TSpace Research Repository tspace.library.utoronto.ca

The longitudinal association between

temperament and physical activity in young children

Daphne J. Korczak MD MSc, Sheri Madigan PhD, Marlena Colasanto MSc, Peter Szatmari MD, Yang Chen MA, MSc,

Jonathon Maguire MD MSc, Patricia Parkin MD, Catherine S. Birken MD MSc

Version

Post-print/Accepted Manuscript

Citation (published version)

Korczak, Daphne J., Sheri Madigan, Marlena Colasanto, Peter Szatmari, Yang Chen, Jonathon Maguire, Patricia Parkin, and Catherine S. Birken. The longitudinal association between temperament and physical activity

in young children. Preventive medicine 111, 342-347 (2018). Doi: 10.1016/j.ypmed.2017.11.021

Copyright/License This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0

International License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

How to cite TSpace items

Always cite the published version, so the author(s) will receive recognition through services that track citation counts, e.g. Scopus. If you need to cite the page number of the author manuscript from TSpace because you cannot access the published version, then cite the TSpace version in addition to the published

version using the permanent URI (handle) found on the record page.

This article was made openly accessible by U of T Faculty. Please tell us how this access benefits you. Your story matters.

The longitudinal association between temperament and physical activity in young children

Daphne J. Korczak1 MD MSc, Sheri Madigan2 PhD, Marlena Colasanto3 MSc, Peter Szatmari4

MD, Yang Chen5 MA, MSc, Jonathon Maguire6 MD MSc, Patricia Parkin7 MD, Catherine S.

Birken8 MD MSc

1 Department of Psychiatry, Hospital for Sick Children, Toronto, ON, Canada; Department of Psychiatry,

Faculty of Medicine, University of Toronto, Canada; daphne.korczak@sickkids.ca 2 Department of Psychology, University of Calgary, Calgary, AB, Canada; Alberta Children’s Hospital

Research Institute, Calgary, AB, Canada; sheri.madigan@ucalgary.ca 3 Department of Psychiatry, Hospital for Sick Children, Toronto, ON, Canada;

marlena.colasanto@sickkids.ca 4 Department of Psychiatry, Hospital for Sick Children, Toronto, ON, Canada; Division of Child and

Youth, Centre for Addiction and Mental Health, Toronto, ON, Canada; Department of Psychiatry,

University of Toronto, ON, Canada; peter.szatmari@utoronto.ca 5 Institute for Health Policy, Management and Evaluation, University of Toronto, ON, Canada;

chenyang@smh.ca 6 Department of Pediatrics, Hospital for Sick Children, Toronto, ON, Canada, Li Ka Shing Knowledge

Institute, and Department of Pediatrics, St Michael’s Hospital, Toronto, ON, Canada, Department of

Pediatrics, Faculty of Medicine, University of Toronto, Toronto, ON, Canada;

jonathon.maguire@utoronto.ca 7 Department of Pediatrics, Hospital for Sick Children, Toronto, ON, Canada; Department of Pediatrics,

Faculty of Medicine, University of Toronto, Toronto, ON, Canada; patricia.parkin@sickkids.ca 8 Department of Pediatrics, Hospital for Sick Children, Toronto, ON, Canada, Li Ka Shing Knowledge

Institute, St Michael’s Hospital, Toronto, ON, Canada, Department of Pediatrics, Faculty of Medicine,

University of Toronto, Toronto, ON, Canada; catherine.birken@sickkids.ca

Address correspondence to: Daphne J. Korczak, Department of Psychiatry, The Hospital for

Sick Children, 555 University Avenue, Toronto, ON, M5G1X8, [daphne.korczak@sickkids.ca],

tel. 416-813-8923; fax. 416 813-5326

Short title: NA predicts PA in young children

Financial Disclosure: The authors have no financial relationships relevant to this article to

disclose.

Funding Source: This study was supported by the Canadian Institutes of Health Research

(CIHR), Sick Kids Foundation, St Michael’s Hospital and CAMH Foundation.

Conflict of interest Statement: The authors have no conflicts of interest relevant to this article

to disclose.

Abbreviations: NA=Negative Affectivity; PA=Physical activity; CVD=Cardiovascular disease;

MDD=Major depressive disorder; CBQ=Children’s behaviour questionnaire;

SES=Socioeconomic status; BMI=Body mass index; ADHD=Attention deficit hyperactivity

disorder

Keywords: depression, physical activity, epidemiology, development, temperament

Word Count abstract = 233

Word Count manuscript = 3,398

Tables = 2 + 1 online only

Figure = 1

ABSTRACT

Objective: To examine the longitudinal association of negative affect and physical activity in a

population of preschool children.

Study Design: Participants included 763 children (53% male) attending scheduled health

supervision visits in their primary care physicians’ offices. Data were collected at two time

points at mean ages 27 (SD = 5.4) and 47 (SD = 6.2) months. Negative affect (NA) was

measured using the Negative Affectivity (frustration/anger, decreased soothability) domain of

the Children’s Behaviour Questionnaire. Physical Activity (PA) was assessed using a parent-

report questionnaire. Multiple regression analyses tested the association between NA and PA,

adjusting for child age, sex, z-BMI, PA at Time 1, maternal education, household income, and

season, and examined for sex differences in the relationship between NA and PA.

