Accepted Manuscript

The longitudinal association between the context of physical activity and mental
health in early adulthood

Isabelle Doré, Jennifer L. O'Loughlin, Mireille E. Schnitzer, Geetanjali D. Datta,

Louise Fournier

PII: S1755-2966(17)30172-2
DOI: 10.1016/j.mhpa.2018.04.001
Reference: MHPA 252

To appear in: Mental Health and Physical Activity

Received Date: 13 November 2017

Revised Date: 27 March 2018
Accepted Date: 9 April 2018

Please cite this article as: Doré, I., O'Loughlin, J.L., Schnitzer, M.E., Datta, G.D., Fournier, L., The
longitudinal association between the context of physical activity and mental health in early adulthood,
Mental Health and Physical Activity (2018), doi: 10.1016/j.mhpa.2018.04.001.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.
 ACCEPTED MANUSCRIPT

Title:

The longitudinal association between the context of physical activity and mental health in
early adulthood

List of authors:

PT
Isabelle Doré, Ph.D. (corresponding author)
Department of Kinesiology and Physical Education, University of Toronto
Department of Social and Preventive Medicine, School of Public Health, University of

RI
Montreal
CHUM Research Centre
55 Harbord Street, Toronto, ON M5S 2W6

C
Email: isabelle.dore@utoronto.ca

US
Jennifer L. O’Loughlin, Ph.D.
Department of Social and Preventive Medicine, School of Public Health, University of
Montreal
CHUM Research Centre
AN
850 rue Saint-Denis, Montréal, QC H2X 0A9
Email: jennifer.oloughlin@umontreal.ca
M

Mireille E. Schnitzer, Ph.D.

Faculty of Pharmacy, University of Montreal
Pavillon Jean-Coutu, 2940, chemin de la Polytechnique, Montréal, QC H3T 1J4
Email: mireille.schnitzer@umontreal.ca
D

Geetanjali D. Datta, Sc.D.

TE

Department of Social and Preventive Medicine, School of Public Health, University of

Montreal
CHUM Research Centre
EP

850 rue Saint-Denis, Montréal, QC H2X 0A9

Email: geetanjali.datta@umontreal.ca

Louise Fournier, Ph.D.

C

Department of Social and Preventive Medicine, School of Public Health, University of

Montreal
AC

CHUM Research Centre

850 rue Saint-Denis, Montréal, QC H2X 0A9
Email: louise.fournier@umontreal.ca

Abstract word count: 250

Word count: 3499
Tables: 3
Figures: 1
 ACCEPTED MANUSCRIPT

Abstract

This study examined the associations between the context in which physical activity (PA)

is undertaken (team sports, informal group, individual PA), and each of positive mental

health, anxiety symptoms and depressive symptoms. It also investigated whether social

PT
connectedness or PA volume mediate these associations. A total of 460 students (62.4%

RI
female; mean age at baseline = 18.5, SD 2.6) completed questionnaires at baseline and at

follow-up 6 months later. Multivariate linear regression was used to model the

C
associations between PA context at baseline and each outcome at follow-up controlling

US
for sex, age, perceived socioeconomic status and baseline values of the outcome.

Mediation analyses used causal inference methods to estimate the controlled direct effect
AN
(CDE), natural direct (NDE) and indirect effects (NIE) of social connectedness and PA

volume. Relative to individual PA, informal group PA and team sports were positively
M

associated with positive mental health (β (95% CI) = 2.24 (0.01, 4.46) and 3.39 (0.74,
D

5.59) respectively), and inversely associated with depressive symptoms (-0.65 (-1.29, -
TE

0.01) and -0.76 (-1.43, -0.09), respectively). A significant CDE was observed for PA

volume on the association between team sports and positive mental health (β (95% CI) =
EP

2.72 (0.03, 5.34)), suggesting that there is no benefit from increasing PA volume and that

it is specifically the context of team sports that provides the benefit on positive mental
C

health. Efforts to promote mental health and prevent depressive symptoms may benefit
AC

from intervention promoting PA in informal group and team sport settings.

Keywords: Physical activity context; mental health; depressive symptoms; social

connectedness; physical activity volume; mediation analysis.

1
 ACCEPTED MANUSCRIPT

1
Introduction

Numerous studies suggest that physical activity (PA) can promote positive mental health

and reduce depression and anxiety among youth in transition to adulthood (Da Silva et

al., 2012; Jewett et al., 2014; Mammen & Faulkner, 2013). Different PA modalities such

PT
as frequency, intensity, domain (i.e., leisure time PA, work-related PA, transport PA,

RI
household PA) and context (i.e., group settings, PA undertaken as an individual) have

been investigated in relation to mental health and mental disorders. The evidence to date

C
suggests that higher PA frequency and intensity are associated with increased mental

US
health benefits (Doré, O'Loughlin, Beauchamp, Martineau, & Fournier, 2016; Kremer et

al., 2014; McPhie & Rawana, 2012). Reasons for participating in PA likely vary
AN
according to the domain in which PA takes place and the experience of PA can differ

across domains, which can impact mental health outcomes differently (Sallis et al., 2006;
M

Shephard, 2003). Studies consistently report that leisure-time PA is more strongly

D

associated with mental health and mental disorder outcomes, than PA in other domains
TE

(Asztalos et al., 2009; Kull, Ainsaar, Kiive, & Raudsepp, 2012), and a recent meta-

analysis suggests that compared to other PA domains, promoting PA during leisure time
EP

could be the most effective method to prevent mental disorders (White et al., 2017).

Leisure-time PA takes place in numerous contexts which vary according to social nature
C

and whether the PA takes place in an organized (i.e., sports team) or non-organized
AC

context. Belonging to a sports team is associated with fewer depressive symptoms

(Brunet et al., 2013; Doré, O'Loughlin, et al., 2016; Sabiston et al., 2016), lower

1
CDE = Controlled Direct Effect, CI = Confidence Interval, GSLTPAQ = Godin-Shephard Leisure-time
Physical Activity Questionnaire, HADS = Hospital Anxiety and Depressive Scale, MVPA = Moderate-
Vigorous Physical Activity, MHC-SF = Mental Health Continuum - Short Form, NDE = Natural Direct
Effect, NIE = Natural Indirect Effect, PA = Physical Activity, ROPAS = Relatedness to others in Physical
Activity Scale, TE = Total Effect.

2
 ACCEPTED MANUSCRIPT

perceived stress and enhanced positive mental health (Doré, O'Loughlin, et al., 2016;

Jewett et al., 2014) compared to participation in individual PA among youth.

