor sleep may include the cancer itself, medical treatments, disease

related pain or treatments, and the emotional impact that a diagnosis carries.

Therefore, we have hypothesized that the prognosticated outcome of cancer therapy is

determined not only by the aggressiveness of the disease and the type of treatment but also by

both the inherent rhythmicity of the biology of cancer and the rhythmic constitution of the

94

patient and their environment. For example, chemotherapy negatively impacts cognitive function

and emotional regulation, and has been linked with increased cognitive dysfunction, known as

chemo brain or chemo fog, a well-established clinical syndrome where patients often report

cognitive deficits at some time during and or following treatment (Wang, Walitt, Saligan, Tiwari,

Cheung & Zhang, 2015). Both cancer patients and caregivers also typically report (for a variety

of reasons) that stress negatively impacts cognitive function, reflected in a reduction of prefrontal

cortex activity underlying many high order cognitive processes (Arnsten, 2009). Cognitive

deficits have been reported by caregivers of ill friends or family members with a range of

physical and or mental conditions, who themselves report elevated stress levels (Jonsdottir et al.,

2013; Oken, Fonareva & Wahbeh, 2011).

In this chapter, we focus on the influence of chronotype and sleep quality on coping with cancer.

Specifically, we look at how the patient and caregiver provide different perspectives on how they

react to the medical situation, and how each views the others’ behavior. We sought to understand

how chronotype and sleep quality influence the judgment of the patient and caregiver, by

assessing how these two factors are related to the subjective recall of sleep quality and coping

ability. Because a cancer diagnosis precipitates a sequence of medical and personal responses,

we attempted to determine how these factors change across the course of cancer therapy.

5.3! Materials and Methods

5.3.1! Participants and Procedures

Prospective patients receiving cancer treatment at Sunnybrook Health Sciences Centre, Odette

Cancer Centre (Toronto, Ontario, Canada) and when applicable, their spousal caregivers, were

identified by their medical oncologist or nurse, and briefed on the general idea of the study. If

patients and caregivers reported interest in participating, they were approached by the study

coordinator, who explained the details and time commitment of the study. Inclusion criteria

indicated that patients had to be female and receiving adjuvant or neoadjuvant chemotherapy for

stage 1, 2, or 3 breast cancer; caregivers had to have a spouse undergoing either of these

treatments who was enrolled in the study. Participants must be 19 or older, be able to read, write

and speak English at least at a basic level. Participants continuing to follow a shiftwork work

95

schedule during the treatment period were excluded. One female caregiver partook in the study,

however, neither her nor her partner returned correctly completed, usable data.

Subjects were informed that their participation was voluntary and would have no bearing on their

treatment. They were informed that while no formal remuneration would be provided, their

participation in the study might provide the opportunity to reflect on personal coping ability, and

factors that influenced their coping. Usable data from 109 participants (86 patients, 23

caregivers) was included in this study. Among those who had signed up but withdrew, the

general reason for withdrawal was that people felt the study would be too time consuming.

5.3.2! Measures

Participants completed baseline, midpoint and endpoint surveys. Baseline questionnaire

packages included background demographics, the Horne-Östberg Morningness Eveningness

Questionnaire (MEQ), Pittsburgh Sleep Quality Index (PSQI), and the University of Toronto

Inventory of Morningness Eveningness (UTIME) (for descriptions, refer to section 2.2

Instruments). Midpoint and endpoint packages included the PSQI and UTIME. Baseline

packages were completed prior to or on the first day of chemotherapy treatment. Midpoint

packages were completed at the halfway session of a patient’s treatment (the treatment session

number varied depending on the particular type of chemotherapy being received). Endpoint

packages were typically completed at the patient’s final chemotherapy session, however, given

that some patients’ final chemotherapy sessions were cancelled upon assessing their health that

day, some endpoint packages were completed at what would have been the final chemotherapy

session.

Actigraphy was attempted as an objective measure of participants’ circadian rhythms. However,

voluntary compliance (e.g., wearing the watch, keeping continuous activity and sleep reports) for

seventeen participants (14 patients; 3 caregivers) was highly variable among participants, and

due to the possibility that actigraphy might have a negative effect on the patients’ treatment, this

approach was dropped from the study.

5.3.3! Statistical Analysis

These data were assessed in three separate analyses: (1) comparing data between patients as an

entire group, (2) comparing data between caregivers as an entire group, and (3) comparing data

96

between patients and their matched caregivers. Descriptive and frequency data were calculated

for each separate analysis. The MEQ was scored according to the criteria presented by Horne and

Östberg (1976). The PSQI was calculated according to the manual, and using software

downloaded from the PSQI website (Buysse, Reynolds III, Monk, Berman, Kupfer, 1988).

UTIME questions were scored according to the criteria presented in Bellicoso (2010). Spearman

correlations were calculated as they are more robust to outliers in non-normally distributed data

than Pearson correlations.

The analyses of patient and caregiver populations were conducted separately, followed by a

comparison of patients with their caregivers. Exploratory analyses to assess whether data were

normally distributed were conducted, and data were winsorized when necessary. For the

population analyses, mixed ANOVAs were performed (with one between subject factor –

chronotype, and one within subject factor – PSQI or UTIME score). Natural logarithm

transformations were not used as they did not generally work well to normalize data, at times

further skewing the data. For the comparison of patients with their caregivers who also took part

in the study, the Mann-Whitney U Test (Wilcoxon Rank Sum Test) was used due to the non-

parametric nature of the data and small sample size. Mixed ANOVAs (with two between subject

factors, and one within subject factor) were used to assess changes in PSQI and UTIME scores

across treatment.

5.4! Results

5.4.1! Descriptive Statistics

5.4.1.1! Overall Demographic Comparisons

Descriptive and frequency statistics for continuous and categorical demographic variables, along

with the data from the MEQ, PSQI, and UTIME questionnaires are presented in Table 5.1. The

average age of the 86 breast cancer patients completing the study was 50.4 yr ± 11.32. Among

the patient participants living with a partner (72), 69 (80.2%) listed their partner as their primary

caregiver. All participating caregivers (23) were living with their patient partners during the

study. The average age of patients with participating partners was not different from that of

patients without, and the average age of the partners was not significantly different from

patients’ age.

97

Overall, about half of the patient population (52.3%) felt that alternate or outside factors would

not influence how they would cope with the disease or treatment. This belief was more strongly

felt by patients who were either married or living with a partner (57.1%). Significantly, among

the patients with caregivers involved in the study, the majority (60.9%) believed the opposite -

that other factors would influence their coping. On the other hand, the majority of caregivers

(60.9%) did not share this belief.

5.4.1.2! Morningness-Eveningness Distribution

The average MEQ score for the overall patient population was 56.3 (i.e., indicative of being N

type). The distribution of patient MEQ scores indicated that 47.7% were M types, 46.5% were N

types, and 5.8% were E types. However, the distribution of scores is different for patients

without caregivers in the study compared with those with caregivers who participated. Although

it is a smaller group, the patients with caregivers in the study were strongly M type (16/23 or

69.4%), and the average MEQ for this group is M type (MEQ average score = 59.96); whereas

the majority of those without caregivers in the study were N type (33/63 or 52.4%) with an

average MEQ score of 55. Furthermore, caregivers in the study were also more strongly M types

(12/23 or 52.2%) (Data are shown in Table 5.2).

5.4.1.3! Sleep Quality Comparison

The average sleep quality score indicated by the PQSI (Table 5.2) for the overall patient

population across treatment indicated poor sleep. Sleep quality worsened from baseline to

endpoint (8.36, to 8.65), though not significantly (p = .384). Between all three phases, this was

the largest change in sleep quality (baseline to midpoint: p = .871; midpoint to endpoint: p =

.422). On the average, sleep quality was higher for patients without caregivers participating than

for those with caregivers. Sleep quality remained essentially identical for all participants

throughout the treatment. Sleep quality was higher for caregivers than for patients throughout the

study.

5.4.1.4! UTIME Performance Results

Responses to seven UTIME scenarios were accumulated from patients and caregivers at three

points in the treatment (beginning, midpoint, end). Midpoint was timed to coincide with any

change in treatment regimen which is part of a normal protocol. Average scores for each group

98

and time are presented graphically in Figure 5.1. Group means did not differ significantly, and

did not change significantly over time. However, there were some consistent results and trends in

responses to the different scenarios. The consistently lowest score was for UTIME #5 with a

score always indicative of peak feeling of fatigue in the second half of the day. The second

lowest score was consistently seen for UTIME #7 (when do you most, average, and least need

time and space to yourself. Relatively high scores were provided consistently in response to

UTIME #1 (when are you able to best, average, and worst cope with the stress associated with

your diagnosis?) and UTIME #4 (when are you most, average, and least alert?). The result

indicates that patients felt that they could cope with the diagnosis and treatment earlier in the day

than other issues that they might face, which corresponds well with their reports of increased

alertness earlier in the day.

In addition, the group of patients whose caregivers participated in the study showed a reduced

score on all but one UTIME question at the midpoint of their treatment. This may reflect a

different outlook or self-perception that is related to an extended presence of the caregiver. The

trend may also be related to the lower sleep quality that was found in this group. The reason for

this is unclear, and is addressed in the Discussion.

99

Table 5.1 Means and frequencies of demographic and questionnaire data for patients and caregivers

Total patient population Patients without study involved caregiver

Patients with study involved caregiver

Total caregiver population

Mean (SD) Freq. (%) Mean (SD) Freq. (%) Mean (SD) Freq. (%) Mean (SD) Freq. (%) Total population size 86 63 23 23 Age 50.38

(11.32) 49.79

(10.81) 52.00

(12.73) 54.22 (13.21)

Relationship Status In relationship, not living together

3 (3.5%) 3 (4.8%) - -

Living together 6 (7.0%) 5 (7.9%) 1 (4.3%) 1 (4.3%) Married, living together 66 (76.7%) 44 (69.8%) 22 (95.7%) 22 (95.7%) Divorced 2 (2.3%) 2 (3.2%) - - Separated 1 (1.2%) 1 (1.6%) - - Single 7 (8.1%) 7 (11.1%) - - Widowed 1 (1.2%) 1 (1.6%) - - Patient with partner as caregiver

Yes 69 (80.2%) 46 (73.0%) 23 (100%) n/a No 3 (3.5%) 3 (4.8%) - n/a Not applicable 14 (16.3%) 14 (22.2%) - n/a Alternate coping influences Yes 40 (46.5%) 26 (41.3%) 14 (60.9%) 8 (34.8%) No 45 (52.3%) 36 (57.1%) 9 (39.1%) 14 (60.9%) Not applicable 1 (1.2%) 1 (1.6%) - 1 (4.3%)

100

Table 5.2 Means and frequencies for chronotype and sleep quality questionnaires for patients and caregivers Total patient population Patients without study

involved caregiver Patients with study involved caregiver

Total caregiver population

Mean (SD) Freq. (%) Mean (SD) Freq. (%) Mean (SD) Freq. (%) Mean (SD) Freq. (%) A.! MEQ All scores 56.33 (9.37) 55.00

(9.707) 59.96

(7.425) 57.22

(8.464)

Morning type 41 (47.7%) 25 (39.7%) 16 (69.4%) 12 (52.2%) Neither type 40 (46.5%) 33 (52.4%) 7 (30.4%) 10 (43.5%) Evening type 5 (5.8%) 5 (7.9%) - 1 (4.3%) B.! PSQI Baseline 8.36 (3.63) 7.92 (3.521) 9.57 (3.727) 6.87 (3.152) Midpoint 8.42 (3.24) 8.29 (3.314) 8.78 (3.089) 6.91 (3.147) Endpoint 8.65 (3.37) 8.32 (3.440) 9.57 (3.057) 6.87 (3.622)

101

Figure 5.1a UTIME scores across treatment for the total patient group

102

Figure 5.1b UTIME scores across treatment for patients without a caregiver involved in the

study

103

Figure 5.1c UTIME scores across treatment for patients with a caregiver involved in the study

104

Figure 5.1d UTIME scores across treatment for caregivers

105

5.4.2! Analysis 1: Total Patient Population Analysis

5.4.2.1! Multifactorial Correlation Analysis

The treatment of breast cancer occurs in stages and is a lengthy and highly personal

(individualized) process. Coping, therefore, is affected by several factors. In this section, we

consider the influences of chronotype on sleep quality and coping performance at each phase of

assessment across chemotherapy treatment.

5.4.2.2! Global Correlations Among Patients’ MEQ, PSQI and UTIME data

Spearman correlations were determined first for the data from the three questionnaires (MEQ,

PSQI, and UTIME) (Table 5.3). Among the 86 breast cancer patients, correlations between

chronotype and sleep quality were not significant, although there was an apparent trend toward

significance as the course of treatment progressed (Table 5.3). However, a distinction emerged

between patients with and without caregivers. Spearman correlations between MEQ and PSQI

for the 23 breast cancer patients whose partner partook in the study were not significantly

correlated, and the non-significant baseline correlation became even less significant by endpoint.

Contrary to this, among the 63 breast cancer patients without a caregiver partaking in the study,

the correlations suggested a significant negative association between chronotype and sleep

quality across treatment.

Spearman correlations between the seven UTIME scores and MEQ and for the complete group

of 86 breast cancer patients were significantly correlated for nearly all questions across treatment

(Table 5.3). The MEQ-UTIME #5 correlation (when are you most, average, and least fatigued

during the day?) never reached significance at any assessment phase in the cancer treatment

period. UTIME #7 showed weak correlations with MEQ and reached significance (p = .003) at

the midpoint of treatment, but was again not significant by endpoint.

Sleep quality and UTIME scores were not significantly correlated across treatment (data not

shown), except for UTIME #5 at midpoint (rs = .218, p = .045), and UTIME #3 at endpoint, (rs =

.268, p = .026); UTIME #3 only applied to patients with a partner (when is your partner able to

best, average, and worst help you cope with the stress of being a cancer patient?). However,

neither correlation reached significance at either of the other two assessment phases.

106

Table 5.3 Spearman correlations for chronotype versus sleep quality and UTIME among patients

Assessment Phase MEQ vs. Baseline Midpoint Endpoint rs p rs p rs p

A.! PSQI Total patient population -.101 .353 -.163 .133 -.170 .118 Patients, caregiver involved .388 .067 .377 .076 .312 .147 Patients, caregiver not involved -.312* .013 -.339** .007 -.328** .009

B.! UTIME UTIME #1 .531** .000 (86) .468** .000 (86) .411** .000 (86) UTIME #2 .578** .000 (70) .583** .000 (67) .481** .000 (69) UTIME #3 .441** .000 (69) .344** .004 (67) .373** .002 (69) UTIME #4 .479** .000 (85) .545** .000 (86) .458** .000 (86) UTIME #5 .097 .374 (86) -.023 .834 (85) .153 .159 (86) UTIME #6 .513** .000 (68) .519** .000 (67) .492** .000 (69) UTIME #7 .202 .063 (86) .317** .003 (85) .173 .114 (85)

Note: numbers in brackets beside MEQ vs. UTIME p-values are n values * p < .05 ** p < .01

107

5.4.2.3! Total Patient Population Mixed Measures ANOVAs

Mixed measures ANOVAs were used to assess the influence of chronotype on changes in sleep

quality and UTIME scores at each assessment phase (baseline, midpoint and endpoint) across

chemotherapy treatments among the total group of 86 breast cancer patients (Table 5.3). The

within subject factor was assessment phase (score at baseline, midpoint, and endpoint). The

between subject factor was chronotype (moderately E type, N type, moderately M type, and

definitely M type); the two largest groups of these between subject factors were N types and

moderately M types.

A standard procedure was used to assess changes in sleep quality or UTIME scores at

assessment times across treatment between chronotypes. Data were assessed for outliers based

on inspection of boxplot values greater than 1.5 box lengths from the edge of the box. Outliers

(detected only among UTIME scores and not PSQI) were corrected for by winsorizing the data

in order to use a standard procedure for each analysis. The Shapiro-Wilk’s test (p > .05) was

used to assess normality of distribution for sleep quality or UTIME score at each between and

within subject level. Homogeneity of variance (p > .05) was assessed by Levene’s test of

homogeneity of variance; any violations were left untransformed and noted. Homogeneity of

covariance (p > .05) was assessed by Box’s test of equality of covariance matrices; any

violations were noted and the interaction term was not interpreted but noted for reference.

Mauchly’s test of sphericity was used to assess whether the assumption of sphericity was met (p

> .05); the Greenhouse-Geisser estimate was used when the assumption was violated. Mixed

ANOVA outputs for sleep quality and UTIME are presented in Table 5.4.

There was no significant interaction between chronotype and assessment phase on either sleep

quality or UTIME score among the total group of 86 breast cancer patients, p > .05. Only

significant between and within subject effects as noted in Table 5.4 are discussed below.

There were no significant main effects at the between or within subject level when assessing

changes in sleep quality between chronotypes or within subjects at different time points.

108

When assessing changes in UTIME #1 scores across assessment phases between chronotypes,

between subjects, the main effect of chronotype showed a significant difference in UTIME #1

score at the different assessment phases p < .001. Definitely M types had a mean UTIME #1

score 0.150 (CI 95%, .002 to .298) points higher than moderately E types, p < .05. Moderately M

types had a mean UTIME #1 score 0.076 (CI 95%, .031 to .122) points higher than N types, p <

05, and 0.122 (CI 95%, .026 to .218) points higher than moderately E types, p < .01. This

indicates that for UTIME #1 (best, average, and worst ability to cope with cancer diagnosis

across the day), patients report reaching their peak coping ability in chronotypically sequential

order, beginning with definitely M types reporting reaching their best coping point earliest in the

day compared to N and E types. Within subjects, the main effect of time showed a statistically

significant difference in UTIME #1 score at the different assessment phases, p = .019. UTIME

#1 score at endpoint showed a statistically significant increase of 0.052 (95% CI, -.090 to -0.014)

points, p < .01, over mean score at midpoint.

When assessing changes in UTIME #2 scores across assessment phases between chronotypes,

between subjects, the main effect of chronotype showed a significant difference in UTIME #2

score at the different assessment phases, p < .001. Definitely M types had a mean UTIME #2

score 0.123 (CI 95%, .005 to .241) points higher than N types, p < .05. Moderately M types had

a mean UTIME #2 score 0.077 (CI 95%, .028 to .127) points higher than N types, p < .001.

Within subjects, the main effect of time showed a significant difference in UTIME #2 score at

the various assessment phases, p = .038. There was a significant increase in UTIME #2 score

across treatment of 0.060 (95% CI, -.118 to -.002), p < .05, from baseline to endpoint.

When assessing changes in UTIME #3 scores across assessment phases between chronotypes,

within subjects, the main effect of time suggested there was a significant difference in UTIME

#3 score at the different assessment phases, p = .044. However, when considering pairwise

comparisons in mean UTIME #3 score within subjects, there were no statistically significant

differences with significance values below p < .05. The closest value to significance was at

endpoint, where mean UTIME #3 scores were 0.064 (CI 95%, -.001 to .130) points above

baseline scores, p < .100 (p = .057).

When assessing changes in UTIME #4 scores across assessment phases between chronotypes,

between subjects, the main effect of chronotype showed a significant difference in UTIME #4

109

score at the different assessment phases, p < .001. Definitely M types had a mean UTIME #4

score 0.177 (CI 95%, .025 to.329) points higher than moderately E types, p < .05. Moderately M

types had a mean UTIME #4 score 0.081 (CI 95%, .033 to .128) points higher than N types, p <

.001, and 0.159 (CI 95%, .060 to .258) points higher than E types, p < .001.

When assessing changes in UTIME #5 scores across assessment phases between chronotypes,

between subjects, the main effect of chronotype suggested there was a significant difference in

UTIME #5 score at the different assessment phases, p = .041. However, when considering

pairwise comparisons in mean UTIME #5 score between various chronotypes, there were no

significant differences with significance values, p < .05. The closest value to significance was

found between N types who had a mean UTIME #5 score -0.060 (CI 95%, -.121 to .001) points

lower than moderately E types, p = .059. While this is not considered statistically significant,

examining Tukey HSD in multiple comparisons shows that N types do have a mean UTIME #5

score -.060 (95% CI, -.119 to -.000) points below moderately E types, a significant difference, p

= .047.

When assessing changes in UTIME #6 scores across assessment phases between chronotypes,

between subjects, the main effect of chronotype showed a significant difference in UTIME #6

score at the different assessment phases, p < .01. Definitely M types had a mean UTIME #6

score 0.104 (CI 95%, .006 to.20) points higher than N types, p < .05. Moderately M types had a

mean UTIME #6 score 0.075 (CI 95%, .034 to .117) points higher than N types, p < .001.

When assessing changes in UTIME #7 scores across assessment phases between chronotypes,

between subjects, the main effect of chronotype showed a significant difference in UTIME #7

score at the different assessment phases, p = .045. Moderately M types had a mean UTIME #7

score of 0.041 (95% CI, 0.000 to 0.081) points higher than N types, p < .05.

110

Table 5.4 Mixed ANOVAs on patient sleep quality and UTIME: main effects and interactions Chronotype

(Between Subject) Assessment phase

Baseline, Midpoint, Endpoint (Within Subject)

Assessment phase x Chronotype (Within Subject)

F df p η2 F df p η2 F df p η2 PSQI 1.014 3, 82 .391 .036 0.168 2, 164 .845 .002 0.401 6, 164 .877 .014 UTIME #1 9.593 3, 82 .000** .260 4.223 1.843, 164 .019* .049 1.422 5.528, 164 .214 .049 UTIME #2 7.328 3, 62 .000** .262 3.366 2, 124 .038* .051 1.950 6, 124 .078 .086 UTIME #3 2.740 3, 61 .051 .119 3.204 2, 122 .044* .050 2.170 6, 122 .050 .050 UTIME #4 11.468 3, 81 .000** .298 0.196 1.788, 162 .798 .002 1.590 5.365, 162 .162 .162 UTIME #5 2.888 3, 81 .041* .097 0.861 2, 162 .425 .011 0.556 6, 162 .765 .020 UTIME #6 9.404 3, 61 .000** .316 1.781 2, 122 .173 .028 1.012 6, 122 .421 .047 UTIME #7 2.083 3, 81 .045* .094 1.635 2, 162 .198 .020 0.981 6, 162 .440 .035

* p < .05 ** p < .01 NOTE: PSQI: Shapiro Wilks violated for N types at baseline (p = .017) UTIME #1: Levene’s test violated at endpoint (p = .016); Box’s test violated (p = .015); Mauchly’s test of sphericity violated, χ2 (2) = 7.229, p = .027 UTIME #2: Levene’s test violated at endpoint (p = .008); Box’s test violated (p = .013) UTIME #3: Shapiro Wilk’s violated for N types at endpoint (p = .022); Box’s test violated (p = .000) UTIME #4: Shapiro Wilk’s violated for N types at baseline (p = .012); Levene’s test violated at baseline (p = .001) and midpoint (p = .000); Mauchly’s test of sphericity violated, χ2 (2) = 10.080, p = .006 UTIME #5: Shapiro Wilk’s violated for moderately M types at baseline (p = .016) and midpoint (p = .044), and definitely M types at midpoint (p = .044); Levene’s test violated at baseline (p = .049); Box’s test violated (p = .023)

111

5.4.3! Analysis 2: Total Caregiver Population Analysis

5.4.3.1! Global Correlations Among Caregivers’ MEQ, PSQI and UTIME data

Spearman correlations indicated no significant association between MEQ and PSQI scores for

the 23 caregivers (Table 5.5). The association was positive, and significance values at each

assessment phase varied greatly from one another. No trend suggested the association was

moving towards or further away from significance. Spearman correlations between MEQ and

UTIME score (Table 5.5) were also non-significant for all questions, except for UTIME #7 at

endpoint, p < 001. Baseline data suggested a positive trend in the association between MEQ and

all UTIME questions. However, while the association remained positive for all other questions at

midpoint and endpoint, MEQ and UTIME #5 showed a negative relationship at midpoint and

endpoint. As in Analysis 1, there were only few significant correlations between PSQI and

UTIME scores (not reported here), however they were not consistently significant across

treatment. The correlation between PSQI and UTIME #2 was only significant at baseline (rs =

.444, p = .034), and endpoint (rs = .441, p = .035). UTIME #3 and UTIME #6 only showed a

significant correlation to PSQI at endpoint (rs = .458, p = .028, and rs = .531, p = .009,

respectively). No significant correlations existed between PSQI and UTIME score at midpoint.

112

Table 5.5 Spearman correlations for caregiver chronotype versus sleep quality and UTIME

Assessment Phase MEQ vs. Baseline Midpoint Endpoint rs p rs p rs p A.! PSQI

.091 .681 .156 .477 .054 .805 B.! UTIME UTIME #1 .130 .553 .215 .325 .374 .078 UTIME #2 .102 .645 .167 .448 .205 .348 UTIME #3 .394 .063 .170 .438 .340 .112 UTIME #4 .165 .451 .107 .626 .282 .192 UTIME #5 .194 .375 -.099 .654 -.168 .442 UTIME #6 .320 .137 .079 .720 .059 .790 UTIME #7 .366 .094 .036 .870 .653** .001

** p < .01

113

5.4.3.2! Caregiver Population Mixed Measures ANOVAs

A standard procedure was used to perform mixed measures ANOVAs to assess changes in sleep

quality or UTIME scores at assessment times across treatment between chronotypes among the

total group of 23 spousal caregivers to breast cancer patients. The within subject factor was

assessment time (score at baseline, midpoint, and endpoint). The between subject factor was

chronotype (divided into two categories: moderately E type and N type versus moderately and

definitely M type). There was only one moderately E type, and one definitely M type individual,

therefore, these scores were joined with N type and moderately M type chronotypes,

respectively.

Data were assessed for outliers based on inspection of boxplot values greater than 1.5 box

lengths from the edge of the box. Outliers (detected only among UTIME #3 - #7) were corrected

for by winsorizing the data in order to use a standard procedure for each analysis. The Shapiro-

Wilk’s test (p > .05) was used to assess normality of distribution for sleep quality or UTIME

score at each between and within subject level. Homogeneity of variance (p > .05) was assessed

by Levene’s test of homogeneity of variance; any violations were left untransformed and noted.

Homogeneity of covariance (p > .05) was assessed by Box’s test of equality of covariance

matrices; any violations were noted and the interaction term was not interpreted but noted for

reference. Mauchly’s test of sphericity was used to assess whether the assumption of sphericity

was met (p > .05); the Greenhouse-Geisser estimate was used when the assumption was violated.

Mixed ANOVA outputs for sleep quality and UTIME are presented in Table 5.6.

Between subjects, the main effect of chronotype did not show a significant difference in sleep

quality or UTIME#1 - #7 scores at the different assessment phases, p > .05. Within subjects, the

main effect of time did not show a significant difference in sleep quality or UTIME#1 - #7 score

at the different assessment phases, p > .05. There were no significant interactions between

chronotype and assessment phase across treatment on sleep quality or UTIME #1-#6. However,

there was a significant interaction between chronotype and assessment phase on UTIME #7

score, p = .010. Specifically, there was a highly significant difference in UTIME #7 score

between chronotypes at the endpoint of treatment, F(1, 21) = 14.649, p = .001, partial η2 = .411.

UTIME #7 score was significantly lower at endpoint among N and moderately E types (M =

114

.441, SE = .025) compared to M types (M = .561, SE = .023), suggesting N and moderately E

types reached their peak time for most needing space to themselves later in the day than M types.

115

Table 5.6 Mixed ANOVAs on caregiver sleep quality and UTIME: main effects and interactions Chronotype

(Between Subject) Assessment phase

Baseline, Midpoint, Endpoint (Within Subject)

Assessment phase x Chronotype (Within Subject)

F df p η2 F df p η2 F df p η2 PSQI 0.140 1, 21 .712 .007 0.004 2, 42 .996 .000 0.406 2, 42 .669 .019 UTIME #1 0.778 1, 21 .388 .036 0.082 2, 42 .922 .004 2.284 2, 42 .114 .098 UTIME #2 0.298 1, 21 .591 .014 1.585 2, 42 .217 .070 0.732 2, 42 .487 .034 UTIME #3 1.240 1, 21 .278 .056 0.942 2, 42 .398 .043 1.139 2, 42 .330 .051 UTIME #4 0.100 1, 21 .754 .005 2.097 2, 42 .135 .091 2.861 2, 42 .068 .120 UTIME #5 1.309 1, 21 .265 .059 0.389 2, 42 .680 .018 1.289 2, 42 .286 .058 UTIME #6 0.090 1, 21 .767 .004 1.945 2, 42 .156 .156 1.335 2, 42 .274 .274 UTIME #7 3.329 1, 21 .083 .143 0.893 2, 42 .417 .043 5.147 2, 42 .010* .205

* p < .05 NOTE: PSQI: Levene’s test violated at baseline (p = .025) UTIME #3: Shapiro Wilk’s violated for M types at midpoint (p = .036); Levene’s test violated at midpoint (p = .006) UTIME #4: Shapiro-Wilk’s violated for E and N types at midpoint (p = .013); Levene’s test violated at midpoint (p = .011); Box’s test violated (p = .018)

116

5.4.4! Analysis 3: Matched Patient and Caregiver Population Analysis

5.4.4.1! Global Associations and Differences Among Matched Patient and Caregiver MEQ, PSQI and UTIME data

The Mann Whitney U (non-parametric test) was used to assess differences between matched

patients and caregivers on their MEQ, PSQI, and UTIME scores (Table 5.7). A non-parametric

test was used due to the small sample size, and the non-normal distribution of data. The only

significant results were the differences between patients and caregivers on all three PSQI scores.

Distributions of the patient and caregiver scores for PSQI at baseline, midpoint, and endpoint

were not similar, as assessed by visual inspection. Sleep quality was statistically significantly

worse among patients at baseline, midpoint, and endpoint (Mdn = 10.0, 9.0, and 10.0) than

among caregivers (Mdn = 8.0, 8.0, and 9.0) (baseline: U = 158.500, z = -2.339, p = .019;

midpoint: U = 172.500, z = -2.032, p = .042; endpoint: U = 136.000, z = -2.837, p = .005). In

these cases, the null hypothesis (H0 = the distribution of scores for the two groups are equal) is

rejected, and the alternate hypothesis (HA = the distribution of scores for the two groups are not

equal), is accepted. For age, chronotype, and UTIME questions assessing memory for their

personal and their partner’s coping abilities, patients and caregivers did not differ significantly.

117

Table 5.7 Median and Mann Whitney U significance values comparing MEQ, PSQI, and UTIME response scores between patients and their caregivers Medians Patient Caregiver U z p Age 52.00 52.00 292.000 0.605 .545 MEQ 61.00 59.00 214.500 -1.100 .271 PSQI

Baseline 10.00 6.00 158.500 -2.339 .019* Midpoint 9.00 7.00 172.500 -2.032 .042* Endpoint 10.00 7.00 136.000 -2.837 .005**

Baseline UTIME #1 .5857 .5329 205.000 -1.307 .191 UTIME #2 .5545 .5659 278.500 0.308 .758 UTIME #3 .5417 .5000 213.000 -0.909 .364 UTIME #4 .5865 .5710 229.000 -0.780 .435 UTIME #5 .4308 .4516 230.500 -0.747 .455 UTIME #6 .5392 .5000 196.000 -1.295 .195 UTIME #7 .5256 .5000 256.500 0.080 .937

Midpoint UTIME #1 .5139 .5256 271.000 0.143 .886 UTIME #2 .5000 .5192 284.000 0.429 .668 UTIME #3 .4900 .5000 302.500 0.836 .403 UTIME #4 .5545 .5929 297.000 0.714 .475 UTIME #5 .4304 .4870 292.000 0.604 .546 UTIME #6 .4910 .5256 278.500 0.308 .758 UTIME #7 .5000 .5000 231.500 -0.726 .468

Endpoint UTIME #1 .5705 .5256 231.500 -0.725 .468 UTIME #2 .5256 .5288 258.000 -0.143 .886 UTIME #3 .5000 .5256 320.000 1.221 .222 UTIME #4 .5604 .6046 351.000 1.901 .057 UTIME #5 .4981 .4402 190.000 -1.638 .101 UTIME #6 .5252 .5321 286.500 0.484 .629 UTIME #7 .5481 .5000 271.000 -1.044 .297

* p < .05 ** p < .01

118

5.4.4.2! Matched Patient and Caregiver Group Mixed Measures ANOVAs

Three-way mixed ANOVAs were run to understand the effects of one’s role (patient or

caregiver), chronotype, and assessment phase across treatment (baseline, midpoint, endpoint) on

sleep quality and UTIME performance scores. The within subject factor was time (score at

baseline, midpoint, and endpoint). The between subject factors were chronotype, (divided into

two categories: moderately E and N type, versus moderately and definitely M type, due to the

small group size), and role (patient versus caregiver). ANOVAs are robust to deviations from

normality, and therefore a good choice when analyzing irregularly distributed data.

Data were assessed for outliers based on inspection of boxplot values greater than 1.5 box

lengths from the edge of the box. Outliers (for sleep quality and UTIME #1-#7) were corrected

for by winsorizing the data in order to use a standard procedure for each analysis. The Shapiro-

Wilk’s test (p > .05) was used to assess normality of distribution for sleep quality or UTIME

score at each between and within subject level. Homogeneity of variance (p > .05) was assessed

by Levene’s test of homogeneity of variance; any violations were left untransformed and noted.

Mauchly’s test of sphericity was used to assess whether the assumption of sphericity was met (p

> .05); the Greenhouse-Geisser estimate was used when the assumption was violated. Mixed

ANOVA outputs for sleep quality and UTIME are presented in Table 5.8, including F-values,

degrees of freedom, levels of significance, and partial eta squared (η2).

When assessing changes in sleep quality at each assessment phase across treatment between

chronotypes and roles, the three-way interaction between treatment time, role, and chronotype

was not statistically significant, p = .848. All two-way interactions were not statistically

significant (p > .05).