Results: The longitudinal association between NA at Time 1 and PA at Time 2 was moderated

by sex (p<.001). After adjusting for covariates, females with greater NA at Time 1

had decreased PA at Time 2 (p=.01), whereas males with greater NA at Time 1

had increased PA at Time 2 (p=.01). Specifically, among females, every 1 unit increase in NA at

Time 1 was associated with a 9.9 minute/day decrease in PA at Time 2 (95% CI: -17.1,-2.8).

Conclusions: NA and PA were associated early in childhood and the effects of NA on PA were

gender specific. These findings underscore the importance of longitudinal and gender-specific

analyses in mood-obesity research.

INTRODUCTION

Cardiovascular disease (CVD) is the leading cause of death for individuals with Major

Depressive Disorder (MDD)1, with a standardized mortality ratio of 1.5-1.91,2 and surpassing all

other causes of death, including suicide. A primary route by which the increased risk of CVD

among depressed individuals may be conferred is through obesity. Moreover, emerging

evidence suggests that the pathways leading from MDD to CVD among adults may be gender-

specific, such that the relationship may be mediated in part by obesity for women but not for

men2,3.

Uncovering mechanisms of the association between MDD and obesity are critical for

providing insight into effective targets of preventive intervention programs. It has been

suggested, for example, that the association between MDD and obesity may be the result of

overlapping illness consequences: depressive amotivation, fatigue and low energy lead to

increased sedentary behaviour, overeating, decreased physical activity, and in turn, weight gain.

Indeed, depression has been associated with low levels of physical activity4. However,

researchers have also noted that depression in childhood and adolescence predicts increased

overweight in adulthood, even after controlling for the persistence of depressive symptoms into

adulthood5. Moreover, evidence suggests that many healthy lifestyle factors, such as engaging in

regular physical activity, are established prior to the onset of MDD and may be circumscribed

early in childhood6,7.

An alternate explanation for the MDD and obesity co-morbidity observed in adult

samples is that common factors, present prior to the onset of either condition, may underlie the

association and predispose a child to the development of both MDD and obesity. Data from the

International Children’s Accelerometry Database (n=20,871)8 and the International Study of

Childhood Obesity, Lifestyle and the Environment (n=6,539)9 confirm that time spent in

physical activity is more strongly associated with obesity than time spent in sedentary behavior,

indicating that decreased physical activity is a key driver of obesity. A growing body of research

now suggests that the processes leading to obesity, including low physical activity10-12, begin in

childhood. Also, compelling research confirms that childhood temperamental factors (i.e. a high

degree of early negative emotionality) predicts future childhood and adult MDD13-20 and is

significantly associated with parental depression, the most established risk factor for future

MDD21. Moreover, a recent systematic review and meta-analysis has found that increased

physical activity among older children and adolescents is associated with decreased

contemporaneous, and to a lesser extent future, depressive symptoms22. Thus, we aimed to

investigate whether child temperament and physical activity are associated in the preschool

period. Understanding the temporal association of these risk factors in this very early

developmental period is critical to the development of interventions that have the potential for

sustained effects. For example, by determining early antecedents to low physical activity,

targeted intervention strategies may prevent the development of a pattern of inactivity before it

becomes entrenched. To our knowledge, this is the first study to examine the longitudinal

relationship of temperament and physical activity in very young children.

The objective of this study was to prospectively investigate the longitudinal association

between child temperament and physical activity in a primary-care based sample of preschool

children. Further, we aimed to determine the temporal association of these factors. Consistent

with the literature regarding MDD and physical activity in older adults, we hypothesized that

greater negative affect (NA) would be associated with decreased physical activity (PA), and that

this relationship would be stronger for girls than for boys.

METHOD

Sample

TARGet Kids! is a primary care-based research network in Toronto, Ontario, Canada

comprised of 10 pediatric or family practice primary care clinics, with 3 to 9 physicians in each

practice23. The practice sites were selected to represent geographical and socioeconomically

diverse areas in Toronto. A complete description of the TARGet Kids! cohort has been

published23. Trained research assistants embedded in the practices obtained data at the time of

children’s scheduled health supervision visits. The cohort is comprised of children followed

prospectively from birth to 5 years of age, with a retention rate of 71% over the first five years of

life23. Temperament and physical activity data were collected at two time points, at least 12

months apart, corresponding to the 3-year and 5-year health supervision visit. Data used in this

analysis were collected from 763 children with non-missing data on childhood temperament at

Time 1 (T1) and physical activity at Time 2 (T2). Exclusion criteria included birth weight <

2500g, presence of a chronic illness that might affect feeding or growth (e.g. cystic fibrosis), or

severe developmental delay. Data were collected between December 2008 and August 2015.