Although there is evidence of an association between PA and both mental health and

PT
mental disorders, the specific mechanisms underpinning these associations are not well-

RI
established (Hallgren et al., 2016; Monshouwer, ten Have, van Poppel, Kemper, &

Vollebergh, 2013). To inform public health programs and policy, we need to better

C
understand how protective influences manifest. It is unlikely that a single mechanism is

US
operative, but rather that a myriad of biological, psychological and social factors are

implicated. These mechanisms might interrelate, they might operate simultaneously

AN
and/or they may exert influence iteratively. The contributions of several physiological

and neurological mechanisms have been studied and suggest that PA influences
M

depressive symptoms by stimulating serotonin (Chaouloff, 1997) and endorphin

D

(Hallgren et al., 2016) secretions, which are known for their antidepressant and analgesic
TE

effects. They may also facilitate hypothalamic-pituitary adrenal axis regulation, which is

responsible for the stress hormone (i.e., cortisol) (Wipfli, Landers, Nagoshi, &
EP

Ringenbach, 2011). PA may also influence psychological factors such as perceived

competence (Valois, Umstattd, Zullig, & Paxton, 2008), self-esteem (McPhie & Rawana,
C

2012), self-mastery (Kwan, Davis, & Dunn, 2012) and body-image (Kamimura et al.,
AC

2014), which in turn enhance mental health. PA in social contexts such as group settings,

clubs or team sports provide increased opportunity for social interaction which may

contribute to mental health by enhancing social connectedness and support (Faulkner &

Carless, 2006; Holt, 2016; McHugh & Lawlor, 2012). Although some studies identify the

3
 ACCEPTED MANUSCRIPT

benefits of sports team or club participation on quality of life (Eime, Harvey, Brown, &

Payne, 2010; Vella, Cliff, Magee, & Okely, 2014), mental health (Doré, O'Loughlin, et

al., 2016; Jewett et al., 2014) and depressive symptoms (Brunet et al., 2013; Sabiston et

al., 2016), no previous study has investigated the mediating role of these social

PT
mechanisms in the association between PA and mental health.

RI
According to Self-Determination Theory, (Deci & Ryan, 2002) social relatedness

C
involves establishing and sustaining meaningful connections with others, which allows a

US
person to be socially connected with, and accepted as important by others. Social

connectedness reflects an internal sense of belonging (Deci & Ryan, 2002; Wilson &
AN
Bengoechea, 2010), and is defined as the subjective awareness of being in a close

relationship with the social world (Lee & Robbins, 1998). High levels of connectedness
M

provide individuals with an enhanced ability to manage their own needs and emotions
D

through cognitive processes (Tesser, 1991), and to protect against anxiety (Lee &
TE

Robbins, 1998) and depression (Armstrong & Oomen-Early, 2009).

EP

Although participation in PA social contexts may increase social connectedness which in

turn, influences mental health positively (see Figure 1), individuals who report
C

involvement in informal group PA and team sports also report higher PA volume than
AC

those who engage in PA individually (Eime, Young, Harvey, Charity, & Payne, 2013). It

is therefore possible that PA context impacts mental health by increasing PA volume

rather than by increasing social connectedness.

4
 ACCEPTED MANUSCRIPT

Our specific objectives were to examine the longitudinal associations between PA

context and each of positive mental health, anxiety symptoms and depressive symptoms

in youth, and to investigate whether social connectedness and/or PA volume mediate

these associations. Three PA contexts were examined including team sports, informal

PT
group PA (i.e., yoga classes, aqua fitness classes, cyclist groups), and PA engaged in

RI
individually. We hypothesized that, relative to individual PA, informal group PA and

team sports are associated with higher levels of positive mental health and fewer anxiety

C
and depressive symptoms, and that social connectedness and/or PA volume mediate these

US
associations. AN
[Insert Figure 1]
M

Methods
D

Study population
TE

A longitudinal study was conducted among CEGEP (Collège d’enseignement général et

professionnel / College of General and Professional Education) students in Quebec,

EP

Canada. In Quebec, CEGEPs provide post-secondary school education including 2-year

pre-university programs and 3-year career programs, the latter typically leading to
C

employment. Baseline data were collected in compulsory physical education classes in a

AC

single CEGEP in October 2013; 1,527 of 1,746 eligible students (87.4%) (58% female;

mean age=18.4, SD=2.4) completed questionnaires during class-time and provided

contact information for follow-up. Six months later (March 2014), 460 of the 1,527

participants (30.1%) completed a web-based questionnaire (62.4% female; mean age at

5
 ACCEPTED MANUSCRIPT

baseline=18.5, SD=2.6). The low response was likely related to online data collection at

follow-up. (2) Data at baseline were collected at school with teachers allotting in-class

time to complete the questionnaire. This strategy yielded an excellent response proportion

compared to online follow-up questionnaires (despite weekly email reminders). The

PT
decision to conduct a 6-month follow-up after baseline was based on several

RI
considerations: first, some students might have recently engaged in a new PA activity

(since the baseline data collection took place approximately one month after beginning

C
the CEGEP semester). Mental health and mental disorder symptoms can vary across

US
relatively short time periods, especially in youth. In order that participants had enough

time to develop and benefit from relations in PA, we chose to conduct a six-month
AN
follow-up. Further it was also important to avoid exam periods (mid- and end of

semester). Finally, because youth attend CEGEP in Quebec for a 2-year period only, it
M

was important to conduct the follow-up during the same academic year as baseline. We
D

believe that a 6-month follow-up was ideal. A higher proportion of females participated
TE

at follow-up, but there were no significant differences between participants with and

without follow-up data in PA context, positive mental health, anxiety or depressive

EP

symptoms, age and perceived socioeconomic status. Although statistically non-

significant, differences in PA context were observed. A higher proportion of participants

C

with follow-up data reported engaging in individual PA only and a lower proportion
AC

participated in team sports (Appendix I). The analytical sample was restricted to

participants who were physically active (n = 430) at the time of the first questionnaire,

since we were interested in mental health and mental disorder symptoms across PA

context, thus necessitating that participants report PA in a PA context.

6
 ACCEPTED MANUSCRIPT

This study was approved by the ethics review boards at the CEGEP de l’Outaouais and

the University of Montreal Faculty of Medicine, Quebec, Canada. Each participant

provided written informed consent.

PT
RI
Measures

PA Context. This variable distinguished three PA contexts with different levels of social

C
interactions: (i) team sports, (ii) informal group PA, and (iii) individual PA. This

US
categorization was based on the hypothesis that PA in team sports involves the highest

levels of social interaction, informal group PA (i.e., yoga classes, running groups),
AN
involves intermediate levels and individual PA involves the lowest levels. To compute

the PA context variable, participants were first asked whether they had been involved in
M

team sports (0, 1, 2, 3 or more teams) since the beginning of the current semester (Fall
D

2013) using the following question: Since the beginning of the current semester, how
TE

many organized sports teams did you belong to (where you practice with teammates or

play against other teams)? Participants involved in at least one team were included in the
EP

“team sports” category. Then, among participants who did not participate in team sports

(i.e., answered 0 to the previous question), data from a modified version of the Godin-
C

Shephard Leisure-Time Physical Activity Questionnaire (GSLTPAQ) (Amireault, Godin,

AC

Lacombe, & Sabiston, 2015) (Appendix II) were used to distinguish participants involved

in individual PA only and those involved in informal group PA. Specifically, among

active participants reporting at least one PA session during a typical week, those engaged

in PA “with someone” or “in a group of ≥ 2 persons” at least once a week were included

in the “informal group PA” category. The remaining active participants, not involved in
7
 ACCEPTED MANUSCRIPT

team sports or in informal group PA (i.e., reporting “alone” in PA session(s) during a

typical week) were included in “individual PA” category.