When assessing changes in UTIME scores at each assessment phase across treatment between

chronotypes and roles, the three-way interaction between assessment phase, role, and chronotype

was only statistically significant for UTIME #4, p = .027, and UTIME #7, p = .002. This

indicates that the interactions of role and chronotype are different at the individual assessment

phases for UTIME #4 and #7 (recalled alertness, and recalled need for time and space to oneself,

respectively).

119

For UTIME #4, significance of a simple two-way interaction was accepted at a Bonferroni-

adjusted alpha level of .017 (i.e., 0.05 ÷ 3 = 0.017, divide by 3 for 3 assessment phases). There

was a significant interaction of role and chronotype at midpoint, F(1, 42) = 21.005, p < .001, but

not at baseline, F(1, 42) = 2.817, p = .101, or endpoint, F(1, 42) = 0.258, p = .614. Significance

of a simple main effect was accepted at a Bonferroni-adjusted alpha level of .025 (i.e., 0.05 ÷ 2 =

0.025, divide by 2 for 2 chronotype groups). There was a statistically significant simple simple

main effect for moderately E and N type chronotypes at midpoint, F(1, 42) = 24.697, p < .001,

but not for moderately and definitely M type, F(1, 42) = 1.215, p = .277. Pairwise comparisons

were performed for significant simple simple main effects. Bonferroni corrections were made

with comparisons within each simple simple main effect considered as a family of comparisons.

Adjusted p-values are reported. Mean UTIME #4 score was higher for moderately E and N type

caregivers (M = 0.601, SD = 0.018) than patients (M = 0.458, SD = 0.023), a mean difference of

.143 [95% CI, 0.085 to 0.201], p < .001. There was also a statistically significant simple simple

main effect for patients at midpoint, F(1, 42) = 19.785, p < .001, but not for caregivers, F(1, 42)

= 3.747, p = .060. All pairwise comparisons were performed for statistically significant simple

simple main effects. Bonferroni corrections were made with comparisons within each simple

simple main effect considered as a family of comparisons. Adjusted p-values are reported. Mean

UTIME #4 scores were higher for patients with moderately and definitely M type chronotype (M

= 0.578, SD = 0.015) than those with moderately E and N type chronotypes (M = 0.458, SD =

0.023), a mean difference of .120 [95% CI, 0.066 to 0.175], p < .001.

For UTIME #7, significance of a simple two-way interaction was accepted at a Bonferroni-

adjusted alpha level of .017 (due to there being three simple two-way interactions, i.e., .05 ÷ 3 =

.017). At p < .017, there was a significant simple two-way interaction of role and chronotype at

endpoint, , F(1, 42) = 6.762, p = .013, but not at baseline, F(1, 41) = 0.366, p = .549, or

midpoint, F(1, 42) = 5.692, p = .022. Significance of a simple simple main effect was accepted at

a Bonferroni-adjusted alpha level of .025 (due to there being two simple simple main effects for

chronotype, i.e., .05 ÷ 2 = .025). There was a significant main effect of chronotype for caregivers

at endpoint, F(1, 42) = 11.414, p = .002. All pairwise comparisons were performed for

statistically significant simple simple main effects. Bonferroni corrections were made with

comparisons within each simple simple main effect considered a family of comparisons.

Adjusted p-values were reported. Among caregivers, UTIME #7 score was higher in moderately

120

and definitely M types (M = 0.561, SD = 0.026) than in moderately E and N types (M = 0.435,

SD = 0.027), a mean difference of 0.126, 95% CI [0.051 to 0.201], p = .002.

At the level of two-way interactions, certain UTIME assessments demonstrated interactions

between chronotype and role, or between assessment phase and role; no significant interactions

were found between assessment phase and chronotype. For UTIME #2, a significant interaction

was found between role and chronotype, p = .034. Significance of a simple main effect was

accepted at a Bonferroni-adjusted alpha level of .025 (due to there being two simple main effects,

i.e., .05/2 = .025). There was a significant main effect of chronotype for patients, F(1, 42) =

11.806, p = .001, η2 = .219, but not for caregivers, F(1, 42) = 0.249, p = .621, η2 =.006. All

pairwise comparisons were performed for significant simple main effects. Bonferroni corrections

were made with comparisons within each simple main effect considered a family of

comparisons. Adjusted p-values are reported. Mean UTIME #2 score was higher in moderately

and definitely M type patients than moderately E and N type patients, a mean difference of .088,

[95% CI, -0.059 to 0.036], p = .001.

It appeared that for UTIME #5 there was a two-way interaction between role and chronotype, p =

.038. However, statistical significance of a simple main effect was accepted at a Bonferroni-

adjusted alpha level of .025; therefore, with this correction there was determined to be no

significant two-way interactions, p > .025.

For UTIME #6, a significant two-way interaction exists between role and chronotype, p = .034.

Significance of a simple main effect was accepted at a Bonferroni-adjusted alpha level of .025.

There was a significant simple main effect of chronotype for patients, F(1, 41) = 9.499, p = .004,

but not for caregivers F(1, 41) = 0.043, p = .837. All pairwise comparisons were performed for

significant simple main effects. Bonferroni corrections were made with comparisons within each

simple main effect considered a family of comparisons. Adjusted p-values are reported. Mean

UTIME #6 score was higher in moderately and definitely M type patients than in moderately E

and N type patients, a mean difference of .081 [95% CI, 0.028 to 0.135].

For UTIME #3, there was a significant two-way interaction between assessment phase and role,

p = .014. Significance of a simple main effect was accepted at a Bonferroni-adjusted alpha level

of .017 (due to there being three simple main effects, i.e., .05/3 = .017). There was a statistically

significant simple main effect of participant role at endpoint, F(1, 41) = 6.275, p = .016, η2 =

121

.133, but not at baseline, F(1, 41) = 0.569, p = .455, η2 = .014, or midpoint, F(1, 41) = 4.114, p =

.049, η2 = .09. All pairwise comparisons were performed for significant simple main effects.

Bonferroni corrections were made with comparisons within each simple main effect considered

as a family of comparisons. Adjusted p-values are reported. Mean UTIME #3 scores were higher

in caregivers, a mean difference of .055, [95% CI, .011 to .099], p = .016.

122

Table 5.8 Mixed ANOVAs on patient AND caregiver sleep quality and UTIME: main effects and interactions

Role x Chronotype (Between Subjects)

Assessment Phase (Within Subjects)

Assessment Phase x Role (Within Subjects)

Assessment Phase x Chronotype (Within Subjects)

Assessment Phase x Role x Chronotype

(Within Subjects) F df p η2 F df p η2 F df p η2 F df p η2 F df p η2 PSQI 0.620 1, 42 .436 .015 1.069 2, 84 .348 .025 1.146 2, 84 .323 .027 0.341 2, 84 .712 .008 0.165 2, 84 .848 .004 UTIME #1 3.115 1, 42 .085 .069 2.302 1.575, 84 .119 .052 1.568 1.575, 84 .219 .036 2.337 1.575, 84 .116 .053 0.384 1.575, 84 .633 .009 UTIME #2 4.793 1, 42 .034* .102 3.349 2, 84 .040 .074 0.189 2, 84 .828 .004 0.371 2, 84 .691 .009 0.461 2, 84 .632 .011 UTIME #3 0.341 1, 41 .563 .008 2.038 2, 82 .137 .047 4.502 2, 82 .014* .099 0.026 2, 82 .974 .001 1.942 2, 82 .150 .045 UTIME #4 6.758 1, 42 .013 .139 0.622 1.683, 84 .513 .015 4.300 1.683, 84 .023 .093 0.593 1.683, 84 .527 .014 4.088 1.683, 84 .027* .089 UTIME #5 4.610 1, 42 .038* .099 0.615 2, 84 .543 .014 2.125 2, 84 .126 .048 0.137 2, 84 .872 .003 1.617 2, 84 .205 .037 UTIME #6 4.784 1, 41 .034* .104 2.078 2, 82 .132 .048 1.280 2, 82 .283 .030 0.023 2, 82 .978 .001 2.192 2, 82 .118 .051 UTIME #7 0.308 1, 41 .582 .007 1.891 1.642, 82 .166 .044 0.220 1.642, 82 .759 .005 0.922 1.642, 82 .386 .022 7.647 1.642, 82 .002** .157

* p < .05 ** p < .01 NOTES: PSQI: Levene’s test violated at baseline (p = .007) UTIME #1: Levene’s test violated at baseline (p = .042); Mauchly’s test of sphericity violated, χ2 (2) = 12.916, p = .002 UTIME #3: Shapiro-Wilk’s violated for E and N type patients at midpoint (p = .016), and M type caregivers at midpoint (p = .036); Levene’s test violated at midpoint (p = .002) UTIME #4: Shapiro-Wilk’s violated for E and N type caregivers at midpoint (p = .013); Levene’s test violated at midpoint (p = .004); Mauchly’s test of sphericity violated, χ2 (2) = 8.547, p = .014 UTIME #5: Levene’s test violated at baseline (p = .006) UTIME #7: Mauchly’s test of sphericity violated, χ2 (2) = 9.838, p = .007

123

5.5! Discussion

Influences of chronotype and sleep quality on coping among cancer patients have been described

previously in only a few separate reports (e.g., Hansen & Lassen, 2012; Ancoli-Israel et al.,

2014). In this study, we sought to build on this background, recognizing that chronotype and

sleep quality are strongly associated with cognitive function and emotional regulation in the

general population (e.g., Clark, Watson & Leeka, 1989; Killgore, 2010; Lara, Madrid & Correa,

2014; Lim & Dinges, 2008). These associations could provide insight into how coping might be

affected by chronobiology either directly or indirectly through temporal regulation of executive

functions. To investigate this, we performed a series of mixed measures ANOVA analyses to

assess correlations among chronotype, sleep quality, and executive functions on the ability to

cope with the disease. Assessments were made at three points in the treatment, as it was possible

that coping strategies might be adjusted as patients dealt with the experience. In addition, the

involvement of caregivers was considered as a potentially strong environmental influence on

coping strategy.

5.5.1! On the Independent Impacts of Sleep and Chronotype

Chronotype has been defined alternatively as either a time of day preference for physical and

mental performance (e.g., Horne & Östberg, 1976) or the timing of sleep (Merrow et al., 2003).

The two metrics, preference and mid-sleep time, are correlated, and both are correlated with

sleep quality in previous work (Keller, Grünewald, Vetter, Roenneberg & Schulte!Körne, 2017;

Rique, Fernandes Filho, Ferreira & de Sousa-Muñoz, 2014). Although our findings did not

produce a significant correlation between chronotype and sleep quality, the results indicated a

trend in the group of 86 patients which became progressively more negative, moving towards

significance over the course of treatment. This may indicate that sleep quality at the outset of

chemotherapy was not associated to chronotype, but that as treatment progressed, patients began

shifting towards following sleep patterns and habits that reflected their own chronotype. While

sleep is poor across treatment in this group, the trend towards a correlation between MEQ and

PSQI may indicate that at the outset of treatment, sleep is poor due to one’s stress stemming

from their diagnosis. As treatment progresses, while sleep quality does not ameliorate, one’s

sleep patterns may be less disorganized due to stress and irregular sleeping and people may be

following sleep schedules that reflect their chronotype.

124

While chronotype does not predict sleep quality as strongly in this total group as it does in the

general population, the negative correlation indicates that better sleep quality was associated

with increasing tendency towards morningness (lower PSQI scores linked to higher MEQ),

which reflects the relationship in the general human population (Soehner, Kennedy & Monk,

2011; Vardar, Vardar, Molla, Kaynak & Ersoz, 2008). Sleep quality among all patients was poor

across treatment, both by PSQI standards (score > 5) and by research standards that have

suggested a PSQI score > 8 among cancer patients indicates poor sleep quality (Carpenter &

Andrykowski, 1998; Vargas, Wohlgemuth, Antoni, Lechner, Holley & Carver, 2010). Several

patients in our study reported that their sleep was of poor quality, and was worsening. Therefore,

the findings suggest there is a common factor shared among patients that affects sleep.

Interestingly, the relationship between chronotype and sleep quality differed among the 23

patients who had caregivers involved in the study and the 63 patients who were either single

(without a spousal caregiver), or had a partner who was not involved in this study. In the former

group, sleep quality was not significantly correlated to chronotype at any assessment phase

across treatment, and the data showed a positive association. Also, the correlations became

increasingly less significant as treatment progressed. For the latter group, there was a significant

negative correlation between chronotype and sleep quality, and the strength of the correlation

increased across treatment. This correlation strongly indicated an association between good sleep

quality and an increased tendency for morningness.

Overall, the change in correlation exhibited by the group without a caregiver involved indicates

that the initial dissociation between chronotype and sleep returns to reflect the correlation found

in the general population over the course of treatment; whereas in patients with a caregiver

involved, sleep quality further separates from chronotype. It appears that an additional factor

affects the typical association between sleep quality and chronotype when a spousal caregiver is

closely involved in the treatment. Future studies should consider whether there are changes in the

types of factors contributing to poor sleep quality (e.g., high worry and anxiety at the outset

which decreases and is maybe replaced by steroid drug side effects such as insomnia with high

alertness as treatment progresses), and what factors may produce a difference in the relationship

between chronotype and sleep quality between patients with and without caregivers closely

involved in their treatment and care.

125

Among the 23 caregivers, MEQ and PSQI showed a positive but non-significant association,

similar to the association in the 23 matched patients. In this group, the positive association

suggests poorer sleep quality (increasing PSQI score) is related to increasing MEQ score

(tendency toward morningness). Like the 23 matched patients, these results do not reflect the

relationship observed in the general population. While it is possible that this caregiver group may

have been too small to yield an accurate statistical result, it is possible that other factors are

affecting caregiver sleep quality independent of chronotype, subsequently masking the effects of

one’s circadian clock. Future studies should examine whether a shared factor negatively

influences sleep and disrupts the sleep-chronotype relationship between patients and their

caregivers closely involved in the treatment and care process. Research should also examine

whether poor patient or caregiver sleep negatively influences sleep quality of the alternate group.

While sleep quality scores across treatment among the total patient and caregiver groups were

poor, the numbers did show slight fluctuations between assessment phases. However, among the

total group of patients and the caregivers, there were no within or between (i.e., chronotype

dependent) subject changes in sleep quality across treatment. This suggests that regardless of

chronotype, patients and caregivers alike experienced poor sleep, and sleep scores were not

significantly different across assessment phases. Future research must consider whether a

particular set of factors consistently influences poor sleep across treatment, and or whether

various chronotypes are more prone to the effects of certain disruptive influences.

While patients and caregivers both have poor sleep quality, all patient PSQI scores (including

those with and without a caregiver) are significantly worse than those of caregivers. This trend

exists from treatment outset, even before chemotherapy drugs have been administered, indicating

that this may not necessarily be a pharmacologically induced difference at all times. Among

caregivers, the poor sleep is not likely pharmacologically induced, unless the caregiver is taking

other unknown medications (drug history was not collected in this study). Between matched

patients and caregivers, patients demonstrate significantly worse sleep across treatment, however

there is no significant chronotype difference. One possible explanation for the poor sleep among

patient caregiver duos may be that patients and caregivers are both cognitively depleted at

similar times as a result of having similar chronotypes, and therefore may ruminate together in

the evening at their off times, subsequently increasing one another’s stress levels. It is important

to assess in future how patients and caregivers affect one another’s sleep quality.

126

5.5.2! Patient Cognitive Function and Emotional Regulation

5.5.2.1! Chronotype-UTIME Correlations

Cognitive based tasks show performance fluctuations across the day. Research has indicated that

M types often show performance declines across the day on executive function and

attention/vigilance tasks, while E types tend to show a later performance peak time in the day on

the same tasks (for review, see Schmidt et al., 2015). One key aspect of this study sought to

assess subjective recall of one’s emotional regulation (an output of cognitive control) in the form

of coping, in relation to one’s chronotype, among cancer patients and their caregivers.

Normalized UTIME scores above .5000 indicate peak performance in the first half of the day,

with increasing scores above .5000 indicating a peak closer to one’s wakeup time, while scores

below .5000 reflect peak performance in the second half of the day. Among patients, all UTIME

scores, except UTIME #5, were significantly positively correlated with chronotype, indicating

that higher MEQ scores (i.e., greater tendency for M) are significantly associated with higher

UTIME scores (i.e., peak performance is earlier in the day). These data support similar findings

in previous studies that chronotype and one’s perception of peak performance time is correlated.

In other words, a positive correlation indicates that higher MEQ scores are associated with

earlier perceived optimal performance times (higher UTIME scores). Among patients, M types

appear to show earlier peaks on the 7-question battery than E types, indicating that participants

typically report reaching their peak emotional regulatory abilities at times reflective of their

chronotype.

Mean UTIME #7 (most need time and space to yourself) scores for baseline and midpoint were

both below .5000, and just slightly above .5000 at endpoint. The baseline and midpoint averages

suggest that while MEQ was correlated with UTIME #7, patients still typically required personal

time or space to themselves later in the day compared to their peak times on other activities. The

positive correlation here indicates that M types, while still showing a peak need for time/space to

themselves in the second half of the day, still require this time earlier than their N and E type

counterparts. Requiring time and space to oneself later in the day may be reflective of the

documented afternoon energy decrease, often known as the “post lunch dip” (Monk, 2005).

UTIME #5, which assessed changes in fatigue across the day, did not show a significant

relationship with MEQ. Research shows that fatigue is one of the most commonly reported

127

symptoms of cancer, and is prevalent during both treatment and recovery. Specifically,

prevalence rates of fatigue have been reported among breast cancer patients at 4% prior to

chemotherapy, and spiking to 91% following surgery and chemotherapy (Carr et al., 2002). The

lack of correlation, and the shift from positive and negative correlation between MEQ and

UTIME #5 scores indicates patients’ chronotype were not associated with their recalled sense of

fatigue across the day. It is important to note that this is the UTIME question patients most

struggled to answer, with many people expressing confusion (note: no quantifiable data available

on how many people had problems with this question). It seems that many patients had difficulty

recalling their fatigue levels across the day. It has been reported that fatigue is difficult to

quantify or interpret in research, as there is no consensus on a standard definition, and varying

criteria are used to define its presence or severity (Institute of Medicine, 2008). Given that

fatigue is a commonly reported symptom, and has no standard against which it can be measured,

patients may have difficulty quantifying shifts in their level of fatigue, particularly if an element

of fatigue persists across the day. This may result in a range of different assessments about

personal fatigue, therefore removing the chance of correlation between MEQ and UTIME #5.

While no correlation exists between MEQ and UTIME #5 scores, mean UTIME #5 scores were

always below .5000, indicating that the average peak fatigue time was in the second half of the

day. This shows a consistent pattern that on average patients recalled to some degree that they

were more fatigued later in the day.

5.5.2.2! Multivariate Analysis of UTIME Correlations in the Patient Population

Mixed ANOVAs were conducted to assess within and between subject changes in cognitive and

emotional regulation across the treatment. Each UTIME item was chosen to address a different

aspect of executive function as it might pertain to the cancer patient’s situation.

UTIME #1: Emotional Regulation, “Coping” – changes in one’s ability to cope with the stress

associated with their cancer diagnosis across the day. Between subjects, chronotype was

significantly related to changes in coping ability across the day, evidenced by definitely and

moderately M types recalling reaching peak coping ability earlier than E types. Moderately M

types also reached peak coping ability significantly earlier than N types. While a significant

difference in coping may not have existed between definitely versus moderately M types, or

between definitely M versus N types, these preliminary data support a chronotype related

128

difference in emotional regulatory abilities. This supports the idea that people have a memory for

their optimal times to face their stressor. Within subjects (i.e., regardless of chronotype), across

treatment, patients collectively showed an earlier peak coping time at endpoint over midpoint,

suggesting people reached their optimal coping ability for the stresses associated with having

cancer earlier in the day. While it is possible that several factors contribute to this shift to earlier

peak emotional regulatory abilities, it seems likely that increasing exhaustion levels as treatment

progressed may impact this trend. There are likely to be multiple factors that contribute to this,

considering especially the novelty of a situation that is outside most peoples’ experience. It

seems reasonable to expect that the initial diagnosis and entry into treatment would be sufficient

to override (mask) chronotypic variation, or to disrupt underlying circadian regulation entirely.

UTIME #2: Prosocial Behaviour – changes in one’s ability to help their partner cope with the

knowledge of the patient’s illness. Prosocial behaviours refer to those actions that benefit or

increase the welfare of others, sometimes at a cost to one’s self (Dunfield, 2014; Gesiarz &

Crockett, 2015), and are believed to be an output of cognitive function, specifically perspective

taking, which in this case is the ability to take the viewpoint of someone in distress (Underwood

& Moore, 1982). This question was assessed among all patients who had a caregiver, including

those with a caregiver partaking in the study, and those whose partner was their main caregiver

but was not involved in the study. N types showed UTIME scores significantly lower than

definitely and moderately M types, indicating that while the mean UTIME #2 score was always

above .5000 and suggestive of best performance in the first half of the day, later chronotypes

(lower MEQ score) showed delayed peak performance times (lower UTIME score) compared to

their earlier chronotype counterparts. This supports the idea that cognitive function is highly

positively correlated to one’s chronotype (Schmidt, Collette, Cajochen & Peigneux, 2007).

UTIME #3: Attitudes on Caregiver Helpfulness – attitude toward caregivers regarding how well

they could help a patient cope. This question was assessed among all patients with a caregiver,

including those with a caregiver partaking in the study, and those whose partner was their main

caregiver but was not involved in the study. No between subject (chronotype) difference was

seen, indicating that patients did not feel they were given assistance by their partner at times

necessarily corresponding to their best or worst coping times in the day. This demonstrates that

one’s partner is not necessarily able to provide assistance depending on the patient’s

chronotypically preferred time of day, and therefore may be providing assistance at their own

129

personal best time of day. A within subject difference (regardless of chronotype) showed that on

average, patients believed their partners were best capable to help them somewhat earlier in the

day at endpoint compared to baseline. This may indicate that overall, caregivers themselves may

have had a tendency towards better ability to cope and provide assistance to others earlier in the

day, suggesting a preference towards morningness.

UTIME #4: Alertness – changes in patients’ degree of alertness across the day. Alertness is

necessary to engage in complex cognitive processing, and carry out cognitive tasks (Killgore,

2010). This includes emotional regulation and coping. As one’s propensity for morning

increased, their corresponding peak period in the day for alertness was also earlier. All

differences between definitely and moderately M types, N types and E types were sequentially

ordered, such that definitely M types had the earliest peak, and E types the latest peak. These

results demonstrate that among individuals faced with a chronic stressor, alertness – an output of

cognitive function – fluctuates simultaneously with chronotype, supporting the overall link

between one’s circadian rhythm and changes in their cognitive abilities. While this question may

have had an increased chance of Type 1 error (a “false positive”), the results relating to UTIME

#4 seem plausible given that they are in line with the chronotype ordering trends seen in other

UTIME questions.

UTIME #5: Fatigue – daily changes in fatigue according to chronotype. E types showed an

earlier peak than all three other groups. Yet the difference in scores between M and N types was

not significant; only E types showed a significantly earlier peak fatigue time compared to N

types. This may be due to either skewed data resulting from a small E type group, or it is

possible that E types are more fatigued earlier before their energy levels increase for the day. No

within subject differences were found for changes in fatigue across treatment. Furthermore, lack

of a clear definition or way to measure fatigue will likely contribute to poor memory for one’s

changes in fatigue across the day. Lack of a clear way to quantify fatigue for one’s own self

likely makes it more difficult to store clear memories for changes in one’s fatigue. In future, a

clearer definition of fatigue along with a more defined benchmark against which patients can

assess fatigue is required to help them quantify their subjective ratings of fatigue.

UTIME #6: Compassion – patients’ level of compassion for their partner as a caregiver.

Compassion is a prosocial behaviour that has been conceptualized in several ways, and can be

130

summarized as the vicarious experience resulting from witnessing the distress of another’s

suffering, that spurs the subsequent desire to help (for review, see Goetz, Keltner & Simon-

Thomas, 2010). Compassion is associated with perspective taking (Gleichgerrcht & Decety,

2013), and therefore requires cognitive reasoning ability. CS and CF were found to be associated

to chronotype among various oncology staff providing patient care in the hospital setting (see

Chapters 3 and 4). This question was assessed among all patients who had a spousal caregiver,

whether he was or was not involved in the study. Similar to UTIME #2 which also assessed

prosocial behaviour, definitely and moderately M types had earlier peak compassion

performance times in the day compared to N types. While E types did not display a significantly

later peak time compared to M or N types, they did exhibit a delayed peak. These data indicate

that when faced with a chronic stressor themselves, individuals still recall showing compassion

to those who work to help them, and report that their compassion is greatest at times

corresponding to their chronotype. No within subject differences were found across treatment,

suggesting that patients remained fairly consistent in the times at which they recalled feeling

compassion towards their caregivers. A significant ordered difference between moderately and

definitely M types, N types, and E types for compassion abilities should be examined in future

studies among larger populations.

UTIME #7: “Private Time” – patients’ need for time and space to themselves. The results

showed a clear statistically significant between subject difference, where earlier chronotypes

recalled an earlier peak time for requiring personal time and space compared to later

chronotypes. Moderately M types reported a significant earlier peak compared to N types across

treatment. Mean normalized UTIME #7 scores were indicative of needing time and space in the

second half of the day at baseline and midpoint, and just slightly before the second half of the

day by endpoint. This suggests that time and space to one’s self is needed later in the day

sometime in the afternoon, possibly when an individual is less cognitively alert. This may

indicate that one’s resources are slightly depleted at this time and therefore an individual requires

some personal space and reduced stimulation. Nonetheless, those with a later chronotype showed

a delay in the time of day they reported personal time and space was required, suggesting that in

line with their later peak performance compared with earlier chronotypes, these individuals also

reach their energy depletion lows later in the day than their counterparts. The somewhat earlier

peak at endpoint may suggest patients reach a point where personal time is required earlier,

131

possibly as a result of frustration with a long treatment process, and or general energy depletion

and cachexia (general wasting syndrome including appetite, energy, muscle and weight loss) that

is often reported by cancer patients (e.g., Aoyagi, Terracina, Raza, Matsubara, Takabe, 2015).

The within subject trend that was visible in some questions looking at changes in cognitive

function peaking earlier in the day by the end of treatment indicates people’s optimal

performance times were shifting. Future research is required to understand what may be causing

this phenomenon. While chronotype was likely a relatively stable and unchanging trait, it is

possible that people’s degree of fatigue was influencing their need to deal with various tasks and

demands earlier in the day before they were too depleted sometime later in the day. Given that

fatigue has been reported in up to 91% of patients undergoing chemotherapy treatment (Carr et

al., 2002), it is possible that patients noticed this change in their ability and adjusted their

performance times accordingly. Future studies with larger patient groups may be able to more

clearly assess this difference.

5.5.3! Caregiver Cognitive Function and Emotional Regulation

Among the 23 caregivers, there were only two statistically significant correlations between MEQ

and UTIME score between baseline, midpoint and endpoint, and none that were consistently

significant. However, as seen with the total group of patients, all associations between MEQ and

UTIME were consistently positive (except for UTIME #5, which again measured fatigue). This

positive association may suggest that a greater tendency for M would result in an earlier peak

performance time across the day, and vice versa for E types, however, due to lack of

significance, no conclusions can be drawn. The change in association direction between MEQ

and UTIME #5 from positive to negative may again be due to the lack of definition and

quantifiable ways to assess fatigue. As seen in the total patient population, caregivers also show

mean UTIME #5 scores below .5000, indicating that they recall their highest fatigue time is in

the second half of the day. It is interesting to note the highly significant positive correlation

between MEQ and UTIME #7 (“Private Time”). The mean normalized UTIME #7 score at

endpoint (.5008) suggests an average of most needing space to one’s self in the midday. This

significant correlation and ANOVA results indicate that M type caregivers recall needing space

earlier than N and E type caregivers. One possible reason this chronotype difference was not

found at baseline or midpoint, aside from small sample size, is that caregivers may have been

132

hesitant earlier in treatment to take time for themselves when they felt it was necessary if they

were already performing other duties related to helping their partner. Given that patients noted

differences in their caregivers’ abilities to provide care, it is possible that caregivers themselves

may have attempted to provide consistent care across the day but were unable to notice changes

in their performance. Alternately, as suggested in Bellicoso (2010), it might be counter to the

heuristic they maintain about their caregiving abilities being consistent across the day to report

fluctuations in their performance. Future studies with a larger caregiver subject group should

further assess these questions to examine whether caregivers are actually not reaching peak and

trough performance times as dictated by their chronotype in order to continue performing tasks

related to helping their spouse, or whether a larger sample size will allow these chronotype based

differences to be observed.

5.5.4! Matched Patient and Caregiver Cognitive Function and Emotional Regulation

Among the 23 pairs of matched patients and caregivers, the two groups were fairly similarly

matched on MEQ and UTIME scores.

Three-way mixed ANOVAs were used to assess differences in UTIME score across treatment

between patients and caregivers split on two chronotype categories (definitely and moderately M

type versus N and E type chronotypes).

UTIME #1: Emotional Regulation, “Coping” – change in ability to cope with the stress

associated with one’s own or their partner’s cancer diagnosis across the day. No significant

within or between subject differences were found, suggesting that among matched patients and

caregivers partaking in the study, participants recalled coping fairly equally across treatment

regardless of their role or chronotype. It is also possible that no significant differences were

found due to the small 23 pair sample size. However, among the total group of patients, a

difference in coping based on chronotype was found to exist, therefore it is possible that the

group size was too small in this analysis to detect a significant difference. Larger studies with

more normal chronotype distributions should reexamine matched patients and caregivers to

assess whether there is in fact no difference in one’s memory for changes in their coping ability,

or whether clear distinctions exist between patient-caregiver and chronotype groups.

133

UTIME #2: Prosocial Behaviour – change in ability to help one’s partner cope with the stress of

their cancer diagnosis/role as a caregiver. As in the previous two groups, this question assessed

changes in prosocial behaviour as an output of shifts in cognitive function across the day. While

caregivers did not show a significant chronotype based difference in prosocial ability across the

day, M type patients showed a significantly earlier peak recalled performance time for prosocial

behaviour compared to N types (there were no E types in the 23 subject patient subset). This

suggests patients recalled aiding their partners at their own personal chronotypically optimal

times, while caregivers did not recall significant differences in their provision of care across the

day. These results among patients reflect those seen in the larger group. It is possible that due to

the small sample size, differences among caregivers’ recalled coping abilities are not uncovered,

however, given that a small sample size for patients yielded a chronotype based difference, it is

possible that caregivers are not forming a memory for differences in their prosocial abilities. As

caregivers, they may feel they are in a position to provide consistent assistance to their partner,

and recalling performance differences may be counter to the heuristic they hold of themselves.

UTIME #3: Attitudes on Helpfulness – assessed differences in perceived helpfulness among

patients or caregivers for when their partner could best help them cope with the stress of being a

patient or caregiver. A significant within subject effect between time and role indicated that at

endpoint, caregivers felt their partners best helped them cope earlier in the day compared to

when patients felt their caregivers assisted them. These data suggest that caregivers recalled

patients reaching an earlier peak performance ability for prosocial behaviours, subsequently

indicating an earlier peak in cognitive abilities. This supports the slightly higher mean

chronotype score found among the 23 patients compared to their 23 caregivers which would

likely lead to differences in peak cognitive performance times. Future studies among larger

groups should reexamine whether this difference exists across treatment.

UTIME #4: Alertness – changes in patients’ and caregivers’ degree of alertness across the day as

a benchmark for changes in their cognitive function. While a significant effect of treatment time

point was seen at midpoint among patients, studies with larger population sizes will be necessary

to further examine this association. Moderately and definitely M type patients reported earlier

alertness peaks compared to N type patients. Larger sample sizes can assess whether this

association exists at isolated time points across treatment, or whether this is normal across

treatment. Nonetheless, this is reflective of the same data seen in the larger total group of patients

134

where chronotype was linked with one’s peak alertness time across the day. Interesting to note is

that among N and E type chronotypes, caregivers recalled a significantly earlier peak in alertness

compared to patients. Whereas caregivers reported peaking on average late in the first half of the

day, patients recalled peaking sometime in the earlier portion of the second half of the day. This

is particularly interesting given that caregivers had a slightly later mean chronotype score

compared to their matched patients. This suggests female patients with a caregiver feel they take

slightly longer to reach peak alertness than their male caregivers who recall reaching their peak

sooner. This needs to be further examined in larger population samples, but these results may

suggest women feel they take longer to reach heightened cognitive function than males. Overall,

the findings relating to MEQ versus UTIME #4 suggest patients and caregivers show differences

in their recalled levels of alertness across treatment, and that earlier chronotypes recall peaking

earlier in the day compared to later chronotypes, continuing to suggest an association between

one’s chronotype and fluctuations in their recalled cognitive ability.

UTIME #5: Fatigue – daily changes in fatigue according to chronotype. Among matched patients

and caregivers, it appeared that a difference existed in one’s memory for fluctuations in their

level of fatigue based on their role or chronotype. However, upon further consideration, there

were no significant differences in this area. The lack of significant difference between recalled

differences in fluctuations in one’s level of fatigue may result from lack of a clear definition of

fatigue, as previously noted. The small sample size may also contribute to not uncovering

significant findings. Finally, among this matched group specifically, it is possible that patients

and caregivers did not focus on their fluctuating fatigue levels across the day, as many may have

felt familial obligations to other members of their household and to one another, and as such,

may not have stopped to reflect on and form a clear memory of their own fatigue.

