University of Groningen Physical activity and depressive … · 2016. 3. 9. · Nikolaos Stavrakakis born on 23 June 1981 in Cholargos, Greece to obtain the degree of PhD at the University

University of Groningen

Physical activity and depressive symptomsStavrakakis, Nikolaos

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite fromit. Please check the document version below.

Document VersionPublisher's PDF, also known as Version of record

Publication date:2015

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA):Stavrakakis, N. (2015). Physical activity and depressive symptoms: is a healthy body necessary for ahealthy mind?. [S.n.].

CopyrightOther than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of theauthor(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

Take-down policyIf you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediatelyand investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons thenumber of authors shown on this cover page is limited to 10 maximum.

Download date: 26-05-2021

https://research.rug.nl/en/publications/physical-activity-and-depressive-symptoms(625e1e97-2fcd-4ee7-b6da-ea62753c8c11).html

Physical Activity and Depressive Symptoms

Is a healthy body necessary for a healthy mind?

Nikolaos Stavrakakis

Physical activity and depressive symptoms


PhD thesis

to obtain the degree of PhD at the

University of Groningen on the authority of the

Rector Magnificus Prof. E. Sterken and in accordance with

the decision by the College of Deans.

This thesis will be defended in public on

Monday 9 March 2015 at 16.15 hours

by


born on 23 June 1981 in Cholargos, Greece

The infrastructure for the TRacking Adolescents' Individual Lives Survey (www.TRAILS.nl) has been financially supported by grants from the Netherlands Organization for Scientific Research NWO (Medical Research Council program grant GB-MW 940-38-011; ZonMW Brainpower grant 100-001-004; ZonMw Risk Behaviour and Dependence grants 60-60600-97-118; ZonMw Culture and Health grant 261-98-710; Social Sciences Council medium-sized investment grants GB-MaGW 480-01-006 and GB-MaGW 480-07-001; Social Sciences Council project grants GB-MaGW 452-04-314 and GB-MaGW 452-06-004; NWO large-sized investment grant 175.010.2003.005; NWO Longitudinal Survey and Panel Funding 481-08-013; NWO Vici 016.130.002; NWO Gravitation 024.001.003), the Dutch Ministry of Justice (WODC), the European Science Foundation (EuroSTRESS project FP-006), Biobanking and Biomolecular Resources Research Infrastructure BBMRI-NL (CP 32), the Gratama foundation, the Jan Dekker foundation, the participating universities, and Accare Centre for Child and Adolescent Psychiatry.

Financial support for the publication of this dissertation was kindly provided by the research school of Behavioural and Cognitive Neurosciences (BCN), the Rijksuniversiteit Groningen and University Medical Center Groningen.

Design and printing METROPOLIS Graphic Arts S.A., Athens, Greece

© 2015 Nikolaos Stavrakakis, Athens, Greece

All rights reserved. No part of this dissertation may be reproduced, stored or transmitted in any form or by any means without the written permission of the author.

ISBN (print): 978-90-367-7600-4ISBN (digital): 978-90-367-7599-1

Physical Activity and Depressive SymptomsIs a healthy body necessary for a healthy mind?

PhD thesis



by


born on 23 June 1981in Cholargos, Greece

to obtain the degree of PhD at theUniversity of Groningenon the authority of the

Rector Magnificus Prof. E. Sterkenand in accordance with


Physical activity and depressive symptoms


PhD thesis

to obtain the degree of PhD at the

University of Groningen on the authority of the

Rector Magnificus Prof. E. Sterken and in accordance with




by


born on 23 June 1981 in Cholargos, Greece

SupervisorsProf A.J. Oldehinkel

Prof P. de Jonge

Co-supervisorDr. A.M. Roest

Assessment committeeProf A. Steptoe

Prof B.W. Penninx

Prof S.A. Reijneveld

ParanymphsBertus F. Jeronimus

Anna-Roos E. Zandstra

Table of Contents

CHAPTER I General Introduction 9

CHAPTER II Bidirectional Prospective Associations between Physical 27 Activity and Depressive Symptoms. The TRAILS Study

CHAPTER III Physical Activity and Onset of Depression in Adolescents: 43 A Prospective Study in the General Population Cohort TRAILS

CHAPTER IV Plasticity Genes do not Modify Associations between Physical 55 Activity and Depressive Symptoms

CHAPTER V Competitive Sport Participation, Sport Competence and 73 Depressive Symptoms in Adolescent Boys and Girls

CHAPTER VI Relative-Age Effects in Adolescents: The TRAILS Study 99

CHAPTER VII The Dynamic Relationship between Physical Activity and 127 Mood in the Daily Life of Depressed and Non-Depressed Individuals: a Within-Subject Time-Series Analysis

CHAPTER VIII Summary and General Discussion 147

Summary in English 171

Summary in Dutch 177

Acknowledgements 183

About the Author 187

List of Publications 191

Recent TRAILS Dissertations 195

“Mens sana in corpore sano”(A healthy mind existing in a healthy body)

Juvenal, Satire X

General Introduction

CHAPTER

13GENERAL INTRODUCTION

GENERAL INTRODUCTION

Physical activity (PA) is considered to be beneficial to general health. However, the benefits

of PA on mental health, and especially depression are less clear. This thesis explores in

depth the relationship between PA and depressive symptoms, in adolescents and adults.

Definition of Physical Activity and Depression

It is important to clarify some of the terms that will be used in this thesis, namely, PA, physi-

cal fitness (PF)1, exercise, sports and depression.

Physical Activity, Physical Fitness, Exercise and Sports

The main two distinctive groups are PA and PF, while exercise and sports are subgroups of

PA. PF is defined as “the ability to carry out daily tasks with vigor and alertness, without

undue fatigue and with ample energy to enjoy leisure-time pursuits and to meet unforeseen

emergencies” (Caspersen et al., 1985; see p. 128). Hence, PF reflects health attributes (for

example, cardiovascular endurance, muscular endurance and strength, flexibility), as well as

skills related to athletic ability (accuracy, spatio-temporal coordination, etc.). PA is defined

as “any bodily movement produced by skeletal muscles that results in energy expenditure”

(Caspersen et al., 1985; see p. 126). The energy expenditure is commonly measured in kilo-

calories (kcal) or kilojoules (kJ). PA can be classified in many different ways. The simplest

way to classify PA is by distinguishing between activity while working and activity during

leisure-time (voluntary activity)2. Further classifications include light, moderate, and inten-

sive activities, team activities, and outdoor activities. The intensity of each activity can be

expressed in the Metabolic Equivalent of Tasks (MET). The MET score assesses the energy

cost of PA and is defined as “the ratio of the associated metabolic rate for the specific activ-

ity divided by the resting metabolic rate (RMR)3” (Ainsworth et al., 1993; see page 72)4.

Although PA and exercise share common characteristics and have been used interchange-

ably in the literature, exercise is a distinct subcomponent of PA. Exercise is defined as “any

physical activity that is planned, structured, repetitive, and purposive in the sense that

improvement or maintenance of one or more components of physical fitness is (the) objec-

tive” (Caspersen et al., 1985; see page 128). The term ‘sports’ does not have a universal

definition, but the majority of definitions on what constitutes a sport tend to converge on a

competitive element. The Oxford dictionary defines sport as: “an activity involving physical

exertion and skill in which an individual or team competes against another or others for

entertainment”5. Thus, the term ‘sports’ is also a subcomponent of PA and shares com-

mon traits with exercise.

CHAPTER I14

Depression

Depression has been defined in diverse ways in the literature (Abela & Hankin, 2008) and is

considered a largely heterogeneous disorder (Hasler, Drevets, Manji, & Charney, 2004; Lux &

Kendler, 2010). Generally, depression is characterized as a state of low mood or a loss of

interest, which affects an individual’s well-being, cognition and behavior (Joorman, 2009).

According to DSM-V criteria for major depressive disorder (MDD), at least one of the symp-

toms of depressed mood and loss of interest or pleasure has to be present for 2 weeks. At

least five of the following nine additional symptoms should also persist almost every day

for at least 2 weeks: a) appetite problems, that is, weight loss or weight gain, or decrease

or increase in appetite; b) sleep difficulties, such as hyper-insomnia or insomnia; c) fatigue

or loss of energy; d) feelings of worthlessness; e) psychomotor agitation; f) cognitive prob-

lems, such as diminished ability to concentrate and indecisiveness; and g) suicidal ideation

or suicide attempts (American Psychiatric Association, 2013). The prevalence rates of MDD

are higher than of all other psychiatric disorders, with approximately 19% of the Dutch

population experiencing a clinically significant episode at least once in their lives6 (de Graaf,

ten Have, van Gool, & van Dorsselaer, 2012).

Humans Evolved to Move

PA has been at the core of every free-living organism’s evolutionary history. The rela-

tionship between caloric acquisition (energy consumed as food) and caloric expenditure,

i.e., PA, has exerted an “ongoing adaptive pressure that affected (the) selection of genes

related to the cardio-respiratory and musculoskeletal systems as well as the internal

metabolism processes of our progenitors” (Eaton & Eaton, 2003; see p. 153). This can be

seen in the evolution of the human species, where the environmental pressures that have

acted in order to produce our unique human nature have taken place immediately after the

‘split’ between our pre-human ancestors with those of the modern day apes (pongids),

an estimated 5-7 million years ago (Cordain et al., 1998; Eaton & Konner, 1985). The ap-

pearance of the Australopithecines7, approx. 4 million years ago8 coincided with gradual

environmental changes in the African continent, which resulted in a ‘drying’ up of tropical

forests to a more open savanna (Behrensmeyer & Cooke, 1985; deMonocal & Bloemen-

dal, 1995; Foley, 1987). The early hominins developed specific structural and functional

characteristics, which helped them exploit their changing environment in ways that their

pongid ancestors could not (see for example: Leonard & Robertson, 1997). The first such

change was the ability to walk straight (bipedalism9), which was still relatively primitive

compared to the abilities of (much) later hominins (i.e., Homo habilis, Homo erectus and

Homo sapiens; McΗenry, 1994).


Bipedalism conferred clear advantages in this new environment, such as facilitated visual

detection of food, water and predators (Shipman, 1986), as well as enabling weapon or tool

use while walking (Cordain et al., 1997; Lovejoy, 1988). The biggest advantage, however, of

bipedalism is considered to be the development of a more complex eccrine sweat gland sys-

tem, which in turn allowed our early ancestors to walk or run over larger distances in hot

climates without overheating (Cordain et al., 1997; Wheeler, 1991)10. Bipedalism paved the

way for further changes that are seen with the rise of the genus Homo. Homo habilis had a

body size similar to the Australopithecines, but with a markedly increased cranial capacity11

(Cordain et al., 1997; Lieberman, 2011; Raichlen & Polk, 2013)12. Homo habilis was succeeded

by Homo erectus who showed more similarities in body size, physique and cranial capacity

to modern humans13 (Brown, Harris, Leakey, & Walker, 1985; McΗenry, 1994; Ruff, 1991). It

is estimated that Homo erectus had better aerobic capacities than the pongids and early

hominins (Australopithecines)(Cordain et al., 1997).

High levels of sustained activity are primarily dependent on an efficient cardiovascular sys-

tem, which can be inferred from the structure of the thoracic cage and an efficient system for

heat dissipation (Aiello & Wheeler, 1995). In Homo the upper part of the rib is able to enlarge

the thorax during inhalation (Aiello & Wheeler, 1995; Cordain et al., 1997) while in Australo-

pithecines and earlier species the rib cages can be narrowed during respiration in order to

accommodate the muscle groups responsible for arboreal locomotion, namely the pectoral

girdle (Aiello & Wheeler, 1995; Schmid, 1991). Since ventilation of the lungs depends on the

movement of the diaphragm, the ability of Homo to enlarge the thorax suggests that they

might have had a more efficient ventilation system and in turn a more efficient cardiovascu-

lar system for prolonged activities compared to pongids and Australopithecines (Aiello &

Wheeler, 1995). Moreover, Homo erectus also showed a relative decrease in maximal pelvic

breadth relative to stature (Ruff, 1991), which would favor better heat dissipation during

intensive physical activities (such as running) even though the absolute body size had in-

creased from earlier hominids (Cordain et al., 1997).

Because of bipedalism and efficient thermoregulation, the increased encephalization14 of

our closest ancestors progressed simultaneously with more complex behaviors and physi-

ological changes (see for example: Bingham, 200015; Dunbar, 2003), such as a larger daily

range for searching for resources16 (Leonard, 2010) and an increase in daily total energy

expenditure17 as well as an increase in size and height (Cordain et al., 1998). Larger brains

inevitably require more energy18 and through improvements in aerobic capacity of early

Homo, larger areas could be forested to obtain high-energy diets. Raichlen and Polk (2013)

have argued that increases in PA and aerobic capacity might have influenced the encepha-

lization of modern humans directly19. This is in accordance with the hypothesis that early

Homo, in order to compete with other scavengers for food, evolved a high specialization in

CHAPTER I16

long-distance running (Bramble & Lieberman, 2004). This adaptation would have ensured

a high quality food supply, which would allow for the development of a larger brain. Re-

gardless of what came first (larger brains or higher activity), it is clear that PA has played

a large part in the evolution of modern humans.

PA is not an important aspect of modern humans’ daily lives, in terms of food security, and

the relationship between energy consumption and energy expenditure of modern humans

has been abrogated (Cordain et al., 1997; Eaton & Eaton, 2003; Katzmarzyk, 2010). The radi-

cal advances of new technologies, the global and organized domestication of animals and

the facilitation of commerce and trading have made food sources more abundant and the

exertion required to obtain food supplies is lower than in any other period in human history.

The estimated ratio of energy consumed over energy expended due to PA (energy efficiency

ratio) is 2.25 (i.e., 1kJ of energy expenditure to consume 2.25kJ of food) for a 57-kg Stone

Age human, and 3.66 for a 64-kg contemporary human (Eaton & Eaton, 2003). Essentially,

modern humans spend less energy and have a higher energy input (i.e., consume more food)

than their predecessors.

Additionally, humans living in modern societies are considerably less physically active than

modern hunter-gatherers (Katzmarzyk, 2010). Studies investigating PF and PA of Eskimo

communities (Igloolik, northern Canada), which have transitioned to a more Western life-

style, show a reduction of PF and PA over time (from 1970 to 1990) in all age groups (Rode &

Shephard, 1984; Rode & Shephard, 1994). Furthermore, a recent study comparing the energy

expenditure of humans living a modern lifestyle and the Old Order Amish population, who

reject modern technology, show large differences, with humans living a modern lifestyle be-

ing much more sedentary than the Amish (Bassett, Schneider, & Huntington, 2004).

In sum, even though PA has played a large part in human evolution, it has been reduced

markedly in modern societies. To illustrate this point, the World Health Organization (WHO)

recommends at least 150 minutes of moderate or light activity or at least 75 minutes of vig-

orous activity per week, but recent reports suggest that the average individual20 does not

adhere to this recommendation (Biddle & Goudas, 1996; Brawley & Rodgers, 1993; Desha et

al., 2007; Sagatun et al., 2007; WHO, 201321). Our genetic evolution is slower than the cultural

changes that have taken place over the years and, as a result, our genes have remained

adapted to the environmental circumstances of the past (Gould, 1980; p. 83). The health

implications of this mismatch between human gene evolution (‘hunter-gatherer genes’22)

and human culture (‘21st century lifestyle’) are numerous and well documented (Cordain et

al., 1997; Cordain et al., 1998; Eaton & Eaton, 2003; Katzmarzyk, Church, Craig, & Bouchard,

2009; Katzmarzyk, 2010).


Overview of Physical Activity and Health Research

From times dating as far back as Hippocrates (460 – c. 370 BC), PA has been considered a key

factor for good health. Galen (129 – c. 210 AD) structured his medical theory around three

concepts, the “naturals” (i.e., physiology), the “non-naturals” (i.e., health) and the “contra-

naturals” (i.e., pathology). He primarily focused on the uses and abuses of the “six things non-

natural”23, which he thought were instrumental for health. These included: a) air; b) diet; c)

sleep and waking; d) exercise (motion) and rest; e) excretions and retentions; and f) passions

of the mind. Galen suggested that practicing the non-naturals (also known as the “six laws of

health”) in moderation would sustain health and prevent illness, whereas if these laws were

not followed, or followed in excess, disease would be the result (Berryman, 1989; Galen, 1991).

The modern scientific field of PA research began in the early 1950s in London. Morris et al.

(1953) investigated the mortality rates of London Transport Executive and Post-Office em-

ployees, and observed lower mortality rates from coronary problems in physically active

employees (bus conductors and post-men) than in their sedentary colleagues (bus-drivers

and telephone switchboard operators)24. Since then, research on the potential health ben-

efits of PA has been flourishing. Early life PA has been shown to promote the formation of

dense, well-mineralized bones (Burr, Ruff, & Thompson, 1990), as well as influence the bone

structural geometry (Larsen, 1997). Moreover, an inverse association exists between PA and

all-cause mortality (Paffenbarger, Hyde, Wing, & Hsieh, 1986; Powell & Blair, 1994; Salonen,

Puska, Kottke, Tuomilehto, & Nissinen, 1983), cardiac events such as myocardial infarction

(Balady, 2002; Myers et al., 2002), hypertension (Fagard, 2001), obesity25 (Prentice & Jebb,

2000), and type II diabetes (Hardman & Stensel, 2003). Furthermore, PA has been associated

with benefits to the immune system26 (Pedersen & Hoffman-Goetz, 2000; Shephard & Shek,

1994) and brain function; since it increases the blood and oxygen flow in the brain (Dishman

et al., 2006). Finally, PA has been positively associated with neurogenesis, gliogenesis and

neuroprotection in some areas of the animal and human brain (van Praag, Christie, Sejnowski,

& Gage, 1999; van Praag, Kempermann, & Gage, 1999; van Praag, 2008), as well as with cogni-

tive improvements in young children, adolescents (Chaddock, Hillman, Buck, & Cohen, 2011;

Hillman, Erickson, & Kramer, 2008), and older adults (Colcombe & Kramer, 2003; Colcombe et

al., 2004). The hypothesized mechanism for these benefits is the activation of neurotrophins

and other growth factors, such as Brain Derived Neurotrophic Factor (BDNF).

The recognition that PA has widespread benefits on (physical) health has led to the explora-

tion of whether the benefits of PA can also extend to well-being27 and mental health28. The

first epidemiological studies showed encouraging results. Stephens (1988), in a large study

of over 55,000 individuals, concluded that there was a clear positive relationship between

PA and well-being, positive mood and lower levels of anxiety and depression. Several later

CHAPTER I18

studies have also shown a positive relationship between PA and mental health (Deslandes

et al., 2009; Paluska & Schwenk, 2000).

Depression imposes a burden on the health of individuals and work place productivity. The

poor physical and social functioning of patients with a depressive disorder or sub-clinical de-

pressive symptoms has been shown to be comparable to (or even worse) than the function-

ing of patients with major chronic medical conditions (Merikangas et al., 2007; Wells et al.,

1989). Studies focusing on the impact of depression on the workplace (Kessler et al., 2006),

in terms of annual salary equivalent costs, have estimated that the cost of depression-

related loss of productivity in the USA is approximately $36.5 million per year. This is likely

to be an underestimation of the true cost to society.

The effectiveness of depression treatments such as anti-depressant medication and be-

havioral therapies is still being widely debated (Cuijpers, van Straten, Bohlmeijer, Hollon, &

Andersson, 2010; Cuijpers, Andersson, Donker, & van Straten, 2011; Khin, Chen, Yang, Yang,

& Laughren, 2011; Kirsch et al., 2008; Turner, Matthews, Linardatos, Tell, & Rosenthal, 2008).

Therefore, there is a need for alternative effective treatments for depression, especially low

cost treatments, with PA being a primary candidate. The UK National Institute for Clinical Ex-

cellence (NICE) guideline for depression recommends structured, supervised exercise three

times a week (45 to 60 minutes) for 10 to 12 weeks for mild depression (National Institute

for Health and Clinical Excellence, 2009). Although these Exercise Referral Schemes have

been implemented, the empirical evidence of their effectiveness on depression is not robust.

For several decades now, researchers have examined the potential use of PA as an effective

treatment for depression (Mead et al., 2009). Recent systematic reviews and meta-analyses

of randomized controlled studies (RCTs) have shown mixed results. Krogh et al. (2010) con-

cluded their meta-analysis by stating that there is a short-term benefit of exercise in clinically

depressed patients, but very little evidence of a long-term benefit. Another meta-analysis of

RCTs by Rimer et al. (2012) concluded that exercise seems to improve depressive symptoms in

clinically depressed patients when compared to no treatment or a control condition. However,

when only methodologically high-quality studies were included, the effect of exercise became

much smaller. Mead et al. (2009) and Cooney et al. (2013) reached a similar conclusion: when

only studies fulfilling three strict methodological criteria (i.e., allocation concealment, blinded

outcome assessment and intention to treat trials) were analyzed, the effect sizes observed

became smaller and not statistically significant. A recent large intervention study (Chalder et

al., 2012) did not find any evidence that PA is beneficial for depressive symptoms29. However,

intervention studies are only valuable in delineating whether a specific treatment is effective

for depression, and do not provide information on the numerous factors that might influence

the risk of developing depression and the progression of the disorder.


Since the effect of PA on depressive symptoms has been reported frequently but seems

to be moderate to weak and may not even exist at all in some populations, there is a need

for a better theoretical understanding of this relationship, in order to derive the best ways

to maximize the potential benefit of PA in treating depression. Epidemiological studies in

representative populations can be useful in elucidating multifactorial causes of the disorder

and can help to provide a better theoretical framework of the relationship between PA and

depressive symptoms.

Research Gaps

Most studies conducted so far have focused on adults. Studies on adolescents have been

lacking; only recently has there been an increase in studies investigating the relationship

between PA and depressive symptoms in adolescents. Adolescence is a period of radical

physical, cognitive, and emotional change, characterized by a high incidence of depression

(Hankin et al., 1998). Youth depression and adult depression are potentially different due

to age-related differences in cognition, emotion, behavior, and physical development; so

adult models of depression cannot be generalized automatically to children and adolescents

(Abela & Hankin, 2008). Therefore, it is important to better understand the development and

trajectory of depressive symptoms and their relationship with PA during adolescence.

Observational cross-sectional and prospective studies have shown mixed results in the

relationship between PA and depressive symptoms in adolescents. In a meta-analysis of

observational studies, Biddle and Asare (2011) showed that PA has a potentially beneficial

effect on depressive symptoms, but the associations were weak and most studies suffered

from serious methodological limitations. Johnson and Taliaferro (2011) reached a similar

conclusion in their meta-analysis. Small to moderate protective effects of PA or sport par-

ticipation on depressive symptoms were observed, but methodological limitations plagued

the majority of these studies.

Besides obvious methodological problems related to this type of research, such as discrep-

ant measures of depressive symptoms and PA and an over-reliance on cross-sectional de-

signs (Johnson & Taliaferro, 2011), other important issues need to be addressed. One issue is

the direction of the association. It is possible that PA helps to alleviate depressive symptoms

(protective hypothesis), but also that depressive symptomatology hinders PA participation

(inhibition hypothesis). Perhaps both, i.e., bidirectional associations, are true. Consideration

of potentially moderating factors such as genetic or psychosocial variables is also vital, as

these might make individuals more prone to benefit from PA, or be more adversely impacted

by physical inactivity.

CHAPTER I20

Aim and Objectives of this Thesis

The overall aim of this thesis is to provide a deeper understanding of the association be-

tween PA and depressive symptoms in adolescents and adults. The study in chapter II in-

vestigates the direction of the relationship between PA and depressive symptoms, while the

study in chapter III investigates the hypothesis that early engagement in PA in adolescence

might prevent an onset of a major depressive episode in early adulthood, and explores the

role of specific characteristics of PA, i.e., the nature, frequency, intensity and duration.

The second part of the thesis focuses on possible moderators of the relationship between

PA and depressive symptoms. Chapter IV examines genetic factors that might influence the

extent to which adolescents benefit from PA, i.e., genes that work on important neuromodu-

lators (such as serotonin and dopamine) and neurotrophins (BDNF). These neuromodulators

and neurotrophins are thought to play an important role in the development and course of

depression and also to be affected by PA. Psychosocial moderators and their influence on

the relationship between PA and depressive symptoms are examined in chapter V and VI.

The study in chapter V investigates whether peer-nominated sport competence and gen-

der moderate the association of sport participation with current depressive symptoms and

symptom changes over time. Chapter VI concerns the relative-age position of adolescents

among their classmates, which might affect the way they are perceived on their athletic

abilities by teachers, and their levels of depressive symptoms.

The final part considers individual differences in the association between PA and mood in

adults. Chapter VII evaluates individual differences in the relationship between PA, as as-

sessed with accelerometers, and repeated daily assessments of positive and negative af-

fect in depressed and non-depressed adults. In this study, the direction of the relationship

between PA and daily mood is investigated at the individual rather than at the group level.

In chapter VIII the main results of the research undertaken are summarized and discussed

further, along with clinical implications and possible directions for future research.

Summary of Samples Used

The TRAILS Sample

Five out of the six empirical studies in this thesis are based on data from the Dutch TRack-

ing Adolescents’ Individual Lives Survey (TRAILS). TRAILS is a prospective population-based

cohort of Northern Dutch adolescents. The general aim of TRAILS was to delineate etiologies

and trajectories of (mental) health problems in the Dutch population. Data collection for the

baseline measurement (T1) started in 2001 and finished in 2002. Data collection for the second

(T2), third (T3) and fourth (T4) measurement waves took place at intervals of approximately

2.5 years. For this thesis, data have been used from all four measurement waves (T1 to T4). At


T1, 2230 adolescents agreed to participate. The response rates at T2, T3 and T4 were 96.4%

(N = 2149; 51% girls, mean age = 13.7, SD = 0.5), 81.4% (N = 1816; 52.3% girls, mean age = 16.3,

SD = 0.7) and 84.3% (N = 1881; 52.3% girls, mean age = 19.1, SD = 0.6) respectively. Detailed in-

formation on the profile, design and procedures of TRAILS have been described elsewhere (de

Winter et al., 2005; Huisman et al., 2008; Nederhof et al., 2012; Ormel et al., 2012).

The MOOVD Sample

The final study of this thesis is based on data from the ongoing ‘Mood and movement in

daily life’ (MOOVD) study, which was set up to investigate the dynamic temporal relationship

between PA and mood in daily life, as well as the role of several biomarkers therein. Data col-

lection of this study started in 2012, and the data from the first 20 individuals to participate

were used (N depressed = 10; 30% males, mean age = 36.4, SD = 10.3 & N non-depressed = 10;

30% males, mean age = 36.7 and SD = 7.9). Participants were intensively monitored in their

natural environment for 30 days, by means of electronic diaries, saliva sampling and contin-

uous actigraphy. Participants were pair-matched on gender, smoking status, age, and Body

Mass Index (BMI). Further details on inclusion and exclusion criteria, procedures, measures,

and ambulatory sampling can be found in Chapter VII.

Notes 1 PF will not be explored further in this thesis. The definition of PF is provided for a better understanding of

the distinct terms related to PA that are sometimes used in the literature. 2 Sometimes the energy expended during sleep, although very low, is still considered part of PA (Caspers-

en, Powell, & Christenson, 1985). In this thesis the energy expended during sleep was not considered as part of PA.

3 RMR is the energy expended during quiet sitting. 4 1MET ≡ 1kcal/kg*h ≡ 4.184kJ/kg*h. 5 See: oxforddictionaries.com/definition/english/sport?q = sport 6 Slightly lower than the prevalence of depression in the USA, which is around 20% (Kessler & Wang, 2009). 7 The Australopithecines are classified as hominins. The terminology has changed recently, but according

to the newest terminology used there are two main distinctions: a) Hominids (hominidae), which are a taxonomic family of all modern and extinct Great Apes including humans; while b) Hominins (Homininae), is a subcategory of the Hominids, including all modern humans, extinct human species and all our im-mediate ancestors (Australopithecines, Homo, Paranthropus and Ardipithecus).

8 Australopithecines walked upright (bipedalism; McΗenry, 1994). The first evidence of bipedalism comes with Australopithecus anamensis, which is assumed to be an ancestor of the better known Australo-pithecus afarensis (Cordain, Gotshall, Eaton, & Eaton, 1998; Leakey, Feibel, McDougall, & Walker, 1995). Although the Australopithecines retained ape like upper limb structural features (McΗenry, 1994; Stern &

CHAPTER I22

Susman, 1983), some evidence from both fossilized footprints and leg bone/pelvic structure suggest that they might have walked with a relatively similar mechanical efficiency to that of contemporary humans (Lovejoy, 1988).

9 It is still disputed whether bipedalism is as efficient, in relation to energy expenditure, as walking in quad-rupeds, but this is not something that will be explored in further detail here (see for example: Halsey & White, 2012).

10 At the same time, the evolution of attenuated body hair size (‘naked’ skin) also improved the evaporative cooling efficiency of the cutaneous sweat glands (Wheeler, 1992).

11 All increases in cranial capacity are discussed in relation to body size. It should be noted, that cranial capacity does not necessarily indicate higher intelligence, since brain structure/volume is more important than actual size.

12 The estimated brain size, relative to body size, of Australopithecines did not exceed 350-600cc, a size much smaller than later Homo species (around 700-1500cc; see figure 4, p. 6784 in McΗenry, 1994) and quite similar in size to modern great apes (around 400cc in chimpanzees and gorillas; Schoenemann, 2006). There is one exception to this, namely Homo floresiensis (Brown et al., 2004) with an estimated brain size equal or smaller than that of chimpanzees. However, it should be noted that the study of brain sizes (paleoneurology) in extinct hominid species is hindered by the incomplete fossil record and as a result these brain sizes are approximations.

13 At the same time of the development of a larger brain, the gut in Homo species started shrinking, simi-larly to what is observed in other carnivores, i.e., a compensation occurring at the expense of the size of the gut to accommodate a more metabolically active brain (Aiello & Wheeler, 1995; Cordain, Gotshall, & Eaton, 1997). Because the gut is not encased by bones, as is the case with the brain, the decrease in size of the gut has been estimated through differences in the anatomy (especially the abdomen and rib cages) of two early species of hominids, namely A. afarensis and Homo ergaster (Aiello & Wheeler, 1995).

14 I.e., developing larger brains.15 The refined ability of humans to throw projectiles in combination with human complex social abilities

(social cooperation) is also another hypothesized factor affecting our current development (for an inter-esting overview of this hypothesis see: Bingham, 1999 and Bingham, 2000).

16 Which was aided by bipedalism and the ability to run large distances (at moderate pace) more efficiently than many quadrupeds (for an interesting discussion see: Bramble & Lieberman, 2004).

17 The estimated energy expenditure of early Homo and modern humans was larger than that of non-human primates (Leonard & Robertson, 1997).

18 The human brain uses around 20% of the body’s energy consumption (Lassen, 1959), a percentage much higher than in other nonhuman primates (Armstrong, 1985; Armstrong, 1990) and other mammals (Arm-strong, 1983).

19 The authors base their hypothesis on the link between proximate mechanisms such as neurotrophins, PA and brain size by: a) reviewing intra and inter-specific and artificial selection studies that suggest that by improving endurance capacity these proximal mechanisms will be altered; and b) studying what is termed ‘athletic species’ (for more information see: Raichlen & Polk, 2013).

20 For example, fewer than 60% of American and Norwegian adolescents adhere to these guidelines (De-sha, Ziviani, Nicholson, Martin, & Darnell, 2007; Sagatun, Sogaard, Bjertness, Selmer, & Heyerdahl, 2007).

21 who.int/dietphysicalactivity/factsheet_inactivity/en/index.html22 For approximately 99% of our existence as a species we have been hunters and food gatherers (Astrand,

1994).23 The works of Hippocrates and especially his three books on regimen, heavily influenced Galen.


24 However these associations might be explained (better) by differences in time spent sitting rather than the activity itself (Hamilton, Hamilton, & Zderic, 2007).

25 However, the relationship between PA and obesity is very complex (Biddle & Mutrie, 2008; p. 22).26 Overtraining, however, may also impair the immune system (Shephard & Shek, 1994).27 Since Hippocrates and Galen, anecdotal evidence or philosophical theories have been proposed on the

benefits of PA on quality of life and well-being. Nonetheless, it should be noted that the terms ‘well-be-ing’ and ‘quality of life’ are rather complex, with various different ways to measure ‘well-being’ and ‘qual-ity of life’, whilst there are different research groups focusing on different aspects of the two (Biddle & Mutrie, 2008).

28 It is important to note here that the distinction between physical and mental health seems to be arbi-trary, since ‘mental health’ problems are multifactorial and probably caused by a dysfunction in both psychological and neurobiological processes (Kendler, 2012). However, this discussion is out of the scope of this thesis and the scientific nomenclature is used for clarity.

29 This study also investigated the cost-effectiveness of a PA intervention and their findings raised doubts on whether PA is a cost-effective treatment for depression.

CHAPTER I24

REFERENCES

Abela, J. R. Z., & Hankin, B. L. (2008). Handbook of depression in children and adolescents. New York: The Guildford Press.

Aiello, L. C., & Wheeler, P. (1995). The expensive-tissue hypothesis - the brain and the digestive-system in human and primate evolution. Current Anthropology, 36(2), 199-221.

Ainsworth, B. E., Haskell, W. L., Leon, A. S., Jacobs, D. R., Montoye, H. J., Sallis, J. F., et al. (1993). Compen-dium of physical activities - classification of energy costs of human physical activities. Medicine and Science in Sports and Exercise, 25(1), 71-80.

American Psychiatric Association (2013). Diagnostic and statistical manual of mental disorders (DSM-5) (5th ed.). Arlington, VA: American Psychiatric Association.

Armstrong, E. (1983). Relative brain size and metabolism in mammals. Science, 220(4603), 1302-1304.

Armstrong, E. (1985). Relative brain size in monkeys and prosimians. American Journal of Physical Anthro-pology, 66(3), 263-273.

Armstrong, E. (1990). Brains, bodies and metabolism. Brain Behavior and Evolution, 36(2-3), 166-176.

Astrand, P. O. (1994). Physical activity and fitness: Evolutionary perspective and trends for the future. In C. Bouchard, R. J. Shephard, & T. Stephens (Eds.), Physical activity, fitness and health (pp. 98-105). Cham-paign, IL: Human Kinetics.

Balady, G. J. (2002). Survival of the fittest - more evidence. New England Journal of Medicine, 346(11), 852-854.

Bassett, D. R., Schneider, P. L., & Huntington, G. E. (2004). Physical activity in an Old Order Amish community. Medicine and Science in Sports and Exercise, 36(1), 79-85.

Behrensmeyer, A. K., & Cooke, H. S. (1985). Paleoenvironments, stratigraphy and taphonomy in the African Pliocene and Early Pleistocene. In E. Delson (Ed.), Ancestors: The hard evidence (pp. 60-62). New York: Wiley-Liss.

Berryman, J. W. (1989). The tradition of the 6 things non-natural - exercise and medicine from Hippocrates through antebellum America. Exercise and Sport Sciences Reviews/series, 17, 515-557.

Biddle, S. J. H., & Mutrie, N. (2008). Psychology of physical activity: Determinants, well-being and interven-tions (2nd ed.). New York, USA: Routledge.

Biddle, S. J. H., & Goudas, M. (1996). Analysis of children’s physical activity and its association with adult encouragement and social cognitive variables. Journal of School Health, 66(2), 75-78.

Biddle, S. J. H., & Asare, M. (2011). Physical activity and mental health in children and adolescents: A review of reviews. British Journal of Sports Medicine, 45(11), 886-895.

Bingham, P. M. (1999). Human uniqueness: A general theory. Quarterly Review of Biology, 74(2), 133-169.

Bingham, P. M. (2000). Human evolution and human history: A complete theory. Evolutionary Anthropology, 9(6), 248-257.

Bramble, D. M., & Lieberman, D. E. (2004). Endurance running and the evolution of Homo. Nature, 432(7015), 345-352.

Brawley, L. R., & Rodgers, W. M. (1993). Social-psychological aspects of fitness promotion. In P. Seraganian (Ed.), Exercise psychology: The influence of physical exercise on psychological processes (pp. 254-298). New York: Wiley.

Brown, F., Harris, J., Leakey, R., & Walker, A. (1985). Early Homo-erectus skeleton from West Lake Turkana, Kenya. Nature, 316(6031), 788-792.

Brown, P., Sutikna, T., Morwood, M. J., Soejono, R. P., Jatmiko, Saptomo, E. W., et al. (2004). A new small-bodied hominin from the Late Pleistocene of Flores, Indonesia. Nature, 431(7012), 1055-1061.


Burr, D. B., Ruff, C. B., & Thompson, D. D. (1990). Patterns of skeletal histologic change through time - com-parison of an archaic Native-American population with modern populations. Anatomical Record, 226(3), 307-313.

Caspersen, C. J., Powell, K. E., & Christenson, G. M. (1985). Physical-activity, exercise, and physical-fitness: Definitions and distinctions for health-related research. Public Health Reports, 100(2), 126-131.

Chaddock, L., Hillman, C. H., Buck, S. M., & Cohen, N. J. (2011). Aerobic fitness and executive control of rela-tional memory in preadolescent children. Medicine and Science in Sports and Exercise, 43(2), 344-349.

Chalder, M., Wiles, N. J., Campbell, J., Hollinghurst, S. P., Haase, A. M., Taylor, A. H., et al. (2012). Facilitated physical activity as a treatment for depressed adults: Randomised controlled trial. British Medical Jour-nal (Clinical Research Ed.), 344, e2758.

Colcombe, S. J., & Kramer, A. F. (2003). Fitness effects on the cognitive function of older adults: A meta-analytic study. Psychological Science, 14(2), 125-130.

Colcombe, S. J., Kramer, A. F., Erickson, K. I., Scalf, P., McAuley, E., Cohen, N. J., et al. (2004). Cardiovascular fitness, cortical plasticity, and aging. Proceedings of the National Academy of Sciences of the United States of America, 101(9), 3316-3321.

Cooney, G. M., Dwan, K., Greig, C. A., Lawlor, D. A., Rimer, J., Waugh, F. R., et al. (2013). Exercise for depres-sion. Cochrane Database of Systematic Reviews, (9), CD004366.

Cordain, L., Gotshall, R. W., Eaton, S. B., & Eaton, S. B. (1998). Physical activity, energy expenditure and fit-ness: An evolutionary perspective. International Journal of Sports Medicine, 19(5), 328-335.

Cordain, L., Gotshall, R. W., & Eaton, S. B. (1997). Evolutionary aspects of exercise. Nutrition and Fitness: Evolutionary Aspects, Children’s Health, Programs and Policies, 81, 49-60.

Cuijpers, P., van Straten, A., Bohlmeijer, E., Hollon, S. D., & Andersson, G. (2010). The effects of psychothera-py for adult depression are overestimated: A meta-analysis of study quality and effect size. Psychologi-cal Medicine, 40(2), 211-223.

Cuijpers, P., Andersson, G., Donker, T., & van Straten, A. (2011). Psychological treatment of depression: Re-sults of a series of meta-analyses. Nordic Journal of Psychiatry, 65(6), 354-364.

de Graaf, R., ten Have, M., van Gool, C., & van Dorsselaer, S. (2012). Prevalence of mental disorders and trends from 1996 to 2009. Results from the Netherlands Mental Health Survey and Incidence Study-2. Social Psychiatry and Psychiatric Epidemiology, 47(2), 203-213.

de Winter, A. F., Oldehinkel, A. J., Veenstra, R., Brunnekreef, J. A., Verhulst, F. C., & Ormel, J. (2005). Evalua-tion of non-response bias in mental health determinants and outcomes in a large sample of pre-adoles-cents. European Journal of Epidemiology, 20(2), 173-181.

deMonocal, P. B., & Bloemendal, J. (1995). Plio-Pleistocene climatic variability in subtropical Africa and the paleoenvironment of hominid evolution: A combined data-model approach. In E. S. Vrba, G. H. Denton, T. C. Patridge, & L. H. Burckle (Eds.), Paleoclimate and evolution with an emphasis on human origins (pp. 262-288). New Haven: Yale University Press.

Desha, L. N., Ziviani, J. M., Nicholson, J. M., Martin, G., & Darnell, R. E. (2007). Physical activity and depres-sive symptoms in American adolescents. Journal of Sport & Exercise Psychology, 29(4), 534-543.

Deslandes, A., Moraes, H., Ferreira, C., Veiga, H., Silveira, H., Mouta, R., et al. (2009). Exercise and mental health: Many reasons to move. Neuropsychobiology, 59(4), 191-198.

Dishman, R. K., Berthoud, H. R., Booth, F. W., Cotman, C. W., Edgerton, V. R., Fleshner, M. R., et al. (2006). Neurobiology of exercise. Obesity, 14(3), 345-356.

Dunbar, R. I. M. (2003). The social brain: Mind, language, and society in evolutionary perspective. Annual Review of Anthropology, 32, 163-181.

Eaton, S. B., & Eaton, S. B. (2003). An evolutionary perspective on human physical activity: Implications for health. Comparative Biochemistry and Physiology A-Molecular & Integrative Physiology, 136(1), 153-159.

CHAPTER I26

Eaton, S. B., & Konner, M. (1985). Paleolithic nutrition - a consideration of its nature and current implications. New England Journal of Medicine, 312(5), 283-289.

Fagard, R. H. (2001). Exercise characteristics and the blood pressure response to dynamic physical training. Medicine and Science in Sports and Exercise, 33(6), S484-S492.

Foley, R. (1987). Another unique species: Patterns in human evolutionary ecology. Harlow: Longman.

Galen, A. (1991). On the therapeutic method (R. J. Hankinson Trans.). Oxford: Clarendon Press.

Gould, S. J. (1980). The panda’s thumb. W.W. Norton & Company.

Halsey, L. G., & White, C. R. (2012). Comparative energetics of mammalian locomotion: Humans are not differ-ent. Journal of Human Evolution, 63(5), 718-22. doi:10.1016/j.Jhevol.2012.07.008.Epub 2012 Sep 7

Hamilton, M. T., Hamilton, D. G., & Zderic, T. W. (2007). Role of low energy expenditure and sitting in obesity, metabolic syndrome, type 2 diabetes, and cardiovascular disease. Diabetes, 56(11), 2655-2667.

Hankin, B. L., Abramson, L. Y., Moffitt, T. E., Silva, P. A., McGee, R., & Angell, K. E. (1998). Development of de-pression from preadolescence to young adulthood: Emerging gender differences in a 10-year longitudinal study. Journal of Abnormal Psychology, 107(1), 128-140.

Hardman, A. E., & Stensel, D. J. (2003). Physical activity and health: The evidence explained. London: Rout-ledge.

Hasler, G., Drevets, W. C., Manji, H. K., & Charney, D. S. (2004). Discovering endophenotypes for major depres-sion. Neuropsychopharmacology, 29(10), 1765-1781.

Hillman, C. H., Erickson, K. I., & Kramer, A. F. (2008). Be smart, exercise your heart: Exercise effects on brain and cognition. Nature Reviews Neuroscience, 9(1), 58-65.

Huisman, M., Oldehinkel, A. J., de Winter, A. F., Minderaa, R. B., de Bildt, A., Huizink, A. C., et al. (2008). Cohort profile: The Dutch ‘TRacking Adolescents’ Individual Lives’ Survey’; TRAILS. International Journal of Epidemiology, 37(6), 1227-1235.

Johnson, K. E., & Taliaferro, L. A. (2011). Relationships between physical activity and depressive symptoms among middle and older adolescents: A review of the research literature. Journal for Specialists in Pedi-atric Nursing: JSPN, 16(4), 235-251.

Joorman, J. (2009). Cognitive aspects of depression. In I. H. Gotlib, & C. L. Hammen (Eds.), Handbook of depression (2nd ed., pp. 298-321). New York: The Guildford Press.

Katzmarzyk, P. T., Church, T. S., Craig, C. L., & Bouchard, C. (2009). Sitting time and mortality from all causes, cardiovascular disease, and cancer. Medicine and Science in Sports and Exercise, 41(5), 998-1005.

Katzmarzyk, P. T. (2010). Physical activity, sedentary behavior, and health: Paradigm paralysis or paradigm shift? Diabetes, 59(11), 2717-2725.

Kendler, K. S. (2012). The dappled nature of causes of psychiatric illness: Replacing the organic-functional/hardware-software dichotomy with empirically based pluralism. Molecular Psychiatry, 17(4), 377-388.

Kessler, R. C., & Wang, P. S. (2009). Epidemiology of depression. In I. H. Gotlib, & C. L. Hammen (Eds.), Hand-book of depression (2nd ed., pp. 5-22). New York: The Guildford Press.

Kessler, R. C., Akiskal, H. S., Ames, M., Birnbaum, H., Greenberg, P., Hirschfeld, R. M. A., et al. (2006). Preva-lence and effects of mood disorders on work performance in a nationally representative sample of U.S. workers. American Journal of Psychiatry, 163(9), 1561-1568.

Khin, N. A., Chen, Y., Yang, Y., Yang, P., & Laughren, T. P. (2011). Exploratory analyses of efficacy data from major depressive disorder trials submitted to the US Food and Drug Administration in support of new drug applications. Journal of Clinical Psychiatry, 72(4), 464-472.

Kirsch, I., Deacon, B. J., Huedo-Medina, T. B., Scoboria, A., Moore, T. J., & Johnson, B. T. (2008). Initial severity and antidepressant benefits: A meta-analysis of data submitted to the Food and Drug Administration. Plos Medicine, 5(2), 260-268.


Krogh, J., Nordentoft, M., Sterne, J. A., & Lawlor, D. A. (2010). The effect of exercise in clinically depressed adults: Systematic review and meta-analysis of randomized controlled trials. The Journal of Clinical Psychiatry, 72(4), 529-538.

Larsen, C. S. (1997). Bioarcheology. Interpreting behavior from the human skeleton. Cambridge: Cambridge University Press.

Lassen, N. A. (1959). Cerebral blood flow and oxygen consumption in man. Physiological Reviews, 39(2), 183-238.

Leakey, M. G., Feibel, C. S., McDougall, I., & Walker, A. (1995). New 4-million-year-old hominid species from Kanapoi and Allia Bay, Kenya. Nature, 376(6541), 565-571.

Leonard, W. R., & Robertson, M. L. (1997). Comparative primate energetics and hominid evolution. American Journal of Physical Anthropology, 102(2), 265-281.

Leonard, W. R. (2010). Size counts: Evolutionary perspectives on physical activity and body size from early hominids to modern humans. Journal of Physical Activity & Health, 7, S284-S298.