Positive mental health. The Mental Health Continuum-Short Form (MHC-SF)

PT
questionnaire comprises 14 items that measure emotional, social and psychological well-

being. Participants indicate how often during the last month they felt a certain way using

RI
a 6-point Likert scale (0-5): never, rarely, a few times, often, most of the time, and all the

C
time. A continuous score (range 0-70) was computed that measures overall well-being.

US
The MHC-SF subscales have good internal consistency and reliability, and sex-

invariance of its structure has been demonstrated; assessment of MHC-SF psychometrics

AN
in the current study has been published earlier (Doré, O’Loughlin, Sabiston, & Fournier,

2016). In this study, we found acceptable internal consistency for all three subscales (α
M

ranging from 0.78 to .90) and for the total scale (α = .91).
D
TE

Anxiety symptoms and depressive symptoms. The Hospital Anxiety and Depressive Scale

(HADS), a brief screening questionnaire referencing the previous seven days, is widely
EP

used to identify probable cases of anxiety and depressive disorders. It comprises two

subscales; 7 items measure anxiety (HADS-A) and 7 items measure depression

C

symptoms (HADS-D). Each item is scored on a 4-point Likert scale (0-3), with scores
AC

ranging from 0 to 21 for each subscale. Higher scores indicate greater distress and a

higher probability of having a disorder. The French-Canadian version of the scale has

good reliability and good discriminant validity, and a two-factor structure reflecting

anxiety and depression factors was identified (Roberge et al., 2012). In the present study,

each subscale demonstrates adequate internal consistency (α = .81 for HADS-A and α =
8
 ACCEPTED MANUSCRIPT

.70 for HADS-D).

Social connectedness in PA. The Relatedness to Others in Physical Activity Scale

(ROPAS) assesses perceived relatedness to others in leisure-time PA, regardless of PA

PT
context (Wilson & Bengoechea, 2010). The ROPAS is a self-administered 6-item scale

RI
with each statement representing feelings people have when they engage in PA; for each

item, participants indicate their level of agreement on a 6-point Likert scale (1-6). A total

C
score (5-30) provides a global assessment of the level of meaningful connection and

US
belonging experienced in PA. The ROPAS was translated into French and then back-

translated by two investigators not involved in this study, to ensure accuracy of the
AN
French version (See Appendix III). The 6 items from the ROPAS were internally

consistent in the present study (α = .91).

M
D

PA Volume. We estimated leisure-time MVPA in metabolic equivalents (METs) for all

TE

participants using the Godin-Shephard Leisure-Time Physical Activity Questionnaire

(GSLTPAQ) (Godin, 2011). Frequency of moderate and/or vigorous leisure-time PA of

EP

more than 15 minutes’ duration in a typical week was used to compute an MVPA score

by multiplying frequency by intensity (5 METS for moderate PA, 9 METS for vigorous
C

PA) and summing across intensities.

AC

Covariates. Based on the literature (Brunet et al., 2013; Jewett et al., 2014; McPhie &

Rawana, 2012; Sabiston et al., 2016), potential confounders of the associations

investigated included sex, age, and perceived socioeconomic status (SES), which was

9
 ACCEPTED MANUSCRIPT

measured by: How do you see your economic situation compared to other people your

age? (affluent, sufficient income, poor or very poor).

Statistical Analysis

PT
Multivariate linear regressions were used to model the longitudinal associations between

RI
PA context at baseline and each mental health outcome at follow-up, adjusting for

covariates and the value of the outcome at baseline. Evidence suggests that mental health

C
and symptoms of mental disorders can be both a cause and a consequence of PA; by

US
using a longitudinal design and adjusting the analysis for the values of the outcomes at

baseline, this study investigated the direct relationship of PA context as an antecedent

AN
cause of mental health outcomes. Confidence intervals were calculated in parametric

regression models. The significance level was set at 0.05.

M
D

Mediation analyses were performed to investigate possible mechanisms on the causal

TE

pathway between the exposure and outcome according to causal inference methods

(VanderWeele & Vansteelandt, 2009), using the value of the mediators at baseline. As in
EP

all methods of mediation analysis, the direct effect represents the component of an effect

that does not pass through a mediator whereas the indirect effect represents the
C

component of an effect passing through the mediator. However, contrary to standard

AC

approaches (i.e. the difference method, the product method (Baron & Kenny, 1986)), the

causal inference approach to mediation analysis estimates the controlled direct (CDE),

natural direct and indirect effects (NDE, NIE), allowing for the assessment of direct and

indirect effects even in the presence of interaction between the exposure and mediator

10
 ACCEPTED MANUSCRIPT

which, if neglected, can lead to severely biased estimates (Valeri & VanderWeele, 2013).

The CDE represents the change in the outcome if the exposure was set at level a=1,

versus level a=0, when the mediator is fixed at a specific value. With continuous

mediators (i.e. PA volume, social connectedness), because no specific threshold is of

PT
interest, we fixed the mediator at the mean for the CDE computation. The NDE expresses

RI
how much the outcome would change if the exposure was set at level a=1, versus level

a=0 with the mediator set at the level it would have been in the absence of the exposure.

C
The NIE expresses how much the outcome would change on average if the exposure was

US
controlled at level a=1 but the mediator was changed from the level it would take if a=0

to the level it would take if a=1. The total effect (TE) decomposes into the sum of the
AN
natural direct and indirect effects. The mediated portion of the association is expressed by

the ratio of the natural indirect effect on the total effect (NIE/TE) (VanderWeele &
M

Vansteelandt, 2009). The CDE, NDE, and NIE were estimated using parametric models
D

that accommodate exposure-mediator interaction (Valeri & VanderWeele, 2013). A first

TE

linear regression on the outcome (Y) included the two levels of exposure, the mediator,

covariates and product terms for the interactions; in the equation, the level of exposure is
EP

represented by a dummy variable where a (individual PA) is the reference, and b

(informal group PA) and c (team sports) are indicator variables (Appendix IV). The
C

second regression models the effect of the two levels of exposure and the covariates on
AC

the mediator. The CDE is estimated by fixing the mediator to the mean for each level of

exposure (i.e. informal group PA and team sports). NDE and NIE are estimated for each

level of the exposure variable using the appropriate equations. Confidence intervals for

the CDE, NDE and NIE estimates were calculated by nonparametric bootstrap

11
 ACCEPTED MANUSCRIPT

resampling. All statistical analyses were performed using R Software Version 1.0.153.

Results

The analytic sample included 430 students who reported at least one PA session per week

PT
(93.5% of 460 participants with follow-up data); inactive students (n = 30) were

RI
excluded. Participants were age 16-39y (M = 18.5, SD = 2.6); 97% were between ages

16-24. According to the PA context variable, 34.4% reported PA involvement in team

C
sports. Among those not involved in team sports, 41.2% reported PA in informal group at

US
least once a week and 24.4% reported individual PA. Table 1 presents descriptive

statistics across PA contexts. Using ANOVA, a statistically significant mean difference

AN
was observed for social connectedness (p<.0005) and MVPA (p<.0005) across PA

contexts. Using bivariate Pearson’s correlations, social connectedness was positively

M

associated with positive mental health (r=.393, p<.0005) and inversely associated with
D

anxiety symptoms (r=-.307, p<.0005) and depressive symptoms (r=-.268, p<.0005).