UTIME #6: Compassion – patients’ and caregivers’ levels of compassion for their partner and

the stresses faced she/he faced as a patient/caregiver. Caregivers did not report recalling a

chronotype dependent difference in their compassion levels towards their partner the patient. As

discussed previously, it is possible that caregivers are accurately recalling being consistent in

their degree of compassion across the day. Alternately, caregivers may not accurately reflect on

and or recall fluctuations in their levels of compassion, as this would have been counter to their

duty to care for and be compassionate towards their partner in their time of sickness. Yet among

patients with caregivers in the study, similar to the pattern seen among the total group of patients,

135

this subset of patients continued to be compassionate to their caregiver in a manner reflective of

their own chronotype; specifically, M types showed an earlier peak compassion time compared

to N type. While caregivers may believe it is their duty to be compassionate towards the patient

at all times, patients may feel that as the sick individual, it is acceptable to limit the degree of

prosocial behaviour they must express towards others, and therefore may be more open to

reflecting on and reporting performance differences in their behaviour. This chronotype based

difference among patients in recalled fluctuations in compassion supports the other observations

relating to chronotype based changes across the day for cognitive domains. These findings on

performance memory differences support the idea that one’s recall for their cognitive

performance ability is closely linked to chronotype, and that patients need to be open to

accepting differences in their performance in order to form a memory.

UTIME #7: “Private Time” – patients’ and caregivers’ need for time and space to themselves.

While no specific difference were found between M versus N type patients for recalled need for

personal time and space, the results found among just caregivers were repeated here. At

endpoint, M types caregivers again recalled an earlier peak time compared to their N and E type

counterparts. While this is a repeated finding, unlike in the total group of patients, this subset of

patients did not report a chronotype related peak timing difference in their memory for requiring

personal time and space. Unlike the other 63 patients, these 23 patients may have recalled a less

differentiated pattern for changes in the need for personal space across the day. It is possible that

these women whose partners were closely involved in the study were accustomed to having

people around and present more often, so they may have learned not to follow a chronotype

dependent schedule for taking personal time and space.

5.5.5! Limitations

At the outset of this study, it was believed that many caregivers would be available for

participation in the study. However, for several reasons (including but not limited to time

constraints, low interest, not being able to attend hospital visits with patients), it was not possible

to recruit the large number of caregivers we had initially hoped for. This reduced the population

size of caregivers, and the number of matched patient-caregiver dyads. Future studies should

include a larger number of patient caregiver dyads in order to better understand how close

simultaneous involvement by the patient and caregiver across chemotherapy affects each

136

individual. Only 14 patients involved in the study were single. Future research may wish to

examine a larger group of single women in comparison to women with a closely involved

caregiver to understand differences in their recall for performing optimally at one time or

another.

Future studies may wish to reassess the use of actigraphy in such a study. While an attempt to

use actigraphy was made in this study, no useable data were obtained. Actigraphy data would

provide an objective measure of patient activity across the day, which may help to better

understand sleep, and perform a more in depth analysis comparing memory recall among those

patients who are more or less active.

5.6! Conclusion

While a significant correlation between sleep quality and chronotype exists among patients who

either have no caregiver, or whose caregiver was not involved in this study, this relationship does

not exist among patients whose caregiver was involved. Nor does it exist among the participating

caregivers. Given the opposite (positive) direction of the correlation between MEQ and PSQI

scores for patients with a caregiver involved in the study as compared to the (negative) direction

of the correlation for patients without a caregiver, and the fact that involved caregivers show the

same positive direction as their spouses, it seems possible that this group of individuals is

influencing sleep quality among one another. Chronotype is a fairly stable trait, while sleep

quality can change, therefore it seems reasonable to suggest sleep quality is being affected

among these individuals to shift the correlation direction. While a comparison statistic was not

calculated, sleep quality scores were consistently worse among patients whose caregiver was

involved in the study, compared to those patients without a caregiver involved. Future studies

should examine whether patients and their caregivers who are closely involved in and go through

the treatment process with little separation from one another are actually causing further

disruption to one another’s sleep. If so, it would be beneficial to suggest taking some time and

space individually to enhance one another’s sleep and possible ability to cope.

Among the total group of patients, while significant clearly ordered differences between

definitely M, moderately M, N and E type chronotypes were not found for all UTIME questions,

137

the general trend showing that earlier chronotypes were consistently recalling reaching peak

performance times earlier than their later counterparts supports the idea that chronotype is

strongly tied to one’s cognitive abilities and their memories for their cognitively based tasks.

Caregivers did not show significant differences in their performance for the various cognitive

and emotional regulatory tasks that were assessed across the day. There are several possible

explanations for this. First, it is possible that the sample size was too small to reveal a significant

difference. Second, as patients and participating caregivers have a different gender

representation, responses based on subjective recall, may be different. For example, as

mentioned, it is possible that caregivers felt they sought to give their best effort to help their

spouse and to remain consistent in their behaviour across the day, and therefore may have

recalled being consistent if they did not stop to actually reflect on changes. Finally, it is possible

that many caregivers were not acting at their own chronotypically optimal time, and instead

aimed to give their best performance for various activities when they believed it was most

necessary by their caregiver.

Overall, the study demonstrated that among cancer patients, and to an extent among caregivers,

chronotype is related to one’s subjective recall of sleep quality and their memory for changes in

cognitive and emotional regulatory performance across the day. Similar patterns have been

reported previously (Bellicoso, 2010), however further examination is needed to understand why

patients with closely involved caregivers and the caregivers themselves show patterns opposite to

the general population. The subject of fatigue which remains poorly defined and without a

standard measure, while still somewhat associated to chronotype, does not have as clear a

relation as found in other tasks. For the emotional regulatory tasks assessed in this study –

personal coping, prosocial behaviour and compassion – patients may have had clearer standards

against which to measure changes in their behaviour compared to the emotional regulatory tasks

used in previous studies that showed little correlation with chronotype (see Bellicoso, 2010).

It is important that cancer patient chronotypes be assessed in order to help them schedule their

day in order to allow them to face trying situations at times when they will be best equipped to

deal with the situation. Caregivers should also be assessed for their chronotype, as a better

understanding of their optimal times in the day may allow them to structure their day accordingly

and take the necessary measures to also help themselves during this difficult time.

138

Chapter 6!

! Coping Behaviour in Chronic Disease

6.1! Abstract

Little is known about the influence of cancer patient and spousal caregiver chronotype and sleep

quality on their self-ratings of coping, or on their specific behaviours used to deal with cancer

related stressors. Given the influence both factors have on cognitive ability, and the lack of

importance society places on sleep, it is important to understand how these factors relate to

changes in coping across treatment. Patient self-rated coping responses, and patient and caregiver

coping behaviours were assessed for potential relationships with chronotype and sleep quality

across chemotherapy treatment. Personality and the difference between patients and caregivers

was also considered when assessing predictive influences on the use of coping behaviours.

Chronotype was related across treatment to in the moment self-rated coping scores, and at

baseline to the use of coping behaviours such as self-distraction, substance use, and into

treatment for the use of religion. Sleep quality was not linked to self-ratings of coping, but was

positively predictive of engagement style coping behaviours by endpoint. Patients and caregivers

typically reported decreased use of coping behaviours as treatment progressed, except for

acceptance which increased from baseline to endpoint. Openness and industriousness had the

most consistent predictive value for various coping behaviours across treatment. When rating

overall coping ability and use of behaviours early on, people rely on chronotype. Chronotype is

not predictive of coping behaviour use as treatment progresses. As in the typical population,

people do not always factor in the influence of sleep quality for changes in the behaviour across

the day.

139

6.2! Introduction

In 2016, breast cancer was predicted to be the most prevalent cancer diagnosed in females in

Canada, comprising 26% of newly diagnosed cancers in women (Canadian Cancer Statistics,

2017). In addition to the necessary medical treatment, many patients depend on the care and

support of family and or friends who act as non-vocational caregivers, a group who similar to

patients, also experience their own stresses related to providing cancer patient care. Historically,

much of the research among breast cancer patients has focused on the role of clinical variables

such as prevention, diagnosis and treatment. However, in recent years, the scope of cancer

related research has broadened to include understanding how patients and caregivers cope with

the stresses associated with having cancer or caring for a cancer patient. Among patients and

caregivers, individuals in both groups show within and between group differences in their coping

habits (Carter & Acton, 2006; Sharma, Chakrabarti & Grover, 2016).

Coping is a complex process, that refers to those “constantly changing cognitive and behavioral

efforts to manage specific external and internal demands that are appraised as taxing or

exceeding the resources of the person” (Lazarus & Folkman, 1984, p. 141). In addition to

appraisal of the situation, one’s assessment of the resources they have available to deal with the

situation will also influence their coping ability. Coping can be looked at in different ways,

including, but not limited to one’s own overall rating of how they feel they are coping, and by

looking at the specific behaviours one uses to deal with a stressor. Separate from their

assessment of the situation and the resources available to deal with the stressor, the number of

variables that might affect one’s ability to cope, are endless. Understanding how even a select

few variables relating specifically to an individual affect coping, including one’s chronotype,

sleep quality, and personality, can provide a better picture of how coping changes across the day,

and give indications for the coping strategies certain people might employ. This will allow for

recognizing signs for those individuals who may have greater difficulty coping at certain times,

or understanding who may be at risk of engaging in harmful coping behaviours.

People facing a major stressor are often asked “how are you doing?” or “how are you coping?”.

Yet, the majority of coping measures that exist do not ask respondents that question, but rather

assess one’s use of coping behaviours (e.g., Brief COPE), strategies (e.g., Cancer Coping

Questionnaire), or perceived ability to cope effectively with life challenges (e.g., Coping Self

140

Efficacy Scale), to name a few (Carver, 1997; Chesney, Neilands, Chambers, Taylor & Folkman,

2006; Moorey, Frampton & Greer, 2003). None of these questionnaires give an indication of

how one’s self-perceived ability to cope changes across the day, or when in the day a person

feels he or she copes best or worst. Furthermore, these questionnaires are retrospective; none of

the measures ask respondents to report how they believe they are coping at a specific moment in

time. Long term memories are influenced by both one’s memory of the event and the emotion

experienced during memory retrieval, therefore it is possible that one’s memory of their coping

may differ from how they actually felt at a specific time or on a specific day. While these

questionnaires provide valuable information, an understanding of how one is coping in the

moment at specific times of the day would give a clearer understanding of one’s ‘ups and downs’

during their waking hours.

Functioning of cognitive and emotional processes vary over the course of the 24 h day.

Appropriately timed wake and sleep cycles that reflect one’s internal biological clock facilitate

maximal cognitive and emotional performance, while wake sleep schedules that do not reflect

one’s biological clock can reduce an individual’s cognitive and or emotional regulatory abilities

(Wright, Lowry, & LeBourgeois, 2012). Chronotype has been linked with changes in

emotionality and mood, such that morning (M) type individuals typically show the quickest rise

in positive affect in the morning between 9 a.m. and noon followed by a dramatic decrease after

9 p.m., while neither (N) and evening (E) types do not demonstrate the same rapid rise in

positive affect in the morning as seen among M types (Clark, Watson & Leeka, 1989). As such,

it is likely that depending on a patient or caregiver’s chronotype, there will be earlier and later

peaks in optimal coping times across they day. A clearer understanding of how one’s chronotype

relates to coping ability and the use of particular coping habits across treatment can help support

workers to understand when in the day patients and caregivers might be more or less able to deal

with information related to the disease, or when in the day additional help might be needed.

Sleep loss negatively alters both cognitive processes such as event memory, judgment and

decision making, and emotional functioning and processing abilities, often resulting in poor

mood and elevating negative response styles in reaction to potentially trying situations (Lim &

Dinges, 2008; 2010; Rosales-Lagarde et al., 2012). Cancer patients and caregivers commonly

report poor sleep quality, suggesting they are more likely to have difficulties with cognitive and

emotional functioning and processing abilities. Avoidance coping has been documented among

141

cancer patients reporting reduced sleep; among caregivers, similar less functional coping

strategies such as venting, self-distraction, and self-blame have been associated with increased

sleep disruptions (Aslan, Sanisoglu, Akyol & Yetkin, 2009; Carter & Acton, 2006; Northouse,

Williams, Given & McCorkle, 2012; Hoyt, Thomas, Epstein & Dirksen, 2009). Conversely, the

use of positive, proactive coping strategies among caregivers has been associated with reduced

numbers of sleep disturbances (Zhang, Yao, Yang, & Zhou, 2014). However, none of these

studies have tracked patients and caregivers across treatment to see whether there is a change in

the relationship between coping and sleep quality, or whether sleep is related to coping at

specific time points. A better understanding of when sleep might be more or less related to one’s

coping can provide suggestions to patients and caregivers on crucial times in treatment when

their sleep is most important in helping them face stressors. Among those reporting poor sleep, it

can be estimated when there may be more or less problems with coping so that additional outside

assistance can be provided.

Personality refers to an individual’s characteristics that influence or determine patterns of

behaviour, feeling, and thought, and has been linked with circadian rhythms and sleep quality

(e.g., Cavallera, Gatto & Boari, 2014; Duggan, Friedman, McDevitt & Mednick, 2014;

Hintsanen et al., 2014; Hsu, Gau, Shang, Chiu & Lee, 2012). Personality has been linked with

coping, such that certain coping styles occur in increasing or decreasing frequency with certain

personality traits. A study among intensive care unit (ICU) nurses found that those reporting

greater conscientiousness, agreeableness, and or openness showed associations with coping

strategies indicative of problem focused coping in order to approach and resolve the workplace

related problem causing stress. Conversely, high neuroticism was strongly related to venting as a

coping strategy to verbally express negative feelings, reflective of an emotion-based coping style

aimed at reducing one’s negative feelings about the threat as opposed to altering the source of

stress itself (Burgess, Irvine & Wallymahmed, 2010). Similarly, when faced with the stress of

their own or their partner’s ill health, cancer patients and their caregivers may exhibit personality

related coping behaviours. Furthermore, as with chronotype and sleep quality, it is important to

understand whether personality has a consistent or varied influence on one’s use of particular

coping habits across treatment. A better understanding of personality’s influence on coping

behaviours will suggest which patients and caregivers may be more likely to use certain coping

142

behaviours, and allow support workers to identify those who may be at risk or require greater

assistance.

As chronotype, sleep quality, and personality each have the potential to influence coping at a

particular instance, it is important to look more closely at how these factors might interact across

time. We hypothesized that one’s chronotype, yet not necessarily their sleep quality, will

influence in the moment coping. Regarding specific coping behaviours, it is believed personality

will have a varied influence across treatment, depending on the various stressors patients are

faced with at different time points across treatment. Chronotype and sleep quality will also be

related to coping behaviours, however they may not exert a simultaneous influence. As predicted

for in the moment coping ratings, given that society often disregards the influence of sleep on

their behaviour therefore operating under a sleep debt, many people may not consider changes in

sleep quality and resulting influence on their abilities when they rate their behaviour.

Conversely, even when tired, people maintain some understanding of their own preference for

morning versus evening.

A greater understanding of how these factors will influence one’s self-perceived in the moment

rating of their own coping and one’s general use of particular coping behaviours over time is

important. For example, it can help support workers provide personalized assistance to patients

and caregivers who might be at risk of using harmful or less positive coping strategies. Not only

can this information be beneficial to cancer patients and their caregivers, it can be tested and

applied in other populations faced with chronic illness.

6.3! Materials and Methods

6.3.1! Participants and Procedures

Prospective patients receiving cancer treatment at Sunnybrook Health Sciences Centre, Odette

Cancer Centre (Toronto, Ontario, Canada) and when applicable, their caregivers, were identified

by their medical oncologist or nurse, and briefed on the general idea of the study. If patients and

caregivers reported interest in participating, they were approached by the study coordinator, who

explained the details and time commitment of the study. Patients receiving adjuvant and

neoadjuvant treatment were included in the study. If patients reported living with a life partner

143

who would be their primary caregiver, the partner was also given the opportunity to partake in

the same study. Inclusion criteria for patients included being 19 or older, receiving chemotherapy

for stage 1, 2, or 3 breast cancer, being able to read, write and speak English at least at a basic

level, and being female. While men can develop breast cancer, only female patient participants

were recruited. Individuals (patients or caregivers) who were going to continue following a

shiftwork work schedule during the treatment period were excluded. Caregiver inclusion criteria

required subjects to be 19 or older, have a spouse undergoing chemotherapy for breast cancer

treatment and enrolled in the study, and able to read, write and speak English at least at a basic

level. One female caregiver partook in the study, however, neither her nor her partner returned

correctly completely, usable data.

Potential participants were informed that their decision to participate or not would have no

bearing on their treatment. They were informed that their decision to take part was voluntary,

and while no formal remuneration would be provided, their participation in the study might

provide the opportunity to reflect on personal coping ability, and factors that influenced their

coping.

A total of 109 participants (86 patients, 23 caregivers) provided usable data that were included in

this study. Among those participants who withdrew, the general reason for withdrawal was that

people felt the study would be too time consuming. Of the 86 patients who took part, 38 patients

provided correctly completed data for their daily coping log. Only 3 caregivers correctly

completed their daily coping log, therefore their data was not assessed due to the small sample

size.

6.3.2! Measures

This section of the study builds on the work from the previous chapter, making use of the Horne-

Östberg Morningness Eveningness Questionnaire (MEQ), Pittsburgh Sleep Quality Index

(PSQI), and the University of Toronto Inventory of Morningness Eveningness (UTIME), while

also incorporating a daily coping log, the Brief COPE as an assessment of coping responses, and

Big Five Aspect Scales (BFAS) as a measure of personality. Follow-up surveys at the treatment

midpoint and endpoint sessions contained the PSQI and UTIME. Descriptions of these

questionnaires can be found in the General Methods section of this dissertation, section 2.2

Instruments. Baseline packages were completed prior to or on the first day of chemotherapy

144

treatment. Midpoint packages were completed at the halfway session of a patient’s treatment (the

treatment session number varied depending on the particular type of chemotherapy being

received). Endpoint packages were generally completed on the day of a patient’s final

chemotherapy session, however, given that some patients’ final chemotherapy sessions were

cancelled upon assessing the patient’s health that day, some final packages were completed at

what would have been the final chemotherapy session. Daily coping logs included 3 Likert scales

per day, one for morning, afternoon, and evening that were completed in the given time period or

left blank if a patient missed responding at the appropriate time. Daily coping logs were kept

across chemotherapy. Of the 86 patients who completed survey data, only 38 correctly

completed coping logs which are used in these data.

6.3.3! Statistics

Statistical analyses were performed using SPSS (Statistical Package for the Social Sciences)

version 23.0 for Mac. Pearson correlations were used to assess the strength and direction of the

association between daily coping log data and one’s MEQ and PSQI scores. Friedman Tests and

Wilcoxon Signed-Rank tests were used to assess whether there were any statistically significant

differences in coping log scores and coping behaviours at baseline, midpoint, and endpoint.

Paired sample t-tests were used to assess differences in pre- and post- chemotherapy treatment

coping scores. Multiple regression analyses were used to determine how well chronotype, sleep

quality, participant role and personality predicted the use of each coping behaviour at baseline,

midpoint, and endpoint.

6.4! Results

Descriptive data in Table 6.1 are mean ± standard deviation. In general, patient’s average daily

coping scores centered around three, which represented a rating of average on a scale of one to

five. From morning to evening as the day progressed, raw coping log scores tended to taper

down slightly, except for on the five days post chemotherapy at baseline, where the evening

coping score (3.19 1 ± .60) was higher than the afternoon coping score (3.14 ± .59), but still

below the morning coping rating (3.20 ± .72). Peak coping times as demonstrated by UTIME

scores occurred earliest in the day on chemotherapy treatment days (baseline, midpoint,

145

endpoint). When comparing an average of the five days prior to chemotherapy, and the five days

after, peak coping time occurred earlier in the day on the five days after treatment. In the five

days prior to treatment, patients showed a slightly later peak coping performance time in the day.

146

Table 6.1 Descriptive data for patient coping logs Mean (SD) Raw Patient Coping Log Scores

7:00 am to 11:00 am Baseline (n = 28) 3.32 (1.09) 1:00 pm to 5:00 pm Baseline (n = 33) 3.08 (.95)

7:00 pm to 11:00 pm Baseline (n = 35) 3.01 (1.13)

7:00 am to 11:00 am Average of 5 Days Post Baseline (n = 38) 3.20 (.72) 1:00 pm to 5:00 pm Average of 5 Days Post Baseline (n = 38) 3.14 (.59)

7:00 pm to 11:00 pm Average of 5 Days Post Baseline (n = 38) 3.19 (.60)

7:00 am to 11:00 am Average of 5 Days Pre Midpoint (n = 37) 3.90 (.82) 1:00 pm to 5:00 pm Average of 5 Days Pre Midpoint (n = 37) 3.86 (.77)

7:00 pm to 11:00 pm Average of 5 Days Pre Midpoint (n = 37) 3.75 (.88)

7:00 am to 11:00 am Midpoint (n = 33) 3.47 (1.20) 1:00 pm to 5:00 pm Midpoint (n = 31) 3.40 (.80)

7:00 pm to 11:00 pm Midpoint (n = 33) 3.10 (.94)

7:00 am to 11:00 am Average of 5 Days Post Midpoint (n = 37) 3.27 (.79) 1:00 pm to 5:00 pm Average of 5 Days Post Midpoint (n = 37) 3.23 (.84)

7:00 pm to 11:00 pm Average of 5 Days Post Midpoint (n = 37) 3.14 (.85)

7:00 am to 11:00 am Average of 5 Days Pre Endpoint (n = 38) 3.93 (.70) 1:00 pm to 5:00 pm Average of 5 Days Pre Endpoint (n = 38) 3.81 (.81)

7:00 pm to 11:00 pm Average of 5 Days Pre Endpoint (n = 38) 3.75 (.82)

7:00 am to 11:00 am Endpoint (n = 33) 3.92 (.84) 1:00 pm to 5:00 pm Endpoint (n = 32) 3.85 (.91)

7:00 pm to 11:00 pm Endpoint (n = 33) 3.61 (.87)

7:00 am to 11:00 am Average of 5 Days Post Endpoint (n = 38) 3.23 (.75) 1:00 pm to 5:00 pm Average of 5 Days Post Endpoint (n = 38) 3.09 (.69)

7:00 pm to 11:00 pm Average of 5 Days Post Endpoint (n = 38) 2.95 (.76)

Peak Coping Time – Patient Coping Log UTIME Scores Baseline (n = 26) .55 (.08)

Average of 5 Days Post Baseline (n = 36) .51 (.06) Average of 5 Days Pre Midpoint (n = 36) .49 (.07)

Midpoint (n = 30) .53 (.06) Average of 5 Days Post Midpoint (n = 33) .51 (.06) Average of 5 Days Pre Endpoint (n = 37) .49 (.07)

Endpoint (n = 34) .54 (.07) Average of 5 Days Post Endpoint (n = 36) .51 (.07)

147

Pairwise comparisons (Friedman tests) were performed to assess changes in raw coping scores

across treatment from baseline to endpoint (Table 6.2). Friedman tests were used instead of one-

way repeated measures ANOVAs due to the non-normal distribution of raw coping scores. Post

hoc analyses with Wilcoxon signed-rank tests were conducted with a Bonferroni correction

applied resulting in a significance level set at p < .017, rounded: p < .02 (i.e., 0.05/3, where three

represents number of comparisons; this allows the 7:00pm to 11:00pm midpoint versus endpoint

comparison to be significant, p = .019) when Friedman test results were significant (Table 6.2).

148

Table 6.2 Median and Friedman test data for changes in raw patient coping log scores across treatment, with Wilcoxon signed-rank tests with Bonferroni correction applied for differences in raw patient coping log scores across treatment

Median Friedman test Wilcoxon signed-rank tests Baseline vs.

Midpoint Baseline vs.

Endpoint Midpoint vs.

Endpoint Baseline Midpoint Endpoint df χ2 p z p z p z p Coping Log Times

7:00 am to 11:00 am 3.00 3.00 4.00 2 6.633 .036* -1.208 .227 -2.624 .009† -2.198 .028 1:00 pm to 5:00 pm 3.00 3.00 4.00 2 12.029 .002* -1.216 .224 -3.216 .001† -2.300 .021

7:00 pm to 11:00 pm 3.00 3.00 4.00 2 10.500 .005* -.436 .663 -2.230 .026 -2.347 .019† * p < .05 † p < .02

149

Pairwise comparisons (paired-samples t-test) were performed to assess whether there was a

statistically significant change in coping log ratings before and after midpoint and endpoint

chemotherapy treatments, in the morning, afternoon, and evening (Table 6.3). Of the six

comparisons, the only two outliers in difference score data were found in a comparison of coping

ratings five days before and after evening coping at chemotherapy midpoint, as assessed by

inspection of a boxplot value greater than 1.5 box-lengths from the edge of the box. The

remaining five pre- and post-chemotherapy treatment coping ratings had no outliers. The outliers

in the evening coping rating data were included after running both paired-sample t-tests with and

without the outliers and yielding similarly statistically significant results. The differences scores

were normally distributed, as assessed by Shapiro-Wilks test for all six comparisons (p > .05).

Across treatment, patients coped better (higher score) in the five days prior to treatment than in

the five days post treatment.

150

Table 6.3 Mean ± standard deviation and paired samples t-test data for pre- and post-chemotherapy coping rating comparisons Pre Chemo

Coping Post Chemo

Coping

Mean (SD) Mean (SD) df t p Midpoint

Morning (7:00am – 11:00am) 3.909 (.826) 3.274 (.803) 35 4.785 .000 Afternoon (1:00pm – 5:00pm) 3.865 (.783) 3.238 (.846) 35 5.352 .000 Evening (7:00pm – 11:00pm) 3.754 (.895) 3.141 (.865) 35 4.985 .000

Endpoint Morning (7:00am – 11:00am) 3.931 (.703) 3.233 (.754) 37 5.477 .000 Afternoon (1:00pm – 5:00pm) 3.809 (.813) 3.090 (.695) 37 5.066 .000 Evening (7:00pm – 11:00pm) 3.753 (.823) 2.951 (.760) 37 5.751 .000

151

Pearson correlations were conducted between patient’s coping log UTIME scores and their MEQ

score, as well as with their PSQI scores across treatment (Table 6.4). Pearson correlations were

used as the majority of data points involved were fairly normally distributed. Correlations

between coping log UTIME scores and MEQ scores were significant on the five days post

baseline and endpoint, and the five days prior to chemotherapy at midpoint and endpoint. While

not statistically significant, correlations between MEQ and coping log UTIME scores were

approaching significant at baseline and endpoint. However, except for one significant correlation

between PSQI and average coping score five days post midpoint treatment, no other associations

are significant. MEQ and PSQI score among this population group also do not show any

significant relationship.

152

Table 6.4 Pearson correlations between MEQ and PSQI and patients’ coping log UTIME scores

MEQ PSQI Baseline Midpoint Endpoint r p r p r p r p Baseline (n = 26) .385 .052† .082 .690 Average of 5 Days Post Baseline (n = 36) .361 .031* .014 .934 Average of 5 Days Pre Midpoint (n = 36) .584 .000** -.205 .230 Midpoint (n = 30) .145 .445 .124 .512 Average of 5 Days Post Midpoint (n = 33) .283 .111 .387 .026* Average of 5 Days Pre Endpoint (n = 37) .569 .000** -.006 .970 Endpoint (n = 34) .332 .055† .140 .431 Average of 5 Days Post Endpoint (n = 36) .376 .024* -.062 .718

† Not far from significance * p < .05 ** p < .01

153

Descriptive data for Brief COPE responses (Table 6.5) indicate that acceptance is the highest

used coping strategy by patients and caregivers at baseline, and continued to be the most highly

engaged in coping strategy at endpoint, at an increased level. Patients started out with slightly

higher levels of acceptance at baseline compared to caregivers, but caregivers had slightly higher

levels of acceptance by endpoint. Among patients, emotional support was the second most used

coping strategy at baseline, and remained the second highest at endpoint, however at a decreased

level. Among caregivers, the second most engaged in coping strategy across treatment was active

coping, however at a decreased level at endpoint compared to baseline. Among patients, active

coping was their third most used coping strategy, yet, was engaged in at a higher level than seen

among caregivers across treatment. The least engaged in coping strategy among patients and

caregivers across treatment was behavioural disengagement, but at a decreased level at endpoint

as compared to baseline. Caregivers consistently engaged in greater substance use than patients

across treatment and showed increased levels across treatment as compared to baseline, whereas

among patients substance use decreased from baseline to endpoint.

154

Table 6.5 Descriptive data for Brief COPE are mean ± standard deviation.

Patients Caregivers Baseline

Mean (SD) Midpoint Mean (SD)

Endpoint Mean (SD)

Baseline Mean (SD)

Midpoint Mean (SD)

Endpoint Mean (SD)

Self-Distraction 2.95 (.87) 2.72 (.88) 2.73 (.82) 2.28 (.74) 2.20 (.75) 2.22 (.81) Active Coping 3.08 (.80) 2.96 (.79) 2.81 (.82) 2.78 (.89) 2.48 (.94) 2.37 (.79)

Denial 1.51 (.72) 1.31 (.60) 1.21 (.42) 1.15 (.35) 1.13 (.46) 1.09 (.25) Substance Use 1.27 (.62) 1.14 (.51) 1.16 (.48) 1.30 (.47) 1.43 (.79) 1.33 (.54)

Emotional Support

3.26 (.81) 3.06 (.81) 2.97 (.81) 2.00 (.67) 1.83 (.86) 1.78 (.75)

Instrumental Support

2.91 (.79) 2.69 (.84) 2.60 (.87) 1.91 (.62) 1.70 (.75) 1.67 (.72)

Behavioural Disengagement

1.19 (.43) 1.11 (.35) 1.06 (.25) 1.13 (.38) 1.00 (.00) 1.04 (.21)

Venting 2.12 (.82) 2.14 (.73) 2.14 (.79) 1.35 (.44) 1.41 (.65) 1.33 (.42) Positive

Reframing 2.80 (.95) 2.89 (.83) 2.88 (.85) 2.39 (.80) 2.57 (.97) 2.11 (.85)

Planning 2.97 (.80) 2.62 (.86) 2.68 (.85) 2.52 (.87) 2.24 (.86) 2.13 (.87) Humour 2.14 (.95) 2.19 (1.04) 2.15 (.89) 1.50 (.75) 1.48 (.86) 1.48 (.68)

Acceptance 3.33 (.63) 3.49 (.54) 3.45 (.57) 3.28 (.64) 3.37 (.68) 3.48 (.79) Religion 2.48 (1.29) 2.42 (1.24) 2.43 (1.19) 1.89 (1.07) 1.76 (1.13) 1.89 (1.11)

Self-Blame 1.64 (.82) 1.41 (.57) 1.40 (.56) 1.17 (.32) 1.26 (.50) 1.26 (.40)

155

Pairwise comparisons (Friedman tests) were performed to assess differences in Brief COPE

scores across treatment from baseline to endpoint (Table 6.6a).

Among patients and caregivers, Friedman tests were used instead of one-way repeated

measures ANOVAs due to the non-normal distribution of Brief COPE scores. Only among

caregivers were some Brief COPE scores normally distributed: self-distraction (across

treatment), and active coping (only normally distributed at midpoint and endpoint). Post hoc

analyses with Wilcoxon signed-rank tests were conducted with a Bonferroni correction applied

resulting in a significance level set at p < .017 (i.e., 0.05/3, where three represents number of

comparisons) when Friedman test results were significant (Table 6.6b). Among patients, there

were significant changes between baseline and midpoint, and baseline and endpoint for various

Brief COPE domains, but no differences between midpoint and endpoint scores. Among

caregivers, only one significant change existed between Brief COPE scores across treatment,

from midpoint to endpoint.

Among patients, there was a significant change across treatment in Brief COPE scores for

Denial, Substance Use, Emotional and Instrumental Support, Behavioural Disengagement,

Planning, Acceptance, and Self-Blame, p < .05. Post hoc analysis were conducted with Wilcoxon

signed-rank tests with Bonferroni correction applied (p < .017). Denial, Planning, and Self-

Blame showed changes between baseline versus midpoint, and midpoint versus endpoint.

Median Denial coping levels for baseline, midpoint, and endpoint were 1.00 (1.00 to 2.00, 1.00

to 1.50, and 1.00 to 1.125 respectively). There were significant decreases in denial coping from

baseline to midpoint (p = .009), and baseline to endpoint (p < .0005). Median Planning scores

for baseline, midpoint, and endpoint were 3.00 (2.50 to 3.50), 2.50 (2.00 to 3.25), and 2.50 (2.00

to 3.50) respectively. Planning scores decreased from baseline to midpoint (p = .001), and from

baseline to endpoint (p = .006). Median Self-Blame scores for baseline, midpoint, and endpoint

were all 1.00. Self-Blame decreased from baseline to midpoint (p = .002), and baseline to

endpoint (p = .002). Substance Use, Emotional and Instrumental Support, Behavioural

Disengagement, and Acceptance only showed changes in scores between two times of treatment.

Median Substance Use coping levels for baseline, midpoint, and endpoint were each 1.00. There

was decreased coping via Substance Use from baseline to midpoint (p = .012). Median

156

Emotional Support coping levels for baseline, midpoint, and endpoint were 3.50 (2.50 to 4.00),

3.00 (2.50 to 4.00), and 3.00 (2.38 to 3.50) respectively. Patients sought less Emotional Support

from baseline to endpoint (p = .002). Median Instrumental Support coping levels for baseline,

midpoint, and endpoint were 3.00 (2.00 to 3.5), 2.50 (2.00 to 3.50), and 2.50 (2.00 to 3.50)

respectively. Use of Instrumental Support to cope decreased from baseline to endpoint (p =

.002). Median Behavioural Disengagement coping levels for baseline, midpoint, and endpoint

were all 1.00. Engaging in Behavioural Disengagement decreased across treatment from baseline

to endpoint (p = .011). Median Acceptance scores for baseline, midpoint, and endpoint were all

3.50 (3.00 to 4.00). Acceptance score changed from baseline to midpoint (p = .015).