Lieberman, D. E. (2011). The evolution of the human head. Cambridge, MA: Harvard University Press.

Lovejoy, C. O. (1988). Evolution of human walking. Scientific American, 259(5), 118-125.

Lux, V., & Kendler, K. S. (2010). Deconstructing major depression: A validation study of the DSM-IV symp-tomatic criteria. Psychological Medicine, 40(10), 1679-1690.

McΗenry, H. M. (1994). Tempo and mode in human-evolution. Proceedings of the National Academy of Sci-ences of the United States of America, 91(15), 6780-6786.

Mead, G. E., Morley, W., Campbell, P., Greig, C. A., McMurdo, M., & Lawlor, D. A. (2009). Exercise for depres-sion. Cochrane Database of Systematic Reviews (Online), (3), CD004366.

Merikangas, K. R., Ames, M., Cui, L., Stang, P. E., Ustun, T. B., Von Korff, M., et al. (2007). The impact of comor-bidity of mental and physical conditions on role disability in the US adult household population. Archives of General Psychiatry, 64(10), 1180-1188.

Morris, Jn., Heady, J. A., Raffle, P. A., Roberts, C. Q., & Parks, J. W. (1953). Coronary heart-disease and physi-cal activity of work. Lancet, 265(6796), 1111-20.

Myers, J., Prakash, M., Froelicher, V., Do, D., Partington, S., & Atwood, J. E. (2002). Exercise capacity and mortality among men referred for exercise testing. New England Journal of Medicine, 346(11), 793-801.

National Institute for Health and Clinical Excellence (2009). Depression: The treatment and management of depression in adults. London: British Psychological Society and Royal College of Psychiatrists.

Nederhof, E., Jorg, F., Raven, D., Veenstra, R., Verhulst, F. C., Ormel, J., et al. (2012). Benefits of extensive recruitment effort persist during follow-ups and are consistent across age group and survey method. The TRAILS study. BMC Medical Research Methodology, 12(1), 93.

Ormel, J., Oldehinkel, A. J., Sijtsema, J., van Oort, F., Raven, D., Veenstra, R., et al. (2012). The TRacking Adolescents’ Individual Lives Survey (TRAILS): Design, current status, and selected findings. Journal of the American Academy of Child and Adolescent Psychiatry, 51(10), 1020-1036.

Paffenbarger, R. S., Hyde, R. T., Wing, A. L., & Hsieh, C. C. (1986). Physical-activity, all-cause mortality, and longevity of college alumni. New England Journal of Medicine, 314(10), 605-613.

Paluska, S. A., & Schwenk, T. L. (2000). Physical activity and mental health: Current concepts. Sports Medi-cine, 29(3), 167-180.

Pedersen, B. K., & Hoffman-Goetz, L. (2000). Exercise and the immune system: Regulation, integration, and adaptation. Physiological Reviews, 80(3), 1055-1081.

Powell, K. E., & Blair, S. N. (1994). The public-health burdens of sedentary living habits - theoretical but realistic estimates. Medicine and Science in Sports and Exercise, 26(7), 851-856.

CHAPTER I28

Prentice, A. M., & Jebb, S. A. (2000). Physical activity and weight control in adults. In C. Bouchard (Ed.), Physical activity and obesity (pp. 247-261). Champaign, IL: Human Kinetics.

Raichlen, D. A., & Polk, J. D. (2013). Linking brains and brawn: Exercise and the evolution of human neurobiol-ogy. Proceedings of the Royal Society B-Biological Sciences, 280(1750), 20122250.

Rimer, J., Dwan, K., Lawlor, D. A., Greig, C. A., McMurdo, M., Morley, W., et al. (2012). Exercise for depression. Cochrane Database of Systematic Reviews, (7), CD004366.

Rode, A., & Shephard, R. J. (1984). Growth, development and acculturation - a 10 year comparison of Cana-dian Inuit children. Human Biology, 56(2), 217-230.

Rode, A., & Shephard, R. J. (1994). Growth and fitness of Canadian Inuit - secular trends, 1970-1990. Ameri-can Journal of Human Biology, 6(4), 525-541.

Ruff, C. B. (1991). Climate and body shape in hominid evolution. Journal of Human Evolution, 21(2), 81-105.

Sagatun, A., Sogaard, A. J., Bjertness, E., Selmer, R., & Heyerdahl, S. (2007). The association between week-ly hours of physical activity and mental health: A three-year follow-up study of 15-16-year-old students in the city of Oslo, Norway. BMC Public Health, 7, 155.

Salonen, J. T., Puska, P., Kottke, T. E., Tuomilehto, J., & Nissinen, A. (1983). Decline in mortality from coronary heart-disease in Finland from 1969 to 1979. British Medical Journal, 286(6381), 1857-1860.

Schmid, P. (1991). The trunk of the Australopithecines. In Y. Coppens, & B. Senut (Eds.), Origine(s) de la bipi-die chez les hominides. Paris: Edition du CNRS.

Schoenemann, P. T. (2006). Evolution of the size and functional areas of the human brain. Annual Review of Anthropology, 35, 379-406.

Shephard, R. J., & Shek, P. N. (1994). Potential impact of physical-activity and sport on the immune-system - a brief review. British Journal of Sports Medicine, 28(4), 247-255.

Shipman, P. (1986). Scavenging or hunting in early hominids - theoretical framework and tests. American Anthropologist, 88(1), 27-43.

Stephens, T. (1988). Physical activity and mental health in the United States and Canada: Evidence from four population surveys. Preventive Medicine, 17(1), 35-47.

Stern, J. T., & Susman, R. L. (1983). The locomotor anatomy of Australopithecus afarensis. American Jour-nal of Physical Anthropology, 60(3), 279-317.

Turner, E. H., Matthews, A. M., Linardatos, E., Tell, R. A., & Rosenthal, R. (2008). Selective publication of antidepressant trials and its influence on apparent efficacy. New England Journal of Medicine, 358(3), 252-260.

van Praag, H., Christie, B. R., Sejnowski, T. J., & Gage, F. H. (1999). Running enhances neurogenesis, learn-ing, and long-term potentiation in mice. Proceedings of the National Academy of Sciences of the United States of America, 96(23), 13427-13431.

van Praag, H., Kempermann, G., & Gage, F. H. (1999). Running increases cell proliferation and neurogenesis in the adult mouse dentate gyrus. Nature Neuroscience, 2(3), 266-270.

van Praag, H. (2008). Neurogenesis and exercise: Past and future directions. Neuromolecular Medicine, 10(2), 128-140.

Wells, K. B., Stewart, A., Hays, R. D., Burnam, M. A., Rogers, W., Daniels, M., et al. (1989). The functioning and well-being of depressed-patients - results from the Medical Outcomes Study. Jama-Journal of the American Medical Association, 262(7), 914-919.

Wheeler, P. E. (1991). The influence of bipedalism on the energy and water budgets of early hominids. Jour-nal of Human Evolution, 21(2), 117-136.

Wheeler, P. E. (1992). The influence of the loss of functional body hair on the water budgets of early homi-nids. Journal of Human Evolution, 23(5), 379-388.

Bidirectional Prospective Associations between Physical Activity and Depressive Symptoms. The TRAILS Study.

“All truly great thoughts are conceived by walking.”

Friedrich Nietzsche, Twilight of the Idols, Or, How to Philosophize with the Hammer, 1889

“If you would get exercise, go in search of the springs of life.”Henry David Thoreau, Walking, 1862

Stavrakakis N, de Jonge P, Ormel J & Oldehinkel AJ

J Adolesc Health 2012; 50(5):503-8

CHAPTER

CHAPTER II30

ABSTRACT

Purpose: Low levels of physical activity have been shown to be associated with depression

in adults. The few studies that focused on adolescents yielded mixed and inconsistent re-

sults. Efforts to examine the direction of this relationship have been inconclusive up to now.

The aims of this study were therefore to investigate (1) the direction of the inverse associa-

tion between physical activity and depressive symptoms over time and (2) whether these

associations are specific to particular clusters of depressive symptoms in adolescents.

Methods: Depressive symptoms and physical activity were assessed in a population sample

of adolescents (N = 2230), who were measured at three waves between age 11 and age 16.

Depressive symptoms were measured by the Affective Problems scale of the Youth Self-Re-

port (YSR) and Child Behavior Checklist (CBCL), while physical activity was operationalized

as the amount of time spent on physical exercise. Structural Equation Modeling was used to

examine bidirectional effects of physical activity and depressive symptoms over time.

Results: We found significant cross-lagged paths from prior physical activity to later de-

pression as well as from prior depression to later physical activity (beta values = -0.033 to

-0.047). After subdividing depression into affective and somatic symptoms, the affective

symptoms were reciprocally related to physical activity, while the paths between somatic

symptoms and physical activity did not reach statistical significance.

Conclusion: An inverse bidirectional association between physical activity and general de-

pressive symptoms was observed. This association was restricted to affective symptoms.

31PHYSICAL ACTIVITY AND DEPRESSIVE SYMPTOMS

INTRODUCTION

Associations between physical activity (PA) and health have been well documented. Stud-

ies have shown that exercise has beneficial effects on medical conditions such as diabetes

and cardiovascular diseases (Hu et al., 2004; Wessel et al., 2004), and also lowers the risk

of developing a variety of cancers (Howard et al., 2008; Bardia et al., 2006). Associations

of PA with mental health, however, have been harder to elucidate. There is some evidence

suggesting that PA is inversely related to depression in adults, but it has been difficult to

establish causality and the strength of the associations varied from study to study (Harris,

Cronkite, & Moos, 2006; Cooper-Patrick, Ford, Mead, Chang, & Klag, 1997).

In observational studies, causality cannot be established, but the temporal order of changes

in PA and depression may provide clues about the direction of the effects. It has been very

difficult to determine temporality so far, however, due to the cross-sectional nature of most

studies. Weyerer (1992) showed a cross-sectional association between low levels of PA and

depression, but could not find a prospective association at 5-year follow-up. Since then, only

a few studies have assessed the direction of the association between depression and PA,

and those were relatively unsuccessful in giving a clear answer (Birkeland, Torsheim, & Wold,

2009; Jerstad, Boutelle, Ness, & Stice, 2010). Intervention trials have yielded ambiguous re-

sults as well. Whereas a number of studies indicated that PA reduces feelings of tension,

anxiety and anger (Babyak et al., 2000), others could not find such clear effects (Salmon, 2001;

Haboush, Floyd, Caron, LaSota, & Alvarez, 2006). In a recent review, Mead and colleagues

(2009) showed that, when only trials with blinded outcome assessments, intention to treat

and allocation concealment were included in the analysis, effect sizes were weaker and not

statistically significant. In adolescents, although cross-sectional and intervention studies

have shown a consistent inverse association between PA and depressive symptoms, results

from prospective studies have been inconsistent and will be briefly described later.

Why are there so many mixed results regarding associations between PA and depression?

Part of the discrepancies can be attributed to methodological limitations such as small

sample sizes and, for prospective studies, short follow-up periods. Another reason might be

that depression is a heterogeneous disorder, consisting of both affective and somatic symp-

toms. Instruments that are used to assess depressive symptoms vary widely with respect

to the symptom profiles they represent, where some instruments do not include somatic

symptoms, while other instruments contain multiple questions on somatic symptoms. It has

been shown that somatic and affective symptoms are differentially associated with cardiac

autonomic (de Jonge et al., 2006) and HPA axis function (Bosch et al., 2009). Since these two

subgroups of symptoms show distinct patterns of association with various physiological

measures, they may be differentially related to PA as well.

CHAPTER II32

Despite these discrepancies, several, not necessarily mutually exclusive, hypotheses have

been proposed to explain the inverse association between PA and depression:

1) Protection hypothesis (direct causal effect): PA decreases depressive symptoms through

biological (elevation of endorphins, serotonin or endocannabinoids) or social (social con-

tact and self-esteem) mechanisms (Allgower, Wardle, & Steptoe, 2001).

2) Inhibition hypothesis (reverse causality): depression negatively influences patterns of PA

(Patten, Williams, Lavorato, & Eliasziw, 2009) through symptoms such as lack of energy,

anhedonia, low mood and social withdrawal (Struder & Weicker, 2001).

3) Common Cause hypothesis: PA and depression share risk factors, for instance genetic or

familial factors such as socioeconomic status (SES), neighborhood and parental rearing

style (Stubbe, de Moor, Boomsma, & de Geus, 2007).

In the present study, associations between PA and depressive symptoms were studied in

a large prospective population cohort of Dutch adolescents. Studying this association in

adolescents is important for at least two reasons. First, evidence regarding the association

between PA and depression is even more ambiguous in adolescents than in adults. While

cross-sectional and intervention studies in adolescents have shown moderate to strong as-

sociations between PA and depression respectively (Monshouwer, ten Have, van Poppel,

Kemper, & Vollebergh, 2009; Kirkcaldy, Shephard, & Siefen, 2002), longitudinal studies have

yielded mixed results varying from strong effects (Patten et al., 2009; Motl, Birnbaum, Ku-

bik, & Dishman, 2004) to weak (Jerstad et al., 2010; Sagatun, Sogaard, Bjertness, Selmer,

& Heyerdahl, 2007) or even no effects at all (Birkeland et al., 2009; Ströhle et al., 2007).

Second, a substantial number of adolescents evince sub-threshold depressive symptoms

(Pine, Cohen, Cohen, & Brook, 1999). Since adolescent depressive symptoms are thought to

predict full-blown depressive disorders later in life (Pine et al., 1999), researching depressive

symptoms in adolescence may help to improve the understanding of the etiology of depres-

sive disorders and help design effective interventions for treating affective disorders.

The first aim of this study was to investigate patterns in the association between PA and

depressive symptoms, in order to determine whether changes in PA precede, follow or co-

occur with changes in depressive symptoms in healthy adolescents. A longitudinal design

is the only way to illuminate such patterns. The second aim was to examine the association

between PA and two subgroups of depressive symptoms, that is, affective symptoms such

as depressed mood and loss of interest, and somatic symptoms such as sleep disturbances

and lack of energy. To the best of our knowledge, PA has not been investigated in relation to

these symptom groups before.


METHODS

Design

Data for this study were collected as part of the Tracking Adolescents' Individual Lives Sur-

vey (TRAILS), a prospective cohort of Northern Dutch adolescents. The study was approved

by the Dutch Central Committee on Research Involving Human Subjects. A detailed descrip-

tion of its objectives and main design, as well as of the sample selection procedure and non-

response, can be found elsewhere (de Winter et al., 2005).

Data collection took part in three assessment waves (T1, T2, T3), which ran from March

2001 to July 2002; September 2003 to December 2004; and September 2005 to August 2007

respectively. Parental written informed consent was obtained after the procedures had been

fully explained. Adolescents gave written informed assent at the second and third assess-

ment wave.

Participants

Of 2935 adolescents initially approached, 2230 (76%; girls = 51%, mean age = 11.11, SD = 0.55)

took part in the first assessment wave (T1). The response rate at the second wave (T2) was

96.4% (N = 2149; 51% girls, mean age = 13.65, SD = 0.53), while the response rate was 81.4%

(N = 1816; 52.3% girls, mean age = 16.27, SD = 0.73) at the third assessment wave (T3).

Measures

Depressive Symptoms

Depressive symptoms were assessed by the Affective Problems scale of the Youth Self Re-

port (YSR; Achenbach, 1991b), which was completed by the adolescents at school. Parents

filled out the parent version of the YSR, the Child Behavior Checklist (CBCL; Achenbach,

1991a) at home. The mean scores of the YSR and CBCL scales were used in the analyses.

The YSR and CBCL are composed of a list of problems that are scored on a three-point

scale (0 = never or not at all true, 1 = sometimes true and 2 = very often or very true). The

Affective Problems scale contains 13 items covering depressive symptoms according to the

DSM-IV (Achenbach, Dumenci, & Rescorla, 2003), that is, sadness, loss of pleasure, crying,

self-harm, suicidal ideation, feelings of worthlessness, guilt, loss of energy, overtiredness,

eating problems and sleeping problems. Because a previous study in the same sample had

shown that omission of one sleep item (‘I sleep more than most kids’) increased the internal

consistency of the scale (Bosch et al., 2009), this item was excluded. Scores on the remain-

ing twelve items were averaged to construct a total depressive symptoms scale with an

internal consistency (Cronbach’s alpha) of 0.75 at T1, of 0.84 at T2 and 0.89 at T3. In addition,

CHAPTER II34

we constructed a somatic symptoms subscale and an affective symptoms subscale, as de-

scribed by Bosch et al. (2009). The affective symptom subscale included loss of pleasure,

crying, self-harm, suicidal ideation, feelings of worthlessness, feelings of guilt and sadness,

while the somatic symptoms subscale included lack of appetite, overtiredness, reduced

sleep, trouble sleeping and lack of energy. The internal consistency for the affective symp-

toms scale was 0.72 at T1, 0.82 at T2 and 0.85 at T3, while the consistencies of the somatic

symptoms scale were 0.57 at T1, 0.69 at T2 and 0.77 at T3.

Physical Activity

At T1, PA was assessed by the question: “How often do you perform physical exercise (for

example, swimming, playing football, horse-riding)?”. The question could be answered on

a five-point scale (0 = never, 1 = once per week, 2 = two to three times per week, 3 = four to

five times per week, and 4 = five to six times per week). At T2 and T3, PA was measured by

two questions assessing the time per week the individual engaged in PA during the summer

(question 1) and the winter (question 2): “How many days in an average week in the summer/

winter do you take part in physical activities?”. These questions were rated on an eight-

point scale (0 = never, to 7 = seven days per week). The questions on summer and winter were

averaged and mean scores were used in the analysis. To achieve a similar metric for the PA

measures at all measurement waves, the T1 data (ranging from 0-4) were recoded into an

eight-point scale using the monotonic R7 transform recommended by Little (in press).

Covariates

Gender and SES were included as covariates. SES was calculated by averaging five stan-

dardized variables (professional occupation and educational attainment for both father and

mother, and household income). Three SES groups were created, where the lowest 25%

of scores were categorized as ‘low SES’, the highest 25% as ‘high SES’ and the remaining

scores were grouped as ‘intermediate SES’.

Analysis

Imputation of Missing Data

A detailed account of attrition rates within the TRAILS study can be found elsewhere (Huis-

man et al., 2008). We performed multiple imputations to complete the dataset (i.e., N = 2230

at all three assessment waves) using STATA version 10 (STATA corp., College station, Texas)

and created five different datasets (Rubin, 1996), which were subsequently imported into

Mplus 5 (Muthén & Muthén, 2007) and used in the analyses.


Cross-Lagged Path Model

Cross-sectional and prospective associations between PA and depressive symptoms were in-

vestigated by means of Structural Equation Modeling (SEM), using Mplus 5 (Muthén & Muth-

én, 2007). Since the depression variables (both total depressive symptoms score and scores

on the two subscales) were positively skewed and slightly kurtotic, they were log trans-

formed prior to analysis. A cross-lagged path model was used to investigate cross-sectional

and prospective associations between PA and depressive symptoms. Cross-lagged panel

designs take into account the time precedence and control for multivariate dependencies of

the antecedent predictor variables. Therefore, besides the cross-lagged paths (interpreted as

a linkage of the level of one variable at the first wave with a relative change in another vari-

able in the subsequent assessment wave) between depression and PA and vice versa (Figure

1, path c), autoregressive paths (interpreted as relative stability over time) within depression

and PA (Figure 1, path b), and cross-sectional covariances (interpreted as correlations at T1

and as correlated change at T2 and T3) between PA and depression (Figure 1, path a) were

also estimated. Correlated change and cross-lagged paths reflect longitudinal relationships.

In order to test the predictive relationships between PA and depressive symptoms we devel-

oped two cross-lagged models. The first model concerned associations between PA and the

total depressive symptoms score, while the second model concerned associations between

b

b

b

b

c c

c c

a a

DepressionT1

PA T1

DepressionT2

PA T2

DepressionT3

PA T3

a

b

b

Figure 1. Model showing Cross-Sectional Paths (a), Autoregressive (b), and Cross-Lagged Longitudinal Paths (c).

PA = physical activity, T1 = first measurement wave, T2 = second measurement wave and T3 = third measurement wave.

CHAPTER II36

PA and the subscales of somatic and affective symptoms (adjusted for each other). Both mod-

els were corrected for gender and SES. As a first step, all cross-lagged and cross-sectional

paths were constrained to be equal over time, and the fit indices of the constrained model

were compared to those of the unconstrained model. If the fit indices were adequate and

not significantly worse than those of the unconstrained model, the more parsimonious, i.e.,

constrained, model was used. The significance of the cross-lagged paths was established by

testing if removal of the paths led to a deterioration of the model fit. To test whether the two

subgroups of depressive symptoms (somatic and affective symptoms) differed with regard to

their cross-sectional and cross-lagged association with PA, we compared a model where the

paths between somatic symptoms and PA and affective symptoms and PA respectively, were

constrained to be equal with a model where these paths were allowed to differ.

A good model fit was defined when the Comparative Fit Index (CFI) and the Tucker-Lew-

is Index (TLI) were greater than 0.95 while the Root Mean Square Error of Approximation

(RMSEA) was lower than 0.05. Ideally, the χ2 should be non-significant (p>0.05) as well, but

larger samples increase the likelihood of obtaining significant p-values (Bentler, 1990).

RESULTS

Descriptive Statistics

Descriptive statistics of the main variables used in this study are shown in Table 1. Depres-

sive symptoms increased slightly over time, mainly due to a rise in somatic symptoms. PA

remained stable over time.

Table 1. Means (Standard Deviations) of the Main Variables.

T1 (mean age 11.1)

T2 (mean age 13.7)

T3 (mean age 16.3)

Total Depressive Symptoms

0.25 (0.20) 0.25 (0.23) 0.31 (0.29)

Affective Symptoms 0.21 (0.20) 0.19 (0.23) 0.23 (0.28)

Somatic Symptoms 0.32 (0.27) 0.33 (0.30) 0.42 (0.36)

Physical Activity 3.38 (2.09) 3.35 (1.73) 3.45 (1.88)

N = 2230 for all measurements since missing data were imputed. T1 = first measurement wave, T2 = second measurement wave and T3 = third measurement wave.


Relationships between Depression and PA

Preliminary analyses revealed that cross-lagged paths and correlated changes were con-

stant over time: the model fit parameters of the models where these paths were constrained

to be equal across time were adequate and the χ2-values of the constrained models were not

significantly worse than the χ2-values of the unconstrained model (χ2difference = 5.555, df = 3,

p>0.05 for the model involving total depressive symptoms; χ2difference = 6.365, df = 6, p>0.05

for the model involving depression subgroups). Because the constrained models were more

parsimonious we used these models in all subsequent analyses.

Relationships between the total depressive symptoms score and PA are shown in Figure 2 (over-

all model fit, χ2 = 13.254, df = 6, p<0.05). Fit indices showed an excellent fit of model (CFI = 0.997,

TLI = 0.988, RMSEA = 0.022), which was significantly better than the fit of the model without

cross-lagged paths from PA to depression (χ2difference = 8.103, df = 1, p<0.05) and the model with-

out cross-lagged paths from depression to PA (χ2difference = 11.556, df = 1, p<0.05). As expected

from these results, the cross-lagged paths were all significant. The autoregressive estimate of

depression was higher than the autoregressive estimate of PA. The cross-sectional association

between PA and depression was significant at T1 but not at later assessment waves.

Figure 2. Associations between Depression (Total Depressive Symptoms Score) and PA over time.

.586* (.015)

.229* (.023)

.556* (.020)

.272* (.023)

-.039* (.016)

-.033* (.010)

-.047* (.015)

-.042* (.017)

-.046 (.026)

-.039 (.022)

DepressionT1

PA T1

DepressionT2

PA T2

DepressionT3

PA T3

-.078* (.022)

.120* (.024)

.135* (.025)

PA = physical activity, T1 = first measurement wave, T2 = second measurement wave and T3 = third measurement wave.The path coefficients reflect model estimated standardized beta values. Statistical insignificant paths shown are indicated by broken lines. The effects are adjusted for gender and socioeconomic status. * p<0.05.

CHAPTER II38

The second model, investigating the relationship between somatic symptoms, affective

symptoms and PA over time, is shown in Figure 3. Indices of approximate fit showed an

excellent fit of model variables (χ2 = 24.170, df = 13, p<0.05, CFI = 0.998, TLI = 0.992, RM-

SEA = 0.020). While cross-lagged relationships between PA and somatic symptoms did not

reach statistical significance, affective symptoms both predicted and were predicted by PA.

This is reflected in the tests comparing the models with and without these cross-lagged

paths: whereas removal of the cross-lagged to and from somatic symptoms did not worsen

the model fit significantly (χ2difference = 2.404, df = 2, p>0.05), removal of the paths to and from

affective symptoms led to a significant decrease of the model fit (χ2difference = 26.755, df = 2,

p<0.05).

.479* (.019)

.228* (.024)

.448* (.023)

.406* (.024)

.272* (.024)

.457* (.035)

.002 (.018)

-.062* (.016)

-.015 (.010)

-.049* (.018)

-.021 (.014)

-.048* (.017)

-.002 (.018)

-.044* (.012)

-.035 (.019)

-.038 (.028)

-.032 (.017)

-.034 (.025)

Somatic T1

PA T1

Affective T1

Somatic T2

PA T2

Affective T2

Somatic T3

PA T3

Affective T3

-.046* (.022)

-.088* (.022)

Figure 3. Association between Somatic Symptoms, Affective Symptoms and PA over time.

PA = physical activity, T1 = first measurement wave, T2 = second measurement wave and T3 = third measurement wave.The path coefficients reflect model estimated standardized beta values. Statistically insignificant paths are indicated by broken lines. The effects are adjusted for gender and socioeconomic status. * p<0.05. Paths between somatic and affective symptoms although defined in the model (cross-sectional, autoregressive between T1 and T3, and cross-lagged) were omitted from this figure for clarity purposes.


DISCUSSION

The results of this study indicate a bidirectional cross-lagged association between depres-

sive symptoms and PA, in which PA precedes a decrease in depressive symptoms and vice

versa. Quite surprisingly, the inverse associations between PA and depressive symptoms

concerned affective symptoms such as depressed mood, loss of pleasure and low self-worth

in particular. Somatic symptoms such as sleep disturbances, eating problems and lack of en-

ergy were not associated with PA. Interestingly, we found that PA remained stable over time,

a finding which is inconsistent with some previous studies, which reported a steady decrease

of PA over time (Riddoch et al., 2004).

Limitations and Strengths of this Study

This study has a number of notable strengths. First, we used a large, population-based

sample of adolescents. Studying adolescents has two main advantages: a) the probability

of confounding by prior depressive episodes is low at this age, and b) the prevalence of so-

matic conditions that may hamper engagement in PA is still relatively low in adolescents

compared to older people. Second, the longitudinal design with three measurements across

6 years made it possible to study changes in PA and depressed mood over a long period of

time. Third, the model fit was excellent, modeling the relationship between PA and depres-

sive symptoms in our data with great accuracy. A final strength is the fact that we subdi-

vided depressive symptoms into somatic and affective symptoms, which, to the best of our

knowledge has not been done before in this context. Depression is a heterogeneous disorder,

with different symptoms observed in different individuals. The use of subgroups does justice

to this heterogeneity and may help to understand individual differences in the association

between PA and depression.

There are also limitations. First, all of the measures used were self-reports and the mea-

surement of PA relied on a single question, unlike validated PA questionnaires (e.g., IPAQ),

which include a series of detailed questions of PA involvement. Self-reports tend to be less

reliable than more objective measures of PA, such as VO2 max, heart rate variability and

calculation of Metabolic Equivalent of Tasks by, for instance, accelerometers. In addition,

at the first measurement wave of this study, PA was assessed slightly different than at the

last two waves, in which separate questions for winter and summer, and more response cat-

egories were used. Furthermore, we did not measure the nature (aerobic-anaerobic, social-

isolated etc.), intensity, or duration of PA, information which may prove useful for a better

understanding of the relationship between PA and depressive symptoms. Concerning the

nature of PA, for instance, aerobic exercises have been reported to have an effect on depres-

sive symptoms while anaerobic exercises do not (Goodwin, 2003). Similarly, differences in the

CHAPTER II40

intensity and the duration of the activity may be related to depressive symptoms (Galper,

Trivedi, Barlow, Dunn, & Kampert, 2006). A final limitation is that the reliability of the so-

matic symptoms subscale was relatively low, which prevents strong associations with other

variables.

The associations observed were bidirectional. The more times per week adolescents en-

gaged in PA, the more likely they were to report a reduction in depressive symptoms. The

opposite also held true: the more depressive symptoms adolescents exhibited at some point

in time, the more likely they were to report a reduction in PA later on. These findings are in

accordance with recent reviews in adults which showed that regular PA decreases the risk

for developing depression (Teychenne, Ball, & Salmon, 2008) and that baseline depression

might play an important role in the development of an inactive lifestyle or decreased level

of PA (Roshanaei-Moghaddam, Katon, & Russo, 2009). In adolescents, longitudinal stud-

ies have yielded mixed results. Whereas Jerstad et al. (2010) demonstrated a bidirectional

relationship between exercise and depression, Birkerland and colleagues (2009) could not.

The bidirectional association between PA and depression found in our study provide support

for both the Protection hypothesis and Inhibition hypothesis, which were mentioned earlier

in the Introduction.

We also showed that PA is differentially associated with subgroups of depressive symp-

toms; the effects are stronger for affective than for somatic depressive symptoms. This lack

of association between PA and somatic symptoms is unexpected. Although it seems plau-

sible that adolescents who exercise more sleep better, eat better and have more energy,

this was not supported by our data. A reason for this lack of association (between PA and

somatic depressive symptoms) may be the before-mentioned low reliability of the somatic

symptoms scale, which prevents strong relationships. The low reliability implies that there

is heterogeneity within the somatic symptom domain. Whether specific somatic symptoms

are differentially related to PA was beyond the scope of this paper and remains to be inves-

tigated in the future.

The inverse associations found between PA and depression were statistically significant

but were relatively weak. This is generally in accordance with previous studies, which rarely

yielded strong effects. This could be due to the measures of PA used, which usually reflect

only a small portion of activities and do not take into account the nature of the activities

(e.g., voluntary, social, competitive or isolated), their duration and their intensity. PA may be

beneficial to mental health only after an intensity/duration threshold is reached, which could

have weakened the associations. Another likely reason for the weak associations is the fact

that both depressed mood and PA are affected by multiple non-shared variables and there-

fore cannot be expected to explain much variance in each other. Finally, it might be possible

that PA and depressive symptoms are not associated in all individuals in the same way; due


to genetic or personality differences, some adolescents may benefit from PA while others do

not. An elucidation of the underlying mechanisms might provide clues about which individu-

als may benefit most from PA.

CONCLUSION

The current prospective study investigated the inverse bidirectional associations between

physical activity and depressive symptoms, in particular the two sub-clusters of depres-

sive symptoms (somatic vs. affective), in a population cohort of Dutch adolescents. Physical

activity has been shown to be related to depressive symptoms and vice versa but only in

relation to affective symptoms. Further research is needed to investigate whether individual

symptoms are differentially associated with physical activity, in order to improve the under-

standing of this complex relation.

CHAPTER II42

REFERENCES

Achenbach, T. M. (1991a). Manual for the child behavior checklist/ 4-18 and 1991 profile. University of Ver-mont, Burlington.

Achenbach, T. M. (1991b). Manual for the youth self report and 1991 profile. University of Vermont, Burlington.

Achenbach, T. M., Dumenci, L., & Rescorla, L. A. (2003). DSM-oriented and empirically based approaches to constructing scales from the same item pools. Journal of Clinical Child and Adolescent Psychology, 32(3), 328-340.

Allgower, A., Wardle, J., & Steptoe, A. (2001). Depressive symptoms, social support, and personal health behaviors in young men and women. Health Psychology, 20(3), 223-227.

Babyak, M., Blumenthal, J. A., Herman, S., Khatri, P., Doraiswamy, M., Moore, K., et al. (2000). Exercise treat-ment for major depression: Maintenance of therapeutic benefit at 10 months. Psychosomatic Medicine, 62(5), 633-638.

Bardia, A., Hartmann, L. C., Vachon, C. M., Vierkant, R. A., Wang, A. H., Olson, J. E., et al. (2006). Recreational physical activity and risk of postmenopausal breast cancer based on hormone receptor status. Archives of Internal Medicine, 166(22), 2478-2483.

Bentler, P. M. (1990). Comparative fit indexes in structural models. Psychological Bulletin, 107(2), 238-246.

Birkeland, M. S., Torsheim, T., & Wold, B. (2009). A longitudinal study of the relationship between leisure-time physical activity and depressed mood among adolescents. Psychology of Sport and Exercise, 10(1), 25-34.

Bosch, N. M., Riese, H., Dietrich, A., Ormel, J., Verhulst, F. C., & Oldehinkel, A. J. (2009). Preadolescents’ so-matic and cognitive-affective depressive symptoms are differentially related to cardiac autonomic func-tion and cortisol: The TRAILS study. Psychosomatic Medicine, 71(9), 944-950.

Cooper-Patrick, L., Ford, D. E., Mead, L. A., Chang, P. P., & Klag, M. J. (1997). Exercise and depression in midlife: A prospective study. American Journal of Public Health, 87(4), 670-673.

de Jonge, P., Ormel, J., van den Brink, R. H., van Melle, J. P., Spijkerman, T. A., Kuijper, A., et al. (2006). Symp-tom dimensions of depression following myocardial infarction and their relationship with somatic health status and cardiovascular prognosis. The American Journal of Psychiatry, 163(1), 138-144.


Galper, D. I., Trivedi, M. H., Barlow, C. E., Dunn, A. L., & Kampert, J. B. (2006). Inverse association between physical inactivity and mental health in men and women. Medicine and Science in Sports and Exercise, 38(1), 173-178.

Goodwin, R. D. (2003). Association between physical activity and mental disorders among adults in the United States. Preventive Medicine, 36(6), 698-703.

Haboush, A., Floyd, M., Caron, J., LaSota, M., & Alvarez, K. (2006). Ballroom dance lessons for geriatric depression: An exploratory study. The Arts in Psychotherapy, 33(2), 89-97.

Harris, A. H., Cronkite, R., & Moos, R. (2006). Physical activity, exercise coping, and depression in a 10-year cohort study of depressed patients. Journal of Affective Disorders, 93(1-3), 79-85.

Howard, R. A., Freedman, D. M., Park, Y., Hollenbeck, A., Schatzkin, A., & Leitzmann, M. F. (2008). Physical activity, sedentary behavior, and the risk of colon and rectal cancer in the NIH-AARP Diet and Health Study. Cancer Causes & Control, 19(9), 939-953.

Hu, G., Lindstrom, J., Valle, T. T., Eriksson, J. G., Jousilahti, P., Silventoinen, K., et al. (2004). Physical activ-ity, body mass index, and risk of type 2 diabetes in patients with normal or impaired glucose regulation. Archives of Internal Medicine, 164(8), 892-896.



Jerstad, S. J., Boutelle, K. N., Ness, K. K., & Stice, E. (2010). Prospective reciprocal relations between physi-cal activity and depression in female adolescents. Journal of Consulting and Clinical Psychology, 78(2), 268-272.

Kirkcaldy, B. D., Shephard, R. J., & Siefen, R. G. (2002). The relationship between physical activity and self-image and problem behaviour among adolescents. Social Psychiatry and Psychiatric Epidemiology, 37(11), 544-550.

Little, T. D. (in press). Longitudinal structural equation modeling: Individual difference panel models. New York: Guildford press.


Monshouwer, K., ten Have, M., van Poppel, M., Kemper, H., & Vollebergh, W. (2009). Low physical activity in adolescence is associated with increased risk for mental health problems. Medicina Sportiva, 13(2), 74-81.

Motl, R. W., Birnbaum, A. S., Kubik, M. Y., & Dishman, R. K. (2004). Naturally occurring changes in physical activity are inversely related to depressive symptoms during early adolescence. Psychosomatic Medi-cine, 66(3), 336-342.

Muthén, L. K., & Muthén, B. O. (2007). Mplus user’s guide: Statistical analysis with latent variables (5th ed.). Los Angeles, CA: Muthén & Muthén.

Patten, S. B., Williams, J. V., Lavorato, D. H., & Eliasziw, M. (2009). A longitudinal community study of major depression and physical activity. General Hospital Psychiatry, 31(6), 571-575.

Pine, D. S., Cohen, E., Cohen, P., & Brook, J. (1999). Adolescent depressive symptoms as predictors of adult depression: Moodiness or mood disorder? The American Journal of Psychiatry, 156(1), 133-135.

Riddoch, C. J., Bo Andersen, L., Wedderkopp, N., Harro, M., Klasson-Heggebo, L., Sardinha, L. B., et al. (2004). Physical activity levels and patterns of 9- and 15-yr-old European children. Medicine and Science in Sports and Exercise, 36(1), 86-92.

Roshanaei-Moghaddam, B., Katon, W. J., & Russo, J. (2009). The longitudinal effects of depression on physi-cal activity. General Hospital Psychiatry, 31(4), 306-315.

Rubin, B. D. (1996). Multiple imputation after 18+ years. Journal of the American Statistical Association, 91(434), 473.

Sagatun, A., Sogaard, A. J., Bjertness, E., Selmer, R., & Heyerdahl, S. (2007). The association between week-ly hours of physical activity and mental health: A three-year follow-up study of 15-16-year-old students in the city of Oslo, Norway. BMC Public Health, 7, 155.

Salmon, P. (2001). Effects of physical exercise on anxiety, depression, and sensitivity to stress: A unifying theory. Clinical Psychology Review, 21(1), 33-61.

Ströhle, A., Hofler, M., Pfister, H., Muller, A. G., Hoyer, J., Wittchen, H. U., et al. (2007). Physical activity and prevalence and incidence of mental disorders in adolescents and young adults. Psychological Medicine, 37(11), 1657-1666.

Struder, H. K., & Weicker, H. (2001). Physiology and pathophysiology of the serotonergic system and its im-plications on mental and physical performance. Part II. International Journal of Sports Medicine, 22(7), 482-497.

Stubbe, J. H., de Moor, M. H., Boomsma, D. I., & de Geus, E. J. C. (2007). The association between exercise participation and well-being: A co-twin study. Preventive Medicine, 44(2), 148-152.

Teychenne, M., Ball, K., & Salmon, J. (2008). Physical activity and likelihood of depression in adults: A re-view. Preventive Medicine, 46(5), 397-411.

CHAPTER II44

Wessel, T. R., Arant, C. B., Olson, M. B., Johnson, B. D., Reis, S. E., Sharaf, B. L., et al. (2004). Relationship of physical fitness vs body mass index with coronary artery disease and cardiovascular events in women. JAMA: The Journal of the American Medical Association, 292(10), 1179-1187.

Weyerer, S. (1992). Physical inactivity and depression in the community. Evidence from the Upper Bavarian Field Study. International Journal of Sports Medicine, 13(6), 492-496.

Physical Activity and Onset of Depression in Adolescents: A Prospective Study in the General Population Cohort TRAILS

“Living with depression is like trying to keep your balance while you dance with a goat—it is perfectly sane to prefer a partner with a better sense of balance”

Andrew Solomon, The Noonday Demon, 2001

Stavrakakis N, Roest AM, Verhulst FC, Ormel J, de Jonge P & Oldehinkel AJ

J Psychiatr Res. 2013; 47(10):1304-8

CHAPTER

CHAPTER III46

ABSTRACT

Purpose: Although it has often been suggested that physical activity and depression are

intertwined, only few studies have investigated whether specific aspects of physical activ-

ity predict the incidence of major depression in adolescents from the general population.

Therefore the aim of this study was to investigate the effects of nature, frequency, duration

and intensity of physical activity during early adolescence on the onset of a major depressive

episode in early adulthood.

Methods: In a population sample of adolescents (N = 1396), various aspects of physical ac-

tivity were assessed at early adolescence (mean age = 13.02, SD = 0.61). Major depressive

episode onset was assessed using the Composite International Diagnostic Interview. A Cox

regression model was performed to investigate whether physical activity characteristics

and their interactions with gender predicted a major depressive episode onset up until mean

age 18.5 (SD = 0.61).

Results: The individual characteristics of physical activity (nature, frequency, duration and

intensity) or their interactions with gender did not predict a major depressive episode onset

(p-values>0.05).

Conclusion: So far, there is no prospective evidence that physical activity protects against

the development of adolescent depressive episodes in either boys or girls.

47PHYSICAL ACTIVITY AND ONSET OF DEPRESSION

INTRODUCTION

Physical activity (PA) has been shown to improve both physical and mental health, espe-

cially depression, in adults (Deslandes et al., 2009; Dunn, Trivedi, & O’Neal, 2001; Penedo &

Dahn, 2005). The possibility that PA can reduce the risk of depression is particularly relevant

in adolescence, a period of life characterized by a high incidence of depression (Hankin et al.,

1998). However, only few studies have investigated the association between PA and depres-

sion in adolescents, and the results have been inconsistent so far (for a detailed review see:

Biddle & Asare, 2011). A possible reason for these inconsistencies is that various aspects

of PA affect the development of depression in diverse ways (Allison et al., 2005; Johnson

& Taliaferro, 2011; Tao et al., 2007). To our knowledge, there are no studies on the specific

effects of nature, duration and frequency of PA on depression in adolescents. A few stud-

ies have investigated the effects of PA intensity on depressive symptoms in adolescents,

but their results are equivocal (Allison et al., 2005; Brand et al., 2010; Goldfield et al., 2011;

Johnson & Taliaferro, 2011; Tao et al., 2007). Allison et al. (2005) found that vigorous PA was

inversely associated with problems in social functioning, but it was not related to depres-

sion or anxiety after adjustment for age, gender and socioeconomic status (SES). Tao et al.

(2007) found that light to moderate PA protected against the development of depressive

symptoms, whereas high intensity PA was a risk factor for suicidal ideation and psychologi-

cal distress. However, Tao et al.’s (2007) findings might be explained by cultural differences,

since the population target was on Chinese adolescents. In contrast, Goldfield et al. (2011)

found that light to moderate intensity PA was not related to depressive symptoms, whereas

high intensity PA was associated with reduced depressive symptoms in boys. Goldfield’s

findings suggest that gender differences might be a source of heterogeneity in effects as

well (Goldfield et al., 2011). Finally, Brand et al. (2010) found that adolescent athletes who

exercised vigorously reported improved sleep and psychological functioning, including lower

depressed mood than non-athletes. Another reason for the lack of clarity regarding the in-

fluence PA may have on adolescent depression is that most associations tested were cross-

sectional (see reviews: Biddle & Asare, 2011 and Johnson & Taliaferro, 2011). One study

found an inverse association between childhood PA and depression in late adulthood (Jacka

et al., 2011) but this finding was based on retrospective reports of childhood PA and did not

involve specific aspects of PA. Therefore, it remains to be elucidated whether specific as-

pects of PA have the potential to prevent depressive episodes in adolescents, and whether

this preventive potential is equal for boys and girls.

CHAPTER III48

Aim

The aim of this study was to investigate the prospective association of the nature (leisure-

time PA), frequency, duration and intensity of PA during early adolescence with the onset

of major depression during adolescence, and gender differences therein. We hypothesized

that the amount of leisure-time PA, frequency, duration and intensity would be negatively

associated with the probability to develop a major depressive episode (MDE) and that the

associations would be stronger in boys than in girls.

METHODS

Design

This study is part of the Tracking Adolescents’ Individual Lives Survey (TRAILS), a Dutch

prospective cohort study of adolescents. The Dutch Central Committee on Research In-

volving Human subjects approved the study to be undertaken. Data collection for the first

measurement wave (T1) started in 2001, while for the remaining waves data collection took

place at intervals of approximately 2.5 years. Previous TRAILS publications (de Winter et al.,

2005; Huisman et al., 2008; Ormel et al., 2012) provide extensive description(s) of TRAILS

main design, data collection and sample selection. Informed consent was collected from the

parents at T1, while for the second (T2) and third wave (T3) informed consent was obtained

from both parents and participants. In the fourth wave (T4), the participants had reached the

age of 18 and parental consent was not requested (Ormel et al., 2012). In this study we used

data from T2 and T4.

Participants

At baseline, 2935 adolescents were requested to participate, of whom 2230 (response rate

76.0%; 51% girls, mean age = 11.1, SD = 0.55) agreed to do so. The response rates at T2 and

T4 were 96.4% (N = 2149; 51% girls, mean age = 13.65, SD = 0.53), and 84.3% (N = 1881; 52.3%

girls, mean age 19.1, SD = 0.60) respectively.

Procedure

During T2 questionnaires were sent through the mail to the parents or guardians, while

the adolescents completed questionnaires at school. During T4, the adolescents filled out

a web-based questionnaire, and were interviewed at home by a trained test assistant (for

more information on recruitment procedures for all waves see: Nederhof et al., 2012).


Measures

Physical Activity

At T2, adolescents were asked to rate how often they engaged in PA during an average week:

“How many days in an average week do you take part in physical activities?”. These ques-

tions were rated on an eight-point scale ranging from 0 = never, up to 7 = 7 days per week.

Furthermore, they could indicate the frequency and amount of time spent on up to four

specific sports. Additional questions concerned the frequency and duration of walking or

cycling to school or work and walking or cycling in free time. The percentage of leisure time

PA (% LTPA) was calculated as a percentage of time spent in leisure time compared to the

time spent on walking and cycling to school or work. Frequency was defined as the number

of times the participant engaged in PA and duration (in minutes) of the PA per time, regard-

less of the nature of the PA. If participants reported walking or cycling both as a sport and in

their free time, we only used the sports information to calculate the frequency and duration

to avoid duplicate answers. Based on visual inspection of the duration data, participants

who reported spending more than 225 minutes on a single activity were considered outliers

and excluded from the analyses. The intensity of PA was expressed in metabolic equivalent

of task (MET) scores, which were assigned to each sport using the MET score 2011 compen-

dium (Ainsworth et al., 2011). Based on these scores we created two intensity measures:

the peak MET score (the highest MET score) and the mean MET score across all activities. In

case a participant did not report any sport, the MET scoring of cycling or walking in leisure

time was used.

Depression

Depression was operationalized as a DSM-IV MDE and assessed using the World Health Or-

ganization Composite International Diagnostic Interview (CIDI) version 3.0 (Kessler & Ustun,

2004), which was administered at T4. The CIDI is a structured diagnostic interview, which

yields the occurrence and age at onset of psychiatric disorders according to DSM-IV criteria,

and has been shown to have good reliability and validity (Kessler et al., 2004). Because we

aimed to investigate the effect of PA on the incidence of depression, adolescents who had

developed an MDE before T2 (n = 103) were excluded from the analyses.