Written informed consent was obtained. The study protocol was approved by the Institutional

Research Ethics Boards.

Measures

Child Temperament

Parent-reported measures of child temperament were collected using the Early Childhood

Behavior Questionnaire24 (ECBQ; T1) and Children’s Behavior Questionnaire – very short

form25 (CBQ; T1 and 2), depending on the age of the child (ECBQ for children less than 3 years

at T1, CBQ for children 3 years of age and older at T1). The ECBQ and the CBQ are

developmentally equivalent measures that have been shown to have satisfactory inter-rater

reliability and longitudinal stability, both within and across measures26, and internal consistency

(0.68 to 0.93)27. Parents are asked to rate their children on a 7-point scale ranging from 1

(never/extremely untrue) to 7 (always/extremely true). Higher scores indicate that the child

exhibits more of the temperamental trait. Factor analyses of the ECBQ and CBQ scales yield a

similar three-factor resolution indicating three broad domains of temperament: Effortful Control,

Surgency/Extraversion, and Negative Affectivity24,26-28. Effortful control is characterized by

increased attentional capacities, inhibitory control, and enjoyment of low-intensity activities.

Surgency/Extraversion is characterized by impulsivity, positive anticipation, smiling and

laughter, enjoyment of intense experiences, and a lack of shyness. Negative affectivity (NA) is

characterized by proneness to sadness, anger, discomfort, fear and a reduced ability to soothe

oneself27. These measures are the most widely used measures of temperament of young children

and validated in samples of children from diverse cultural backgrounds29-32.

Physical Activity

Physical activity in children was measured by parent report. Parents were asked “Aside

from time in daycare and preschool, on a typical weekday, how much time does your child spend

outside in unstructured free play?” Parents were instructed to provide their response in minutes

per day. This question has been used to assess regular physical activity of children in the

Canadian Community Health Survey33, a national survey of population health, and has been

shown to correlate with accelerometry in the current sample34.

Covariates

Variables hypothesized to be associated with NA and PA were included as covariates,

including household income, maternal education, child age and sex, child body mass index

(BMI), and season at the time of physical activity measure. Covariates were determined a priori

and are presented as is, without model selection based on the p-value, in order maintain the

interpretation of the p-values in our hypothesis testing. Weight and height were measured at the

time of primary clinic visit. BMI was calculated as weight in kilograms divided by height in

meters squared to allow for computation of BMI z-scores, which indicate relative weight

adjusted for child age and sex, according to reference standards35.

Statistical analyses

Baseline (T1) assessments of the relationships between covariates, temperament domains and

PA were performed using t-tests, ANOVA, or chi-squared tests, as appropriate. Pearson product-

moment, point-biserial, and phi correlations were computed to assess the relationship between

continuous-continuous, dichotomous-continuous, and dichotomous-dichotomous variables,

respectively (see Table 3). After checking for collinearity, multiple linear regression analyses were

conducted to examine the relationship between each temperament domain (e.g. NA) and PA. For each

temperament domain examined, the T1 measure of PA was included as a covariate in order to control

for degree of PA at the earlier time point. Covariates included child age, sex, z-BMI, maternal

education, household income, and season (fall/winter or spring/summer). As previous research

suggests that the association between depression and cardiovascular risk factors may be gender-

specific2-4, the a priori decision was made to further examine the interaction of temperament domain

and child sex on PA. Determination of the interaction variable was conducted after computing

centered variables for inclusion in the regression analysis. Centering, or subtracting the mean from a

variable and leaving deviation scores, was performed under these circumstances to improve the

interpretability of the regression coefficients and reduce multicollinearity among predictor variables36.

Covariate data was missing for 78 (9%) participants. Comparison of T1 demographic factors,

temperament domain score and PA level between participants with missing and with complete data

revealed that participants with missing data were younger than those with complete data (p < .05).

Participants with missing and complete data did not differ on other factors examined. Data was

analyzed using the Statistical Package for the Social Sciences (SPSS) for Windows Version 23.0.

RESULTS

Demographic factors, physical activity, and temperament characteristics of the 763

participants (404 boys, 358 girls) are presented in Table 1. Participants were a mean age 27

months (SD = 5.4) at Time 1, and 47 months (SD = 6.2) at Time 2. At T1, participants’ mean z-

BMI was equal to 0.3 (SD=1). The amount of time that children spent participating in physical

activity decreased over the study period from 64 minutes/day (T1) to 53 minutes/day (T2) [t

(762)=4.96, p<.001]. Significant changes in child temperament were reported over the course of

the study, with older children demonstrating greater scores on NA [t (762)=-18.4, p<.001] and

Effortful Control [t (762)= -9.0, p<.001], and lower scores on Surgency/Extraversion [t

(762)=10.1, p<.001] at T2 compared with reports at younger ages (T1) .