Caregivers only exhibited significant changes in Brief COPE Positive Reframing scores across

treatment, p < .05. Post hoc analysis were conducted with Wilcoxon signed-rank tests with

Bonferroni correction applied (p < .017). Median Positive Reframing scores for baseline,

midpoint, and endpoint were 2.00 (2.00 to 3.00), 2.50 (2.00 to 3.50), and 2.00 (1.50 to 3.00),

respectively. Caregiver decreased Positive Reframing from midpoint to endpoint (p = .008).

157

Table 6.6a Median and Friedman test data for changes in Brief COPE scores across treatment among patients and caregivers Baseline Midpoint Endpoint Friedman test Mdn Mdn Mdn df χ2 p Patients

Self-Distraction 3.00 2.50 2.50 2 3.556 .169 Active Coping 3.00 3.00 3.00 2 3.508 .173

Denial 1.00 1.00 1.00 2 26.406 .000* Substance Use 1.00 1.00 1.00 2 12.822 .002*

Emotional Support 3.50 3.00 3.00 2 9.653 .008* Instrumental Support 3.00 2.50 2.50 2 12.531 .002*

Behavioural Disengagement

1.00 1.00 1.00 2 11.955 .003*

Venting 2.00 2.00 2.00 2 0.422 .810 Positive Reframing 3.00 3.00 3.00 2 1.033 .596

Planning 3.00 2.50 2.50 2 14.171 .001* Humour 2.00 2.00 2.00 2 0.575 .750

Acceptance 3.50 3.50 3.50 2 7.557 .023* Religion 2.00 2.00 2.25 2 0.182 .913

Self-Blame 1.00 1.00 1.00 2 9.329 .009* Caregivers

Self-Distraction 2.00 2.50 2.50 2 0.033 .983 Active Coping 3.00 2.50 2.50 2 5.040 .080

Denial 1.00 1.00 1.00 2 2.000 .368 Substance Use 1.00 1.00 1.00 2 1.040 .595

Emotional Support 2.00 2.00 2.00 2 3.085 .214 Instrumental Support 2.00 1.50 1.50 2 2.576 .276

Behavioural Disengagement

1.00 1.00 1.00 2 3.500 .174

Venting 1.00 1.00 1.00 2 0.136 .934 Positive Reframing 2.00 2.50 2.00 2 6.694 .035*

Planning 2.50 2.00 2.00 2 4.946 .084 Humour 1.00 1.00 1.00 2 1.067 .587

Acceptance 3.50 3.50 4.00 2 4.105 .128 Religion 1.50 1.00 1.50 2 3.211 .201

Self-Blame 1.00 1.00 1.00 2 3.379 .185 * p < .05

158

Table 6.6b Wilcoxon signed-rank tests with Bonferroni correction applied for differences in Brief COPE scores across treatment

Baseline vs. Midpoint Baseline vs. Endpoint Midpoint vs. Endpoint z p z p z p Patients

Denial -2.604 .009* -3.630 .000* -1.655 .098 Substance Use -2.503 .012* -1.745 .081 -0.741 .459

Emotional Support -1.720 .085 -3.149 .002* -1.242 .214 Instrumental Support -2.169 .030 -3.062 .002* -0.848 .396

Behavioural Disengagement

-1.761 .078 -2.551 .011* -0.828 .408

Planning -3.184 .001* -2.741 .006* -0.613 .540 Acceptance -2.443 .015* -1.530 .126 -0.875 .382 Self-Blame -3.100 .002* -3.084 .002* -0.367 .714

Caregivers Positive Reframing -1.053 .292 -1.467 .142 -2.632 .008*

* p < .017

159

Multiple regression analyses were run to predict individual Brief COPE styles across the course

of treatment from a participant’s role in the study (patient or caregiver), chronotype (MEQ

score), and sleep quality (PSQI score) at baseline midpoint or endpoint (Table 6.7). Log

transformations (Lg10) were carried out (for data in Table 6.7) for Denial, Substance Use,

Behavioural Disengagement, and Self Blame data at baseline, midpoint, and endpoint. Reverse

log transformations (Reverse_Lg10) were carried out for Acceptance data at baseline, midpoint,

and endpoint. There was linearity as assessed by partial regression plots and a plot of studentized

residuals against the predicted values. The Durbin-Watson statistic was used to assess

independence of residuals. The assumption of homoscedacity was met as assessed by visual

inspection of plots of studentized residuals versus unstandardized predicted values. No evidence

of multicollinearity was found as assessed by tolerance values greater than 0.1. There were a

minimal number of studentized deleted residuals ± 3 standard deviations and leverage values

greater than 0.2, and no values for Cook’s distance above 1. The assumptions of normality were

met as assessed by Q-Q plots. At different time points in the study, participant role, chronotype,

and sleep quality each contributed differently to the use of particular Brief COPE styles.

Regression coefficients and standard errors are presented in Table 6.7.

Note that the multiple regression analysis performed for behavioural disengagement at baseline

may not have independence of errors (residuals), based on assessment by the Durbin-Watson

statistic (.042). According to Field (2009) values less than one or greater than three are cause for

concern as these are on the fringes of the data range for the Durbin-Watson test statistic. A value

of approximately 2 for the Durbin-Watson test statistic indicates that no correlation exists

between residuals. At midpoint and endpoint, the Durbin-Watson test statistics (2.067 and 2.185,

respectively) indicates independence of errors. Caution should be taken when interpreting the

multiple regression analyses for behavioural disengagement at baseline.

Across treatment, multiple regression models consistently significantly predicted the use of Self-

Distraction, Emotional Support, Instrumental Support, Venting, and Humour (Self-Distraction:

baseline: F(3, 105) = 8.055, p < .0005 , adj. R2 = .164, midpoint: F(3, 104) = 3.494, p < .018, adj.

R2 = .065, endpoint: F(3, 105) = 5.466, p < .002, adj. R2 = .110; Emotional Support: baseline:

F(3, 105) = 16.594, p < .0005, adj. R2 = .302, midpoint: F(3, 104) = 13.738, p < .0005, adj. R2 =

.263, endpoint: F(3, 105) = 15.471, p < .0005 , adj. R2 = .287; Instrumental Support: baseline:

160

F(3, 105) = 10.331, p < .0005, adj. R2 = .206, midpoint: F(3, 104) = 8.920, p < .0005, adj. R2 =

.182, endpoint: F(3, 105) = 8.061, p < .0005 , adj. R2 = .164; Venting: baseline: F(3, 105) =

7.367, p < .0005, adj. R2 = .150, midpoint: F(3, 104) = 7.447, p < .0005, adj. R2 = .153, endpoint:

F(3, 105) = 9.291, p < .0005, adj. R2 = .187; Humour: baseline: F(3, 105) = 3.319, p < .050, adj.

R2 = .061; midpoint: F(3, 104) = 3.194, p < .050 , adj. R2 = .058; endpoint: F(3, 105) = 4.211, p <

.010, adj. R2 = .082). In each of these models, participant role was a significant predictor, p < .05,

both by itself, and when combined with other significant factors; patients consistently engaged in

greater Self-Distraction and Venting, sought more Emotional and Instrumental Support, and used

more Humour to cope than caregivers. For the use of Self-Distraction, chronotype was

significantly predictive at baseline (p < .05); those with higher chronotype scores (i.e., tendency

towards being M type) more likely report greater use of self-distraction to cope. While

chronotype was no longer a significant predictor of Self-Distraction at endpoint, sleep quality

significantly contributed to the model, p < .05; higher PSQI scores, (i.e., increasingly poor sleep

quality) predict greater use of Self-Distraction to cope. Similarly, sleep quality was also a

significant predictor for the use of Venting at endpoint (p < .05) and a trend moving towards

significance for this variable had been present since baseline; however unlike Self-Distraction,

chronotype was not a significant contributor to the multiple regression model at baseline. Sleep

quality and chronotype do not significantly predict Humour at any time, however it appears that

the significance levels for chronotype and sleep quality across treatment run counter to one

another, reflecting a similar pattern to that seen in the multiple regression models for Self-

Distraction and Venting.

For Active Coping, the multiple regression model was not significantly predictive at baseline,

F(3, 105) = .792, p = .501, adj. R2 = -.006. At midpoint, the multiple regression model indicated

an overall shift towards significance, but the model did not reach significance, F(3, 104) = 2.428,

p = .070 , adj. R2 = .039. While the midpoint model was not significant, the value of participant

role reached significance, p < .05, suggesting patients engage in more Active Coping than

caregivers. At endpoint, the multiple regression model was significantly predictive of Active

Coping, F(3, 105) = 3.985, p < .010, adj. R2 = .077; sleep quality acted as a significant predictor,

p < .05, such that higher PSQI scores, (i.e., increasingly poor sleep quality) predict greater use of

Active Coping.

161

Across treatment, multiple regression models were not significantly predictive of Denial or

Acceptance (Denial: baseline: F(3, 105) = 2.681, p = .051, adj. R2 = .045, midpoint: F(3, 104) =

.849, p = .470 , adj. R2 = -.004, endpoint: F(3, 105) = 1.446, p = .234 , adj. R2 = .012;

Acceptance: baseline: F(3, 105) = .443, p = .723 , adj. R2 = -.016, midpoint: F(3, 104) = 1.430, p

= .238 , adj. R2 = .012, endpoint: F(3, 105) = 1.800, p = .340 , adj. R2 = .004). Neither coping

behaviour was significantly predicted by multiple regression models, yet certain factors reached

or approached significance: for Denial, participant role was significant at baseline, p < .05; for

Acceptance, chronotype was not a significant predictor, but it predictive value approached

significance as treatment progressed (baseline: p = .290, midpoint: p = .130, endpoint: p = .080).

For Substance Use, the multiple regression model was not significantly predictive at baseline

(F(3, 105) = 1.409, p = .244, adj. R2 = .011) or endpoint (F(3, 105) = 1.066, p = .367, adj. R2 =

.002). At midpoint, the multiple regression model did significantly predicted Substance Use, F(3,

104) = 3.157, p < .05 , adj. R2 = .057; participant role acted as a significant predictor, p < .05,

suggesting caregivers engaged in greater Substance Use than patients at midpoint.

At baseline, midpoint, and endpoint, the multiple regression model was not statistically

significantly predictive of Behavioural Disengagement, F(3, 105) = .156, p = .925 , adj. R2 = -

.024; F(3, 104) = 1.766, p = .158, adj. R2 = .021; F(3, 105) = 2.611, p = .055 , adj. R2 = .041,

respectively. While neither participant role, chronotype, or sleep quality were significant

predictors across treatment (p < .05), it is important to note that chronotype and sleep quality

each approached significance as treatment progressed. For chronotype, the significance value

progressed from p = .869 at baseline, p = .168 at midpoint, and p = .069 at endpoint. For sleep

quality, significance value progressed from p = .886 at baseline, p = .267 at midpoint, and p =

.072 at endpoint. Participant role did not approach significance from baseline to endpoint.

Multiple regression models for Positive Reframing and Planning were only significant at

endpoint (Positive Reframing: baseline: F(3, 105) = 1.451, p = .232, adj. R2 = .012, midpoint:

F(3, 104) = 0.926, p = .431, adj. R2 = -.002, endpoint: F(3, 105) = 6.251, p = .001, adj. R2 =

.127; Planning: baseline: F(3, 105) = 2.010, p = .117 , adj. R2 = .027, midpoint: F(3, 104) =

2.195, p = .093 , adj. R2 = .032, endpoint: F(3, 105) = 5.559, p < .001, adj. R2 = .112). For both

models, at endpoint participant role was a significant predictor; patients engaged in greater use of

162

both strategies to cope. While neither model was significantly predictive at baseline, participant

role appeared as a significant factor in each, p < .05. Sleep quality was a significant predictor of

Planning at endpoint; worse sleep quality (i.e., higher PSQI score) predicted greater use of

Planning to cope. It appears there was a trend moving towards significance across treatment.

For Religion, the multiple regression model was significant at baseline, midpoint and endpoint,

F(3, 105) = 3.530, p < .017, adj. R2 = .066; F(3, 104) = 4.105, p < .008, adj. R2 = .080; F(3, 105)

= 3.684, p < .014 , adj. R2 = .069, respectively. Participant role was a significant predictor at

baseline and midpoint, p < .05, but not at endpoint, p > .05; patients more likely used Religion to

cope than caregivers. Chronotype was a significant predictor across treatment, p < .05; greater

preference for morningness predicted increased use of Religion to cope. Sleep quality was not a

significant predictor at any time point.

Multiple regression models significantly predicted the use of Self-Blame at baseline, F(3, 105) =

4.316, p < .007, adj. R2 = .084, and endpoint, F(3, 105) = 3.008, p < .034, adj. R2 = .053. At

midpoint, the multiple regression model did not significantly predict coping via Self-Blame, F(3,

104) = 1.036, p < .380, adj. R2 = .001. At baseline, participant role was the only significant

predictor, p < .05. At endpoint, participant role was no longer significant, but sleep quality had

become a statistically significant predictor, p < .05. Sleep quality did not show a progressive

trend towards significance across treatment. At baseline, while chronotype did not reach

significance, it was relatively close to a significant p value (p = .058).

163

Table 6.7 Summary of Multiple Regression Analyses for Brief-COPE Scores Across Treatment, assessing the predictive value of participant role, chronotype, and sleep quality

Baseline Midpoint Endpoint Brief COPE Included

Covariates B (SE) ß p Adjusted

R2 B (SE) ß p Adjusted

R2 B (SE) ß p Adjusted

R2 Self-Distraction .164 .065 .110 Constant

Participant role MEQ PSQI

1.269 (.531) -0.687 (.192) 0.029 (.008) 0.007 (.022)

-.319 .300 .027

.019 .001* .001*

.762

1.658 (.586) -0.500 (.203) 0.015 (.009) 0.030 (.026)

-.235 .153 .108

.006 .015*

.109

.262

1.985 (.541) -0.399 (.190) 0.003 (.008) 0.067 (.023)

-.195 .033 .276

.000 .038*

.717 .004**

Active Coping -.006 .039 .077 Constant

Participant role MEQ PSQI

3.340 (.599) -0.296 (.197) 0.000 (.009) 0.002 (.023)

-.147 .005 .008

.000

.136

.962

.937

2.540 (.572) -.446 (.198) .003 (.009) .027 (.026)

-.218 .038 .102

.000 .026**

.694

.297

1.601 (.546) -.373 (.192) .014 (.008) .047 (.023)

-.184 .157 .198

.004

.055

.095 .040*

Denial .045 -.004 .012 Constant

Participant role MEQ PSQI

.075 (.119) -.090 (.039) .002 (.002) .004 (.005)

-.219 .119 .082

.529 .024*

.212

.394

.034 (.101) -.049 (.035) .001 (.002) .002 (.004)

-.138 .037 .047

.734

.166

.706

.640

.135 (.077) -.029 (.027) -.002 (.001) .002 (.003)

-.106 -.131 .068

.083

.279

.179

.489

Substance Use .011 .057 .002 Constant

Participant role MEQ PSQI

-.134 (.105) .022 (.035) .003 (.002) .002 (.004)

.062 .183 .058

.205

.524

.060

.555

-.124 (.091) .082 (.032) .002 (.001) .004 (.004)

.248 .151 .092

.177 .011*

.115

.343

.008 (.086) .052 (.030) .001 (.001) .001 (.004)

.169 .040 .019

.930

.088

.680

.852

Emotional Support

.302 .263 .287

Constant Participant role MEQ PSQI

3.678 (.565) -1.247 (.186)

.013 (.008)

.013 (.021)

-.548 .124 .052

.000 .000*

.126

.530

2.758 (.570) -1.249 (.197)

.006 (.009) -.007 (.025)

-.535 .061

-.023

.000 .000*

.469

.787

1.877 (.539) -1.125 (.189)

.013 (.008)

.042 (.022)

-.494 .127 .157

.001 .000*

.124

.064

Instrumental Support

.206 .182 .164

Constant Participant role MEQ PSQI

3.939 (.554) -1.012 (.182)

.001 (.008) -.001 (.021)

-.483 .014

-.047

.000 .000*

.870

.592

2.282 (.574) -.992 (.198) .006 (.009) .009 (.026)

-.445 .060 .031

.000 .000*

.495

.734

2.103 (.574) -.877 (.202) .004 (.009) .032 (.024)

-.391 .040 .122

.000 .000*

.657

.181

Behavioural Disengagement

-.024 .024 .048

Constant .089 (.083) .289 -.061 (.058) .296 .065 (.048) .178

164

Participant role MEQ PSQI

-.018 (.027) .000 (.001) .000 (.003)

-.065 -.016 -.014

.511

.869

.886

-.030 (.020) .001 (.001) .003 (.003)

-.145 .134 .109

.139

.168

.267

.002 (.017) -.001 (.001) .004 (.002)

.009 -.175 .177

.928

.069

.072 Venting .150 .153 .187 Constant

Participant role MEQ PSQI

2.004 (.497) -.712 (.180) -.003 (.008) .036 (.021)

-.357 -.037 .156

.000 .000*

.680

.088

1.947 (.491) -.671 (.170) -.002 (.008) .038 (.022)

-.358 -.026 .159

.000 .000*

.771

.085

1.639 (.492) -.745 (.173) .002 (.008) .043 (.021)

-.382 .027 .187

.001 .000*

.755 .039*

Positive Reframing

.012 -.002 .127

Constant Participant role MEQ PSQI

2.878 (.608) -.444 (.220) .002 (.010)

-.021 (.025)

-.196 .018

-.081

.000 .046*

.854

.405

2.642 (.605) -.317 (.209) .003 (.009) .010 (.027)

-.149 .031 .038

.000

.132

.751

.705

2.096 (.579) -.705 (.203) .007 (.009) .043 (.024)

-.319 .074 .166

.000 .001*

.414

.077

Planning .027 .032 .112 Constant

Participant role MEQ PSQI

2.596 (.539) -.424 (.195) .004 (.009) .016 (.022)

-.209 .046 .068

.000 .032*

.627

.483

1.840 (.592) -.324 (.204) .007 (.009) .043 (.026)

-.153 .078 .160

.002

.116

.417

.105

1.611 (.568) -.435 (.200) .008 (.009) .069 (.024)

-.202 .088 .271

.005 .031*

.340 .005*

Humour .061 .080 .082 Constant

Participant role MEQ PSQI

2.747 (.606) -.633 (.219) -.011 (.010) -.002 (.025)

-.273 -.102 -.006

.000 .005*

.280

.947

2.600 (.702) -.702 (.243) -.008 (.011) .004 (.031)

-.276 -.069 .013

.018 .002*

.499

.644

1.071 (.586) -.622 (.206) -.003 (.009) .027 (.024)

-.285 -.208 .106

.001 .003*

.760

.266

Acceptance -.016 .012 .006 Constant

Participant role MEQ PSQI

.300 (.107) .017 (.039)

-.002 (.002) -.001 (.004)

.043

-.104 -.018

.006

.664

.290

.868

.253 (.104) .036 (.036)

-.002 (.002) .004 (.005)

.097

-.148 .095

.017

.326

.130

.338

.325 (.111) -.018 (.039) -.003 (.002) .001 (.005)

-.046 -.171 .016

.004

.644

.080

.869

Religion .066 .080 .069 Constant

Participant role MEQ PSQI

.825 (.806) -.637 (.291) .032 (.013)

-.016 (.033)

-.206 .229

-.044

.308 .031* .016*

.644

.613 (.824) -.697 (.285) .032 (.013)

-.002 (.037)

-.231 .240

-.005

.459 .016* .012*

.958

.377 (.785) -.501 (.276) .031 (.012) .037 (.033)

-.173 .237 .108

.632

.072 .013*

.263

Self-Blame .084 .001 .053 Constant

Participant role MEQ PSQI

.317 (.117) -.099 (.042) -.044 (.002) .006 (.005)

-.219 -.177 .122

.008 .021*

.058

.196

.247 (.103) -.047 (.036) -.002 (.002) -.004 (.005)

-.130 -.103 -.082

.019

.190

.293

.413

.176 (.093) -.017 (.033) -.002 (.001) .008 (.004)

-.051 -.149 .203

.062

.597

.117 .038*

Note. *p < .05; B = unstandardized regression coefficient, SEB = standard error of the coefficient; ß = standardized coefficient; Participant role: patient = 0, caregiver = 1

165

Comparing personality facets, descriptive data are mean ± standard deviation (Table 6.8).

There was homogeneity of variances for all personality facets, as assessed by Levene’s test for

equality of variances (p > .05). Among patients and caregivers, volatility was the lowest

expressed trait, however patients show a slightly higher average volatility level than caregivers,

though not significantly. The second lowest expressed trait among both groups is withdrawal,

which is again slightly less in caregivers than among patients, though not significantly. Among

patients, the highest expressed trait is compassion, while among caregivers, politeness was the

highest expressed trait. Patients were significantly more compassionate than caregivers, 0.43

(95% CI, 0.19 to 0.66), t(107) = 3.601, p < .000). There was no significant difference in

politeness between patients and caregivers. Patients were significantly more enthusiastic than

caregivers, 0.35 (95% CI, 0.05 to 0.53), t(107) = 2.293, p = .024. The only two traits expressed

among caregivers at a slightly higher level than among patients were industriousness and

assertiveness, yet there was no significant difference.

166

Table 6.8 Descriptive data and independent samples t-test for BFAS between men and women. Descriptive data are mean ± standard deviation, for patients and caregivers Patients

(n = 86) Caregivers

(n = 23) Independent Samples t-test

Mean (SD) Mean (SD) df t p Withdrawal 2.68 (.71) 2.45 (.60) 107 1.391 .167

Volatility 2.56 (.72) 2.40 (.69) 107 .949 .345 Compassion 4.33 (.51) 3.90 (.50) 107 3.601 .000**

Politeness 4.18 (.44) 4.02 (.49) 107 1.495 .138 Industriousness 3.75 (.67) 3.77 (.54) 107 -.099 .921

Orderliness 3.76 (.65) 3.53 (.62) 107 1.494 .138 Enthusiasm 3.84 (.68) 3.49 (.56) 107 2.293 .024*

Assertiveness 3.50 (.69) 3.55 (.57) 107 -.326 .745 Intellect 3.62 (.67) 3.60 (.62) 107 .122 .903

Openness 3.60 (.63) 3.33 (.65) 107 1.761 .081 * p < .05; ** p < .01

167

Multiple regression analyses were run to predict individual Brief COPE styles across the course

of treatment from a participant’s role in the study (patient or caregiver), chronotype (MEQ

score), sleep quality (PSQI score) at baseline midpoint or endpoint, and personality (BFAS)

(Table 6.9). Log transformations (Lg10) were carried out (for data in Table 6.9) for Denial,

Substance Use, Behavioural Disengagement, and Self Blame data at baseline, midpoint, and

endpoint. Reverse log transformations (Reverse_Lg10) were carried out for Acceptance data at

baseline, midpoint, and endpoint. There was linearity as assessed by partial regression plots and

a plot of studentized residuals against the predicted values. The Durbin-Watson statistic was used

to assess independence of residuals. The assumption of homoscedacity was met as assessed by

visual inspection of plots of studentized residuals versus unstandardized predicted values. No

evidence of multicollinearity was found as assessed by tolerance values greater than 0.1. There

were a minimal number of studentized deleted residuals ± 3 standard deviations and leverage

values greater than 0.2, and no values for Cook’s distance above 1. The assumptions of normality

were met as assessed by Q-Q plots. At different time points in the study, participant role,

chronotype, sleep quality and personality each contributed differently to the use of particular

Brief COPE styles. Regression coefficients and standard errors are presented in Table 6.9.

For Self-Distraction, the multiple regression model was significantly predictive, at baseline,

F(13, 95) = 2.277, p < .012, adj. R2 = .133, and endpoint F(13, 95) = 1.938, p < .035, adj. R2 =

.101; the model was not significant at midpoint, F(13, 94) = 1.369, p = .189, adj. R2 = .043.

Participant role and chronotype were significant predictors at baseline, p < .05, while only sleep

quality was a significant predictor at endpoint, p < .05. This mimics the same trend, shifting from

an influence of participant role and chronotype at outset, to a predictive influence of sleep quality

by endpoint as seen when personality factors were not considered. The models suggested that at

baseline, patients and or those reporting higher preference for morningness, and at endpoint,

those reporting poorer sleep quality (i.e., higher PSQI score) more likely engaged in Self-

Distraction to cope. While no personality factors were significant predictors of Self-Distraction,

politeness appeared to be approaching significance as treatment progressed (baseline: p < .809,

midpoint: p < .176, endpoint: p < .080).

The multiple regression model only significantly predicted Active Coping at baseline, F(13, 95)

= 2.602, p < .004, adj. R2 = .162. Orderliness significantly predicted Active Coping at baseline; a

168

greater tendency towards orderliness predicts increased Active Coping. At midpoint and

endpoint, the multiple regression model was not significantly predictive of Active Coping, F(13,

94) = 1.264, p = .249, adj. R2 = .031, and F(13, 95) = 1.605, p = .097 , adj. R2 = .068,

respectively. While the model was not predictive of Active Coping at midpoint, participant role

did reach significance, p < .05; patients are more likely than caregivers to engage in Active

Coping at midpoint. At endpoint, Active Coping did not reach significance, p = .097.

At baseline and midpoint, the multiple regression model significantly predicted Denial, F(13, 95)

= 2.142, p < .018, adj. R2 = .121, and F(13, 94) = 2.031, p < .026, adj. R2 = .111, respectively.

There were no consistent significant predictors across treatment. At baseline, participant role and

openness were significant predictors; patients and or those reporting lower openness were more

likely to engage in Denial. At midpoint, withdrawal was the only statistically significant

predictor for the multiple regression model; higher withdrawal signals increased use of Denial to

cope. By endpoint, the multiple regression model did not predict Denial, F(13, 95) = 1.184, p =

.303, adj. R2 = .022.

The multiple regression model was significantly predictive of engaging in Substance Use to cope

at baseline F(13, 95) = 1.946, p < .034, adj. R2 = .102 and midpoint, F(13, 94) = 2.542, p < .005,

adj. R2 = .158. Chronotype and industriousness predicted engaging in Substance Use to cope at

baseline and midpoint; greater tendency towards morningness (i.e., higher MEQ score), and

lower propensity for industriousness predicts increased use. At midpoint, participant role also

became a significant predictor; caregivers were more likely to engage in Substance Use to cope.

At endpoint, the multiple regression model was not significantly predictive of engaging in

Substance Use, F(13, 95) = 1.741, p = .065, adj. R2 = .082; while the model narrowly missed

being significant, participant role continued to be a significant predictor.

Across treatment, multiple regression models consistently significantly predicted the use of

Emotional and Instrumental Support, and Venting as means to cope (Emotional Support:

baseline: F(13, 95) = 4.042, p < .0005, adj. R2 = .268, midpoint: F(13, 94) = 4.037, p < .0005,

adj. R2 = .270, endpoint: F(13, 95) = 3.846, p < .0005, adj. R2 = .255; Instrumental Support:

baseline: F(13, 95) = 3.265, p < .0005, adj. R2 = .214, midpoint: F(13, 94) = 2.711, p < .003, adj.

R2 = .172, endpoint: F(13, 95) = 2.565, p < .004, adj. R2 = .158; Venting: baseline: F(13, 95) =

3.546, p < .0005, adj. R2 = .235, midpoint: F(13, 94) = 3.805, p < .0005, adj. R2 = .254, endpoint:

169

F(13, 95) = 4.671, p < .0005, adj. R2 = .306). Participant role was a significant predictor across

treatment for Emotional and Instrumental Support, and Venting models, p < .01, such that

patients reporting engaging in greater use of these methods to cope than caregivers; for

Instrumental Support, it was the sole significant predictor. For Emotional Support, intellect was a

significant predictor at midpoint, p < .05; higher intellect was predictive of less need for

Emotional Support. For Venting, withdrawal was a significant predictor at baseline and midpoint

(p < .05), but not at endpoint; reporting higher withdrawal predicted greater venting at baseline

and midpoint. By endpoint, withdrawal was no longer significantly predicted Venting; it

appeared that as treatment progressed withdrawal became a progressively less significant

predictor. The presence of participant role as a significant predictor in each of these models

across treatment reflects the same trend observe when personality was not included, suggesting

one’s role as a patient or caregiver plays a consistent role in determining their use of these

coping behaviours even when considering a range of different cofactors.

The multiple regression model was significant of Behavioural Disengagement at baseline, F(13,

95) = 4.197, p < .0005, adj. R2 = .278, and endpoint, F(13, 95) =1.857 , p < .045, adj. R2 = .093,

but not midpoint, F(13, 94) = 1.395, p = .177, adj. R2 = .046. At baseline, increased propensity

for withdrawal, and decreased levels of openness predicted greater use of Behavioural

Disengagement; assertiveness and orderliness narrowly missed reaching significance (p = .052

and p = .076, respectively). The model is not significantly predictive of Behavioural

Disengagement at midpoint, however it should be noted that both participant role and chronotype

score narrowly miss reaching statistical significance (p = .053 and p = .054, respectively) as

predictors. At endpoint, while the model was statistically significant, there were no significant

predictors. The two predictors with p-values closest to significance were enthusiasm (p = .076)

and sleep quality (p = .099). There was no prior trend at baseline or midpoint that indicated a

shift towards significance in these two predictors.

Positive Reframing was significantly predicted by the multiple regression model across treatment

(baseline: F(13, 95) = 2.207, p < .015, adj. R2 = .127, midpoint: F(13, 94) = 2.075, p < .023, adj.

R2 = .115, endpoint: F(13, 95) = 2.800, p < .002, adj. R2 = .178). Openness consistently predicted

positive reframing across treatment, p < .05; increased propensity for openness predicted greater

use of Positive Reframing. At baseline, greater tendency for industriousness predicted increased

Positive Reframing. At midpoint, tendency towards compassion predicted increased Positive

170

Reframing. At endpoint, increased ratings of enthusiasm and or being a patient predicted higher

levels of Positive Reframing. While orderliness consistently predicted Positive Reframing, these

other factors should not be ignored.

The multiple regression model was significantly predictive of Planning as a coping strategy at

endpoint, F(13, 95) = 2.390, p < .008, adj. R2 = .143. Higher intellect and poor sleep quality (i.e.,

higher PSQI score) significantly predicted Planning at endpoint. When personality was not

included in the multiple regression model, PSQI score also significantly predicted Planning

behaviours only at endpoint. The multiple regression model was not significantly predictive of

Planning at baseline, F(13, 95) = 1.407, p = .170, adj. R2 = .047, or midpoint, F(13, 94) = .997, p

=. 461, adj. R2 = .000. While the model was not significant at baseline, intellect appeared as a

significant predictor variable, p < .05.

At baseline and midpoint, multiple regression models did not significantly predict the use of

Humour as a coping strategy, F(13, 95) = 1.653, p = .084, adj. R2 = .073, and F(13, 94) = 1.632,

p = .090, adj. R2 = .071, respectively. While the models were not statistically significant at

baseline or midpoint, at both time points, participant role appeared as a significant predictor, p <

.05; patients were more likely than caregivers to use Humour to cope. By endpoint, the multiple

regression model significantly predicted using Humour to cope, F(13, 95) = 2.513, p < .005, adj.

R2 = .154; intellect was the only significant predictor – higher intellect predicts greater use of

Humour. Participant role (p = .078) was the next most significant predictor at endpoint; patients

more likely to used Humour to cope. Intellect did not approach significance at baseline or

midpoint as a predictor.

Across treatment, the multiple regression models were not significant predictors of Acceptance

(baseline: F(13, 95) = 1.710, p = .071, adj. R2 = .079, midpoint: F(13, 94) = .542, p = .892, adj.

R2 = -.059, endpoint: F(13, 95) = 1.017, p = .441, adj. R2 = .002). While the model was not

significant at baseline, withdrawal and volatility were significant predictors, p < .05. This is the

only time in this study where volatility had some predictive value on a coping behaviour. While

enthusiasm was not a significant predictor of Acceptance across treatment, it appeared to be

approaching significance as treatment progressed; this trend suggested that the higher one’s level

of enthusiasm, the less likely they engaged in Acceptance to cope. The lack of a significant

model reflects the pattern observed when personality was not included in the multiple regression,

171

suggesting that neither participant role, chronotype, sleep quality, or personality strongly predict

one’s use of Acceptance to cope with the stress being a cancer patient or caregiver.

Using Religion to cope was significantly predicted across treatment by multiple regression

models (baseline: F(13, 95) = 2.149, p < .018, adj. R2 = .121, midpoint: F(13, 94) = 2.384, p <

.008, adj. R2 = .144, endpoint: F(13, 95) = 2.252, p < .013, adj. R2 = .131). At baseline,

chronotype and assertiveness were significant predictors, p < .05; preference for morningness

and increased assertiveness predicted greater use of religion. Openness narrowly missed being a

significant baseline predictor, p = .052. At midpoint, participant role, industriousness, and

openness were significant predictors, p < .05, while assertiveness (p = .069) and chronotype (p =

.056) were not; being a patient, and reporting higher industriousness and openness increased

propensity for using Religion to cope at midpoint. At endpoint, industriousness and assertiveness

were significant predictors, p < .05, while participant role (p = .064), chronotype (p = .064), and

assertiveness (p = .068) were not significant; high industriousness and assertiveness predicted

greater use of Religion to cope. The lack of consistent significance of chronotype as a predictor

across treatment differs from when personality was not included in the model.