TE

Similar findings were observed for MVPA, suggesting a positive association with

positive mental health (r=.192, p<.0005) and negative associations with anxiety (r=-.161,
EP

p=.001) and depressive symptoms (r=-.100, p=.037).

C

[Insert Table 1]
AC

The multivariate models (Table 2), which do not include the hypothesized mediators,

describe the total effect of PA context on the mental health outcomes (i.e., the effect that

passes through all known and unknown mediating factors). PA in informal group and

team sports was associated with higher positive mental health scores (β (95% CI) = 2.24
12
 ACCEPTED MANUSCRIPT

(0.01, 4.46) and 3.39 (0.74, 5.59), respectively) and with fewer depressive symptoms (-

0.65 (-1.29, -0.01) and -0.76 (-1.43, -0.09), respectively), after adjustment for covariates

and baseline values of the outcome. No association was observed between PA context

and anxiety symptoms. Mediation paths between these variables were not tested, as direct

PT
and indirect effects would need to go in opposite directions (and be of a similar

RI
magnitude) for their combined effect to result in the absence of a total effect, which

seems unlikely.

C
US
[Insert Table 2] AN
Table 3 presents the CDE, NDE and NIE estimates for the mediated effects by social

connectedness and PA volume. The estimates for models with social connectedness as a
M

potential mediator were not statistically significant (Model 1), suggesting that social
D

connectedness is not a mediator of the association between PA contexts and positive

TE

mental health or depressive symptoms. In models with PA volume as a potential

mediator, only the CDE for the association between team sports and positive mental
EP

health was significant (Model 2). This result suggests that if PA volume was fixed at the

mean (27.5 (20.5) METs) for all participants, assigning individuals to exercise in a team
C

sport context would produce a mean positive mental health score that is 2.72 points (0.03,
AC

5.34) higher than assigning individuals to individual PA. This finding suggests that there

is no benefit from increasing PA volume; it is rather the selection of team sports context

that provides the benefit on positive mental health. The NDEs for the team sport and

positive mental health and depression associations indicate that if PA volume was fixed

13
 ACCEPTED MANUSCRIPT

at values they would have taken if each individual exercised in a team sport context, the

assignment of all participants to team sports would produce a higher mean positive

mental health score (3.39 (0.52, 6.09)) and a mean lower depressive symptom score (–

0.89 (-1.59, -0.17)) than the assignment of all participants to individual PA. These results

PT
are expressed assuming no unmeasured confounding of the exposure-outcome and

RI
mediator-outcome associations, and the NDE and NIE rely on the assumptions of no

unmeasured confounding of the exposure-mediator association and absence of

C
confounders affected by the exposure (Valeri & VanderWeele, 2013).

[Insert Table 3]
US
AN
Discussion
M

In this study, PA in team sports or in informal group was longitudinally associated with
D

enhanced positive mental health and with fewer depressive symptoms, compared to
TE

individual PA among youth. Further, in bivariate analysis, PA in team sports or in

informal group was associated with higher levels of social connectedness and PA
EP

volume, and social connectedness and PA volume were positively associated with

positive mental health and inversely associated with anxiety and depressive symptoms.
C

Although neither social connectedness nor PA volume appears to mediate the

AC

associations of interest, the CDE estimate in the mediation model for PA volume suggests

that team sports are associated with positive mental health regardless of PA volume. This

controlled direct relationship suggests that features of the team sport context, excluding

PA volume, explain its effect on positive mental health.

14
 ACCEPTED MANUSCRIPT

To date, relatively few studies distinguish PA contexts in relation to mental health or

investigate PA group contexts other than team sport. Informal group PA in particular has

received little empirical attention, although recent studies consistently report decreases in

PT
team sport participation with age, and increases in less structured group PA (Eime et al.,

RI
2015). This current study also expands on previous research by examining possible

mechanisms underpinning the PA context mental health association, using a measure of

C
social connectedness specific to PA.

US
Our findings corroborate previous studies suggesting increased benefits of team sports or
AN
clubs over individual PA on positive mental health (Jewett et al., 2014), health-related

quality of life (Eime et al., 2010; Vella et al., 2014), depressive symptoms (Brunet et al.,
M

2013; Sabiston et al., 2016) and other psychological and social outcomes (Eime, Young,
D

Harvey, Charity, & Payne, 2013). The positive longitudinal effect of PA in social
TE

contexts (e.g. informal group, team sports) on mental health among youth detected in this

study, aligns with the protective effect of team sports for health-related quality of life
EP

previously observed among children (Vella et al., 2014). In addition, this is the first study

in youth to demonstrate that, compared to individual PA, a non-organized PA context

C

(i.e., informal group PA) is associated with better mental health and fewer depressive
AC

symptoms six months later. Contrary to previous studies, no association was found

between team sports and anxiety symptoms (Eime et al., 2013). Divergent findings may

relate to the wide spectrum of anxiety disorders. Sthröle (2009) found an inverse

association between PA and agoraphobia, specific phobia and posttraumatic stress

15
 ACCEPTED MANUSCRIPT

disorder, but no association with panic, obsessive-compulsive or generalized anxiety

disorders (Ströhle, 2009).

To provide better understanding of possible causal mechanisms, we investigated the

PT
possible mediating role of social connectedness and PA volume in the associations of

RI
interest. Relative to those who engaged in individual PA, being active in informal group

PA or sports team was associated with higher levels of social connectedness and higher

C
MVPA volume. In mediation analysis, the hypothesis that PA context influences mental

US
health indirectly through social connectedness was not confirmed.
AN
The hypothesis that PA volume mediates the association between PA context and mental

health was also not supported, although team sport was associated with positive mental
M

health when PA volume was fixed at the sample mean. Using mediation analyses in a
D

causal inference framework, the controlled direct effect identified in this study represents
TE

a unique contribution by suggesting that components of the team sports context other

than PA volume, might underpin these associations. Since PA volume could reflect
EP

biological mechanisms, we can hypothesize that other mechanisms are implicated. For

example, team sports might influence mental health by promoting psychosocial attributes
C

known for their benefits on mental health including higher levels of self-confidence
AC

(Zeng, 2003), better self-esteem (McHale et al., 2005), higher social support (Babiss &

Gangwisch, 2009) or a wide and supportive social network (McNeill, Kreuter, &

Subramanian, 2006).