Socioeconomic Status

A measure of SES was obtained by averaging five standardized variables, describing the

educational attainment and professional occupation of both parents and household income.

CHAPTER III50

Statistical Analysis

All analyses were conducted using the Statistical Package for the Social Sciences (SPSS Inc.,

Chicago, IL.) version 20. Two-tailed tests with an alpha-level of 0.05 were performed. Cox re-

gression analysis was used to investigate whether the T2 PA measures (%LTPA, frequency,

duration and intensity) predicted the onset of an MDE between T2 and T4. SES and gender

were included as covariates. The proportional hazards (PH) assumption was tested using

the Schoenfield residuals. In case this test was significant for any of the predictor variables

an extended Cox model was used to examine whether that variable interacted significantly

with the time variable. If the interaction was not significant a normal Cox regression model

was used. First, we tested a model including the main effects of the PA measures (%LTPA,

frequency, duration, intensity) and covariates as predictors. After that, we added interactions

of each of the PA measures with gender one by one. If the interaction term was significant,

it was kept in the model. Finally, we examined to what extent the exclusion of adolescents

due to outliers had influenced the results by repeating the analyses including these cases.

RESULTS


Of all T4 participants (n = 1881), 408 were excluded because they either did not complete the

CIDI interviews or had developed an MDE before T2. A further 18 participants were excluded

from the final analysis because the values reported for the average duration per activity

were considered outliers. Finally, 59 participants had missing values on the PA and SES

questions and therefore were also excluded from the analysis. The final sample consisted of

1396 participants (mean age at T2 = 13.02, SD = 0.61; mean age at T4 = 18.53, SD = 0.61).

739 (52.9%) girls and 657 boys, of whom 101 (13.6%) and 39 (5.9%) respectively, reported an

MDE onset between T2 and T4. Table 1 shows the total number of minutes spent on PA per

week, by gender, SES category, and MDE. The percentage spent on LTPA, the weekly fre-

quencies of PA, the duration in minutes spent per activity, and the mean and peak intensity

of activities for boys and girls are shown in table 2.

Prediction of Depression Onset by Physical Activity

The PA frequency variable violated the proportional hazards assumption (Schoenfield re-

sidual = 0.17, p<0.05), but the extended Cox regression model indicated that the interaction

effect of frequency and time was not significant (p-value = 0.18), so we performed a regular

Cox regression analysis. The results can be seen in Table 3. The main effect of gender was

significant (with adolescent girls being more likely to develop an MDE than boys) but neither


Table 1. Minutes per week PA, by Gender, SES, and MDE.

Variables (N) Minutes per week of General PA (Median and Interquartile range)

GenderFemale (N = 739)Male (N = 657)

360 (210-735)410 (245-815)

SESLow SES (N = 262)Interm. SES (N = 701)High SES (N = 433)

350 (190-752)375 (225-775)420 (245-763)

MDE after T2Yes (N = 140)No (N = 1256)

380 (200-774)385 (226-765)

PA = physical activity, SES = socioeconomic status, MDE = major depressive episode.

Table 2. PA Nature (LTPA), Frequency, Duration and Intensity according to Gender and in the Total Sample.

Sample size LTPA (%)

PA Frequency (times per week)

PA Duration (time)

PA Intensity (peak)

PA Intensity(mean)

Males (n = 657)MedianInterq. Range

8873-95

11 8-15

4026-59

76-8

32-4

Females (n = 739)MedianInterq. Range

8467-94

108-14

3723-55

64-7

32-4

Total (n = 1396)MedianInterq. Range

8669-94

118-14

3824-58

75-7

32-4

PA = physical activity, %LTPA = percentage of time spent on leisure time physical activities.

CHAPTER III52

any of the PA measures nor their interactions with gender (p-value between 0.13 and 0.74)

significantly predicted MDE onset.

As mentioned earlier, 18 subjects were excluded from the analysis because their average

duration per activity was considered unrealistic (over 4 hours per activity). Including these

subjects in the analyses did not change the results (data upon request).

DISCUSSION

To our knowledge, this is the first study on the prospective relationship between specific

aspects of PA and the risk of developing an MDE in adolescents. Contrary to our expecta-

tions we did not find an effect of any of these aspects, nor interaction effects between PA

aspects and gender.

Studies in adults suggest that PA may help prevent depression (Farmer et al., 1988; Kritz-

Silverstein, Barrett-Connor, & Corbeau, 2001; Strawbridge, Deleger, Roberts, & Kaplan, 2002),

but only few studies investigated the preventive role of PA in adolescents. Adolescence is a

critical period with a high incidence of psychiatric disorders (Hankin et al., 1998). Prospective

studies with assessments before the onset can help to understand the mechanisms that

Table 3. Relationship of PA and MDE.

B SE p-value Hazard Ratio 95% CI

%LTPA -0.16 0.11 0.12 0.85 0.69-1.05

PA Frequency 0.08 0.12 0.50 1.09 0.86-1.38

PA Duration -0.02 0.10 0.81 0.98 0.81-1.18

PA MET Peak -0.20 0.13 0.11 0.82 0.64-1.05

PA MET Mean 0.21 0.16 0.19 1.23 0.90-1.68

Gender1 -0.84 0.19 <0.001 0.43 0.30-0.63

SES2 -0.07 0.11 0.53 0.93 0.75-1.16

n = 1396, all PA variables were z-transformed prior to analysis. PA = physical activity, %LTPA = percentage of leisure-time physical activity, MDE = major depressive episode, MET = metabolic equivalent of tasks, SES = socioeconomic status, SE = standard error, CI = confidence intervals.1 males are the reference group. 2 SES was used as a continuous variable in analyses.


lead to the development of mental health problems. In a previous study involving the same

sample (Stavrakakis, de Jonge, Ormel, & Oldehinkel, 2012), we studied prospective relation-

ships between general levels of PA and depressive symptoms between age 11 and 16, and

found weak but significant negative associations, in both directions. In the present study, we

focused on onset of depressive disorder and specific aspects of PA, and we did not find evi-

dence of such effects. This might indicate that the association between PA and depressive

symptoms found is mainly caused by the potential of PA to reduce symptoms in those suf-

fering from depression (Mota-Pereira et al., 2011; Pilu et al., 2007; Silveira et al., 2013), or that

PA is more strongly associated with sub-threshold depressive symptoms than with clinical

depression. However, it is still possible that PA can prevent an onset of MDE in late adult-

hood, as reported by Jacka et al. (2011), even though we did not find an effect on adolescent

depression. Alternatively, the different results of these two studies might be due to meth-

odological differences between the two. Although Jacka et al.’s study (2011) had a larger

sample than ours (n = 2152 vs. n = 1396), they only used self-reports for defining depression

and not a validated and reliable measure such as the CIDI. Furthermore, they used a sample

of older persons (median age for males = 55; median age for females = 57) who were asked to

retrospectively calculate their PA levels at the age of 15. This might have introduced error in

their results, since activity levels across the lifespan have been reported to be only weakly

correlated (Friedman et al., 2008). In general, the relationship between PA and depression or

depressive symptoms is a complex one. Networks of genetic and biological factors, person-

ality, SES, and lifestyle habits interact and jointly affect depressive symptoms. In combina-

tion with specific other factors, PA might prevent the onset of a depressive episode.

Although there are indications that males and females benefit differently from different lev-

els of PA participation (Asztalos, De Bourdeaudhuij, & Cardon, 2010; Mikkelsen et al., 2010),

we did not observe an interaction between PA aspects and gender on MDE onset. For adults,

gender differences in depression risk have been suggested with regard to the intensity of

PA; women have been reported to benefit more from low intensity activities and men benefit

more from higher intensity activities. In our study, PA intensity did not prevent the onset of

depression in either gender.

It is important to elucidate the true role of PA on depression, since its effect may have been

overestimated in the literature (Daley & Jolly, 2012). Reviews have suggested that PA is

beneficial in alleviating depression (Dunn et al., 2001; Penedo & Dahn, 2005) but most of

the reviewed findings concerned cross-sectional associations in observational studies,

small samples of clinically depressed patients, short follow-up periods or non-clinical vol-

unteers in intervention studies (Daley & Jolly, 2012). In other words, far-reaching conclusions

have been drawn concerning the ability of PA to treat and prevent depression, while in fact

the empirical basis is less firm than often thought. Our study by no means excludes the

CHAPTER III54

possibility that PA can help to reduce depressive symptoms in healthy individuals as shown

previously (Brand et al., 2010; Sigfusdottir, Asgeirsdottir, Sigurdsson, & Gudjonsson, 2011;

Stavrakakis et al., 2012), but clearly suggests that PA cannot prevent or postpone the de-

velopment of incident depressive episodes in adolescents.

A strength of this study is its large sample size, which provides sufficient power to detect

relevant differences and avoid type II errors. A further advantage is the young age of the

sample; the probability of confounding by prior depressive episodes or somatic conditions is

relatively low in adolescence; and the (retrospective) dating of first MDE onsets is less liable

to recall bias than in older samples. Furthermore, the longitudinal design, with a follow-up

period of over 5 years, made it possible to investigate prospectively whether early PA en-

gagement may protect against MDE onset. Finally, the use of the CIDI interviews, adminis-

tered by trained test assistants, provided standardized measures of DSM-IV depression and

its onset.

However, the results of this study have to be interpreted with some caution due to sev-

eral limitations. First, all PA measures were based on self-reports. Although self-reports

can provide an accurate measure of an individual’s engagement in PA (Haskell, 2012; Mil-

ton, Clemes, & Bull, 2012), more objective measures such as accelerometers are preferred

when possible. Self-reports can lead to an over-estimation of the levels of PA (due to social

pressures) in some participants and therefore partly explain the lack of associations found.

Second, we did not use a validated questionnaire, such as the International Physical Ac-

tivity Questionnaire (IPAQ), and therefore cannot be certain on the trustworthiness of the

responses. Finally, the measure of the nature of PA (%LTPA) is quite difficult to be used

accurately in adolescence, since engagement in occupational PA is not that clear compared

to adulthood. This measure was broad and only measured approximately the percentage of

time spent on leisure-time compared to non-leisure time activities (commuting in this case).

Therefore, conclusions about differential associations of PA nature (leisure-time vs. occupa-

tional activities) on depression onset should be tentative.

CONCLUSION

We did not find any evidence that the nature, frequency, duration or intensity of PA influenced

the probability to develop a depressive episode in adolescence in neither boys nor girls.


REFERENCES

Ainsworth, B. E., Haskell, W. L., Herrmann, S. D., Meckes, N., Bassett, D. R.,Jr, Tudor-Locke, C., et al. (2011). 2011 compendium of physical activities: A second update of codes and MET values. Medicine and Science in Sports and Exercise, 43(8), 1575-1581.

Allison, K. R., Adlaf, E. M., Irving, H. M., Hatch, J. L., Smith, T. F., Dwyer, J. J., et al. (2005). Relationship of vigorous physical activity to psychologic distress among adolescents. Journal of Adolescent Health, 37(2), 164-166.

Asztalos, M., De Bourdeaudhuij, I., & Cardon, G. (2010). The relationship between physical activity and men-tal health varies across activity intensity levels and dimensions of mental health among women and men. Public Health Nutrition, 13(8), 1207-1214.


Brand, S., Gerber, M., Beck, J., Hatzinger, M., Puehse, U., & Holsboer-Trachsler, E. (2010). High exercise levels are related to favorable sleep patterns and psychological functioning in adolescents: A comparison of athletes and controls. Journal of Adolescent Health, 46(2), 133-141.

Daley, A., & Jolly, K. (2012). Exercise to treat depression. British Medical Journal (Clinical Research Ed.), 344, e3181.


Deslandes, A., Moraes, H., Ferreira, C., Veiga, H., Silveira, H., Mouta, R., et al. (2009). Exercise and mental health: Many reasons to move. Neuropsychobiology, 59(4), 191-198.

Dunn, A. L., Trivedi, M. H., & O’Neal, H. A. (2001). Physical activity dose-response effects on outcomes of depression and anxiety. Medicine and Science in Sports and Exercise, 33(6 Suppl), S587-97; discussion 609-10.

Farmer, M. E., Locke, B. Z., Moscicki, E. K., Dannenberg, A. L., Larson, D. B., & Radloff, L. S. (1988). Physical activity and depressive symptoms: The NHANES I epidemiologic follow-up study. American Journal of Epidemiology, 128(6), 1340-1351.

Friedman, H. S., Martin, L. R., Tucker, J. S., Criqui, M. H., Kern, M. L., & Reynolds, C. A. (2008). Stability of physical activity across the lifespan. Journal of Health Psychology, 13(8), 1092-1104.

Goldfield, G. S., Henderson, K., Buchholz, A., Obeid, N., Nguyen, H., & Flament, M. F. (2011). Physical activity and psychological adjustment in adolescents. Journal of Physical Activity & Health, 8(2), 157-163.


Haskell, W. L. (2012). Physical activity by self-report: A brief history and future issues. Journal of Physical Activity & Health, 9 Suppl 1, S5-10.


Jacka, F. N., Pasco, J. A., Williams, L. J., Leslie, E. R., Dodd, S., Nicholson, G. C., et al. (2011). Lower levels of physical activity in childhood associated with adult depression. Journal of Science and Medicine in Sport / Sports Medicine Australia, 14(3), 222-226.


CHAPTER III56

Kessler, R. C., Abelson, J., Demler, O., Escobar, J. I., Gibbon, M., Guyer, M. E., et al. (2004). Clinical calibration of DSM-IV diagnoses in the World Mental Health (WMH) version of the World Health Organization (WHO) Composite International Diagnostic Interview (WMHCIDI). International Journal of Methods in Psychiat-ric Research, 13(2), 122-139.

Kessler, R. C., & Ustun, T. B. (2004). The World Mental Health (WMH) survey initiative version of the World Health Organization (WHO) Composite International Diagnostic Interview (CIDI). International Journal of Methods in Psychiatric Research, 13(2), 93-121.

Kritz-Silverstein, D., Barrett-Connor, E., & Corbeau, C. (2001). Cross-sectional and prospective study of ex-ercise and depressed mood in the elderly: The Rancho Bernardo Study. American Journal of Epidemiol-ogy, 153(6), 596-603.

Mikkelsen, S. S., Tolstrup, J. S., Flachs, E. M., Mortensen, E. L., Schnohr, P., & Flensborg-Madsen, T. (2010). A cohort study of leisure time physical activity and depression. Preventive Medicine, 51(6), 471-475.

Milton, K., Clemes, S., & Bull, F. (2012). Can a single question provide an accurate measure of physical activ-ity? British Journal of Sports Medicine, 47(1), 44-48.

Mota-Pereira, J., Silverio, J., Carvalho, S., Ribeiro, J. C., Fonte, D., & Ramos, J. (2011). Moderate exercise im-proves depression parameters in treatment-resistant patients with major depressive disorder. Journal of Psychiatric Research, 45(8), 1005-1011.



Penedo, F. J., & Dahn, J. R. (2005). Exercise and well-being: A review of mental and physical health benefits associated with physical activity. Current Opinion in Psychiatry, 18(2), 189-193.

Pilu, A., Sorba, M., Hardoy, M. C., Floris, A. L., Mannu, F., Seruis, M. L., et al. (2007). Efficacy of physical activ-ity in the adjunctive treatment of major depressive disorders: Preliminary results. Clinical Practice and Epidemiology in Mental Health: CP & EMH, 3, 8.

Sigfusdottir, I. D., Asgeirsdottir, B. B., Sigurdsson, J. F., & Gudjonsson, G. H. (2011). Physical activity buffers the effects of family conflict on depressed mood: A study on adolescent girls and boys. Journal of Ado-lescence, 34(5), 895-902.

Silveira, H., Moraes, H., Oliveira, N., Coutinho, E. S., Laks, J., & Deslandes, A. (2013). Physical exercise and clinically depressed patients: A systematic review and meta-analysis. Neuropsychobiology, 67(2), 61-68.

Stavrakakis, N., de Jonge, P., Ormel, J., & Oldehinkel, A. J. (2012). Bidirectional prospective associations between physical activity and depressive symptoms. The TRAILS study. Journal of Adolescent Health, 50(5), 503-508.

Strawbridge, W. J., Deleger, S., Roberts, R. E., & Kaplan, G. A. (2002). Physical activity reduces the risk of subsequent depression for older adults. American Journal of Epidemiology, 156(4), 328-334.

Tao, F. B., Xu, M. L., Kim, S. D., Sun, Y., Su, P. Y., & Huang, K. (2007). Physical activity might not be the pro-tective factor for health risk behaviours and psychopathological symptoms in adolescents. Journal of Paediatrics and Child Health, 43(11), 762-767.

Plasticity Genes do not Modify Associations between Physical Activity and Depressive Symptoms

“Let us understand what our own selfish genes are up to, because we may then at least have a chance to upset their designs, something that no other species has ever aspired to do.”

Richard Dawkins, The Selfish Gene, 1989

Stavrakakis N, Oldehinkel AJ, Nederhof E, Oude Voshaar RC, Verhulst FC, Ormel J & de Jonge P

Health Psychol. 2013; 32(7):785-92

CHAPTER

CHAPTER IV58

ABSTRACT

Purpose: Physical activity is inversely associated with depression in adolescents but the

overall associations are fairly weak, suggesting individual differences in the strength of the

associations. The aim of this study was to investigate whether plasticity genes modify the

reciprocal prospective associations between physical activity and depressive symptoms

found previously.

Methods: In a prospective population-based study (N = 1196), physical activity and depres-

sive symptoms were assessed three times, around the ages of 11, 13.5 and 16. Structural

Equation Modeling was used to examine reciprocal effects of physical activity and depressive

symptoms over time. The plasticity genes examined were 5-HTTLPR, DRD2, DRD4, MAOA,

TPH1, 5-HTR2A, COMT, and BDNF. A cumulative gene plasticity index consisting of three

groups (low, intermediate and high) according to the number of plasticity alleles carried by

the adolescents was created. Using a multi-group approach we examined if the associations

between physical activity and depressive symptoms differed between the three cumulative

plasticity groups as well as between the individual polymorphisms.

Results: We found significant cross-sectional and cross-lagged paths from physical activ-

ity to depressive symptoms and vice versa. Neither the cumulative plasticity index nor the

individual polymorphisms modified the strengths of these associations.

Conclusion: Associations between adolescents’ physical activity and depressive symptoms

are not modified by plasticity genes.

59GENES, PHYSICAL ACTIVITY AND DEPRESSIVE SYMPTOMS

INTRODUCTION

Individuals carrying specific genotypes have a higher sensitivity to environmental influences

than others (Belsky et al., 2009). Although research initially focused on the negative conse-

quences of these genotypes in the context of the classic diathesis-stress framework, more

recent work stresses the possibility that individuals who suffer most from negative experi-

ences might also benefit most from environmental support or the absence of adversity (Bel-

sky & Beaver, 2011; Ellis, Boyce, Belsky, Bakermans-Kranenburg, & van Ijzendoorn, 2011).

This ‘For better and for worse’ manner in which individuals are more responsive to environ-

mental influences suggests that individuals who carry specific genotypes might benefit more

from positive environmental influences (such as family or social support) in early life, while

they might also be more vulnerable to negative environmental influences (such as child mal-

treatment and negative life events) in developing psychopathology later on. Genes thought

to be responsible for this differential susceptibility have been labeled plasticity genes.

Belsky and Pluess (2009) proposed several genes to operate as plasticity genes including

the serotonin transporter gene (SLC6A4), the dopamine receptor 2 and 4 genes (DRD2 &

DRD4), the dopamine transporter gene (DAT1), the serotonin 2a receptor gene (5-HTR2A),

the catechol-O-methyl transferase gene (COMT), the monoamine oxidase gene (MAOA), and

the tryptophan hydroxylase 1 gene (TPH1). Although these genes are not the only likely

candidates, they have been shown most consistently to act as plasticity genes in the lit-

erature (for a review see Belsky & Pluess, 2009). Belsky and Pluess proposed a cumulative

genetic plasticity model, in which the susceptibility to particular environmental influences is

hypothesized to increase with increasing numbers of plasticity alleles (Belsky et al., 2009;

Belsky & Beaver, 2011).

In adults, inverse associations between physical activity (PA) and depressive symptoms have

been shown previously (Penedo & Dahn, 2005), supporting the idea that engaging in PA can

at least prevent the development of depressive symptoms. In adolescents, fewer studies

have investigated this relationship, and the results have been inconsistent so far (Johnson &

Taliaferro, 2011). A recent prospective study in older adults demonstrated reciprocal inverse

relationships between PA and depressive symptoms (Lindwall, Larsman, & Hagger, 2011).

They found that there was a stronger relationship between PA predicting later depressive

symptoms than between depressive symptoms predicting later engagement in PA. We also

have reported reciprocal prospective associations between PA and depressive symptoms in

a cohort of Dutch adolescents (Stavrakakis et al., 2012). Associations were measured using

a cross-lagged path analysis model involving three measurements. The associations found

were significant but rather weak, indicating a high amount of unexplained variance. Part of

this unexplained variance may be due to individual differences in both the potential benefits

CHAPTER IV60

of PA in alleviating depressive symptoms and the negative effects of depressive symptoms

on PA levels. A better understanding of these individual differences is needed in order to

design effective interventions for treating affective disorders.

We hypothesized that the plasticity genes might underlie part of the assumed individual

differences in the reciprocal associations between PA and depressive symptoms. In their

review, Belsky and Pluess (2009), proposed a theoretical framework in which they suggest

that these individuals might have lower thresholds for experiencing pleasure and/or displea-

sure, which might result in differential susceptibility to environmental influences. Although

this is speculative, two major neurobiological mediating mechanisms have been identified,

namely the serotonergic (experience of displeasure) and dopaminergic systems (reward sen-

sitivity and sensation seeking). These two systems have been implicated in the aetiology

of depression (Lapin & Oxenkrug, 1969; van Praag & Korf, 1970) but also have been shown

to be influenced by PA (Dishman et al., 2006). Alterations in serotonin and noradrenaline

levels have been suggested to be the main reason for the beneficial effects of PA on stress-

resilience and mood (Dishman et al., 2006). Evidence for an effect of PA on the dopaminergic

system comes mainly from animal research where treadmill running increased the release

(Meeusen, Piacentini, & De Meirleir, 2001) and turnover of dopamine (Hattori, Naoi, & Nishino,

1994) in the striatum of rats. Increasing dopamine levels is known to be an effective way

to improve mood (Lemke, Brecht, Koester, & Reichmann, 2006). Therefore, these plastic-

ity genes might modify the reciprocal association between PA and depressive symptoms

through their differential action on the serotonergic and dopaminergic systems.

To date, only three studies have tried to elucidate whether specific genotypes may be as-

sociated with the effects of PA on depressive symptoms in humans. Rethorst, Landers, Na-

goshi and Ross (2010) showed that after 30 minutes of exercise carriers of a long allele of

the serotonin-transporter-linked polymorphic region (5-HTTLPR) exhibited greater reduc-

tion in depressive symptoms than individuals with two short alleles. In a cross-sectional

study, Rethorst and colleagues (2011) demonstrated that the 5-HTTLPR gene modified the

relationship between PA and depressive symptoms in a sample of 170 students. Finally,

Mata, Thompson and Gotlib (2010) conducted a study in adolescent girls, and found that

PA involvement reduced depressive symptoms in carriers of the Brain Derived Neurotrophic

Factor (BDNF) met allele, but not in carriers of the val/val polymorphism. BDNF is a growth

factor that is encoded by the BDNF gene and is abundant throughout the human body. Bel-

sky et al. did not include the BDNF gene in their list of plasticity genes. However, because

BDNF seems to be related to both PA (Knaepen et al., 2010; Goekint et al., 2010) and depres-

sive symptoms (Hashimoto, 2010; Verhagen et al., 2010), the BDNF gene was considered a

plausible modifier of the reciprocal association between PA and depressive symptoms, and

therefore was included in our study as well.


The aim of the present study was to investigate whether plasticity genes modified the

reciprocal prospective associations between PA and depressive symptoms for several rea-

sons. First, it is important to extensively test robustness of gene environment interac-

tions (GxE), as this field seems to be plagued by non-confirmations. For example, the GxE

between the short 5-HTTLPR allele and adversity on depression first reported by Caspi et

al. (2003) has not been confirmed in two meta-analyses (Risch et al., 2009; Munafo et al.,

2009), but has been justified in a later qualitative review (Caspi et al., 2010) and in the most

recent meta-analysis (Karg et al., 2011). Thus, for the advancement in the field of GxE re-

garding PA and depressive symptoms, confirmation studies are of utmost importance. Sec-

ond, previous investigations focused only on the unidirectional effects of PA on depressive

symptoms (Mata et al., 2010; Rethorst et al., 2010; Rethorst et al., 2011). We extend this

line of research by investigating possible GxE in the reciprocal association between PA and

depressive symptoms using previously developed cross-lagged models in a multi-group

analysis. Third, we extend the GxE literature regarding PA and depressive symptoms by

using an index of cumulative plasticity (Belsky & Pluess, 2009; Belsky et al., 2009; Belsky &

Beaver, 2011). Summarizing, in order to test if the associations between PA and depressive

symptoms were moderated by genetic polymorphisms, we performed multi-group analyses

with regard to both cumulative plasticity index and individual polymorphisms on cross-

lagged models of PA and depressive symptoms.

METHODS

Design

Data were used from the TRacking Adolescents’ Individual Lives Survey (TRAILS), a pro-

spective cohort (three assessment waves) of Dutch adolescents. Permission for the study

was obtained from the Dutch Central Committee on Research Involving Human Subjects.

Further information on TRAILS objectives, main design, data collection, sample selection

and non-response can be found elsewhere (de Winter et al., 2005; Huisman et al., 2008).

Parents gave informed consent in the first wave (T1), while for the second (T2) and third

wave (T3) informed consent was obtained from both the parents and the participants

themselves.

Procedure

During T1, parents or guardians were visited by TRAILS interviewers in their homes and were

administered an interview. In addition, they were also asked to complete a questionnaire.

During T2 and T3, parents had to complete a questionnaire, which was sent through the

CHAPTER IV62

mail. The adolescents filled out self-report questionnaires at school. Further information on

TRAILS procedures can be found elsewhere (de Winter et al., 2005; Huisman et al., 2008).

Participants

At T1, 2230 adolescents from a possible 2935 (51% females, mean age = 11.1, SD = 0.55)

agreed to take part in TRAILS. The response rates at T2 and T3 were 96.4% (N = 2149;

51% girls, mean age = 13.65, SD = 0.53) and 81.4% (N = 1816; 52.3% girls, mean age = 16.27,

SD = 0.73) respectively. For the present study only participants with complete genetic data

were included (N = 1196; 52.3% girls). Individuals that did not have both parents born in the

Netherlands were also excluded from the final analysis, since these SNPs can differ between

ethnic groups. This focus sample was still representative of a normal population of adoles-

cents and the problems observed therein.

Measures

Depressive Symptoms

Adolescents filled out the Affective Problems scale of the Youth Self Report (YSR; Achen-

bach, 1991b) for each assessment wave at school. Parents completed the parent version of

the YSR the Child Behavior Checklist (CBCL; Achenbach, 1991a) at home. The mean scores

of the YSR and CBCL scales were used in the analyses. The Affective Problem scale con-

tains 13 items, which correspond to DSM-IV depressive symptoms (van Lang et al., 2005).

These items include sadness, loss of pleasure, crying, self-harm, suicidal ideation, feelings

of worthlessness, guilt, loss of energy, overtiredness, eating problems, and sleeping prob-

lems; and are scored on a three-point scale (0 = never or not at all true, 1 = sometimes true

and 2 = very often or very true). One of the sleep items (‘I sleep more than most kids’) was ex-

cluded because exclusion of this item increased the internal consistency of the scale (Bosch

et al., 2009). Scores on the remaining twelve items were averaged in order to construct a

total depressive symptoms scale with an internal consistency (Cronbach’s alpha) of 0.74 at

T1, of 0.81 at T2 and 0.86 at T3.

Physical Activity

At T1 adolescents were asked to rate how many times per week they performed physi-

cal exercise (for example, swimming, playing football, horse-riding) on a five-point scale

(0 = never, 1 = once per week, 2 = two to three times per week, 3 = four to five times per week,

and 4 = five to six times per week). At T2 and T3, separate questions were asked for PA during

the summer and the winter: “How many days in an average week in the summer/winter do

you take part in physical activities?” and the responses were scored on an eight-point scale


(0 = never, to 7 = seven days per week). The answers to these two questions were averaged

and mean scores were used in the analysis. T1 responses (ranging from 0-4) were recoded

into an eight-point scale using the monotonic R7 transform recommended by Little in order

to achieve a similar metric for the PA measures at all measurement waves (Little, in press).

Plasticity Genes

The list of plasticity genes as proposed by Belsky and Pluess (2009) include DRD2, DRD4,

5-HTTLPR, MAOA, TPH1, 5-HTR2A, COMT and the DAT1. In our study DAT1 was excluded be-

cause it was not genotyped, while the BDNF gene was included for reasons explained in the

Introduction. Following the criteria set by Belsky and Pluess (2009), the A1 allele of DRD2,

the long version of DRD4 (from 7 repeats to 10 repeats), the short allele of 5-HTTLPR, the

2R/3R alleles of MAOA, the A allele of TPH1, the T allele of the 5-HTR2A, the val (G) allele of

COMT and the met allele of BDNF were defined as plasticity alleles.

DNA Extraction

Blood samples (n = 1190) or buccal swabs (Cythobrush®) (n = 275) were collected for DNA

extraction using a manual salting out procedure as described by Miller et al. (1988).

Genotyping of BDNF, DRD2, COMT, TPH1 and 5-HTR2A

The BDNF single nucleotide polymorphism (rs6265), DRD2/TaqIA (rs1800497), COMT/val-

158met (rs4680), TPH1 (rs1799913) and 5-HTR2A (rs6313) were genotyped on the Golden

Gate Illumina BeadStation 500 platform (Illumina, San Diego, California) according to the

manufacturer’s protocol. Call rates were: 81% for BDNF, 100% for DRD2, 95% for COMT,

100% for TPH1 and 100% for 5-HTR2A. All DNA samples could be amplified and concordance

between DNA replicates (n = 53) showed a 100% genotyping accuracy (Nederhof et al., 2010).

Data cleaning was completed according to recommended procedures (Steptoe, 2010). All

SNPs were well within Hardy-Weinberg equilibrium, with HWE p-values ranging between

0.42 and 0.52.

Genotyping Length Polymorphisms of 5-HTTLPR, DRD4 and MAOA

Genotyping of the length polymorphisms (LP) DRD4, MAOA, HTTLPR and SNP rs25531

(A/G SNP in L HTTLPR) was done at the Research lab for Multifactorial Diseases within the

Human Genetics department of the Radboud University Nijmegen Medical Centre in Nijme-

gen, the Netherlands. Genotyping of the HTTLPR polymorphism in the promoter region of

SLC6A4 (5-HTT, SERT) gene was performed by simple sequence length analysis. Call rate

was 91.6%. A custom-made TaqMan assay (Applied Biosystems) was utilized to genotype

the single nucleotide substitution (A to G) which is present in the HTTLPR long (l) allele

CHAPTER IV64

(rs25531). Call rate was 96.5%. Concordance between DNA replicates showed an accuracy of

100%. All lg alleles were recoded into s’, because it has been shown that this polymorphism

represents low serotonin expression comparable to the s’ allele, while la was recoded as l’

(Nederhof et al., 2010). The 48 bp direct repeat polymorphism in exon 3 of DRD4 was geno-

typed on the Illumina BeadStation 500 platform (Illumina Inc., San Diego, CA, USA). Three

percent blanks as well as duplicates between plates were taken along as quality controls

during genotyping. Determination of the length of the alleles was performed by direct analy-

sis on an automated capillary sequencer (ABI3730, Applied Biosystems, Nieuwerkerk a/d

IJssel, the Netherlands) using standard conditions. Call rate for DRD4 was 99.4%. The 30bp

variable number of tandem repeat polymorphism (called MAOA-LPR or MAOA-uVNTR) was

also genotyped on the Illumina BeadStation 500 platform (Illumina Inc., San Diego, CA, USA).

Three percent blanks as well as duplicates between plates were taken along as quality con-

trols during genotyping. Call rate was 100% for MAOA. All polymorphisms were well within

Hardy-Weinberg equilibrium (HWE p-values ranged from 0.77 to 0.87).

Cumulative Genetic Plasticity Index

Each polymorphism was assigned one point if a plasticity genotype was present and these

values were summed to create a cumulative index. Girls who were heterozygous for the

MAOA gene, which is located on the X chromosome, were categorized into the low activ-

ity (i.e., high plasticity) genotype as proposed by Belsky and Beaver (2011). Three groups

were created according to the number of plasticity genotypes present: a low group (high

plasticity in 0-3 genes), intermediate group (high plasticity in 4-5 genes) and a high group

(high plasticity in 6-8 genes). Initially the low group was defined as individuals who were

not carrying any plasticity genotypes, but because of insufficient power (n = 22) we merged

this group with the individuals carrying one to three plasticity genotypes. The three groups

were of unequal size, with the intermediate group having the most individuals. We expected

the intermediate group to be the most heterogeneous with regard to each of the individual

genes, therefore we deemed necessary to create relatively equally sized groups for the high

and lowest plasticity groups.

Covariates

Gender and socioeconomic status (SES) were included as covariates. Five standardized vari-

ables (professional occupation and educational attainment for both father and mother, and

household income) were averaged and used to calculate SES. The lowest 25% of scores were

considered as ‘low SES’, the highest 25% as ‘high SES’ and the remaining were grouped as

‘intermediate SES’.


Analysis

Imputation of Missing Data

A detailed account of attrition rates within the TRAILS study can be found elsewhere (Huis-

man et al., 2008). Multiple imputations were performed for all variables except for the miss-

ing genotyping data (N = 1196 at all three measurement waves) using STATA version 10

(STATA corp., College station, Texas). Five different datasets (Rubin, 1996) were created and

were subsequently imported into Mplus 5 (Muthén & Muthén, 2007) in order to be used in

the analysis.

Cross-Lagged Path Model

A cross-lagged path model (Structural Equation Modeling) using Mplus was created to in-

vestigate cross-sectional and prospective associations between PA and depressive symp-

toms. Using this model we estimated the direction of the prospective effect (cross-lagged)

between PA and depressive symptoms but also the autoregressive paths (relative stability

over time) and the cross-sectional covariances (interpreted as correlations at T1 and as

correlated change at T2 and T3). Correlated change and cross-lagged paths reflect longi-

tudinal relationships. Due to positive skewness of the depression variables, they were log

transformed prior to analysis. The cross-lagged path model was also adjusted for gender

and SES. A Comparative Fit Index (CFI) and Tucker-Lewis Index (TLI) greater than 0.95 and

a Root Mean Square Error of Approximation (RMSEA) lower than 0.05 were considered a

good model fit.

Cumulative Plasticity Index and Individual Genes Comparisons

A cross-lagged path model was specified for the three groups based on the cumulative

plasticity index. To test whether there was a group difference in the reciprocal associations

between PA and depressive symptoms, we compared a model where the cross-sectional

paths at T1 and the cross-lagged paths were constrained to be equal between groups with

a model where these paths were allowed to differ. To test whether the individual polymor-

phisms affect the reciprocal association between PA and depressive symptoms a similar de-

sign was used for each of the individual genes. A chi-square difference test was performed to

test whether the differences between the models reached statistical significance. To avoid

possible chance findings due to multiple testing a significant chi-square difference score at

the 0.01 level was considered a significant difference between groups.

CHAPTER IV66

RESULTS


Descriptive statistics of the main variables used in this study are shown in Table 1. Both depres-

sive symptoms and PA remained stable over time. Frequencies for the cumulative plasticity

index and the frequencies for the different genotypes in individual genes are shown in Table 2.

Relationships between Depression and PA

Relationships between the total depressive symptoms score and PA are shown in Figure

1 (overall model fit, χ2 = 6.445, df = 6, p>0.05). Fit indices showed an excellent fit of model

(CFI = 0.999, TLI = 0.999, RMSEA = 0.008), which was significantly better than the fit of the

model without cross-lagged paths from PA to depressive symptoms (χ2difference = 6.825,

df = 1, p<0.05) and the model without cross-lagged paths from depressive symptoms to PA

(χ2difference = 8.333, df = 1, p<0.05). As in our previous paper with more participants, all cross-

lagged paths were significant, indicating that depressive symptoms at any wave predict-

ed PA at the next and vice versa. The autoregressive estimate of depression was higher

than the autoregressive estimate of PA. The cross-sectional association (and correlational

change) between PA and depression was significant at all assessment waves.

Multi-Group Analysis

The multi-group analysis for the three different putative plasticity groups (low, intermediate

and high) is shown in Table 3. Because the middle group was expected to be more heteroge-

neous with regard to each of the individual genes, we first compared the highest group against

the lowest group (χ2unconstrained = 9.456, df = 12, χ2

constrained = 16.897, df = 16) and then compared

all groups with each other (χ2unconstrained = 16.709, df = 18, χ2

constrained = 24.728, df = 26). The fit

indices for both models were excellent (CLI = 0.998, TLI = 0.995, RMSEA = 0.015). As seen in

the table, the χ2difference test was not significant at the 0.01 level.

Table 1. Means (Standard Deviations) of the Main Variables (N = 1196).

T1(mean age 11.1)

T2(mean age 13.5)

T3 (mean age 16.3)

Total Depressive Symptoms

0.21 (0.19) 0.19 (0.21) 0.21 (0.25)

Physical Activity 3.52 (2.09) 3.38 (1.68) 3.51 (1.85)

T1 = first measurement wave, T2 = second measurement wave and T3 = third measurement wave.


Table 2. Frequencies of the Plasticity Index and Individual Genes.

Groups (N = 1196)

Low Group(0-3 plasticity alleles)

Intermediate Group (4-5 plasticity alleles)

High Group(6-8 plasticity alleles)

Plasticity Index 28.7% 52.6% 18.7%

Individual Genes Genotype Distribution (%)

BDNF Val/Val62.5%

Val/Met*34.4%

Met/Met*3.1%

DRD2 G/G 63%

A/G*33%

A/A*

4%

COMT A/A30.2%

A/G*

50.1%

G/G* 19.7%

DRD4 s/s63%

s/l*32.3%

l/l*4.7%

5-HTR2A C/C41%

C/T*44.6%

T/T*14.4%

5-HTTLPR l/l 26.2%

l/s*

50%

s/s*

23.8%

TPH1 C/C 36.5%

C/A*

48.7%

A/A*

14.8%

MAOA girls(n = 625)

High19.4%

Intermediate*55.2%

Low*25.4%

MAOA boys(n = 571)

High 37.1%

Low*

62.9%N/A

* Indicates which genotypes were used as the plasticity genotype(s).

CHAPTER IV68

Individual Polymorphisms

We carried out the same multi-group analysis approach for each of the individual polymor-

phisms, comparing the putative plasticity alleles against the non-plasticity alleles. Table 3

depicts the χ2difference test for each individual polymorphism. No significant differences were

found, just a trend (p = 0.03) for MAOA.

DISCUSSION

The results of this study demonstrate reciprocal associations between PA and depressive

symptoms consistent with previous studies (Stavrakakis et al., 2012; Lindwall et al., 2011)

in larger samples. This suggests that early engagement in PA precedes a decrease in later

depressive symptoms and vice versa.

This was the first study using a cumulative plasticity index as proposed by Belsky et al. as

a moderator in the association between PA and depressive symptoms. According to Bel-

sky and Pluess (2009), the cumulative plasticity index assumes an additive effect of these

genes, where the susceptibility to environmental influences is hypothesized to increase with

increasing numbers of plasticity alleles. These genes play a role, through the translation and

transcription of specific proteins, in the formation and degradation of serotonin and dopa-

Figure 1. Associations between Depressive Symptoms and PA over time.

.584* (.020)

.269* (.028)

.558* (.025)

.246* (.030)

-.035*(.015)

-.056*(.020)

-.055*(.026)

DepressionT1

PA T1

DepressionT2

PA T2

DepressionT3

PA T3

-.074*(.029)

-.049*(.024)

-.057*(.020)

-.048*(.020)

PA = physical activity, T1 = first measurement wave, T2 = second measurement wave and T3 = third measurement wave.The path coefficients reflect model estimated standardized beta values. The effects are adjusted for gender and socioeconomic status (SES). * p<0.05.


mine. Both of these neurotransmitters are affected by PA and physical inactivity. They also

have been implicated in the etiology of depression, therefore making them primary candi-

dates for modifying the reciprocal association between PA and depressive symptoms. How-

ever, in the current study, we did not find support for any modifying effects of the cumulative

plasticity index on the associations between PA and depressive symptoms.

We then proceeded in investigating the role of individual polymorphisms on the association

of PA and depressive symptoms. To date, there are only three published studies investigat-

ing modifying effects of specific individual polymorphisms on the association between PA

and depressive symptoms (Mata et al., 2010; Rethorst et al., 2010; Rethorst et al., 2011). The

two studies by Rethorst and colleagues investigated the 5-HTTLPR polymorphism, while

the study by Mata and colleagues investigated the BDNF polymorphism. All three studies

found an influence of specific genotypes (the carriers of met allele for BDNF and the carriers

of the short allele for 5-HTTLPR ) on the relationship between PA and depressive symptoms.

None of these findings were confirmed in our study. Both the BDNF and 5-HTTLPR geno-

types as well as the other individual genes that we investigated did not modify the reciprocal

association between PA and depressive symptoms in our sample.

Table 3. Multi-Group Analysis comparing the three different Cumulative Plasticity Groups (low, intermediate and high) and Individual Polymorphisms.

Polymorphism x2diff = x2

con. – x2uncon. dfdiff = dfcon-dfuncon

p-values

Cumulative Plasticity (high vs. low of 3 groups)

7.44 4 0.11

Cumulative Plasticity(3 groups)

8.02 8 0.43

Individual Polymorphisms x2diff = x2

con. – x2uncon. dfdiff = dfcon-dfuncon

p-values

BDNF 3.03 4 0.55

5-HTTLPR 7.12 4 0.13

DRD2 2.50 4 0.64

MAOA 10.75 4 0.03

5-HTR2A 3.64 4 0.46

COMT 5.82 4 0.21

DRD4 4.23 4 0.38

TPH1 8.41 4 0.08

CHAPTER IV70

This lack of modifying effect could be down to a number of reasons. First, perhaps these

plasticity genes are not influencing the relationship between PA and depressive symptoms

at all, contrary to our expectations. It is possible that other genes acting on different neu-

rotransmitter systems, such as the endorphin system (Hegadoren et al., 2009; Fox, 1999) or

the endocannabinoid system (Sparling et al., 2003), might be playing a more important role

in influencing the relationship between PA and depressive symptoms. Second, it is possible

that plasticity genes play a role on the association between PA and depressive symptoms

but the design chosen in this study was not appropriate to reveal these effects. This could

be either because of measurement error (see later section on strengths and limitations for

further information) or because the measurement waves were not close enough to capture

the processes involved in the interrelationships of plasticity, PA and depressive symptoms.

In animal research the ability to remain active (through free wheel running) is thought to be

an enriching environment with numerous benefits on behavior and physiology. In humans, PA

can be considered a supportive or an enriching experience, especially when individuals start

exercising from a very early age and find it a pleasurable experience, but that is still a matter

of debate. In our study, we did not find an effect of early engagement in exercise (at the age of

11) on depressive symptoms and vice versa in individuals carrying more plasticity genotypes

compared to individuals carrying less. However, in order to investigate this plasticity theory

in depth, the effects of even earlier (early childhood) exercise participation on depressive

symptoms should be investigated in individuals carrying these plasticity genotypes. Finally,

the exact functioning of certain polymorphisms is not entirely clear. For example, with re-

spect to the TPH1 gene previous studies have produced mixed results of which allele might

be acting as the plasticity allele. Belsky et al. (2009) assumed that carrying the A allele is

the putative plasticity genotype while other studies have suggested that the A allele might

be playing a protective role against MDD irrespective of positive environmental influences

(Viikki et al., 2010). MAOA is another example of this complexity. The MAOA gene is only

found at the X chromosome, which makes the grouping of females more troublesome than

the grouping for males.

Further research on the exact function of these genes on the neurotransmitter systems and

the effects of exercise on the human biology is needed in order to discover important genetic

and biological markers of the association between PA and depressive symptoms. Epistatic in-

teractions (gene-gene) must be also studied extensively. Although studies on epistasis might

be complicated and difficult to conduct, the results might be informative in relation to indi-

vidual differences behind the reciprocal associations between PA and depressive symptoms.

There are some notable strengths in this study. First, the sample size, which was consider-

ably larger than previous studies mentioned. Second, the longitudinal design (3 measure-

ments) allowed for the detection and investigation of changes between PA and depressive


symptoms over a long period of time (6 years). Finally, the model fit was excellent, model-

ing the reciprocal relationship between PA and depressive symptoms in our data with great

accuracy.

However, there are also limitations. First, most of the analysis relied on self-reports. Self-

reports are not as reliable as more objective measurements of PA, such as, calculation of

Metabolic Equivalent of Task (MET) in minutes by accelerometers. In addition, PA was as-

sessed by a single question, which was slightly different at T1 than at the last two waves,

unlike validated PA questionnaires (e.g., IPAQ) that include a series of detailed questions of

PA involvement. At T2 and T3 separate questions for winter and summer were asked and

more response categories were used compared to T1. Moreover, information on the nature

(leisure-time, voluntary or spontaneous), intensity or duration of PA were not collected. This

type of information may prove useful for a better understanding of the influence of specific

genes on the association between PA and depressive symptoms, since different types, in-

tensities and durations of PA have been shown to be differentially associated with depres-

sive symptoms. Furthermore, due to insufficient power, we merged the group without any

plasticity alleles (n = 22) with the group carrying one to three alleles, probably leading to

some information loss. Finally, it would be optimal to investigate changes in neurotrans-

mitter levels as a result of engaging in PA or increases/decreases in depressive symptoms

before investigating genetic modification, in order to advance the knowledge of how these

effects (PA and depressive symptoms) affect human biology.

CONCLUSION

Although gene-environment and studies on epistasis can produce conflicting results and

seem overly complicated, there is still optimism that these studies can shed light on impor-

tant genetic factors, which might explain individual differences in the associations between

PA and depressive symptoms. This elucidation of genetic differences between individuals is

necessary for designing effective treatments for affective disorders. However, at this time

we did not find support for the notion that plasticity genes may moderate the reciprocal as-

sociation between depressive symptoms and PA.