Multiple regression models significantly predicted engaging in Self-Blame at baseline, F(13, 95)

= 3.228, p < .0005, adj. R2 = .211, and endpoint, F(13, 95) = 2.375, p < .008, adj. R2 = .142. At

midpoint, the multiple regression model narrowly missed significance, F(13, 94) = 1.780, p =

.058, adj. R2 = .087. At baseline, withdrawal and industriousness significantly predicted

engaging in Self-Blame; lower propensity for withdrawal and higher industriousness predict

greater use of Self-Blame to cope. While the model was not significantly predictive at midpoint,

it should be noted that withdrawal narrowly missed being a significant predictor, p < .05. By

endpoint, industriousness and openness were significant predictors; lower industriousness and

higher openness predicted greater use of Self-Blame. While at baseline industriousness positively

predicted Self-Blame, by endpoint, industriousness negatively predicted engaging in Self-Blame;

withdrawal and openness, while not significant across treatment, also switched the direction to

which they contributed from baseline to endpoint.

172

Table 6.9 Summary of Multiple Regression Analyses for Brief-COPE Scores Across Treatment, assessing the predictive value of participant role, chronotype, sleep quality and personality

Baseline Midpoint Endpoint Brief COPE Included

Covariates B (SE) ß p Adjusted

R2 B (SE) ß p Adjusted

R2 B (SE) ß p Adjusted

R2 Self-Distraction .133 .043 .101 Constant

Participant role MEQ PSQI BFAS

Withdrawal Volatility

Compassion Politeness

Industriousness Orderliness Enthusiasm

Assertiveness Intellect

Openness

.736 (1.503) -.655 (.220) .029 (.009)

-.001 (.024)

.033 (.172)

.020 (.148) -.360 (.241) .057 (.233)

-.016 (.195) .198 (.159) .205 (.153)

-.135 (.183) .215 (.172) .183 (.142)

-.305 .299

-.003

.026

.016 -.217 .029

-.012 .146 .156

-.102 .160 .134

.625 .004* .003*

.978

.848

.893

.139

.809

.934

.215

.186

.463

.214

.201

3.221 (1.581) -.418 (.230) .017 (.010) .038 (.030)

.053 (.186)

-.019 (.156) -.076 (.253) -.331 (.243) -.083 (.207) .093 (.166) .289 (.165)

-.240 (.193) -.032 (.181) .022 (.151)

-.196 .176 .140

.042

-.016 -.046 -.172 -.060 .069 .221

-.183 -.024 .016

.044

.072

.091

.202

.775

.901

.764

.176

.690

.575

.083

.216

.860

.882

1.797 (1.258) -.277 (.213) .006 (.009) .052 (.026)

.236 (.173)

-.019 (.144) .117 (.230)

-.396 (.223) .027 (.195) .177 (.154) .055 (.149)

-.249 (.179) .103 (.168) .150 (.141)

-.135 .060 .214

.196

-.016 .074

-.214 .020 .137 .044

-.198 .081 .115

.221

.195

.549 .048*

.176 .897 .614 .080 .892 .254 .711 .167 .541 .291

Active Coping .162 .031 .068 Constant

Participant role MEQ PSQI BFAS

Withdrawal Volatility

Compassion Politeness

Industriousness Orderliness Enthusiasm

Assertiveness Intellect

Openness

.996 (1.383) -.168 (.202) -.009 (.009) .015 (.022)

-.314 (.158) .206 (.136) .090 (.222)

-.008 (.214) .091 (.180) .342 (.146) .023 (.141) .052 (.169) .122 (.158) .068 (.131)

-.083 -.098 .064

-.264 .178 .058

-.004 .070 .270 .019 .042 .097 .053

.473

.408

.305

.506

.050* .133 .686 .970 .615

.021* .870 .758 .440 .606

3.362 (1.530) -.497 (.222) .003 (.010) .037 (.029)

-.305 (.180) .071 (.151)

-.117 (.245) -.026 (.235) -.147 (.200) .123 (.160) 019 (.159)

.033 (.186)

.196 (.175)

.023 (.146)

-.242 .029 .141

-.252 .060

-.073 -.014 -.112 .094 .015 .026 .153 .017

.030 .028*

.783

.201

.094

.641

.634

.911

.463

.447

.904

.860

.266

.877

1.029 (1.471) -.359 (.215) .014 (.009) .019 (.026)

.255 (.175)

-.145 (.145) .052 (.233)

-.210 (.225) .147 (.197)

-.133 (.156) .074 (.150) .228 (.181)

-.112 (.170) .231 (.142)

-.177 .151 .078

.213

-.124 .033

-.114 .113

-.104 .059 .183

-.089 .178

.486

.097

.142

.477

.149

.319

.823

.355

.458

.397

.626

.210

.511

.109

Denial .121 .111 .022 Constant

Participant role -.062 (.289)

-.091 (.042)

-.221 .831

.034* .203 (.252)

-.051 (.037)

-.143 .423 .172

.286 (.206) -.031 (.030)

-.111

.168

.310

173

MEQ PSQI BFAS

Withdrawal Volatility

Compassion Politeness

Industriousness Orderliness Enthusiasm

Assertiveness Intellect

Openness

.003 (.002)

.002 (.005)

.048 (.033)

.020 (.028) -.012 (.046) .026 (.045)

-.029 (.038) .018 (.030) .011 (.029) .054 (.035)

-.034 (.033) -.056 (.027)

.170

.033

.198

.085 -.039 .071

-.109 .069 .044 .215

-.132 -.214

.086

.739

.149

.478

.792

.559

.449

.559

.707

.127

.306 .043*

.000 (.002)

.001 (.005)

.068 (.030) -.038 (.025) -.024 (.040) -.036 (.039) .042 (.033) .007 (.026) .043 (.026)

-.002 (.031) -.035 (.029) -.041 (.024)

.021

.030

.325 -.185 -.087 -.111 .187 .033 .197

-.009 -.160 -.181

.834

.779

.025* .134 .557 .361 .203 .780 .109 .951 .225 .092

-.001 (.001) .022 (.004)

.015 (.024)

-.007 (.020) -.011 (.033) .005 (.032)

-.021 (.027) .018 (.022) .005 (.021) .013 (.025)

-.038 (.024) -.011 (.020)

-.111 .052

.095

-.047 -.053 .019

-.116 .102 .028 .075

-.218 -.061

.291

.641

.528

.713

.729

.881

.457

.414

.825

.613

.116

.586 Substance Use .102 .158 .082 Constant

Participant role MEQ PSQI BFAS

Withdrawal Volatility

Compassion Politeness

Industriousness Orderliness Enthusiasm

Assertiveness Intellect

Openness

-.160 (.253) .043 (.037) .005 (.002) .000 (.004)

.046 (.029)

-.017 (.025) .028 (.041)

-.016 (.039) -.078 (.033) -.005 (.027) .012 (.026)

-.005 (.031) .002 (.029) .019 (.024)

.120 .305

-.011

.219 -.083 .102

-.051 -.343 -.024 .053

-.025 .009 .083

.528

.253 .003*

.911

.115

.498

.492

.679 .019*

.844

.655

.859

.948

.431

.020 (.230) .074 (.033) .004 (.001) .005 (.004)

-.003 (.027) .021 (.023)

-.010 (.037) -.007 (.035) -.073 (.030) -.006 (.024) -.012 (.024) .012 (.028)

-.001 (.026) -.003 (.022)

.224 .283 .128

-.016 .108

-.038 -.024 -.345 -.029 -.058 .061

-.007 -.014

.930 .029* .004*

.212

.909

.365

.793

.842 .017*

.804

.626

.662

.959

.894

-.095 (.222) .074 (.032) .002 (.001) .000 (.004)

.009 (.026) .030 (.022) .040 (.035)

-.034 (.034) -.050 (.030) .016 (.023)

-.001 (.023) .005 (.027)

-.022 (.026) .011 (.021)

.240 .117

-.012

.051

.166

.169 -.121 -.254 .082

-.004 .028

-.116 .055

.668 .024*

.250

.909

.725

.179

.255

.326

.093

.500

.975

.849

.385

.612

Emotional Support

.268 .270 .255

Constant Participant role MEQ PSQI BFAS

Withdrawal Volatility

Compassion Politeness

Industriousness Orderliness

2.035 (1.462) -1.255 (.214)

.012 (.009)

.011 (.023)

.020 (.167)

.076 (.144) -.187 (.235) .168 (.227) .066 (.190)

-.008 (.154)

-.551 .115 .042

.015 .058

-.107 .082 .045

-.005

.167 .000*

.202

.643

.904

.598

.428

.461

.730

.959

3.463 (1.512) -1.137 (.220)

.006 (.009) -.008 (.028)

.143 (.178)

-.042 (.149) .256 (.242)

-.292 (.232) .040 (.198)

-.061 (.159)

-.487 .057

-.026

.104 -.032 .141

-.138 .027

-.041

.024 .000*

.526

.782

.423

.773

.293

.211

.841

.702

2.950 (1.477) -1.057 (.215)

.015 (.009)

.028 (.026)

.275 (.176) -.230 (.146) .203 (.233)

-.301 (.226) -.085 (.197) .036 (.156)

-.464 .151 .104

.205

-.175 .116

-.146 -.058 .025

.049 .000*

.099

.286

.121

.117

.386

.187

.667

.818

174

Enthusiasm Assertiveness

Intellect Openness

.110 (.149)

.160 (.178)

.100 (.167)

.007 (.138)

.079

.115

.071

.005

.465

.371

.549

.961

.088 (.158)

.268 (.184) -.365 (.173) -.120 (.144)

.061

.187 -.250 .081

.579

.149 .038*

.406

.080 (.151)

.092 (.181) -.063 (.171) -.005 (.143)

.058

.066 -.044 -.003

.596

.613

.712

.972 Instrumental Support

.214 .172 .158

Constant Participant role MEQ PSQI BFAS

Withdrawal Volatility

Compassion Politeness

Industriousness Orderliness Enthusiasm

Assertiveness Intellect

Openness

3.506 (1.393) -1.094 (.204)

.003 (.009) -.014 (.022)

.056 (.159)

-.102 (.137) -.264 (.224) .056 (.216)

-.151 (.181) .255 (.147) .019 (.142) .149 (.170) .251 (.159)

-.125 (.132)

-.523 .036

-.058

.045 -.085 -.164 .029

-.113 .193 .015 .116 .191

-.093

.014 .000*

.701

.534

.726

.456

.240

.797

.406

.086

.892

.383

.119

.347

4.908 (1.539) -1.028 (.224)

.009 (.010)

.009 (.029)

.200 (.181) -.172 (.152) -.007 (.246) -.213 (.236) -.033 (.201) -.017 (.161) .023 (.160) .113 (.187)

-.304 (.176) -.033 (.146)

-.461 .088 .030

.152

-.134 -.004 -.106 -.023 -.012 .017 .083

-.219 -.023

.002 .000*

.360

.764

.273

.258

.976

.369

.869

.917

.886

.547

.088

.823

3.513 (1.543) -.807 (.225) .008 (.010) .025 (.027)

.197 (.184)

-.182 (.152) -.158 (.244) -.266 (.236) -.217 (.206) .193 (.163) .284 (.158)

-.182 (.190) .071 (.178) .090 (.149)

-.360 .084 .093

.149

-.141 -.092 -.132 -.151 .136 .208

-.132 .050 .063

.025 .001*

.387

.369

.287

.235

.518

.263

.296

.241

.074

.340

.693

.547

Behavioural Disengagement

.278 .046 .093

Constant Participant role MEQ PSQI BFAS

Withdrawal Volatility

Compassion Politeness

Industriousness Orderliness Enthusiasm

Assertiveness Intellect

Openness

-.025 (.177) -.017 (.026) .001 (.001)

-.003 (.003)

.057 (.020)

.023 (.017)

.023 (.028)

.023 (.027) -.024 (.023) -.033 (.019) -.009 (.018) .042 (.022)

-.031 (.020) -.039 (.017)

-.062 .110

-.100

.347

.147

.105

.090 -.138 -.191 -.051 .249

-.178 -.222

.886

.509

.219

.265

.006* .180 .430 .412 .289 .076 .635 .052 .130

.021*

.128 (.154) -.044 (.022) .002 (.001) .003 (.003)

.001 (.018) .001 (.015)

-.035 (.025) .017 (.024)

-.023 (.020) .010 (.016)

-.008 (.016) -.008 (.019) .008 (.018)

-.007 (.015)

-.211 .200 .109

.009 .005

-.219 .092

-.175 .077

-.060 -.063 .059

-.052

.409

.053

.054

.319

.953

.967

.154

.464

.249

.532

.633

.666

.666

.640

.075 (.124) -.001 (.018) -.001 (.001) .004 (.002)

.002 (.015) .020 (.012) .004 (.020) .008 (.019) .011 (.017)

-.012 (.013) -.023 (.013) -.022 (.015) -.005 (.014) .004 (.012)

-.007 -.145 .179

.021 .197 .033 .052 .102

-.108 -.214 -.102 -.049 .037

.547

.947

.153

.099

.884

.110

.822

.672

.497

.369

.076

.477

.712

.735

Venting .235 .254 .306 Constant

Participant role MEQ

-.915 (1.311) -.581 (.192) .002 (.008)

-.291 .021

.487 .003*

.819

1.226 (1.229) -.633 (.179) .005 (.008)

-.337 .061

.321 .001*

.503

1.855 (1.219) -.758 (.178) .012 (.008)

-.389 .134

.131 .000*

.131

175

PSQI BFAS

Withdrawal Volatility

Compassion Politeness

Industriousness Orderliness Enthusiasm

Assertiveness Intellect

Openness

.021 (.021)

.499 (.150)

.100 (.129)

.263 (.210) -.021 (.203) .149 (.170)

-.133 (.138) -.009 (.134) .256 (.160) .058 (.150)

-.124 (.124)

.093

.423

.087

.171 -.012 .117

-.106 -.007 .209 .047

-.097

.314

.001* .441 .214 .918 .383 .337 .949 .113 .697 .321

.030 (.023)

.296 (.145)

.152 (.121) -.065 (.196) -.203 (.189) -.024 (.161) -.070 (.129) .131 (.128) .253 (.150)

-.057 (.141) .037 (.117)

.124

.268

.140 -.045 -.119 -.020 -.059 .114 .219

-.049 .031

.200

.044* .213 .741 .285 .881 .586 .308 .095 .685 .754

.025 (.022)

.279 (.145)

.155 (.120) -.192 (.193) -.272 (.187) -.062 (.163) -.062 (.129) .155 (.125) .076 (.150) .069 (.141) .100 (.118)

.109

.243

.138 -.128 -.155 -.050 -.050 .130 .064 .057 .080

.250

.057

.201

.320

.148

.702

.633

.217

.613

.624

.400 Positive Reframing

.127 .115 .178

Constant Participant role MEQ PSQI BFAS

Withdrawal Volatility

Compassion Politeness

Industriousness Orderliness Enthusiasm

Assertiveness Intellect

Openness

1.531 (1.590) -.347 (.233) -.010 (.010) -.018 (.025)

-.124 (.182) .205 (.156)

-.316 (.255) -.094 (.247) .532 (.207) .097 (.168) .311 (.162)

-.174 (.194) -.101 (.182) .364 (.150)

-.153 -.098 -.069

-.092 .157

-.181 -.046 .367 .068 .224

-.125 -.072 .251

.338

.138

.318

.483

.498

.192

.218

.704 .012*

.562

.058

.372

.578 .018*

-.307 (1.515) -.164 (.220) -.004 (.009) -.002 (.028)

-.095 (.179) .316 (.149)

-.147 (.242) .030 (.233) .284 (.198) .030 (.159) .226 (.158)

-.002 (.185) -.016 (.174) .433 (.144)

-.077 -.038 -.007

-.076 .258

-.089 .015 .209 .022 .173

-.001 -.012 .319

.840

.459

.702

.950

.595 .037*

.545

.898

.155

.849

.156

.993

.927 .003*

-.806 (1.505) -.542 (.220) .000 (.009) .014 (.027)

.131 (.179) .153 (.148)

-.132 (.238) .063 (.231) .362 (.201) .030 (.159) .309 (.154)

-.096 (.185) .058 (.174) .322 (.146)

-.245 -.002 .052

.100 .120

-.078 .032 .255 .022 .228

-.071 .042 .228

.594 .015*

.982

.610

.467

.305

.579

.784

.075

.850 .048*

.605

.739 .029*

Planning .047 .000 .143 Constant

Participant role MEQ PSQI BFAS

Withdrawal Volatility

Compassion Politeness

Industriousness Orderliness Enthusiasm

.523 (1.482) -.315 (.217) .006 (.009) .016 (.024)

-.031 (.169) .242 (.146)

-.032 (.238) -.034 (.230) -.006 (.193) .071 (.156) .128 (.151)

-.156 .066 .068

-.026 .207

-.020 -.019 -.005 .056 .103

.725

.150

.520

.509

.854

.100

.894

.881

.974

.649

.401

.811 (1.603) -.157 (.233) .009 (.010) .037 (.030)

.096 (.189) .045 (.158) .286 (.256)

-.350 (.246) -.007 (.210) .111 (.168) .024 (.167)

-.074 .090 .138

.077 .037 .174

-.183 -.005 .082 .018

.614

.501

.395

.219

.612

.777

.267

.159

.972

.512

.887

-.017 (1.496) -.299 (.218) .014 (.009) .055 (.026)

.296 (.178)

-.006 (.147) .136 (.236)

-.360 (.229) -.003 (.200) .025 (.158) .168 (.153)

-.139 .143 .216

.233

-.005 .082

-.186 -.002 .019 .127

.991

.173

.147 .041*

.099 .968 .565 .119 .986 .874 .275

176

Assertiveness Intellect

Openness

-.111 (.181) .377 (.169) .090 (.140)

-.090 .298 .070

.540 .028*

.522

-.104 (.195) .093 (.184) .158 (.153)

-.080 .071 .117

.597

.612

.302

-.133 (.184) .377 (.173) .073 (.145)

-.101 .280 .053

.470 .032*

.617 Humour .073 .071 .154 Constant

Participant role MEQ PSQI BFAS

Withdrawal Volatility

Compassion Politeness

Industriousness Orderliness Enthusiasm

Assertiveness Intellect

Openness

3.772 (1.673) -.526 (.245) -.007 (.010) .014 (.027)

-.221 (.191) .212 (.164) .431 (.269)

-.406 (.260) -.314 (.217) .146 (.177)

-.006 (.171) .072 (.204) .040 (.191)

-.188 (.158)

-.227 -.070 .051

-.161 .159 .242

-.194 -.212 .100

-.004 .050 .028

-.127

.027 .034*

.490

.611

.251

.201

.112

.121

.152

.410

.971

.726

.835

.238

2.152 (1.859) -.542 (.270) -.002 (.012) .024 (.035)

-.085 (.219) .247 (.183) .415 (.297)

-.297 (.285) -.318 (.243) .055 (.295) .074 (.194) .180 (.226) .062 (.213)

-.177 (.177)

-.213 -.015 .074

-.056 .169 .210

-.129 -.195 .034 .048 .115 .039

-.109

.250 .048*

.881

.494

.700

.180

.166

.301

.194

.776

.702

.429

.770

.320

-.125 (1.508) -.392 (.220) .005 (.009) .029 (.027)

.110 (.179) .186 (.149) .358 (.238)

-.213 (.231) -.330 (.202) .146 (.160) .140 (.154)

-.011 (.185) .385 (.174)

-.136 (.146)

-.180 .048 .113

.085 .148 .213

-.108 -.236 .106 .105

-.008 .282

-.097

.934

.078

.621

.278

.541

.214

.136

.360

.105

.362

.365

.955 .030*

.355

Acceptance .079 -0.059 .002 Constant

Participant role MEQ PSQI BFAS

Withdrawal Volatility

Compassion Politeness

Industriousness Orderliness Enthusiasm

Assertiveness Intellect

Openness

.412 (.283) .014 (.041) .000 (.002) .001 (.005)

.086 (.034)

-.071 (.028) .028 (.045)

-.007 (.043) -.052 (.038) -.040 (.030) -.011 (.029) .053 (.035)

-.026 (.033) -.027 (.027)

.036 .013 .014

.370

-.313 .093

-.020 -.206 -.161 -.044 .218

-.106 -.109

.149

.736

.896

.897

.013*

.013* .532 .871 .173 .187 .716 .134 .429 .319

.231 (.288) .029 (.042)

-.002 (.002) .005 (.005)

-.005 (.034) .015 (.028)

-.052 (.046) .029 (.044)

-.017 (.038) .009 (.030) .023 (.030)

-.016 (.035) .023 (.033)

-.011 (.027)

.078

-.125 .104

-.025 .070

-.181 .087

-.072 .040 .101

-.069 .098

-.047

.425

.491

.251

.364

.873

.601

.263

.512

.650

.755

.448

.654

.493

.689

.657 (.297) -.026 (.043) -.003 (.002) .000 (.005)

-.014 (.035) -.003 (.029) -.063 (.047) -.011 (.046) -.018 (.040) -.020 (.031) .060 (.030)

-.014 (.036) -.031 (.034) .033 (.029)

-.065 -.155 .004

-.060 -.014 -.207 -.030 -.072 -.080 .248

-.058 -.123 .131

.029

.554

.144

.975

.694

.914

.182

.813

.649

.525

.050

.700

.376

.250

Religion .121 .144 .131 Constant

Participant role MEQ PSQI BFAS

Withdrawal

-.865 (2.171) -.597 (.317) .027 (.013)

-.027 (.035)

.226 (.248)

-.194 .199

-.076

.124

.691

.063 .045*

.439

.365

-.064 (2.120) -.767 (.308) .025 (.013)

-.022 (.040)

.092 (.250)

-.254 .188

-.058

.052

.976 .015*

.056

.576

.713

-.922 (2.032) -.556 (.296) .024 (.013)

-.022 (.036)

.402 (.242)

-.191 .184

-.065

.234

.651

.064

.064

.539

.100

177

Volatility Compassion

Politeness Industriousness

Orderliness Enthusiasm

Assertiveness Intellect

Openness

.017 (.213) -.091 (.348) -.238 (.337) .393 (.282)

-.105 (.229) .097 (.222) .537 (.264)

-.393 (.248) .404 (.205)

.010 -.038 -.086 .199

-.054 .052 .284

-.204 .205

.937

.793

.480

.167

.648

.661 .045*

.116

.052

.181 (.209) -.114 (.339) -.251 (.325) .587 (.277)

-.079 (.222) -.132 (.221) .475 (.258)

-.459 (.243) .402 (.202)

.104 -.048 -.092 .303

-.041 -.071 .255

-.243 .208

.388

.738

.443 .037*

.723

.552

.069

.062 .049*

-.061 (.200) .019 (.321)

-.369 (.311) .580 (.272)

-.035 (.215) -.154 (.208) .461 (.250)

-.372 (.235) .438 (.197)

-.036 .009

-.141 .312

-.019 -.087 .259

-.205 .236

.762

.953

.238 .035*

.871

.460

.068

.116 .028*

Self-Blame .211 .087 .142 Constant

Participant role MEQ PSQI BFAS

Withdrawal Volatility

Compassion Politeness

Industriousness Orderliness Enthusiasm

Assertiveness Intellect

Openness

.628 (.204) .043 (.030) .000 (.001)

-.001 (.003)

-.052 (.023) -.004 (.020) .002 (.033)

-.019 (.032) .068 (.026)

-.012 (.022) -.021 (.021) -.004 (.025) -.001 (.023) -.005 (.019)

.140

-.018 -.020

-.286 -.022 -.007 -.068 .349

-.064 -.115 -.022 -.003 -.018

.003

.154

.843

.833

.029* .850 .960 .554

.011* .566 .305 .871 .980 .779

.116 (.264) -.028 (.038) .000 (.002)

-.007 (.005)

.062 (.031) -.002 (.026) -.013 (.042) -.006 (.040) -.046 (.034) -.003 (.028) .021 (.027)

-.009 (.032) .006 (.030) .033 (.025)

-.078 .005

-.142

.287 -.010 -.044 -.018 -.199 -.014 .094

-.040 .025 .141

.660

.460

.964

.184

.050

.938

.767

.885

.182

.909

.447

.783

.850

.194

.122 (.238) .004 (.035)

-.001 (.001) .007 (.004)

.027 (.028)

-.018 (.023) .001 (.038)

-.003 (.036) -.080 (.032) .006 (.025) .016 (.024)

-.014 (.029) .012 (.027) .049 (.023)

.011

-.044 .167

.133

-.093 .002

-.010 -.364 .030 .077

-.069 .057 .224

.607

.915

.656

.113

.345

.435

.986

.931 .014*

.797

.510

.623

.657 .036*

Note. * p < .05; B = unstandardized regression coefficient, SEB = standard error of the coefficient; ß = standardized coefficient; Participant role: patient = 0, caregiver = 1

178

Pearson correlations were run between UTIME coping log scores at baseline, midpoint and

endpoint, and the respective Brief COPE scores, to see whether any particular coping behaviours

were associated with how patients’ coping changes across the day. Of the 14 correlations at each

time point, only substance use at endpoint showed a significant correlation to one’s coping

across the day (r = .352, p = .041). None of the other correlations were significant. Substance use

did not show any significant correlation to coping score across the day at baseline (r = -.013, p =

.948), or midpoint (r = .015, p = .938).

6.5! Discussion

Various studies have individually investigated chronotype, sleep quality, coping and personality

among cancer patients (e.g., Fiorentino & Ancoli-Israel, 2007; Papantoniou et al, 2015;

Porcerelli, Bornstein, Porcerelli & Arterbery, 2015; Zaza, Sellick & Hillier, 2005). These topics

have also been studied in numerous combinations among the general population (e.g., Afshar et

al., 2015; Vitale et al., 2015). We assessed coping two ways: based on one’s personal assessment

of how well they were coping, and based on one’s use of various behaviours as mechanisms for

coping with the situation. Chronotype and sleep quality were compared to self-rated coping, and

to coping behaviours, to understand how one’s rhythm and quality of sleep relate to and predict

coping. Personality was also examined in relation to one’s coping behaviours to further identify

people likely to engage in various coping behaviours over the course of chemotherapy treatment.

6.5.1! Personal Coping Assessment

On self-reported coping logs, patients consistently reported decreased coping ability as the day

progressed. However, on treatment days, patients reported increased coping ability at midpoint

and endpoint versus baseline. Increased self-reported coping ability may be the result of various

factors, including learning what to expect (i.e., from the treatment itself, side-effects, personal

recovery time, etc.), and realizing that one is better prepared to handle the situation than

originally thought.

Comparing coping in the days pre- and post- chemotherapy, patients reported poorer coping in

the days following chemotherapy, suggesting people are likely feeling an onset of negative

179

treatment related side effects in this time. Conversely, in the days prior to treatment, most

patients have likely recovered from the negative symptoms resulting from their previous

chemotherapy infusion. Anecdotally, in the days prior to a chemotherapy treatment session,

some patients report being happy that they are one step closer to completing the entire treatment

process, which may also influence more positive coping ratings. This result does not indicate that

patients actually cope more or less pre- or post- treatment, but rather there is a difference in the

extent to which they feel they are coping.

Plotting coping scores graphically indicates that patients reach their peak coping times earlier in

the day on treatment days and the five days post treatment (prior to midday), and later in the day

(early to midafternoon) in the five days prior to treatment. Treatment days and the five days post

chemotherapy are likely more demanding on patients (i.e., spending several hours out of the

house, waiting in the hospital, and facing side effects), compared to the five days pre-

chemotherapy when many patients are able to follow a more regular routine, without large

blocks of time spent in a hospital and or side effects. Prior to chemotherapy, having less external

stressors and being in better health may allow patients to function longer at slightly higher levels

before feeling they are less able to cope, resulting in a later peak coping time.

Significant positive correlations between patients’ peak coping scores across treatment and their

MEQ scores indicates that changes in coping were associated with one’s chronotype, as

predicted at the study’s outset; earlier chronotypes reached their peak coping times before later

(less morning preference) individuals. While significance for correlations between MEQ score

and UTIME coping at baseline and endpoint was not achieved, the correlation was approaching

significance; future studies with larger or more normally distributed groups across the

chronotype spectrum may achieve a significant correlation. Pre-chemotherapy correlations were

significant, suggesting that when patients have typically recovered from most of their previous

treatment’s side effects and are preparing for their next chemotherapy session, their coping

across the day is reflects their chronotype. The lack of significant correlation between MEQ

score and peak coping at midpoint and in the days post midpoint suggests people’s coping is not

reflective of their chronotype at this time. Several factors may influence this outcome. For

example, many patients, reaching their treatment midpoint may lead people to wonder whether

their treatment is working, and or they may feel fatigued and realize there is still another half of

their treatment course ahead which may be daunting. For several patients their chemotherapy

180

drug regiment changed at midpoint, possibly leaving many people worrying about new side

effects, or with many new thoughts and emotions in their mind which may have affected their

coping across the day. These thoughts and emotions may override the effects of chronotype and

cognitive functioning on coping ability.

Lack of many significant correlations between sleep quality and peak daily coping score reflects

the same trend observed in Chapter 5, where recalled coping ability and feelings across the day

did not correlate significantly with sleep quality. The associations between sleep quality and

peak coping were not uniformly positive or negative, unlike the consistently positive association

between MEQ and peak coping scores. Chronotype and sleep quality were not related in this

population, unlike in the general population (Vitale et al., 2015). The sole significant correlation

between sleep quality and peak coping in the five days post chemotherapy at midpoint is likely a

false positive (Type 1 error). The lack of correlation between peak coping and sleep quality

suggests sleep does not influence patients’ ratings of their own ability to cope in this population.

This is important to consider since poor sleep quality is associated with reduced cognitive

function, which has been linked to less positive or adaptive coping (Byun, Gay & Lee, 2016;

Goretti, Portaccio, Zipoli, Razzolini & Amato, 2010). In our society, sleep debt is often

overlooked as many people try to schedule as many activities in their day, and many people

mistakenly believe they can easily recover from sleep debt (Cohen et al., 2010). Therefore,

patients likely continue to disregard the effects of their sleep debt on their coping ability.

6.5.2! Coping Behaviours: Changes in Coping Scores

Patients and caregivers completed the Brief COPE at baseline, midpoint, and endpoint. Among

patients, use of the majority of coping behaviours changed across treatment. Conversely,

caregivers use of various coping strategies across treatment remained relatively consistent. Of

the fourteen Brief COPE behaviours, with the exception of self-distraction, patients and

caregivers reported that their most often used coping strategies were adaptive (i.e., acceptance,

emotional support, active coping, planning, instrumental support, positive reframing, religion,

and humour). The remaining coping strategies were maladaptive or less solution based. While

patients and caregivers typically reported engaging in more proactive coping behaviours, they

did not engage in the same behaviours at the same rate. Except for acceptance, the coping

behaviours that demonstrated significant changes in use across treatment typically displayed

181

decreased use. This reflects previously reported results that found many coping reactions are

more prominent early on when dealing with a crisis compared to later (Carver et al., 1993).

Self-distraction was the only coping behaviour predicted by chronotype and sleep quality,

regardless of personality trait inclusion in multiple regression. This relationship is modeled in

part in other coping behaviours, however there is a clear shift in predictive influence for self-

distraction across treatment. Self-distraction has been classified as engagement coping (focuses

on dealing with the stressor or distressing emotions) and accommodative coping (aims to adapt

to effects of the stressor) (Carver & Connor-Smith, 2010). While literature on self-distraction

and chronotype is sparse, some research suggests evening types (i.e., lower MEQ score) show

less emotionally adaptive coping (Ottoni, Antoniolli & Lara, 2012), suggesting a similar

relationship to that observed between chronotype and self-distraction. While literature on sleep

quality and coping styles is also sparse, research suggests maladaptive behaviours increase with

poor sleep (Lenjavi, Ahuja, Touchette & Sandman, 2010), which indicates an opposite

relationship to that found between sleep quality and self-distraction in the multiple regression

models. One possible explanation might be that poor sleep quality results in greater self-

distraction as individuals utilize greater self-distractive thought and behaviours when unable to

sleep. The shift in predictive value across treatment from chronotype to sleep quality is

important, as it highlights that at the outset of treatment, many people are likely still following a

schedule based on their chronotypic demands. However, as treatment progresses and fatigue (a

commonly reported feeling by cancer patients and their caregivers) increases, people may be

sleeping at odd hours, not necessarily reflective of their chronotype. In which case, their ability

to perform becomes dependent on the sleep they have obtained, instead of an innate rhythm.

Patients and caregivers most reported using acceptance across treatment. Both groups reported

consistent increases in acceptance as treatment progressed. Patients may demonstrate greater

acceptance as treatment progresses, as side effects are experienced, and the diagnosis and

treatment become more of a reality. Interestingly, acceptance was not significantly predicted by

either of the multiple regression models at any time across treatment. While the multiple

regression model xincluding personality was not significant at baseline, the predictive roles of

withdrawal and volatility did reach significance; these are the two opposite components of

neuroticism. Volatility refers to high emotional reactivity, and difficulty controlling emotional

impulses, and withdrawal encompasses traits of negative affect, such as anxiety and depression.

182

It is believed that these are part of a joint system governing sensitivity to threat and punishment

(DeYoung, Quilty & Peterson, 2007). Neuroticism has been negatively correlated to acceptance

in previous research (Afshar et al., 2015). While heightened withdrawal and decreased volatility

contributed to increased acceptance at baseline, at midpoint and endpoint, withdrawal and

volatility negatively contributed to greater acceptance. It is possible that at baseline, while low

volatility may have helped someone accept the diagnosis, increased withdrawal may have

contributed to greater acceptance by reducing the amount of questioning a person did, instead

making them submit to the situation. Future research with larger populations should reexamine

this relationship to assess whether the previously documented inverse relationship between

neuroticism and acceptance exists across treatment with withdrawal and volatility.