16
 ACCEPTED MANUSCRIPT

Our findings can inform mental health programs and policy. The transition to adulthood

is a critical life period requiring adaptation to changes at the personal, interpersonal,

academic and professional levels, and it represents a period when mental disorders often

present. Social relationships and a sense of belonging have powerful effects on mental

PT
health (Berkman, 2001; Hendry & Reid, 2000) but lack of knowledge on how to promote

RI
social connectedness among youth is a major challenge. Our data support that PA may be

an effective intervention pathway and provide evidence on how PA guidelines could be

C
adapted to improve mental health. The Canadian Society for Exercise Physiology

US
currently recommends that adults engage in ≥150 minutes of aerobic MVPA per week to

achieve health benefits. These guidelines focus on PA volume but may also need to
AN
incorporate PA social context to optimize mental health. Clubs, team sports and informal

group PA opportunities could be promoted in diverse settings targeting all youth

M

regardless of PA skill level, who want the benefits of PA social contexts. Moreover,
D

because PA levels in childhood track to adolescence and young adulthood (Bélanger et

TE

al., 2015), our findings suggest that it may be important from a public health perspective,

to encourage children to engage in social PA contexts. More research is needed to

EP

elucidate which components of group PA experiences (in team sport, informal group or

other social contexts) explain the positive impact on mental health. Future research
C

should query specific components of PA in social contexts such as the size of the group,
AC

frequency of sessions, characteristics of leaders, coaches and trainers, opportunities for

social interaction outside the PA context, and social support provided by team or group

members. This kind of evidence will help identify which components of PA in social

contexts should be incorporated into PA interventions that aim to improve mental health

17
 ACCEPTED MANUSCRIPT

in youth. Level of competitiveness and training intensity should also be taken into

account because there is evidence that these aspects might threaten mental health (Tao et

al., 2007; Tobar, 2012).

PT
Strengths of this study include a longitudinal design, use of a comprehensive measure of

RI
positive mental health and use of a measure of social connectedness specific to PA.

Previous studies used an assessment of positive mental health restricted to emotional

C
(Hinkley et al., 2014) or psychological well-being (Edwards, Ngcobo, Edwards, &

US
Palavar, 2005) and they included only general measures of social support (McHugh &

Lawlor, 2012). Use of causal inference methods for mediation analysis provides more
AN
accurate assessment of the influence of potential mediators by allowing for interaction

between the exposure and mediator, and for decomposition of the total effect into natural
M

direct and indirect effects (VanderWeele & Vansteelandt, 2009). Other studies
D

investigating mechanisms did not use methods specifically intended to assess indirect
TE

effects (e.g., multivariate regression adjusting for the potential mediator) or used

mediation analysis methods subject to bias (e.g., difference or product methods) (Babiss
EP

& Gangwisch, 2009; Monshouwer et al., 2013).

C

Limitations include that the temporality between exposure and mediator variables could
AC

not be established since both variables were measured at baseline. The relatively small

sample size may have limited our ability to detect significant effects. The sample was

restricted to post-secondary students in Quebec, which may limit the generalizability of

the findings. Because of possible selection bias related to the low response at follow-up,

18
 ACCEPTED MANUSCRIPT

the results need replication in other settings and populations. We used self-report

measures, which are subject to misclassification (eg. self-report PA might overestimate

MVPA) (James et al., 2016). Finally, we did not study the validity of the PA context

variable, which might not differentiate levels of social interaction in PA accurately.

PT
RI
Conclusion

The results of this study indicate that promoting PA in organized (i.e., team sports) and

C
non-organized (i.e., informal group PA) social contexts, is likely more effective than

US
individual PA, in enhancing positive mental health and reducing depressive symptoms

among youth in transition to adulthood. Although replication of these findings is

AN
necessary, our results provide support for the development of PA interventions in social

contexts that promote social interactions among youth which in turn, may influence
M

mental health positively. The benefit of team sports on positive mental health and
D

depressive symptoms is not explained by increased PA volume so that more research is

TE

needed to identify the contribution of other (social) mechanisms implicated in these

associations.
EP

Conflict of interest The authors declare that there are no conflicts of interest.
C
AC

Authors contributors' statement ID designed the study, managed data collection,

performed the analysis, interpreted results and wrote the manuscript. MS contributed to

the statistical analysis. All authors contributed to the interpretation of the results,

reviewed and revised the manuscript, and approved the final manuscript as submitted.

19
 ACCEPTED MANUSCRIPT

Acknowledgements For the duration of the study, ID was supported by doctoral awards

from the 4P Strategic Training Program funded by the Canadian Institutes of Health

Research and the Réseau de recherche en santé des populations du Québec, the Fonds de

PT
recherche du Québec - Société et culture, the School of Public Health and the Faculté des

RI
études supérieures et postdoctorales of the Université de Montréal. ID currently holds a

postdoctoral fellowship from the Fond de recherche du Québec – Santé. JOL holds a

C
Canada Research Chair in the Early Determinants of Adult Chronic Disease. MS holds a

US
new investigator salary award from the Canadian Institutes of Health Research. GDD is

supported by a cancer prevention salary award from the Canadian Cancer Society (award
AN
#703946). The authors thank CEGEP de l’Outaouais staff, Marc Martineau and Guy

Beauchamp for valuable help with data collection.

M
D

Funding This research did not receive any specific grant from funding agencies in the
TE

public, commercial, or not-for-profit sectors.

EP

References

Amireault, S., Godin, G., Lacombe, J., & Sabiston, C. M. (2015). The use of the Godin-
C

Shephard Leisure-Time Physical Activity Questionnaire in oncology research: a

systematic review. BMC Medical Research Methodology, 15(1), 1.
AC

Armstrong, S., & Oomen-Early, J. (2009). Social connectedness, self-esteem, and

depression symptomatology among collegiate athletes versus nonathletes. Journal of
American College Health, 57(5), 521-526.

Asztalos, M., Wijndaele, K., De Bourdeaudhuij, I., Philippaerts, R., Matton, L.,
Duvigneaud, N., . . . Cardon, G. (2009). Specific associations between types of physical
activity and components of mental health. Journal of Science and Medicine in Sport,
12(4), 468-474.

20
 ACCEPTED MANUSCRIPT

Babiss, L. A., & Gangwisch, J. E. (2009). Sports participation as a protective factor

against depression and suicidal ideation in adolescents as mediated by self-esteem and
social support. Journal of Developmental & Behavioral Pediatrics, 30(5), 376.

Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in

social psychological research: Conceptual, strategic, and statistical considerations.
Journal Of Personality And Social Psychology, 51(6), 1173.

PT
Bélanger, M., Sabiston, C. M., Barnett, T. A., O’Loughlin, E., Ward, S., Contreras, G., &
O’Loughlin, J. (2015). Number of years of participation in some, but not all, types of
physical activity during adolescence predicts level of physical activity in adulthood:

RI
Results from a 13-year study. International Journal of Behavioral Nutrition and Physical
Activity, 12(1), 76.

C
Berkman, L. F. (2001). Social ties and mental health. Journal of Urban health, 78(3),
458-467.

US
Brunet, J., Sabiston, C. M., Chaiton, M., Barnett, T. A., O'Loughlin, E., Low, N. C., &
O'Loughlin, J. L. (2013). The association between past and current physical activity and
depressive symptoms in young adults: a 10-year prospective study. Annals of
AN
Epidemiology(23), 25-30.

Chaouloff, F. (1997). The serotonin hypothesis. In M. W. P. (Ed.), Physical Activity And

M

Mental Health (pp. 179-198). Washington, D.C: Taylor and Francis.