CHAPTER IV72

REFERENCES

Achenbach, T. M. (1991a). Manual for the child behavior checklist/ 4-18 and 1991 profile. University of Ver-mont, Burlington.

Achenbach, T. M. (1991b). Manual for the youth self report and 1991 profile. University of Vermont, Burlington.

Belsky, J., & Beaver, K. M. (2011). Cumulative-genetic plasticity, parenting and adolescent self-regulation. Journal of Child Psychology and Psychiatry, and Allied Disciplines, 52(5), 619-626. doi:10.1111/j.1469-7610.2010.02327.x

Belsky, J., Jonassaint, C., Pluess, M., Stanton, M., Brummett, B., & Williams, R. (2009). Vulnerability genes or plasticity genes? Molecular Psychiatry, 14(8), 746-754. doi:10.1038/mp. 2009.44

Belsky, J., & Pluess, M. (2009). Beyond diathesis stress: Differential susceptibility to environmental influ-ences. Psychological Bulletin, 135(6), 885-908. doi:10.1037/a0017376

Bosch, N. M., Riese, H., Dietrich, A., Ormel, J., Verhulst, F. C., & Oldehinkel, A. J. (2009). Preadolescents’ somatic and cognitive-affective depressive symptoms are differentially related to cardiac auto-nomic function and cortisol: The TRAILS study. Psychosomatic Medicine, 71(9), 944-950. doi:10.1097/PSY.0b013e3181bc756b

Caspi, A., Hariri, A. R., Holmes, A., Uher, R., & Moffitt, T. E. (2010). Genetic sensitivity to the environment: The case of the serotonin transporter gene and its implications for studying complex diseases and traits. The American Journal of Psychiatry, 167(5), 509-527. doi:10.1176/appi.ajp. 2010.09101452

Caspi, A., Sugden, K., Moffitt, T. E., Taylor, A., Craig, I. W., Harrington, H., & Poulton, R. (2003). Influence of life stress on depression: Moderation by a polymorphism in the 5-HTT gene. Science, 301(5631), 386-389. doi:10.1126/science.1083968


Dishman, R. K., Berthoud, H. R., Booth, F. W., Cotman, C. W., Edgerton, V. R., Fleshner, M. R., & Zigmond, M. J. (2006). Neurobiology of exercise. Obesity, 14(3), 345-356. doi:10.1038/oby.2006.46

Ellis, B. J., Boyce, W. T., Belsky, J., Bakermans-Kranenburg, M. J., & van Ijzendoorn, M. H. (2011). Differential susceptibility to the environment: An evolutionary–neurodevelopmental theory. Development and Psy-chopathology, 23(1), 7-28. doi:10.1017/S0954579410000611

Fox, K. R. (1999). The influence of physical activity on mental well-being. Public Health Nutrition, 2(3A), 411-418.

Goekint, M., Roelands, B., de Pauw, K., Knaepen, K., Bos, I., & Meeusen, R. (2010). Does a period of detrain-ing cause a decrease in serum brain-derived neurotrophic factor? Neuroscience Letters, 486(3), 146-149. doi:10.1016/j.neulet.2010.09.032

Hashimoto, K. (2010). Brain-derived neurotrophic factor as a biomarker for mood disorders: An historical overview and future directions. Psychiatry and Clinical Neurosciences, 64(4), 341-357. doi:10.1111/j.1440-1819.2010.02113.x

Hattori, S., Naoi, M., & Nishino, H. (1994). Striatal dopamine turnover during treadmill running in the rat: Relation to the speed of running. Brain Research Bulletin, 35(1), 41-49.

Hegadoren, K. M., O’Donnell, T., Lanius, R., Coupland, N. J., & Lacaze-Masmonteil, N. (2009). The role of beta-endorphin in the pathophysiology of major depression. Neuropeptides, 43(5), 341-353. doi:10.1016/j.npep. 2009.06.004

Huisman, M., Oldehinkel, A. J., de Winter, A. F., Minderaa, R. B., de Bildt, A., Huizink, A. C., & Ormel, J. (2008). Cohort profile: The Dutch ‘TRacking Adolescents’ Individual Lives’ Survey’; TRAILS. International Jour-nal of Epidemiology, 37(6), 1227-1235. doi:10.1093/ije/dym273


Johnson, K. E., & Taliaferro, L. A. (2011). Relationships between physical activity and depressive symptoms among middle and older adolescents: A review of the research literature. Journal for Specialists in Pedi-atric Nursing: JSPN, 16(4), 235-251. doi:10.1111/j.1744-6155.2011.00301.x

Karg, K., Burmeister, M., Shedden, K., & Sen, S. (2011). The serotonin transporter promoter variant (5-HT-TLPR), stress, and depression meta-analysis revisited: Evidence of genetic moderation. Archives of Gen-eral Psychiatry, 68(5), 444-454. doi:10.1001/archgenpsychiatry.2010.189

Knaepen, K., Goekint, M., Heyman, E. M., & Meeusen, R. (2010). Neuroplasticity - exercise-induced response of peripheral brain-derived neurotrophic factor: A systematic review of experimental studies in human subjects. Sports Medicine, 40(9), 765-801. doi:10.2165/11534530-000000000-00000

Lapin, I. P., & Oxenkrug, G. F. (1969). Intensification of the central serotoninergic processes as a possible determinant of the thymoleptic effect. Lancet, 1(7586), 132-136.

Lemke, M. R., Brecht, H. M., Koester, J., & Reichmann, H. (2006). Effects of the dopamine agonist pramipex-ole on depression, anhedonia and motor functioning in Parkinson’s disease. Journal of the Neurological Sciences, 248(1-2), 266-270. doi:10.1016/j.jns.2006.05.024

Lindwall, M., Larsman, P., & Hagger, M. S. (2011). The reciprocal relationship between physical activity and depression in older European adults: A prospective cross-lagged panel design using SHARE data. Health Psychology, 30(4), 453-462. doi:10.1037/a0023268

Little, T. D. (in press). Longitudinal structural equation modeling: Individual difference panel models. New York: Guildford press.

Mata, J., Thompson, R. J., & Gotlib, I. H. (2010). BDNF genotype moderates the relation between physical activity and depressive symptoms. Health Psychology, 29(2), 130-133. doi:10.1037/a0017261

Meeusen, R., Piacentini, M. F., & De Meirleir, K. (2001). Brain microdialysis in exercise research. Sports Medi-cine, 31(14), 965-983.

Miller, S. A., Dykes, D. D., & Polesky, H. F. (1988). A simple salting out procedure for extracting DNA from hu-man nucleated cells. Nucleic Acids Research, 16(3), 1215.

Munafo, M. R., Durrant, C., Lewis, G., & Flint, J. (2009). Gene X environment interactions at the serotonin transporter locus. Biological Psychiatry, 65(3), 211-219. doi:10.1016/j.biopsych.2008.06.009

Muthén, L. K., & Muthén, B. O. (2007). Mplus user’s guide: Statistical analysis with latent variables (5th ed.). Los Angeles, CA: Muthén & Muthén.

Nederhof, E., Bouma, E. M., Oldehinkel, A. J., & Ormel, J. (2010). Interaction between childhood adversity, brain-derived neurotrophic factor val/met and serotonin transporter promoter polymorphism on depres-sion: The TRAILS study. Biological Psychiatry, 68(2), 209-212. doi:10.1016/j.biopsych.2010.04.006

Penedo, F. J., & Dahn, J. R. (2005). Exercise and well-being: A review of mental and physical health benefits associated with physical activity. Current Opinion in Psychiatry, 18(2), 189-193.

Rethorst, C. D., Landers, D. M., Nagoshi, C. T., & Ross, J. T. (2010). Efficacy of exercise in reducing depressive symptoms across 5-HTTLPR genotypes. Medicine and Science in Sports and Exercise, 42(11), 2141-2147. doi:10.1249/MSS.0b013e3181de7d51

Rethorst, C. D., Landers, D. M., Nagoshi, C. T., & Ross, J. T. (2011). The association of 5-HTTLPR genotype and depressive symptoms is moderated by physical activity. Journal of Psychiatric Research, 45(2), 185-189. doi:10.1016/j.jpsychires.2010.05.007

Risch, N., Herrell, R., Lehner, T., Liang, K. Y., Eaves, L., Hoh, J., & Merikangas, K. R. (2009). Interaction be-tween the serotonin transporter gene (5-HTTLPR), stressful life events, and risk of depression: A me-ta-analysis. JAMA: The Journal of the American Medical Association, 301(23), 2462-2471. doi:10.1001/jama.2009.878

CHAPTER IV74

Rubin, B. D. (1996). Multiple imputation after 18+ years. Journal of the American Statistical Association, 91(434), 473.

Sparling, P. B., Giuffrida, A., Piomelli, D., Rosskopf, L., & Dietrich, A. (2003). Exercise activates the endocan-nabinoid system. Neuroreport, 14(17), 2209-2211. doi:10.1097/01.wnr.0000097048.56589.47

Stavrakakis, N., de Jonge, P., Ormel, J., & Oldehinkel, A. J. (2012). Bidirectional prospective associations between physical activity and depressive symptoms. Τhe TRAILS study. Journal of Adolescent Health, 50(5), 503-8. doi:10.1016/j.jadohealth.2011.09.004

Steptoe, A., Nolte I. M., McCaffery, J. M., & Snieder, H. (2010). Candidate gene and genome-wide association studies in behavioral medicine. In A. Steptoe (Ed.), Handbook of behavioral medicine: Methods and ap-plications (1st ed., ). New York: Springer.

van Lang, N. D., Ferdinand, R. F., Oldehinkel, A. J., Ormel, J., & Verhulst, F. C. (2005). Concurrent validity of the DSM-IV scales affective problems and anxiety problems of the youth self-report. Behaviour Research and Therapy, 43(11), 1485-1494. doi:10.1016/j.brat.2004.11.005

van Praag, H. M., & Korf, J. (1970). L-tryptophan in depression. Lancet, 2(7673), 612.

Verhagen, M., van der Meij, A., van Deurzen, P. A., Janzing, J. G., Arias-Vasquez, A., Buitelaar, J. K., & Franke, B. (2010). Meta-analysis of the BDNF Val66Met polymorphism in major depressive disorder: Effects of gender and ethnicity. Molecular Psychiatry, 15(3), 260-271. doi:10.1038/mp. 2008.109

Viikki, M., Kampman, O., Illi, A., Setala-Soikkeli, E., Anttila, S., Huuhka, M., et al. (2010). TPH1 218A/C poly-morphism is associated with major depressive disorder and its treatment response. Neuroscience Let-ters, 468(1), 80-84. doi:10.1016/j.neulet.2009.10.069

Competitive Sport Participation, Sport Competence and Depressive Symptoms in Adolescent Boys and Girls

“Serious sport has nothing to do with fair play. It is bound up with hatred, jealousy, boastfulness, disregard of all rules and sadistic pleasure in witnessing violence: in other words it is war minus the shooting.”

George Orwell, in The Sporting Spirit, 1945

“Certification from one source or another seems to be the most important thing to people all over the world. A piece of paper from a school that says you’re smart, a pat on the head from your parents that says you’re good or some reinforcement from your peers that makes you think what you’re doing is worthwhile. People are just waiting around to get certified.”

Frank Zappa, in Oui interview, 1979

Stavrakakis N, Roest AM, Verhulst FC, Veenstra R, Ormel J, de Jonge P & Oldehinkel AJ

Under preparation

CHAPTER

CHAPTER V76

ABSTRACT

Purpose: Physical activity is inversely but moderately related to depressive symptoms in

adolescents. This might indicate that not all types of physical activity are associated with

depressive symptoms, or that the association is present only in a subgroup. We set out to

investigate: a) whether participation in sports with a clear competitive element is associ-

ated with current depressive symptoms and symptom changes over time in adolescents; b)

whether these associations are modified by gender and sport competence.

Methods: Data came from the Dutch TRacking Adolescents’ Individual Lives Survey

(TRAILS), a longitudinal population-based study in adolescents (N = 2149; girls = 51%, mean

age = 13.65, SD = 0.53). Depressive symptoms were measured by the Affective Problems

scale of the Youth Self-Report. Based on participants’ reported participation in sports, we

selected sports with a clear competitive element. Sport competence was assessed by peer

nominations.

Results: Participation in competitive sports was significantly associated with lower cur-

rent depressive symptom scores (p = 0.002) but not with positive changes in depressive

symptoms (p = 0.23). After adjustment for sport competence the association between sport

participation and current depressive symptoms was no longer significant (p = 0.41). No three-

way interactions between competitive sports, gender, and sport competence were observed.

Conclusion: Participation in competitive sports was related to lower current depressive

symptoms, which could be explained by sport competence, but not with changes in depres-

sive symptoms. Finally, we found no evidence that associations between sport participation

and depressive symptoms are modified by gender and sport competence.

77COMPETITIVE SPORT PARTICIPATION AND DEPRESSIVE SYMPTOMS

INTRODUCTION

Physical activity (PA) has received significant attention over the last years as being poten-

tially beneficial in alleviating depressive symptoms (Vilhjalmsson & Thorlindsson, 1992;

Sanders et al., 2000; Salmon, 2001; Gore et al., 2001; Dunn et al., 2005; Rimer et al., 2012). Re-

views and meta-analyses of intervention studies investigating the relationship between PA

and depressive symptoms in adults have also shown that PA can reduce depressive symp-

toms, but their conclusions should be taken with caution because of the low methodological

quality of most studies included (Rimer et al., 2012; Cooney et al., 2013; Rosenbaum et al.,

2014). Two meta-analyses of intervention studies in adolescents also showed that PA was

inversely related to depressive symptoms but the effect sizes were small, and again there

was a lack of high methodological quality studies (Larun et al., 2006; Brown et al., 2013).

Because depressive symptoms in adolescence have been shown to predict the development

of a depressive disorder in adulthood (Hankin et al., 1998; Lewinsohn et al., 1999; Fergusson

et al., 2005; Monshouwer et al., 2012), it is pertinent to investigate the relationship between

PA and depressive symptoms at an early age. According to recent reviews, the majority of

observational studies in adolescents, regardless of design, show an inverse association be-

tween PA and depressive symptoms, with weak to moderate effect sizes (Biddle & Asare,

2011; Johnson & Taliaferro, 2011). Two possible reasons that might explain the relatively

weak associations observed in the relationship between PA and depressive symptoms were

investigated in the current study. First, the definition of PA is broad, which could mean that

not all types of PA are associated with depressive symptoms, resulting in an overall small-

sized association. Second, PA might be inversely associated with depressive symptoms only

in specific subgroups of individuals and not in others, also resulting in an overall modest

association. For example, a recent study showed that PA was inversely associated with de-

pressive symptoms in males but not in females (Edman et al., 2014).

PA is defined as any bodily muscle movement that leads to energy expenditure, and includes

a broad range of activities regardless of the environment (e.g., leisure-time vs. occupational)

or intensity of the activity (see: Caspersen et al., 1985). Sport participation is a more specific

definition relying on a competitive element of the activity. Previous research has shown an

inverse association between sport participation and depressive symptoms (Vilhjalmsson &

Thorlindsson, 1992; Sanders et al., 2000; Gore et al., 2001), with effect sizes similar to the

association between PA and depressive symptoms (see for a review: Johnson & Taliaferro,

2011). However, these studies did not define sports according to the competitive element

of the activity, but used an unspecified definition of sports, or categorized sports according

to their social aspect (club, individual, or team) (Vilhjalmsson & Thorlindsson, 1992; Sand-

ers et al., 2000; Gore et al., 2001). Because participation in sports is suggested to have an

CHAPTER V78

evolutionary origin (Lombardo, 2012), we hypothesized that in particular the competitive

element of sports might play a crucial role in the association with depressive symptoms.

According to Lombardo (2012), one of the best ways to achieve social status is through

sports and especially those with a highly competitive element, because sport competitions

in the modern world closely resemble hunting activities and primitive warfare. Competing

in sports gives an opportunity for males to rank each other according to their abilities and

therefore be able to select capable allies or avoid capable enemies. This ranking is assumed

to result in improved or diminished chances of being selected by females (Lombardo, 2012).

This can be seen in both ancient (e.g., Olympians in Classical Greece, see: Poliakoff, 1987

and Golden, 2008) and modern athletes, who tend to gain status fast, especially if they are

champions (Loy, 1972; Chase & Dummer, 1992; Holland & Andre, 1994; Buss, 2007; Chase &

Machida, 2011). This suggests that individuals who are athletically competent will be more

likely to achieve high status than their less competent peers.

The social status that the individual attains or maintains has also been hypothesized to

influence the development of mental disorders. Price’s Social Competition Hypothesis of De-

pression (Price et al., 1994) suggests that depressed mood is a result of inhibitory mecha-

nisms that force the individual to concede defeat and downplay their ambitions and desires

in situations where it is impossible or unlikely for them to succeed. This involuntary defense

mechanism, is hypothesized to help individuals to psychologically deal with ‘what would

otherwise be unacceptably low social status’ (Price et al., 1994; in abstract). Status can be

operationalized in many ways (Linton, 1936; Price, 1972; Price et al., 1994; Anderson et al.,

2001). In this study, we will focus on status in terms of achievements (Price, 1972), which

mainly depends on the ranking of individuals according to their competency levels as per-

ceived by their peers. We focus on peer-perceived sport competence because of the impor-

tance given to sport competence and sport participation in adolescence, especially in boys

(Gill, 1992; Kavussanu & Harnisch, 2000; Sallis et al., 2000).

There are reasons to assume that the above-described mechanisms apply to males in par-

ticular. Boys and men attach more importance to sport participation and are conspicuously

more competitive than girls and women (Price, 1988; Balafoutas et al., 2012; Deaner et al.,

2012), as their achievement status will directly influence their chances of being selected by

women (Fieder & Huber, 2012; Lombardo, 2012; Li et al., 2013). Furthermore, several reports

suggest that achievement status is more strongly associated with well-being in men than in

women (Brendgen et al., 2005; Tiffin et al., 2005; Oldehinkel et al., 2007). Taken together, this

indicates that the way competition for status influences social hierarchy formation and sub-

sequently the development of mental health problems may be different for boys and girls.

The aims of this study were to investigate: 1) whether sport participation is inversely asso-


ciated with current depressive symptoms and depressive symptom change in late adoles-

cence, and 2) whether these associations are modified by peer-perceived sport competence

and gender. We hypothesized that peer-perceived sport competence modifies the relation-

ship between sport participation and depressive symptoms, and that sport-competent boys

will benefit more from performing competitive sports than girls and less competent boys.

METHODS

Design

Data used for this study were part of a large population cohort of adolescents in the Neth-

erlands called TRacking Adolescents’ Individual Lives Survey (TRAILS). The aim of TRAILS

was to delineate the development of mental health problems from childhood to adulthood.

Approval of the study was obtained from the Dutch Committee on Research Involving Hu-

man Subjects. Data for the first measurement wave (T1) were collected in 2001, and for the

second (T2) and third (T3) measurement waves data collection took place between 2003

and 2007, with approximate intervals of 2.5 years between each wave. Informed consent was

collected from the parents at T1, while both parents and participants gave informed consent

at T2 and T3. Extensive descriptions of TRAILS design, data and sample collection can be

found elsewhere (de Winter et al., 2005; Huisman et al., 2008; Ormel et al., 2012). The pres-

ent study was based on data from T2 and T3.

Participants and Procedure

The total sample at T1 included 2230 adolescents (50.8% girls, mean age = 11.1, SD = 0.6).

The response rates at T2 and T3 were 96.4% (N = 2149; 51% girls, mean age = 13.7, SD = 0.5)

and 81.4% (N = 1816; 52.3% girls, mean age = 16.3, SD = 0.7) respectively (Ormel et al., 2012).

Detailed information on sampling procedures can be found elsewhere (de Winter et al., 2005;

Nederhof et al., 2012). During T2 and T3, parents received a questionnaire by mail. Adoles-

cents and their teachers were asked to complete questionnaires at school.

Outcome Variables

Depressive Symptoms

Depressive symptoms were assessed using the Affective Problem Scale of the Youth Self

Report (YSR; Achenbach & Rescorla, 2001). The Affective Problem Scale consists of 13 items,

which correspond to DSM-IV depressive symptoms (van Lang et al., 2005). These items are

scored on a three-point scale (0 = never or not at all true, 1 = sometimes true and 2 = very of-

ten or very true) and include questions on sadness, loss of pleasure, crying, self-harm, suicidal

CHAPTER V80

ideation, feelings of guilt and worthlessness, loss of energy, eating problems, overtiredness, and

sleeping problems (3 items). The internal consistency (Cronbach’s alpha) for the 13 items was

0.77 for T2 and 0.78 for T3. A mean score of all 13 items for T2 and T3 was used in the analyses.

Predictors

Sport Participation

According to the Oxford dictionary a sport is: “an activity involving physical exertion and skill

in which an individual or team competes against another or others for entertainment” (Ox-

ford dictionaries; http://oxforddictionaries.com/definition/english/sport?q=sport). Through-

out this manuscript we will use this definition of sports, because we were interested in the

competitive element of activities. During T2, adolescents were asked to indicate whether

they participate in any sports and what types of sports they participate in (maximum of four

sports). From these responses, we created a binary variable indicating whether at least one

of the activities reported, corresponded to the definition of sports we used. If none of the

activities met the criterion for a sport the participant was coded as a non-sporter. Activities

such as ‘running’, ‘walking’ or ‘dancing’ were not considered as sports per se, because these

activities do not necessarily imply direct competition with other individuals or teams. In con-

trast, activities such as “running in competitions” were considered a sport. For responses

that had an ambiguous competitive element (such as gymnastics and roller-blading), the

frequencies per week (more or less often than two times per week) were used to distinguish

between sporters and non-sporters. In total, 2089 adolescents had valid responses on the

sport participation measure. The 141 (6.3%) individuals that did not respond to the sport

participation and the general PA questions were considered missing. More information on

the sport participation measure can be found in the Appendix, Table F.

Peer Nominations of Sport Competence

At T2, peer nominations were assessed with regard to various characteristics, including

sport competence. These nominations were assessed in classes with at least three regular

TRAILS participants. Peer nominations were assessed in 172 classes of 34 schools in the

first year (72 school classes) and second year (100 school classes) of secondary education,

each including 7 to 30 participating adolescents per class (mean = 18.4, SD = 5.99). In total,

3312 students (1675 boys; 1637 girls) including 1007 regular TRAILS participants filled out

the peer nomination questionnaire (mean age = 13.7, SD = 0.66). For the purpose of this study,

we selected the question on sport competence (‘Who is good at sports?’), and used the per-

centage of nominations as a continuous measure in the analyses. Further details on the peer

nomination measures in TRAILS and the representativeness of the peer nominations sample

have been described elsewhere (Dijkstra et al., 2008).


Teacher Evaluations of Sport Competence

At T2, the teachers were asked to rate the athletic competence of each TRAILS participant

in their class. This question was rated on a five-point scale, with 1 = inadequate, 2 = more

or less adequate, 3 = adequate, 4 = good, and 5 = outstanding. Information on the teacher’s

evaluation of sport competence was collected for 1455 adolescents.


Socioeconomic status (SES) of the family was obtained by averaging five standardized vari-

ables measured at T1, i.e., professional occupation and educational attainment of both par-

ents/guardians, and household income. The internal consistency of the SES scale was 0.84.

The scale captures 61.2% of the variance in the five items. SES was used as a continuous

measure in the analysis. At T1, 16.8% of the TRAILS participants lived in families with a

disposable income annual up to € 13620, which is below the at-risk-of-poverty-threshold.

Further information on each of these variables and the aggregated SES can be found else-

where (Amone-P’Olak et al., 2010).

Analyses

Missing Data

Differences between responders and non-responders at T3, and between the adolescents

that completed the peer evaluations and those that did not, for all predictor variables were

estimated using independent t-tests and chi-square difference tests depending on whether

the variables in question were continuous or dichotomous. Non-responders at T3 were more

likely to be males (x2 = 15.7, df = 1, p = 0.001), participate in sports (x2 = 9.5, df = 1, p = 0.002),

report more depressive symptoms at T2 (t = -2.7, p = 0.01) and have lower SES (t = -10.1,

p = 0.001) than responders. There was no significant difference between responders and

non-responders at T3 regarding the peer evaluations of sport competence (t = -0.6, p = 0.53).

Adolescents who had completed the peer nominations were more likely to have higher SES

(t = -6.3, p = 0.001) and participate in sports (x2 = 8.6, df = 1, p = 0.004) than those who did not,

while they did not differ regarding gender (x2 = 0.7, df = 1, p = 0.42), depressive symptoms at T2

(t = 0.8, p = 0.40) and depressive symptoms at T3 (t = 0.5, p = 0.64).

Correlations

All analyses were performed using SPSS (version 20, SPSS Inc., Chicago, Illinois). The as-

sociations between the predictors, outcome variables and covariates were calculated using

Pearson’s correlations (r). For the sport participation and gender variables, the correlations

can be interpreted as point-biserial correlations because of the dichotomous nature of these

variables (Field, 2009).

CHAPTER V82

Linear Regressions

The hypotheses were tested by two-tailed tests with an alpha-level of 0.05. Because the

residuals of the outcome variable (i.e., depressive symptoms) were positively skewed, we

performed linear regressions with bootstrapping, that is, estimations of confidence intervals

based on repeated sample extractions from the original dataset, which do not rely on a priori

distributional assumptions (for more information see Field, 2009 p. 163; Hayes, 2009). Boot-

strapping of 1000 samples with Bias Corrected and Accelerated confidence intervals (CIs)

was performed for all analyses.

In order to test associations with current depressive symptoms, sport participation was

used as the predictor, and gender and SES as covariates. In order to predict depressive

symptom change, baseline (T2) depressive symptoms were included as an additional co-

variate. In total, 2054 participants were included in the cross-sectional analysis, while 1614

adolescents were included in the depressive symptom change analysis. We repeated these

analyses including sport competence. Because the peer perceptions of sport competence

were collected from a subsample of TRAILS, a total of 960 adolescents were included in the

cross-sectional analysis, while 785 adolescents were included in the depressive symptom

change analysis. In order to test interactions between sport participation, sport competence,

and gender, we created a model including all possible two-way and three-way interactions

among these variables, and removed any non-significant interactions from the model, pro-

viding the model was still hierarchical.

Finally, in order to examine to what extent the results were affected by the measures used,

the analyses were repeated using the teacher evaluations of sport competence and a broad-

er definition of sports. In this broader definition, all reported physical activities (including, for

example, walking in the park, dancing, or running) were categorized as a sport.

RESULTS


From a total of 2089 participants (51% girls) with valid responses in the sport participation mea-

sure, 1177 (56.3%) reported at least one competitive sport, while 912 (43.7%) did not take part in

any activity that fit our definition. Within the sample that participated in the peer nominations,

134 (13.3%) adolescents received no nominations for being good at sports from their peers, 391

(38.8%) received between 1 and 25% nominations, 245 (24.3%) between 26 and 50%, and 237

(23.6%) between 51 and 100%. Table 1 shows the frequencies of all predictors according to sport

participation. Depressive symptoms at T2 (mean age = 13.7) had a mean score of 0.27 (SD = 0.26)

while depressive symptoms at T3 (mean age = 16.3) had a mean score of 0.30 (SD = 0.27).


Correlations

As can be seen from table 2, for boys sport participation was significantly correlated with

depressive symptoms at both T2 and T3, whereas for girls sport participation was only sig-

nificantly correlated with depressive symptoms at T2. Sport competence was significantly

associated with T2 and T3 depressive symptoms and with sport participation for both boys

and girls (see also table A in appendix).

Linear Regressions

In the cross-sectional analysis, sport participation was significantly associated with depres-

sive symptoms at T2 after adjustment for gender and SES (B = -0.05, CIs = -0.07 to -0.02,

p = 0.002), but its effect was no longer significant after inclusion of sport competence (table

3). This model explained around 4% of the variance in the depressive symptoms. Neither

sport participation nor sport competence was significantly associated with change in de-

pressive symptoms between T2 and T3 (table 4). This model explained approximately 30%

of the variance, due to the inclusion of T2 depressive symptoms as a covariate.

The three-way interaction between sport participation, sport competence, and gender did

not predict current depressive symptoms significantly (SES-adjusted B = 0.06, CIs = -0.03 to

0.16, p-value = 0.17), and was therefore removed from the model. In the remaining model there

were no significant two-way interactions (B ranges = -0.04 to -0.00, p-value ranges = 0.31 to

0.95), hence a model with only main effects described the data best. The same was true for

the prediction of change in depressive symptoms between T2 and T3 (B ranges = -0.04 to

0.01, p-value ranges = 0.30 to 0.98).

Repeating the analyses with the teacher evaluations of sport competence (tables B and C

in the appendix) yielded similar results, with only one small difference: the cross-sectional

association between sport participation and depressive symptoms remained marginally sig-

nificant (B = -0.03, CIs = -0.06 to 0.001, p = 0.054) after adjustment for the teacher evaluations

of sport competence. When using a broader, non-competitive, definition of sport participa-

tion, sport participation remained a significant predictor of current depressive symptoms

after adjustment for (peer-perceived) sport competence as well. Apart from this, the results

were very similar to the results obtained when sports was defined as being competitive (see

appendix, tables D and E).

CHAPTER V84

Table 1. Gender, SES, and Peer-Perceived Sport Competence by Sport Participation.

Variable Sport Participation

Yes Percentage % No Percentage %

Gender (N = 2089)

BoysGirls

745 432

63.3 36.7

273 639

29.9 70.1

Indices of SES (N = 2058)

Maternal Education

Elementary 52 4.5 67 7.6

Lower-Track Secondary 332 28.9 293 33.3

Higher-Track Secondary 413 35.9 315 35.8

Senior Vocational 252 21.9 157 17.8

University 100 8.8 49 5.5

Paternal Education

Elementary 49 4.8 46 6.2

Lower-Track Secondary 259 25.1 214 28.8

Higher-Track Secondary 326 31.7 243 32.7

Senior Vocational 232 22.5 151 20.3

University 164 15.9 89 12

Maternal Occupation

Low Level 144 16.7 122 18.6

Intermediate Level 462 53.6 349 53.1

High Level 314 29.7 122 28.3

Paternal Occupation

Low Level 266 27.1 240 33.9

Intermediate Level 307 31.3 213 30

High Level 409 41.6 256 36.1


Variable Sport Participation

Yes Percentage % No Percentage %

Family Income

Low (less than €2,500 per month)

139 12.9 159 19.6

Intermediate (€2,500-5,500 per month)

597 55.6 482 59.4

High (more than € 5,500 per month)

338 31.5 171 21

Aggregate SES (N = 2058)a

Low SES 249 21.5 251 27.9

Intermediate SES 560 48.3 466 51.8

High SES 350 30.2 182 20.2

Percentage of Peer Nominations of Sport Competence (N = 1007)a

No Nominations (0%) 46 7.9 85 21.6

1 to 25% of Nominations 166 28.5 210 53.4




SES = socioeconomic status. a For SES and sports competence, categories were created for descriptive purposes; in the analyses continuous measures were used.

Table 1. continued.

CHAPTER V86

Table 2. Pearson Correlations according to Gender.

Variables 1. Sport Participation

2. Sport Competence

3. Depressive Symptoms T2


5. SES

1. Sport Participationa

— 0.36** -0.07* -0.05 0.15**

2. Sport Competence

0.40** — -0.12** -0.10** 0.12**


-0.12** -0.15** — 0.51** -0.05


-0.12** -0.14** 0.53** — -0.10**

5. SES 0.13** 0.11* -0.02 -0.02 —

The top half shows correlations for girls while the bottom half shows correlations for boys.SES = socioeconomic status, T2 = second measurement wave and T3 = third measurement wave. Two-tailed significance **p<0.01, *p<0.05. a Point-biserial correlation was conducted for sport participation because of the binary nature of this variable.

Table 3. Prediction of Depressive Symptoms at T2 (N = 960).

Variable Adjusted R2 B (SE) p-value CI (95%)

Sport Participation 0.04 -0.02 (0.02) 0.41 -0.06 to 0.02

Gender -0.05 (0.02) 0.01 -0.09 to -0.01

Sport Competence -0.04 (0.01) 0.001 -0.05 to -0.02

SES -0.01 (0.01) 0.41 -0.02 to 0.01

SES = socioeconomic status, CI = Bias Corrected and Accelerated confidence intervals, SE = standard error, T2 = second measurement wave. All continuous/interval predictors were z-transformed prior to analysis.


DISCUSSION

Three findings emerged from the current study. First, participation in sports was associated

with current depressive symptoms but not depressive symptom change. Second, when peer-

perceived sport competence was included in the model, the association between participa-

tion in competitive sports and current depressive symptoms was no longer statistically sig-

nificant. Finally, interactions between sport participation, perceived sport competence, and

gender were associated with neither current depressive symptoms nor depressive symptom

change in adolescence.

Participation in both competitive and more broadly defined sports was associated with

low current depressive symptom levels. This suggests that the association with depres-

sive symptoms cannot be ascribed to the competitive element of some physical activities.

The effect sizes were relatively weak and comparable with the ones observed in a previ-

ous study investigating the relationship between general PA and depressive symptoms in

the same sample (Stavrakakis et al., 2012). Therefore, our hypothesis that focusing on the

competitive element of sports might strengthen the association with depressive symp-

toms was not supported.

The reason why we evaluated this question was that it has been suggested that sports

have an evolutionary origin (Lombardo, 2012), and that the competitive element of activities

plays a role in subsequent hierarchy formations. The positive correlations observed in the

Table 4. Prediction of Depressive Symptoms at T3, adjusting for Depressive Symptoms at T2 (N = 785).

Variable Adjusted R2 B (SE) p-value CI (95%)

Sport Participation 0.30 0.02 (0.02) 0.28 -0.02 to 0.06

Gender -0.08 (0.02) 0.001 -0.11 to -0.05

Sport Competence -0.01 (0.01) 0.24 -0.03 to 0.01

SES -0.02 (0.01) 0.05 -0.03 to -0.00

Depressive Symptoms T2 0.14 (0.01) 0.001 0.12 to 0.16

SES = socioeconomic status, CI = Bias Corrected and Accelerated confidence intervals, SE = standard error, T2 = second measurement wave and T3 = third measurement wave. All continuous/interval predictors were z-transformed prior to analysis.

CHAPTER V88

current study between sport participation and peer-perceived sport competence allude to

this, i.e., peers regarded the adolescents participating in competitive sports as being more

competent than non-sporters. The perceived competence can be considered an indication of

social status, which has been linked to depressive symptoms (Price et al., 1994; Gilbert & Al-

lan, 1998; Oldehinkel et al., 2007). Consistent with these findings, peer-perceived sport com-

petence was cross-sectionally associated with depressive symptoms in the current study,

and explained most of the variance observed in this association. This suggests that being

perceived as not being very competent by peers might increase the likelihood of depressive

symptoms regardless of participation in competitive sports. Interestingly, while the asso-

ciation between participation in competitive sports and depressive symptoms became non-

significant after including peer-perceived sport competence in the model, the effect of more

broadly defined PA remained significant. Hence, the ranking achieved according to peer-per-

ceived sport competence may be a more important mediator of the effect of participation in

competitive sports than of the effect of PA in general, with regard to depressive symptoms.

Not only the measurement of sport participation, but also that of perceived sport compe-

tence affected the results: when sport competence was assessed by teacher reports, it ex-

plained less of the effect of participation in competitive sports on depressive symptoms

than when assessed by peer nominations. These diverging results for peer nominations and

teacher evaluations could be due to sample size differences: peer nominations were col-

lected in only part of the sample. Another explanation is that these two measures cover

different aspects of perceived sport competence, as suggested by the only moderate cor-

relations between the two. Perhaps teachers’ evaluations are a more accurate and objective

measure of adolescents’ athletic competence than perceptions of peers. On the other hand,

adolescents spend a lot of time with each other, so may have a better notion of their peers’

actual competence than teachers. Because we were particularly interested in the hierarchy

formation among adolescents, we consider peer nominations of sport competence as a bet-

ter measure than teacher evaluations within the context of this study. Additionally, we did

not know whether the peer perceptions of sport competence were based on school sports

only or also on observations made outside the school setting. Considering the discrepancy

between the peer perceptions and the teacher evaluations of sport competence observed

in this study, it is possible that the perceptions of the peers and the teachers were partly

based on different situations. It would be interesting for future studies to explore the influ-

ence of their teammates perceptions of sport competence on depressive symptoms, since

they might differ from the perceptions of their schoolmates.

Regardless of its definition, sport participation did not predict change in depressive symp-

toms; apparently its benefits, if any, were not strong enough to affect mood in the longer

term. The lack of prospective associations between sport participation and depressive


symptoms might indicate a reciprocal association between sport participation and depres-

sive symptoms. Indeed, previous studies have shown not only that physically active indi-

viduals will be less likely to suffer from depressive symptoms, but also that individuals

who suffer from depressive symptoms will be less physically active (Jerstad et al., 2010;

Lindwall et al., 2011; Stavrakakis et al., 2012). Therefore, any long-term effects of sport

participation on depressive symptoms might be counteracted by the fact that adolescents

with many depressive symptoms are less likely to participate in sports (reverse causality/

inhibition hypothesis).

The Social Competition Hypothesis (Price et al., 1994) provides an interesting theoretical

framework to predict depressive problems, and we tried to extend this theory to investi-

gate whether hierarchy formations through ritual agonistic behaviors could be a possible

explanation of the relationship between sport participation and depressive symptoms. We

expected that adolescents who participated in competitive sports would report fewer de-

pressive symptoms than non-sporters, and that the beneficial effects of sport participa-

tion would be especially strong for boys who were perceived as highly sport competent.

However, the lack of interactions between sport participation, gender and perceived sport

competence on depressive symptoms did not support this expectation. Perhaps other fac-

tors, such as the results of competitions (winning or losing), self-esteem and self-worth,

or biological mechanisms (e.g., neurotrophin levels) are more important moderators of the

relationship between sport participation and depressive symptoms than the external per-

ceptions of an individuals’ sport competence (Dishman et al., 2006; Buss, 2007; Laske et

al., 2010). A further point that needs to be acknowledged is that our assumption that peer-

perceived sport competence is an indication of social status is an oversimplification of a

complex construct. Social status is multidimensional and can be operationalized in diverse

ways (see: Linton, 1936; Gilbert et al., 1995; Anderson et al., 2001), and modern social hier-

archies do not depend as strictly on ritual agonistic fighting as suggested in Price’s Social

Competition Hypothesis. Social hierarchies can also be formed by how much social atten-

tion is given to an individual by others (for example see: Baumeister & Leary, 1995; Leary

& Baumeister, 2000), or through capacities that do not rely on physical abilities, such as

intelligence (Nettle, 2003). Having a high status in non-sport domains might counteract the

effect of a low peer-perceived sport competence status on depressive symptoms in adoles-

cents and especially in boys, regardless of their sport participation levels. Some support for

this counteracting of other social hierarchy dimensions has been shown in previous studies

(Seroczynski et al., 1997; Oldehinkel et al., 2007).

CHAPTER V90

Limitations

Our study has some limitations. First, some of the sports reported were uninformative with

regard to the competitive element. Activities such as ballroom dancing, horse riding, or gym-

nastics can be competitive in specific circumstances, but are not necessarily so. Because we

did not have information on the actual competitions that adolescents took part in, we used

the frequency of these hard-to-categorize activities, assuming that competitive engagement

requires a frequency of more than twice a week. This limitation could have led to an over-

estimation or underestimation of the proportion of competitive sporters. However, because

using a broader definition of sports in the analysis hardly altered the results, this potential

source of bias is probably not very influential. Second, peer evaluations of sport competence

were not assessed at T3. Therefore, we could not investigate whether adolescents who regu-

larly participated in sports improved their actual athletic competence and subsequently the

perceptions of their peers and how these changes might have affected depressive symptom

changes over time. Moreover, participants with missing data at T3 were more likely to be

males, to participate in sports and to have more depressive symptoms at T2. Missing data

were not accounted for in this study and it is possible that non-response might be directly

caused by the adolescents’ previous depressive symptoms. Therefore, our findings might not

be accurately generalizable to the population. Furthermore, we based our hypotheses on a

combination of two distinct evolutionary hypotheses, namely the evolution of sports and an

evolutionary perspective of the development of depressive symptoms. The measurements

used in TRAILS were not selected from an evolutionary perspective per se, and perhaps were

not optimal to test the full extent of these theories. Therefore the negative findings observed

in the current study might be explained by the less than optimal measures. Finally, we did

not take into account the frequency, duration or intensity of sport participation. It is possible

that only individuals who engage moderately or frequently in sports or PA experience fewer

depressive symptoms (Sanders et al., 2000; Dunn et al., 2005; Goldfield et al., 2011). Although

we cannot exclude this possibility, it has been previously shown that the frequency, duration

and intensity of PA did not predict the onset of a first major depressive episode in adolescents

(Stavrakakis et al., 2013), so a dose-response influence of PA or sport participation on depres-

sive symptom levels seems unlikely.

CONCLUSION

Sport participation was inversely related to current depressive symptoms. This association

might be explained by peer-perceived sport competence. However, we found no evidence

that the interactions between peer-perceived sport competence, sport participation and

gender predicted current depressive symptoms or depressive symptom change over time.


REFERENCES

Achenbach, T. M., & Rescorla, L. A. (2001). Manual for the ASEBA school-age forms and profiles. Burlington, VT: University of Vermont, Research center for children, youth and families.

Amone-P’Olak, K., Ormel, J., Oldehinkel, A. J., Reijneveld, S. A., Verhulst, F. C., & Burger, H. (2010). Socioeco-nomic position predicts specialty mental health service use independent of clinical severity: The TRAILS study. Journal of the American Academy of Child and Adolescent Psychiatry, 49(7), 647-655.

Anderson, C., John, O. P., Keltner, D., & Kring, A. M. (2001). Who attains social status? Effects of personality and physical attractiveness in social groups. Journal of Personality and Social Psychology, 81(1), 116-132.

Balafoutas, L., Kerschbamer, R., & Sutter, M. (2012). Distributional preferences and competitive behavior. Journal of Economic Behavior & Organization, 83(1), 125-135.

Baumeister, R. F., & Leary, M. R. (1995). The need to belong - desire for interpersonal attachments as a fundamental human-motivation. Psychological Bulletin, 117(3), 497-529.


Brendgen, M., Wanner, B., Morin, A. J. S., & Vitaro, F. (2005). Relations with parents and with peers, tempera-ment, and trajectories of depressed mood during early adolescence. Journal of Abnormal Child Psychol-ogy, 33(5), 579-594.

Brown, H. E., Pearson, N., Braithwaite, R. E., Brown, W. J., & Biddle, S. J. H. (2013). Physical activity inter-ventions and depression in children and adolescents: A systematic review and meta-analysis. Sports Medicine, 43(3), 195-206.

Buss, D. M. (2007). Evolutionary psychology: The new science of the mind (3rd ed.). New York: Allyn and Bacon.


Chase, M. A., & Dummer, G. M. (1992). The role of sports as a social-status determinant for children. Re-search Quarterly for Exercise and Sport, 63(4), 418-424.

Chase, M. A., & Machida, M. (2011). The role of sport as a social status determinant for children: Thirty years later. Research Quarterly for Exercise and Sport, 82(4), 731-739.



Deaner, R. O., Geary, D. C., Puts, D. A., Ham, S. A., Kruger, J., Fles, E., et al. (2012). A sex difference in the predisposition for physical competition: Males play sports much more than females even in the contem-porary US. Plos One, 7(11), e49168.

Dijkstra, J. K., Lindenberg, S., & Veenstra, R. (2008). Beyond the class norm: Bullying behavior of popular adolescents and its relation to peer acceptance and rejection. Journal of Abnormal Child Psychology, 36(8), 1289-1299.

Dishman, R. K., Hales, D. P., Pfeiffer, K. A., Felton, G., Saunders, R., Ward, D. S., et al. (2006). Physical self-concept and self-esteem mediate cross-sectional relations of physical activity and sport participation with depression symptoms among adolescent girls. Health Psychology, 25(3), 396-407.

Dunn, A. L., Trivedi, M. H., Kampert, J. B., Clark, C. G., & Chambliss, H. O. (2005). Exercise treatment for de-pression: Efficacy and dose response. American Journal of Preventive Medicine, 28(1), 1-8.

CHAPTER V92

Edman, J. L., Lynch, W. C., & Yates, A. (2014). The impact of exercise performance dissatisfaction and physi-cal exercise on symptoms of depression among college students: A gender comparison. Journal of Psy-chology, 148(1), 23-35.

Fergusson, D. M., Horwood, L. J., Ridder, E. M., & Beautrais, A. L. (2005). Subthreshold depression in adoles-cence and mental health outcomes in adulthood. Archives of General Psychiatry, 62(1), 66-72.

Fieder, M., & Huber, S. (2012). An evolutionary account of status, power, and career in modern societies. Human Nature, 23(2), 191-207.

Field, A. (2009). Discovering statistics using SPSS (3rd ed.). SAGE publications Ltd.

Gilbert, P., & Allan, S. (1998). The role of defeat and entrapment (arrested flight) in depression: An explora-tion of an evolutionary view. Psychological Medicine, 28(3), 585-598.

Gilbert, P., Price, J. S., & Allan, S. (1995). Social-comparison, social attractiveness and evolution - how might they be related. New Ideas in Psychology, 13(2), 149-165.

Gill, D. L. (1992). Gender and sport behaviour. Advances in sport psychology (pp. 143-160). Champaign, IL: Human Kinetics.

Golden, M. (2008). Greek sport and social status. Austin TX: University of Texas Press.


Gore, S., Farrell, F., & Gordon, J. (2001). Sports involvement as protection against depressed mood. Journal of Research on Adolescence, 11(1), 119-130.


Hayes, A. F. (2009). Beyond Baron and Kenny: Statistical mediation analysis in the new millennium. Com-munication Monographs, 76(4), 408-420.

Holland, A., & Andre, T. (1994). Athletic participation and the social-status of adolescent males and females. Youth & Society, 25(3), 388-407.




Kavussanu, M., & Harnisch, D. L. (2000). Self-esteem in children: Do goal orientations matter? The British Journal of Educational Psychology, 70(2), 229-242.

Larun, L., Nordheim, L. V., Ekeland, E., Hagen, K. B., & Heian, F. (2006). Exercise in prevention and treatment of anxiety and depression among children and young people. The Cochrane Database of Systematic Reviews, (3), CD004691-CD004691.