While patients’ acceptance levels increased, denial decreased across treatment. The increase in

acceptance appears to plateau after midpoint, yet denial continues to decrease across treatment,

suggesting patients continued coming to terms with their cancer diagnosis. Kreitler (1999) found

that cancer patients reporting denial in the early phases of coping following diagnosis seemed to

report reduced anxiety compared to those with lower denial during this period. Furthermore,

Kreitler (1999) found that patterns of denial tended to be highest earlier in treatment, while

longitudinal studies showed that denial decreased as time passed for a period of up to two years

post treatment. This decreasing pattern is reflected in our own research. Future research in the

same population should consider whether denial is directed towards one’s emotions or the

information relating to the diagnosis given that emotional denial may provide some positive

benefit while information denial may lead to a worse prognosis or delays in treatment. Multiple

regression models for denial were only predictive when personality was included, then only at

baseline and midpoint. While the multiple regression model was not significant across treatment

when personality was not included, participant role appeared as a significant predictor at baseline

only, in models with and without personality, suggesting at this time, patients show greater

denial at baseline than caregivers. Lower openness at baseline contributed to increased denial,

reflecting previously reported results (Carver & Connor-Smith, 2010), while higher withdrawal

at midpoint also predicted increased denial.

Patients sought emotional support as the second most used coping strategy at baseline, and the

second highest at endpoint, but at a significantly reduced level than at baseline. Good emotional

support protects against poor emotional and mental health (Clough-Gorr, Ganz and Silliman,

183

2007). As treatment progressed, patients may have sought less emotional support for many

reasons, including having learned to deal with some of the stressors they faced, or feeling that

after many months, they could no longer turn to others to continuing seeking support. By

endpoint, patients may have required less emotional support as their acceptance for the situation

generally increased. Instrumental support was sought at a lesser extent than emotional support,

and its use decreased across treatment. The multiple regression models predicting emotional and

instrumental support were significant at all time points across treatment, with and without the

inclusion of personality. For both types of support, in multiple regression analyses with and

without personality, being a patient consistently predicted seeking greater emotional support

across treatment. When including personality in the model, only intellect significantly predicted

emotional support at midpoint; lower intellect predicted greater need for emotional support.

Individuals higher in intellect may have greater emotional intelligence (i.e., ability to recognize

and analyze emotional information and conduct general problem solving), and better understand

and process their own emotions instead of seeking emotional support from others; this reflects

previous work which suggests high emotional intelligence is negatively related to perceptions of

acute and chronic stress (Singh & Sharma, 2012). The lack of predictive value of sleep quality

and chronotype, along with personality on emotional and instrumental support suggest other

factors not explored here contribute to the use of these coping behaviours.

Caregivers reported fairly consistent use of various coping behaviours across treatment. Given

the small group of caregivers in this study, changes may not be as easily observed as in a large

population. Alternately, if caregivers are consistent in their coping strategies across treatment, it

may be due to their stressor remaining relatively consistent: dealing consistently with a partner

facing a chronic health condition; conversely, patients deal with many changes in their

wellbeing, their acceptance of the diagnosis, and their ability to handle daily life. Caregivers only

reported a significant decrease in positive reframing, from midpoint to endpoint; patients

reported no changes in positive reframing across treatment. Since positive reframing is typically

aimed at managing distressful emotions rather than dealing with the stressor itself, the overall

decline in caregiver use of this coping strategy may be due to learning to handle the stressful

situation (i.e., partner’s cancer diagnosis and treatment) as time progresses. The slight non-

significant increase in positive reframing from baseline to midpoint, may result from caregivers

feeling unable to help their partner as treatment progresses, and therefore using this behaviour to

184

deal with their emotions and cope with the situation. As chemotherapy treatment nears

completion and the stressor diminishes, caregivers may be less dependent on positive reframing.

Future research should examine this by assessing changes in caregiver ratings of the stressor.

While caregivers reported significant changes in positive reframing, patients displayed higher

levels overall, and by endpoint, multiple regression models, with and without personality,

demonstrated this difference: participant role predicted a significant difference in endpoint use of

positive reframing. Chronotype and sleep quality were not significant contributors. Across

treatment, high openness significantly predicted greater use of positive reframing, likely because

it allowed individuals to consider the possibility of ‘finding the silver lining’ in an otherwise

negative situation. At baseline, industriousness increased positive reframing, and while it was not

significantly predictive at midpoint or endpoint, it hovered near significance. Volatility only

significantly predicted positive reframing at midpoint, suggesting that this association should be

reexamined in future studies to verify its validity. Enthusiasm was only significantly predictive at

endpoint, however it also hovered near significance at baseline. Enthusiasm is a component of

extraversion, which has been positively related to positive reframing (Afshar et al., 2015). Future

studies should reexamine the predictive role of industriousness and enthusiasm, as well as

volatility, to understand their predictive value on positive reframing and identify individuals at

risk for low use of this coping strategy.

Overall, patients reported using more planning behaviours (i.e., understanding and coming up

with strategies on how to deal with the problem) to cope than caregivers. While caregivers use of

planning was fairly consistent across treatment, patients reported significantly reduced midpoint

and endpoint planning compared to baseline. Planning is an active and adaptive coping strategy

(Burker, Evon, Loiselle, Finkel & Mill, 2005; Carver, Scheier & Weintraub, 1989). Planning

behaviours may decrease as time passes once patients establish steps and patterns to face their

illness (e.g., begun treatment, adopted strategies to overcome stress, progression toward recovery

is occurring, etc.). Multiple regression models predicting planning were only significant at

endpoint. With and without personality, endpoint sleep quality significantly predicted planning

such that worse sleep quality (i.e., higher sleep scores) predicted greater planning. It is possible

that planning behaviours occur with sleeplessness, as people are awake, and may ruminate over

possible strategies to overcome their stressor. As treatment progresses, the significance of sleep

quality as a predictor increases. Future studies should assess whether ruminating about planning

185

strategies interferes with sleep quality, or whether poor sleep quality leads to greater planning.

Chronotype did not have a predictive role for planning behaviour across treatment. Intellect

predicted planning behaviour at baseline and endpoint, not midpoint. While intellect appeared to

predict use of planning at baseline, the model overall was not significant, suggesting the model

prevented against Type I error (i.e., incorrect rejection of a null hypothesis), yet the role of

intellect should not be disregarded, especially due to its significant predictive value at endpoint

when the model is significant. Retesting these findings in a larger population may confirm

whether intellect plays a significant predictive role across treatment among cancer patients and

caregivers for their use of planning behaviour. Previous studies have suggested that intellect is a

predictive trait for one’s use of planning as a coping behaviour (e.g., Lawson, Bundy, Belcher &

Harvey, 2013), therefore it is important to reconsider the results of this study in a larger

population experiencing the same stresses. This would clarify whether intellect is a strong

general predictor across the population of planning behaviours when dealing with a stressor, and

if there are specific times in treatment or whether across treatment intellect predicts planning.

Active coping was the second most engaged in behaviour by caregivers, and the third most used

patient coping behaviour, yet patients reported more active coping than caregivers. Neither

patients nor caregivers demonstrated significant changes in active coping across treatment.

Patients may report greater active coping than caregivers because they are facing the actual stress

of the disease itself, whereas caregivers face the stress of their partner being sick. Dealing with

the disease itself and faced with one’s own mortality daily may make an individual more likely

to engage in actions to alleviate the associated stresses. While caregivers report engaging in

active coping, their rate of engagement may be lower if they feel they cannot do anything to treat

the disease itself or the effects of the disease on their partner. When not including personality,

the multiple regression model for active coping was only significant at endpoint, and was

predicted by sleep quality (poorer sleep quality increases active coping). When including

personality, the multiple regression model was only significant at baseline, suggesting that higher

orderliness predicted greater active coping. Midpoint multiple regression models with and

without personality were non-significant, however participant role appeared as the only

significant predictor, suggesting participant role was related to active coping use. The model

again appears to be preventing against Type I error, which may have resulted due to having

multiple non-significant predictors included. However, all predictors were included in the

186

standard multiple regression in an attempt to model how all factors interacted together to create

an influence in real life.

Patients and caregivers remained relatively consistent in their use of humour and religion to help

cope across treatment, with no significant changes in their use. Previous research with cancer

patients identified humour as a valuable therapeutic tool for decreasing anxiety and discomfort,

while laughter specifically has been noted as a healthy method to relieve stress, let patients feel a

sense of control, and aide in body relaxation (e.g., Christie & Moore, 2005). Humour may have

been used less among caregivers than patients because while patients can make light of their own

poor health, other individuals, including caregivers, may see it as inappropriate to make light of

the patient’s ill health. In multiple regression models, when not considering personality,

participant role was the only significant factor predicting one’s use of humour, while chronotype

and sleep quality were not significantly predictive. When personality was included, the only

significant multiple regression model was at endpoint, where only intellect significantly

positively predicted one’s use of humour; participant role was close to but not statistically

significant. However, at baseline and endpoint, when considering personality, while neither

model was significant, participant role contributed to one’s use of humour. Future studies with

larger population groups and a more equal distribution of patients and caregivers must reexamine

this difference to verify whether participant role actually does contribute to a significant

predictive model across treatment.

Female patients also used religion to cope more than male caregivers, reflecting findings in the

general population which suggest that across the age span, women consistently engage in more

religious practices and have more spiritual beliefs than men (Jones et al., 2011). As a coping tool

in other chronic illnesses and conditions, women have reported greater use of religion to cope

than men (e.g., Dunn & Horgas, 2004). It is important to note that the religious affiliation of

people in this study was not surveyed, therefore it is unclear if people were predominantly of one

particular religious background, or held a variety of religious beliefs. Some studies suggest that

in the Christian faith, women are more religious than men, but that among other religions such as

Judaism or Hinduism, women are less involved in the religion (Loewenthal, McLeod &

Cinnirella, 2000). Future research should examine participants’ faith to understand if women do

rely on more religion-based coping than men, or if this is specific to women of a particular faith.

When personality was not included in multiple regression, participant role predicted significant

187

differences in one’s use of religion to cope at baseline and midpoint, while chronotype was

significantly predictive across treatment. When personality was included, chronotype was only

significant at baseline, and approached significance at midpoint and endpoint, while participant

role was only significant at midpoint and approached significance at baseline and endpoint.

Personality factors played significant predictive roles in the use of religion across treatment as

well: industriousness and openness significantly positively predicted use of religious coping at

midpoint and endpoint, while only openness approached significance at baseline; assertiveness

positively predicted religious coping at baseline and neared significance at midpoint and

endpoint. These findings support those of previous studies suggesting a positive relationship

between religious beliefs and openness, industriousness – which is part of conscientiousness, and

assertiveness – which is part of extraversion (Khoynezhad, Rajaei & Sarvarazemy, 2012;

Saroglou, 2002) Future studies with larger populations must reexamine the predictive influence

of personality and participant role on one’s use of religion as a means to cope, particularly to

verify whether the variables that neared significance are actually significant across treatment.

Patients and caregivers demonstrated relatively consistent levels of venting across treatment,

with patients reporting greater venting behaviour than caregivers. Venting can be positive –

allowing one to discuss their feelings or play out their thoughts and devise a solution for a

problem, or negative – allowing one dwell on the negative aspects of the situation and impede

actively coping with the stressor (Carver, Sheier & Weintraub, 1989). Future research on venting

and what causes it should also examine if people are using it as a positive or negative behaviour,

to devise appropriate strategies to address the individual’s concerns. Multiple regression models

for venting were significant across treatment, with and without personality. Participant role was a

consistent predictor of venting, supporting previous research that suggests women engage in

venting more than men (Sullivan, 2002). Without personality in the multiple regression, sleep

quality was a significant predictor at endpoint, such that poorer sleep predicted greater venting;

at baseline and midpoint sleep quality was not significant but was not far off. However, when

personality was included in the model, sleep quality was no longer a significant predictor.

Withdrawal significantly contributed to baseline and midpoint venting, interestingly suggesting

that the more withdrawn an individual is, the greater their venting. Withdrawal became a less

significant predictor as treatment progressed, and was no longer significant at endpoint. While it

might seem that withdrawal would be negatively predictive of venting since a person high in

188

withdrawal would be more apt to keep to themselves, these results support previous research that

found individuals high in neuroticism (of which withdrawal is a component) engage in greater

venting of emotions (Vollrath, Torgersen & Alnaes, 1998). Chronotype was not a significant

predictor in either model of venting across treatment.

Across treatment, patients and caregivers consistently reported low behavioural disengagement

as a coping strategy. Only at baseline did personality traits of withdrawal and openness

significantly predict behavioural disengagement levels. While the model appeared to be

significant at endpoint when considering personality, no predictors were significant at p < .05,

suggesting that some of the almost significant variables (sleep quality, p = .099; enthusiasm, p =

076) allow the model to appear significant. When personality was not included, sleep quality also

neared significance. It is possible that with all variables in the model, the true effect of sleep

quality is not seen; future research should reassess the relationship between sleep quality and

behavioural disengagement.

Substance use was reported to be low among patients and caregivers. Substance use is

considered a maladaptive coping strategy (Carver, Scheier & Weintraub, 1989). Literature on

drug and alcohol use suggests women typically abuse drugs and alcohol at a lesser rate than men,

yet the number of women using these substances is on the rise (Becker & Hu, 2008). Among

caregivers, research indicates that those who experience higher social and emotional burdens as a

result of caregiving are at increased risk for overconsumption of alcohol (Rospenda, Minich,

Milner & Richman, 2010). Individuals displaying risk factors for demonstrating such

maladaptive coping strategies should be monitored. While only participant role was a significant

predictor when not considering personality, once personality was added to the multiple

regression model, chronotype also became a significant predictor of substance use at baseline

and midpoint (i.e., increased tendency for morningness predicted increased substance use, and

vice versa). This is interesting given that substance use is typically more commonly reported

among individuals reporting later/evening chronotypes, even independently of delayed sleep

phase disorder (Reid et al., 2012; Tavernier, Munroe & Willoughby, 2015). It may be that this

difference is a result of engaging in substance use recreationally versus to cope with stress (with

participants engaging in substance use to cope with stress). Industriousness also predicted

substance use (i.e., higher industriousness predicted decreased substance use, and vice versa), as

189

seen in previous research (Turiano et al., 2012). Poor sleep quality is also known to relate to

substance use (e.g., Wong, Roberson & Dyson, 2015), however that was not observed here.

Patients’ self-blame decreased across treatment, suggesting patients came to realize or accept

that their cancer diagnosis was not their fault. Denial scores were significantly positively

correlated to self-blame scores across treatment (data not shown). While it is unknown whether

denial was directed at information or emotions in this study, this correlation suggests that while

patients may have denied their feelings and or the information being received, an element of self-

blame still existed. As denial diminished, it seems patients felt less blame. Interestingly, no

significant correlation between self-blame and acceptance exists, but overall the decrease in

patient self-blame suggests subjects accepted that the diagnosis and or resulting stress was not

their fault. When not considering personality, sleep quality was associated with self-blame at

endpoint. It is possible that self-blame led to poor sleep, or sleeplessness led to rumination and

self-blame. When personality was included in the multiple regression model, this relationship no

longer appeared significant, however the significance of sleep quality over the course of

treatment did increase in value. Industriousness was significantly predictive at baseline and

endpoint, however the relationship changed directions. The negative endpoint relationship

reflects previous findings (e.g., O’Brien & DeLongis, 1996). At baseline, it is possible that

before fully processing the situation, individuals high in industriousness wondered how when

they are typically organized, diligent people, they could have developed cancer, resulting in a

positive relationship with self-blame.

6.5.3! In the Moment Coping vs. Coping Behaviours

Daily coping logs did not correlate with one’s use of coping behaviours. This suggests that while

patients report having a memory for their use of various coping behaviours (indicated by their

completion of the Brief COPE), they did not rely on their memory for use of these behaviours

when rating how they felt they were coping in the moment. Given that chronotype does relate to

one’s rating of their in the moment coping, there is some element of rhythmicity related to one’s

sense of their coping ability across the day. While further research is needed to validate this, it is

possible that one’s in the moment rating is based on how they are feeling depending on their

cognitive and emotional regulatory abilities. Comparing one’s in the moment coping ratings

across the day to their performance on an emotional regulatory task across the day could shed

190

light on whether one’s rating of their coping ability is in fact tied to one’s emotional and

cognitive control capacities across the day. Conversely, one’s recall for use of various coping

behaviours may be based on salient memories for behaviours that an individual recalls using. To

verify whether one’s memory for use of particular coping behaviours is correct, it would be

important to have the patient, and possibly an additional rater (such as a family member who is

frequently with the patient) keep a record for one’s use of various behaviours. One’s memory for

the use of various coping behaviours may vary from a patients’ actual use of particular

behaviours. If this is correct, it is also possible that the gaps in one’s memory for their use of

particular coping behaviours might explain why certain coping behaviours showed varied

relationships with chronotype, sleep quality and personality across treatment, due to poor

memory for the use of particular behaviours.

6.5.4! Limitations

While many patients reported having a partner, many partners did not attend appointments and or

chemotherapy treatments regularly or even at all. Of the partners who did attend, many were not

interested in participating in studies, often remarking that they were “not the sick one”. It proved

to be more difficult than expected to recruit caregivers for study participation. It is possible that

the commonly echoed comment “I’m not the sick one” suggests caregivers did not want to incur

attention and take away from their partner. Alternately, it is possible that caregivers often felt

they were already too overwhelmed with the responsibilities of their caregiving goal to take on

another commitment (i.e., study participation). In future, it will be important to recruit a greater

sample size of caregivers to better examine coping (both one’s personal reflection of their own

coping abilities and one’s use of various behaviours to cope) and how it relates to chronotype

and sleep quality.

Participants were divided based on their role – patient or caregiver. Given the nature of breast

cancer, it is most common to recruit female participants. A single male participant was initially

recruited but unable to participate as his disease had progressed to stage 4. Future studies need to

examine coping in both male and female patients and caregivers. This will provide a more clear

understanding of how chronotype and sleep quality, together with one’s personality, influence

the way different genders cope with their role of patient or caregiver. This will allow for a better

191

understanding of who is at risk of developing maladaptive coping strategies and to look for

solutions to prevent this from happening.

In this study, participants rated their own daily coping, use of various coping strategies, and

personality. We recognized that study participants may be biased insofar as they believe that

their answers are socially desirable to those assessing their responses (social desirability bias). In

future, it would be beneficial to obtain these same ratings from an independent observer (e.g.,

family member, friend). In addition, previous research has suggested individuals are not

particularly accurate at rating their emotional behaviour fluctuations across the day (Bellicoso,

2010); an independent observation in some cases, particularly when fluctuations in one’s

emotional abilities run counter to one’s image or sense or of self, might provide a more realistic

description. These discrepancies between how one views themselves versus their actual outward

actions can also be seen for personality. For example, a person might see themselves in a more

favourable light than others around them might, thereby biasing a personality score, and vice

versa. Alternately, an individual might see themselves as consistently emotionally stable across

the day, whereas their outward responses in reaction to experiences and individuals throughout

the day might be different. It would help to compare participants’ responses to how other’s view

their coping and personality. While one might argue that in having caregiver reports on the

patient and the patient’s self-reports it is difficult to identify the ‘truth’, these two ratings could

be beneficial in several ways. For example, if patient and caregiver both provide very different

coping ratings, it is important to understand what each individual is assessing when providing a

rating; it is possible that each individual is focusing on specific, yet different, aspects of

behaviour. Having self-ratings and independent ratings would allow a more detailed picture of a

participant’s coping and personality. Furthermore, understanding caregivers’ perception of how

their patient is coping may shed light on why caregivers believe they must act in certain ways in

their own role.

In multiple regression models, to avoid masking the influence of certain variables, all variables

were included in standard multiple regression analyses. There, some models appear non-

significant while components of the model still appear as statistically significant predictors.

When several variables are included in a model to attempt to predict an outcome variable, it is

possible that too many predictors are contributing to the outcome, therefore diluting the

importance of individual variables. In this case, the model itself is not significant, which prevents

192

against Type 1 error (which would cause a false positive, i.e., incorrect rejection of a null

hypothesis). It is still important to make note of those variables that are significant predictors, as

in smaller models they may play a significant role. Alternately, for behavioural disengagement at

endpoint when considering personality, the model appears significant, but there are no significant

predictors. This is possible when predictor variables are highly correlated (multicollinearity).

While the tolerance values for this model were all above .100, they were still lower than

tolerance values found for other models in this study. It is possible that while tolerance values

were not below an acceptable level, the fact that they were low combined with having several

predictors in the model diluted the overall significance any one predictor could hold on its own.

It should be noted that tolerance values for all three models predicting behavioural

disengagement across treatment, while not below .100, were all low.

6.6! Conclusion

Chronotype, sleep quality, and participant role each contribute in some degree to the use of

various coping behaviours across treatment. Most interesting is that chronotype and sleep quality

did not simultaneously influence coping behaviours at the three time points across treatment.

While chronotype and sleep quality do not contribute heavily to the use of all coping strategies,

there appears to be a clear pattern that chronotype affects use of particular coping strategies at

baseline, while sleep quality distinctly only significantly contributes to various coping

behaviours at endpoint. Chronotype did not significantly influence very many behaviours, yet its

consistent role predicting self-distraction at baseline, regardless of one’s personality, should be

noted. It also had a relatively consistent influence on the use of religious coping. Self-distraction

is considered an engagement or accommodative type coping strategy; religion is not necessarily

an engagement or disengagement strategy, and instead can be seen in either way, depending if

someone turns to their faith to seek strength to face the problem, or relies on their faith to ignore

the issue. Coping behaviours relating to sleep are engagement type strategies (Carver & Connor-

Smith, 2010). It is possible that either people are thinking how to best deal with the stressor,

which keeps them awake resulting in poor sleep, or alternatively, that people are unable to sleep

and instead spend their time ruminating and partaking in these engagement style coping

behaviours.

193

While various personality factors contributed to using particular coping behaviours, openness

had the most consistent predictive value, positively relating to engagement based behaviours, and

negatively contributing to disengagement style behaviours. Industriousness also significantly

influenced coping, but less consistently than openness. It followed a similar pattern for

predicting engagement and disengagement coping as openness. It is possible that including

personality did not allow the roles of chronotype and sleep quality to be significant. Future

studies should only include significant predictors in the model to reexamine the relationship.

The significant relationship between one’s chronotype and their self-rating of coping ability

across the day suggests people were aware of fluctuations in their ability across the day, and may

have accepted these changes as part of their overall preference for performing tasks at one time

or another. The lack of correlation between sleep quality and coping log scores across treatment

suggests patients did not consider their sleep debt. It is possible that patients may have accepted

that a sleep debt would be experienced during cancer treatment, and therefore did not factor in its

outputs, (e.g., fatigue) when rating their coping. This is particularly interesting considering the

distinction between chronotype contributing to coping behaviours earlier in treatment, and sleep

quality contributing to use of behaviours at endpoint. While people may believe their daily

schedule and functionality is unaffected by sleep debt, it appears that as treatment progresses and

poor sleep quality persists thereby increasing sleep debt, people engage in certain behaviours,

possibly out of habit, instead of actively thinking of using them at one time of day or another.

This distinction suggests it is important to use self-rated coping measures, and behaviour

assessments, to distinguish between what people believe versus what they are actually doing.

These results should be considered to help patients and caregivers follow schedules that optimize

their performance based on chronotype and cognitive function especially for stressful situations,

and to maximize the possibility of having a full night sleep based on their body’s rhythm.

194

Chapter 7!

! Conclusion The main objective of this dissertation was to assess the degree to which coping is influenced by

chronotype and sleep quality in a prevalent model of chronic stress. In breast cancer, we

examined both patients as well as caregivers and oncology staff, as the issue of coping with

chronic disease depends on the attitudes and interactions of the numerous parties that are

involved with the patient. Coping was examined in several ways, both retrospectively and in the

moment. Among patients and spousal caregivers, coping was assessed using one’s recall of their

ability to cope across the day (including feelings of how well they coped, ability to show

compassion to their ill partner or caregiver, and their need for time and space to themselves), and

based on their use of various coping behaviours over time. Patient coping was also assessed in

the moment across the day. Among oncology staff, coping was assessed based on levels of

compassion satisfaction (CS), compassion fatigue (CF), and burnout. CS, CF, and burnout were

used as coping measures based on the idea that one’s ability to manage the specific and

potentially taxing external and internal demands of their job would produce various levels of

each of these domains (with inability for or low coping leading to low CS and high CF and

burnout, and vice versa). In addition to chronotype and sleep quality, personality was also

examined as a secondary factor that would influence one’s ability to cope with chronic stress.

In this population, chronotype and sleep quality influenced various measures of coping when

faced with chronic stress. Chronotype and sleep quality were not significantly correlated except

specifically in patients without a caregiver involved in the study. Sleep timing and chronotype

are correlated in a world-wide study, and it is possible that the size of the sample taken in our

study was too small to detect this. Furthermore, the inclusion of both cancer patients and their

caregivers in one group may have introduced considerably more variance in sleep habits than

would be found in a general population.

This study examined individuals predominantly working or being treated at Sunnybrook Health

Science’s Centre (SH) in the Odette Cancer Centre; a smaller population of nurses from Princess

Margaret Hospital (PMH) were also included. These are both teaching hospitals affiliated with

the University of Toronto. Although SH and PMH provide care to individuals from all over the

195

greater Toronto area, there are other major healthcare centers across the city. Therefore, the

target subject pool as well as the hospital environment and working conditions, may have been

limited. When nurses at SH and PMH were compared, they showed no significant differences on

the domains being examined. Nonetheless, it is possible that differences exist in populations that

are either working or receiving cancer treatment, in different facilities, cities, and or countries.

7.1! Chronotype

Chronotype was related to recalled coping assessments among patients and oncology staff,

however not significantly among caregivers. Patient reports also indicated that one’s chronotype

was linked with their in the moment coping ratings. Among patients, on recalled and in the

moment coping, people reported reaching their optimal coping ability in sequential order, with M

types showing peak times earlier than N types, who were earlier than E types. Among M types,

distinctions were also reported with definitely M types reaching their peak performance before

moderately M types. Recalled fatigue levels did not correlated with Horne-Östberg Morningness

Eveningness Questionnaire (MEQ) scores, however among patients it appeared that E types

reached their peak fatigue level earlier than M and N types (who did not differ significantly in

their peak times), suggesting E types do feel more fatigued earlier in the day, which may

subsequently delay their peak performance time for other coping behaviours. In the moment

coping also showed a chronotype dependent relationship, indicating that patients do realize

changes in their ability across the day. Unlike previous studies (e.g., Bellicoso, 2010) where

memory for emotional regulatory behaviours was poorly correlated to chronotype, it is possible

that continually facing the same chronic stressor on a daily basis has allowed people to be aware

of their behavioural abilities in the moment and retrospectively upon recall.

Among oncology staff, chronotype was also related to coping as measured by burnout (using the

Copenhagen Burnout Inventory, CBI), and Professional Quality of Life (including CS, and CF -

consisting of a different burnout measure, and secondary traumatic stress, STS). Among nurses

only, even when considering other factors such as work stressfulness and job satisfaction,

chronotype was a significant predictor of one’s burnout score; M types following dayshift

schedules showed lower burnout than N and E types on the same work schedule. However, when

burnout and secondary traumatic stress (STS) were examined using the ProQoL among a mixed

196

group of oncology caregiving staff, chronotype did not contribute to the onset of burnout, or

STS. When considered alone, at the univariate level chronotype did relate to coping as measured

by CS, suggesting staff with earlier chronotypes working dayshift schedules more likely coped

better and were able to enjoy the positive outcomes of providing care. However, when other

factors such as sleep quality, job satisfaction, and personality were included in a model with

chronotype as predictors of CS, the predictive role of chronotype was no longer significant. None

the less, the small role of chronotype in helping people obtain positive feelings and benefits from

providing care should not be dismissed. It is possible that in a larger population, or among

specific subgroups of oncology staff, chronotype plays a stronger role in mediating one’s level of

CS.

Among spousal caregivers, chronotype was not strongly linked to one’s recalled coping ability.

Anecdotally many caregivers would report that they were ‘not the sick one’ and ‘were doing

fine’, often saying suggesting they saw how sick their partner was and comparatively speaking

were doing fine. While caregivers did report a chronotype dependent need for time and space to

themselves at endpoint, there were no other chronotype dependent differences in recalled coping

performance across the day. It is possible that male caregivers do not pay strong attention to

performance fluctuations across the day and therefore do not form strong memories. It is also

possible however that caregivers typically felt they had to be “on duty” and perform optimally

around the clock when caring for their partner, leading them to believe their performance did not

fluctuate across the day. While it is possible that their performance actually did not fluctuate, this

is highly unlikely, given previous research indicating that even regardless of one’s awareness,

performance does fluctuate. Letting caregiver’s know that it is acceptable to take a break for

themselves or to have fluctuations in their caregiving abilities across the day may also reduce

stress and ameliorate their poor sleep quality.

When considering recalled coping behaviours, chronotype appeared to be a more prominent

predictor earlier in treatment (baseline and midpoint), but not often at endpoint. While

chronotype showed a fairly consistent relationship with performance fluctuations and one’s

ratings of their in the moment coping among patients, chronotype did not show a strong

relationship with use of specific coping behaviours among patients and caregivers. When asked

about how they were coping at a particular time of day, in the moment or retrospectively,

participants could consider a broad overview of how they were coping, instead of considering a

197

specific coping behaviour. Reflecting on one’s overall coping may not necessarily cause an

individual to think of the specific behaviours they are engaging in – it is possible that there are

too many different behaviours people use to face a chronic stressor, and instead people may store

a summative memory of their coping overall across the day. Future studies should assess what

factors people consider when rating their overall coping, and whether the individual contributors

relate differently to chronotype.

Overall, the role of chronotype in predicting coping should be considered in future studies.

Awareness for the fact that people are not always able to give their best coping performance for

chronic stressors is important in reducing employee burnout and providing additional help when

necessary to patients and caregivers before they may engage in destructive coping behaviours or

develop problems such as depression. Among caregivers, reducing CF and or increasing CS will

enhance the overall employee experience, and likely lead to more productive staff, and lower

turnover. Positive patient coping during cancer leads to better coping during survival (e.g., Jim,

Richardson, Golden-Kreutz & Andersen, 2006). Future studies should further examine the long-

term influence of positive coping with a chronic stressor such as cancer, together with an

understanding and acceptance that coping behaviour does fluctuate as predictors of greater

emotional recovery upon treatment completion.

7.2! Sleep Quality

Sleep quality was related to various recalled coping assessments among patients,

caregivers and oncology staff, but did not correlate strongly with in the moment measures of

coping. It is possible that in the moment coping was not strongly linked with sleep quality

because society often disregards the importance of sleep, believing and reporting that they can

function well on little sleep. Future studies examining sleep wake times as opposed to sleep

quality may discover different results, possibly showing peak coping times at specific distances

from one’s wake up or sleep time. The results of this study suggest patients disregard the

influence of their sleep quality (which was typically reported as poor) when immediately

examining their behaviour. However, long term, people may store a memory of their chronic

poor sleep, and relate that with fluctuations in their performance.

198

Among oncology nurses, and oncology staff as a collective group, sleep quality significantly

predicted burnout, both with and without the inclusion of work stressfulness and or job

satisfaction. Among the collective group of staff, when not considering job satisfaction, sleep

quality also significantly predicted the onset of CS, indicating that sleep quality influences

various aspects of coping (i.e., mediating stress perceptions leading to burnout, and helping to

find a positive effect of providing care in a chronically stressful environment). While poor sleep

quality can reduce CS and or increase burnout, it is important to note that several steps can be

taken to remedy poor sleep quality. Overall, good sleep quality has the potential to serve as a

protective factor influencing positive coping when facing chronic stress.

The intergroup differences between patients with and without a caregiver partaking in the study

are interesting to note and should be examined in future. Patients and caregivers who partook in

the study together did not show a significant correlation between their chronotype score and

sleep quality. Instead, patients without a caregiver involved in the study (i.e., either they had no

life partner, or their partner – who may or may not have been their primary caregiver and or may

or may not have attended treatments with them – did not enroll in the study), showed a highly

significant relationship between sleep quality and chronotype. Their results suggest that later

chronotypes had worse sleep quality, similar to other research findings (e.g., Merikanto,

Paavonen, Saarenpää-Heikkilä, Paunio & Partonen, 2017; Yun, Ahn, Jeong, Joo & Choi, 2015).

Instead, patients and caregivers involved in the study together showed no significant relationship

between chronotype and sleep quality, and suggested M types would likely have worse sleep

quality than E types, which is counter to many published findings, as well as the correlation

direction reported among nurses and oncology staff as a whole in this study. These results

suggest patients and caregivers participating together may be interrupting one another’s sleep,

possibly ruminating together over worries and fears as both are tired and cognitively depleted.

While these results do not suggest that it is necessarily better to not have a caregiver, it is

possible that having some space to oneself is beneficial.

Patient and caregiver sleep quality did not differ significantly by chronotype or across treatment.

However, patients’ and caregivers’ sleep quality did predict use of certain engagement style

coping behaviours, but only at endpoint. While sleep quality was reported as poor among both

groups from treatment outset, they were likely less cognitively depleted and sleep quality did not

influence coping behaviour choice or use earlier on. Chronic poor sleep quality reduces cognitive

199

abilities, particularly attentional and executive control (e.g., Alhola & Polo-Kantola, 2007;

Stavitsky, Neargarder, Bogdanova, McNamara & Cronin-Golomb, 2012). The results from our

study support this previous work, with sleep quality significantly influencing highly cognitive

coping behaviours such as self-distraction, active coping, planning, venting and self-blame at

endpoint. By endpoint, patients and caregivers alike have reported chronic poor sleep quality,

which begins to impact their coping. Future studies should examine sleep quality’s influence on

coping behaviours into the radiation therapy portion of treatment to assess whether the effect

continues. Research should also examine the specific components of one’s sleep that are

contributing to poor sleep quality, and whether particular components are influencing coping.

