Original Article

From the *Jewish General Hospital,

Montr�eal, Qu�ebec, Canada; †IngramSchool of Nursing, McGill University,

Montr�eal, Qu�ebec, Canada.

Address correspondence to C�eline

G�elinas, RN, PhD, Ingram School of

Nursing, Wilson Hall, 3506

University Street. Montr�eal, Qu�ebec,Canada H3A 2A7. E-mail: celine.

gelinas@mcgill.ca

Received August 15, 2016;

Revised April 7, 2017;

Accepted April 7, 2017.

1524-9042/$36.00

� 2017 by the American Society for

Pain Management Nursing

http://dx.doi.org/10.1016/

j.pmn.2017.04.005

Family Perspectives ofTraumatically Brain-Injured Patient PainBehaviors in theIntensive Care Unit

--- Brandy L. Vanderbyl, RN, MSc, MScA(N),*

and C�eline G�elinas, RN, PhD*,†

- ABSTRACT:Behavioral scales allow for the pain assessment of vulnerable critically

ill patients who are unable to self-report. However, validity of the use

of such scales is limited in traumatic brain injury patients with an

altered level of consciousness as a result of the different way that these

patients express pain. Family participation is considered as an

important component of pain assessment for those unable to self-

report, but research in this area is minimal so far. This study aimed to

describe what behaviors family caregivers deemed relevant to pain for

patients with a traumatic brain injury with an altered level of con-

sciousness in the intensive care unit. Using a mixed-method descrip-

tive design, semistructured interviews were conducted and behaviors’

relevance was quantitatively rated by seven family caregivers of

nonverbal patients with traumatic brain injury in the intensive care

unit of a tertiary trauma center in Montreal, Canada. Family caregivers

were able to provide rich descriptions of a number of behaviors they

observed in their loved ones that were perceived to be relevant indi-

cators of pain, such as muscle tension and key facial expressions and

body movements. Several factors influenced how behaviors were in-

terpreted by family, including personal medical beliefs and intimate

knowledge of the patient’s history. The pain behaviors determined by

family caregivers can be useful in the pain assessment process of

traumatic brain injury patients with an altered level of consciousness.

Their input could also be helpful in further development of pain

assessment tools.

� 2017 by the American Society for Pain Management Nursing

INTRODUCTION

It is estimated that 1.7 million people annually sustain a traumatic brain injury inthe United States, and 275,000 are hospitalized (Faul, Xu, Wald, & Coronado,

Pain Management Nursing, Vol -, No - (--), 2017: pp 1-12

2 Vanderbyl and G�elinas

2010). Many of those individuals will be hospitalized in

an intensive care unit (ICU) and will likely suffer from

pain but may be unable to self-report their pain as a

result of their critical condition. Despite recent up-

dates of guidelines and the progressing validation of

pain scales, there is still a substantial problem in deter-

mining when pain is present for critically ill patientswho are unable to communicate, particularly those

with a brain injury. No current scale can adequately

determine the presence of pain in those with serious

brain injuries such as traumatic brain injury (TBI)

with altered level of consciousness (LOC; Arbour &

G�elinas, 2014; G�elinas & Arbour, 2009; Le, G�elinas,Arbour, & Rodrigue, 2013). Family participation is

considered an important component of painassessment for those unable to self-report, but research

in this area is minimal so far (Herr, Coyne, McCaffery,

Manworren, & Merkel, 2011). The aim of this study

was to describe pain behaviors as perceived by family

caregivers of critically ill TBI patients with an altered

LOC in the ICU.

Literature ReviewCritically ill patients experience moderate to severe

pain that can occur at rest or during routine care(Arbour et al., 2014; Chanques et al., 2007; Puntillo

et al., 2014). In large international studies involving

critically ill adults, including those with trauma, pain

was commonly reported during procedures such as

turning, wound care, and tube or drain removal

(Puntillo et al., 2001, 2014). Moreover, in a study of

255 patients who had been admitted in the ICU, 28%

of whom had suffered a TBI, posttraumatic stresssymptoms 4-6 weeks after discharge correlated with

a patient’s memory of pain and an inability to

express needs during their ICU stay (Myhren et al.,

2009). Unfortunately, many critically ill patients are un-

able to communicate their pain as a result of several

factors, including an altered LOC, mechanical ventila-

tion, and the administration of sedative agents or

neuromuscular blocking agents. This inability to pro-vide a self-report of pain leaves the patient particularly

vulnerable to underrecognition and undertreatment of

pain (Herr et al., 2011).

When self-report is impossible to obtain, behav-

ioral indicators should be taken into account as part

of the pain assessment process (Herr et al., 2011). In

the Society of Critical Care Medicine practice guide-

lines (Barr et al., 2013), two behavioral pain scaleshave been suggested for clinical use: the Critical-Care

Pain Observation Tool (CPOT) and the Behavioral

Pain Scale (BPS). According to the available evidence

analyzed in the Society of Critical Care Medicine guide-

lines and in a psychometric review (G�elinas, Puntillo,

Joffe, & Barr, 2013), these scales were found to be

the most valid and reliable for use in adult ICU patients

with nonbrain trauma, postoperative or medical

causes. Some studies have also found that their clinical

application may be effective in improving patient out-

comes and pain management practices (Chanques

et al., 2006; G�elinas, Arbour, Michaud, Valliant, &Desjardins, 2011; Payen et al., 2009; De Jong et al.

2013; Rose, Haslam, Dale, Knechtel, & McGillion,

2013).

Although scales such as the CPOT and the BPS

have been developed specifically for nonverbal criti-

cally ill ICU patients, recent studies have reported

that in fact patients with brain injury, including TBI,

react differently to painful stimuli when they have analtered LOC (Arbour et al., 2014; G�elinas & Arbour,

2009; Le et al., 2013; Roulin & Ramelet, 2014). For

example, it has been observed that brain-injured ICU

patients with an altered LOC exhibit different behav-

iors (such as eye opening, eye weeping, face flushing)

and are less likely to grimace or become rigid in

response to painful procedures as opposed to other

critically ill ICU patients (Arbour et al., 2014; G�elinas& Arbour, 2009; Le et al., 2013). Therefore,

descriptions of behaviors included in existing pain

scales may not be adapted to ICU patients with a

brain injury and altered LOC.