Da Silva, M. A., Singh-Manoux, A., Brunner, E. J., Kaffashian, S., Shipley, M. J.,
Kivimäki, M., & Nabi, H. (2012). Bidirectional association between physical activity and
D

symptoms of anxiety and depression: the Whitehall II study. European Journal of

Epidemiology, 27(7), 537-546.
TE

Deci, E. L., & Ryan, R. M. (2002). Handbook of self-determination research: University

Rochester Press.
EP

Doré, I., O'Loughlin, J. L., Beauchamp, G., Martineau, M., & Fournier, L. (2016).
Volume and social context of physical activity in association with mental health, anxiety
and depression among youth. Preventive Medicine, 91, 344-350.
C

Doré, I., O’Loughlin, J. L., Sabiston, C. M., & Fournier, L. (2016). Psychometric
AC

Evaluation of the Mental Health Continuum–Short Form (MHC-SF) in French Canadian

Young Adults. The Canadian Journal of Psychiatry, 0706743716675855.

Edwards, S. D., Ngcobo, H. S., Edwards, D. J., & Palavar, K. (2005). Exploring the
relationship between physical activity, psychological well-being and physical self-
perception in different exercise groups. South African Journal for Research in Sport,
Physical Education and Recreation, 27(1), 75-90.

21
 ACCEPTED MANUSCRIPT

Eime, R. M., Harvey, J. T., Brown, W. J., & Payne, W. R. (2010). Does sports club
participation contribute to health-related quality of life? Medicine and Science in Sports
and Exercise, 42(5), 1022-1028.

Eime, R. M., Young, J. A., Harvey, J. T., Charity, M. J., & Payne, W. R. (2013). A
systematic review of the psychological and social benefits of participation in sport for
children and adolescents: informing development of a conceptual model of health

PT
through sport. International Journal of Behavioral Nutrition and Physical Activity,
10(98), 1.

Eime, R. M., Sawyer, N., Harvey, J., Casey, M., Westerbeek, H., & Payne, W. (2015).

RI
Integrating public health and sport management: sport participation trends 2001–2010.
Sport Management Review, 18(2), 207-217.

C
Faulkner, G., & Carless, D. (2006). Physical activity in the process of psychiatric
rehabilitation: theoretical and methodological issues. Psychiatric Rehabilitation Journal,

US
29(4), 258-266.

Godin, G. (2011). The Godin-Shephard leisure-time physical activity questionnaire. The

Health & Fitness Journal of Canada, 4(1), 18-22.
AN
Hallgren, M., Herring, M. P., Owen, N., Dunstan, D., Ekblom, Ö., Helgadottir, B., . . .
Forsell, Y. (2016). Exercise, Physical Activity, and Sedentary Behavior in the Treatment
M

of Depression: Broadening the Scientific Perspectives and Clinical Opportunities.

Frontiers in Psychiatry, 7.

Hendry, L. B., & Reid, M. (2000). Social relationships and health: the meaning of social
D

‘‘connectedness’’ and how it relates to health concerns for rural Scottish adolescents.
Journal of Adolescence, 23, 705-719.
TE

Hinkley, T., Teychenne, M., Downing, K. L., Ball, K., Salmon, J., & Hesketh, K. D.
(2014). Early childhood physical activity, sedentary behaviors and psychosocial well-
being: a systematic review. Preventive Medicine, 62, 182-192.
EP

Holt, N. L. (2016). Positive Youth Development Through Sport: Routledge.

C

James, P., Weissman, J., Wolf, J., Mumford, K., Contant, C. K., Hwang, W.-T., . . .
Glanz, K. (2016). Comparing GPS, Log, Survey, and Accelerometry to Measure Physical
AC

Activity. American Journal of Health Behavior, 40(1), 123-131.

Jewett, R., Sabiston, C. M., Brunet, J., O'Loughlin, E. K., Scarapicchia, T., &
O'Loughlin, J. (2014). School sport participation during adolescence and mental health in
early adulthood. Journal of Adolescent Health, 55(5), 640-644.

Kamimura, A., Christensen, N., Al-Obaydi, S., Solis, S. P., Ashby, J., Greenwood, J. L.,
& Reel, J. J. (2014). The relationship between body esteem, exercise motivations,
depression, and social support among female free clinic patients. Women's Health Issues,
24(6), 656-662.

22
 ACCEPTED MANUSCRIPT

Kremer, P., Elshaug, C., Leslie, E., Toumbourou, J. W., Patton, G. C., & Williams, J.
(2014). Physical activity, leisure-time screen use and depression among children and
young adolescents. Journal of Science and Medicine in Sport, 17(2), 183-187.

Kull, M., Ainsaar, M., Kiive, E., & Raudsepp, L. (2012). Relationship between low
depressiveness and domain specific physical activity in women. Health care for women
international, 33(5), 457-472.

PT
Kwan, B. M., Davis, K. J., & Dunn, A. L. (2012). Physical Activity for the Prevention of
Depression. Physical Activity Across the Lifespan, 97-119.

RI
Lee, R. M., & Robbins, S. B. (1998). The relationship between social connectedness and
anxiety, self-esteem, and social identity.

C
Mammen, G., & Faulkner, G. (2013). Physical activity and the prevention of depression:
a systematic review of prospective studies. American Journal of Preventive Medicine,

US
45(5), 649-657.

McHale, J. P., Vinden, P. G., Bush, L., Richer, D., Shaw, D., & Smith, B. (2005).
Patterns of personal and social adjustment among sport-involved and noninvolved urban
AN
middle-school children. Sociology of Sport Journal, 22(2), 119-136.

McHugh, J. E., & Lawlor, B. A. (2012). Exercise and social support are associated with
M

psychological distress outcomes in a population of community-dwelling older adults.

Journal of Health Psychology, 17(6), 833-844.

McNeill, L. H., Kreuter, M. W., & Subramanian, S. (2006). Social environment and
D

physical activity: a review of concepts and evidence. Social Science & Medicine, 63(4),
1011-1022.
TE

McPhie, M. L., & Rawana, J. S. (2012). Unravelling the relation between physical
activity, self-esteem and depressive symptoms among early and late adolescents: A
mediation analysis. Mental Health and Physical Activity, 5(1), 43-49.
EP

Monshouwer, K., ten Have, M., van Poppel, M., Kemper, H., & Vollebergh, W. (2013).
Possible mechanisms explaining the association between physical activity and mental
C

health findings from the 2001 Dutch Health Behaviour in School-Aged Children survey.
Clinical Psychological Science, 1(1), 67-74.
AC

Roberge, P., Doré, I., Menear, M., Chartrand, É., Ciampi, A., Duhoux, A., & Fournier, L.
(2012). A psychometric evaluation of the French Canadian version of the Hospital
Anxiety and Depression Scale in a large primary care population. Journal of Affective
Disorders.

23
 ACCEPTED MANUSCRIPT

Sabiston, C. M., Jewett, R., Ashdown-Franks, G., Belanger, M., Brunet, J., O’Loughlin,
E., & O'Loughlin, J. (2016). Number of Years of Team and Individual Sport Participation
During Adolescence and Depressive Symptoms in Early Adulthood. Journal of Sport and
Exercise Psychology, 38(1).