7.3! Personality

Like chronotype, personality is considered a relatively stable trait, that has the ability to change

over time, but is typically stable for extended periods. While the main factors for assessment in

this study were chronotype and sleep quality, we believed personality should also be examined

as it is present in all individuals and would undoubtedly influence certain coping decisions or

styles. While personality cannot be changed, it can be screened to identify individuals who may

have greater difficulty coping or those who may do very well on their own without additional

help. In our findings, when personality was included in a model it typically diminished the

effects of chronotype and sleep quality. However, those instances where chronotype and sleep

quality persisted in exerting an influence on coping even when considering personality are

important to note as it suggests a strong predictive value for chronotype and sleep quality. In

those cases where chronotype and sleep quality were no longer significant once personality was

included, the roles of chronotype and sleep quality should not be dismissed – it is possible that in

a larger population their roles might become more pronounced.

Among the total group of oncology staff included in this study, personality traits were

consistently predictive of CS, burnout, and STS. Emotional stability consistently predicted CF,

with and without the inclusion of job satisfaction, while agreeableness consistently predicted CS.

Burnout was also consistently predicted by traits shared with each CS and STS, suggesting that

burnout is possibly influenced by one’s positive and negative take on their work experiences and

stresses. When job satisfaction was no longer included as a predictive variable for ProQoL, while

200

the influence of sleep quality increased, not all personality variables increased their contribution

to CS, burnout, and STS. Those variables that were predictive with and without JSS increased in

their predictive value, however without the JSS, some variables lost their predictive significance.

These variables may have been related to job satisfaction and without it in the model were no

longer significantly predictive. Among oncology staff, openness was significantly predictive of

CS and burnout when job satisfaction was included in the model (i.e., high openness led to

increased CS and diminished burnout, and vice versa), yet was replaced by agreeableness as

dually predictive when JSS was not included. Emotional stability was predictive of burnout and

STS regardless of the inclusion of JSS, suggesting one’s level of emotional stability should be

screened when hiring individuals to work in this field in order to prevent CF.

The most commonly predictive personality trait among patients and caregivers was openness,

which unlike chronotype or sleep quality exhibited predictive value at various points across

treatment. When personality was not included in the model, sleep quality was predictive of

greater number of coping behaviours; upon including personality, sleep quality was only

predictive of self-distraction and planning behaviours. Openness was significantly predictive of

engagement style coping behaviours, suggesting a higher degree of openness predisposes an

individual to use more behaviours aimed at diminishing the stress of being a patient or caregiver.

In a larger group of participants, openness may demonstrate a more consistent predictive role

across treatment for the use of particular coping strategies.

7.4! Future Directions

Many of the patients who participated in this study were able to take time off work, often with

continued pay and benefits. This is not always possible for many individuals, and may vary

depending on the types of jobs held by patients. While chronotype, sleep quality, and personality

were each found to play significant roles influencing coping (both recalled and in the moment)

among patients, caregivers, and oncology staff, it is possible that the influence of these factors

may show greater variation depending on the demographics of the population studied. Given

cancer’s ability to affect anyone, regardless of age, race, gender, ethnicity, or socioeconomic

status, individuals from different walks of life will have their own mediating factors that

influence coping. For example, inner city hospitals (e.g., St. Michael’s Hospital) may show a

201

varied patient population in terms of socioeconomic status, where some patients may come from

affluent households, while others may come from low income areas, or even be homeless.

Individuals in each situation will have their own personal stresses that contribute to the overall

stress they feel from the situation. Future studies must examine the influence of chronotype and

sleep quality together with a broader description of one’s living situation and socioeconomic

status in order to provide a complete picture of results that is more generalized to the entire

population. In future, making the study available on an app or on computer with reminders to

participants for when to complete specific components of the study would reduce the time

constraint on the researcher and allow for study implementation at a greater number of hospitals.

Implementing the study at a greater number of hospitals would allow for a broader sample of the

population and how they cope.

Sunnybrook Health Sciences Centre and Princess Margaret Hospital are teaching hospitals,

servicing large populations. While nurses at the two sites did not show significant differences on

the domains examined, future research should examine oncology staff working at a range of

hospitals, including teaching and non-teaching hospitals, in and out of Toronto. It is important to

understand whether all oncology staff are feeling similar levels of stress, and understanding how

chronotype and sleep quality are relating to their coping as measured by ratings of their ProQoL.

Finally, staff, patients, and familial caregivers dealing with the stresses of other chronic illnesses

need to be examined to understand how stress and burnout related to chronotype, sleep quality,

and personality. For example, heart disease is also a chronic illness, with potentially fatal side

effects, as is liver or kidney disease. These chronic illnesses can also require regular hospital

visits, medication, and severe side effects. Working with such a population can be stressful for

staff, however it remains to be understood whether this would be the similar type of stress faced

by oncology staff, and whether their coping is similar. Patient and caregiver coping for such

illnesses – in the moment and recalled – is also susceptible to the influence of chronotype and

sleep quality. Overall, it is important to understand whether the results from the current study can

be generalized to other populations facing chronic health issues. A better understanding of

whether people faced with chronic illness cope similarly or not would help develop necessary

tools to ameliorate coping among staff, patients, and caregivers, and provide outside help when

necessary.

202

Bibliography

Achermann, P. (2004). The two-process model of sleep regulation revisited. Aviation, Space, and Environmental Medicine, 75, A37-A43. Adan, A. & Natale, V. (2002). Gender differences in morninness-eveningnes preference. Chronobiology International, 19, 709-720. Afshar, H., Roohafza, H. R., Keshteli, A. H., Mazaheri, M., Feizi, A., & Peyman, A. (2015). The association of personality traits and coping styles according to stress level. Journal of Research in Medical Sciences, 20, 353-358. Alhola, P. & Polo-Kantola, P. (2007). Sleep deprivation: Impact on cognitive performance. Neuropsychiatric Disease and Treatment, 3, 553-567. Ancoli-Israel, S., Liu, L., Rissling, M., Natarajan, L., Neikrug, A. B., Palmer, B. W., … Maglione, J. (2014). Sleep, fatigue, depression and circadian activity rhythms in women with breast cancer before and after treatment: A 1-year longitudinal study. Supportive Care in Cancer, 22, 2535–2545. doi:10.1007/s00520-014-2204-5 Ancoli-Israel, S., Moore, P. J., & Jones, V. (2001). The relationship between fatigue and sleep in cancer patients: A review. European Journal of Cancer Care, 10, 245-255. Antoniou, A., Pharoah, P. D. P., Narod, D., Risch, H. A., Eyfjord, J. E., Hopper, J. L., … Easton, D. F. (2003). Average risks of breast and ovarian cancer associated BRCA1 or BRCA2 mutations detected in case series unselected for family history: A combined analysis of 22 studies. American Journal of Human Genetics, 72, 1117-1130. doi: 10.1086/375033 Aoyagi, T., Terracina, K. P., Raza, A., Matsubara, H., & Takabe, K. (2015). Cancer cachexia, mechanism and treatment. World Journal of Gastrointestinal Oncology, 7, 17-29. doi: 10.4251/wjgo.v7.i4.17 Araghi, M.H., Jagielski, A., Neira, I., Brown, A., Higgs, S., Thomas, G.N., & Taheri, S. (2013). The complex associations among sleep quality, anxiety-depression, and quality of life in patients with extreme obesity. Sleep, 36, 1859-1865. doi: http://dx.doi.org/10.5665/sleep.3216 Arendt, J. (2006). Melatonin and human rhythms. Chronobiology International, 23, 21-37. Arnsten, A. F. T. (2009). Stress signaling pathways that impair prefrontal cortex structure and function. National Review of Neuroscience, 10, 410-422. doi: 10.1038/nrn2648 Armitage, R., Trivedi, M., Hoffmann, R., & Rush, A., J. (1997). Relationship between objective and subjective sleep measures in depressed patients and healthy controls. Depression and Anxiety, 5, 97-102.

203

Arpino, G., Weiss, H., Lee, A. V., Schiff, R., De Placido, S., Osborne, C. K, & Elledge, R. M. (2005). Estrogen receptor-positive, progesterone receptor-negative breast cancer: Association with growth factor receptor expression and tamoxifen resistance. Journal of the National Cancer Institute, 97 (17), 1254-1261. doi: 10.1093/jnci/dji249 Aslan, O., Sanisoglu, Y., Akyol, M., & Yetkin, S. (2009). Quality of sleep in Turkish family caregivers of cancer patients. Cancer Nursing, 32, 370-377. doi: 10.1097/NCC.0b013e3181a87fb9 Atkinson, G. & Reilly, T. (1996). Circadian variations in sports performance. Sports Medicine, 21, 292-312.

Austenfeld, J. L., & Stanton, A. L. (2004). Coping through emotional approach: A new look at emotion, coping, and health-related outcomes. Journal of Personality, 7, 1335-1363. doi: 10.1111/j.1467-6494.2004.00299.x Bahk, Y.-C., Han, E., & Lee, S.-H. (2014). Biological rhythm differences and suicidal ideation in patients with major depressive disorder. Journal of Affective Disorders, 168, 294- 297. doi: 10.1016/j.jad.2014.07.001 Bakker, A. B., Van der Zee, K. I., Lewig, K. A., & Dollard, M. F. (2006). The relationship between the Big Five personality factors and burnout: A study among volunteer counselors. Journal of Social Psychology. 146, 31–50. doi: 10.3200/SOCP.146.1.31-50 Banks, S., & Dinges, D. F. (2007). Behavioral and physiological consequences of sleep restriction. Journal of Clinical Sleep Medicine, 3, 519–28. Barre, V. P., Padmaja, G., Saxena, R. K., & Rana, S. (2015). Impact of medical intervention on stress and quality of life in patients with cancer. Indian Journal of Palliative Care, 21 (2), 203-208. doi: 10.4103/0973-1075.156503 Bartley, C. E. & Roesch, S. C. (2011). Coping and daily stress: The role of conscientiousness. Personality and Individual Differences, 50, 79–83. doi: 10.1016/j.paid.2010.08.027 Baum, A. (1990). Stress, intrusive imagery, and chronic distress. Health Psychology, 9 (6), 653- 675. Baum, A., Friedman, A. L., & Zakowski, S. G. (1997). Stress and genetic testing for disease risk. Health Psychology, 16, 8-19. Beach, S. R., Schulz, R., Williamson, G. M., Miller, S., Weiner, M. F., & Lance, C. E. (2005). Risk factors for potentially harmful informal caregiver behavior. 53, 255-261. doi: 10.1111/j.1532-5415.2005.53111.x Becker, J. B. & Hu, M. (2008). Sex differences in drug abuse. Frontiers in Neuroendocrinology, 29, 36-47. doi: 10.1016/j.yfrne.2007.07.003

204

Bellicoso, D. (2010). A task’s cognitive demands influence self-reported performance variances throughout the day (Master’s Thesis). Available from University of Toronto Research Repository. https://tspace.library.utoronto.ca/bitstream/1807/25428/1/Bellicoso_Daniela_201011_M A_thesis.pdf Bellicoso, D., Ralph, M. R., Trudeau, M. E. (2014) Burnout among oncology nurses: The influence of chronotype and sleep quality. Journal of Nursing Education and Practice, 4, 80-89. doi: https://doi.org/10.5430/jnep.v4n8p80 Ben-Zur, H. & Michael, K. (2007). Burnout, social support, and coping at work among social workers, psychologists, and nurses: The role of challenge/control appraisals. Social Work in Health Care, 45, 63-82. doi: 10.1300/J010v45n04_04 Berger, A. M., Parker, K. P., Young-McCaughan, S., Mallory, G. A., Barsevick, A., Beck, S. L., … Hall, M. (2005). Sleep/Wake disturbances in people with cancer and their caregivers: State of the science. Oncology Nursing Forum, 32, E98-E126. doi: 10.1188/05.ONF.E98-E126 Bersagliere A, Raduazzo ID, Nardi S, Gatta A, Amodio P, Achermann P, & Montagnese, S. (2012). Induced hyperammonemia may compromise the ability to generate restful sleep in patients with cirrhosis. Hepatology, 55, 869-878. doi: 10.1002/hep.24741 Biss, R. & Hasher, L. (2012). Happy as a lark: Morning-type younger and older adults are higher in positive affect. Emotion, 12, 437-441. doi: 10.1037/a0027071 Blatter, K. & Cajochen, C. (2007). Circadian rhythms in cognitive performance: Methodological constraints, protocols, and theoretical underpinnings. Physiology and Behavior, 90, 196-208. doi: 10.1016/j.physbeh.2006.09.009 Bonnet, M. H. & Arand, D. L. (1995). We are chronically sleep deprived. Sleep, 18, 908-911. Bower, J. E. (2008). Behavioral symptoms in breast cancer patients and survivors: Fatigue, insomnia, depression, and cognitive disturbance. Journal of Clinical Oncology, 26, 768- 777. doi: 10.1200/JCO.2007.14.3248 Brand, S., Beck, J., Hatzinger, M., Harbaugh, A., Ruch, W., Holsboer-Trachsler, E. (2010). Associations between satisfaction with life, burnout-related emotional and physical exhaustion, and sleep complaints. The World Journal of Biological Psychiatry, 11, 744- 754. doi: 10.3109/15622971003624205 Brown, S. A., Kunz, D., Dumas, A., Westermark, P. O., Vanselow, K., Tilmann-Wahnschaffe, … Kramer, A. (2008). Molecular insights into human daily behavior. PNAS, 105, 1602- 1607. doi: 10.1073/pnas.0707772105 Brown, F. M., Neft, E. E., & LaJambe, C. M. (2008). Collegiate rowing crew performance varies by morningness-eveningness. Journal of Strength and Conditioning Research, 22, 1894- 1900. doi: 10.1519/JSC.0b013e318187534c

205

Bultz, B. D. & Carlson, L. E. (2006). Emotional distress: The sixth vital sign – future directions in cancer care. Psycho-Oncology, 15, 93-95. doi: 10.1002/pon.1022 Burgess, H. J. & Fogg, L. F. (2008). Individual differences in the amount and timing of salivary melatonin secretion. PLoS One, 3, e3055. doi: 10.1371/journal.pone.0003055 Burgess, L., Irvine, F., & Wallymahmed, A. (2010). Personality, stress and coping in intensive care nurses: A descriptive exploratory study. Nursing in Critical Care, 15, 129-140. doi: 10.1111/j.1478-5153.2009.00384.x Burker, E. J., Evon, D., Loiselle, M. M., Finkel, J., & Mill, M. (2005). Planning helps, behavioral disengagement does not: coping and depression in the spouses of heart transplant candidates. Clinical Transplant, 19, 653-658. doi: 10.1111/j.1399-0012.2005.00390.x Buysse, D. J., Reynolds III, C. F., Monk, T. H., Berman, S. R. & Kupfer, D. J. (1989). The Pittsburgh sleep quality index: A new instrument for psychiatric practice and research. Psychiatry Research, 28, 193-213. Byun, E., Gay C. L., & Lee, K. A. (2016). Sleep, fatigue and problems with cognitive function in adults living with HIV. The Journal of the Association of Nurses in AIDS Care, 27, 5-16. doi: 10.1016/j.jana.2015.10.002 Canadian Breast Cancer Foundation (2016). Breast Cancer in Canada, 2016. Retrieved at: http://www.cbcf.org/central/AboutBreastCancerMain/FactsStats/Pages/Breast-Cancer- Canada.aspx Canadian Cancer Society’s Advisory Committee on Cancer Statistics. Canadian Cancer Statistics 2016. Toronto, ON: Canadian Cancer Society; 2016a. Retrieved at: http://www.cancer.ca/~/media/cancer.ca/CW/cancer%20information/cancer%20101/Cana dian%20cancer%20statistics/Canadian-Cancer-Statistics-2016-EN.pdf?la=en Canadian Cancer Society (2016b). Survival statistics for breast cancer. Retrieved at: http://www.cancer.ca/en/cancer-information/cancer-type/breast/prognosis-and- survival/survival-statistics/?region=on Carlson, L. E. & Bultz, B. D. (2004). Efficacy and medical cost offset of psychosocial interventions in cancer care: Making the case for economic analyses. Psycho- Oncology, 13, 837-849. doi: 10.1002/pon.832 Carpenter, J. S. & Andrykowski, M. A. (1998). Psychometric evaluation of the Pittsburgh sleep quality index. Journal of Psychosomatic Research, 45, 5-13. Carr, D., Goudas, L., Lawrence, D., Pirl, W., Lau, J., DeVine, D., … & Miller, K. (July 2002). Management of Cancer Symptoms: Pain, Depression, and Fatigue. Evidence Report/Technology Assessment No. 61 (Prepared by the New England Medical Center Evidence-based Practice Center under Contract No 290-97-0019). AHRQ Publication No. 02-E032. Rockville, MD: Agency for Healthcare Research and Quality.

206

Carter, P. A., & Acton, G. J. (2006). Personality and coping: Predictors of depression and sleep problems among caregivers of individuals who have cancer. Journal of Gerontological Nursing, 32, 45-53. Carver, C. S. (1997). You want to measure coping but your protocol’s too long: Consider the Brief-COPE. International Journal of Behavioral Medicine, 41, 92-100. Carver, C. S. (2005). Enhancing adaptation during treatment and the role of individual differences. Cancer, 104 (11 Suppl.), 2602–2607. doi: 10.1002/cncr.21247 Carver, C. S. & Connor-Smith, J. (2010). Personality and coping. Annual Review of Psychology, 61, 679-704. doi: 10.1146/annurev.psych.093008.100352 Carver, C. S., Pozo, C., Harris, S. D., Noriega, V., Scheier, M. F., Robinson, D. S., … Clark, K. C. (1993). How coping mediates the effect of optimism on distress: A study of women with early stage breast cancer. Journal of Personality and Social Psychology, 65, 375- 390. Carver, C. S., Scheier, M. F., & Weintraub, J. K. (1989). Assessing coping strategies: A theoretically based approach. Journal of Personality and Social Psychology, 56, 267-283. Cavallera, G. M., Gatto, M., & Boari, G. (2014). Personality, cognitive styles, and morningness- eveningness disposition in a sample of Yoga trainees. Medical Science Monitor, 13, 238- 246. doi: 10.12659/MSM.889030 Chaudhury, D. & Colwell, C. S. (2002). Circadian modulation of learning and memory in fear conditioned mice. Behavioural Brain Research, 133, 95-108. Chen, S. & Parmigiani, G. (2007). Meta-analysis of BRCA1 and BRCA2 Penetrance. Journal of Clinical Oncology, 25, 1329-1333. doi: 10.1200/JCO.2006.09.1066 Chen, S.M. & McMurray, A. (2001). “Burnout” in intensive care nurses. Journal of Nursing Research, 9, 152-164. Chesney, M., Neilands, T. B., Chambers, D. B., Taylor, J. M. & Folkman, S. (2006). A validity and reliability study of the coping self-efficacy scale. British Journal of Health Psychology, 11, 421-437. doi: 10.1348/135910705X53155 Choi, J., Yoon, I., Kim, H., Chung, S., & Yoo, H. J. (2010). Differences between objective and subjective sleep measures in children with attention deficit hyperactivity disorder. Journal of Clinical Sleep Medicine, 6, 589-595. Christie, W. & Moore, C. (2005). The impact of humor on patients with cancer. Clinical Journal of Oncology Nursing, 9, 211-218. doi: 10.1188/05.CJON.211-218 Clark, L. A., Watson, D., & Leeka, J. (1989). Diurnal variation in the positive affects. Motivation and Emotion, 13, 205-234.

207

Clough-Gorr, K. M., Ganz, P. A., & Silliman, R. A. (2007). Older breast cancer survivors: Factors associated with change in emotional well-being. Journal of Clinical Oncology, 25, 1334-1340. doi: 10.1200/JCO.2006.09.8665 Cohen, D. A., Wang, W., Wyatt, J. K., Kronauer, R. E., Dijk, D. J., Czeisler, C. A., & Klerman, E. B. (2010). Uncovering residual effects of chronic sleep loss on human performance. Science Translational Medicine, 2, 14ra3. doi: 10.1126/scitranslmed.3000458 Connor-Smith, J. & Flachsbart, C. (2007). Relations between personality and coping: A meta- analysis. Journal of Personality and Social Psychology, 93, 1080-1107. doi: 10.1037/0022-3514.93.6.1080 Costa-Martins, J.M., Moura-Ramos, M., Cascais, M. J., da Silva, C. F., Costa-Martins, H., Pereira, M., Coelho, R., Tavares, J. (2016). Adult attachment style and cortisol responses in women in late pregnancy. BMC Psychology, 4, 1. doi: 10.1186/s40359-016-0105-8 Craig, C. D. & Sprang, G. (2010). Compassion satisfaction, compassion fatigue, and burnout in a national sample of trauma treatment therapists. Anxiety, Stress & Coping, 23 (3), 319- 339. doi: 10.1080/10615800903085818 Cubrilo-Turek, M., Urek, R., & Turek, S. (2006). Burnout syndrome – assessment of a stressful job among intensive care staff. Collegium Antropologicum, 30, 131-135. Davies, K. A., Macfarlane, G. J., Nicholl, B. I., Dickens, C., Morriss, R., Ray, D. & McBeth, J. (2008). Restorative sleep predicts the resolution of chronic widespread pain: Results from the EPIFUND study. Rheumatology, 47, 1809–1813. doi: 10.1093/rheumatology/ken389 Davis-Ali, S., Chesler, M. A., & Chesney, B. K. (1993). Recognizing cancer as a family disease: Worries and support reported by patients and spouses. Social Work in Health Care, 19, 45-65. Demytteraera, K., Bonte, L., Gheldof, M., Vervaeke, M., Meuleman, C., Vanderschuerem, D., & D’Hooghe, T. (1998). Coping style and depression level influence outcome in in vitro fertilization. Fertility and Stress, 69, 1026-1033. doi: https://doi.org/10.1016/S0015- 0282(98)00089-2 Dent, R., Trudeau, M., Pritchard, K. I., Hanna, W. M., Kahn, H. K., Sawka, C. A., … Narod, S. A. (2007). Triple negative breast cancer: Clinical features and patterns of recurrence. Clinical Cancer Research, 13 (15), 4429-4434. doi: 10.1158/1078-0432.CCR-06-3045 DeYoung, C. G., Quilty, L. C., Peterson, J. B. (2007). Between facets and domains: 10 aspects of the Big Five. Journal of Personality and Social Psychology, 93, 880–96. doi: 10.1037/0022-3514.93.5.880 Di Milia, L., Adan, A., Natale, V., & Randler, C. (2013). Reviewing the psychometric properties of contemporary circadian typology measures. Chronobiology International, 30, 1261- 1271. doi: 10.3109/07420528.2013.817415

208

Dogru, S., Dogru, H. Y., Tapar, T., Karaman, S., Arici, S., Ozsoy, A. Z., … Karakis, A. (2016). Do urgent caesarian sections have a circadian rhythm? Turkish Journal of Anesthesiology and Reanimation, 44, 124-7. doi: 10.5152/TJAR.2016.57614 Drury, V., Francis, K., & Aoun, S. (2014). Compassion satisfaction, compassion fatigue, anxiety, depression and stress in registered nurses in Australia: Phase 2 results. Journal of Nursing Management, 22, 519-531. Duggan, K. A., Friedman, H. A., McDevitt, E. A., & Mednick, S.C. (2014). Personality and healthy sleep: The importance of conscientiousness and neuroticism. PLoS ONE, 9, e90628. doi: doi: 10.1371/journal.pone.0090628 Dunfield, K. (2014). A construct divided: Prosocial behavior as helping and comforting subtypes. Frontiers in Psychology, 5, 1-13. doi: 10.3389/fpsyg.2014.00958 Dunn, K. S. & Horgas, A. L. (2004). Religious and nonreligious coping in older adults experiencing chronic pain. Pain Management Nursing, 5, 19-28. doi:10.1016/S1524- 9042(03)00070-5 Dworak, McCarley, Kim, Kalinchuk & Basheer, 2010). Sleep and brain energy levels: ATP changes during sleep. The Journal of Neuroscience, 30, 9007-9016. doi: 10.1523/JNEUROSCI.1423-10.2010 Eddy, R. (2005). Sleep deprivation among physicians. BC Medical Journal, 47, 176-180. Edge, S. B., Byrd, D. R., Compton, C. C., Fritz, A. G., Charlotte, F. I., & Trotti, A. (Eds.) (2010). Breast. In AJCC Cancer Staging Handbook, 7th Ed., pp. 347-376. New York, NY: Spring. El-Bar, N., Levy, A., Wald, H. S., & Biderman, A. (2013). Compassion fatigue, burnout and compassion satisfaction among family physicians in the Negev area – a cross sectional study. Isreal Journal of Health Policy, 15, 1. doi: 10.1186/2045-4015-2-31. Engstron, C. A., Strohl, R. A., Rose., L., Lweandowski, L. & Stefanek, M. E. (1999). Sleep alterations in cancer patients. Cancer Nursing, 22 (2), 143-148. Ferrara, M. & De Gennaro, L. (2001). How much sleep do we need? Sleep Medicine, 5, 155-179. doi: 10.1053/smrv.2000.0138 Fiorentino, L. & Ancoli-Israel, S. (2007). Sleep dysfunction in patients with cancer. Current Treatment Options in Neurology, 9, 337-346. Folkman, S. & Lazarus, R. S. (1980). An analysis of coping in a middle-aged community sample. Journal of Health and Social Behavior, 21, 219-239. Fortner, B. V., Stepanski, E. J., Wang, S. C., Kasprowicz, S., & Heith Durrence, H. (2002). Sleep and quality of life in breast cancer patients. Journal of Pain and Symptom Management, 24, 471-480.

209

Friedman, L., Brooks, J. O., Bliwise, D. L., Yesavage, J. A., & Wicks, D. S. (1995). Perceptions of life stress and chronic insomnia in older adults. Psychology and Aging, 10, 352-357. Fujiwara, Y., Habu, Y., Ashida, K., Kusano, M., Higuchi, K., & Arakawa, T. (2013). Sleep disturbances and refractory gastroesophageal reflux disease symptoms in patients receiving once-daily proton pump inhibitors and efficacy of twice-daily Rabeprazole treatment. Digestion, 88, 145-152. doi: http://dx.doi.org/10.1159/000354071 Gau, A. S.-F., Shang, C.-Y., Merikangas, K. R., Chiu, Y.-N., Soong, W.-T., & Cheng, A. T.- A. (2007). Association between morningness-eveningness and behavioral/emotional problems among adolescents. Journal of Biological Rhythms, 22, 268-274. doi: 10.1177/0748730406298447 Gesiarz, F. & Crockett, M. J. (2015). Goal-directed, habitual and Pavlovian prosocial behavior. Frontiers in Behavioral Neuroscience, 9, 1-18. doi: 10.3389/fnbeh.2015.00135 Gleichgerrcht, E. & Decety, J. (2013) Empathy in clinical practice: How individual dispositions, gender, and experience moderate empathic concern, burnout, and emotional distress in physicians. PLOS One, 8, e61526. doi:10.1371/journal.pone.0061526 Goetz, J. L., Keltner, D., & Simon-Thomas, E. (2010). Compassion: An evolutionary analysis and empirical review. Psychological Bulletin, 136, 351-374. doi:10.1037/a001880 Goretti, B., Portaccio, E., Zipoli, V., Razzolini, L., & Amato, M. P. (2010). Coping strategies, cognitive impairment, psychological variables and their relationship with quality of life in multiple sclerosis. Neurological Sciences, 31, S227-30. doi: 10.1007/s10072-010-0372-8 Gosling, S.D., Rentfrow, P.J., & Swann, W.B. Jr. (2003). A very brief measure of the big five personality domains. Journal of Research in Personality, 37, 504–28. doi: 10.1016/S0092-6566(03)00046-1 Hahn, C., Cowell, J. M., Wiprzycka, U. J., Goldstein, D., Ralph, M., Hasher, L., & Zelazo, P. D. (2012). Circadian rhythms in executive function during the transition to adolescence: The effect of synchrony between chronotype and time of day. Developmental Science, 15, 408-16. doi: 10.1111/j.1467-7687.2012.01137.x Hall, M., Buysse, D. J., Dew, M. A., Prigerson, H. G., Kupfer, D. J., & Reynolds, C. F. (1997). Intrusive thoughts and avoidance behavior are associated with sleep disturbances in bereavement-related depression. Depression and Anxiety, 6, 106-112. Hamilton, J. G., Lobel, M., & Moyer, A. (2009). Emotional distress following genetic testing for hereditary breast and ovarian cancer: A meta-analytic review. Health Psychology, 28 (4) 510-518. doi: 10.1037/a0014778 Hammen, C., Kim, E. Y., Eberhart, N. K. & Brennan, P. A. (2009). Chronic and acute stress and the prediction of major depression in women. Depression and Anxiety, 26, 718-723. doi: 10.1002/da.20571

210

Hansen, J. & Lassen, C. F. (2012). Nested case-control study of night shift work and breast cancer risk among women in the Danish military. Occupational and Environmental Medicine, 69 (8), 551 – 556. doi: 10.1136/oemed-2011-100240. Hasher, L., Chung, C., May, C. P., & Foong, N. (2002). Age, time of testing, and proactive interference. Canadian Journal of Experimental Psychology, 56, 200-207. Hasher, L., Zacks, R. T., & Rahhal, T. A. (1999). Timing, instructions, and inhibitory control: Some missing factors in the age and memory debate. Gerontology, 45, 355-357. Hasler, B. P., Allen, J. J. B., Sbarra, D. A., Bootzin, R. R., & Bernert, R. A. (2010). Morningness-eveningness and depression: Preliminary evidence for the role of the behavioral activation system and positive affect. Psychiatry Research, 176, 166-173. doi: 10.1016/j.psychres.2009.06.006 Hede, K. (2009). Cancer and the circadian clock: Has the time finally come? Journal of the National Cancer Institute, 101, 550-553. doi: 10.1093/jnci/djp085 Heinzelmann, M., Lee, H., Rak, H., Livingston, W., Barr, T., Baxter, T., Scattergood-Keepper, L., Mysliwiec, V. & Gill, J. (2014). Sleep restoration is associated with reduced plasma C reactive protein and depression symptoms in military personnel with sleep disturbance after deployment. Sleep Medicine, 15, 1565-1570. doi: 10.1016/j.sleep.2014.08.004 Hicks, D. G. & Kulkarni, S. (2008). Review of biologic relevance and optimal use of diagnostic tools. American Journal of Clinical Pathology, 129, 263-273. Hidalgo, M. P., Caumo, W., Posser, M., Coccaro, S. B., Camozzato, A. L., & Chaves, M. L. (2009). Relationship between depressive mood and chronotype in healthy subjects. Psychiatry and Clinical Neurosciences, 63, 283-290. doi: 10.1111/j.1440- 1819.2009.01965.x Hintsanen, M., Puttonen, S., Smith, K., Toornroos, M., Jokela, M., Pulkki-Råback, L., Hintsa, T., Merjonen, P., Dwyer, T., Raitakari, O. T., Venn, A. & Keltikangas-Järvinen, L. (2014). Five factor personality traits and sleep: Evidence from two population based cohort studies. Health Psychology, 33, 1214-1223. doi: 10.1037/hea0000105 Hofman, M., Ryan, J. L., Figueroa-Moseley, C. D., Jean-Pierre, P., & Morrow, G. R. (2007). Cancer-related fatigue: The scale of the problem. The Oncologist, 12, 4-10. doi: 10.1634/theoncologist.12-S1-4 Horne, J. A. & Östberg, O. (1976). A self-assessment questionnaire to determine morningness- eveningness in human circadian rhythms. International Journal of Chronobiology, 4, 97- 110. Horowitz, M., Wilner, N., & Alvarez, W. (1979). Impact of events scale: A measure of subjective stress. Psychosomatic Medicine, 41, 209-218.