Multiple linear regression was conducted to examine the association of temperament at

T1 with Physical Activity (PA) at T2, controlling for age, sex, maternal education, household

income, season, z-BMI, and PA at T1. As the interaction of NA and sex was significant (p<.001;

R2= 0.112) in predicting PA (see Figure 1), further analyses were stratified by sex.

Among girls, NA was significantly associated with PA at T2 (p=.011) and explained

10.9% of the variance in PA (see Table 2). Similarly among boys, NA was significantly

associated with PA at T2 (p=.01) and explained 15.3% of the variance in PA (Table 2).

However, the direction of the association was different for girls and for boys. Among girls, for

every unit increase in NA score, PA decreased by 9.9 minutes per day (95% CI = -17.1, -2.8),

whereas among boys, for every unit increase in NA score, PA increased by 9.0 minutes per day

(95% CI = 2.0, 14.3). Sensitivity analyses conducted for temperament measure subgroups (ie.

CBQ at T1 and T2; n=482) yielded similar results (data available upon request). Examination of

the relationship between variables found that covariates in the model were not highly correlated

(Table 3). Factors of moderate strength that were significantly correlated were decreased

physical activity and winter season (r = -.27, p<.01), level of PA at baseline and at follow-up

(r=.27, p<.01), and degree of NA at T1 and T2 (r=.45, p<.01). Neither surgency/extraversion (p=

.43) nor effortful control (p = .76) temperament domains predicted PA. Controlling for potential

effects of concurrent temperament in the model, by including the T2 measure of the temperament

domain (for example, NA at T2), had no appreciable change on these results (data not shown).

DISCUSSION

This study examined the longitudinal relationship of early childhood temperament and later

physical activity among preschool aged children, and found significant associations between NA

and physical activity which were moderated by child sex. Among girls, greater early NA

predicted decreased future physical activity, whereas among boys, greater NA was associated

with later increased physical activity. These findings remained after controlling for child age,

BMI, family income, season, or previous physical activity level. Neither child surgency nor

effortful control domains were associated with future PA.

Our findings with respect to girls are consistent with those reported on the relationship

between depression and physical activity in adolescent and adult populations, though few studies

have examined the relationship in children, and to our knowledge, no previous study has

examined this association in preschoolers22. A systematic review of adult studies found that

depression was associated with decreased future physical activity levels, though sex effects were

not examined37. Similarly, in a follow up study of adolescents who had been depressed in

childhood, adolescents with child-onset (mean age 9 years) depression had lower PA levels

compared with their never-depressed siblings and controls, even though the majority of the

sample (85%) was not depressed at follow up38. However, as in the systematic review above37

sex differences in the relationship between depression and low physical activity were not

examined. Our findings are, however, consistent with emerging literature reporting sex

differences in the association of youth-onset depression with future CVD, in which the

potentially mediating role of obesity may be of greater importance for women than for men2,3.

Biological maturation during puberty, social, racial/ethnic, and socioeconomic influences have

all been reported to contribute to a general decrease in level of physical activity among girls

relative to boys in the adolescent age group39-44.

In the preschool age group, the mechanism by which increased NA might differentially

influence physical activity among boys and girls is less clear. One possible explanation may be

that girls demonstrate greater homotypic continuity of NA with depressive symptoms over time.

That is, that NA is a developmental precursor of depression, with similar phenotypic expression

over time. Decreased experience of enjoyment, difficulty overcoming fear associated with novel

physical activities, persistence in the discomfort that may accompany outdoor play, or belief in

one’s ability to compare favorably with one’s peers may be present earlier and with greater

stability for girls. In contrast, boys have been reported to demonstrate heterotypic continuity of

psychopathological symptoms45. That is, the earliest manifestations of adolescent or adult

depressive risk among boys are more likely to be disruptive and hyperactive behaviours at

younger ages45. Thus, boys with greater temperamental NA may also demonstrate increased

behavioural hyperactivity, with resultant increased physical activity at younger ages when

compared with girls. Further support for this explanation comes from population-based studies

of Attention Deficit Hyperactivity Disorder (ADHD), a disorder that affects twice as many boys

as girls46, which confirm that boys with ADHD in early childhood are at increased risk of

depression at adolescent and older ages46 (an example of heterotypic continuity). Other plausible

explanations for the discordant gender findings in our study include potential differential

parenting responses to girls and boys who are distressed or difficult to soothe47, different

parenting approaches for boys and girls regarding physical activity48,49, or more direct effects of

a common underlying etiological basis for both depression and obesity-related cardiometabolic

risk that are detectable early in the lifecourse. Several studies, for example, have implicated

common pathophysiological pathways in the etiology of both conditions, including disturbances

in the HPA axis and norepinephrine systems50, potentially shared genetic vulnerabilities51, and

increased immuno-inflammatory activity52,53.

This study found that while many of the variables considered in our model were weakly or

not at all correlated, stronger associations in the weak to moderate range were found for PA at T1

and T2, and for PA with season of measurement. These findings are consistent with previous

studies and with the experience of life in Canada, in which young children receive less outdoor

physical activity during the colder and shorter winter days compared with the warmer seasons.