Because of their unique point of view, family care-

givers should be encouraged to take part in pain assess-

ment of those who cannot self-report (Herr et al.,

2011). Family caregivers possess the valuable perspec-tive of being familiar with the patient before and after a

TBI and thus may be well-equipped to identify subtle

behaviors in their loved one that could be indicative

of pain. So far, little research has been done to describe

family caregivers’ perceptions of pain behaviors of

nonverbal ICU patients. Indeed, the only study con-

ducted in the ICU was led by Puntillo and colleagues,

who compared 245 ICU patients’ self-report of theirpain and distress to the family member’s, nurse’s and

physician’s reports of perceived pain and distress

(Puntillo et al., 2012). Findings supported that family

members determined pain intensity and distress in

the patient more accurately than the nurses or physi-

cians. However, Puntillo’s study was conducted with

patients able to self-report, and little evidence exists

about the family members’ perceptions of the pain ex-pressed by patients unable to self-report.

METHODS

Research QuestionThis study aimed to answer the following research

question: What behaviors are deemed relevant to

3Family Perspectives of Patient Pain Behaviors

pain by family caregivers during the hospital stay of

nonverbal TBI patients with an altered level of con-

sciousness in the intensive care unit?

DesignBoth quantitative questionnaire and qualitative inter-

views were employed in this mixed-method descrip-

tive study design.

Sample and SettingParticipants were recruited using convenience sam-

pling from the ICU of a tertiary trauma center in Mon-

treal, Canada. To be eligible, participants needed to be

adult, English-speaking family caregivers of a TBI pa-

tient in the ICU. A family caregiver was defined as

someone living with the patient, a family memberwho the patient saw regularly, or an individual who

played a significant role in the patient’s life for a mini-

mum of 1 year (Lefebvre, Cloutier, & Levert, 2008). The

participant must have visited the patient’s bedside at

least three times during the patient’s current ICU

stay, allowing the participant time for behavior obser-

vation. Participant eligibility criteria also related to

the characteristics of patients associated with the fam-ily caregivers. Patient criteria included being presently

admitted to the ICU after a TBI for a minimum of 3 days

and a maximum of 4 weeks. These patients also

needed to be unable to self-report pain for reasons

such as an altered LOC (defined by a Glasgow Coma

Scale score < 13 (Teasdale & Jennett, 1974)) at the

time of family caregiver recruitment. To ensure patient

movements were possible during family observation,patients could not have had suspected brain death,

been quadriplegic, or have been under the effects of

neuromuscular blocking agents. Family caregivers of

patients who were hospitalized because of a suicide

attempt were also excluded.

ProceduresStudy procedures, including participant screening and

recruitment, began after ethics approval by the hospi-tal’s ethics board. Eligible participants were identified

with the support of the assistant nurse managers and

were met by a research team member to present the

study and obtain written consent. Consented partici-

pants were interviewed individually in a private

meeting room in the ICU at a day and time that were

convenient to them. Interviews performed in English

lasted up to 60 minutes and were audio-recorded. Atthe end of the interview, so to not influence previous

responses, participants were asked to fill out a ques-

tionnaire that required them to evaluate relevance of

behavioral indicators on a checklist, as described later.

After having completed the questionnaire, participants

were asked to make any additional comments about

behaviors they had observed in their hospitalized rela-

tives during painful events and to describe them in

further detail.

InstrumentsInterview Guide. A semistructured interview guide

was used to cover several topics: events that family

caregivers believe caused their relatives pain in theICU; related behaviors they have witnessed; beliefs

regarding the timing, specificity, and intensity of the

behaviors; and participant confidence about the rele-

vance of those behaviors. The interview (see Table 1

for interview guide) followed the ‘‘PQRSTU’’ mne-

monic commonly used in pain assessment that guides

questioning on possible palliating/provoking factors,

quality, region, severity, timing, and understanding ofa symptom (Powell, Downing, Ddungu, & Mwangi-

Powel, 2010). Although this mnemonic is normally

used for direct patient questioning and was used in a

previous study (G�elinas, 2007), it was adapted in our

interview for use in proxy reporting of patient pain be-

haviors by family caregivers. Our interview guide was

trialed with two colleagues and an expert in qualitative

research, and questions were edited accordinglybefore use with participants.

Questionnaire about Relevance of Behaviors. Astructured questionnaire was used to evaluate the rele-

vance of 50 behaviors combining indicators from the

CPOT (G�elinas, Fillion, Puntillo, Viens, & Fortier,

2006), the Behavior Observation Tool used in Thunder

Project II (Puntillo et al., 2004), and newly developed

items (Le et al., 2013). This behavioral checklist wasused in a recent observational study to describe behav-

iors in critically ill TBI patients (Arbour et al., 2014).

The behaviors were clustered into five categories:

facial expressions, body movements, muscle tension,

compliance with the ventilator (for intubated pa-

tients), and vocalization (for nonintubated patients).

Each behavior’s relevance to pain was rated by the

participant on a 4-point descriptive scale (1 ¼ not rele-vant, 2 ¼ somewhat relevant, 3 ¼ quite relevant,

4 ¼ very relevant) specifically for their hospitalized

relative. Extra space was also given for participants

to suggest new behaviors they observed that were

not already included in the list, allowing for compre-

hensiveness of the behavioral questionnaire to be

addressed.

Sociodemographic Information of Family Care-giver and Information Related to the Patient.Basic sociodemographic information (e.g., gender,

age) for the participant and patient was gathered. Fam-

ily caregivers were also asked to report on basic patient

medical information (e.g., TBI cause, type, number of

4 Vanderbyl and G�elinas

days in ICU) and on information about their relation-

ship with the patient (e.g., type of familial relationship,

length of relationship, time family member spent at

bedside).

Data AnalysisQuantitative and qualitative data sets were linked, a

process of separately analyzing each type of datawith their traditional techniques and combining the re-

sults at the interpretation stage to facilitate data trian-

gulation (Sandelowski, 2000).

Qualitative Data. Data analysis occurred as soon as

possible after interviews took place. All audio-

recordingswere transcribedverbatimby the interviewer

(B. V.) conducting the interviews, and descriptive field

notes taken during the interview were appended tothe transcript at the corresponding time points.