211

Howlader, N., Noone, A. M., Krapcho, M., Garshell, J., Miller, D. Altekruse, S. F., … Cronin, K. A. (eds). SEER Cancer Statistics Review, 1975-2011, National Cancer Institute. Bethesda, MD, http://seer.cancer.gov/csr/1975_2011/ , based on November 2013 SEER data submission, posted to the SEER web site, April 2014. Retrieved at: https://www.cancer.gov/about-cancer/causes-prevention/genetics/brca-fact-sheet#r6 Hoyt, M. A., Thomas, K. S., Epstein D. R., & Dirksen, S. R. (2009). Coping style and sleep quality in men with cancer. Annals of Behavior Medicine, 37, 88-93. doi: 10.1007/s12160-009-9079-6 Hsu, C., Gau, S. S., Shang, C., Chiu, Y. & Lee, M. (2012) Associations between chronotypes, psychopathology, and personality among incoming college students. Chronobiology International, 29, 491-501. doi: 10.3109/07420528.2012.668995 Injeyan, M. C., Shuman, C., Shugar, A., Chitayat, D., Atenafu, E. G., & Kaiser, A. (2011). Personality traits associated with genetic counselor compassion fatigue: The roles of dispositional optimism and locus of control. Journal of Genetic Counselling, 20 (5), 526- 540. doi: 10.1007/s10897-011-9379-4 Innominato, P. F., Levi, F. A., & Bjarnason, G. A. (2010). Chronotherapy and the molecular clock: Clinical implication in oncology. Advanced Drug Delivery Reviews, 62, 979-1001. doi: 10.1016/j.addr.2010.06.002 Institute of Medicine (IOM). (2008). Cancer care for the whole patient: Meeting psychosocial health needs. Nancy E. Adler and Ann E. K. Page, eds. Washington, DC: The National Academies Press. Jim, Richardson, Golden-Kreutz & Andersen, 2006). Strategies used in coping with a cancer diagnosis predict meaning in life for survivors. Health Psychology, 25, 753-761. doi: 10.1037/0278-6133.25.6.753 Jones, J. M., St. Peter, J. R., Fernandes, S. J., Herrenkohl, T. I., Kosterman, R., & Hawkins, J. D. (2011). Ethnic and gender variation in religious involvement: Patterns of expression in young adulthood. Review of Religious Research, 53, 207–225. doi:10.1007/s13644-011- 0006-5 Jonsdottir, I. H., Nordlund, A., Ellbin, S., Ljung, T., Glise, K., Währborg, P., & Wallin, A. (2013). Cognitive impairment in patients with stress-related exhaustion. Stress, 16, 181- 190. doi: 10.3109/10253890.2012.708950 Kahn-Greene, E., Lipizzi, E. L., Conrad, A. K., Kamimori, G. H. & Killgore, W. D. S. (2006). Sleep deprivation adversely affects interpersonal responses to frustration. Personality and Individual Differences, 41, 1433-1443. doi: 10.1016/j.paid.2006.06.002 Kalmbach, D. A., Schneider, L. D., Cheung, J., Bertrand, S. J., Kariharan, T., Pack, A., & Gehrman, P. R. (2017). Genetic basis of chronotype in humans: Insights from three landmark GWAS. Sleep, 40, 1-10. doi: 10.1093/sleep/zsw048

212

Karagozoglu, S. & Bingöl, N. (2008). Sleep quality and job satisfaction of Turkish nurses. Nursing Outlook, 56, 298-307, 307.e1-.e3. doi: 10.1016/j.outlook.2008.03.009 Keller, L. K., Grünewald, B., Vetter, C., Roenneberg, T., & Schulte!Körne, G. (2017). Not later, but longer: Sleep, chronotype and light exposure in adolescents with remitted depression compared to healthy controls.European Child and Adolescent Psychiatry, Epub ahead of print. doi: 10.1007/s00787-017-0977-z Khoynezhad, G., Rajaei, A. R., & Sarvarazemy, A. (2012). Basic religious beliefs and personality traits. Iranian Journal of Psychiatry, 7, 82-86. Killgore, W. D. S. (2010). Effects of sleep deprivation on cognition. Progress in Brain Research, 185, 105-129. doi: 10.1016/B978-0-444-53702-7.00007-5 Kleitman, N. (1963). Sleep and wakefulness. Chicago, Il: University of Chicago Press. Knutson, Spiegel, Penev & Van Cauter, 2007). The metabolic consequences of sleep deprivation. Sleep Medicine Reviews, 11, 163-178. doi: 10.1016/j.smrv.2007.01.002 Korczak, D., Huber, B., & Kister, C. (2010). Differential diagnostic of the burnout syndrome. GMS Health Technology Assessment, 6, 1-9. doi: 10.3205/hta000087 Koscec, A., Radosevic-Vidacek, B., & Bakotic, M. (2014). Morningness-eveningness and sleep patterns of adolescents attending school in two rotating shifts. Chronobiology International, 31, 52-63. doi: 10.3109/07420528.2013.821128 Kreitler, S. (1999). Denial in cancer patients. Cancer Investigation, 17, 514-534. Krishnan, H. C. & Lyons, L. C. (2015) Synchrony and desynchrony in circadian clocks: Impacts on learning and memory. Leaning & Memory (Cold Springs Harbor, N.Y.), 22 (9), 426- 437. Kristensen, T.S., Borritz M., Villadsen, E., & Christensen, K.B. (2005). The Copenhagen burnout inventory: a new tool for the assessment of burnout. Work Stress, 19, 192-207. doi: http://dx.doi.org/10.1080/02678370500297720 Kurtz, M. E., Given, B., Kurtz J. C., & Given, C. W. (1994). The interaction of age, symptoms, and survival status on physical and mental health of patients with cancer and their families. Cancer, 74(Suppl 7), 2071-2078. Lack, L., Bailey, M., Lovato, M., & Wright, H. (2009). Chronotype differences in circadian rhythms of temperature, melatonin, and sleepiness as measured in a modified constant routine protocol. Nature and Science of Sleep, 1, 1-8. Laerd Statistics (2015). Statistical tutorials and software guides. Retreived from: https://statistics.laerd.com

213

Lara, T., Madrid, J. A., & Correa, A. (2014). The vigilance decrement in executive function is attenuated when individual chronotypes perform at their optimal time of day. PLoS ONE, 9, e88820. doi: 10.1371/journal.pone.0088820 Lawson, V. L., Bundy, C., Belcher, J., & Harvey, J. N. (2013). Changes in coping behavior and the relationship to personality, health threat communication and illness perceptions from the diagnosis of diabetes: a 2-year prospective longitudinal study. Health Psychology Research, 1, 93-99. doi: 10.4082/hpr.2013.e20 Lazarus R. & Folkman S. (1984) Stress, appraisal, and coping. New York: Springer Lemma, S., Berhane, Y., Worku, A., Gelaye, B., Williams, M.A. (2014). Good sleep quality is associated with better academic performance among university students in Ethiopia. Sleep & Breathing, 18, 257-263. doi: 10.1007/s11325-013-0874-8 Lee, C. Y., Chen,H. C., Tseng, M. C., Lee,H. C., & Huang, L.H. (2015). The relationships among sleep quality and chronotype, emotional disturbance, and insomnia vulnerability in shift nurses. The Journal of Nursing Research, 23, 225-235. doi: 10.1097/jnr.0000000000000095 Lenaert, B., Barry, T. J., Schruers, K., Vervliet, B., & Hermans, D. (2016). Emotional attentional control predicts changes in diurnal cortisol secretion following exposure to a prolonged psychosocial stressor. Psychoneuroendocrinology, 63, 291-5. doi: 10.1016/j.psyneuen.2015.10.013 Lenjavi, M. R., Ahuja, M. A., Touchette, P. E., & Sandman, C. A. (2010). Maladaptive behaviors are linked with inefficient sleep in individuals with developmental disabilities. Journal of Neurodevelopmental Disorders, 2, 174-180. doi: 10.1007/s11689-010-9048-1 Leproult, R., Copinschi, G., Buxton, O. & Van Cauter, E. (1997). Sleep loss results in an elevation of cortisol levels the next evening. American Sleep Disorders Association and Sleep Research Society, 20, 865-870. Lerman, C. & Croyle, R. T. (1996). Emotional and behavioral responses to genetic testing for susceptibility to cancer. Oncology, 10, 191-195. Lerman, C. & Schwartz, M. (1993). Adherence and psychological adjustment among women at high risk for breast cancer. Breast Cancer Research and Treatment, 28, 145–155. Levi, F. (2001). Circadian chronotherapy for human cancers. The Lancet Oncology, 2, 307-315. doi: http://dx.doi.org/10.1016/S1470-2045(00)00326-0 Levi, F. (2006). The circadian timing system, a coordinator of life processes. Implications for the rhythmic delivery of cancer therapeutics. Conference Proceedings: Annual International Conference of the IEEE Engineering in Medicine and Biology Society, Suppl: 6736-6739. doi: 10.1109/IEMBS.2006.260934

214

Lim, J. & Dinges, D. F. (2008). Sleep deprivation and vigilant attention. Annals of the New York Academy of Science, 1129, 305-322. doi: 10.1196/annals.1417.002 Lim & Dinges (2010). A meta-analysis of the impact of short-term sleep deprivation on cognitive variables. Psychology Bulletin, 136, 375-389. doi: 10.1037/a0018883 Loewenthal, K. M., McLeod, A. K., & Cinnirella, M. (2000). Are women more religious than men? Gender differences in religious activity among different religious groups in the UK. Personality and Individual Differences, 32, 133-139. Lowrey, P. L. & Takahashi, J. S. (2011). Genetics of circadian rhythms in mammalian model organisms. Advances in Genetics, 74, 175-230. doi: 10.1016/B978-0-12-387690-4.00006- 4 Luo, J. Zhu, G., Zhao, Q., Guo, Q., Meng, H., Hong, Z. & Ding D. (2013). Prevalence and risk factors of poor sleep quality among Chinese elderly in an urban community: Results from the Shanghai aging study. PLoS ONE, 8, e81261. doi: 10.1371/journal.pone.0081261 Matchock, R. L. & Mordkoff, J. T. (2009). Chronotype and time-of-day influences on the alerting, orienting, and executive components of attention. Experimental Brain Research, 192, 189-198. doi: 10.1007/s00221-008-1567-6 May, C. P., Hasher, L., & Stoltzfus, E. R. (1993). Optimal time of day and the magnitude of age differences in memory. Psychological Science, 4, 326-330. McClung, C. R. (2011). Circadian rhythms: Lost in post translation. Current Biology, 21 (10), R400-R402. doi: 10.1016/j.cub.2011.04.024 McEwen, B. S. & Karatsoreos, I. N. (2015). Sleep deprivation and circadian disruption: Stress, allostastis and allostatic load. Sleep Medicine Clinics, 10, 1-10. doi: https://doi.org/10.1016/j.jsmc.2014.11.007 Merckaert, I., Libert, Y., Lieutenant, F., Moucheux, A., Farvacques, C., Slachmuyider, J.-L., & Razavi, D. (2013). Desire for formal psychological support among caregivers of patients with cancer: Prevalence and implications for screening their needs. Psycho-Oncology, 22, 1389-1395. doi: 10.1002/pon.3153 Merikanto, I., Kronholm, E., Peltonen, M., Laatikainen, T., Lahti, T., Partonen, T. (2012). Relation of chronotype to sleep complaints in the general Finnish population. Chronobiology International, 29, 311-317. doi: 10.3109/07420528.2012.655870 Merikanto, I., Paavonen, E. J., Saarenpää-Heikkilä, O., Paunio, T., & Partonen, T. (2017). Eveningness associates with smoking and sleep problems among pregnant women. Chronobiology International, 34, 650-658. doi: 10.1080/07420528.2017.1293085 Miyata, S., Noda, A., Iwamoto, K., Kawano, N., Okuda, M., & Ozaki, N. (2013). Poor sleep quality impairs cognitive performance. Journal of Sleep Research, 22, 535-541. doi: http://dx.doi.org/10.1111/jsr.12054

215

Monk, T. H. (2005). The post-lunch dip in performance. Clinics in Sports Medicine, 24, e15-23, xi-xii. doi: 10.1016/j.csm.2005.12.002 Moorey, S., Frampton, M., & Greer, S. (2003). The cancer coping questionnaire: A self-rating scale for measuring the impact of adjuvant psychological therapy on coping behaviour. Psycho-Oncology, 12, 331-344. doi: 10.1002/pon.646 Morin, L. P. & Allen, C. N. (2005). The circadian visual system, 2005. Brain Research Reviews, 51, 1-60. Muscatell, K. A., Slavich, G. M., Monroe, S. M., & Gotlib, I. H. (2009). Stressful life events, chronic difficulties, and the symptoms of clinical depression. Journal of Nervous Mental Disorders, 197, 154-160. doi:10.1097/NMD.0b013e318199f77b Natale, V., Alzani, A., & Cicogna, P. (2003). Cognitive efficiency and circadian typologies: A diurnal study. Personality and Individual Differences, 35, 1089-1105. National Research Council (US) Committee on Recognition and Alleviation of Distress in Laboratory Animals. Recognition and Alleviation of Distress in Laboratory Animals. Washington (DC): National Academies Press (US); 2008. Committee on Recognition and Alleviation of Distress in Laboratory Animals. Neilsen, T. (2010). Nightmares associated with the eveningness chronotype. Journal of Biological Rhythms, 25, 53-62. doi: 10.1177/0748730409351677 Nordstokke, D. W., Zumbo, B. D., Cairns, S. L., & Saklofske, D. H. (2011). The operating evaluation data. Practical Assessment, Research & Evaluation, 16, 1-8. Available: http://pareonline.net/getvn.asp?v=16&n=5 Northouse, L., Williams, A. L., Given, B., & McCorkle, R. (2012). Psychosocial care for family caregivers of patients with cancer. Journal of Clinical Oncology, 30, 1227-1234. doi: 10.1200/JCO.2011.39.5798 O’Brien, T. B. & DeLongis, A. (1996). The interactional context of problem-, emotion-, and relationship-focused coping: The role of the big five personality factors. Journal of Personality, 64, 775-813. Oken, B. S., Fonareva, I., & Wahbeh, H. (2011). Stress-related cognitive dysfunction in dementia caregivers. Journal of Geriatric Psychiatry and Neurology, 24, 191-198. doi:10.1177/0891988711422524 Opitz, P. C., Lee, I. A., Gross J. J., & Urry, H. L. (2014). Fluid cognitive ability is a resource for successful emotional regulation in older and younger adults. Frontiers in Psychology, 5, 1-13. doi: 10.3389/fpsyg.2014.00609

216

Orzechowska, Zajaczkowska, Talarowska & Galecki, 2013). Depression and ways of coping with stress: A preliminary study. Medical Science Monitor, 19, 1050-1056. doi: 10.12659/MSM.889778 Osse, B. H. P., Vernooij-Dassen, M. J. F. J., Schadé, E., & Grol R. P. T. M. (2006). Problems experienced by the informal caregivers of cancer and their needs for support. Cancer Nursing, 29, 378-388. Ostacoli, L., Cavallo, M., Zuffranieri, M., Negro, M., Sguazzotti, E., Picci, R. L., … Furlan, P. M. (2010). Comparison of experienced burnout symptoms in specialist oncology nurses working in hospital oncology units or in hospices. Palliative & Supportive Care, 8, 427- 432. doi: 10.1017/S1478951510000295 Ottoni, G. L., Antoniolli, E., & Lara, D. R. (2012). Circadian preference is associated with emotional and affective temperaments. Chronobiology International, 29, 786-193. doi: 10.3109/07420528.2012.679329 Pabst, S. R., Negriff, S., Dorn, L. D., Susman, E. J., & Huang, B. (2009). Depression and anxiety in adolescent demales: The impact of sleep preference and body mass index. Journal of Adolescent Health, 44, 554-560. doi: 10.1016/j.jadohealth.2008.11.012 Papantoniou, K., Castaño-Vinyals, G., Espinosa, A., Aragonés, N., Pérez-Gómez, B., Burgos, J., Gómez-Acebo, I., Llorca, J., Peiró, R., Jimenez-Moleón, J. J., Arredondo, F., Tardón, A., Pollan, M., & Kogevinas, M. (2015). Night shift work, chronotype and prostate cancer risk in the MCC-Spain case-control study. International Journal of Cancer, 137, 1147- 1157. doi: 10.1002/ijc.29400 Peres, I., Vetter, C., Blautzik, J., Reiser, M., Pöppel, E., Meindl, T., Roenneberg, T., & Gutyrchik, E. (2011). Chronotype predicts activity patterns in the neural underpinnings of the motor system during the day. Chronobiology International, 28, 883-889. doi: http://dx.doi.org/10.3109/07420528.2011.619084 Portaluppi, F., Riseo, R., Smolensky, M. H., Hermida, R. C., Ayala, D. R., & Fabbian, F. (2012). Circadian rhythms and cardiovascular health. Sleep Medicine Reviews, 16, 151-66. doi: 10.1016/j.smrv.2011.04.003 Potter, P., Deshields, T., Divanbeigi, J., Berger, J., Cipriano, D., Norris, L., & Olsen, S. (2010). Compassion fatigue and burnout: Prevalence among oncology nurses. Clinical Journal of Oncology Nursing, 14 (5), E45-E62. doi: http://dx.doi.org/10.1188/10.CJON.E56-E62 Radloff, L. S. (1977). The CES-D scale: A self-report depression scale for research in the general population. Applied Psychological Measurement, 1, 385-401. Ralph, M. R., Foster, R.G., Davis, F. C., & Menaker, M. (1990). Transplanted suprachiasmatic nucleus determines circadian period. Science, 247, 975-978. Ray, S. L., Wong, C., White, D. & Heaslip, K. (2013). Compassion satisfaction, compassion fatigue, work life conditions, and burnout among frontline mental health care

217

professionals. Traumatology, 19 (4), 255-267. Reid, K. J., Jaksa, A. A., Eisengart, J. B., Baron, K. G., Lu, B., Kane, P., Kang, J., & Zee, P. C (2012). Systematic evaluation of Axis-I DSM diagnoses in delayed sleep phase disorder and evening-type circadian preference. Sleep Medicine, 13, 1171-1177. doi:10.1016/j.sleep.2012.06.024 Renzi, C., Tabolli, S., Ianni, A., Di Pietro, C., & Puddu, P. (2005). Burnout and job satisfaction comparing healthcare staff of a dermatological hospital and a general hospital. Journal of the European Academy of Dermatology & Venereology, 19, 153–7. doi: 10.1111/j.1468-3083.2005.01029.x Reppert, S. M. & Weaver, D. R. (2001). Molecular analysis of mammalian circadian rhythms. Annual Review of Physiology, 63, 647-676. doi: 10.1146/annurev.physiol.63.1.647 Riley, L. A. & Esser, K. A. (2017). The role of the molecular clock in skeletal muscle and what it is teaching us about muscle-bone crosstalk. Current Osteoporosis Reports, Epub ahead of print. doi: 10.1007/s11914-017-0363-2 Rique, G. L., Fernandes Filho, G. M., Ferreira A. D., & de Sousa-Muñoz, R. L. (2014). Relationship between chronotype and quality of sleep in medical students at the Federal University of Paraiba, Brazil. Sleep Science, 7, 96-102. doi: http://dx.doi.org/10.1016/j.slsci.2014.09.004 Roenneberg, T., Kuehnle, T., Juda, M., Kanermann, T., Allebrandt, K., Gordijn, M., & Merrow, M. (2007). Epidemiology of the human circadian clock. Sleep Medicine Reviews, 11, 429-438. doi: 10.1016/j.smrv.2007.07.005 Roenneberg, T., Wirz-Justice, A., & Merrow, M. (2003). Life between clocks: Daily temporal patterns of human chronotypes. Journal of Biological Rhythms, 18, 80-90. doi: 10.1177/0748730402239679 Rosales-Lagarde, A., Armony, J. L., del Rio-Portilla, Y., Trejo-Martinez, D., Conde R., & Corsi- Cabrera, M. (2012). Enhanced emotional reactivity after selective REM sleep deprivation in humans: An fMRI study. Frontiers in Behavioral Neuroscience, 6, 1-13. doi: 10.3389/fnbeh.2012.00025 Roscoe, J. A., Kaufman, M. E., Matteson-Rusby, S. E., Palesh, O. G., Ryan, J. L., Kohli, S., Perlis, M. L., & Morrow, G. R. (2007). Cancer-related fatigue and sleep disorders. The Oncologist, 12, 35-42. doi: 10.1634/theoncologist.12-S1-35 Rose, D., Gelaye, B., Sanchez, S., Castañeda, B., Sanchez, E., Yanez, D., & Williams, M. A. (2015). Morningness/eveningness chronotype, poor sleep quality, and daytime sleepiness in relation to common mental disorders among Peruvian college students. Psychology of Health Medicine, 20, 345-352. doi: 10.1080/13548506.2014.951367

218

Rospenda, K. M., Minich, L. M., Milner, L. A., & Richman, J. A. (2010). Caregiver burden and alcohol use in a community sample. Journal of Addictive Diseases, 29, 314–324. doi:10.1080/10550887.2010.489450. Russo, J. & Russo, I. H. (2006). The role of estrogen in the initiation of breast cancer. The Journal of Steroid Biochemistry and Molecular Biology, 102 (1-5), 89-96. doi: 10.1016/j.jsbmb.2006.09.004 Ruggieri, V., Zeppegno, P., Gramaglia, C., Gill, S., Deantonio, L., & Krengili, M. (2014). A survey of Italian radiation oncologists: Job satisfaction and burnout. Tumori, 100, 307- 314. doi: 10.1700/1578.17212 Rzeszutek M, Schier K. (2014). Temperament traits, social support, and burnout symptoms in a sample of therapists. Psychotherapy (Chicago, III), 51, 574–9. doi: 10.1037/a0036020 Saita, E., Acquati, C., & Kayser, K. (2015). Coping with early stage breast cancer: Examining the influence of personality traits and interpersonal closeness. Frontiers in Psychology, 6, 1-8. doi: 10.3389/fpsyg.2015.00088 Sardiwalla, N., VandenBerg, H., & Esterhuyse, K. G. F. (2007). The role of stressors and coping strategies in the burnout experienced by hospice workers. Cancer Nursing, 30, 488-497. doi: 10.1097/01.NCC.0000300159.67897.c7 Saroglou, V. (2002). Religion and the five factors of personality: A meta-analytic review. Personality and Individual Differences, 32, 15-25. Savard, J. & Morin, C. M. (2001). Insomnia in the context of cancer: A review of a neglected problem. Journal of Clinical Oncology, 19, 895-908. doi: 10.1200/JCO.2001.19.3.895 Schaufeli, W.B. & Greenglass, E. R. (2001). Introduction to special issue on burnout and health. Psychology & Health, 16, 501-510. doi: 10.1080/08870440108405523 Scheer, F. A. J. L., Morris, C. J., & Shea, S. A. (2013). The internal circadian clock increases hunger and appetite in the evening independent of food intake and other behaviors. Obesity (Silver Spring), 21, 421-423. doi: 10.1002/oby.20351 Shehmar, M. & Gupta, J. K. (2010). The influence of psychological factors on recovery from hysterectomy. Journal of the Royal Society of Medicine, 103, 56-59. doi: 10.1258/jrsm.2009.090198 Schlarb, A. A., Sopp, R., Ambiel, D., & Grünwald, J. (2014). Chronotype-related differences in childhood and adolescent aggression and antisocial behavior – A review of the literature. Chronobiology International, 31, 1-16. doi: 10.3109/07420528.2013.829846 Schmidt, C., Collette, F., Cajochen, C., & Peigneux, P. (2007). A time to think: Circadian rhythms in human cognition. Cognitive Neuropsychology, 24, 755-789. doi: 10.1080/02643290701754158

219

Schmidt, C., Collette, F., Reichert, C. F., Maire, M., Vandewalle, G., Peigneux, P., & Cajochen, C. (2015). Pushing the limits: Chronotype and time of day modulate working memory- dependent cerebral activity. Frontiers in Neurology, 6, 1-9. doi: 10.3389/fneur.2017.00149 Selvi, Y., Aydin, A., Atli, A., Boysan, M., Selvi, F., & Besiroglu, L. (2011). Chronotype differences in suicidal behavior and impulsivity among suicide attempters. Chronobiology International, 28, 170-175. doi: 10.3109/07420528.2010.535938 Severn, M. S., Searchfield, G. D., & Huggard, P. (2012) Occupational stress among audiologists: Compassion satisfaction, compassion fatigue, and burnout. International Journal of Audiology, 51, 3-9. doi: 10.3109/14992027.2011.602366. Sharma, N., Chakrabarti S., & Grover, S. (2016). Gender differences in caregiving among family – caregivers of people with mental illness. World Journal of Psychiatry, 6, 7-17. doi: 10.5498/wjp.v6.i1.7 Sharpe, J. P., Martin, N. R., & Roth, K. A. (2011). Optimism and the big five factors of personality: Beyond neuroticism and extraversion. Personality and Individual Differences, 51, 946–51. doi: https://doi.org/10.1016/j.paid.2011.07.033 Shi, L., Xia, T.-S., Wei, X.-L., Zhou, W., Xue, J., Cheng, L., … Ding, Q. (2015). Estrogen receptor (ER) was regulated by RNPC1 stabilizing mRNA in ER positive breast cancer. Oncotarget, 6, 12264-12278. doi: 10.18632/oncotarget.3654 Shin, H., Park, Y. M., Ying, J. Y., Kim, B., Noh, H., & Lee, S. M. (2014). Relationships between coping strategies and burnout: A meta-analytic approach. Professional Psychology: Research and Practice, 45, 44-56. doi: 10.1037/a0035220 Simpkin, C. T., Jenni, O. G., Carskadon, M. A., Wright Jr., K. P., Akacem, L. D., Garlo, K., & LeBourgeois, M. K. (2014). Chronotype is associated with the timing of the circadian clock and sleep in toddlers. Journal of Sleep Research, 23, 397-405. doi: 10.1111/jsr.12142 Singh, Y. & Sharma, R. (2012). Relationship between general intelligence, emotional intelligence, stress levels and stress reactivity. Annals of Neurosciences, 19, 107-111. doi: 10.5214/ans.0972.7531.190304 Sklenarova, H., Krümpelmann, A., Haun, M. W., Friederich, H.-C., Huber, J., Thomas, M., Winkler, E. C., Herzog, W., & Hartmann, M. (2015). When do we need to care about the caregiver? Suppotive care needs, anxiety, and depression among informal caregivers of patients with cancer and cancer survivors. Cancer, 121, 1513-1519. Sodeke-Gregson, E. A., Holttum, S., & Billings, J. (2013). Compassion satisfaction, burnout, and secondary traumatic stress in UK therapists, who work with adult trauma clients. Journal of Psychotraumatology, 4, eCollection. doi: 10.3402/ejpt.v4i0.21869.

220

Söderström, M., Ekstedt, M., Åkerstedt, T., Nilsson, J., & Axelsson, J. (2004). Sleep and sleepiness in young individuals with high burnout scores. Sleep, 27, 1369-1377. Soehner, A. M., Kennedy, K. S., & Monk, T. H. (2011). Circadian preference and sleep-wake regularity: Associations with self-report sleep parameters in daytime-working adults. Chronobiology International, 28, 802-809. doi:10.3109/07420528.2011.613137 Spahni, S., Morselli, D., Perrig-Chiello, P. & Bennet, K. M. (2015). Patterns of psychological adaptation to spousal bereavement in old age. Gerontology, 61, 456-68. doi: 10.1159/000371444 Spector PE. (1985). Measurement of human service staffsatisfaction: Development of the job satisfaction survey. American Journal of Community Psychology, 13, 693–713 Stamm, B. H. (2002). Measuring compassion satisfaction as well as fatigue: Developmental history of the compassion satisfaction and fatigue test. In C. R. Figley (Ed.). Treating compassion fatigue (pp. 107-119). New York: Brunner-Routledge. Stamm, B. H. (2010). The Concise ProQol Manual, 2nd Ed. Pocatello, ID: ProQOL.org. Staner, L. (2003). Sleep and anxiety disorders. Dialogues in Clinical Neuroscience, 5, 249-258. Stavitsky, K., Neargarder, S., Bogdanova, Y., McNamara, P., & Cronin-Golomb, A. (2012). The impact of sleep quality on cognitive functioning in Parkinson’s disease. Journal of the International Neuropsychological Society, 18, 108–117. doi: 10.1017/S1355617711001482. Stepanski, E. J. (2002). The effect of sleep fragmentation on daytime function. Sleep, 25, 268- 276. Stephan, F. K., Schwann, J. M, & Sisk, C. L. (1979). Anticipation of 24-hour feeding schedules in rats with lesions in the suprachiasmatic nucleus. Behavioral and Neural Biology, 25 (3), 346-363. Sullivan, A. (2002). Gender differences in coping strategies of parents of children with Down syndrome. Down Syndrome Research and Practice, 8, 67-73. Tavernier, R., Munroe, M., & Willoughby, T. (2015). Perceived morningness-eveningness predicts academic adjustment and substance use across university, but social jetlag is not to blame. Chronobiology International, 32, 1233-1245. doi: 10.3109/07420528.2015.1085062 Tempesta, D., Couyoumdjian, A., Curcio, G., Moroni, F., Marzano, C., De Gennaro, L., and Ferrara, M. (2010). Lack of sleep affects the evaluation of emotional stimuli. Brain Research Bulletin, 82, 104-108. doi: 10.1016/j.brainresbull.2010.01.014

221

Thomas, K., Bower, J., Hoyt, M. A., & Sepah, S. (2010). Disrupted sleep in breast and prostate cancer patients undergoing radiation therapy: The role of coping processes. Psychooncology, 19, 767-776. doi: 10.1002/pon.1639 Tsai, F. J., Huang, W. L. & Chan, C. (2009). Occupational stress and burnout of lawyers. Journal of Occupational Health, 51(5), 443-450. Turiano, N. A., Whiteman, S. D., Hampson, S. E., Roberts, B. W., & Mroczek, D. (2012). Personality and substance use in midlife: Conscientiousness as a moderator and the effects of trait change. Journal of Research in Personality, 46, 295-305. doi:10.1016/j.jrp.2012.02.009 Underwood, B. & Moore, B. (1982). Perspective-taking and altruism. Psychological Bulletin, 91, 143-173. Vaessen, T. J. A., Overeem, S., & Sitskoorn, M. M. (2015). Cognitive complaints in obstructive sleep apnea. Sleep Medicine Reviews, 19, 51-58. doi: 10.1016/j.smrv.2014.03.008 Vahey, D. C., Aiken, L. H., Sloane, D. M., Clarke, S. P., & Vargas, D. (2004). Nurse burnout and patient satisfaction. Medical Care, 42 (2 Suppl), II57-66. doi: 10.1097/01.mlr.0000109126.50398.5a Vandekerckhove, M. & Cluydts, R. (2010). The emotional brain and sleep: An intimate relationship. Sleep Medicine Reviews, 14, 219-226. doi: 10.1016/j.smrv.2010.01.002 Vanwesenbeeck, I. (2005). Burnout among female indoor sex workers. Archives of Sexual Behavior, 34 (6), 627-639. doi: 10.1007/s10508-005-7912-y Van Dongen, H. P. A., Maislin, G., Mullington, J. M., & Dinges, F. (2003). The cumulative cost of additional wakefulness: Dose-response effects on neurobehavioral dunctions and sleep physiology from chronic sleep restriction and total sleep deprivation. Sleep, 26, 117-126. van Litsenburg, R.R., Huisman, J., Hoogerbrugge, P. M., Egeler, R. M., Kaspers, G. J., & Gemke, R. J. (2011). Impaired sleep affects quality of life in children during maintenance treatment for acute lymphoblastic lymphoma: an exploratory study. Health and Quality of Life Outcomes, 9, 25-31. doi: 21496357 http://dx.doi.org/10.1186/1477-7525-9-25 Vardar, E.,Vardar, S. A., Molla, T., Kaynak, C., & Ersoz, E. (2008). Psychological symptoms and sleep quality in young subjects with different circadian preferences. Biological Rhythms Research, 39, 493-500. doi: 10.1080/09291010701736884 Vargas, S., Wohlgemuth, W. K., Antoni, M. H., Lechner, S. C., Holley H. A., & Carver, C. S. (2010). Sleep dysfunction and psychosocial adaptation among women undergoing treatment for non-metastatic breast cancer. Psychooncology, 19, 669-673, doi: 10.1002/pon.1603 Vela-Bueno, A., Moreno-Jiménez, B., Rodríguez-Muñoz, A., Olavarrieta-Bernardino, S., Fernández-Mendoza, J., De la Cruz-Troca, J. J. Bixler, E. O., & Vgontzas, A. N. (2008).

222

Insomnia and sleep quality among primary care physicians with low and high burnout levels. Journal of Psychosomatic Research, 64, 435-442. doi: 10.1016/j.jpsychores.2007.10.014 Vitale, J. A., Colagiuri, G., & Weydahl, A. (2013). Influence of chronotype on responses to a standardized, self-paced walking task in the morning vs. afternoon: A pilot study. Perceptual & Motor Skills: Exercise & Sport, 116, 1020-1028. doi: 10.2466/06.19.PMS.116.3.1020-1028 Vitale, J. A., Roveda, E., Montaruli, A., Galasso, L., Weydahl, A., Caumo, A., & Carandente, F. (2015). Chronotype influences activity circadian rhythm and sleep: differences in sleep quality between weekdays and weekend. Chronobiology International, 32 (3), 405-415. Vollrath, M., Torgersen, S., & Alnaes, R. (1998). Neuroticism, coping and change in MCM-II clinical syndromes: Test of a mediator model. Scandinavian Journal of Psychology, 39, 15-24. Wang, X. M., Walitt, B., Saligan, L., Tiwari, A. F. Y., Cheung C. W., & Zhang, Z. J. (2015). Chemobrain: A critical review and causal hypothesis of link between cytokines and epigenetic. Cytokine, 72, 86-96. doi: 10.1016/j.cyto.2014.12.006 Westermann, C., Kozak, A., Harling, M., & Nienhaus, A. (2014). Burnout intervention studies for inpatient elderly care nursing staff: systematic literature review. International Journal of Nursing Studies, 51, 63-71. doi: 10.1016/j.ijnurstu.2012.12.001 Wong, M. W., Roberson, G., & Dyson, R. (2015). Prospective relationship between poor sleep and substance-related problems in a national sample of adolescents. Alcoholism, Clinical and Experimental Research, 39, 355-362. doi: 10.1111/acer.12618 Wright, K. P., Lowry, C. A., & LeBourgeois, M. K. (2012). Circadian and wakefulness-sleep modulation in humans. Frontiers in Molecular Neuroscience, 5, 1-12. doi: 10.3389/fnmol.2012.00050 Yun, J. A., Ahn, Y. S., Jeong, K. S., Joo, E. J., & Choi, K. S. (2015). The relationship between chronotype and sleep quality in Korean firefighters. Clinical Psychopharmacology and Neuroscience, 13, 201-208. doi: http://dx.doi.org/10.9758/cpn.2015.13.2.201 Zaza, C., Sellick, S. M., & Hillier, L. M. (2005). Coping with cancer: What do patients do? Journal of Psychosocial Oncology, 23, 55-73, doi:10.1300/J077v23n01_04 Zhang, Q., Yao, D., Yang, J., & Zhou, Y. (2014). Factors influencing sleep disturbances among caregivers of cancer patients in northeast China. PLOS One, e108614. doi: 10.1371/journal.pone.0108614 Zhang, L., Zhao, J., Xiao, H., Zheng, H., Xiao, Y., Chen, M. & Chen, D. (2014). Mental health and burnout in primary and secondary school teachers in the remote mountain areas of Guangdong Province in the People’s Republic of China. Neuropsychiatric Disease and Treatment, 10, 123-130. doi: 10.2147/NDT.S56020