The strongest association was observed between NA at both time points, indicating the relative

stability of temperament over time26,54,55. That parents of children in our sample reported

relatively greater effortful control, and decreased surgency/extraversion, at T2 as compared with

T1 is also consistent with current developmental knowledge regarding children’s increased

ability to regulate their reactions and responses in different contexts as they mature56,57.

This study found that a one unit increase in NA is associated with a 10 minute decrease in

PA among girls. In our sample of preschool-aged children, a one unit rise in NA corresponds to

an increase of just over one standard deviation, or approximately 35%. As temperament

measures do not yield clinical diagnoses, it cannot be determined whether such an increase

would result in a clinical diagnosis of depression. However, research has reported that higher

temperamental negative emotionality is associated with a concurrent15 and future17 diagnosis of

depression among very young children and increased risk of future depression13,16,18-20.

Similarly, our results suggest that the magnitude of the decrease in physical activity for children

with high NA is also of clinical relevance. In keeping with our finding of decreased PA at T2,

recent research has reported that among children 3 to 5 years of age, younger children are more

likely to meet physical activity guidelines58. Moreover, these data report that 5 year old girls in

Canada are less physically active than 5 year old boys (68 vs 81 minutes/day of moderate-

vigorous physical activity, p<.05)58. A decrease in PA of 10 minutes/day represents a potential

further 15% decline among girls with higher temperamental NA. Thus, the results of this study

may have clinical implications when defining a population of preschool aged girls at risk for

both negative mental and physical health outcomes.

As the first study to report on the longitudinal association of early childhood temperament

with physical activity, this study has several strengths including recruitment from population

based primary care settings, inclusion of measured height and weight, and an analytical approach

that accounted for previous PA level. However, our findings must also be interpreted within the

context of the limitations of this study. Children and families that participated in the study were

primarily of middle to higher socioeconomic status (SES), which may limit the ability to

generalize our findings. However, previous research has found that the proportion of children

meeting current PA guidelines in the TARGet Kids! Cohort is similar to other Canadian studies

of this age group59, and self-reported income levels of our cohort are consistent with population-

based measures of Canadian family incomes60. Research regarding the association of SES with

child physical activity has yielded inconsistent findings61-63, thus whether or not the present

study’s findings would be altered in a sample with greater heterogeneity is unclear. Similarly,

although we found the demographic characteristics and physical activity levels in our sample to

be comparable with those of other Canadian studies, we cannot eliminate the possibility of

selection bias in our analyses. We were also unable to account for the potential role of siblings in

This study used the most widely studied, validated and reliable measure of childhood

temperament, however, the measure was completed by a single informant. Use of multiple

informants or independent observer rated temperament may decrease the potential for rater bias,

though parent report of temperament has been reported to correlate with that of a familiar,

unrelated adult (teacher) in this age group64. Future studies would benefit from employing

multiple raters of child temperament in this age group, to minimize the potential for rater bias.

Also, this study investigated the level of children’s PA outside of daycare and as such, cannot

address the potential moderating role of childcare settings on PA. However, an underestimation

of PA level in the current study based on daycare attendance is unlikely as a growing body of

literature shows that children are not physically active in many child care settings65-67 and we

have found that parent-report of unstructured free play demonstrated the strongest correlation

with accelerometry data, irrespective of daycare attendance34. Finally, this study employed

report-based measurement of PA for young children. Accelerometry may provide more precise

information regarding PA, although is also subject to inaccuracy if removed for activities (e.g.

swimming) or if adhered to inconsistently. Accuracy may also vary with the specific device

employed68. The use of accelerometry in the TARGet Kids! sample may similarly underestimate

child PA in this age group34. This would not be expected to alter the findings of the present

study, however, as patient-related factors did not predict differences in PA-accelerometry

correlation, such that we would not expect variation in correlation across subgroups.

Conclusions

This study suggests that early NA is associated with future PA and that the relationship is

moderated by sex. Moreover, this study suggests that the putative psychological risk factors for

depression are observable prior to those that may ultimately lead to increased cardiometabolic

risk. These findings underscore the importance of including longitudinal and gender-specific

analyses in mood-obesity research, to detect changes in symptom emergence throughout key

developmental periods and ensure interventions are aimed at the right targets, in the right

population, during the right period of development.

REFERENCES

1 Osby, U., Brandt, L., Correia, N., Ekbom, A. & Sparen, P. Excess mortality in bipolar and unipolar disorder in Sweden. Arch Gen Psychiatry 58, 844-850 (2001).

2 Markkula, N. et al. Mortality in people with depressive, anxiety and alcohol use disorders in Finland. Br J Psychiatry 200, 143-149, doi:10.1192/bjp.bp.111.094904 (2012).

3 Appleton, A. A., Loucks, E. B., Buka, S. L., Rimm, E. & Kubzansky, L. D. Childhood emotional functioning and the developmental origins of cardiovascular disease risk. J Epidemiol Community Health 67, 405-411, doi:10.1136/jech-2012-201008 (2013).