Content analysis followed the approach of

Graneheim and Lundman (2004). The transcripts and

field notes were read by the interviewer (B. V.) and

the study investigator (C. G.) to gain a thorough under-

standing of the content. A code was assigned to each

word or phrase identifying or describing patient behav-

iors and their interpreted meaning. Behaviors wereclassified using the main categories found in the ques-

tionnaire: facial expressions, body movements, muscle

tension, compliance with the ventilator (for intubated

patients), vocalizations (for nonintubated patients),

and other. Two additional research trainees not

involved in this study also reviewed a subset of the

interview transcripts and the developed coding frame-

work to ensure codes and themes were not missed bythe study team. In addition, new behaviors that partic-

ipants identified that were not already included in the

structured behavioral questionnaire were compiled

and compared with behaviors discussed in the

interview.

Quantitative Data. Descriptive statistics were calcu-

lated using SPSS version 22.0 (IBM Corp., Armonk, NY,

USA). Frequencies were calculated for nominal data(i.e., gender, level of education, relationship with pa-

tient, cause and severity of TBI), and medians and

ranges were calculated for interval data (i.e., age, esti-

mated number of ICU visits, date of admission, and

length of ICU stay). Frequencies of relevance scores

for each behavior in the questionnaire were also

calculated.

RESULTS

Sample CharacteristicThe sample included seven participants who were fam-

ily caregivers of patients admitted to the ICU for a TBI

(Table 2). The participants were reporting on ICU

patients who were their spouses, parents, siblings, or

children, with whom they had had a relationship for

a median of 37 years. The participants consisted of fam-

ily caregivers who were highly present during the hos-

pitalization; they had visited their relative at the ICU

bedside for an average of 30 hours at the time of inter-

view, which took place at least 4 days after patientadmission. The ICU patients being discussed in the in-

terviews were predominantly male patients diagnosed

with a recent severe TBI (86%) that was acquired from

a fall (57%). All patients were intubated with an altered

level of consciousness (Glasgow coma scale scores

ranging from 7 to 10 out of 15), and thus were unable

to self-report. However, two patients (29%) had been

able to self-report on the presence or absence of painduring the first day of their ICU hospitalization before

being intubated.

Pain BehaviorsFamily caregivers observed several behaviors that they

associated with patients’ pain, and this was a major

theme coming from our analysis. Behaviors rated on

the questionnaire as quite or very relevant to pain for

their hospitalized relative by three or more partici-

pants are listed in Table 3. To facilitate triangulationof quantitative and qualitative data, quotations from

the interviews are included in Table 3 to help further

describe behaviors that were rated as most relevant

to patients’ pain. Novel behaviors not listed in the

questionnaire, or those discussed by two or fewer par-

ticipants, are described in the text. Pain behaviors fit

into several themes: facial expressions, body move-

ments, muscle tension, and ventilator compliance.One of the participants did not believe that her family

member had experienced any pain during ICU hospi-

talization and thus did not consider or rate any of the

behaviors observed as relevant to pain on the question-

naire or during the interview. Although participants

were also questioned on pain behavior severity, timing,

and specificity of pain behaviors, no common themes

across participants relating to these factors were foundduring the analysis.

Facial Expressions. Facial expressions that were

believed to be relevant to patient pain were diverse.

These behaviors involved the eyes and eyebrows, the

mouth, or more generally the face.

Expressions involving the eyes/eyebrows. The

most common facial behavior observed and deemed

relevant to pain by five participants was tearing or eyeweeping. Despite its high rating of relevance to pain,

many participants noted in the interview that this

behavior was sometimes hard to differentiate from

other emotions: ‘‘[H]e opened his eyes, and I don’t

know if it’s normal or not, but we could see a bit of

TABLE 1.

Interview Guide

Provoking factors: (situation surrounding pain) Tell me about a specific event you observed in the ICU duringwhich you believe your loved one experienced pain.

Do you believe they have had any pain during a procedure?Do you believe they have had any pain while being moved or

repositioned?Do you believe they have had any pain at rest?Why do you believe they experienced pain during these

events?Palliating factors: (factors or strategies relieving pain) What seemed to help relieve their pain? (i.e., medication,

rest)Why do you believe these factors or strategies relieved pain?

Quality/region: (specific behaviors associated with pain) Did your loved one react in any way to the painfulexperiences you described in the ICU?

Could you describe in detail what these behaviors looked orsounded like?

Were there any other facial expressions, bodymovements, orsounds made?

Severity: (intensity of pain behaviors) Did your loved one react differently to painful procedures thatseem to be more or less intense? If yes, please describehow these reactions differ.

Specificity: (specificity of behaviors with pain) Do you believe the pain behaviors you listed earlier are onlypresent in the case of pain?

In your opinion, has your loved one experienced pain duringan instance at the ICU but not shown any of thesebehaviors?

Have you seen your loved one also do any of the painbehaviors you listed when you believe they are notexperiencing pain?

Timing: (when and how long behaviors lasted) When have you observed these behaviors?How long did they last?How frequently did you see these behaviors?How long did these behaviors last after the procedure/event?

Understanding: (confidence) Why do you believe the behaviors you have observed arerelated to pain?

How well do you think you are able to detect pain in yourloved one?

Participant fills out Content Validation Questionnaire of the behavioral checklistInterview review and debrief: Now that you have looked at a detailed list of behaviors that

could be related to pain in different types of patients, canyou think of any other instances in the ICU that your lovedone showed these signs in response to pain? Pleasedescribe.

Those are all the questions I have for you today. Is thereanything you would like to add that we did not cover?

ICU ¼ intensive care unit.

5Family Perspectives of Patient Pain Behaviors

water coming out. Was it related to pain, or was itbecause he was seeing us there?’’ (04). Brow lowering,

an expression that involves lowering of the eyebrows,

was also believed to be a sign of pain.Other facial behav-

iors that involve the eyes were described in the qualita-

tive interview as relevant to pain—for example, eye

opening (n¼2), squinting/wincing (n¼2), andmoving

the eyes under closed eyelids (n ¼ 2).

Expressions involving the mouth. Two partici-pants described the patient biting his or her

endotracheal tube and how that can be indicative ofpain. A curling or moving of the lips was also seen as

a sign of pain (n ¼ 2): ‘‘And you know when you’re

in pain you’re like ‘Uhhh!’.His lip goes up’’ (01).

Flushing. A reddening of the patient’s face was

seen by two participants as a sign of pain, which could

be combined or not with other emotions: ‘‘Well I know

when they brought him out of sedation he reacted very

violently to it. His face was red.like he was really un-comfortable’’ (06).

TABLE 2.