Laske, C., Banschbach, S., Stransky, E., Bosch, S., Straten, G., Machann, J., et al. (2010). Exercise-induced normalization of decreased BDNF serum concentration in elderly women with remitted major depression. The International Journal of Neuropsychopharmacology, 13(5), 595-602.

Leary, M. R., & Baumeister, R. F. (2000). The nature and function of self-esteem: Sociometer theory (pp. 1-62). San Diego, CA, US: Academic Press.


Lewinsohn, P. M., Rohde, P., Klein, D. N., & Seeley, J. R. (1999). Natural course of adolescent major depressive disorder: I. Continuity into young adulthood. Journal of the American Academy of Child and Adolescent Psychiatry, 38(1), 56-63.

Li, N. P., Yong, J. C., Tov, W., Sng, O., Fletcher, G. J., Valentine, K. A., et al. (2013). Mate preferences do predict attraction and choices in the early stages of mate selection. Journal of Personality and Social Psychol-ogy, 105(5), 757-776.

Lindwall, M., Larsman, P., & Hagger, M. S. (2011). The reciprocal relationship between physical activity and depression in older European adults: A prospective cross-lagged panel design using SHARE data. Health Psychology, 30(4), 453-462.

Linton, R. (1936). The study of man: An introduction. Appleton Century Crofts, Inc.

Lombardo, M. P. (2012). On the evolution of sport. Evolutionary Psychology, 10(1), 1-28.

Loy, J. W. (1972). Social origins and occupational mobility patterns of a selected sample of American ath-letes. International Review for the Sociology of Sport, (7), 5-25.

Monshouwer, K., Smit, F., Ruiter, M., Ormel, J., Verhulst, F. C., Vollebergh, W., et al. (2012). Identifying target groups for the prevention of depression in early adolescence: The TRAILS study. Journal of Affective Disorders, 138(3), 287-294.


Nettle D. (2003). Intelligence and class mobility in the British population. British Journal of Psychology, 94(4), 551-61.

Oldehinkel, A. J., Rosmalen, J. G. M., Veenstra, R., Dijkstra, J. K., & Ormel, J. (2007). Being admired or being liked: Classroom social status and depressive problems in early adolescent girls and boys. Journal of Abnormal Child Psychology, 35(3), 417-427.


Poliakoff, M. B. (1987). Combat sports in the ancient world: Competition, violence and culture. New Haven, CT: Yale University Press.

Price, J. S. (1972). Genetic and phylogenetic aspects of mood variations. International Journal of Mental Health, (1), 124-144.

Price, J. S. (1988). Alternative channels for negotiating asymmetry in social relationships. In M. R. A. Chance (Ed.), Social fabrics of the mind (pp. 157-195). Erlbaum.

Price, J. S., Sloman, L., Gardner, R., Gilbert, P., & Rohde, P. (1994). The social competition hypothesis of de-pression. British Journal of Psychiatry, 164, 309-315.

Rimer, J., Dwan, K., Lawlor, D. A., Greig, C. A., McMurdo, M., Morley, W., et al. (2012). Exercise for depression. Cochrane Database of Systematic Reviews, (7), CD004366.

Rosenbaum, S., Tiedemann, A., Sherrington, C., Curtis, J., & Ward, P. (2014). Physical activity interventions for people with mental illness: A systematic review and meta-analysis. Journal of Clinical Psychiatry, 31, 1555-2101.

Sallis, J. F., Prochaska, J. J., & Taylor, W. C. (2000). A review of correlates of physical activity of children and adolescents. Medicine and Science in Sports and Exercise, 32(5), 963-975.


Sanders, C. E., Field, T. M., Diego, M., & Kaplan, M. (2000). Moderate involvement in sports is related to lower depression levels among adolescents. Adolescence, 35(140), 793-797.

CHAPTER V94

Seroczynski, A. D., Cole, D. A., & Maxwell, S. E. (1997). Cumulative and compensatory effects of competence and incompetence on depressive symptoms in children. Journal of Abnormal Psychology, 106(4), 586-597.

Stavrakakis, N., de Jonge, P., Ormel, J., & Oldehinkel, A. J. (2012). Bidirectional prospective associations be-tween physical activity and depressive symptoms. The TRAILS study. The Journal of Adolescent Health, 50(5), 503-508.

Stavrakakis, N., Roest, A. M., Verhulst, F. C., Ormel, J., de Jonge, P., & Oldehinkel, A. J. (2013). Physical activ-ity and onset of depression in adolescents: A prospective study in the general population cohort TRAILS. Journal of Psychiatric Research, 47(10), 1304-1308.

Tiffin, P. A., Pearce, M. S., & Parker, L. (2005). Social mobility over the lifecourse and self reported mental health at age 50: Prospective cohort study. Journal of Epidemiology and Community Health, 59(10), 870-872.

van Lang, N. D., Ferdinand, R. F., Oldehinkel, A. J., Ormel, J., & Verhulst, F. C. (2005). Concurrent validity of the DSM-IV scales affective problems and anxiety problems of the youth self-report. Behaviour Research and Therapy, 43(11), 1485-1494.

Vilhjalmsson, R., & Thorlindsson, T. (1992). The integrative and physiological-effects of sport participation – a study of adolescents. Sociological Quarterly, 33(4), 637-647.


SUPPLEMENTARY MATERIAL

Table A. Correlations in the Total Sample.

Variable 1. 2. 3. 4. 5. 6. 7. 8.

1. Competitive Sport —

2. Broader Definition of Sports

0.54** —

3. Peer Nomination of Sport Competence

0.42** 0.26** —

4. Teacher Evaluations of Sport Competence

0.22** 0.22** 0.27** —

5. Gender 0.33** 0.03 0.31** 0.01 —

6. Depresive Symptoms T2

-0.14** -0.10** -0.17** -0.10** -0.19** —


-0.14** -0.08** -0.18** -0.14** -0.24** 0.54** —

8. SES 0.12** 0.18** 0.09** 0.11** -0.03 -0.04 -0.06* —

SES = socioeconomic status, T2 = second measurement wave and T3 = third measurement wave. Shown here are Pearson correlations between the variables. Point-biserial correlations were conducted for sport participation (both competitive and broad definitions) and gender because of the binary nature of these variables. Because the teacher’s evaluations of sport competence are in an ordinal scale, Spearman Rho correlations were calculated. The correlations between the teacher evaluations of sport competence and dichotomous variables (both sport participation and gender) were converted to rank-biserial correlations using the following formula: rb = (2/n)*(Y1-Y0) where Y1 and Y0 are the mean scores for the group with 1 and 0 respectively (Kraemer, 1982).

REFERENCES

Kraemer, H. C. (1982). Biserial correlation. Encyclopaedia of Statistical Sciences, 1, 276-279.

CHAPTER V96

Table B. Prediction of Depressive Symptoms at T2 by Competitive Sport Participation, Teacher Evaluations of Sport Competence and Gender (N = 1396).

Step Variable Adjusted R2 B (SE) p-value CI (95%)

1 Sport Participation 0.04 -0.03 (0.02) 0.05 -0.06 to 0.001

Gender -0.07 (0.02) 0.001 -0.10 to -0.04


SES -0.004 (0.01) 0.63 -0.02 to 0.01

2 Sport Participation* Gender* Sport Competence

0.04 -0.01 (0.03) 0.81 -0.06 to 0.05

SES = socioeconomic status, CI = Bias Corrected and Accelerated confidence intervals, SE = standard error, T2 = second measurement wave.All continuous/interval predictors were z-transformed prior to analysis. Step 1: shows the main effects of the predictors without interactions (adjusted for SES) with current depressive symptoms. Step 2: shows the three-way interaction of sport participation, teacher evaluations of sport competence and gender (adjusted for SES) with current depressive symptoms. In this model the three-way interaction, all two-way interactions (not shown here) and main effects of the predictors (not shown here) were included.


Table C. Prediction of Depressive Symptoms at T3 by Competitive Sport Participation, Teacher Evaluations of Sport Competence and Gender, adjusting for Depressive Symptoms at T2 (N = 1134).



Gender -0.09 (0.02) 0.001 -0.12 to -0.06


SES -0.02 (0.01) 0.02 -0.03 to -0.001



0.33 0.02 (0.04) 0.65 -0.05 to 0.08

SES = socioeconomic status, CI = Bias Corrected and Accelerated confidence intervals, SE = standard error, T2 = second measurement wave and T3 = third measurement wave.All continuous/interval predictors were z-transformed prior to analysis. Step 1: shows the main effects of the predictors without interactions (adjusted for SES and baseline depressive symptoms) with depressive symptom change. Step 2: shows the three-way interaction of competitive sport participation, teacher evaluations of sport competence and gender (adjusted for SES and baseline depressive symptoms) with depressive symptom change. The three-way interaction, all two-way interactions (not shown here) and main effects of the predictors (not shown here) were included in this model.

CHAPTER V98

Table D. Prediction of Depressive Symptoms at T2 by Sport Participation according to a Broader Definition, Peer Nominations of Sport Competence and Gender (N = 960).


1 Sport Participation 0.05 -0.06 (0.03) 0.01 -0.11 to -0.02


Gender -0.06 (0.02) 0.003 -0.09 to -0.01

SES -0.001 (0.01) 0.68 -0.02 to 0.01


0.05 0.11 (0.07) 0.13 -0.02 to 0.24

SES = socioeconomic status, CI = Bias Corrected and Accelerated confidence intervals, SE = standard error, T2 = second measurement wave.All continuous/interval predictors were z-transformed prior to analysis. Step 1: shows the main effects of sport participation, sport competence and gender (adjusted for SES) with current depressive symptoms. Step 2: shows the three-way interaction of sport participation, peer nominations of sport competence and gender (unadjusted) with current depressive symptoms. In this model the three-way interaction, all two-way interactions (not shown here) and main effects of the predictors (not shown here) were included.


Table E. Prediction of Depressive Symptoms at T3 by Sport Participation according to a Broader Definition, Peer Nominations of Sport Competence and Gender, adjusting for Depressive Symptoms at T2 (N = 785).



Sport Competence -0.01 (0.01) 0.52 -0.02 to 0.01

Gender -0.08 (0.02) 0.001 -.011 to -0.05

SES -0.01 (0.01) 0.12 -0.03 to 0.01



0.30 0.01 (0.06) 0.89 -0.11 to 0.13

SES = socioeconomic status, CI = Bias Corrected and Accelerated confidence intervals, SE = standard error, T2 = second measurement wave and T3 = third measurement wave.All continuous/interval predictors were z-transformed prior to analysis. Step 1: shows the main effects of sport participation (adjusted for baseline depressive symptoms and SES) with depressive symptom change. Step 2: shows the three-way interaction of sport participation, peer nominations of sport competence and gender (adjusted for baseline depressive symptoms at T2 and SES) with depressive symptom change. The three-way interaction, all two-way interactions (not shown here) and main effects of the predictors (not shown here) were included in this model.

Table F. All Sports (sport1, sport2, sport3, sport4) classified according to decision on Competitiveness, Non-Competitiveness and Uncertain Competitiveness.

Decision # of Sport Activities

Frequencies of Participants for each of Four Sports

Sport1 Sport2 Sport3 Sport4

Sum Competitive 48 1008 332 117 74

Sum Non-Competitive 17 647 541 280 135

Sum Ambiguous 65 59 27 15 9

Total 130 1714 900 412 218

Relative-Age Effects in Adolescents: The TRAILS Study

“I can honestly say that insecurity was something formerly unknown to me. I was always the best of my grade. The tallest, the fastest – I thought I was Superman. It turned out this was mainly because I was born in January, thus older than my peers”.

Gert Verhulst, in De Volkskrant interview, 2013

Jeronimus BF, Stavrakakis N, Veenstra R & Oldehinkel AJ

Submitted

CHAPTER

CHAPTER VI102

ABSTRACT

Purpose: This study examined intra-classroom age position (or relative-age) effects on ado-

lescents’ school progress and performance (as rated by teachers), physical development,

temperament, and depressive symptoms, all adjusted for age at the time of measurement.

Methods: Data were derived from three waves of Tracking Adolescents’ Individuals Lives

Survey (TRAILS) of 2230 Dutch adolescents (51% girls, baseline mean age = 11.1, SD = 0.6).

Results: Relative-age predicted school progress (grade retention ORs = 0.83, skipped grade

OR = 1.47, both p<0.001), but substantial effects in adolescents with a normative school tra-

jectory were absent, in contrast to most literature.

Conclusion: Adolescents who had repeated a grade showed inverse relative-age effects on

physical development and school performance, as well as on depressive symptoms, favoring

the relative young.

103RELATIVE-AGE EFFECTS IN ADOLESCENTS

INTRODUCTION

Relative-Age

In most countries, children at school are assorted in same-age groups based on the month

and year of birth (Gledhill, Ford, & Goodman, 2002; Verachtert, De Fraine, Onghena, & Ghes-

quiere, 2010). Consequently, within a single classroom children may differ in age by up to 11

months. Relatively older children have a slightly more developed physique and mind than

their younger classmates (Martin, Foels, Clanton, & Moon, 2004; Verachtert et al., 2010).

These physical and psychological advantages may become catalyzed into different de-

velopmental trajectories through favorable peer-contrast effects (Bedard & Dhuey, 2006;

Gladwell, 2008). We define developmental differences due to the age position driven peer

contrast effects as relative-age effects.

Taller and more mature children tend to have more prestige (Harris, 2009; Henrich & Gil-

White, 2001), which in turn affects friendship formation (Dijkstra, Cillessen, & Borch, 2013;

Newcomb, Bukowski, & Pattee, 1993), and enhances learning opportunities (Carruthers,

Laurence, & Stich, 2006). Relative-age has been associated with higher intelligence (Lawlor,

Clark, Ronalds, & Leon, 2006), school success (Cobley, McKenna, Baker, & Wattie, 2009),

identity formation (Allen, 2008), peer-perceived competence and leadership (Dhuey & Lip-

scomb, 2008), success in sports (Musch & Grondin, 2001), and positive self-perception and

self-esteem (Fenzel, 1992; Thompson, Barnsley, & Battle, 2004). Children’s internal working

models are based on self- and other representations (“looking glass self”), which emerge

and crystallize relatively early in development (Fraley, 2002; Harris, 1995): five-year olds

have already stable and clearly established classroom hierarchies (Boyce et al., 2012).

Self-Fulfilling Prophecy

Children’s relative-age position can become a self-fulfilling prophecy (“learning by being”), cata-

lyzed by reciprocal feedback loops between the developing phenotypes and their environments

(Allen, 2008; Pierson, Addona, & Yates, 2014), analogous to the corresponsive principle (Caspi &

Shiner, 2006) and Dick-Flynn model (Beam & Turkheimer, 2013; Flynn, 2009). Part of this effect

may be driven by external adult evaluations, based on unfavorable intra-cohort contrast effects

(Verachtert et al., 2010). Relatively old children are granted special opportunities for success,

such as extra coaching in sports (Gladwell, 2008; Musch & Grondin, 2001; Persico, Postlewaite,

& Silverman, 2004), whereas relatively young children meet lower expectations and have an

increased risk to repeat a grade (Cobley et al., 2009; Doornbos, 1971; Martin et al., 2004).

Though classroom hierarchies are established in childhood, their consequences may be

particularly salient in adolescence. Adolescents become increasingly able to influence their

CHAPTER VI104

environment (while parental socialization wanes), and select a rapidly expanding peer net-

work and a first romantic partner (Collins, Welsh, & Furman, 2009; Cyranowski, Frank,

Young, & Shear, 2000). Furthermore, earlier work associated being relatively young with

victimization (Muhlenweg, 2010), psychiatric problems (Goodman, Gledhill, & Ford, 2003;

Morrow et al., 2012), and suicide before age 20 (Thompson, Barnsley, & Dyck, 1999). Be-

cause unfavorable relative-age contrast effects modulate children’s self-perception and

self-esteem (Thompson et al., 2004), which are known risk factors for affective disor-

ders (Abela, Fishman, Cohen, & Young, 2012), being relatively young may increase risk

of depression, which has a high incidence in adolescence (Hankin et al., 1998; Oldehinkel,

Wittchen, & Schuster, 1999). Because low self-perception and self-esteem can have a per-

sistent impact on affect, the effect of relative-age might also be discernible in tempera-

mental negative affect.

Finally, relatively older children tend to play more sports – partly driven by selection effects

(Gladwell, 2008; Pierson et al., 2014), which may explain observed relative-age effects on

multiple indices of physical growth in adolescence (Sandercock et al., 2013). Timing of pu-

berty onset seems also influenced by environmental factors (up to 12% variance), including

experiences unshared by twins (Eaves et al., 2004) and neighborhood characteristics (Dear-

dorff et al., 2012). Hence, small relative-age effects on body mass and rate of maturation

seem speculative, but not impossible.

Ability Streaming

Though relative-age effects have been observed in multiple countries (Bedard & Dhuey,

2006), most literature is based on samples from the United States of America (USA) or United

Kingdom (UK) (Musch & Grondin, 2001; Verachtert et al., 2010). One of the challenges in iso-

lating relative-age effects is that their manifestation is contingent on mechanisms to group

children in classes, which differ over time and place (Dalton, 2012; Cascio and Schanzenbach,

2007). For example, the Dutch cohort under study was allocated over classes based upon

an annual birthdate cutoff (pre/post October), and all children in each grade attended all

courses together. In other systems children attend courses (e.g., language, math, or sports)

subdivided on base of ability, called ‘setting’ or ‘banding’, which may alleviate relative-age

effects (Pierson et al., 2014).

The influence of apparently innocuous institutional differences becomes manifest in interna-

tional comparisons, e.g., in the 2006 Program for International Student Assessment (PISA)

up to half of the Dutch (or Belgian, Austrian, or Czech) 15-year olds were in a different grade

than expected in a normative trajectory, compared to 12% in the USA, 1% in the UK, and

none in Japan or Finland (Dalton, 2012). Relative-age effects may thus manifest themselves

via grade progression, that is, the possibility to allocate children to a higher or lower grade


than the normative one, based on their abilities (Doornbos, 1971; Dalton, 2012). We expect

that especially the (less qualified) relative young are retained, whereas the (highly qualified)

relative old are accelerated, which we call ability streaming.

The Present Study

A demonstration of persistent relative-age effects in adolescence, bestowed upon children

by an adult-imposed structuring of their worlds, could lead to renewed awareness and pre-

vention strategies of teachers, parents, and psychiatrists. We therefore aimed to quantify

associations between relative-age position and multiple outcome domains in early and

middle adolescence. Relative-age effects were defined as the effects of intra-cohort age

position adjusted for the actual age at the time of measurement. Only after adjustment for

the additional developmental time granted to the relatively old (a methodological artifact)

the alleged accumulating benefits and disadvantages driven by adolescents’ relative-age

position can be observed.

We hypothesized unfavorable outcomes, in multiple domains, for the relatively young com-

pared to the relatively old adolescents. More specifically, we expected to replicate relative-

age effects on school progress (H1), and tested whether relative-age predicted whether

adolescents had repeated or skipped a grade. For both the adolescents with a normative

progress and the group who repeated a grade we tested whether relative-age predicted

weight (H2a), pubertal status (H2b), school performance (H3), sport competence (H4), and

peer status in terms of peer rejection (H5a) and popularity (H5b). Innovatively, we tested

whether relative-age effects predicted depressive symptoms (H6a) and temperamental fear

and frustration (negative affect, H7a); as well as change in depressive symptoms (H6b) and

fear and frustration (H7b) between age 11 and 16. Because earlier work suggested associa-

tions between maternal socioeconomic status (SES) and month of birth (Bobak & Gjonca,

2001; Buckles & Hungerman, 2013; Martin, 2013), we ran all analyses without and with ad-

justment for family SES.

METHODS

Design and Participants

Data were collected as part of the TRacking Adolescents' Individual Lives Survey (TRAILS),

a large ongoing prospective cohort of Dutch adolescents followed to study the psychologi-

cal, social and physical development of children towards adulthood (Ormel et al., 2012). The

core aim was to unravel developmental pathways to psychological (ill) health. The study

was approved by the Dutch Central Committee on Research Involving Human Subjects. The

CHAPTER VI106

first measurement wave (T1) started in 2001, whereas data collection for the remaining

waves (T2 to T3) took place at intervals of approximately 2.5 years. Informed consent was

collected from the parents at T1, whereas for T2 and T3 informed consent was obtained from

both parents and adolescents. The TRAILS design, sample selection, and data collection are

described extensively elsewhere (de Winter et al., 2005; Huisman et al., 2008; Ormel et al.,

2012). Briefly, participants were selected from five municipalities in the North of the Nether-

lands. From a total of 2935 children, 2230 agreed to take part at T1 (response rate 76%; 51%

girls, mean age 11.1, SD = 0.6). The response rates were 96% at T2 (N = 2149; 51% girls, mean

age 13.7, SD = 0.5) and 81% at T3 (N = 1816; 52% girls, mean age 16.3, SD = 0.7). Non-response

associated slightly with low socioeconomic background, male gender, low IQ and school per-

formance, non-western ethnicity, and externalizing problems, but not with other emotional

and behavioral problems (de Winter et al., 2005). At T1, the parents or guardians were inter-

viewed at their homes, and handed in a previously sent questionnaire at that occasion. At T2

and T3, the questionnaire was sent to the parents or guardians by mail. At all three waves,

the adolescents and their teachers completed the questionnaires at school.

Measures

Relative-Age

As outlined, when the TRAILS children entered school, children born between the first of Oc-

tober and the 30th of September of the next year were allocated in the same age group in the

Netherlands. Consequently, in a normative situation, children born in September were the

youngest in a given grade (month 1), while children born in October were the oldest (month

12). This relative-age measure (1-12) was used as a continuous measure in our analysis.

School Progress

Information on school progress was collected from the schools before sample selection.

Four categories were distinguished: (1) normal progression, (2) children who repeated a

grade, (3) children who skipped a grade, and (4) children in special education. The groups of

adolescents who had repeated one or two grades were merged because only nine adoles-

cents had repeated a grade twice.

Physical Development

At T2, the length in centimeters (cm) and the weight in kilograms (kg, without shoes and

heavy clothing) were measured. Body Μass Ιndex (BMI) was calculated by dividing the

weight (kg) by the square of the height (m2).


Pubertal Status

At T2, pubertal status was measured with the Pubertal Development Scale (PDS), previously

shown to be reliable, with a Chronbach’s alpha of 0.77 for boys and 0.81 for girls (Shirtcliff,

Dahl, & Pollak, 2009). The PDS assesses development on five (Tanner) characteristics, includ-

ing growth spurt in height, skin changes, body hair in both boys and girls, breast development

and menarche in girls, and voice change and facial hair growth in boys (Janssens et al., 2011).

Each item was rated on a four-point scale (0 = not yet started, 1 = just started, 2 = going on for

a while, 3 = passed that). In our analyses we used the mean of the four item scores.

School Performance

At T2, teachers provided school ratings on history, geography, math, and natural sciences.

Adolescent’s school performance was operationalized as the composite of these marks

rated on a five point scale, ranging from 1 = inadequate to 5 = outstanding. Cronbach’s alpha

was 0.86.

Sport Competence

At T2, teachers were asked to rate the sport competence of each adolescent on a five-point

scale (1 = inadequate, 2 = hardly adequate, 3 = adequate, 4 = good, and 5 = outstanding).

Peer Status (being popular or rejected)

At T2, social status was assessed using a sociometric nomination procedure in classrooms

with at least three TRAILS respondents (Oldehinkel, Rosmalen, Veenstra, Dijkstra, & Ormel,

2007). Adolescents could nominate an unlimited number of classmates on a total of 18 ques-

tions, covering a wide range of issues and behaviors. For the purpose of this study, we used

the proportions of the peer nominations for rejection (“being disliked”) and popularity (“being

someone others want to be associated with”), and contrasted these nominations against

the adolescents without this specific peer nomination (i.e., rejected/popular status adoles-

cents vs. all adolescents without this peer nomination).

Depressive Symptoms

Depressive symptoms were assessed at T1 and T3 with the Affective Problem scales of

the Youth Self Report (YSR; Achenbach, 1991b) and parent-reported Child Behavior Checklist

(CBCL; Achenbach, 1991a), which cover depressive symptoms according to DSM-IV criteria

with 13 items, including information on sadness, loss of pleasure, crying, self-harm, suicidal

ideation, feelings of worthlessness, guilt, loss of energy, overtiredness, eating problems and

sleeping problems (Achenbach, Dumenci, & Rescorla, 2003; Ferdinand, 2008). Both scales are

rated on a three-point scale ranging from 0 = never or not at all true to 2 = very often or very

CHAPTER VI108

true. Cronbach’s alpha for the YSR was 0.72 at T1 and 0.78 at T3 and for the CBCL 0.68 at T1

and 0.76 at T3. In the analyses we used the combined mean scores of the CBCL and YSR scales.

Temperamental Fear and Frustration

Temperamental negative affectivity was assessed at T1 and T3 with the Dutch parent ver-

sion (Hartman, 2000) of the revised Early Adolescent Temperament Questionnaire (EATQ-R;

Putnam, Ellis, & Rothbart, 2001), which is based on the temperamental model by Rothbart et

al. (2011; 2000). Earlier work in TRAILS showed the EATQ factor structure of the parent ver-

sion to be superior to the child version (Oldehinkel, Hartman, de Winter, Veenstra, & Ormel,

2004). Fear and frustration were measured with five questions rated on a five-point scale

(ranging from 1 = almost never to 5 = almost always true). Cronbach’s alpha was 0.63 (T1) and

0.66 (T3) for Fear and 0.74 (T1) and 0.75 (T3) for Frustration.


SES of the family of origin was the composite of five standardized variables measured at

T1, i.e., professional occupation and educational attainment of both parents/guardians, and

household income.

Plan of Analyses

Data cleaning, calculation of descriptives, and all analyses were performed in SPSS (version

20, SPSS Inc., Chicago, Illinois). To test whether relatively young or relatively old adolescents

were more likely to repeat or skip a grade (H1), we performed bootstrapped multinomial

regressions with school progress as outcome (1 = normal progression, 2 = repeated grade,

3 = skipped grade, and 4 = special education). Normal school progress was used as reference

category, and relative-age effects on school progress were expressed in odds ratios (ORs).

Hypotheses H2 to H7 were tested with partial Pearson’s correlations (rp) adjusted for age at

testing. In addition, we performed a series of linear regression analyses in which relative-age

predicted, respectively, length and weight (H2a), pubertal status (H2b), school performance

(H3), sport competence (H4), depressive symptoms (H6a), fear and frustration (H7a), and

change in depressive symptoms (H6b) or temperament (H7b), adjusted for age at testing.

To obtain change scores for depressive symptoms and temperamental fear and frustration

we subtracted T1 from T3 scores. To test relative-age effects on being rejected (H5a) and

popular (H5b), we applied binary logistic regression analyses adjusted for age at testing.

All hypotheses were tested in adolescents with a normal school progress and in those

who had repeated a grade (with the exception of H5 because of insufficient peer nomination

data). We lacked power to test effects in the group that skipped a grade (see appendix for

calculations). To ensure the robustness of our results, and because height, weight and BMI


were non-normally distributed, we bootstrapped all regression analyses (k = 10000 with bias

corrected confidence intervals), see (Li, Wong, Lamoureux, & Wong, 2012). We calculated

the power for our regression analyses, and to enable comparison with other literature, we

converted some results to Cohen’s d (standardized effect sizes), based on formulas derived

from Borenstein et al. (2009), and Peterson and Brown (2005). To reduce family-wise alpha

inflation, we only interpreted correlations that were significant at p<0.01.

Finally, we performed two robustness checks. First, we repeated all analyses adjusted for

family SES. Second, to circumvent the strong positive association between relative-age

and biological age, we applied an alternative approach to adjust for relative-age, and com-

bined the relative old children from grade 7 (month 7-12) and relative young from grade 8

(month 1-6). In this control sample of 794 children with a normative school progress rel-

ative-age showed an inverse association with biological age, and was used to repeat our

analyses (H2-7).

RESULTS


Descriptive statistics of all variables used in this study are reported in table 1. The correlations

shown in table 2 indicate that relative old adolescents were still somewhat larger, heavier,

and in a more advanced pubertal development stage than the relative young adolescents.

School Progress

Relative-age effects on school progress are reported in table 3. For each additional month,

relative-age was associated with 17% lower odds of grade repetition and a 47% increased

odds of skipping a grade, see also figure 1. The relatively young quartile (July to Sept) was

almost four times more likely to repeat a grade (29.6% vs. 8.2%) and over twenty times less

likely to skip a grade (0.3% vs. 7%) than the relatively old quartile (Oct to Dec); all details

can be found in the appendix (table A1). Compared to adolescents with a normative school

progress adolescents who repeated a grade were on average 8 weeks younger and those

who skipped a grade were on average 14 weeks older (75% of the latter were relatively old).

As shown in table 3, no association was found for special education (d≈0.03; 95% CI = -0.06 to

0.01). Notably, after adjustment for SES, the effects became slightly stronger, and the effect

on special education significant (see appendix, table A3).

Adolescents with a Normative School Progress (75.4%, n = 1681)

Recall that all results in the following section were adjusted for age at testing. Partial cor-

relations and linear regression models (table 4) showed that relative-age effects predicted

CHAPTER VI110

temperamental frustration (d≈0.22); but this effect disappeared after adjustment for SES (see

appendix, table A2). We further observed that relative-age predicted social rejection (OR = 1.12,

p<0.001), but was unrelated to popularity (OR = 0.99, see appendix, table A4). This rather small

relative-age effect on rejection followed a u-shape, favoring the middle quartiles (9.3%, 6.1%,

6.5%, 9.8%, respectively; all details are given in the appendix, table A5). All other associations

were absent. Because relative-age correlated ( r = 0.53) with biological age (see table 2), we

calculated a control relative-age sample (see method section), which correlated ( r = -0.28)

with biological age (see table 2), but led to similar results (see table 4). In sum, there were no

substantial relative-age effects in the adolescents with a normative school progress.

Adolescents who had Repeated a Grade (16.9%, n = 377)

Though adolescents who had repeated a grade were inherently relative old compared to

their new peers (with a normative school progress), relative-age effects might still play a

role for adolescents who repeated a grade. Partial correlations, presented in table 4, showed

slightly lower intellectual ability and more depressive symptoms for the relatively old ado-

lescents who repeated a grade; a reversed relative-age effect. As shown in table 4, linear re-

gression models showed that relative older adolescents were heavier (d≈0.39), had a higher

BMI (d≈0.42), lower intellectual ability (d≈0.35), and reported more depressive symptoms

(d≈0.50). These results persisted after adjustment for SES (see appendix, table A2).

Perc

enta

ge

Relative-Age in Months

01Sept

02Aug

03Jul

04Jun

05May

06Apr

07March

08Feb

09Jan

10Dec

11Nov

12Oct

100%

80%

60%

40%

20%

0%

Figure 1. School Progress stratified over Relative-Age Positioning.

Special Education Normative Progress Skipped Grade Repeated GradeSchool Progress:


Table 1. Descriptive Statistics of the Variables.

Variable Wave N Range Mean SD

Relative-Age 2230 1 to 12 6.20 3.44

Relative-Age (control) 794 1 to 12 8.02 2.99

Age 1 2230 10.01 to 12.58 11.11 0.55

Fear 1 1982 1 to 5 2.42 0.73

Frustration 1 1983 1 to 4.80 2.79 0.66

Depressive Symptoms 1 2024 0 to 2.31 0.48 0.36

SES 1 2188 -1.94 to 1.73 -0.05 0.80

Age in Years 2 2149 12.15 to 15.15 13.65 0.53

Length (cm) 2 2041 131 to 195 164.85 8.24

Weight (kg) 2 2030 29 to 134 52.84 11.08

BMI 2 2028 12.23 to 40.20 19.00 3.21

Physical Development 2 2087 1 to 20 9.34 3.38

Intellectual Development 2 1534 1 to 20 11.64 3.71

Social Status 2 1007 1 to 5 3.40 1.31

Sport Competence 2 1455 1 to 5 3.48 0.64

Age 3 1819 14.69 to 18.69 16.27 0.73

∆ Fear 1 to 3 1396 -3.57 to 4.12 0.05 1.04

∆ Frustration 1 to 3 1397 -4.07 to 3.11 0.02 0.99

∆ Depressive Symptoms 1 to 3 1343 -3.63 to 4.94 0.00 1.06

N = 2230 (50.8% women), T1 = baseline wave, T3 = third measurement wave, ∆ = change score between T1 and T3, BMI = body mass index, cm = centimeter, k = number of categories, kg = kilogram, N = number of participants, SD = standard deviation.

CHAPTER VI112

Table 2. Pearson Correlations among all Study Variables.

Wave 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16.

1. Relative-Age —

2. Length 2 0.15*** —

3. Weight 2 0.13*** 0.62*** —

4. BMI 2 0.08*** 0.19*** 0.88*** —

5. Physical Dev. 2 0.15*** 0.46*** 0.43*** 0.29*** —

6. Intellectual Dev. 2 -0.01 0.02 -0.01 -0.03 0.03 —

7. Popular Status 2 0.00 -0.04 -0.08 -0.08 0.08 0.10* —

8. Rejected Status 2 0.06 0.04 0.08 0.10 0.06 0.02 — —

9. Sport Competence 2 -0.02 -0.05 -0.16*** -0.17*** -0.02 0.28*** 0.12* -0.06 —

10. Fear 1 -0.02 -0.03 0.03 0.06* 0.03 -0.04 0.10* 0.02 -0.03 —

11. Frustration 1 0.01 0.04 0.07** 0.07** 0.02 -0.05 -0.06 0.11** -0.05 0.31*** —

12. Depressive Sx 1 -0.01 -0.03 0.03 0.04 0.01 -0.02 -0.10* 0.17*** -0.12*** 0.29*** 0.32*** —

13. ΔFear 1-3 -0.01 -0.02 -0.02 0.01 0.03 -0.06* -0.02 -0.02 -0.03 -0.51*** -0.08** -0.07* —

14. ΔFrustration 1-3 0.03 -0.02 -0.02 -0.02 0.02 -0.07* 0.04 -0.06 -0.05 -0.11*** -0.49*** -0.08** 0.26*** —

15. Δ Depressive Sx 1-3 0.03 0.00 0.05 0.07** 0.11*** 0.00 0.06 -0.08 -0.04 -0.08** -0.08** -0.52*** 0.19*** 0.18***

16. Rel. Age (control) 1.00*** 0.02 0.07 0.08* 0.02 0.01 0.03 -0.07*** 0.01 -0.01 0.05 0.03 -0.04 0.04 0.01

17. Biological Age 0.53*** 0.15*** 0.14*** 0.09*** 0.16*** -0.04 -0.02 -0.02 -0.03 -0.07** -0.04 -0.05* 0.08** 0.06* 0.07* -0.28***

N = 2230 (50.8% women). Rel. Age = Relative-Age, T1 = baseline wave, T3 = third measurement wave, ∆ = change score between T1 and T3. Table 1 gives details (e.g., ages). For popular and rejected social status we report biserial correlations (e.g., being popular or rejected or not), because the scale was artificially dichotomous. Partial correlations adjusted for real age at testing, which show the relative-age effects, are presented in table 4. Significance ***p<0.001, **p<0.01, *p<0.05, two-tailed.


Table 2. Pearson Correlations among all Study Variables.

Wave 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16.

1. Relative-Age —

2. Length 2 0.15*** —

3. Weight 2 0.13*** 0.62*** —

4. BMI 2 0.08*** 0.19*** 0.88*** —

5. Physical Dev. 2 0.15*** 0.46*** 0.43*** 0.29*** —

6. Intellectual Dev. 2 -0.01 0.02 -0.01 -0.03 0.03 —

7. Popular Status 2 0.00 -0.04 -0.08 -0.08 0.08 0.10* —

8. Rejected Status 2 0.06 0.04 0.08 0.10 0.06 0.02 — —

9. Sport Competence 2 -0.02 -0.05 -0.16*** -0.17*** -0.02 0.28*** 0.12* -0.06 —

10. Fear 1 -0.02 -0.03 0.03 0.06* 0.03 -0.04 0.10* 0.02 -0.03 —

11. Frustration 1 0.01 0.04 0.07** 0.07** 0.02 -0.05 -0.06 0.11** -0.05 0.31*** —

12. Depressive Sx 1 -0.01 -0.03 0.03 0.04 0.01 -0.02 -0.10* 0.17*** -0.12*** 0.29*** 0.32*** —

13. ΔFear 1-3 -0.01 -0.02 -0.02 0.01 0.03 -0.06* -0.02 -0.02 -0.03 -0.51*** -0.08** -0.07* —

14. ΔFrustration 1-3 0.03 -0.02 -0.02 -0.02 0.02 -0.07* 0.04 -0.06 -0.05 -0.11*** -0.49*** -0.08** 0.26*** —

15. Δ Depressive Sx 1-3 0.03 0.00 0.05 0.07** 0.11*** 0.00 0.06 -0.08 -0.04 -0.08** -0.08** -0.52*** 0.19*** 0.18***

16. Rel. Age (control) 1.00*** 0.02 0.07 0.08* 0.02 0.01 0.03 -0.07*** 0.01 -0.01 0.05 0.03 -0.04 0.04 0.01

17. Biological Age 0.53*** 0.15*** 0.14*** 0.09*** 0.16*** -0.04 -0.02 -0.02 -0.03 -0.07** -0.04 -0.05* 0.08** 0.06* 0.07* -0.28***

CHAPTER VI114

DISCUSSION

In this study we tested effects of intra-classroom relative-age position on multiple domains

of functioning and well-being in adolescence. Three key observations merit further discussion.

First, we observed substantial relative-age effects on school progress; relative young adoles-

cents repeated a grade about four times more often than the relatively old, who in turn were

over 20 times more likely to skip a grade. These observations align with our first hypothesis

(H1), and replicate earlier studies (Cobley et al., 2009; Doornbos, 1971; Verachtert et al., 2010).

Second, in adolescents with a normative school progress (75.4%), no substantive relative-

age effects were observed. Although we observed a small effect on peer rejection (H5a;

which might be a chance finding), we could not replicate previously reported relative-age

effects on weight (H2a), pubertal status (H2b), school performance (H3), sport-competence

(H4), or peer status in terms of popularity (H5b). Neither did we observe substantial effects

on depressive symptoms (H6a), temperamental negative affectivity (H7a), and changes in

depressive symptoms (H6b) or negative affectivity (H7b) between age 11 and 16. Third, in

the subgroup of adolescents who had repeated a grade (16.9%), inverse relative-age effects

were observed; the relatively young were thinner (weight and BMI), had higher school marks,

and reported less depressive symptoms than their relatively older peers.

Relative-Age Effects

The absence of substantial relative-age effects in adolescents with a normative school prog-

ress is surprising, given the existing literature (Doornbos, 1971; Goodman et al., 2003; Martin

et al., 2004; Verachtert et al., 2010), which also covers adolescents between age 11 and

17 (Allen, 2008; Bell & Daniels, 1990; Musch & Grondin, 2001). Some studies reported that

Table 3. Relative-Age Effects on School Progress (Normative Development is reference).

Binary Outcome OR CI (95%)

Repeated Grade 0.83*** (0.80 to 0.86)

Skipped Grade 1.47*** (1.30 to 1.67)

Special Education 0.96 (0.91 to 1.01)

N = 2230 (50.8% women). Regression estimates are based on bootstrapping (k = 10000). CI = Bias Corrected and Accelerated confidence intervals, OR = Ratio of the probability (odds) of an event occurring in one group to the odds of it occurring in another group.***p<0.001, **p<0.01, *p<0.05, two-tailed.


Normative School Progress

Repeated Class Control Sample (robustness check)

Variable Wave rp B CI (95%) rp B CI (95%) rp B CI (95%)

Weight (kg) 2 -.03 -0.10 -0.28 to 0.08 .09 0.57* 0.07 to 1.09 .01 0.05 -0.21 to 0.30

BMI 2 -.01 -0.01 -0.06 to 0.04 .11 0.19* 0.04 to 0.36 .00 0.04 -0.03 to 0.11

Pubertal Status

2 -.01 -0.04 -0.10 to 0.02 .09 0.07 -0.07 to 0.19 -.02 -0.03 -0.11 to 0.05

Intellectual Development

2 .04 0.02 -0.00 to 0.04 -.14* -0.05* -0.10 to -0.01 .05 0.04 -0.07 to 0.15

Sport Competence

2 .01 0.00 -0.02 to 0.02 -.02 -0.01 -0.05 to 0.04 .02 0.00 -0.00 to 0.00

Fear 1 -.02 0.01 -0.01 to 0.02 .10 0.04 0.00 to 0.08 -.03 -0.01*** -0.01 to -0.01

Frustration 1 .05* 0.01* 0.01 to 0.02 .05 0.02 -0.03 to 0.06 .03 0.01 -0.01 to 0.02

Depressive Symptoms

1 .02 0.00 -0.01 to 0.01 .18*** 0.07** 0.02 to 0.11 .03 0.00 -0.01 to 0.01

Δ Fear 1-3 -.05 -0.02 -0.04 to 0.00 -.11 -0.06 -0.13 to 0.02 -.06 -0.02*** -0.02 to -0.02

Δ Frustration 1-3 -.01 -0.00 -0.23 to 0.02 -.04 -0.03 -0.09 to 0.04 .02 0.01* 0.01 to 0.01

Δ Depressive Symptoms

1-3 .01 0.00 -0.02 to 0.02 -.08 -0.03 -0.09 to 0.00 -.01 -0.01*** -0.01 to -0.01

Table 4. Relative-Age Effects, adjusted for Actual Age, as predictor of Multiple Domains, for Adolescents with a Normative School Progress (n = 1681) and Adolescents who had Repeated a Grade (n = 377). Finally, the Control Sample (n = 794) comprised Relative Old Children from Grade 7 and Relative Young from Grade 8 with a Normative School Progress.

T1 = baseline wave, T3 = third measurement wave, ∆ = change between T1 (age 11) and T3 (Age 16), BMI = body mass index, CI = Bias Corrected and Accelerated confidence intervals, rp = partial correlations between relative-age and outcome, adjusted for real age at time of testing. Regression estimates were bootstrapped (k = 10000, bias corrected intervals), and indicate change in outcome per month in relative-age, after adjustment for age at testing. Note that for change variables we also adjusted for change in age between T1 and T3. Details on all measures and procedures can be found in the method section. All correlations between all variables are given in table 2, and SES-adjusted regression estimates in the appendix (table A2). Significance ***p<0.001, **p<0.01, *p<0.05, two-tailed.

CHAPTER VI116

relative-age effects reverse for relatively young adolescents who managed to stay in their

initial cohort (Dalton, 2012; McDonald, 2001), called the extended Akerlof/Kranton model

(Allen, 2008), but we neither replicated this. It has been commonly argued that relative-age

advantages erode when full maturity in the specific system has been reached, in contrast to

advantages conferred by genes (Allen, 2008). Some studies indeed reported this dissipation

of differences (Hutchison & Sharp, 1999; Verachtert et al., 2010).

Many studies reported relative-age effects on health, educational attainment, earnings, and

mortality that persisted into adulthood (Bedard & Dhuey, 2006; Cobley et al., 2009; Persico et

al., 2004). This suggests that relative-age advantages modulate personal and social devel-

opment such that perpetuating mechanisms “get under the skin”, e.g., via identity-formation

(Fenzel, 1992; Thompson et al., 2004), opportunity costs (Bedard & Dhuey, 2006; Flynn,

2009; Gladwell, 2008; Persico et al., 2004), and/or development of (soft) skills (Dhuey & Lip-

scomb, 2008; Musch & Grondin, 2001). We may interpret the substantial relative-age effects

on school progress along similar lines, an amplification of small differences.

Ability Streaming

The absence of substantive relative-age effects in adolescents with a normative school prog-

ress may be a Dutch cultural artifact. In our sample 20% of the adolescents at age 11 had

repeated or skipped a grade (another 6% went to special education), which is already more

than the 1% in the UK and 12% in the USA at age 15 (Dalton, 2012). Ability streaming might

have diluted the relative-age effects in the adolescents with a normative school progress. Al-

ternative (and convergent) explanations are that the Dutch environment is less competitive,

which is a known moderator of relative-age effects (Musch & Grondin, 2001), or that combined

classrooms (with multiple grades) dilute effects. Finally, it may be that positive spillover from

relative older peers result in a net zero effect for the relative young.

This ability streaming hypothesis aligns in a slightly unexpected way with the hypothesized

reciprocal feedback loops in which individuals shape (select/evoke) their environments to

their propensities, in analogy to the corresponsive principle (Caspi & Shiner, 2006) and Dick-

Flynn model (Beam & Turkheimer, 2013; Flynn, 2009). The corresponsive principle predicts

that small intrinsic differences can become amplified by intra-classroom relative-age position

effects, and this process may become catalyzed by strong ability streaming (via its outcome,

that is, a change of grade) which results in a qualitative different classroom environment.

We may understand the reversed relative-age effects in adolescents who repeated a grade

(lower school performance and more depressive symptoms for the relative old than the rela-

tive young) also in terms of the ability-streaming hypothesis; with increasing relative-age

low innate potential (or maturity) rather than chronological age (the cultural relative-age


artifact) becomes the preeminent reason for grade retention. In addition, relatively young

retenders also lost their undesirable intra-cohort position (they became relatively old),

whereas the relatively old became conspicuously old (Doornbos, 1971). Negative long-term

effects of grade retention have been reported before (Wu, West, & Hughes, 2010).

Strengths and Limitations

Our results should be interpreted in light of the following strengths and limitations. A no-

table strength of this study is the large sample of adolescents from the general Dutch popu-

lation. We are quite confident that substantial relative-age effects are absent in our sample,

because our power calculations (in the appendix) for our specific regression analyses in-

dicated that we would observe effects that explained 1% of the variance in each outcome

variable in adolescents with normative school progress (≥d = 0.18) and 3% in the group who

repeated a grade (≥d = 0.34). Moreover, we repeated analyses with a control relative-age

sample, which, despite a negative association with biological age (r = -0.28 versus r = 0.53),

showed similar results. Nevertheless, our adjusted regression analyses lack the rigor of an

experimental design.