Sallis, J. F., Cervero, R. B., Ascher, W., Henderson, K. A., Kraft, M. K., & Kerr, J.
(2006). An ecological approach to creating active living communities. Annu Rev Public

PT
Health, 27, 297-322.

Shephard, R. J. (2003). Limits to the measurement of habitual physical activity by

RI
questionnaires. British Journal of Sports Medicine, 37(3), 197-206.

Ströhle, A. (2009). Physical activity, exercise, depression and anxiety disorders. Journal
of Neural Transmission, 116(6), 777-784.

C
Tao, F. B., Xu, M. L., Kim, S. D., Sun, Y., Su, P. Y., & Huang, K. (2007). Physical

US
activity might not be the protective factor for health risk behaviours and
psychopathological symptoms in adolescents. Journal of Paediatrics and Child Health,
43(11), 762-767.
AN
Tesser, A. (1991). Social versus clinical approaches to self psychology: The self-
evaluation maintenance model and Kohutian object relations theory. The relational self:
Theoretical convergences in psychoanalysis and social psychology, 257-281.
M

Tobar, D. A. (2012). Trait anxiety and mood state responses to overtraining in men and
women college swimmers. International Journal of Sport and Exercise Psychology,
D

10(2), 135-148.

Valeri, L., & VanderWeele, T. J. (2013). Mediation analysis allowing for exposure–
TE

mediator interactions and causal interpretation: Theoretical assumptions and

implementation with SAS and SPSS macros. Psychological Methods, 18(2), 137.

Valois, R. F., Umstattd, M. R., Zullig, K. J., & Paxton, R. J. (2008). Physical Activity
EP

Behaviors and Emotional Self‐Efficacy: Is There a Relationship for Adolescents?

Journal of School Health, 78(6), 321-327.
C

VanderWeele, T., & Vansteelandt, S. (2009). Conceptual issues concerning mediation,

interventions and composition. Statistics and its Interface, 2, 457-468.
AC

Vella, S. A., Cliff, D. P., Magee, C. A., & Okely, A. D. (2014). Sports participation and
parent-reported health-related quality of life in children: longitudinal associations. The
Journal of Pediatrics, 164(6), 1469-1474.

White, R. L., Babic, M. J., Parker, P. D., Lubans, D. R., Astell-Burt, T., & Lonsdale, C.
(2017). Domain-specific physical activity and mental health: a meta-analysis. American
Journal of Preventive Medicine, 52(5), 653-666.

24
 ACCEPTED MANUSCRIPT

Wilson, P. M., & Bengoechea, E. G. (2010). The relatedness to others in physical activity
scale: evidence for structural and criterion validity. Journal of Applied Biobehavioral
Research, 15(2), 61-87.

Wipfli, B., Landers, D., Nagoshi, C., & Ringenbach, S. (2011). An examination of
serotonin and psychological variables in the relationship between exercise and mental
health. Scandinavian Journal of Medicine & Science in Sports, 21(3), 474-481.

PT
Zeng, H. Z. (2003). The differences between anxiety and self-confidence between team
and individual sports college varsity athletes. International Sports Journal, 7(1), 28.

C RI
US
AN
M
D
TE
C EP
AC

25
 ACCEPTED MANUSCRIPT

Figure 1. Directed Acyclic Graph of the Hypothesized Associations. PA context affects positive mental
health, anxiety symptoms and depressive symptoms through the creation or enhancement of social
connectedness in PA and/or through increasing PA volume. Sex, age and perceived socioeconomic status
are portrayed as confounders.

PA Volume

PT
Social
connectedness

RI
Positive mental health
PA Context Anxiety symptoms

C
Depressive symptoms

US
Sex, age, perceived
socioeconomic
AN
status
M
D
TE
C EP
AC

26
 ACCEPTED MANUSCRIPT

Table 1. Selected Characteristics of Participants by Physical Activity (PA) Context, Québec, Canada (n=430)

PA Context
Total Team sports Informal group Individual
n=430 n=148 n=177 n=105
Male (%) 35.5 43.9 33.9 30.5
Age, y, (%)

PT
16-17 38.1 43.9 35.6 34.3
18-19 46.5 48.6 45.2 45.7
≥20 15.3 7.4 19.2 20.0
SES perception, (%)

RI
Affluent 36.5 42.6 36.7 27.6
Sufficient income 50.7 47.3 49.2 58.1
Poor/very poor 12.8 10.1 14.1 14.3

C
MVPA, METs (mean, SD) 27.5 (20.5) 37.0 (21.5) 24.1 (17.4) 19.7 (18.8)
Social connectedness (mean, SD) 26.9 (6.7) 29.9 (4.9) 26.9 (6.4) 22.9 (7.3)

US
Positive mental health (mean, SD) 46.6 (12.8) 49.4 (12.4) 46.3 (12.2) 42.9 (13.5)
Anxiety symptoms (mean, SD) 7.2 (4.2) 6.3 (3.9) 7.6 (4.3) 7.9 (4.4)
Depressive symptoms (mean, SD) 4.1 (3.2) 3.6 (3.0) 4.0 (3.0) 4.9 (3.8)
AN
SES = socioeconomic status, SD = Standard deviation, MVPA = Moderate-to-vigorous physical activity, METs =
metabolic equivalents.
M

Table 2. Beta Coefficients and 95% Confidence Intervals (CI) for Positive Mental Health, Anxiety Symptoms
D

and Depressive Symptoms by Physical Activity (PA) Context, Québec, Canada (n =430)
TE

Positive Mental Anxiety Depressive

Health Symptoms symptoms
PA context β β β
(95% CI) (95% CI) (95% CI)
EP

Individual Ref Ref ref

Informal group 2.24 (0.01, 4.46) 0.14 (-0.61, 0.89) -0.65 (-1.29, -0.01)
Team sports 3.39 (0.74, 5.59) -0.47 (-1.27, 0.32) -0.76 (-1.43, -0.09)
C

Bold indicates statistically significant results (p<0.05)

Models adjusted for values of the outcome at baseline, age, sex and perceived socioeconomic status,
AC

CI = Confidence Intervals, METs = metabolic equivalents.