Sample Characteristics

Participants (N ¼ 7) Total Median (Range)

Age, yr 54 (32-83)Gender 4 women, 3 menEducation level 5 college, 2 high schoolCountry of origin 6 Canada, 1 PortugalParticipant-Patient Relationship

Relationship to patient 3 spouses/partners2 children1 sibling1 parent

Length of relationship, yr 37 (8-60)Number of visits to ICU during currentadmission

10 (4-12)

Visit hours spent in the ICU 24 (11-75)Patients

Age, yr 55 (28-8)Gender 6 men, 1 womanTBI cause 4 falls

1 assault1 pedestrian accident1 motor vehicle accident

TBI severity 6 severe TBI1 moderate TBI

No. days since ICU admission, days 9 (4-11)Glasgow Coma Scale score 8 (7-10)

ICU ¼ intensive care unit; TBI ¼ traumatic brain injury.

6 Vanderbyl and G�elinas

General facial expressions. Three family caregivers

reported a general pain expression that they could not

describe further: ‘‘But we could tell that he felt things.

You know, like, you know when you struggle and your

facial expressions change’’ (06).

Body Movements. In terms of body movements,

touching or attempting to reach a pain site and general

restlessness were rated as relevant to patient pain onthe behavioral questionnaire by most participants.

Some participants also believed that when they

observed pulling of tubes or intravenous (IV) lines,

limb flailing, or restlessness, this was a sign their family

member was in pain. Other novel pain-related behav-

iors that were not included in the questionnaire were

instead described by participants in the qualitative

interview:Fighting restraints. Three participants noted that

another sign of pain was when the patients resisted re-

straint by equipment, staff, or family members: ‘‘Like he

didn’t want to be restrained because I could tell.he

was trying to really put up his head, and fight it!’’ (06).

Twitching or flinching. Three participants com-

mented on observing quick movements soon after a

pain stimulus: ‘‘He sort of jerked! Yah I think he feltthat! That was pain!.Like a twitch’’ (05).

General agitation. Two participants used agita-

tion to describe the pain behavior of their loved

one, which involved violently moving the limbs.

‘‘His two legs were moving, his arms were mo-

ving.And even one time, he was so agitated that

one of the nurses told me that his feet were totally

out of the bed’’ (04).

Other. Other behaviors described as relevant topain by two participants include lifting or turning the

head and twisting or turning the body. One participant

observed his family member exhibit behaviors compa-

rable to decortication and decerebration postures at

different time points after the TBI. The participant

thought these behaviors were pain related, despite

never having been witness to similar movements

before.Muscle Tension. Three participants rated moderate

or strong muscle tension, defined on the behavioral

questionnaire by how difficult it is to move the patient,

as relevant to pain in their loved one. During the inter-

view, however, four family caregivers instead described

muscle tension in terms of a visible tension they could

see without touching the patient’s body: ‘‘When he lifts

up his shoulder a bit, and his head, we saw a contrac-tion when he did that’’ (03).

TABLE 3.

Behavior Relevance to Pain

Facial Expressions Quotation in Interview

Rated Quite orVery Relevant to

Pain (n [%])

Eye weeping ‘‘The newest thing I’ve noticed is, I don’t know if they’re tears, or ifthey put something in her eye. So we’re not sure, but myDad assumes that she is trying to cry, or she has a tear.’’ (07)

5 [71]

Frowning/brow lowering ‘‘At the beginning I’d see the nurse rub his sternum, or touchpressure points on his feet. And you can see he reacts. Hefrowns.’’ (01)

‘‘While he was closing his eyes.his eyebrows went down, likethat.’’ (04)

3 [43]

Body MovementsAttempting to reach pain site ‘‘Well, like when she (the nurse) rubs his sternum, he goes to

reach.’’ (01)‘‘She would try to go towards her arm (where the IVs are), or go

towards her head (while intubated).’’ (07)

5 [71]

Touching pain site ‘‘It keeps on being the same arm, it keeps on going up to thebelly..I think he’s uncomfortable.’’ (06)

‘‘He took his left arm to where the pain is.’’ (03)

5 [71]

Restlessness ‘‘I know she was restless last night. maybe she feelsuncomfortable and she wants to move.’’ (07)

‘‘[H]e’s not comfortable. Even when people are not touching him,he’s moving.’’ (06)

5 [71]

Pulling tubes ‘‘If he’s not attached he’s going to pull everything out. At one pointall the intravenous in his arm had to be put in his chest becausehe had removed everything.. Either [it’s because the IVs areuncomfortable] or because of the pain, the general pain, and notknowing.where he is.’’ (04)

3 [43]

Flailing ‘‘[H]e can’t really flail with the right side, so it’s a lot with the left!.The whole arm!’’ (05)

3 [43]

Flexion ‘‘He flexes his toes in a way, well for me, I’m sure it’s because he’sin pain.’’ (01)

3 [43]

Muscle TensionMuscle resistance ‘‘Yah he sort of like tensed, you know?’’ (05) 3 [43]

IV ¼ intravenous.

Note: Only behaviors rated as relevant to pain by three or more participants on the questionnaires are shown.

7Family Perspectives of Patient Pain Behaviors

Ventilator Compliance or Vocalization Behav-iors. Behaviors having to do with ventilator compli-

ance (e.g., coughing) were not commonly reported

by participants as relevant to pain, and vocalization be-

haviors were not listed because all patients were intu-bated at the time of interview.

Painful Stimuli Leading to Expression of Behav-iors. The pain-related behaviors that participants

described were related to specific stimuli that they

perceived to be causing pain to the patient. These pain-

ful stimuli were related to the patients’ current medical

situation, their treatment or assessment, or their body

positioning.Medical situation. Most participants (n ¼ 5)

believed that an aspect related to the patient’s in-

juries or medical diagnosis was currently causing

the patient pain, even at rest. ‘‘Especially because

of all the multiple fractures that he has. Only there,

I’m sure he’s in pain’’ (04). While explaining behav-

iors of a patient with a recent craniotomy, one partic-

ipant stated that ‘‘.I’m sure it’s because he’s in pain.

But I think it’s just general, especially after his sur-

gery’’ (01).Treatments or assessments. Certain interventions

and physical assessments were viewed as painful by

those participants who believed their loved one had

experienced pain during his or her ICU admission.