4 Jerstad, S. J., Boutelle, K. N., Ness, K. K. & Stice, E. Prospective reciprocal relations between physical activity and depression in female adolescents. J Consult Clin Psychol 78, 268-272, doi:10.1037/a0018793 (2010).

5 Korczak, D. J., Lipman, E., Morrison, K., Duku, E. & Szatmari, P. Child and adolescent psychopathology predicts increased adult body mass index: results from a prospective community sample. J Dev Behav Pediatr 35, 108-117, doi:10.1097/dbp.0000000000000015 (2014).

6 Belanger, M. et al. Number of years of participation in some, but not all, types of physical activity during adolescence predicts level of physical activity in adulthood: Results from a 13-year study. Int J Behav Nutr Phys Act 12, 76, doi:10.1186/s12966-015-0237-x (2015).

7 Kwon, S., Janz, K. F., Letuchy, E. M., Burns, T. L. & Levy, S. M. Developmental Trajectories of Physical Activity, Sports, and Television Viewing During Childhood to Young Adulthood: Iowa Bone Development Study. JAMA Pediatr 169, 666-672, doi:10.1001/jamapediatrics.2015.0327 (2015).

8 Ekelund, U. et al. Moderate to vigorous physical activity and sedentary time and cardiometabolic risk factors in children and adolescents. JAMA 307, 704-712 (2012).

9 Katzmarzyk, P. T. et al. Physical Activity, Sedentary Time, and Obesity in an International Sample of Children. Med Sci Sports Exerc. 47 2062-2069 (2015).

10 Dunton, G. et al. Organized physical activity in young school children and subsequent 4-year change in body mass index. Arch Pediatr Adolesc Med 166, 713-718, doi:10.1001/archpediatrics.2012.20 (2012).

11 Saldanha-Gomes, C. et al. Prospective associations between energy balance-related behaviors at 2 years of age and subsequent adiposity: the EDEN mother-child cohort. Int J Obes (Lond), doi:10.1038/ijo.2016.138 (2016).

12 Telama, R. et al. Tracking of physical activity from early childhood through youth into adulthood. Med Sci Sports Exerc 46, 955-962, doi:10.1249/mss.0000000000000181 (2014).

13 Bruder-Costello, B. et al. Temperament among offspring at high and low risk for depression. Psychiatry Res 153, 145-151, doi:10.1016/j.psychres.2007.02.013 (2007).

14 Bufferd, S. J. et al. Predictors of the onset of depression in young children: a multi-method, multi-informant longitudinal study from ages 3 to 6. J Child Psychol Psychiatry 55, 1279-1287, doi:10.1111/jcpp.12252 (2014).

15 Dougherty, L. R. et al. Preschoolers' observed temperament and psychiatric disorders assessed with a parent diagnostic interview. J Clin Child Adolesc Psychol 40, 295-306, doi:10.1080/15374416.2011.546046 (2011).

16 Lieb, R., Isensee, B., Hofler, M., Pfister, H. & Wittchen, H. U. Parental major depression and the risk of depression and other mental disorders in offspring: a prospective-longitudinal community study. Arch Gen Psychiatry 59, 365-374 (2002).

17 Reinfjell, T., Karstad, S. B., Berg-Nielsen, T. S., Luby, J. L. & Wichstrom, L. Predictors of change in depressive symptoms from preschool to first grade. Dev Psychopathol 28, 1517-1530, doi:10.1017/s0954579415001170 (2016).

18 Verstraeten, K., Vasey, M. W., Raes, F. & Bijttebier, P. Temperament and risk for depressive symptoms in adolescence: mediation by rumination and moderation by effortful control. J Abnorm Child Psychol 37, 349-361, doi:10.1007/s10802-008-9293-x (2009).

19 Weissman, M. M. et al. Onset of major depression in early adulthood. Increased familial loading and specificity. Arch Gen Psychiatry 41, 1136-1143 (1984).

20 Weissman, M. M. et al. Offspring of depressed parents: 20 years later. Am J Psychiatry 163, 1001-1008, doi:10.1176/appi.ajp.163.6.1001 (2006).

21 Olino, T. M., Klein, D. N., Dyson, M. W., Rose, S. A. & Durbin, C. E. Temperamental emotionality in preschool-aged children and depressive disorders in parents: associations in a large community sample. J Abnorm Psychol 119, 468-478, doi:10.1037/a0020112 (2010).

22 Korczak, D. J., Madigan, S. & Colasanto, M. Children's Physical Activity and Depression: A Meta-analysis. Pediatrics 139, doi:10.1542/peds.2016-2266 (2017).

23 Carsley, S. et al. Cohort Profile: The Applied Research Group for Kids (TARGet Kids!). Int J Epidemiol, doi:10.1093/ije/dyu123 (2014).

24 Putnam, S. P., Gartstein, M. A. & Rothbart, M. K. Measurement of fine-grained aspects of toddler temperament: the early childhood behavior questionnaire. Infant Behav Dev 29, 386-401, doi:10.1016/j.infbeh.2006.01.004 (2006).