The biggest limitation is our lack of knowledge about why children repeated a grade. For

example, some studies reported that some high SES parents deliberately delay their relative

young child’s entry into school (“academic redshirting”) to create a favorable relative old age

position (Bedard & Dhuey, 2006; Graue & DiPerna, 2000; Musch & Grondin, 2001). This sug-

gests cumulative disadvantages for relatively young children from a low SES background

(Dhuey & Lipscomb, 2008). However, we did not observe more relatively young adolescents

from families in the low SES quartile (see appendix table A1), and our results remained un-

changed after adjustment for family SES. Notably, alternative explanations like season-of-

birth effects have been refuted in earlier work, because relative-age effects were observed

in both hemispheres (Bedard & Dhuey, 2006; Musch & Grondin, 2001) and remained after sta-

tistical adjustment for seasonal effects (Lawlor et al., 2006). However, future studies might

explore gender effects, because most studies suggest that relative-age effects are stronger

in males (Musch & Grondin, 2001; Cascio & Schanzenbach, 2007; Sandercock et al., 2013).

CONCLUSION

We studied the relative-age effect bestowed upon children by an adult-imposed structuring

of their worlds, a cultural artifact that may modulate the development of their innate abili-

ties. Our preeminent observation is that intra-classroom position influenced school prog-

ress, and rendered relatively young adolescents more likely to repeat a grade. Second, we

CHAPTER VI118

could not replicate substantial relative-age effects on physical and psychosocial develop-

ment and well-being in adolescents with a normative progress. We argued that this absence

of relative-age effects in adolescents with a normative school progress may reflect a “wash-

out effect” due to ability streaming. This ability streaming (grade retention) may distinguish

the Dutch sample from earlier studies in Britain and the USA. Third, in the subgroup of ado-

lescents who repeated a grade, reverse relative-age effects were observed; the relative old

adolescents performed worse in school and reported more depressive symptoms than their

relatively younger peers did. Future studies might explore in more detail the mechanisms by

which relative-age effects may get internalized, and models that can explain alleged cultural

differences in their effects.


REFERENCES

Abela, J. R. Z., Fishman, M. B., Cohen, J. R., & Young, J. F. (2012). Personality predispositions to depression in children of affectively-ill parents: The buffering role of self-esteem. Journal of Clinical Child and Ado-lescent Psychology, 41(4), 391-401. doi:10.1080/15374416.2012.654463

Achenbach, T. M. (1991a). Manual for the child behavior Checklist/4-18 and 1991 profiles. Burlington, VT: Department of Psychiatry, University of Vermont.

Achenbach, T. M. (1991b). Manual for the youth self-report and 1991 profile. Burlington: University of Ver-mont.

Achenbach, T. M., Dumenci, L., & Rescorla, L. A. (2003). DSM-oriented and empirically based approaches to constructing scales from the same item pools. Journal of Clinical Child and Adolescent Psychology, 32(3), 328-340. doi:10.1207/S15374424JCCP3203_02

Allen, J. (2008). Relative age, identity and schooling: An extension of the Akerlof/Kranton model which solves a puzzle. The Journal of Socio-Economics, 37(1), 343-352. doi:http://dx.doi.org/10.1016/j.so-cec.2007.06.011

Beam, C. R., & Turkheimer, E. (2013). Phenotype-environment correlations in longitudinal twin models. De-velopment and Psychopathology, 25(1), 7-16. doi:10.1017/S0954579412000867

Bedard, K., & Dhuey, E. (2006). The persistence of early childhood maturity: International evidence of long-run age effects. The Quarterly Journal of Economics, 121(4), 1437-1472. doi:10.1093/qje/121.4.1437

Bell, J. F., & Daniels, S. (1990). Are summer-born children disadvantaged - the birth-date effect in education. Oxford Review of Education, 16(1), 67-80. doi:10.1080/0305498900160106

Bobak, M., & Gjonca, A. (2001). The seasonality of live birth is strongly influenced by sociodemographic fac-tors. Human Reproduction, 16(7), 1512-1517. doi:10.1093/humrep/16.7.1512

Borenstein, M., Hedges, L. V., Higgins, J. P. T., & Rothstein, H. R. (2009). Introduction to meta-analysis. United Kingdom: John Wiley & Sons Ltd.

Boyce, W. T., Obradovic, J., Bush, N. R., Stamperdahl, J., Kim, Y. S., & Adler, N. (2012). Social stratification, classroom climate, and the behavioral adaptation of kindergarten children. Proceedings of the National Academy of Sciences, 109, 17168-17173.

Buckles, K. S., & Hungerman, D. M. (2013). Season of birth and later outcomes: Old questions, new answers. Review of Economics and Statistics, 95(3), 711-724. doi:10.1162/REST_a_00314

Carruthers, P., Laurence, S., & Stich, S. (Eds.). (2006). The innate mind: Culture and cognition (vol. 2). Lon-don: Oxford University Press.

Cascio, E., & Schanzenbach, D. W. (2007). First in the class? Age and the education production function. (NBER Working Paper No. 13663). Cambridge, MA: National Bureau of Economic Research.

Caspi, A., & Shiner, R. (2006). Personality development. In N. Eisenberg (Ed.), Handbook of child psychology (Vol. 3 ed., pp. 300). Hobokon, New Jersey: John Wiley & Sons, Inc.

Cobley, S., McKenna, J., Baker, J., & Wattie, N. (2009). How pervasive are relative age effects in secondary school education? Journal of Educational Psychology, 101(2), 520-528. doi:10.1037/a0013845

Collins, W. A., Welsh, D. P., & Furman, W. (2009). Adolescent romantic relationships. Annual Review of Psy-chology, 60(1), 631-652.

Cyranowski, J., Frank, E., Young, E., & Shear, M. (2000). Adolescent onset of the gender difference in lifetime rates of major depression - A theoretical model. Archives of General Psychiatry, 57(1), 21-27. doi:10.1001/archpsyc.57.1.21

Dalton, B. (2012). Grade level and science achievement: US performance in cross-national perspective. Comparative Education Review, 56(1), 125-154. doi:10.1086/660745

CHAPTER VI120

Deardorff, J., Fyfe, M., Ekwaru, J. P., Kushi, L., Greenspan, L., & Yen, I. (2012). Does neighborhood environ-ment influence girls’ pubertal onset? Findings from a cohort study. BMC Pediatrics, 12(1), 27.

de Winter, A. F., Oldehinkel, A. J., Veenstra, R., Brunnekreef, J. A., Verhulst, F. C., & Ormel, J. (2005). Evalua-tion of non-response bias in mental health determinants and outcomes in a large sample of pre-adoles-cents. European Journal of Epidemiology, 20(2), 173-181. doi:10.1007/s10654-004-4948-6

Dhuey, E., & Lipscomb, S. (2008). What makes a leader? Relative age and high school leadership. Economics of Education Review, 27(2), 173-183. doi:10.1016/j.econedurev.2006.08.005

Dijkstra, J. K., Cillessen, A. H. N., & Borch, C. (2013). Popularity and adolescent friendship networks: Selec-tion and influence dynamics. Developmental Psychology, 49(7), 1242–1252. doi:10.1037/a0030098

Doornbos, K. (1971). Geboortemaand en schoolsucces: Een onderwijskundig onderzoek naar de samen-hang tussen de leeftijdspositie binnen de jaargroep en het schoolsucces van leerlingen in het basis, voortgezet en buitengewoon onderwijs. [Birth month and school success: An educational study about the association between intra-class relative-age position and the education success of pupils in pri-mary and secondary education]. PhD thesis, Rijksuniversiteit Groningen.

Eaves, L., Silberg, J., Foley, D., Bulik, C., Erkanli, A., Angold, A., et al. (2004). Genetic and environmental influ-ences on the relative timing of pubertal change. Twin Research, 7(5), 471.

Fenzel, L. M. (1992). The effect of relative age on self-esteem, role strain, GPA, and anxiety. The Journal of Early Adolescence, 12(3), 253-266. doi:10.1177/0272431692012003002

Ferdinand, R. F. (2008). Validity of the CBCL/YSR DSM-IV scales anxiety problems and affective problems. Journal of Anxiety Disorders, 22(1), 126-134. doi:http://dx.doi.org/10.1016/j.janxdis.2007.01.008

Flynn, J. R. (2009). What is intelligence? (2nd ed.). Cambridge: Cambridge University Press.

Fraley, R. (2002). Attachment stability from infancy to adulthood: Meta-analysis and dynamic modeling of developmental mechanisms. Personality and Social Psychology Review, 6(2), 123-151. doi:10.1207/S15327957PSPR0602_03

Gladwell, M. (2008). Outliers: The story of success. USA: Little, Brown and Company.

Gledhill, J., Ford, T., & Goodman, R. (2002). Does season of birth matter? The relationship between age within the school year (season of birth) and educational difficulties among a representative general popu-lation sample of children and adolescents (aged 5-15) in Great Britain. Research in Education, 68(1), 41-47. doi:10.7227/RIE.68.4

Goodman, R., Gledhill, J., & Ford, T. (2003). Child psychiatric disorder and relative age within school year: Cross sectional survey of large population sample. British Medical Journal, 327(7413), 472-475. doi:10.1136/bmj.327.7413.472

Graue, M., & DiPerna, J. (2000). Redshirting and early retention: Who gets the “gift of time” and what are its outcomes? American Educational Research Journal, 37(2), 509-534. doi:10.2307/1163532

Hankin, B., Abramson, L., Moffitt, T., Silva, P., McGee, R., & Angell, K. (1998). Development of depression from preadolescence to young adulthood: Emerging gender differences in a 10-year longitudinal study. Journal of Abnormal Psychology, 107(1), 128-140. doi:10.1037//0021-843X.107.1.128

Harris, J. R. (1995). Where is the childs environment - A group socialization theory of development. Psycho-logical Review, 102(3), 458-489. doi:10.1037/0033-295X.102.3.458

Harris, J. R. (2009). The nurture assumption: Why children turn out the way they do (2nd ed.). New York, USA: Free Press, Simon & Schuster, Inc.

Hartman, C. A. (2000). Dutch translation of the early adolescent temperament questionnaire. Internal Re-port, Department of Psychiatry, University of Groningen, the Netherlands.

Henrich, J., & Gil-White, F. (2001). The evolution of prestige - freely conferred deference as a mechanism for enhancing the benefits of cultural transmission. Evolution and Human Behavior, 22(3), 165-196. doi:10.1016/S1090-5138(00)00071-4


Huisman, M., Oldehinkel, A. J., de Winter, A. F., Minderaa, R. B., de Bildt, A., Huizink, A. C., et al. (2008). A cohort profile: The Dutch ‘TRacking Adolescents’ Individual Lives’ Survey’; TRAILS. International Journal of Epidemiology, 37(6), 1227-35. doi:10.1093/ije/dym273

Hutchison, D., & Sharp, C. (Sept, 1999). A lasting legacy? The persistence of season of birth effects. British Educational Research Association Conference, National Foundation for Educational Research, 18-10-2013. doi: http://www.leeds.ac.uk/educol/documents/000001095.htm

Janssens, K. A. M., Rosmalen, J. G. M., Ormel, J., Verhulst, F. C., Hunfeld, J. A. M., Mancl, L. A., et al. (2011). Pubertal status predicts back pain, overtiredness, and dizziness in American and Dutch adolescents. Pediatrics, 128(3), 553-559. doi:10.1542/peds.2010-2364

Lawlor, D. A., Clark, H., Ronalds, G., & Leon, D. A. (2006). Season of birth and childhood intelligence: Findings from the Aberdeen children of the 1950s cohort study. British Journal of Educational Psychology, 76, 481-499. doi:10.1348/000709905X49700

Li, X., Wong, W., Lamoureux, E. L., & Wong, T. Y. (2012). Are linear regression techniques appropriate for analysis when the dependent (outcome) variable is not normally distributed? Investigative Ophthalmol-ogy & Visual Science, 53(6), 3082-3083. doi:10.1167/iovs.12-9967

Martin, R. (2013). How we do it. The evolution and future of human reproduction. New York: Basic Books.

Martin, R., Foels, P., Clanton, G., & Moon, K. (2004). Season of birth is related to child retention rates, achievement, and rate of diagnosis of specific LD. Journal of Learning Disabilities, 37(4), 307-317. doi:10.1177/00222194040370040301

McDonald, G. (2001). Comparing school systems to explain enduring birth date effects. Compare: A Journal of Comparative and International Education, 31(3), 381-391. doi:10.1080/03057920120098509

Morrow, R. L., Garland, E. J., Wright, J. M., Maclure, M., Taylor, S., & Dormuth, C. R. (2012). Influence of relative age on diagnosis and treatment of attention-deficit/hyperactivity disorder in children. Canadian Medical Association Journal, 184(7), 755-762. doi:10.1503/cmaj.111619

Muhlenweg, A. M. (2010). Young and innocent international evidence on age effects within grades on vic-timization in elementary school. Economics Letters, 109(3), 157-160. doi:10.1016/j.econlet.2010.08.032

Musch, J., & Grondin, S. (2001). Unequal competition as an impediment to personal development: A review of the relative age effect in sport. Developmental Review, 21(2), 147-167. doi:10.1006/drev.2000.0516

Newcomb, A. F., Bukowski, W. M., & Pattee, L. (1993). Childrens peer relations – a meta-analytic review of popular, rejected, neglected, controversial, and average sociometric status. Psychological Bulletin, 113(1), 99-128. doi:10.1037//0033-2909.113.1.99

Oldehinkel, A. J., Hartman, C. A., de Winter, A. F., Veenstra, R., & Ormel, J. (2004). Temperament profiles associated with internalizing and externalizing problems in preadolescence. Development and Psycho-pathology, 16(2), 421-440. doi:10.1017/S0954579404044591

Oldehinkel, A. J., Wittchen, H., & Schuster, P. (1999). Prevalence, 20-month incidence and outcome of unipo-lar depressive disorders in a community sample of adolescents. Psychological Medicine, 29(3), 655-668. doi:10.1017/S0033291799008454

Oldehinkel, A. J., Rosmalen, J. G. M., Veenstra, R., Dijkstra, J. K., & Ormel, J. (2007). Being admired or being liked: Classroom social status and depressive problems in early adolescent girls and boys. Journal of Abnormal Child Psychology, 35(3), 417-427. doi:10.1007/s10802-007-9100-0

Ormel, J., Oldehinkel, A. J., Sijtsema, J., van Oort, F., Raven, D., Veenstra, R., & Verhulst, F. C. (2012). The TRacking Adolescents’ Individual Lives Survey (TRAILS): Design, current status, and selected findings. Journal of the American Academy of Child & Adolescent Psychiatry, 51(10), 1020-1036. doi:10.1016/j.jaac.2012.08.004

Persico, N., Postlewaite, A., & Silverman, D. (2004). The effect of adolescent experience on labor market outcomes: The case of height. National Bureau of Economic Research Working Paper Series, No. 10522.

CHAPTER VI122

Peterson, R. A., & Brown, S. P. (2005). On the use of beta coefficients in meta-analysis. Journal of Applied Psychology, 90(1), 175-181. doi:10.1037/0021-9010.90.1.175

Pierson, K., Addona, V., & Yates, P. (2014). A behavioural dynamic model of the relative age effect. Journal of Sports Sciences, 32(8), 776-784. doi:10.1080/02640414.2013.855804

Putnam, S. P., Ellis, S. K., & Rothbart, M. K. (2001). The structure of temperament from infancy through adolescence. In A. Eliasz, & A. Engleneiter (Eds.), Advances in research on temperament (pp. 165-182). Berlin: Pabst Scientist Publicer.

Rothbart, M. K. (2011). Becoming who we are. New York, NY: Guilford Press.

Rothbart, M. K., Ahadi, S. A., & Evans, D. E. (2000). Temperament and personality: Origins and outcomes. Journal of Personality and Social Psychology, 78(1), 122-135. doi:10.1037/0022-3514.78.1.122

Sandercock, G. R., Taylor, M. J., Voss, C., Ogunleye, A. A., Cohen, D. D., & Parry, D. A. (2013). Quantification of the relative age effect in three indices of physical performance. Journal of Strength & Conditioning Research, 27(12), 3293.

Shirtcliff, E. A., Dahl, R. E., & Pollak, S. D. (2009). Pubertal development: Correspondence between hormonal and physical development. Child Development, 80(2), 327-337. doi:10.1111/j.1467-8624.2009.01263.x

Thompson, A., Barnsley, R., & Battle, J. (2004). The relative age effect and the development of self-esteem. Educational Research, 46(3), 313-320. doi:10.1080/0013188042000277368

Thompson, A., Barnsley, R., & Dyck, R. (1999). A new factor in youth suicide: The relative age effect. Cana-dian Journal of Psychiatry-Revue Canadienne De Psychiatrie, 44(1), 82-85.

Verachtert, P., De Fraine, B., Onghena, P., & Ghesquiere, P. (2010). Season of birth and school suc-cess in the early years of primary education. Oxford Review of Education, 36(3), 285-306. doi:10.1080/03054981003629896

Wu, W., West, S. G., & Hughes, J. N. (2010). Effect of grade retention in first grade on psychosocial outcomes. Journal of Educational Psychology, 102(1), 135-152. doi:10.1037/a0016664


SUPPLEMENTARY MATERIAL

Table A1. Frequencies for School Progress and Socioeconomic Status (SES) stratified over Relative-Age Distribution in Quartiles.

Relative Young

2nd Quartile

3rd Quartile

Relative Old

Relative-Age: 1 to 3 4 to 6 7 to 9 10 to 12

Birth Month: Sept, Aug, July

June, May, April

March, Feb, Jan

Dec, Nov, Oct

Total Sample

2230 100.0% 624 100% 569 100% 537 100% 500 100%

Normative Development

1681 75.4% 398 63.8% 441 77.5% 443 82.5% 399 79.8%

Repeated Grade

377 16.9% 185 29.6% 90 15.8% 61 11.4% 41 8.2%

Skipped Grade

48 2.2% 2 0.3% 4 0.7% 7 1.3% 35 7.0%

Special Education

124 5.5% 39 6.3% 34 6.0% 26 4.8% 25 5.0%

Low SES Quartile

547 25.0% 135 24.7% 142 26.0% 149 27.2% 121 22.1%

High SES Quartile

547 25.0% 150 27.4% 141 25.8% 139 25.4% 117 21.4%

CHAPTER VI124

Table A2. Relative-Age Effects on Multiple Domains, adjusted for Actual Age at testing and Socioeconomic Status (SES), for Adolescents with a Normative School Progress (n = 1681), and Adolescents who Repeated a Grade (n = 377).

Normal School ProgressRelative-Age Effect

Repeated GradeRelative-Age Effect

Variable Wave rp B CI (95%) rp B CI (95%)

Length (cm) 2 -.04 -0.10 (-0.24 to 0.04) -.01 0.09 (-0.25 to 0.42)

Weight (kg) 2 -.04 -0.13 (-0.30 to 0.05) .08 0.60* (0.02 to 1.11)

BMI 2 -.02 -0.02 (-0.07 to 0.03) .10 0.19* (0.03 to 0.36)

Pubertal Status 2 -.02 -0.04 (-0.10 to 0.02) .08 0.07 (-0.07 to 0.19)

Intellectual Development

2 .04 0.06 (-0.01 to 0.13) -.12 -0.19* (-0.36 to -0.02)

Sport Competence

2 .01 0.00 (-0.01 to 0.02) -.01 -0.00 (-0.03 to 0.03)

Fear 1 .01 0.01 (-0.01 to 0.02) .09 0.04 (-0.00 to 0.08)

Frustration 1 .05 0.02 (-0.00 to 0.03) .04 0.02 (-0.03 to 0.06)

Depressive Symptoms

1 .01 0.00 (-0.02 to 0.02) .18*** 0.02** (0.02 to 0.11)

Δ Fear 1-3 -.06a -0.02 (-0.04 to 0.00) -.11 -0.06 (-0.13 to 0.02)

Δ Frustration 1-3 -.01 -0.00 (-0.03 to 0.02) -.04 -0.03 (-0.09 to 0.04)

Δ Depressive Symptoms

1-3 .01 0.00 (-0.02 to 0.02) -.08 -0.03 (-0.09 to 0.04)

a = p was 0.052. T1 = baseline wave, T3 = third measurement wave, ∆ = change between T1 (age 11) and T3 (Age 16), BMI = body mass index, CI = Bias Corrected and Accelerated confidence intervals, rp = partial correlations between relative-age and outcome adjusted for real age at time of testing. Regression estimates were bootstrapped (k = 10000, bias corrected intervals), and indicate change in outcome per month in relative-age, after adjustment for age at testing. Note that for change variables we also adjusted for change in age between T1 and T3. Details on all measures and procedures can be found in the method section. All correlations between all variables are given in table 2.***p<0.001, **p<0.01, *p<0.05, two-tailed.


Table A3. Relative-Age Effects on School Progress (Normative Development is reference).

Adj. for Real Agea Adj. for Real Age & SESb

Binary Outcome OR CI (95%) OR CI (95%)

Repeated Grade 0.83*** (0.80 to 0.86) 0.82*** (0.79 to 0.86)

Skipped Grade 1.47*** (1.30 to 1.67) 1.66*** (1.44 to 1.92)

Special Education 0.96 (0.91 to 1.01) 0.81*** (0.75 to 0.87)

N = 2230 (50.8% women). a The odds are based on bootstrapping (k = 10000). b Odds are not based on bootstrapping. CI = Bias Corrected and Accelerated confidence intervals, OR = Ratio of the probability that an event will happen to all possible cases for that event. ***p<0.001, **p<0.01, *p<0.05, two-tailed.

CHAPTER VI126

Table A4. Relative-Age as the predictor of the Odds of Being Popular or Rejected (Normative Development is reference) in Adolescents with a Normative School Progress and Adolescents who Repeated a Grade, adjusted for Real Age at testing.

School Progress

Social Status

rp OR CI (95%) p B CI (95%)

Normative Popular -.00 1.00 (0.94 to 1.06) 0.90 -0.00 (-0.07 to 0.06)

Rejected .09*** 1.12 (1.05 to 1.19) 0.00 0.11 (0.04 to 0.18)

Repeated a Grade

Popular .04 1.11 (0.83 to 1.49) 0.55 0.11 (-0.53 to 0.66)

Rejected .02 1.04 (0.85 to 1.28) 0.67 0.04 (-0.20 to 0.24)

N = 2230 (50.8% women). In total 137 (8.1%) adolescents were rated as popular and 132 (7.9%) as rejected. Regression estimates are based on bootstrapping (k = 10000). CI = Bias Corrected and Accelerated confidence intervals, OR = Ratio of the probability that an event will happen to all possible cases for that event, rp = partial correlations between relative-age and outcome, adjusted for real age at time of testing. The values for the presented correlations and regression remained almost identical after additional adjustment for family SES. ***p<0.001, **p<0.01, *p<0.05, two-tailed.


Table A5. Social Status stratified over Relative-Age Position in Quartiles and School Progress.

Normative School Progress Repeated a Grade

Rejected n Popular n Rejected n Popular n

Relatively Young

9.3% 37 7.5% 30 2.7% 5 1.6% 3

Second 6.1% 27 9.5% 42 3.3% 3 1.1% 1

Third 6.5% 29 8.4% 37 4.9% 3 1.6% 1

Relatively Old

9.8% 39 7.0% 28 4.9% 2 2.4% 1

Total 7.9% 132 8.1% 137 3.4% 13 1.6% 6

n = number of subjects.

Table A6. Quartiles of Socioeconomic Status (SES) stratified over School Progress.

Normative Repeated a Grade Skipped a Grade Special Education

% n % n % n % n

Low SES 65.3% 357 22.7% 124 0.9% 5 11.2% 61

Second Q 72.0% 394 21.4% 117 0.7% 4 5.9% 32

Third Q 77.1% 422 15.2% 83 3.8% 21 3.8% 21

High SES 87.6% 479 8.6% 47 3.1% 17 0.7% 4

n = number of subjects, Q = Quartile.

CHAPTER VI128

A7. Power Calculations

We calculated the power of all specific linear regression analyses (Selya, Rose, Dierker, He-

deker, & Mermelstein, 2012). In the adolescents with a normative school progress reasonably

precise estimations (given 80% power) were guaranteed beyond Cohen’s f 2 = 0.01 (~R2 = 0.01,

d = ~0.18, 2 predictors, α = 0.05, and N≥1100 for all outcomes). In the group of adolescents

who had repeated a grade we could estimate effects beyond Cohen’s f 2 = 0.03 (~R2 = 0.03,

d = ~0.34, given 80% power, 2 predictors, α = 0.05, N≥320 for all outcomes). However, we con-

cluded we lacked the power to test for relative-age effects in the group that skipped a grade

(N = 48, Cohen’s f 2 = 0.24, R2 = 0.19, d = ~0.98). These post-hoc power calculations indicate

that we obtained reasonably precise estimations in our study, and we are therefore confi-

dent that substantial relative-age effects are indeed absent.

REFERENCES

Selya, A. S., Rose, J. S., Dierker, L. C., Hedeker, D., & Mermelstein, R. J. (2012). A practical guide to calculating Cohen’s f2, a measure of local effect size, from PROC MIXED. Frontiers in Psychology, 3, 111. doi:10.3389/fpsyg.2012.00111

The Dynamic Relationship between Physical Activity and Mood in the Daily Life of Depressed and Non-Depressed Individuals: a Within-Subject Time-Series Analysis

“I get my exercise running to the funerals of my friends who exercise.”

Barry Gray, in New York Magazine, 1980

“Getting fit is all about mind over matter. I don’t mind, so it doesn’t matter.”

Adam Hargreaves, in Mr Lazy's Guide to Fitness, 2000

Stavrakakis N, Booij SH, Roest AM, de Jonge P, Oldehinkel AJ & Bos EH

Under preparation

CHAPTER

CHAPTER VII130

ABSTRACT

Purpose: The association between physical activity and mood found in longitudinal observa-

tional studies is generally small to moderate. It is unknown how this association generalizes

to individuals. The aim of the present study was to investigate inter-individual differences in

the bidirectional dynamic relationship between physical activity and mood, in depressed and

non-depressed individuals, using time-series analysis.

Methods: Twenty pair-matched depressed and non-depressed participants (30% males,

mean age = 36.6, SD = 8.9) wore accelerometers and completed electronic questionnaires

three times a day for 30 days. Physical activity was operationalized as the total energy ex-

penditure (EE) per day segment (i.e., 6 hours). The multivariate time series (T = 90) of every

individual were analyzed using vector autoregressive modeling (VAR), with the aim to assess

direct as well as lagged (i.e., over 1 day) associations between EE and positive and negative

affect.

Results: We observed large inter-individual differences in the strength, direction and tem-

poral aspects of the relationship between physical activity and positive and negative affect.

An exception was the direct (but not the lagged) influence of physical activity on positive

affect, which was positive in nearly all individuals.

Conclusion: This study showed that the association between physical activity and affect

varied considerably across individuals. Thus, while at the group level the effect of physical

activity on mood may be small, in some individuals the effect may be clinically relevant.

131PHYSICAL ACTIVITY AND MOOD IN DAILY LIFE

INTRODUCTION

Physical activity or exercise is considered to be an effective treatment for depression (Dunn,

Trivedi, Kampert, Clark, & Chambliss, 2005; Ströhle et al., 2007; Ströhle, 2009). Although this

proposition is appealing, since physical activity can be implemented easily and with rela-

tive low costs, the empirical evidence of the effectiveness of physical activity is not robust

(Chalder et al., 2012; Cooney et al., 2013; Krogh, Nordentoft, Sterne, & Lawlor, 2010). Simi-

larly, observational studies have shown an inverse association between physical activity

and depressive symptoms (Dunn, Trivedi, & O’Neal, 2001; Farmer et al., 1988; Salmon, 2001;

Strawbridge, Deleger, Roberts, & Kaplan, 2002), with small to moderate effect sizes in dif-

ferent populations (Biddle & Asare, 2011; Johnson & Taliaferro, 2011). Although large cohort

studies with a few measurements are useful to detect general patterns and traits in the

population, they are not suitable for studying dynamic relationships between variables that

fluctuate in daily life (Hamaker, 2012; Hamaker, Dolan, & Molenaar, 2005), which is the case

for physical activity and depressive symptoms.

Some studies have adopted a more ecologically valid approach to investigate the associa-

tion between depressive symptoms and physical activity. Ecological momentary assess-

ment (EMA) studies address the dynamic nature of factors by investigating them repeat-

edly over time in participants’ daily life. Since slight changes in positive and negative affect

might accumulate over time to result in depressed mood, it is important to investigate how

physical activity influences daily affect levels (Rosmalen, Wenting, Roest, de Jonge, & Bos,

2012). A recent literature overview of EMA studies investigating the relationship between

physical activity and momentary affective states in daily life (Kanning, Ebner-Priemer, &

Schlicht, 2013) showed a generally consistent positive association between physical activ-

ity and positive affect across studies, which became stronger when only high-quality stud-

ies were included. In contrast, the influence of physical activity on negative affect, although

observed in some studies (see for example: Dunton, Liao, Intille, Wolch, & Pentz, 2011 and

LePage & Crowther, 2010), was non-existent in the majority of studies reviewed (see: Kan-

ning et al., 2013).

The above EMA studies all used multilevel analysis to estimate dynamic associations. Al-

though multilevel analysis allows for inter-individual differences when estimating the rela-

tionship between physical activity and mood, it is in principle a between-subject (group) ap-

proach; individual regression terms are averaged across the group and the same regression

model is imposed on all subjects rather than modeling the dynamic relationship for each

subject individually. Only under strict conditions can findings at the inter-individual level be

generalized to the intra-individual level (see: Hamaker, 2012), and these conditions rarely

apply to the study of dynamic psychological processes (Hamaker et al., 2005; Molenaar &

CHAPTER VII132

Campbell, 2009). In addition, a within-subject time-series approach may be more suitable for

investigating bidirectional relationships than multilevel analysis, because these models are

built at the individual level and all variables in the model can be used as both predictors and

outcomes (Brandt & Williams, 2007). The added value of such an approach was illustrated

by a recent time-series study investigating the dynamic relationship between physical activ-

ity and depressive symptoms in four post-myocardial infarction patients (Rosmalen et al.,

2012). Inter-individual differences in the magnitude as well as the direction of the association

were shown, in spite of the relatively homogeneous study sample.

The present study aimed to further elucidate inter-individual differences in the bidirec-

tional association between physical activity and mood. Adopting an intensive time-series

approach, we studied the dynamic relationship between directly measured physical activ-

ity (i.e., accelerometers) and affect in the daily life of pair-matched depressed and non-de-

pressed individuals, and individual differences therein.

METHODS

Participants

The sample was part of the ‘Mood and movement in daily life’ (MOOVD) study, which was

set up to investigate the dynamic relationship between physical activity and mood in daily

life, and the role of several biomarkers therein. Participants (age 20-50) were intensively

monitored in their natural environments for 30 days, by means of electronic diaries, saliva

sampling, and continuous actigraphy. The multiple repeated measurements per individual

(T = 90) allowed assessing within-person temporal relationships between variables at the

individual level. For the current study, 10 depressed and 10 non-depressed participants, who

were pair-matched on gender, smoking status, age, and Βody Μass Ιndex (BMI), were in-

cluded (the first 20 participants that finished the study).

Depressed individuals were recruited from the psychiatry department of the University Med-

ical Center Groningen (UMCG) and the Center for Integrative Psychiatry (CIP) in Groningen,

the Netherlands. Non-depressed participants were recruited from the general population by

means of posters and advertisements in local newspapers. Participants were screened for

depressive symptoms with the Beck Depression Inventory version 2 (BDI-II; Beck, Erbaugh,

Ward, Mock, & Mendelsohn, 1961) and for several other health conditions with a general

health questionnaire. Individuals with a BDI-II score of >14 (i.e., depressed group) and indi-

viduals with a BDI-II score of <9 (i.e., non-depressed group) were invited for the inclusion

interview. The inclusion interview covered the Composite International Diagnostic Interview

(CIDI; World Health Organization, 1990), several questionnaires, and a briefing in which the


use of equipment and the electronic diary was explained. Depressed individuals were includ-

ed if they met DSM-IV criteria for a major depressive disorder (current episode or in remission

for no longer than 8 weeks). Non-depressed individuals were included if they were free of

mood disorders at the moment of inclusion. Individuals who reported chronic somatic illness

or medication use that may directly influence functioning of the Hypothalamic Pituitary Ad-

renal (HPA) axis or the Autonomic Nervous System (ANS) were excluded. Other reasons for

exclusion were: pregnancy, significant hearing or visual impairments, and having a current

or recent (within the last two years) psychotic or bipolar disorder as assessed with the CIDI

interview. Participants received a fee for participation and a personal report of their daily

mood and activity patterns. The MOOVD study design was approved by the Medical Ethical

Committee and all participants gave written informed consent.

Ambulatory Sampling

Participants completed questionnaires on an electronic diary, the PsyMate (PsyMate BV,

Maastricht, the Netherlands) (Myin-Germeys, Birchwood, & Kwapil, 2011) for a total of 32

days, with the first two days used to get familiar with the device. The PsyMate was pro-

grammed to generate beeps at three predetermined moments a day with equidistant in-

tervals: in the morning (mean≈10:00 AM), six hours later in the afternoon (mean≈4:00 PM),

and six hours later in the evening (mean≈10:00 PM). This fixed interval design was chosen

to allow the application of time-series analysis. To capture most of participants’ daily life

without intruding with their sleep habits, beeps were planned at the end of the morning,

afternoon, and evening, with the exact time points depending on the individual participant’s

sleep-wake schedule. This schedule was assessed by the Munich Chronotype Questionnaire

(MCTQ; Roenneberg et al., 2007) and an extra question regarding the time they would go to

bed if they were to go to bed ‘on time’. After every alarm beep, participants were asked to

fill out the electronic diary. They were instructed to do so immediately after the beep, but a

time window of one hour was used for situations in which this was not possible. Through-

out the study period, participants wore the ActiCal® (Respironics, Bend, OR, USA), an om-

nidirectional water-resistant accelerometer, on the wrist of the non-writing arm. They were

instructed never to remove the ActiCal except when entering a sauna (high temperatures).

The ActiCal recorded data over 1-min sampling intervals. For more details about the ActiCal,

see Heil (2006).

Measures

The electronic diary questionnaire contained 60 items on mood, cognition, and daily activi-

ties. Positive and negative affect scores were computed from mood items adopted from

Bylsma, Taylor-Clift, and Rottenberg (2011), rated on a 7-point Likert scale. For positive

CHAPTER VII134

affect, seven positive mood items were averaged, namely feeling talkative, enthusiastic,

confident, cheerful, energetic, satisfied, and happy (range 1-7). For negative affect, seven

negative mood ratings were averaged, namely feeling tense, anxious, distracted, restless,

irritated, depressed, and guilty (range 1-7).

Energy expenditure (EE) data from the ActiCal was used as a measure for physical activity.

For each participant, the amount of EE (kcal/min) across all sampling minutes over one day

segment (i.e., 360 minutes per segment) was summed. This resulted in total EE per day seg-

ment (kcal/day segment).

Person characteristics including age, gender, educational level, smoking status, coffee and

alcohol intake, and the average amount of exercise per week (min) were obtained from ques-

tionnaires filled out at the baseline assessment. BMI was calculated from the weight and

height of each individual.

Statistical Analysis

Missing diary data were imputed by means of Expectation-Maximization imputation in IBM

SPSS Statistics 20. For every participant, all variables used in the time-series model as well

as auxiliary variables that were significantly associated with these variables were used as

predictor variables in the imputation. On average, participants had 8 missing values on the

diary variables (8.8%). Of the 20 participants, three participants had missing actigraphy data

at the last 9 or 10 days of their study, because of technical problems. These relatively large

periods without activity records were discarded and not imputed, leaving 60-63 valid data

points for these three participants, which is still considered sufficient for the application of

time-series analysis (Lütkepohl, 2007; Rosmalen et al., 2012).

The multiple time series of every individual were analyzed using vector autoregressive (VAR)

modeling. A VAR model is a multivariate autoregressive model that consists of a set of re-

gression equations for a system of two or more variables, in this case positive affect, nega-

tive affect, and EE (Brandt & Williams, 2007). All three variables in the system were treated

as endogenous, which means that they could be both determinant and outcome. Each of

the three endogenous variables was regressed on its own p lagged values and the p lagged

values of the other variable. The errors of the VAR model should be serially uncorrelated but

can be contemporaneously correlated. VAR analysis is especially suitable for investigating

the dynamic relationships between two or more variables: inferences can be made about the

temporal order of effects, which can involve bidirectional effects and feedback loops (Brandt &

Williams, 2007). Moreover, by analyzing all variables in one system, the most important ef-

fects will be identified. For more details about VAR modeling and a non-technical introduc-

tion, see Brandt & Williams (2007) and Rosmalen et al. (2012), respectively.


The number of lags included in the models was a priori set to three, which is equivalent to a

period of one day. A fixed number of lags was chosen to ease comparison of results for dif-

ferent participants. Considering the study’s exploratory nature, a relatively high number of

lags was chosen, allowing for the investigation of relatively long-term lagged effects (i.e.,

over 3 time intervals = 1 day). To account for structural lower morning values for total EE,

compared to afternoon and evening values – participants spent part of the morning lying in

bed – as well as diurnal rhythms in mood, dummy variables for morning and evening were in-

cluded (afternoon served as a reference category). Variables for time and the square of time

were included into the model if this was necessary to render the series stationary. Total EE

was divided by 100 to accommodate the difference in scaling between EE and positive and

negative affect. Maximum likelihood estimation with a degrees-of-freedom adjustment ad-

vocated for small samples was used for estimating the VAR coefficients (Lütkepohl, 2007).

VAR model assumptions, namely stability of the model, independence, homoscedasticity

and normality of residuals, were assessed using diagnostic checks (Lütkepohl, 2007). If one

of the tests indicated a violation of the model assumptions, models were adjusted, re-esti-

mated, and re-evaluated, in an iterative model-building process, until all assumptions were

met. Outlier modeling was applied only if this was required to meet model assumptions,

using the criterion of residual SD>3 (stepwise lowering to SD>2, if necessary; Field, 2009).

Log transformation of the variables was applied if the residuals remained skewed. If scat-

terplots of skewed variables with other variables suggested curvilinear relationships, non-

skewed variables were left untransformed. In other cases, all endogenous variables were

log-transformed.

To assess the direct effect of EE on positive and negative affect, we examined the contempo-

raneous correlations between the variables, which can be retrieved from the residuals of the

final models. These correlations represent the simultaneous correlations between the vari-

ables, i.e., between the scores at the same assessment point. To assess the overall lagged

effects of EE on positive and negative affect, and vice versa, the effect size coefficients of

lags 1 to 3 were averaged into one net effect size coefficient, and standardized so that they

could be compared across participants. To test the significance of the overall lagged effects

the Granger causality Wald test was used. This is a test for the directionality of the influence

between two time series (Granger, 1969; Lütkepohl, 2007). In view of the exploratory nature

of the study, a significance level of 0.05 was used for all tests. All analyses were done in

STATA 11 using the suite of VAR commands (StataCorp., 2009).

CHAPTER VII136

RESULTS


The demographic and clinical characteristics of the depressed and non-depressed partici-

pants are presented in Table 1.

Analysis

Short-Term Effect of Physical Activity on Positive and Negative Affect

As depicted in Figure 1a, the majority of depressed (red bars) and non-depressed (blue bars)

participants showed a positive contemporaneous association between EE and positive af-

fect. As the EE scores covered the period directly before the mood rating, this can be inter-

preted as a direct effect of EE on positive affect. For five individuals (three non-depressed

and two depressed) these positive associations were significant (p<0.05). The size of the

correlations was small to moderate, according to the suggestions provided by Cohen (1992).

Figure 1b shows the direct association between EE and negative affect for depressed (red

bars) and non-depressed (blue bars) participants. Eight of them showed a positive (non-sig-

nificant) contemporaneous association between EE and negative affect, while nine partici-

pants showed a negative association, without any apparent differences between the groups

(depressed and non-depressed). For two individuals (one depressed and one non-depressed)

the negative association between EE and negative affect was significant (p<0.05). The size of

these correlations was again small to moderate.


Table 1. Demographic and Clinical Characteristics for the Depressed and Non-Depressed Group.

Depressed Group (n = 10) Non-Depressed Group (n = 10)

Mean SD Min Max Mean SD Min Max

Demographics

Age 36.4 10.3 22 49 36.7 7.9 24 46

Gender (% male) 30.0 — — — 30.0 — — —

Body Mass Index (BMI) 23.8 4.9 17.4 34.9 22.2 2.3 19.5 26.9

Educational Level (0-3) 2.6 0.5 2 3 2.6 0.5 2 3

Lifestyle

Coffee Intake (per day) 2.2 2.5 0 8 1.8 1.4 0 4

Alcohol Intake (per month) 10.8 14.5 0 40 8.4 14.4 0 48

Self-Reported Exercise Frequency (per week)

1.7 1.5 0 4 2 1.5 0 4

Self-Reported Duration of Exercise (min. per week)

83 89 0 240 133 109 0 240

Average EE over one day segment

233 77 126 385 258 69 120 369

Mood and Depressive Symptoms

Pre-BDI-II 32.5 10 20 51 2.8 3.4 0 10

Post-BDI-II 25.9 15.8 2 47 4.4 4.6 0 13

Average PA (1-7) 3.1 1.3 1.1 4.5 4.6 1.3 1.7 6.1

Average NA (1-7) 3.6 1.2 1.4 5.3 1.6 0.6 1.0 2.8

BDI-II = Beck Depression Inventory, EE = energy expenditure, PA = positive affect, NA = negative affect, SD = standard deviation.

CHAPTER VII138

Figure 1a and 1b. Direct Association between Physical Activity and Positive (1a) and Negative (1b) Affect for the Depressed (red bars) and Non-Depressed (blue bars) Individuals.

EE = energy expenditure. * significant at the 0.05 level. Values on y-column depict correlation coefficients (r).

5 4 16 1 6 12 9 17

815

7

113 13 19 2

1018

Participant Number

1b. EE - Negative Affect

0.4

0.3

0.2

0.1

0

-0.1

-0.2

-0.3

-0.4

r

12 85 19 2 16 1 9 17

Participant Number

1a. EE - Positive Affect

0.4

0.3

0.2

0.1

0

-0.1

-0.2

-0.3

-0.4

r

6143182013117154


Lagged Αssociation between Physical Activity and Positive, and Negative Affect

The lagged associations between previous values of EE (i.e., the joint effects of lag 1, lag 2,

and lag 3) and current values of positive affect varied greatly between individuals (Figure

2a), without any apparent differences between the groups (depressed versus non-de-

pressed). Seven participants showed a positive lagged association between EE and posi-

tive affect, but only for two (one depressed and one non-depressed individual) this effect

was significant. For the remaining 12 participants a negative lagged association of EE on

subsequent positive affect was observed, and for four individuals (all non-depressed) this

association was significant. Mixed effects (i.e., both positive and negative associations on

different lags) were present in two of them (participant 15 and 16). Figure 2b shows the

results for EE and negative affect. The majority of participants (eleven compared to seven

participants) showed a negative lagged association between EE and subsequent negative

affect, which was significant in three participants (one depressed and two non-depressed).

Overall, the effect sizes of the lagged associations between EE and positive and negative

affect were small.

Lagged Association between Positive and Negative Affect and Physical Activity

With regard to the lagged association between positive affect and subsequent EE (Figure

3a), the majority of the participants (n = 15) showed a positive association between positive

affect and subsequent EE. For two participants (both depressed) this association was sig-

nificant. Four participants (two depressed and two non-depressed) showed a non-significant

negative association. The majority of participants also showed a positive lagged association

between negative affect and subsequent EE (eleven compared to six participants) (see Fig-

ure 3b). For only one (depressed) participant this association reached statistical significance

(p<0.05). Overall, the effect sizes of the lagged associations between positive and negative

affect and EE were small.

CHAPTER VII140

EE = energy expenditure; * significant at the 0.05 level; ~significant association with mixed sign of coefficients at different lags. Values on y-column depict standardized coefficients (beta).

16 9 12 7 19 31

11 15 134 8

6 2 10 518 17

Participant Number

2b. EE (lagged) - Negative Affect

0.3

0.2

0.1

0

-0.1

-0.2

-0.3

Bet

aFigure 2a and 2b. Lagged Associations between Physical Activity and Positive (2a) and Negative (2b) Affect in the Depressed (red bars) and Non-Depressed (blue bars) Individuals.

14 17 11 9 6 133

1912

2 157 8

5 16 18 20 1 4

Participant Number

2a. EE (lagged) - Positive Affect

0.3

0.2

0.1

0

-0.1

-0.2

-0.3

Bet

a


Figure 3a and 3b. Lagged Associations between Positive (3a) and Negative (3b) Affect and Physical Activity in the Depressed (red bars) and Non-Depressed (blue bars) Individuals.

11 12 18 4 16 21714156191398

317 5 20

Participant Number

3a. Positive Affect (lagged) - EE

0.3

0.2

0.1

0

-0.1

-0.2

-0.3

Bet

a

EE = energy expenditure. * significant at the 0.05 level. Values on y-column depict standardized coefficients (beta).

3b. Negative Affect (lagged) - EE

5 3 19 12 4 2161876110

13 1517 9 8 11

Participant Number

0.3

0.2

0.1

0

-0.1

-0.2

-0.3

Bet

a

CHAPTER VII142

DISCUSSION

To our knowledge, this is the first study to use a within-subject time-series approach to

investigate the dynamic relationship between physical activity and affective states in de-

pressed and non-depressed individuals. The associations found were not consistent across

individuals, ranging from negative to positive, bidirectional or not, and with varying effect

sizes. An exception was the direct association between physical activity and positive affect,

which was positive in nearly all individuals, albeit with varying effect sizes.

A positive relationship between physical activity and positive affect has also been found in

the majority of group-based EMA studies (for example see: Giacobbi, Hausenblas, & Frye,

2005 and Wichers et al., 2012). However, the delayed association between physical activity

and positive affect was not consistently found in the current study; in some individuals it

was positive, while in others it was negative. Furthermore, the strength of the relationship

was small to negligible for most individuals. Thus, the results suggest that physical activity

enhances positive affect directly while only in some individuals there is also a delayed effect.

The increase in positive affect following physical activity did not seem to differ between de-

pressed and non-depressed individuals, although any potential differences between the two

groups were not evaluated statistically due to the small sample size. A recent EMA study

comparing patients with a depressive disorder and non-depressed controls (Mata et al.,

2012) showed similar improvements in positive affect after physical activity in both groups.

Hence, physical activity seems to (directly) improve positive affect in individuals regardless

of their depression status.