Four participants had seen their relative be subject to

stimuli intended to cause pain as part of a neurologicexamination, such as rubbing the sternum, applying

nail bed pressure, or pinching, and they believed the

patient had indeed experienced these stimuli as pain-

ful. Two participants had observed endotracheal suc-

tioning, which they believed induced pain: ‘‘I can tell

8 Vanderbyl and G�elinas

you that what we saw that caused him pain was

when.they took out the secretions’’ (03). Two partic-

ipants believed that needles for intravenous infusions

or blood glucose testing were also causing pain and

inducing pain-related behaviors in patient. Two partic-

ipants observed that having the pain site touched by a

clinician as part of a physical assessment (e.g., palpa-tion) also caused their relative to exhibit pain

behaviors.

Positioning. Transferring or regular turning and re-

positioning of the patient were also perceived as pain-

ful by three of the family caregivers interviewed. On

the contrary, some participants (n ¼ 2) felt that their

hospitalized relative’s pain behaviors may have been

caused by sustained uncomfortable positioning orlying in bed for long periods.

Understanding Behaviors: Influencing andLimiting FactorsAlthough not addressed directly through the interview

questions, several themes related to influences on fam-

ily interpretations of pain emerged throughout the

participant interviews. The family caregivers’ under-

standing of behaviors as related to pain or not was

affected by many factors. They were influenced by in-formation they received from nurses and doctors, per-

sonal medical beliefs, and the personal history of the

patient.

Influencing Factors.Interpretations from doctors or nurses. Families

adopted the interpretation that doctors or nurses

gave them of certain observed behaviors that were dis-

cussed: ‘‘.When he put the hand inside [referring todecerebrate posturing], they [the nurses] told us that

[this behavior was] ‘Ohhh, no good’’’ (03).

Personal medical beliefs. The family caregiver’s

medical beliefs, particularly those stemming from pre-

vious experiences with people experiencing a TBI,

influenced participants’ interpretation of their loved

one’s situation: ‘‘I don’t think when you’re heavily

sedated like that you’re going to feel anything. Let’sface it! It’s like, you’re pretty much knocked out!’’

(05). Several participants tried to put themselves in

the shoes of their loved ones in imagining what they

would be feeling or why they would behave in certain

ways: ‘‘Like when you have a belly ache, especially us

women when we have cramps or whatever, we try

to rub our belly. I think that’s what he was trying to

do!’’ (06); ‘‘Or was he not comfortable in his position?You know? Imagine lying, like I know they move him

every couple hours, but it must be.I would be so rest-

less! So was it restlessness?’’ (05).

Personal history of the patient. With a median of

37 years of relationship between family caregivers

and their hospitalized relatives, participants had a great

deal of personal history to draw on to help decipher

the patient’s actions.

Pain history or previous self-report. Two patients

were not intubated immediately after their accident

and thus had a brief opportunity to self-report on the

presence or absence of pain during hospitalization.Family caregivers’ assessments of pain during the

time that patients were unable to report remained

congruent with the patient’s previous statements in

the ICU: ‘‘Cause I know she mentioned [pain from

her IVs] when she was responsive. And then later on

she kept on yanking the same place, so wewould think

it’s the same thing’’ (07). Remembering their relative’s

previous experiences with pain before the accidentalso helped all seven participants interpret behaviors

they recognized: ‘‘At home, he often had stomach-

aches, and he’d rub.[H]e probably has his stomach-

ache [while in the hospital] and he’s trying to rub it’’

(01). Two participants believed their relative’s previ-

ous high threshold to pain meant that any current

sign of pain was likely quite significant: ‘‘He could sus-

tain pain, that’s for sure!’’ (04).Patient personality. All participants considered

their relative’s personality before the accident when

considering the meaning of the behaviors they were

observing: ‘‘He likes to look around; very restless! So

I wasn’t surprised to see that [he was moving a lot in

the ICU]. I told my Mom, if he wasn’t tied to this

bed, he would’ve ran out of this hospital in a heartbeat.

That kind of person. Very restless.’’ (06); ‘‘[The pa-tient] is a relaxed person. If it was me, I know I’d

maybe feel it. But he, that’s his personality. He’s not a

high-strung person like I am. His personality is very

quiet.’’ (02).

Factors Impeding Observation or Interpretationof Behaviors. Some influencing factors actually pre-

vented family caregivers from or limited them in assess-

ing pain in ICU patients. Factors that impeded theobservation of pain-related behaviors included those

that were specific to the family member’s situation

and those that were related to the specific medical sit-

uation of the patient.

Family-related factors. Families did not often wit-

ness painful procedures or patient care because they

were not present at the bedside at that time, because

of either personal or staff preference. Many partici-pants were actively asked to leave the room by staff

to shield them from observing the possibly painful in-

terventions. For example, one participant said,

‘‘When I came in [the medical staff] asked me to leave

because [the patient] was in crisis and I came in by er-

ror’’ (04). In some cases this was in line with the partic-

ipants’ wish to not see their loved one in pain: ‘‘I

9Family Perspectives of Patient Pain Behaviors

always walk out of the room, because I don’t want to

see [endotracheal suctioning]. I choose to leave the

room. It’s not a very nice thing to see’’ (05). The read-

iness to observe unpleasant patient care may differ be-

tween family caregivers, as highlighted by one

participant: ‘‘But when they put something in [the pa-

tient’s] mouth to take out of the secretions, my wifedoesn’t like to see that. She turns around. Me, I want

to see’’ (03). Some participants also highlighted the

fact that they are not at the hospital at all times to be

able to observe for pain: ‘‘No. I’ve never seen it [refer-

ring to the patient neurological assessment]. Either

they do it when we’re not here, I don’t know. I’m

sure they do it, but we’re not here all the time’’ (02).

Family caregivers lack assessment knowledge. Par-ticipants noted that they did not always have the med-

ical knowledge to be able to assess for signs of pain

even when they did observe painful stimuli. ‘‘I don’t

know, I’m not a doctor’’ (05). Although technological

signals such as respirator alarms can be used by clinical

teams in pain assessment, these signals were not

salient or interpretable to family caregivers: ‘‘I haven’t

noticed any alarms. There’s things that beep, I don’tknow where they come from’’ (01).