25 Putnam, S. P. & Rothbart, M. K. Development of short and very short forms of the Children's Behavior Questionnaire. J Pers Assess 87, 102-112, doi:10.1207/s15327752jpa8701_09 (2006).

26 Putnam, S. P., Rothbart, M. K. & Gartstein, M. A. Homotypic and Heterotypic Continuity of Fine-grained Temperament during Infancy, Toddlerhood, and Early Childhood. Inf. Child. Dev. 17, 387-405 (2008).

27 Rothbart, M. K., Ahadi, S. A., Hershey, K. L. & Fisher, P. Investigations of temperament at three to seven years: the Children's Behavior Questionnaire. Child Dev 72, 1394-1408 (2001).

28 Rothbart, M. K. & Ahadi, S. A. Temperament and the development of personality. J Abnorm Psychol 103, 55-66 (1994).

29 Lim, J. Y. & Bae, Y. J. Validation Study of Korean Version of the Rothbart's Children's Behavior Questionnaire. Korean Journal of Human Ecology 24, 477-497 (2015).

30 Klein, V. C., Putnam, S. P. & Martines Linhares, M. B. Assessment of temperament in children: Translation of instruments to Portuguese (Brazil) Language. Interamerican Journal of Psychology 43, 442-447 (2009).

31 Sleddens, E. F. C., Kremers, S. P. J., Candel, M. J. J. M., DeVries, N. N. K. & Thijs, C. Validating the Children's Behavior Questionnaire in Dutch Children: Psychometric Properties and a Cross-Cultural Comparison of Factor Structures. Psychological Advancement, doi:10.1037/a0022111 (2011).

32 Ahadi, S. A., Rothbart, M. K. & Ye, R. Children's Temperament in the US and China: Similarities and Differences. European Journal of Personality 7, 359-377 (1993).

33 Canadian Community Health Survey 2010, <http://www.hc-sc.gc.ca/fn-an/surveill/nutrition/commun/index-eng.php> (

34 Sarker, H. et al. Validation of parent-reported physical activity and sedentary time by accelerometry in young children. BMC Res Notes 8, 735, doi:10.1186/s13104-015-1648-0 (2015).

35 Secker, D. Promoting optimal monitoring of child growth in Canada: using the new WHO growth charts. Can J Diet Pract Res 71, e1-3 (2010).

36 Aiken, L. & West, S. Multiple regression: Testing and interpreting interactions. . (Sage, 1991). 37 Roshanaei-Moghaddam, B., Katon, W. J. & Russo, J. The longitudinal effects of depression on

physical activity. Gen Hosp Psychiatry 31, 306-315, doi:10.1016/j.genhosppsych.2009.04.002 (2009).

38 Rottenberg, J. et al. The association between major depressive disorder in childhood and risk factors for cardiovascular disease in adolescence. Psychosom Med 76, 122-127, doi:10.1097/psy.0000000000000028 (2014).

39 Bradley, C. B., McMurray, R. G., Harrell, J. S. & Deng, S. Changes in common activities of 3rd through 10th graders: the CHIC study. Med Sci Sports Exerc 32, 2071-2078 (2000).

40 Cumming, S. P., Standage, M., Gillison, F. & Malina, R. M. Sex differences in exercise behavior during adolescence: is biological maturation a confounding factor? J Adolesc Health 42, 480-485, doi:10.1016/j.jadohealth.2007.10.005 (2008).

41 Gordon-Larsen, P., McMurray, R. G. & Popkin, B. M. Adolescent physical activity and inactivity vary by ethnicity: The National Longitudinal Study of Adolescent Health. J Pediatr 135, 301-306 (1999).

42 Grieser, M. et al. Physical activity attitudes, preferences, and practices in African American, Hispanic, and Caucasian girls. Health Educ Behav 33, 40-51, doi:10.1177/1090198105282416 (2006).

43 Richmond, T. K., Field, A. E. & Rich, M. Can neighborhoods explain racial/ethnic differences in adolescent inactivity? Int J Pediatr Obes 2, 202-210, doi:10.1080/17477160701305781 (2007).

44 Vu, M. B., Murrie, D., Gonzalez, V. & Jobe, J. B. Listening to girls and boys talk about girls' physical activity behaviors. Health Educ Behav 33, 81-96, doi:10.1177/1090198105282443 (2006).

45 Boylan, K., Georgiades, K. & Szatmari, P. The longitudinal association between oppositional and depressive symptoms across childhood. J Am Acad Child Adolesc Psychiatry 49, 152-161 (2010).

46 Larson, K., Russ, S. A., Kahn, R. S. & Halfon, N. Patterns of comorbidity, functioning, and service use for US children with ADHD, 2007. Pediatrics 127, 462-470, doi:10.1542/peds.2010-0165 (2011).

47 Bezirganian, S. & Cohen, P. Sex differences in the interaction between temperament and parenting. J Am Acad Child Adolesc Psychiatry 31, 790-801, doi:10.1097/00004583-199209000-00004 (1992).