While the direct association between physical activity and positive affect was quite consis-

tent across individuals, this was not the case for negative affect. Both positive and nega-

tive associations between physical activity and negative affect were observed, though we

found the strongest effect sizes for a negative association. A similar pattern emerged for the

lagged association between physical activity and subsequent negative affect. So, in some

individuals there seems to be a negative effect of physical activity on negative affect, but in

most individuals this relationship is weak or non-existent. When aggregating data of indi-

viduals with opposing effects of physical activity on negative affect, the group-level effect

will be small and inconsistent across different study samples. Hence, individual differences

may explain inconsistent results in EMA studies regarding the relationship between physical

activity and negative affect (Kanning et al., 2013).

Observational (between-subject) studies investigating the direction of the relationship

between physical activity and depressive symptoms have found a reciprocal relationship

between the two variables under investigation (Jerstad, Boutelle, Ness, & Stice, 2010;


Lindwall, Larsman, & Hagger, 2011; Stavrakakis, de Jonge, Ormel, & Oldehinkel, 2012). We

extended these findings by showing that the reciprocal relationship differed between indi-

viduals. In some individuals increases in positive mood were followed by increases in their

physical activity levels, while in others it was the other way around. Interestingly, in only a

few individuals there appeared to be a noticeable bidirectional relationship. If these data

were aggregated, we might have concluded that there was a bidirectional relationship at the

group level, while in most individuals the relationship was unidirectional, albeit in different

directions.

The effect sizes in the present study were generally small, but considering the non-experi-

mental daily-life design and the short intervals between the measurements this could have

been expected (Wichers et al., 2012). Nonetheless, small effects can accumulate over time

and become larger, because of the way these changes are disseminated through the sys-

tem and mutually enhanced through possible feedback loops. This can especially be true

if physical activity becomes a daily habit, for instance through regular sport participation,

which could result in systematic long-term improvements in positive mood. Future research

using intensive time-series designs should explore this putative accumulating effect in more

detail, in order to elucidate whether the short-term benefits of physical activity on mood can

accumulate into larger benefits over time, and if so, in which individuals.

Clinical Relevance

Two findings from the current study might be of interest to clinicians. First, our results sug-

gest that physical activity benefits positive affect in the short-term. A recent meta-analysis

of randomized controlled studies showed evidence of a short-term benefit of physical activ-

ity on depression (Krogh et al., 2010). Our results support and extend this finding by showing

that these short-term benefits might be acting through an elevation of positive mood rather

than through a decrease in negative mood for the majority of individuals. The mechanisms

behind this short-term benefit of physical activity on positive affect are currently not under-

stood. Some hypotheses have been proposed implicating the endorphin system (Dishman &

O’Connor, 2009; Thoren, Floras, Hoffmann, & Seals, 1990) and increases in self-efficacy

(Bodin & Martinsen, 2004), but future research is needed to explore these in more depth.

Second, in some individuals high levels of positive or negative affect were associated with

increases in physical activity in the current study. Feeling good might increase the level of

physical activity engagement, but apparently also feeling bad might have such an effect

in some individuals. This concurs with the suggestion that physical activity might be per-

ceived as an efficient strategy to repair negative mood, i.e., ‘walking off’ negative affectivity

(Clark & Isen, 1982). Our study cannot address this hypothesis properly, but suggests inter-

CHAPTER VII144

individual differences in the way affect influences subsequent activity levels that need to

be taken into account.

Strengths & Limitations

This study has several notable strengths. First, the time-series design allows the investiga-

tion of the associations at the individual level, thanks to the multiple repeated measure-

ments over time, and also allows for the exploration of temporal patterns and the direction

of the relationship. Second, although obtaining multiple repeated measurements can be

troublesome, time-consuming, and demanding, especially for participants with depression,

the compliance of the participants did not seem to deteriorate significantly in this study.

There was no indication that the participants were unable or found it difficult to comply with

the study protocol, and any systematic missing values were due to technical problems with

the accelerometers. Even in these cases the number of obtained measurements per indi-

vidual (>60) was still adequate to perform time-series analysis.

The results from this study should be interpreted with some caution due to several limita-

tions. First, the generalizability of within-subject studies to the population at large is lim-

ited, and this is the trade-off that comes with the increased specificity of within-subjects

time-series studies. Therefore, a combination of within- and between-subject approaches

is warranted in order to improve our understanding of the complex relationship between

physical activity and daily mood. Second, the models applied in this study assume that the

relationship between physical activity and affect is linear. It is possible that the relationship

between the two is curvilinear and a specific threshold (i.e., frequency, duration or inten-

sity) of physical activity is needed to benefit individuals with regards to their affect, while

exceeding this threshold might result in negative consequences. Another potential limita-

tion is that the strength of an association depends on the degree of variability in the data.

Therefore, it is possible that for some individuals the variability between the measures was

not large enough to detect an association. Finally, we considered the temporal ordering of

the contemporaneous associations to be from physical activity to affect, given the fact that

physical activity covered each time period as a whole and affect was assessed at the end of

each time period. However, it is possible that reverse effects (affect to physical activity) were

acting here as well, which might have introduced bias in our interpretation. However, due to

the design of the study we could not assess this eventuality further.


CONCLUSION

We explored inter-individual differences in the magnitude and direction of the relationship

between physical activity and positive and negative affect using a within-subjects time-se-

ries design. Our findings provided evidence that the strength and direction of the relationship

between physical activity and affect differs across individuals. Future research investigating

this dynamic relationship could try to identify subgroups of individuals that might benefit

most from physical activity. These findings could be informative for both research and clini-

cal care intended to improve the mood of depressed individuals.

CHAPTER VII146

REFERENCES

Beck, A. T., Erbaugh, J., Ward, C. H., Mock, J., & Mendelsohn, M. (1961). An inventory for measuring depres-sion. Archives of General Psychiatry, 4(6), 561-71.


Bodin, T., & Martinsen, E. W. (2004). Mood and self-efficacy during acute exercise in clinical depression. A randomized, controlled study. Journal of Sport & Exercise Psychology, 26(4), 623-633.

Brandt, P. T., & Williams, J. T. (2007). Multiple time series models. Sage, Inc.

Bylsma, L. M., Taylor-Clift, A., & Rottenberg, J. (2011). Emotional reactivity to daily events in major and minor depression. Journal of Abnormal Psychology, 120(1), 155-167.

Chalder, M., Wiles, N. J., Campbell, J., Hollinghurst, S. P., Searle, A., Haase, A. M., et al. (2012). A pragmatic randomised controlled trial to evaluate the cost-effectiveness of a physical activity intervention as a treatment for depression: The treating depression with physical activity (TREAD) trial. Health Technol-ogy Assessment, 16(10), 1-+.

Clark, M. S., & Isen, A. M. (1982). Toward understanding the relationship between feeling states and social behavior. In A. H. Hastorf, & A. M. Isen (Eds.), Cognitive social psychology (pp. 73-108). New York: Elsevier.

Cohen, J. (1992). A power primer. Psychological Bulletin, 112(1), 155-159.


Dishman, R. K., & O’Connor, P. J. (2009). Lessons in exercise neurobiology: The case of endorphins. Mental Health and Physical Activity, 1(2), 4-9.


Dunn, A. L., Trivedi, M. H., & O’Neal, H. A. (2001). Physical activity dose-response effects on outcomes of de-pression and anxiety. Medicine and Science in Sports and Exercise, 33(6 Suppl), S587-97; discussion 609-10.

Dunton, G. F., Liao, Y., Intille, S., Wolch, J., & Pentz, M. A. (2011). Physical and social contextual influences on children’s leisure-time physical activity: An ecological momentary assessment study. Journal of Physi-cal Activity & Health, 8, S103-S108.

Farmer, M. E., Locke, B. Z., Moscicki, E. K., Dannenberg, A. L., Larson, D. B., & Radloff, L. S. (1988). Physical activity and depressive symptoms: The NHANES I epidemiologic follow-up study. American Journal of Epidemiology, 128(6), 1340-1351.

Field, A. (2009). Discovering statistics using SPSS (3rd ed.). SAGE publications Ltd.

Giacobbi, P. R., Hausenblas, H. A., & Frye, N. (2005). A naturalistic assessment of the relationship between personality, daily life events, leisure-time exercise, and mood. Psychology of Sport and Exercise, 6(1), 67-81.

Granger, C. W. J. (1969). Investigating causal relations by econometric models and cross-spectral methods. Econometrica, 37(3), 424-38.

Hamaker, E. L. (2012). Why researchers should think “within-person”. In M. Mehl, & T. Conner (Eds.), Hand-book of research methods for studying daily life (pp. 43-61). New York: The Guildford Press.

Hamaker, E. L., Dolan, C. V., & Molenaar, P. C. M. (2005). Statistical modeling of the individual: Rationale and application of multivariate stationary time series analysis. Multivariate Behavioral Research, 40(2), 207-233.

Heil, D. P. (2006). Predicting activity energy expenditure using the actical (R) activity monitor. Research Quarterly for Exercise and Sport, 77(1), 64-80.

Jerstad, S. J., Boutelle, K. N., Ness, K. K., & Stice, E. (2010). Prospective reciprocal relations between physical activity and depression in female adolescents. Journal of Consulting and Clinical Psychology, 78(2), 268-272.



Kanning, M. K., Ebner-Priemer, U. W., & Schlicht, W. M. (2013). How to investigate within-subject associa-tions between physical activity and momentary affective states in everyday life: A position statement based on a literature overview. Frontiers in Psychology, 4, 187.

Krogh, J., Nordentoft, M., Sterne, J. A., & Lawlor, D. A. (2010). The effect of exercise in clinically depressed adults: Systematic review and meta-analysis of randomized controlled trials. The Journal of Clinical Psychiatry, 72(4), 529-538.

LePage, M. L., & Crowther, J. H. (2010). The effects of exercise on body satisfaction and affect. Body Image, 7(2), 124-130.


Lütkepohl, H. (2007). New introduction to multiple time series analysis. Springer.

Mata, J., Thompson, R. J., Jaeggi, S. M., Buschkuehl, M., Jonides, J., & Gotlib, I. H. (2012). Walk on the bright side: Physical activity and affect in major depressive disorder. Journal of Abnormal Psychology, 121(2), 297-308.

Molenaar, P. C. M., & Campbell, C. G. (2009). The new person-specific paradigm in psychology. Current Direc-tions in Psychological Science, 18(2), 112-117.

Myin-Germeys, I., Birchwood, M., & Kwapil, T. (2011). From environment to therapy in psychosis: A real-world momentary assessment approach. Schizophrenia Bulletin, 37(2), 244-247.

Roenneberg, T., Kuehnle, T., Juda, M., Kantermann, T., Allebrandt, K., Gordijn, M., et al. (2007). Epidemiology of the human circadian clock. Sleep Medicine Reviews, 11(6), 429-438.

Rosmalen, J. G. M., Wenting, A. M., Roest, A. M., de Jonge, P., & Bos, E. H. (2012). Revealing causal hetero-geneity using time series analysis of ambulatory assessments: Application to the association between depression and physical activity after myocardial infarction. Psychosomatic Medicine, 74(4), 377-386.


Stavrakakis, N., de Jonge, P., Ormel, J., & Oldehinkel, A. J. (2012). Bidirectional prospective associations between physical activity and depressive symptoms. The TRAILS study. Journal of Adolescent Health, 50(5), 503-508.

Strawbridge, W. J., Deleger, S., Roberts, R. E., & Kaplan, G. A. (2002). Physical activity reduces the risk of subsequent depression for older adults. American Journal of Epidemiology, 156(4), 328-334.

Ströhle, A., Hofler, M., Pfister, H., Muller, A. G., Hoyer, J., Wittchen, H. U., et al. (2007). Physical activity and prevalence and incidence of mental disorders in adolescents and young adults. Psychological Medicine, 37(11), 1657-1666.

Ströhle, A. (2009). Physical activity, exercise, depression and anxiety disorders. Journal of Neural Trans-mission, 116(6), 777-784.

Thoren, P., Floras, J. S., Hoffmann, P., & Seals, D. R. (1990). Endorphins and exercise - physiological-mecha-nisms and clinical implications. Medicine and Science in Sports and Exercise, 22(4), 417-428.

Wichers, M., Peeters, F., Rutten, B. P. F., Jacobs, N., Derom, C., Thiery, E., et al. (2012). A time-lagged mo-mentary assessment study on daily life physical activity and affect. Health Psychology, 31(2), 135-144.

World Health Organization (1990). Composite international diagnostic interview (CIDI), ver. 1.0. Geneva: WHO.

Summary and General Discussion

“The great tragedy of science – the slaying of a beautiful hypothesis by an ugly fact”

Thomas Henry Huxley, in Collected Essays, Vol. 8, 1870

CHAPTER

151DISCUSSION

Summary and General Discussion

Overview of the Research Objectives and Key Results

Physical activity (PA) has played a very important role in human evolution. Humans living in

modern societies are relatively inactive compared to our early ancestors and modern hunter-

gatherers, which clearly has led to health problems. However, the association between being

physically active and developing depressive symptoms is not that clear. The overall aim of

this thesis was to investigate the nature of the relationship between PA and depressive

symptoms and to provide a better understanding of this relationship.

The first part of this thesis (chapters II and III) investigated the strength and direction of

the relationship between PA and depressive symptoms in adolescents. The data used came

from the TRacking Adolescents’ Individual Lives Survey (TRAILS). More specifically, the re-

lationship between PA and depressive symptoms was investigated prospectively (over three

measurement waves spanning approximately 6 years) using Structural Equation Modeling

(SEM) in chapter II. The results showed a weak bidirectional relationship between the two

factors. This bidirectional relationship was observed for cognitive but not for somatic symp-

toms of depression. Adolescents who were more physically active were less likely to experi-

ence depressive symptoms over time, and vice versa, adolescents with more depressive

symptoms were less likely to be physically active over time.

The study described in chapter III focused on the protective hypothesis of PA, by investigat-

ing its potential to prevent the onset of depression. Using a sample of adolescents, followed

up for approximately 5 years, this study found no evidence that PA protected against the on-

set of a first major depressive episode in early adulthood in either boys or girls. Adolescents

who were physically active at an early age were, regardless of the frequency, duration and

intensity of the activity, equally likely to develop a first episode of depression later in life as

their more sedentary counterparts.

The overall aim of the second part of this thesis was to explore possible genetic (chapter IV)

and psychosocial (chapter V and VI) modifiers of the relationship between PA and depressive

symptoms. The samples used were derived from TRAILS as well.

In chapter IV, the role of so-called plasticity genes in the prospective reciprocal relation-

ship between PA and depressive symptoms was investigated. Specific alleles of these genes

have been suggested to confer a susceptibility to environmental influences. This susceptibil-

ity has been hypothesized to increase with increasing numbers of plasticity alleles (Belsky

& Beaver, 2011). Therefore, a cumulative plasticity index was created, in order to investigate

its putative modifying effects on the relationship between PA and depressive symptoms.

The individual polymorphisms were investigated as well. Overall, there was no evidence that

CHAPTER VIII152

either the cumulative plasticity gene index or the individual polymorphisms modified the

bidirectional prospective association between PA and depressive symptoms. Adolescents

with genotypes assumed to reflect high plasticity did not benefit more from PA with regards

to their depressive symptoms and did not exhibit more depressive symptoms after being

physically inactive than individuals without these susceptibility alleles.

The study in chapter V investigated the effects of sport participation from an evolutionary

perspective. We explored whether participating in competitive sports was inversely related to

baseline depressive symptoms and changes in depressive symptoms over time. In addition, we

examined moderating effects of peer-perceived sport competence and gender. Overall, sport

participation was inversely associated with current depressive symptoms, but not with symp-

tom changes over time. The association between sport participation and current depressive

symptoms was largely explained by sport competence. Finally, there was no evidence that

sport competence and gender moderated the relationship between sport participation and de-

pressive symptoms: boys who participated in sports and were considered to be highly athleti-

cally competent did not exhibit fewer depressive symptoms than girls and less competent boys.

The study in chapter VI explored the effects of classroom positioning according to birth date

in multiple domains of psychological development and other outcomes, including baseline

depressive symptoms, depressive symptom changes during adolescence, and teacher-rated

sport competence. In a large sample of adolescents followed for approximately 6 years, sub-

stantial relative-age effects were observed in relation to school progress: relatively younger

adolescents were four times more likely to repeat a grade, while relatively older adolescents

were up to 20 times more likely to skip a grade. In the subgroup that repeated a grade,

the relatively young adolescents reported fewer depressive symptoms than their relatively

older peers. However, the relative-age effects did not influence the evaluation of sport com-

petence in either the normative group or the subgroup that repeated a grade.

The final part of this thesis (chapter VII)30 examined temporal patterns in the association

between PA and affect, using a within-subject approach (time-series analysis), with the aim

to provide a better understanding of the strength and direction of the relationship between

PA and affect within individuals31. A sample of depressed and non-depressed adults from the

Mood and Movement in Daily Life (MOOVD) study was used. The results showed inter-indi-

vidual differences in the strength and the sign of the relationships between PA and affect. An

exception was formed by the correlations between concurrent PA and positive affect, which

were positive in nearly all individuals. This suggests a consistent pattern of enhancement of

positive affect during and immediately after PA, but no consistent improvement of negative

affect. The longer-term (delayed) effects of PA on affect varied greatly across individuals,

regardless of their depression status.

153DISCUSSION

The Direction and Strength of the Relationship between PA and Depressive Symptoms

In this section, the direction of the relationship between PA and depressive symptoms will

be discussed, followed by some considerations on the strength of the association, and how

the findings from the studies of this thesis relate to previous research.

Direction

Chapter II describes a weak bidirectional relationship between PA and depressive symptoms

in adolescents. Thus, physically active adolescents were less depressed than sedentary

adolescents and adolescents with many depressive symptoms were more sedentary than

those with few depressive symptoms, both cross-sectionally and over time. This finding is

consistent with the results from previous studies in adolescent girls (Jerstad, Boutelle, Ness,

& Stice, 2010) and older adults (Lindwall et al., 2011)32. Birkeland et al. (2009) did not find

any evidence of a bidirectional relationship in a longitudinal study of adolescents between

the ages of 13 and 23. Although they observed a relationship between PA and depression at

age 13, changes in PA did not predict changes in depressive symptoms and vice versa. The

reasons for this discrepancy are not quite clear33. Nonetheless, the findings from chapter II

support both the protective and the inhibition hypotheses previously proposed. Bidirectional

relationships between PA and positive and negative affect were also found in the MOOVD

study described in chapter VII, but with substantial individual differences. Some individuals

exercised more when they felt better, which resulted in feeling even better, while for others

this spiral was reversed or not the case at all. Individual differences in the direction of the

relationship between PA and depressive symptoms are of great interest to both researchers

and clinicians and should be examined in more detail, since most observational studies ag-

gregate results across individuals and ignore differences among them systematically.

A number of mechanisms have been proposed to explain the putative benefits of PA on de-

pressive symptoms (protective hypothesis). The predominant models include biological and

psychosocial factors that are altered through PA and in turn influence the trajectory of de-

pressive symptoms. It is important to point out that the proposed biological and psychoso-

cial mechanisms are not in any way mutually exclusive.

The main biological mechanisms involve neuromodulators such as serotonin (5-HT; Dishman

et al., 2006a), dopamine (DA; Bliss & Ailion, 1971), adrenaline and noradrenaline (NA; Dish-

man et al., 2006a; Helmich et al., 2010), endorphins (Boecker et al., 2008), and neurotrophins

such as the Brain Derived Neurotrophic Factor (BDNF; Ploughman, 2008). Levels of these

biomarkers are altered during and after exercise, which in turn triggers a cascade of changes

in brain functioning that might explain the alleviation of depressive symptoms (Dishman

et al., 2006a; Helmich et al., 2010; Struder & Weicker, 2001; Toups et al., 2011; van Praag,

CHAPTER VIII154

Christie, Sejnowski, & Gage, 1999; van Praag, Kempermann, & Gage, 1999; van Praag, 2008).

For example, the monoamine hypothesis of depression postulates that imbalances in 5-HT

and DA34 underlie features and symptoms of depression (Lapin & Oxenkrug, 1969; van Praag

& Korf, 1970). Although this hypothesis has been questioned recently (Delgado, 2000; Leo &

Lacasse, 2008; Luscher, Shen, & Sahir, 2011 but also see Cowen, 2008), evidence from imag-

ing studies supports a dysfunction in 5-HT1A receptors in depressed individuals (Drevets

et al., 2007), and suggests that this dysfunction is not merely an artifact of antidepres-

sant medication (Hirvonen et al., 2008). Likewise, decreased DA35 activity has been linked

to motoric and cognitive symptoms of depression (Nestler & Carlezon, 2006; Willner, 1995),

but also to motivational problems and anhedonia in patients with major depressive disorder

(MDD; Dunlop & Nemeroff, 2007). Because of interactions between the DA and 5-HT sys-

tems (Dremencov et al., 2004; Sasaki-Adams & Kelley, 2001), dysfunctions in one of these

systems can initiate secondary changes in others, such as the GABA or adrenaline systems

(Luscher et al., 2011; Thase, 2009).

The psychosocial mechanisms that have been proposed to explain the association between

PA and depressive symptoms are based on self-determination and self-efficacy theories

(Biddle & Mutrie, 2008; Deci & Ryan, 1985; Deci & Flaste, 1995; Sallis & Owen, 1999; Salmon,

2001). The rationale is that by engaging in PA, individuals become more confident in their

abilities, and feel more in control. The increase in self-esteem induces an improvement in

mood (Dishman et al., 2006b). There is also evidence that activities in which individuals form

social networks provide a feeling of belonging and empowerment, which might result in alle-

viation of depressive symptoms as well (McGale, McArdle, & Gaffney, 2011). Finally, a recent

review (Thompson-Coon et al., 2011) suggests that particularly outdoor activities benefit

mental health. However, this area of research is relatively underdeveloped and although the

initial results seem promising, future research is needed to explain the particular impact of

outdoor activities on individuals’ mood and depressive symptoms.

The inhibition hypothesis posits that depressed mood, at least to some degree, demoti-

vates the individual from being physically active (Birkeland et al., 2009). Even sub-thresh-

old depressive symptoms, i.e., symptoms that do not reach the threshold for clinical di-

agnosis, have been associated with psychosocial and functional impairments (Georgiades,

Lewinsohn, Monroe, & Seeley, 2006; Lewinsohn, Solomon, Seeley, & Zeiss, 2000). These

symptoms may trigger a negative feedback loop in which it is difficult for individuals to be

energetic and motivated enough to exercise36 (Goodwin, 2003). The majority of studies to

date have focused on the benefits of PA on depressive symptoms, while only very few have

focused on the inhibition hypothesis (Roshanaei-Moghaddam, Katon, & Russo, 2009). The

findings of chapter II and VII indicate that there is a need to acknowledge the inhibition hy-

pothesis, at least for certain individuals.

155DISCUSSION

Strength of the Αssociation

Overall, whenever a statistically significant association between PA and depressive symp-

toms was observed in the studies described in this thesis, the effects were weak. PA ex-

plained approximately 1 to 2% of the variance in depressive symptoms, and sport participa-

tion, regardless of the competitive element of each sport, slightly more, approximately 3

to 5%. Besides methodological limitations, which will be discussed later, there are several

possible reasons why the effects were weak.

First, the weak effects might indicate that the relationship between PA and depressive

symptoms is not likely to be causal. Although a plethora of hypotheses have been proposed

explaining the relationship between PA and depressive symptoms, a genetic study in a large

sample of monozygotic and dizygotic twins by de Moor et al. (2008) has raised doubts on the

causality of the relationship. They observed cross-sectional and prospective inverse asso-

ciations between leisure-time PA and anxiety and depressive symptoms, with small effect

sizes. However, monozygotic twins that exercised a lot did not report fewer anxiety and de-

pressive symptoms than their more sedentary co-twins (de Moor et al., 2008). Moreover, in

the prospective analyses, changes in PA were not related to changes in anxiety and depres-

sive symptoms, which contradicts one of the predictions made by the causal hypothesis37.

This indicates that the association between PA and depressive symptoms might be caused

by the effects of common genes acting independently, i.e., influencing the propensity to ex-

ercise while at the same time also protecting against symptoms of depression, but without

a causal link between the two.

If a long-term causal relationship exists, it may be expected that PA will exhibit clear ben-

efits in preventing the onset of depression. This has been shown in a recent review of pro-

spective studies in adults, adolescents and children, where PA seemed to prevent the devel-

opment of depression (Mammen & Faulkner, 2013). However, the findings from chapter III do

not indicate such benefits. This is contrary to the findings of a previous study (Jacka et al.,

2011), based on a large Australian study in over 2000 adults (median age 56 years), in which

PA during childhood and depression in later life were retrospectively measured. In this study,

low levels of self-reported childhood PA increased the odds of reporting depression in adult-

hood by 35%. Even though these findings might be affected by retrospective bias, it is still

possible that PA prevents depression onset in middle to late adulthood where the average

onset of MDD is estimated to be between the ages of 20 and 3038 (Kessler & Wang, 2009).

A second reason for the weak associations between PA and depressive symptoms could be

that the relationship between PA and depressive symptoms is modified by other factors.

For example, puberty, school transitions and other important life changes, biological fac-

tors, social factors, sleep and diet, and personality traits may all affect both one’s mood and

CHAPTER VIII156

engagement in PA, and perhaps also moderate the relationship between the two. Chapters

IV, V and VI investigated some of these factors in more detail and their impact is discussed

in the next section.

Another possible reason for the weak associations is that PA may have a differential relation-

ship with different clusters of depressive symptoms. Chapter II showed that PA was inversely

associated with the cognitive symptom but not with the somatic symptom subscale of de-

pression39. This difference might have been due to the low reliability of the somatic symptom

subscale of depressive symptoms40. Possibly, individual somatic symptoms are differentially

related to PA, for instance, PA could be inversely related to sleep problems and positively

related to energy levels. Currently, there is a lack of studies examining this possibility.

Finally, the weak effects might reflect that PA is beneficial in reducing depressive symptoms

for some individuals but not for all41. In large-scale epidemiological studies like TRAILS,

information is aggregated from groups of individuals (between-subject design) and group av-

erages of the association between PA and depressive symptoms are presented42. Even large

effects at the group level, however, may not apply to fluctuations within individuals, since

a between-subject correlation is not the same as a within-subject correlation (Hamaker,

Dolan, & Molenaar, 2005; Molenaar & Campbell, 2009; Rosmalen, Wenting, Roest, de Jonge,

& Bos, 2012). This between-subject variability needs to be taken into account when studying

the relationship between PA and depressive symptoms, in order to avoid unjustified infer-

ences of group-level findings with regard to temporal patterns within individuals (de Jonge &

Bos, 2013; Hamaker, 2012; Molenaar & Campbell, 2009). To learn how PA and affect influence

each other over time within persons, studies using time-series analysis43 with a large number

of repeated assessments per person offer a great research tool. The results of the study in

chapter VII suggest that PA had a consistent positive short-term influence on positive affect

for the majority of participants, but not on negative affect. This finding is in line with the

majority of ecological momentary assessment (EMA) studies, which have generally shown

that the relationship between PA and positive affect is more consistent than the relationship

between PA and negative affect (Kanning et al., 2013). However, even though the majority of

individuals showed a consistent positive relationship between PA and positive affect, only

for a few individuals (5 out of 19) was this relationship significant and of at least moderate

strength. This indicates that, even though for some individuals the relationship was moder-

ate, a group aggregation approach would most likely dilute the effects and be weak over all

participants. This was even more pronounced in the relationship between PA and negative

affect, where the associations were very heterogeneous among individuals, that is, some

participants showed a negative association, a few others a positive one44. Therefore, the

results from the study in chapter VII indicate that the weak effects found in prior chapters are

probably due to individual heterogeneity in the relationship between PA and affect. Future

157DISCUSSION

studies should use a combination of between- and within-subject approaches, in order to

identify commonalities between individuals and elucidate why some individuals benefit or

are harmed by PA in relation to their mood.

Genetic and Psychosocial Modifiers

The second part of this thesis (chapters IV, V and VI) explored possible genetic and social modi-

fiers of the PA-depressive symptom relationship. Both will be discussed in the following part.

Genetic Modifiers

Belsky et al. (2009) identified a number of genes that are thought to be involved to suscep-

tibility to environmental influences, and have been labeled plasticity genes. The plasticity

genes hypothesis extends the classic diathesis-stress model45, by stating that individuals

with a specific genotype will not only be at increased risk when exposed to negative life

events and experiences, but also benefit more from positive ones, such as social support46

(Belsky & Pluess, 2009; Belsky et al., 2009; Ellis, Boyce, Belsky, Bakermans-Kranenburg, &

van Ijzendoorn, 2011). These putative plasticity genes, which were investigated in chapter

IV, influence mainly the serotonergic and dopaminergic system, and have been shown to act

in a ‘For better and for worse’ manner consistently in the literature (Belsky & Pluess, 2009;

Belsky et al., 2009). In the study described in chapter IV, there was no evidence to suggest

that either the cumulative plasticity index or the individual polymorphisms modified the bidi-

rectional relationship between PA and depressive symptoms prospectively.

This is contrary to findings from three recent studies. Two studies on the serotonin trans-

porter polymorphism (5-HTTLPR; Rethorst, Landers, Nagoshi, & Ross, 2010 and Rethorst,

Landers, Nagoshi, & Ross, 2011) concluded that carriers of the short allele of 5-HTTLPR had

a larger reduction in depressive symptoms following exercise than carriers of the long allele.

Another study, which concerned the BDNF val/met polymorphism, showed that carriers of

the met allele that exercised more exhibited fewer depressive symptoms than carriers of the

val allele (Mata, Thompson, & Gotlib, 2010). Both were cross-sectional studies conducted

in adults with relatively small sample sizes, and investigated the role of these genes in

the unidirectional effect of PA on depressive symptoms. In contrast, the study in chapter IV

was prospective and investigated the moderating effects of these genes on the bidirectional

relationships between PA and depressive symptoms in a large population sample of adoles-

cents. Therefore, differences in the design and the population sample might account for the

discrepant findings.

Mixed findings concerning the 5-HTTLPR gene and especially the gene-environment inter-

action of the short allele on depression are ubiquitous in the literature (Caspi et al., 2003;

CHAPTER VIII158

Karg, Burmeister, Shedden, & Sen, 2011; Munafo, Durrant, Lewis, & Flint, 2009; Risch et al.,

2009). Therefore, the influence of the short-allele of the 5-HTTLPR as a moderator of the

relationship between PA and depressive symptoms remains inconclusive. The discrepant

findings concerning the BDNF gene are more surprising, considering the large body of re-

search showing that BDNF levels are altered by PA (Toups et al., 2011; van Praag et al., 1999;

van Praag, 2008) and changes in BDNF levels are implicated with depression47 (for a review

see Hashimoto, 2010). It is possible that this discrepant finding is due to methodological

limitations of our study: since the number of participants homozygous to the met allele was

small, we merged this group with the individuals carrying the met/val alleles, which may

have introduced bias in the results.

The BDNF and other plasticity genes have numerous polymorphisms, of which only a small

selection were investigated in our study. Investigation of additional polymorphisms may help

to identify genetically based subgroups of individuals that will benefit from PA with regard to

their depressive symptoms48. However, considering the difficulties and inconsistencies with

candidate genes replications in MDD (Flint & Kendler, 2014), among other disorders, this is

speculative. Depressive disorders are multifactorial and due to the inability to identify clear

candidate genes for these ailments, we cannot exclude the possibility that multiple genes

and their interactions influence their onset and trajectory. This polygenic effect of multiple

loci of genes interacting has been recently investigated in several psychopathologies, in-

cluding MDD, and yielded some promising results (Peyrot et al., 2014; Smoller et al., 2013).

Future research is needed to elucidate the complex interactions between genes and their

influence on behavior, but there are reasons for optimism in this line of research.

Psychosocial Modifiers

Two possible psychosocial modifiers of the relationship between PA and depressive symp-

toms have been explored, namely sport competence as perceived by peer evaluations and

classroom positioning according to the month of birth.

PA , Spor t Par ticipation a nd Spor t Co m petence

The competitive element of sport participation has been argued to have an evolutionary ori-

gin (Lombardo, 2012) for several reasons. First, sports resemble primitive hunting and war-

fare of our ancestors. Second, sports are universal and can be found in almost every human

group throughout history. Finally, athletic competitions play a very influential role in society.

Lombardo argued that the competitive element of sports influences hierarchy formations

in males. This hierarchy formation will in turn affect the chances of males being selected by

females as sexual partners (Lombardo, 2012).

159DISCUSSION

In turn, this hierarchy formation has been linked to depression49. Price et al. (1994) suggest-

ed in their Social Competition Hypothesis that depression might be an adaptive strategy,

similar to hibernation in some animals50. Hence, individuals who are likely to lose in ritual

agonistic competitions may develop depression in order to be protected from further dam-

age and loss of their position within a social group51. In other words, the Social Competition

Hypothesis predicts that lower status individuals are more likely to suffer from depression.

The study in chapter V investigated whether the Social Competition Hypothesis extends to

sport participation and depressive symptoms. Overall, there was a positive association be-

tween adolescents who participate in competitive sports and social status, supporting the

hypothesis that sport participation acts as a hierarchy formation in adolescents52. This is

consistent with other studies showing that sport participation, especially in boys, is impor-

tant for adolescent hierarchy formation (Gill, 1992). Moreover, sport competence was also

related to concurrent depressive symptoms, as was suggested by Price et al. (1994), i.e., rel-

atively incompetent individuals exhibited more depressive symptoms53. Sport competence

explained the observed relationship between sport participation and depressive symptoms,

which indicates that sport competence may have a larger influence on depressive symp-

toms than participation in sports. There was no evidence that sport participation and sport

competence were related to changes in depressive symptoms. Furthermore, boys who were

considered to be highly sport competent did not benefit more from sport participation than

girls or less competent boys.

PA has had an important role in the evolution of humans, thus any beneficial effects of PA

on depressive symptoms could be put into a larger evolutionary theoretical framework. Our

study supported some predictions based on these evolutionary perspectives, but not the

whole postulated relationship between PA, sport participation and depressive symptoms. It

is important to investigate the relationship between PA or sport participation and depres-

sive symptoms from an evolutionary perspective, because through this larger framework

important insights can be achieved. Evolutionary perspectives offer the possibility for new

hypotheses to be tested, which can lead to a better understanding of the relationship be-

tween PA and mental health.

Classroo m Position in g a nd Relative-Age Ef fects

Children are allocated into classrooms according to their date of birth54. This might result

in classrooms encompassing children of up to a year older being in the same grade with

relatively younger children. Previous research has shown relative unfavorable outcomes

in certain domains for the relatively young compared to the relatively old children (see

for example: Bedard & Dhuey, 2006 and Morrow et al., 2012). These favorable or unfavor-

able outcomes might be a result of the evaluation of adults or their peers. Relative older

CHAPTER VIII160

adolescents will be slightly taller, stronger and physically more mature, which would result

in a higher likelihood to be selected by coaches and excel in sports (Gladwell, 2008)55. If being

perceived or being actually competent in sports influences social status (as was shown, to

some extent, in chapter V) then the relative older adolescents might report fewer depressive

symptoms through this attainment of higher status (as shown in chapter V as well). Thus,

the relative-age effects might moderate the relationship between PA and depressive symp-

toms. Although not directly explored here, the possibility that relative-age effects influence

hierarchy formations and so result in better opportunities for the relative older individuals

appears to be a promising line for future research. The findings in chapter VI indicated that

there are no relative-age effects (favorable or unfavorable) in adolescents with a norma-

tive school progress. However, within adolescents who repeated a grade, the relative young

ones reported fewer depressive symptoms than their relatively older peers. These relative-

age effects did not translate into differences in the evaluation of sport competence by teach-

ers, in either group. The possibility that these relative-age effects influence peer perceptions

of sport competence was not considered. Chapter V showed a difference in the perceptions

of sport competence between teachers and peers. Hence, it would be interesting to explore

how peer perceptions might be affected by relative-age differences.

In sum, most of the moderation analyses performed in the studies in this thesis have not

yielded clear factors that might help identify individuals that can benefit or be harmed by PA,

regarding depressive symptomatology. The study in chapter VII, however, showed individual

differences in the relationship between PA and affect, which still suggests that there might

be common factors between individuals that influence whether PA can be beneficial for men-

tal health. Future research using a combination of within and between subject designs might

prove imperative in identifying these common factors. The elucidation of the reasons why

some individuals benefit from PA while others do not, might prove helpful in the develop-

ment of individual-targeted PA interventions that will be effective in treating depression.

Strengths & Limitations

The longitudinal design and the large sample sizes used in most chapters is an important

strength of this thesis. In most cases, sample sizes of more than 1000 adolescents were used,

which provided adequate power to detect relatively small effect sizes. Additionally, we used two

or more measurement waves, covering a relatively long period of time. This facilitates the exclu-

sion of time-invariant unobserved individual differences and informs of the temporal order of

events, thus making the exploration of the direction of the effects under investigation possible.

There are also limitations that need to be acknowledged. In most studies in this thesis,

self-report measures of PA and depressive symptoms were used, with different degrees of

161DISCUSSION

specificity. Self-reports typically depend on the memory of participants, who estimate their

depressive symptoms or PA based on an average week. The affective scale of the Youth Self-

Report, which assesses depressive symptoms corresponding to DSM-IV criteria (Achenbach

& Rescorla, 2001), is an empirically based and well-validated questionnaire with an accept-

able test-retest reliability (Achenbach, Dumenci, & Rescorla, 2003). However, the use of

self-reports to estimate PA has been criticized by researchers (see for example: Johnson &

Taliaferro, 2011), since these might be prone to over or under-estimation of the levels of PA

engagement. Although acceptable test-retest reliability was shown in independent samples

(Booth, Okely, Chey, & Bauman, 2001; Haskell, 2012)56, recent reviews comparing self-re-

ports to more direct measures of PA, such as accelerometers, revealed only a moderate to

weak correlation between the two (Adamo, Prince, Tricco, Connor-Gorber, & Tremblay, 2009;

Prince et al., 2008). For example, Adamo et al. (2009) observed that self-reports underes-

timated light or moderate PA, while vigorous PA was consistently over-estimated, when

compared to more direct measures in children and adolescents. Prince et al. (2008) found

that studies using self-reports generally overestimated the levels of PA participation when

compared to direct measures, especially for women. Another limitation is that, throughout

this thesis, the relationship between PA and depression was modeled using linear regression

techniques. It is possible that PA is beneficial up to a certain point (curvilinear relationship),

while exceeding this threshold results in adverse mental health consequences, as studies on

the negative effects of overtraining on mood have shown (Bresciani et al., 2011; Armstrong &

VanHeest, 2002). In most chapters of this thesis, the effects of duration and intensity of PA

were not considered. It is possible that only individuals who engage at a specific frequency,

duration and intensity of PA experience fewer depressive symptoms (Dunn, Trivedi, Kam-

pert, Clark, & Chambliss, 2005; Goldfield et al., 2011; Sanders, Field, Diego, & Kaplan, 2000).

A study in Chinese adolescents (Tao et al., 2007), for example, showed that only low to mod-

erate PA intensity but not high intensity was related to depressive symptoms. Therefore,

it is possible that the different intensities of PA might have a differential association with

depression. However, in the study in chapter III we examined the dose-response effect of PA

and found no evidence that different frequencies, durations and intensities of activities pro-

tected against an onset of depression. To conclude, it is still possible that the different dura-

tions and intensities of activities differentially relate to depressive symptoms, but based on

the results in this thesis, this seems unlikely.

Clinical Relevance

Throughout this thesis, the effect sizes observed were small. Still, weak effects observed

in population studies (epidemiology) do not necessarily imply that the use of PA or exercise

in a clinical trial cannot alleviate such symptoms. Findings from population studies cannot

CHAPTER VIII162

be used to justify exercise as a treatment without relying on randomized controlled trials

(RCTs), regardless of the strength of the relationship observed in these studies. Even small

effect sizes found in population studies can translate to substantial clinical benefits in cer-

tain individuals. To illustrate this point, Rosenthal showed that prescribing aspirin resulted

in a substantial decrease in heart attacks (85 out of approximately 11000 patients), despite

the estimated effect being weak (r = -.03; Rosenthal, 1990)57.

Concerning the relationship between PA and depression, initially promising results (in both

observational and intervention studies) have been prematurely echoed and exaggerated58,

mainly because of the elegant and intuitively appealing hypothesis that PA might be used

as a treatment for depression. Some support for this hypothesis was observed in the study

in chapter VII59. The results suggested that, although higher levels of PA resulted in lower

negative affect in a few subjects60, there were also individuals in whom high levels of nega-

tive affect were associated with more activity. This may indicate that for some individuals PA

may be an efficient strategy to alleviate negative mood. However, a recent high quality RCT

(Chalder et al., 2012) and meta-analyses of RCTs (Cooney et al., 2013; Mead et al., 2009) have

not found sufficient support for this hypothesis at the group level61.

Finally, the results in chapter VII indicate that PA has a direct effect on elevating positive

mood for the majority of individuals, regardless of whether they are depressed or not, but

the delayed effect of PA on positive mood seems to be much less consistent. The reasons

for a short-term effect of PA on positive affect, and for why this does not carry over to longer

time periods, are currently not understood. Cogent and elegant hypotheses implicating bio-

logical factors such as the endorphin system62 (Dishman & O’Connor, 2009; Thoren, Floras,

Hoffmann, & Seals, 1990) have been proposed, but future research is needed to investigate

these mechanisms in further detail. By identifying the reasons why these benefits do not

carry over the longer term, clinicians might be able to manipulate factors and identify spe-

cific individuals that might be helped by PA in terms of their depressive symptoms. Until a

better understanding is achieved, however, any conclusions about this relationship should

be tentative and conservative until further studies demonstrate substantial clinical benefits

that are specific to PA.

Concluding Remarks

The relationship between PA and depressive symptoms has been investigated extensively

over the last few years. Initial studies bore the promise that PA might reduce depressive

symptoms, which would make it a very elegant and potentially cost-effective treatment

for depression. Major organizations such as the WHO have implemented exercise referral

schemes for the treatment of depression. However, some authors have questioned the cau-

163DISCUSSION

sality of the relationship between PA and depressive symptoms (Birkeland et al., 2009; de

Geus & de Moor, 2008; de Moor et al., 2008), while others have argued that the protective

effects of exercise might have been overestimated (Daley & Jolly, 2012).

This thesis illustrates that the relationship between PA and depressive symptoms is bi-

directional, not robust and more complex than what was previously envisioned. This sug-

gests that PA might not be directly related to mental health but instead represents a small

fraction in a large dynamic network of factors interacting with each other. Additionally, this

thesis demonstrated large inter-individual differences in the relationship between PA, affect

and depressive symptoms, where some individuals benefited from PA, while others did not.

It was also evident that for some individuals depressed mood hindered them from being

physically active, while for others this negative mood motivated them to ‘walk it off’.

Future research should focus on delineating other factors that influence the relationship

between PA and depressive symptoms, in order to accurately identify individuals that might

benefit from PA. This accurate identification of individuals will hopefully help to devise per-

sonal exercise interventions for treating depression.

Notes30 This study has many differences from all other studies included in this thesis and therefore has been in-

cluded in a separate section. However, because this study explored the strength and the direction of the relationship between PA and affect in individuals, the findings will be discussed in parts of the discussion that are most relevant.

31 Affect is thought to be the most elementary form of a conscious feeling and is defined as “a neurophysi-ological state that is consciously accessible as a simple, non-reflective feeling that is an integral blend of hedonic (pleasure-displeasure) and arousal (sleepy-activated) values” (see: Russell, 2003, table 1). A prolonged core affect leads to what is termed mood. Affect is divided into positive affect and negative affect. Long lasting changes on affect and especially increases in negative affect can lead to a depressed mood (Clark & Watson, 1991), and consequently a prolonged depressed mood can lead to further exacer-bation of depressive symptoms and subsequently to the development of depression. This study explored the influence of PA on positive and negative affect in depressed and non-depressed individuals because of the close connection between affect and depressed mood.

32 Lindwall et al. (2011) established that the direction of the relationship was stronger from PA to depres-sive symptoms than the other way around, but the model fit suggested that the relationship was bidi-rectional.

33 Numerous reasons could be behind this discrepancy, but most of these are speculative. For instance, it could be that Birkerland et al.'s results reflected a false negative. Perhaps the protective and inhibi-tion hypotheses are true only in adulthood and not in adolescence. It is also possible that the results of Birkerland et al. were due to a narrow definition of PA. Birkerland et al. focused on leisure-time PA, which, according to some authors, should be considered as exercise and not PA (Caspersen, Powell, & Christen-son, 1985). Finally, the relationship between PA and depressive symptoms might not be a causal one (this point will be elaborated further in later parts of this thesis).

CHAPTER VIII164

34 NA has also been implicated in depression (Thase, 2009).35 It should be noted that most antidepressants do not seem to directly impact DA levels. However, MAOIs,

buproprion and sertraline have been shown to have some effect on dopaminergic neuromodulation in animal models of depression (Thase, 2009; Willner, 1995; Willner, 1997). Moreover, in animals these anti-depressants countermand dysfunctions in DA systems, especially in conditions of chronic stress (Cuadra, Zurita, Gioino, & Molina, 2001; Nestler & Carlezon, 2006). Therefore, DA and its possible effect on depres-sion is of great interest for future research.

36 Similarly, a positive mood may result in higher energy levels and motivation, which can influence engage-ment in PA.

37 The causal hypothesis on the relationship between PA and depressive symptoms, with the protective and inhibition hypotheses as two specifications, predicts that increases in frequency and intensity of PA over time would reduce depressive symptoms over time, and conversely, that decreases in PA levels over time would increase depressive symptoms over time, through a causal chain.

38 With a later peak between the ages of 30 and 40 (see: http://www.health.am/psy/major-depressive-disorder/).

39 Cognitive symptoms subscale includes: lack of interest; feelings of worthlessness; loss of pleasure; cry-ing; self-harm; suicidal ideation; and feelings of guilt and sadness. The somatic symptoms subscale in-cludes: lack of appetite; overtiredness; trouble with sleeping and lack of energy.

40 This finding is counter-intuitive, because it would be expected that PA, if anything, improves sleep qual-ity, increases appetite and (unless overtraining) increases energy levels (Biddle & Mutrie, 2008). Fur-thermore, a recent study has shown that PA reduced somatic symptoms of depression in patients with chronic heart failure (Redwine et al., 2012).

41 Note here that the modifying analyses in this thesis explored this idea further, without conclusive find-ings. The study in chapter VII, however, showed individual differences in the relationship between PA and affect more clearly and these findings will be discussed in this section.

42 Between-subject designs are also called nomothetic approaches, while within-subject designs are also called idiographic approaches. Because the terms ‘nomothetic’ and ‘idiographic’ are not used in the origi-nally intended way (Lamiell, 1998), the terms between-subject and within-subject designs are used in-stead, in order to avoid any further confusion.