Medical-related factors. Medical equipment,

illness-related factors, and sedatives limited the obser-

vation of pain behaviors by relatives. Patient restraint

use was mentioned by all participants; although these

measures are used for patient safety, they can limit the

distinction between specific patient behaviors, such as

localizing pain or performing random movements.Respirator equipment was also discussed by some par-

ticipants as limiting the observation of facial expres-

sions related to pain. ‘‘Yeah, like he was squinting

and that’s mainly the only thing he could see because

he had so much equipment in the face’’ (04). Possible

paralysis could inhibit some behavior observations, as

one participant noted that her relative ‘‘hasn’t been

moving his left side at all’’ (01), whereas seizures or se-vere edema could mask some pain-related behaviors or

muscle tension. Two participants commented on the

fact that sedation may act to limit pain behaviors but

not pain itself: ‘‘She’s sedated, so I don’t know, maybe

she doesn’t have the strength to move. Maybe if it is

pain, she wouldn’t have the strength to move I don’t

think!’’ (07).

Interpreting Behaviors. The pain observations andvarious influences affected the meaning that pain be-

haviors held for family members.

Pain vs. discomfort. All participants used both the

terms pain and discomfort when describing behaviors

they interpreted. For some family caregivers these

words were used interchangeably. However, others

distinguished between the two terms, reporting

some behaviors as being indicative of pain and others

signaling that the patient was instead uncomfortable.

Pain behaviors as a positive sign. Patient pain be-

haviors were not perceived as entirely negative to rel-

atives, because these behaviors represented signs of

patient mobility and a positive prognosis: ‘‘. he

even had to be restrained because he tries to grab athis tubes, you know. So that’s in a way a positive sign

because, God, we didn’t want him to be paralyzed’’

(06); ‘‘For me, it’s a good sign ’cuz she’s moving’’

(07). Conversely, a lack of behaviors was seen as nega-

tive by some participants, to the extent that they actu-

ally preferred seeing pain behaviors to a lack of

activity:

[We have] suspicions that he’s in a coma. I meanwe don’t know, we’re not medical people, but for a

man to be out three days and never opens his eyes,

and looks like he’s sleeping constantly, there’s some-

thing not well. (02).

Emotional Reactions to Pain BehaviorsObservation of pain behaviors coming from the criti-

cally ill relatives had an emotional impact on family

caregivers. All participants who believed their relativewas experiencing pain during hospitalization (n ¼ 6)

also expressed the emotional difficulties they had in

observing their loved one in pain: ‘‘I’m having trouble

seeing him.I don’t want to see him in pain, you

know!.It’s really hard. It’s a nightmare!’’ (05).

DISCUSSION

The present study is the first of its kind reporting onfamily caregivers’ perspectives of pain behaviors pro-

duced by nonverbal TBI patients in the ICU. Family

caregivers were able to describe several behaviors per-

formed by their critically ill relatives that they observed

and considered relevant to pain. Although this was the

primary aim of the study, other themes emerged when

family caregivers related their experience of observing

and understanding patient behaviors.The most commonly listed patient pain behaviors

by family caregivers could be sorted into three broad

categories: facial expressions, body movements, and

muscle tension. These three categories are already

included in many behavioral pain scales that have

been developed for critically ill adults (G�elinas et al.,

2013). Although body movements were the most

commonly described pain behaviors by family care-givers in this study, facial expressions such as brow

lowering or eye weeping were believed to be the

most compelling signs of pain.

Many of the specific behaviors that family care-

givers were able to describe in detail correspond to

10 Vanderbyl and G�elinas

behaviors described elsewhere in the critical care liter-

ature on pain (e.g., frowning/brow lowering or reach-

ing for pain site; G�elinas & Arbour, 2009; Puntillo et al.,

2004). It appears that family caregivers’ observations

may be appropriate and pertinent because they

overlap with some behaviors that ICU nurses

observe. This is not surprising considering the recentwork that has found that family members are fairly

accurate proxy reporters of patient symptoms in the

ICU (Puntillo et al., 2012), providing ratings of pain

more consistent to patient report than even doctors

or nurses. However, grimace, one of the most preva-

lent pain behaviors observed in critically ill patients

in other studies (Puntillo et al., 2004) and a behavior

commonly used in behavioral pain scales, was notablynot described by family caregivers of TBI patients in

our study. This result fits with a recent study by

Arbour and colleagues (2014), wherein few critically

ill TBI patients (<10%) were found to grimace when

observed undergoing a nociceptive procedure such

as turning.

Most importantly, family caregivers also describe

some of the ‘‘atypical’’ pain behaviors that TBI patientsexhibit that are not classically observed in other ICU

populations. For example, they noted that TBI patients

displayed eye weeping, limb flexion, and face flushing

in association with painful stimuli, some of the same

behaviors that were detected by trained observers in

two recent studies of TBI patients in the ICU by our

group (Arbour et al., 2014; Le et al., 2013). This may

help validate these atypical behaviors as related topain, because they have been identified by different

stakeholders as associated with pain. One notable

difference is that Arbour and colleagues (2014) found

that muscle tension is a behavior that, while normally

displayed in response to noxious stimuli by critically

ill patients, is rarely exhibited by TBI patients. Howev-

er, three of the family caregivers in our study reported

that they saw muscle tension in response to a stimulusthey perceived as pain related.

Furthermore, some novel pain behaviors that

were not listed on the behavioral questionnaire

used in this study were identified by family care-

givers, such as flinching/twitching or lip curling.

One participant described how the participant’s

loved one exhibited decorticate and decerebrate

posturing at different points during the ICU stay, be-haviors that the family caregiver believed to be indic-

ative of pain. Decorticate and decerebrate postures

are well-described signs of brain damage of the cor-

tex or midbrain, respectively, and thus are behaviors

that are very specific to neurologic injury (Davis &

Davis, 1982). The results from our study highlight

the fact that the content of existing behavioral pain

scales (G�elinas et al., 2013) may not be sufficient

to study possible behaviors of those with a brain

injury. The data support the idea that family care-

givers are able to offer insight into behaviors that

are possibly relevant to pain in this population of

TBI patients.

LimitationsThis study was subject to limitations, some of which

are related to the sample recruited. First, saturation

of data for the qualitative portion of the study was

not achieved with seven participants. The range of be-

haviors that can be observed in ICU patients in

response to painful stimuli is vast and also influenced

by the family caregivers’ intimate knowledge of the pa-

tient’s history and personality, making it challenging toachieve data saturation. Second, although we had

participation from family caregivers with a variation

of demographic characteristics, the patients they

were reporting on were more homogenous (i.e.,

mostly male patients with a severe TBI). Because all pa-

tients described were intubated, we are unable to

generalize our findings to nonintubated TBI patients

who cannot provide a self-report of pain for reasonsother than intubation.