48 Jago, R. et al. Parenting styles, parenting practices, and physical activity in 10- to 11-year olds. Prev Med 52, 44-47, doi:10.1016/j.ypmed.2010.11.001 (2011).

49 St George, S. M. & Wilson, D. K. A qualitative study for understanding family and peer influences on obesity-related health behaviors in low-income African-American adolescents. Child Obes 8, 466-476, doi:10.1089/chi.2011.0067 (2012).

50 Gold, P. W. The organization of the stress system and its dysregulation in depressive illness. Mol Psychiatry 20, 32-47, doi:10.1038/mp.2014.163 (2015).

51 Krabbe, K. S. et al. Brain-derived neurotrophic factor (BDNF) and type 2 diabetes. Diabetologia 50, 431-438, doi:10.1007/s00125-006-0537-4 (2007).

52 Chaves Filho, A. J. M. et al. IDO chronic immune activation and tryptophan metabolic pathway: A potential pathophysiological link between depression and obesity. Prog Neuropsychopharmacol Biol Psychiatry 80, 234-249, doi:10.1016/j.pnpbp.2017.04.035 (2018).

53 Rethorst, C. D., Bernstein, I. & Trivedi, M. H. Inflammation, obesity, and metabolic syndrome in depression: analysis of the 2009-2010 National Health and Nutrition Examination Survey (NHANES). J Clin Psychiatry 75, e1428-1432, doi:10.4088/JCP.14m09009 (2014).

54 Komsi, N. et al. Continuity of temperament from infancy to middle childhood. Infant Behav Dev 29, 494-508, doi:10.1016/j.infbeh.2006.05.002 (2006).

55 Neppl, T. K. et al. Differential stability of temperament and personality from toddlerhood to middle childhood. J Res Pers 44, 386-396, doi:10.1016/j.jrp.2010.04.004 (2010).

56 Kochanska, G. & Knaack, A. Effortful control as a personality characteristic of young children: antecedents, correlates, and consequences. J Pers 71, 1087-1112 (2003).

57 Posner, M. I., Rothbart, M. K., Sheese, B. E. & Tang, Y. The anterior cingulate gyrus and the mechanism of self-regulation. Cogn Affect Behav Neurosci 7, 391-395 (2007).

58 Statistics Canada. Physical activity and sedentary behaviour of Canadian children aged 3 to 5: Sample characteristics, household population aged 5, <http://www.statcan.gc.ca/pub/82-003-x/2016009/article/14653/tbl/tbl03-eng.htm> (2009 to 2013).

59 Borkhoff, C. M. et al. Objectively measured physical activity of young Canadian children using accelerometry. Appl Physiol Nutr Metab 40, 1302-1308, doi:10.1139/apnm-2015-0164 (2015).

60 Statistics Canada. Average income after tax by family types (2007-2011). (2013). 61 Matsudo, V. K. et al. [Socioeconomic status indicators, physical activity, and overweight/obesity

in Brazilian children]. Rev Paul Pediatr, doi:10.1016/j.rpped.2015.04.003 (2016). 62 Ritterman Weintraub, M. L., Fernald, L. C., Goodman, E., Guendelman, S. & Adler, N. E. Obesity-

Related Behaviors among Poor Adolescents and Young Adults: Is Social Position Associated with Risk Behaviors? Front Public Health 3, 224, doi:10.3389/fpubh.2015.00224 (2015).

63 Villagran Perez, S., Novalbos-Ruiz, J. P., Rodriguez-Martin, A., Martinez-Nieto, J. M. & Lechuga-Sancho, A. M. Implications of family socioeconomic level on risk behaviors in child-youth obesity. Nutr Hosp 28, 1951-1960, doi:10.3305/nutr hosp.v28in06.6848 (2013).

64 Bishop, G., Spence, S. H. & McDonald, C. Can parents and teachers provide a reliable and valid report of behavioral inhibition? Child Dev 74, 1899-1917 (2003).

65 Carsley, S. et al. The impact of daycare attendance on outdoor free play in young children. J Public Health (Oxf), doi:10.1093/pubmed/fdw006 (2016).

66 Reilly, J. J. Low levels of objectively measured physical activity in preschoolers in child care. Med Sci Sports Exerc 42, 502-507, doi:10.1249/MSS.0b013e3181cea100 (2010).

67 Tandon, P. S., Saelens, B. E. & Christakis, D. A. Active play opportunities at child care. Pediatrics 135, e1425-1431, doi:10.1542/peds.2014-2750 (2015).

68 Hnatiuk, J. A., Salmon, J., Hinkley, T., Okely, A. D. & Trost, S. A review of preschool children's physical activity and sedentary time using objective measures. Am J Prev Med 47, 487-497, doi:10.1016/j.amepre.2014.05.042 (2014).

Figure 1. NA and child sex interact to predict physical activity

Legend: Child sex moderates the association between NA and physical activity.