27
 ACCEPTED MANUSCRIPT

Table 3. Controlled Direct Effect (CDE), Natural Direct Effect (NDE), and Natural Indirect Effect
(NIE) estimates and 95% Confidence Intervals (CI) for Social Connectedness and Physical Activity (PA)
Volume as Potential Mediators of the Associations Between PA Context and Positive Mental Health and
Depressive Symptoms, Québec, Canada (n =430)

PA Context CDE NDE NIE

PT
β (95% CI) β (95% CI) β (95% CI)
Model 1 - Mediator = Social connectedness
Positive mental health Individual ref ref ref

RI
Informal group 1.74 (-0.64, 4.41) 1.73 (-0.60, 4.30) 0.52 (-0.30, 1.42)
Team sports 2.38 (-0.52, 5.29) 2.11 (-0.54, 4.94) 1.28 (-0.68, 3.15)
Depressive symptoms Individual ref ref ref

C
Informal group -0.40 (-1.56, 0.35) -0.53 (-1.26, 0.25) -0.11 (-0.37, 0.11)
Team sports -0.41 (-1.29, 0.36) -0.51 (-1.63, 0.62) -0.24 (-0.97, 0.51)

US
Model 2 - Mediator = PA volume
Positive mental health Individual ref ref ref
Informal group 1.61 (-0.61, 3.85) 1.96 (-0.20, 4.19) 0.14 (-0.11, 0.56)
AN
Team sports 2.72 (0.03, 5.34) 3.39 (0.52, 6.09) -0.15 (-1.04, 0.78)
Depressive symtoms Individual ref ref ref
Informal group -0.55 (-1.28, 0.12) -0.59 (-1.28, 0.07) -0.05 (-0.16, 0.04)
Team sports -0.71 (-1.47, 0.12) -0.89 (-1.59, -0.17) 0.15 (-0.12, 0.38)
M

Bold indicates statistically significant results (p<0.05)

CDE = Controlled Direct Effect, NDE = Natural Direct Effect, NIE = Natural Indirect Effect, CI = Confidence
Intervals, PA = physical activity
D
TE
C EP
AC

28
 ACCEPTED MANUSCRIPT

Appendix I. Comparison of selected baseline characteristics of participants with and

without follow-up data

Without follow- With follow-up

up data data p-value
(n=1,067) (n=460)
Male (%) 43.9 37.0 0.01

PT
Age, y (%)
16-17 38.9 37.4 0.79
18-19 44.9 46.7
14.8 15.4

RI
≥20
SES (%)
Affluent 38.3 36.7 0.74

C
Sufficient income 47.1 49.6
Poor/very poor 12.7 13.3
PA context (%)

US
Inactive 5.2 3.7 0.06
Individual 17.8 23.3
Informal group 38.6 39.1
AN
Team sports 37.2 33.7
MVPA, METs (mean, SD) 28.3 (22.9) 26.7 (21.5) 0.23
Social connectedness (mean, SD) 27.0 (7.2) 26.8 (6.7) 0.63
Positive mental health (mean, SD) 46.8 (11.8) 46.2 (11.9) 0.39
M

Anxiety symptoms (mean, SD) 7.3 (4.0) 7.2 (4.1) 0.73

Depressive symptoms (mean, SD) 4.0 (3.0) 3.8 (3.1) 0.19
D

Bold indicates statistically significant results (p<0.05)

SES = socioeconomic status, PA = physical activity, SD = Standard deviation, MVPA = Moderate-to-vigorous
physical activity.
TE
C EP
AC

29
 ACCEPTED MANUSCRIPT

Appendix II. Godin-Shepherd Leisure-Time Physical Activity Questionnaire (GSLTPAQ) –

Modified Version

During a typical 7-day period (week), how many times on average do you do the following kinds of exercises for more
than 15 minutes during your free time?

Record the TOTAL number of times per week you do each type of physical activity. Of this total, record how many

PT
times you do this type of physical activity ALONE, WITH SOMEONE and IN A GROUP (3 or more people).

Please fill in all the blanks.

blanks.
If you do not do this type of physical activity, write 0.
0.

RI
STRENUOUS EXERCISE

C
(Heart beats rapidly) ___ times per week ALONE
(e.g., running, jogging, cross country

US
skiing, basketball, soccer, vigorous ___ TOTAL times per week ___ times per week WITH SOMEONE
swimming, vigorous long distance
biking, vigorous gym exercises, ___ times per week in a GROUP
aerobics, kickboxing, bootcamp…)
AN
MODERATE EXERCISE
___ times per week ALONE
(Not exhausting)
(e.g., fast walking, tennis, badminton, ___ TOTAL times per week ___ times per week WITH SOMEONE
M

dance, easy bicycling, alpine skiing…)

___ times per week in a GROUP
D

LIGHT EXERCISE
___ times per week ALONE
(Minimal effort)
(e.g., easy walking, yoga, golf, curling…) ___ TOTAL times per week ___ times per week WITH SOMEONE
TE

___ times per week in a GROUP

C EP
AC

30
 ACCEPTED MANUSCRIPT

Appendix III. Relatedness to Others in Physical Activity Scale (ROPAS) – English and
French versions

Original (English) version

The following statements represent different feelings people have when they engage in physical activity. Please answer

PT
the following question by considering how YOU TYPICALLY feel when participating in physical activity using the scale
provided.

More More

RI
Mostly Mostly
False false than true than True
false true
true false
I feel like I have developed a close bond

C
1 2 3 4 5 6
with others
I feel like I fit in well with others 1 2 3 4 5 6

US
I feel like I am included by others 1 2 3 4 5 6

I feel like I am part of a group who share

1 2 3 4 5 6
my goals
AN
I feel like I am supported by others in this
1 2 3 4 5 6
activity
I feel like others want me to be involved
1 2 3 4 5 6
with them
M
D

French Version
TE

Les énoncés suivants représentent différents sentiments que les gens ont lorsqu’ils pratiquent des activités physiques.
Indiquez comment vous vous sentez de MANIÈRE HABITUELLE lorsque vous pratiquez des activités physiques dans vos
temps libres.
EP

Plutôt Plus faux Plus vrai Plutôt

Faux Vrai
faux que vrai que faux vrai
Je sens que j’ai développé des liens étroits
1 2 3 4 5 6
avec les autres
C

Je sens que j’ai ma place avec les autres 1 2 3 4 5 6

AC

Je me sens inclus(e) par les autres 1 2 3 4 5 6

Je sens que je fais partie d’un groupe qui

1 2 3 4 5 6
partage mes objectifs
Je me sens supporté(e) par les autres dans
1 2 3 4 5 6
cette (ces) activité(s)
Je sens que les autres veulent que je
1 2 3 4 5 6
m’implique avec eux

31
 ACCEPTED MANUSCRIPT

Appendix IV. Equations for regression and computation of Controlled Direct Effect (CDE),
Natural Direct Effect (NDE) and Natural Indirect Effect (NIE)

Exposure variable categories (physical activity contexts):

a = individual
b = informal group
c = team sport

PT
Regression 1:
| , = , = = + + + + + + ’

RI
Regression 2:

C
| , = = + + + ’

US
CDE equations:
, ( = !") = ( + )( − !)
AN
, ( = !") = ( + )( − !)

NDE equations:
M

% , (0) = ( + ( + !+ ’ ))( − !)
% , (0) = ( + ( + !+ ’ ))( − !)
D

NIE equations:
TE

%' , (1) = ( + )( − !)
%' , (1) = ( + )( − !)
C EP
AC

32
 ACCEPTED MANUSCRIPT

Highlights

• Group PA and team sports are associated with increased mental health

• Group PA and team sports are associated with decreased depressive symptoms

• Social connectedness and PA volume significantly different across PA contexts

PT
• Benefit of participation in team sports on mental health is not affected by the volume

C RI
US
AN
M
D
TE
C EP
AC