Other limitations are related to methodologic is-

sues. The behavioral questionnaire used in this study

did not allow us to distinguish between two scenarios:

when a family caregiver observed a behavior but

believed it was irrelevant to pain, or when the behavior

was just not witnessed at all. Also, this questionnaire

was completed retrospectively by family caregiversduring a stressful period of their lives, a factor that

may have affected memory recall and the accuracy of

their responses. Family caregivers also appeared to

have different understandings of pain; for some this

was synonymous with discomfort, and for others these

were two different concepts. Clarification with partic-

ipants at the beginning of the interview on their inter-

pretation of the two terms would have allowed forbetter analysis of behaviors associated with pain and

other related concepts.

Family Perspectives and Implications forNursingCurrent pain assessment guidelines (Herr et al., 2011)

recommend that family caregivers be involved in the

assessment of pain in those who cannot self-report

on symptoms because of their unique point of viewof the patient situation. Family caregivers in our study

indeed had a personal and intimate perspective that

was implicated in their pain observations and assess-

ment. Although not the primary aim of our study, we

found that family caregivers used intimate knowledge

11Family Perspectives of Patient Pain Behaviors

of the patient, including personal and medical histories

and personality characteristics, to understand the

meaning of certain behaviors. These informal ob-

servers were also influenced by their own factors,

such as their medical knowledge or ability to empa-

thize with the patient. Family caregivers were also

influenced by the interpretations that nurses and doc-tors gave them on the possible meaning of specific be-

haviors; therefore, clinicians should be aware that

providing their perspectives may have an impact on

family understanding of the behaviors they observe.

Family caregivers also had particular limitations to

their observations of pain in the ICU. Some of these

limitations would affect any observer, such as patient

medical equipment blocking the view of facial expres-sions that could indicate pain. However, other limita-

tions may be particular to the family caregiver as an

observer. Families lacked medical knowledge and

were not often present to observe painful stimuli

such as procedures or transfers, either because of their

own decision to leave the bedside or at the request of

the medical team. Indeed, it is remarkable how much

family caregivers were able to add to the discussionon patient pain considering that the participants

were not often present in the patient room during

nociceptive stimuli. Many of the pain behaviors

observed in our study were detected by family care-

givers when the patient was actually at rest. During

validation of pain scales by researchers or clinical staff,

ICU patients at rest normally exhibit neutral behaviors

or an absence of muscle activity (G�elinas & Arbour,2009). Although this represents a difference between

this study and the literature, it may indicate that family

caregivers are particularly astute observers of

nonverbal patient pain. Nonetheless, the practice of

performing procedures when families are absent

from the bedside likely inhibits family caregivers’ abil-

ity to observe and interpret behaviors of critically ill pa-

tients. If we are to follow best practice guidelines that

suggest we engage family caregivers in pain assessment

of those patients who are nonverbal, we must recon-

sider the practice of shielding those family members

who are willing to observe painful instances even

when it is unpleasant. This can be further justified by

the unexpected finding that many family caregivers re-

ported finding some solace in observing pain behav-iors because they interpreted any type of patient

mobility or activity as a positive prognostic sign.

CONCLUSIONS

Properly assessing pain is the first step in the adequatetreatment of pain. Having family input to determine

what behaviors are related to pain is an important

step in following current pain assessment guidelines

for patients who cannot communicate. Furthermore,

frequent or novel pain behaviors determined by family

caregivers in this study should be considered when

revising current behavioral pain scales to make ver-

sions applicable and relevant to critically ill TBI patientswith an altered LOC. By enacting the collaboration of

family caregivers during pain assessment for this popu-

lation, ICU nurses may be better able to detect the pres-

ence of pain, adjust analgesic medication according to

the individual patient needs, and appraise the success

of pain management interventions.

Acknowledgments

The authors would like to thank Dr. Margaret Purden from

McGill University Ingram School of Nursing for helpful com-

ments on the study development and manuscript. We would

also like to acknowledge the nurse manager and the assistant

nurse managers of the ICU for supporting participant recruit-

ment and data collection. Finally, this study was supported by

an operating grant from the Fonds de recherche du Qu�ebec–Sant�e (FRQS grant #25094) and a Master’s Degree Award

from the Minist�ere de l’�Education, du Loisir et du Sport

(MELS).

REFERENCES

Arbour, C., Choini�ere, M., Topolovec-Vranic, J.,

Loiselle, C., Puntillo, K., & G�elinas, C. (2014). Detecting painin traumatic brain injured patients with different levels ofconsciousness exposed to common procedures in the ICU:Typical or atypical behaviors? The Clinical Journal of Pain,30(11), 960–969.

Arbour, C., & G�elinas, C. (2014). Behavioral and physio-logic indicators of pain in nonverbal patients with a trau-matic brain injury: An integrative review. Pain Management

Nursing, 15(2), 506–518.Barr, J., Fraser, G. L., Puntillo, K., Ely, E. W., G�elinas, C.,

Dasta, J. F., Davidson, J. E., Devlin, J. W., Kress, J. P.,Joffe, A. M., Coursin, D. B., Herr, D. L., Tung, A.,Robinson, B. R., Fontaine, D. K., Ramsay, M. A., Riker, R. R.,

Sessler, C. N., Pun, B., Skrobik, Y., & Jaeschke, R. (2013).Clinical practice guidelines for the management of pain,agitation, and delirium in adult patients in the intensive careunit. Critical Care Medicine, 41(1), 263–306.Chanques, G., Jaber, S., Barbotte, E., Violet, S.,

Sebbane, M., Perrigault, P. F., Mann, C., Lefrant, J. Y., &Eledjam, J. J. (2006). Impact of systematic evaluation of painand agitation in an intensive care unit. Critical Care Medi-

cine, 34(6), 1691–1699.Chanques, G., Sebbane, M., Barbotte, E., Viel, E.,

Eledjam, J. J., & Jaber, S. (2007). A prospective study of painat rest: Incidence and characteristics of an unrecognizedsymptom in surgical and trauma versus medical intensivecare unit patients. Anesthesiology, 107(5), 858–860.

12 Vanderbyl and G�elinas

Davis, R. A., & Davis, L. (1982). Decerebrate rigidity inhumans. Neurosurgery, 10(5), 635–642.