43 Which is a within-subject approach measuring the same individual over multiple measurement points. Although this approach cannot compare effects between individuals, it can elucidate the pattern of the relationship in individuals over time.

44 It should be noted here, that these results could be interpreted differently. It could be argued that the relationship between PA and negative affect was rather homogeneous, since the majority of participants did not show an effect of PA on negative affect or vice versa. However, the fact that some individu-als showed positive while others showed negative associations between PA and negative affect, even though a minority, should make it clear that there is some heterogeneity in the association between these two factors.

45 Individuals with specific genotypes (diathesis) might be vulnerable to developing mental health problems after important negative life events (stressors) have occurred.

46 This is referred to as ‘For better and for worse’.47 BDNF is targeted by a variety of antidepressants (Hashimoto, 2010). In MDD and Bipolar Disorder (BPD)

studies, antidepressant medication increased the expression of BDNF mRNA in limbic structures in re-sponse to chronic treatment but not acute treatment (see for example: Hashimoto, 2010). Finally, the time that it takes to increase BDNF expression coincides with the time typically required for antidepres-sants to work in depressed patients (Hashimoto, 2010).

165DISCUSSION

48 Or conversely increase vulnerability to depression that in turn can affect PA engagement levels.49 This hypothesis is consistent with gender differences observed in depression and has been dealt with in

previous publications from the authors (Price, 1988). Briefly, they suggest that females still compete for ranking, but female competition is more inconspicuous and less common than male competition.

50 Notably, not all evolutionary perspectives converge on the idea that depression might be adaptive (Al-len & Badcock, 2006; Nettle, 2004). Some theories propose that personality traits, such as neuroticism, might be adaptive and therefore survived evolution. Individuals, who are on the far side of this affective reactivity distribution will suffer detrimental effects, i.e., develop depression (Nettle, 2004).

51 Throughout the early human evolutionary history, this loss would have likely resulted in physical harm, lower chances of reproduction or death.

52 And as such, supported the prediction made by the evolutionary perspective of sport participation (Lom-bardo, 2012), i.e., more athletically competent individuals will attain higher status.

53 Therefore also supporting one of the predictions of the Social Competition Hypothesis (Price et al., 1994) that lower status individuals will suffer more from depressive symptoms than higher status individuals.

54 There is considerable variation to this rule in different cultures (Bedard & Dhuey, 2006).55 Especially since in childhood and adolescence even small differences in age (a year or less) result in large

differences in physique and abilities.56 Even a single item is considered to provide an accurate representation of an individual’s engagement in

PA (Milton, Clemes, & Bull, 2012).57 This might seem as a rather small proportion of people being protected, but this clinical trial was termi-

nated prematurely, because the evidence that aspirin protected from heart-attacks was so overwhelm-ing that it was deemed unethical to continue giving placebo medication to the control group (see Rosen-thal, 1990).

58 There are intervention studies that show a benefit of PA on depressive symptoms (see for example: Blu-menthal et al., 1999 and Brown, Pearson, Braithwaite, Brown, & Biddle, 2013) but have been criticized for low quality methodological designs including: small samples, short follow-ups, poor randomization, no ad-equate control group and reliance on non-clinical volunteers (Daley & Jolly, 2012). Adding to these problems the possibility of publication bias, it is imperative to design intervention studies of high methodological quality in order to conclusively demonstrate that PA can be used as a treatment for depression.

59 Although not an intervention study.60 And therefore supporting (to an extent) the bidirectional relationship observed in prospective (observa-

tional) studies.61 Also see a review on EMA studies by Kanning et al. (2013).62 Briefly, this theory suggests that PA increases endorphin levels, resulting in what colloquially is termed

‘Runner’s high’. These changes have been shown to last only for a very brief period of time.

CHAPTER VIII166

REFERENCES

Achenbach, T. M., & Rescorla, L. A. (2001). Manual for the ASEBA school-age forms and profiles. Burlington, VT: University of Vermont, Research center for children, youth and families.

Achenbach, T. M., Dumenci, L., & Rescorla, L. A. (2003). DSM-oriented and empirically based approaches to constructing scales from the same item pools. Journal of Clinical Child and Adolescent Psychology, 32(3), 328-340.

Adamo, K. B., Prince, S. A., Tricco, A. C., Connor-Gorber, S., & Tremblay, M. (2009). A comparison of indirect versus direct measures for assessing physical activity in the pediatric population: A systematic review. International Journal of Pediatric Obesity, 4(1), 2-27.

Allen, N. B., & Badcock, P. B. T. (2006). Darwinian models of depression: A review of evolutionary accounts of mood and mood disorders. Progress in Neuropsychopharmacology & Biological Psychiatry, 30(5), 815-826.

Armstrong, L. E., & VanHeest, J. L. (2002). The unknown mechanism of the overtraining syndrome - clues from depression and psychoneuroimmunology. Sports Medicine, 32(3), 185-209.

Bedard, K., & Dhuey, E. (2006). The persistence of early childhood maturity: International evidence of long-run age effects. Quarterly Journal of Economics, 121(4), 1437-1472.

Belsky, J., & Beaver, K. M. (2011). Cumulative-genetic plasticity, parenting and adolescent self-regulation. Journal of Child Psychology and Psychiatry, and Allied Disciplines, 52(5), 619-626.

Belsky, J., Jonassaint, C., Pluess, M., Stanton, M., Brummett, B., & Williams, R. (2009). Vulnerability genes or plasticity genes? Molecular Psychiatry, 14(8), 746-754.

Belsky, J., & Pluess, M. (2009). Beyond diathesis stress: Differential susceptibility to environmental influ-ences. Psychological Bulletin, 135(6), 885-908.

Biddle, S. J. H., & Mutrie, N. (2008). Psychology of physical activity: Determinants, well-being and interven-tions (2nd ed.). New York, USA: Routledge.

Birkeland, M. S., Torsheim, T., & Wold, B. (2009). A longitudinal study of the relationship between leisure-time physical activity and depressed mood among adolescents. Psychology of Sport and Exercise, 10(1), 25-34.

Bliss, E. L., & Ailion, J. (1971). Relationship of stress and activity to brain dopamine and homovanillic acid. Life Science Part 1 Physiology & Pharmacology, 10(20), 1161-9.

Blumenthal, J. A., Babyak, M. A., Moore, K. A., Craighead, W. E., Herman, S., Khatri, P., et al. (1999). Effects of exercise training on older patients with major depression. Archives of Internal Medicine, 159(19), 2349-2356.

Boecker, H., Sprenger, T., Spilker, M. E., Henriksen, G., Koppenhoefer, M., Wagner, K. J., et al. (2008). The run-ner’s high: Opioidergic mechanisms in the human brain. Cerebral Cortex, 18(11), 2523-2531.

Booth, M. L., Okely, A. D., Chey, T., & Bauman, A. (2001). The reliability and validity of the physical activ-ity questions in the WHO health behaviour in schoolchildren (HBSC) survey: A population study. British Journal of Sports Medicine, 35(4), 263-267.

Bresciani, G., Cuevas, M. J., Molinero, O., Almar, M., Suay, F., Salvador, A., et al. (2011). Signs of overload after an intensified training. International Journal of Sports Medicine, 32(5), 338-343.

Brown, H. E., Pearson, N., Braithwaite, R. E., Brown, W. J., & Biddle, S. J. H. (2013). Physical activity inter-ventions and depression in children and adolescents: A systematic review and meta-analysis. Sports Medicine, 43(3), 195-206.


Caspi, A., Sugden, K., Moffitt, T. E., Taylor, A., Craig, I. W., Harrington, H., et al. (2003). Influence of life stress on depression: Moderation by a polymorphism in the 5-HTT gene. Science, 301(5631), 386-389.

167DISCUSSION

Chalder, M., Wiles, N. J., Campbell, J., Hollinghurst, S. P., Searle, A., Haase, A. M., et al. (2012). A pragmatic randomised controlled trial to evaluate the cost-effectiveness of a physical activity intervention as a treatment for depression: The treating depression with physical activity (TREAD) trial. Health Technol-ogy Assessment, 16(10), 1-+.

Clark, L. A., & Watson, D. (1991). Tripartite model of anxiety and depression - psychometric evidence and taxonomic implications. Journal of Abnormal Psychology, 100(3), 316-336.


Cowen, P. J. (2008). Serotonin and depression: Pathophysiological mechanism or marketing myth? Trends in Pharmacological Sciences, 29(9), 433-436.

Cuadra, G., Zurita, A., Gioino, G., & Molina, V. (2001). Influence of different antidepressant drugs on the effect of chronic variable stress on restraint-induced dopamine release in frontal cortex. Neuropsychopharma-cology, 25(3), 384-394.

Daley, A., & Jolly, K. (2012). Exercise to treat depression. British Medical Journal (Clinical Research Ed.), 344, e3181.

de Geus, E. J. C. & de Moor, M. H. M. (2008). A genetic perspective on the association between exercise and mental health. Mental Health and Physical Activity, 1, 53-61.

de Jonge, P., & Bos, E. H. (2013). Tobacco smoking and depression: Time to move on to a new research para-digm in medicine? BMC Medicine, 11, 138.

de Moor, M. H. M., Boomsma, D. I., Stubbe, J. H., Willemsen, G., & de Geus, E. J. C. (2008). Testing causality in the association between regular exercise and symptoms of anxiety and depression. Archives of General Psychiatry, 65(8), 897-905.

Deci, E. L., & Flaste, R. (1995). Why we do what we do: Understanding self-motivation. New York: Penguin.

Deci, E. L., & Ryan, R. M. (1985). Intrinsic motivation and self-determination in human behavior. New York, USA: Plenum Press.

Delgado, P. L. (2000). Depression: The case for a monoamine deficiency. Journal of Clinical Psychiatry, 61, 7-11.

Dishman, R. K., Berthoud, H. R., Booth, F. W., Cotman, C. W., Edgerton, V. R., Fleshner, M. R., et al. (2006a). Neurobiology of exercise. Obesity, 14(3), 345-356.

Dishman, R. K., Hales, D. P., Pfeiffer, K. A., Felton, G., Saunders, R., Ward, D. S., et al. (2006b). Physical self-concept and self-esteem mediate cross-sectional relations of physical activity and sport participation with depression symptoms among adolescent girls. Health Psychology, 25(3), 396-407.

Dishman, R. K., & O’Connor, P. J. (2009). Lessons in exercise neurobiology: The case of endorphins. Mental Health and Physical Activity, 1(2), 4-9.

Dremencov, E., Gispan-Herman, I., Rosenstein, M., Mendelman, A., Overstreet, D. H., Zohar, J., et al. (2004). The serotonin-dopamine interaction is critical for fast-onset action of antidepressant treatment: In vivo studies in an animal model of depression. Progress in Neuropsychopharmacology & Biological Psychia-try, 28(1), 141-147.

Drevets, W. C., Thase, M. E., Moses-Kolko, E. L., Price, J. S., Frank, E., Kupfer, D. J., et al. (2007). Serotonin-1A receptor imaging in recurrent depression: Replication and literature review. Nuclear Medicine and Biol-ogy, 34(7), 865-877.

Dunlop, B. W., & Nemeroff, C. B. (2007). The role of dopamine in the pathophysiology of depression. Archives of General Psychiatry, 64(3), 327-337.


CHAPTER VIII168

Ellis, B. J., Boyce, W. T., Belsky, J., Bakermans-Kranenburg, M. J., & van Ijzendoorn, M. H. (2011). Differential susceptibility to the environment: An evolutionary–neurodevelopmental theory. Development and Psy-chopathology, 23(1), 7-28.

Flint, J., & Kendler, K. S. (2014). The genetics of major depression. Neuron, 81(3), 484-503. doi: 10.1016/ j.Neuron.2014.01.027.Epub 2014 Feb 5

Georgiades, K., Lewinsohn, P. M., Monroe, S. M., & Seeley, J. R. (2006). Major depressive disorder in adoles-cence: The role of subthreshold symptoms. Journal of the American Academy of Child and Adolescent Psychiatry, 45(8), 936-944.

Gill, D. L. (1992). Gender and sport behaviour. Advances in sport psychology (pp. 143-160). Champaign, IL: Human Kinetics.

Gladwell, M. (2008). Outliers: The story of success. USA: Little, Brown and Company.


Goodwin, R. D. (2003). Association between physical activity and mental disorders among adults in the United States. Preventive Medicine, 36(6), 698-703.

Hamaker, E. L. (2012). Why researchers should think “within-person”. In M. Mehl, & T. Conner (Eds.), Hand-book of research methods for studying daily life (pp. 43-61). New York: The Guildford Press.

Hamaker, E. L., Dolan, C. V., & Molenaar, P. C. M. (2005). Statistical modeling of the individual: Rationale and application of multivariate stationary time series analysis. Multivariate Behavioral Research, 40(2), 207-233.

Hashimoto, K. (2010). Brain-derived neurotrophic factor as a biomarker for mood disorders: An historical overview and future directions. Psychiatry and Clinical Neurosciences, 64(4), 341-357.

Haskell, W. L. (2012). Physical activity by self-report: A brief history and future issues. Journal of Physical Activity & Health, 9 Suppl 1, S5-10.

Helmich, I., Latini, A., Sigwalt, A., Carta, M. G., Machado, S., Velasques, B., et al. (2010). Neurobiological alterations induced by exercise and their impact on depressive disorders. Clinical Practice and Epidemi-ology in Mental Health: CP & EMH, 6, 115-25.

Hirvonen, J., Karlsson, H., Kajander, J., Lepola, A., Markkula, J., Rasi-Hakala, H., et al. (2008). Decreased brain serotonin 5-HT1A receptor availability in medication-naive patients with major depressive disorder: An in-vivo imaging study using PET and [carbonyl-C-11]WAY-100635. International Journal of Neuropsy-chopharmacology, 11(4), 465-476.

Jacka, F. N., Pasco, J. A., Williams, L. J., Leslie, E. R., Dodd, S., Nicholson, G. C., et al. (2011). Lower levels of physical activity in childhood associated with adult depression. Journal of Science and Medicine in Sport / Sports Medicine Australia, 14(3), 222-226.



Kanning, M. K., Ebner-Priemer, U. W., & Schlicht, W. M. (2013). How to investigate within-subject associa-tions between physical activity and momentary affective states in everyday life: A position statement based on a literature overview. Frontiers in Psychology, 4, 187.

Karg, K., Burmeister, M., Shedden, K., & Sen, S. (2011). The serotonin transporter promoter variant (5-HT-TLPR), stress, and depression meta-analysis revisited: Evidence of genetic moderation. Archives of Gen-eral Psychiatry, 68(5), 444-454.

169DISCUSSION

Kessler, R. C., & Wang, P. S. (2009). Epidemiology of depression. In I. H. Gotlib, & C. L. Hammen (Eds.), Hand-book of depression (2nd ed., pp. 5-22). New York: The Guildford Press.

Lamiell, J. T. (1998). ‘Nomothetic’ and ‘idiographic’ - contrasting Windelband’s understanding with contem-porary usage. Theory & Psychology, 8(1), 23-38.

Lapin, I. P., & Oxenkrug, G. F. (1969). Intensification of the central serotoninergic processes as a possible determinant of the thymoleptic effect. Lancet, 1(7586), 132-136.

Leo, J., & Lacasse, J. R. (2008). The media and the chemical imbalance theory of depression. Society, 45(1), 35-45.

Lewinsohn, P. M., Solomon, A., Seeley, J. R., & Zeiss, A. (2000). Clinical implications of “subthreshold” de-pressive symptoms. Journal of Abnormal Psychology, 109(2), 345-351.


Lombardo, M. P. (2012). On the evolution of sport. Evolutionary Psychology, 10(1), 1-28.

Luscher, B., Shen, Q., & Sahir, N. (2011). The GABAergic deficit hypothesis of major depressive disorder. Mo-lecular Psychiatry, 16(4), 383-406.

Mammen, G., & Faulkner, G. (2013). Physical activity and the prevention of depression: A systematic re-view of prospective studies. American Journal of Preventive Medicine, 45 (5), 649-57. doi:10.1016/j.amepre.2013.08.001

Mata, J., Thompson, R. J., & Gotlib, I. H. (2010). BDNF genotype moderates the relation between physical activity and depressive symptoms. Health Psychology, 29(2), 130-133.

McGale, N., McArdle, S., & Gaffney, P. (2011). Exploring the effectiveness of an integrated exercise/CBT inter-vention for young men’s mental health. British Journal of Health Psychology, 16, 457-471.


Milton, K., Clemes, S., & Bull, F. (2012). Can a single question provide an accurate measure of physical activ-ity? British Journal of Sports Medicine, 47(1), 44-8.

Molenaar, P. C. M., & Campbell, C. G. (2009). The new person-specific paradigm in psychology. Current Direc-tions in Psychological Science, 18(2), 112-117.

Morrow, R. L., Garland, E. J., Wright, J. M., Maclure, M., Taylor, S., & Dormuth, C. R. (2012). Influence of relative age on diagnosis and treatment of attention-deficit/hyperactivity disorder in children. Canadian, Medical Association Journal, 184(7), 755-62.

Munafo, M. R., Durrant, C., Lewis, G., & Flint, J. (2009). Gene X environment interactions at the serotonin transporter locus. Biological Psychiatry, 65(3), 211-219.

Nestler, E. J., & Carlezon, W. A. (2006). The mesolimbic dopamine reward circuit in depression. Biological Psychiatry, 59(12), 1151-1159.

Nettle, D. (2004). Evolutionary origins of depression: A review and reformulation. Journal of Affective Dis-orders, 81(2), 91-102.

Peyrot, W. J., Milaneschi, Y., Abdellaoui, A., Sullivan, P. F., Hottenga, J. J., Boomsma, D. I., et al. (2014). Effect of polygenic risk scores on depression in childhood trauma. British Journal of Psychiatry, 205(2), 113-9.

Ploughman, M. (2008). Exercise is brain food: The effects of physical activity on cognitive function. Develop-mental Neurorehabilitation, 11(3), 236-240.

Price, J. S. (1988). Alternative channels for negotiating asymmetry in social relationships. In M. R. A. Chance (Ed.), Social fabrics of the mind (pp. 157-195). Erlbaum.

CHAPTER VIII170

Price, J. S., Sloman, L., Gardner, R., Gilbert, P., & Rohde, P. (1994). The social competition hypothesis of de-pression. British Journal of Psychiatry, 164, 309-315.

Prince, S. A., Adamo, K. B., Hamel, M. E., Hardt, J., Gorber, S. C., & Tremblay, M. (2008). A comparison of direct versus self-report measures for assessing physical activity in adults: A systematic review. International Journal of Behavioral Nutrition and Physical Activity, 5, 56.

Redwine, L. S., Tsuang, M., Rusiewicz, A., Pandzic, I., Cammarata, S., Rutledge, T., et al. (2012). A pilot study exploring the effects of a 12-week T’ai chi intervention on somatic symptoms of depression in patients with heart failure. Journal of Alternative and Complementary Medicine, 18(8), 744-748.

Rethorst, C. D., Landers, D. M., Nagoshi, C. T., & Ross, J. T. (2010). Efficacy of exercise in reducing depressive symptoms across 5-HTTLPR genotypes. Medicine and Science in Sports and Exercise, 42(11), 2141-2147.

Rethorst, C. D., Landers, D. M., Nagoshi, C. T., & Ross, J. T. (2011). The association of 5-HTTLPR genotype and depressive symptoms is moderated by physical activity. Journal of Psychiatric Research, 45(2), 185-189.

Risch, N., Herrell, R., Lehner, T., Liang, K. Y., Eaves, L., Hoh, J., et al. (2009). Interaction between the sero-tonin transporter gene (5-HTTLPR), stressful life events, and risk of depression: A meta-analysis. JAMA: The Journal of the American Medical Association, 301(23), 2462-2471.

Rosenthal, R. (1990). How are we doing in soft psychology. American Psychologist, 45(6), 775-777.

Roshanaei-Moghaddam, B., Katon, W. J., & Russo, J. (2009). The longitudinal effects of depression on physi-cal activity. General Hospital Psychiatry, 31(4), 306-315.

Rosmalen, J. G. M., Wenting, A. M., Roest, A. M., de Jonge, P., & Bos, E. H. (2012). Revealing causal hetero-geneity using time series analysis of ambulatory assessments: Application to the association between depression and physical activity after myocardial infarction. Psychosomatic Medicine, 74(4), 377-386.

Russell, J. A. (2003). Core affect and the psychological construction of emotion. Psychological Review, 110(1), 145-172.

Sallis, J. F., & Owen, N. (1999). Physical activity and behavioral medicine. Thousand Oaks: SAGE publi-cations, Inc.


Sanders, C. E., Field, T. M., Diego, M., & Kaplan, M. (2000). Moderate involvement in sports is related to lower depression levels among adolescents. Adolescence, 35(140), 793-797.

Sasaki-Adams, D. M., & Kelley, A. E. (2001). Serotonin-dopamine interactions in the control of conditioned reinforcement and motor behavior. Neuropsychopharmacology, 25(3), 440-452.

Smoller, J. W., Craddock, N., Kendler, K., Lee, P. H., Neale, B. M., Nurnberger, J. I., et al. (2013). Identifica-tion of risk loci with shared effects on five major psychiatric disorders: A genome-wide analysis. Lancet, 381(9875), 1371-1379.

Struder, H. K., & Weicker, H. (2001). Physiology and pathophysiology of the serotonergic system and its im-plications on mental and physical performance. Part II. International Journal of Sports Medicine, 22(7), 482-497.

Tao, F. B., Xu, M. L., Kim, S. D., Sun, Y., Su, P. Y., & Huang, K. (2007). Physical activity might not be the pro-tective factor for health risk behaviours and psychopathological symptoms in adolescents. Journal of Paediatrics and Child Health, 43(11), 762-767.

Thase, M. E. (2009). Neurobiological aspects of depression. In I. H. Gotlib, & C. L. Hammen (Eds.), Handbook of depression (2nd ed., pp. 187-217). New York: The Guildford Press.

Thompson-Coon, J., Boddy, K., Stein, K., Whear, R., et al. (2011). Does participating in physical activity in outdoor natural environments have a greater effect on physical and mental wellbeing than physical ac-tivity indoors? A systematic review. Environmental Science & Technology, 45(5), 1761-72.

171DISCUSSION

Thoren, P., Floras, J. S., Hoffmann, P., & Seals, D. R. (1990). Endorphins and exercise - physiological-mecha-nisms and clinical implications. Medicine and Science in Sports and Exercise, 22(4), 417-428.

Toups, M. S. P., Greer, T. L., Kurian, B. T., Grannemann, B. D., Carmody, T. J., Huebinger, R., et al. (2011). Effects of serum brain derived neurotrophic factor on exercise augmentation treatment of depression. Journal of Psychiatric Research, 45(10), 1301-1306.

van Praag, H. M., Christie, B. R., Sejnowski, T. J., & Gage, F. H. (1999). Running enhances neurogenesis, learn-ing, and long-term potentiation in mice. Proceedings of the National Academy of Sciences of the United States of America, 96(23), 13427-13431.

van Praag, H. M., Kempermann, G., & Gage, F. H. (1999). Running increases cell proliferation and neurogen-esis in the adult mouse dentate gyrus. Nature Neuroscience, 2(3), 266-270.

van Praag, H. M., & Korf, J. (1970). L-tryptophan in depression. Lancet, 2(7673), 612.

van Praag, H. M. (2008). Neurogenesis and exercise: Past and future directions. Neuromolecular Medicine, 10(2), 128-140.

Willner, P. (1995). Dopaminergic mechanisms in depression and mania. In F. E. Bloom, & D. J. Kupfer (Eds.), Psychopharmacology: The fourth generation of progress (pp. 614-620). New York: Raven Press.

Willner, P. (1997). The mesolimbic dopamine system as a target for rapid antidepressant action. Interna-tional Clinical Psychopharmacology, 12, S7-S14.

Summary in English

175SUMMARY IN ENGLISH

Depression is a global public health problem, mainly because of the relatively high lifetime

prevalence and the substantial disability of individuals afflicted with this disorder. In the gen-

eral population, an estimated one out of 20 people will suffer from major depression at any

given point in time. The causes of the disorder include a complex interaction of psychological,

biological and behavioral factors, which makes depression difficult to treat. The effective-

ness of current interventions to treat depression, such as cognitive behavioral therapy and

antidepressant medication, are intensely debated. In the last few decades, research has

focused on the potential benefits of physical activity in treating depression. Early studies

suggested that physical activity might be effective in alleviating depressive symptoms, thus

providing new grounds for optimism and potentially offering a valuable intervention for the

treatment of depression. However, the evidential base for this optimism is not very strong,

especially concerning depressive symptoms in adolescents.

The aim of this dissertation was to explore in depth the relationship between physical

activity and depressive symptoms in adolescents and adults. A better understanding of

the direction and strength of this relationship and identification of possible moderators

may contribute to an effective prevention and treatment of depression. The research de-

scribed in this thesis was part of the studies Tracking Adolescents’ Individual Lives Survey

(TRAILS) and Mood and Movement in Daily Life (MOOVD). TRAILS has followed a large

cohort of Dutch adolescents from early adolescence up until adulthood. MOOVD examines

the temporal relationship between physical activity and mood in daily life in depressed and

non-depressed adults.

In chapter II, the direction of the relationship between physical activity and depressive

symptoms was explored and discussed. Previous research has primarily focused on the ef-

fects of physical activity on depressive symptoms, and largely ignored the potential effects

of depressive symptoms on physical activity. We observed a bidirectional prospective re-

lationship. Adolescents who were active exhibited fewer depressive symptoms over time

than more sedentary adolescents, and the opposite was also true, that is, adolescents with

many depressive symptoms became more sedentary over time than adolescents with fewer

depressive symptoms. Cognitive symptoms of depression such as loss of interest and de-

pressed mood were more likely to be associated with physical activity than somatic symp-

toms such as loss of appetite, sleep and energy.

The study described in chapter III investigated the protective hypothesis of physical ac-

tivity, which states that early engagement in physical activity might prevent the onset of

depression. This was explored regarding the period from adolescence to early adulthood,

which spanned approximately 6 years. Effects of specific characteristics of physical activity,

i.e., its nature, duration, frequency and intensity, were examined in the analysis. Physical

SUMMARY IN ENGLISH176

activity did not protect against a first episode of depression in this study, so the hypoth-

esis that early physical activity might prevent the onset of depression was not supported

empirically.

Chapter IV concerns the potential role of so-called plasticity genes on the (bidirectional)

relationship between physical activity and depressive symptoms. The term plasticity genes

refers to polymorphisms that have been reported to influence the actions of some neuro-

modulating and neurotrophin systems. These systems, among which are the serotonergic

and dopaminergic systems, can be altered by physical activity and depression. Overall, ado-

lescents with these polymorphisms assumed to reflect high plasticity did not differ from

other adolescents with regard to any of the associations between physical activity and de-

pressive symptoms. In other words, we found no evidence that the proposed genes directly

influence the relationship between physical activity and depressive symptoms.

The study in chapter V explored whether adolescents participating in competitive sports

reported fewer concurrent depressive symptoms and developed fewer depressive symp-

toms over time. In addition, we examined the influence of peer-perceived sport competence

and gender on the relationship between competitive sport participation and depressive

symptoms. Overall, both boys and girls who took part in competitive sports reported fewer

concurrent depressive symptoms, but did not differ from others with regard to symptom

changes over time. The association between sport participation and concurrent depressive

symptoms was largely explained by peer-perceived sport competence. This indicates that

athletically competent adolescents are less likely to exhibit depressive symptoms, regard-

less of their competitive sport participation.

Chapter VI concerns the effects of classroom positioning according to birth dates on multiple

outcomes, including depressive symptoms, depressive symptom changes during adoles-

cence, and teacher-rated sport competence. Substantial relative-age effects were observed

in relation to school progress: relatively young adolescents were four times more likely to

repeat a grade than the relatively old adolescents, while relatively old adolescents were up

to 20 times more likely to skip a grade. In the subgroup of adolescents who had repeated a

grade, the younger ones exhibited fewer depressive symptoms than relatively older peers

who had repeated a grade. Finally, the relative-age effects did not influence the evaluation

of sport competence in either the normative group or the subgroup that repeated a grade.

The final empirical chapter of the thesis (chapter VII) examined the relationship between

physical activity and affect in depressed and non-depressed adults, using a within-subject

approach. During a month, physical activity and affect (positive and negative) were mea-

sured three times a day by means of accelerometers and diaries. The results showed indi-

vidual differences in both the direction and the strength of the relationships. Cross-sectional

177SUMMARY IN ENGLISH

correlations between physical activity and positive affect were positive in nearly all individu-

als, though not always significant, which suggests that positive affect was enhanced during

and immediately after physical activity. No consistent improvement was observed in nega-

tive affect. The long-term effects of physical activity on positive and negative affect varied

greatly across individuals, regardless of their depression status. While for a few individuals

physical activity improved (positive) affect in the long term, for the majority of individuals

the relationship did not exist or was even reversed.

In conclusion, this thesis confirms the existence of a relationship between physical activity

and depressive symptoms, but it is rather weak, and shows substantial individual differ-

ences. Some individuals tend to feel better after having been physically active, while others

feel worse or show no relationship between mood and activity. The reasons behind these

differences are still unknown, and we do not know yet which individuals might benefit from

an active lifestyle. This knowledge is needed to design effective personalized interven-

tions for the treatment of depression. Up to this point, however, the evidence that physical

activity can help treat depressive symptoms is relatively weak, and additional research is

required before physical activity can be implemented as an effective strategy for the treat-

ment of depression.

Summary in Dutch

181SUMMARY IN DUTCH

Depressie is een wereldwijd gezondheidsprobleem. Dit komt voornamelijk doordat relatief

veel mensen een depressie doormaken in hun leven, maar ook omdat depressie gepaard

kan gaan met fors functieverlies. Naar schatting 1 op de 20 mensen krijgt gedurende de le-

vensloop een depressieve stoornis. De oorzaken van depressieve stoornissen omvatten een

complexe interactie van psychologische, biologische, en gedragsmatige factoren, hetgeen

depressie lastig te behandelen maakt. De effectiviteit van de beschikbare interventies voor

depressie, zoals cognitieve gedragstherapie en antidepressieve medicatie, zijn onderwerp

van hevig debat. Wetenschappers hebben zich de afgelopen decennia gericht op de poten-

tiële voordelen van lichamelijke beweging als alternatieve behandeling van depressie. Tot

dusver hebben enkele studies laten zien dat lichamelijke beweging effectief kan zijn in het

verminderen van depressieve symptomen en dus een waardevolle aanvulling kan zijn. Er is

echter nog weinig bewijs voor dit optimisme, in het bijzonder met betrekking tot depressieve

symptomen in de adolescentie.

Het doel van dit proefschrift is om de associatie tussen lichamelijke activiteit en depressieve

symptomen uit te diepen. Een beter begrip van de richting en sterkte van de associatie tus-

sen lichamelijke (in)activiteit en depressieve symptomen en identificatie van mogelijke mo-

deratoren kan bijdragen aan het voorkomen en behandelen van depressie. Het proefschrift

beschrijft onderzoek dat is uitgevoerd in de studies TRAILS (‘Tracking Adolescents’ Indivi-

dual Lives Survey’) en MOOVD (Mood and Movement in Daily Life). TRAILS volgt een groot

cohort van Nederlandse adolescenten van de vroege adolescentie tot in de volwassenheid.

MOOVD onderzoekt de temporele relatie tussen lichamelijke activiteit en stemming in het

dagelijks leven in depressieve en niet-depressieve volwassenen.

In hoofdstuk II wordt de richting van de verbanden tussen lichamelijke activiteit en depres-

sieve symptomen verkend en bediscussieerd. Eerder onderzoek was voornamelijk gericht

op de effecten van lichamelijke activiteit op depressieve symptomen, terwijl de potentiële

effecten van depressieve symptomen op lichamelijke activiteit grotendeels werden gene-

geerd. In onze studie werd een wederkerige relatie geobserveerd. Adolescenten die actief

waren ervoeren minder depressieve symptomen op het volgende meetmoment dan ado-

lescenten die minder bewogen, maar het tegenovergestelde was ook waar: adolescenten

met veel depressieve symptomen bewogen vervolgens minder dan adolescenten met minder

depressieve symptomen. Cognitieve symptomen van depressie, zoals verlies van interesse

en depressieve stemming, hingen sterker samen met lichamelijke activiteit dan lichamelijke

symptomen zoals verlies van eetlust, slaap en energie.

De studie beschreven in hoofdstuk III onderzocht de hypothese dat lichamelijke activiteit be-

schermt tegen het ontstaan van een depressie. Dit werd in kaart gebracht voor een ongeveer

6 jaar durende periode die liep van de adolescentie tot de vroege volwassenheid. Effecten

SUMMARY IN DUTCH182

van de aard, duur, frequentie en intensiteit van beweging werden nagegaan in de analyse.

Lichamelijke activiteit voorkwam een eerste episode van depressie niet in deze studie; de

hypothese dat vroege lichamelijke activiteit beschermt tegen het ontstaan van een depres-

sie werd dus niet ondersteund.

Hoofdstuk IV betreft de potentiële rol van de zogenaamde plasticiteitsgenen in de (weder-

kerige) relatie tussen lichamelijke activiteit en depressieve symptomen. De term plastici-

teitsgenen verwijst naar genvarianten waarvan bekend is dat ze de activiteit van bepaalde

neuromodulatoren en neurotropische systemen beïnvloeden. Deze systemen, waaronder

het serotonerge en dopaminerge systeem, kunnen worden beïnvloed door fysieke activi-

teit en depressie. Adolescenten met genvarianten waarvan verondersteld wordt dat ze een

hoge mate van plasticiteit weerspiegelen lieten geen sterkere (of zwakkere) verbanden

tussen depressieve symptomen en lichamelijke inactiviteit zien dan adolescenten zonder

deze varianten. Met andere woorden, er werd geen bewijs gevonden dat de voorgestelde

genen een directe invloed hadden op de associatie tussen lichamelijke activiteit en depres-

sieve symptomen.

De studie in hoofdstuk V onderzocht of adolescenten die deelnamen aan competitieve spor-

ten minder depressieve symptomen hadden en minder depressieve symptomen ontwikkel-

den in de loop der jaren. Daarnaast testten we de invloed van geslacht en sportcompetentie

volgens klasgenoten op de associatie tussen deelname aan competitieve sporten en de-

pressieve symptomen. Zowel jongens als meisjes die deelnamen aan competitieve sporten

rapporteerden minder depressieve symptomen, maar lieten geen verschil zien met betrek-

king tot toekomstige veranderingen in symptoomniveaus. De associatie tussen deelname

aan sport en depressieve symptomen werd grotendeels verklaard door sportcompetentie

zoals waargenomen door klasgenoten. Dit suggereert dat atletisch competente adolescen-

ten minder depressieve symptomen rapporteren, ongeacht hun deelname aan competitieve

sporten.

Hoofdstuk VI betreft de effecten van de relatieve leeftijd van adolescenten in hun schoolklas

op verschillende uitkomsten, waaronder depressieve symptomen, veranderingen in depres-

sieve symptomen tijdens de adolescentie en door de docent waargenomen sportcompeten-

tie. Substantiële relatieve leeftijdseffecten werden geobserveerd in relatie tot de kans om

zitten te blijven of een jaar over te slaan: adolescenten die relatief jong waren in hun klas

hadden een vier keer zo grote kans dat ze een jaar moesten overdoen dan hun relatief oudere

klasgenoten; terwijl de relatief oude adolescenten een 20 keer zo grote kans hadden dat ze

een klas mochten overslaan. In de subgroep van adolescenten die een jaar moest overdoen

hadden de relatief jonge adolescenten minder depressieve symptomen dan relatief oudere

leeftijdsgenoten die ook een jaar over moesten doen. De relatieve leeftijdseffecten hadden

183SUMMARY IN DUTCH

geen invloed op de waargenomen sport competentie in zowel de normatieve groep als de

subgroep die een jaar bleef zitten.

In het laatste empirische hoofdstuk van het proefschrift, hoofdstuk VII, werd de relatie tus-

sen lichamelijke activiteit en stemming in depressieve en niet-depressieve volwassenen

onderzocht door middel van een zogenaamde binnen-persoon benadering. Gedurende een

maand werden lichamelijke activiteit en affect (positief en negatief) driemaal daags gemeten

met een accelerometer en vragenlijsten. We vonden individuele verschillen in zowel de rich-

ting als de sterkte van de verbanden. Bijna alle individuen rapporteerden meer positief affect

tijdens of onmiddellijk na lichamelijke activiteit (positief cross-sectioneel verband), maar de

associatie was niet altijd significant. Er werd echter geen consistente verbetering in nega-

tief affect geobserveerd. De lange-termijn effecten van lichamelijke activiteit op positief en

negatief affect varieerde sterk tussen mensen, onafhankelijk van hun depressieve status.

Voor sommige personen voorspelde lichamelijke activiteit een verbetering van de stemming

over de tijd, maar voor de meerderheid van de deelnemers bestond dit verband niet of was

de relatie zelfs omgekeerd.

Samengevat bevestigt dit proefschrift het bestaan van een relatie tussen lichamelijke ac-

tiviteit en depressieve symptomen, maar het verband is zwak en vertoont forse individuele

verschillen. Sommige mensen voelen zich beter na lichamelijke inspanning, terwijl er voor

andere personen geen relatie bestaat tussen stemming en activiteit en weer anderen zich

juist slechter voelen na lichamelijke inspanning. De redenen voor deze verschillen zijn tot

dusver onbekend; en we weten nog niet welke mensen baat kunnen hebben van een actieve

leefstijl. Deze kennis is nodig om effectieve gepersonaliseerde interventies voor de behan-

deling van depressie te ontwikkelen. Tot dusver is het bewijs dat lichamelijke activiteit kan

helpen om depressieve symptomen te behandelen relatief zwak. Kortom, meer onderzoek is

nodig voordat lichamelijke activiteit kan worden geïmplementeerd als een effectieve strate-

gie voor de behandeling van depressie.

Acknowledgements

187ACKNOWLEDGEMENTS

The writing of this book has been a great adventure. I have always argued that the message

is more important than the messenger, however I have to thank all of the people that have

helped with the delivery of this message. I have to express my gratitude to Peter de Jonge

for giving me the opportunity to undertake this project. Without his keen eye for talent and

without him taking a risk with me, this book would have not materialized. I also want to

thank him for the support and guidance throughout these years. A special thank you to

Tineke Oldehinkel. Every new scientist and author should have the experience of working

with Tineke. Tineke’s insights, honesty, skeptical mind and amazing ability with details have

been a revelation for me. Tineke, thank you for your support and for being the best educator

I had over those years. I would also like to thank Annelieke Roest, who although came late

in the project, she has been a great guidance through patchy moments. Thank you Annelieke

for being an empathic individual who encouraged me and provided me with new perspectives

that I was not previously aware of.

I am very thankful to Liesbeth, Margo, Gerry and Martha (the lovely secretaries), for mak-

ing me feel like home in a foreign land. Without your cheerful personalities, incredible help

and guidance, life at the ICPE would have been very difficult for me and rather boring. Thank

you ladies.

I would like to thank Hans (Jans!) Ormel for all the critical comments and sharp insights. I

know Jans that you can make life very difficult for your collaborators, but life and science

would not be as much fun if they were easy. A big thank to Elske Bos for some truly great

statistical guidance and philosophical discussions. Keep dancing Elske! Dennis Raven, thank

you for being patient with me but also for your great guidance with the TRAILS dataset. I am

sure you will miss me appearing at your door with all those sticky notes.

I would like to thank Esther Nederhof, Richard Oude Voshaar, René Veenstra and Frank Ver-

hulst for their input throughout my project. I have enjoyed working with you, even if brief

in some cases, and hopefully have learned a few important skills from our collaborations.

Thanks to all volunteers who have participated in this project. A special thank you also to

Catharina Hartman, who although was not directly involved in this project, she always had

her door open for me with whatever I needed. Working with you and Xiaochen has taught me

a lot about myself, and what is needed to be a good teacher.

Many friends and relatives have been supportive during this project. Thank you Ma and Pa

(pappou!) for keeping me in line when facing tough times. Pappou: eagles should not be

feeding on flies, and you might be right. A special thank goes to my sister. Andriana, I am

extremely grateful for your time and input on this project. Jez, Kosta, Jan and Harm Jan (my

dear friends) thank you for your support, intriguing discussions and funny moments. Thanks

for being there for me whenever I needed you guys.

ACKNOWLEDGEMENTS188

Thanks to my roomies for five years. Sanne, Nynke and Bertus, three great young minds that

have helped me, supported me and inspired me to finish. Funny that I managed to finish first,

hey? Thank you guys for putting up with my absurdity, fidgety behavior and noisiness. Sanne

and Bertus it has been a great pleasure for me to collaborate with you on two papers of this

thesis. Nynke thank you for helping me with the student seminars and fruitful discussions

when I was stuck (damn statistics!). Above all, I thank you guys for being amazing friends.

Also thanks to the lovely girls next door, Arunima, Elise, Anna-Roos, Charlotte and Petra

for the sometimes absurd and mind opening discussions. Arunima thank you for giving me

feedback on early versions of the discussion of this thesis. Anna-Roos you have become a

great friend over the years, even though our initial moments were shaky (that crappy bike!).

I feel immense gratitude that you provided me with a house to stay for four months and

introduced me to some great friends of yours.

Finally, many thanks to the wonderful people that I have failed to mention here, but who

nonetheless played an important part in my life in Groningen.

Nikolaos

About the Author

191ABOUT THE AUTHOR

On the 23rd of June 1981 Nikolaos Stavrakakis was born in Athens, Greece. He finished sec-

ondary education at the International Baccalaureate Moraitis school in Athens in 1999. In

the same year he started his study in psychology at Reading University in the UK. After he

received his bachelor’s degree, Nikolaos undertook a master’s degree in Neuroscience in Im-

perial College in London in 2004. There he worked as an internship at the Cyclotron building

of Hammersmith hospital in London investigating emotional regulation and the serotonin

transporter gene. From the beginning of 2010 until august 2014 Nikolaos has worked on the

research presented in this dissertation. Currently, Nikolaos is living in Athens.

List of Publications

195LIST OF PUBLICATIONS

Stavrakakis N, de Jonge P, Ormel J, Oldehinkel AJ. “Bidirectional prospective associations

between physical activity and depressive symptoms. The TRAILS Study.” Journal of Adoles-

cent Health 2012; 50(5):503-8.

Stavrakakis N, Oldehinkel AJ, Nederhof E, Oude Voshaar RC, Verhulst FC, Ormel J, de Jonge

P. “Plasticity genes do not modify associations between physical activity and depressive

symptoms.” Health Psychology 2013; 32(7):785-92.

Stavrakakis N, Roest AM, Verhulst F, Ormel J, de Jonge P, Oldehinkel AJ. “Physical activ-

ity and onset of depression in adolescents: Α prospective study in the general population

cohort TRAILS.” Journal of Psychiatric Research 2013; 47(10):1304-8.

Rhodes RA, Murthy NV, Dresnet MA, Selvaraj S, Stavrakakis N, Babar S, Cowen PJ, Grasby

PM. “Human 5-HT transporter availability predicts amygdala reactivity in vivo.” Journal of

Neuroscience 2007; 27(34):9233-7.

Submitted or under preparation

Jeronimus BF, Stavrakakis N, Veenstra R, Oldehinkel AJ. “Relative Age Effects in Adoles-

cents: The TRAILS study” (under review).

Stavrakakis N, Roest AM, Verhulst FC, Veenstra R, Ormel J, de Jonge P, Oldehinkel AJ.

“Competitive sport participation, sport competence and depressive symptoms in adolescent

boys and girls” (under preparation).

Stavrakakis N, Booij SH, Roest AM, de Jonge P, Oldehinkel AJ, Bos EH. “The dynamic rela-

tionship between physical activity and mood in the daily life of depressed and non-depressed

individuals: a within-subject time-series analysis” (under preparation).

Xiaochen L, Stavrakakis N, Penninx BW, Bosker FJ, Nolen WA, Boomsma DI, de Geus EJC,

Smit JH, Snieder H, Nolte IM, Hartman CA. “Does refining the phenotype improve replication

rates? A candidate gene study on Major Depressive Disorder and Chronic Major Depressive

Disorder” (under preparation).

Recent TRAILS Dissertations

199RECENT TRAILS DISSERTATIONS

Some of the work described in this thesis is part of the Tracking Adolescents’ Individual

Lives Survey (TRAILS) and was performed at the Interdisciplinary Centre Psychopathology

and Emotion regulation (ICPE) of the University Medical Centre Groningen (UMCG), Univer-

sity of Groningen, the Netherlands. More information concerning TRAILS and the research

within can be found in the dedicated internetsite: www.trails.nl

Recent TRAILS theses

Mathyssek, CM (2014). The development of anxiety symptoms in adolescents. Supervisors:

Prof. dr. F.C. Verhulst, dr. F.V.A. van Oort.

Prince van Leeuwen, AL (2013). Blunt vulnerabilities. Identifying risks for initiation and con-

tinued use of cannabis in a Dutch adolescent population. Supervisors: Prof. dr. A.C. Huizink,

Prof. dr. F.C. Verhulst, dr. H.E. Creemers.

Verboom, CE (2012). Depression and role functioning. Their relation during adolescence and

adulthood. Supervisors: Prof. dr. J. Ormel, Prof. dr. W.A. Nolen, Prof. dr. B.W.J.H. Pennix, dr.

J.J. Sijtsema.

Marsman, H (2013). HPA-axis, genes and environmental factors in relation to externalizing

behaviors. Supervisors: Prof. dr. J.K. Buitelaar.

Griffith-Lendering, MFH (2013). Cannabis use, cognitive functioning and behaviour prob-

lems. Supervisors: Prof. dr. H. Swaab, dr. S.C.J. Huijbregts.

Vink, NM (2013). The role of stress in the etiology of asthma. Supervisors: Prof. dr. H.M.

Boezen, Prof. dr. J.G.M. Rosmalen, Prof. dr. D.S. Postma.

Laceulle, OM (2013). Programming effects of adversity on adolescent adaptive capacity. Su-

pervisors: Prof. dr. J. Ormel, Prof. dr. M.A.G. van Aken, dr. E. Nederhof.

Documents

University of Groningen Physical activity and depressive … · 2016. 3. 9. · Nikolaos Stavrakakis born on 23 June 1981 in Cholargos, Greece to obtain the degree of PhD at the University