De Jong, A., Molinari, N., De Lattre, S., Gniadek, C., Carr, J.,Conseil, M., Susbielles, M. P., Jung, B., Jaber, S., &Chanques, G. (2013). Decreasing severe pain and seriousadverse events while moving intensive care unit patients: Aprospective interventional study (the NURSE-DO project).Critical Care, 17(2), R74.

Faul, M., Xu, L., Wald, M. M., & Coronado, V. G. (2010).Traumatic Brain Injury in the United States: Emergency

Department Visits, Hospitalizations and Deaths 2002–

2006. Atlanta: Centers for Disease Control and Prevention,National Center for Injury Prevention and Control.

G�elinas, C. (2007). Management of pain in cardiac surgeryICU patients: Have we improved over time? Intensive &Critical Care Nursing, 23, 298–303.

G�elinas, C., & Arbour, C. (2009). Behavioral and physio-logic indicators during a nociceptive procedure inconscious and unconscious mechanically ventilated adults:Similar or different? Journal of Critical Care, 24(4).628.e7–628.e17.

G�elinas, C., Arbour, C., Michaud, C., Valliant, F., &Desjardins, S. (2011). Implementation of the Critical-CarePain Observation Tool on pain assessment/managementnursing practices in an intensive care unit with nonverbalcritically ill adults: A before and after study. InternationalJournal of Nursing Studies, 48(12), 1495–1504.

G�elinas, C., Fillion, L., Puntillo, K., Viens, C., & Fortier, M.(2006). Validation of the Critical-Care Pain Observation Toolin adult patients. American Journal of Critical Care, 15(4),420–427.

G�elinas, C., Puntillo, K. A., Joffe, A., & Barr, J. A. (2013). Avalidated approach to evaluating psychometric properties ofpain assessment tools for use in non-verbal critically illadults. Seminars in Respiratory and Critical CareMedicine,

34(2), 153–168.Graneheim, U. H., & Lundman, B. (2004). Qualitative

content analysis in nursing research: Concepts, proceduresand measures to achieve trustworthiness. Nurse EducationToday, 24(2), 105–112.

Herr, K., Coyne, P. J., McCaffery, M., Manworren, R., &Merkel, S. (2011). Pain assessment in the patient unable toself-report: Position statement with clinical practice recom-mendations. Pain Management Nursing, 12(4), 230–250.

Le, Q., G�elinas, C., Arbour, C., & Rodrigue, N. (2013).Description of behaviours in nonverbal critically ill patientswith a when exposed to common procedures in the inten-sive care unit: A pilot study. Pain Management Nursing,

14(4), e251–e261.Lefebvre, H., Cloutier, G., & Jos�ee Levert, M. (2008). Per-

spectives of survivors of traumatic brain injury and theircaregivers on long-term social integration. Brain Injury,

22(7-8), 535–543.Myhren, H., Toien, K., Ekeberg, O., Karlsson, S.,

Sandvik, L., & Stokland, O. (2009). Patients’ memory andpsychological distress after ICU stay compared with

expectations of the relatives. Intensive Care Medicine,

35(12), 2078–2086.Payen, J. F., Bosson, J. L., Chanques, G., Mantz, J.,

Labarere, J., & DOLOREA Investigators (2009). Painassessment is associated with decreased duration of me-chanical ventilation in the intensive care unit: A post hocanalysis of the DOLOREA study. Anesthesiology, 111(6),1308–1316.Powell, R. A., Downing, J., Ddungu, H., & Mwangi-

Powel, F. N. (2010). Guide to pain management inlow-resource settings. In A. Kopf, & N. B. Patel (Eds.), Painhistory and pain assessment. Seattle: International Associ-ation for the Study of Pain.Puntillo, K. A., Max, A., Timsit, J. F., Vignoud, L.,

Chanques, G., Robleda, G., Roche-Campo, F., Mancebo, J.,Divatia, J. V., Soares, M., Ionescu, D. C., Grintescu, I. M.,Vasiliu, I. L., Maggiore, S. M., Rusinova, K., Owczuk, R.,Egerod, I., Papathanassoglou, E. D., Kyranou, M., Joynt, G. M.,Burghi, G., Freebairn, R. C., Ho, K. M., Kaarlola, A.,Gerritsen, R. T., Kesecioglu, J., Sulaj, M. M., Norrenberg, M.,Benoit, D. D., Seha, M. S., Hennein, A., Periera, F. J.,Benbenishty, J. S., Abroug, F., Aquilina, A., Monte, J. R., An, Y.,& Azoulay, E. (2014). Determinants of procedural pain in-tensity in the intensive care unit: The Europain study.American Journal of Respiratory & Critical Care Medicine,

189(1), 39–47.Puntillo, K. A., Morris, A. B., Thompson, C. L., Stanik-

Hutt, J., White, C. A., & Wild, L. R. (2004). Pain behaviorsobserved during six common procedures: Results fromThunder Project II. Critical Care Medicine, 32(2), 421–427.Puntillo, K. A., Neuhaus, J., Arai, S., Paul, S. M.,

Gropper, M. A., Cohen, N. H., & Miaskowski, C. (2012).Challenge of assessing symptoms in seriously ill intensivecare unit patients: Can proxy reporters help? Critical CareMedicine, 40(10), 2760–2767.Puntillo, K. A., White, C., Morris, A. B., Perdue, S. T., Sta-

nik-Hutt, J., Thompson, C. L., & Wild, L. R. (2001). Patients’perceptions and responses to procedural pain: Results fromThunder Project II. American Journal of Critical Care,

10(4), 238–251.Rose, L., Haslam, L., Dale, C., Knechtel, L., & McGillion, M.

(2013). Behavioral pain assessment tool for critically ill adultsunable to self-report pain. American Journal of Critical

Care, 22(3), 246–254.Roulin, M. J., & Ramelet, A. S. (2014). Behavioral

changes in brain-injured critical care adults with differentlevels of consciousness during nociceptive stimulation:An observational study. Intensive Care Medicine, 40,1115–1123.Sandelowski, M. (2000). Focus on research methods:

Combining qualitative and quantitative sampling, datacollection, and analysis techniques in mixed-methodsstudies. Research in Nursing & Health, 23, 246–255.Teasdale, G., & Jennett, B. (1974). Assessment of coma and

impaired consciousness. A practical scale. Lancet, 2(7872),81–84.