CONTEXT:

Research regarding the protective effects of early physical activity on depression has yielded conflicting results.

OBJECTIVE:

Our objective was to synthesize observational studies examining the association of physical activity in childhood and adolescence with depression.

DATA SOURCES:

Studies (from 2005 to 2015) were identified by using a comprehensive search strategy.

STUDY SELECTION:

The included studies measured physical activity in childhood or adolescence and examined its association with depression.

DATA EXTRACTION:

Data were extracted by 2 independent coders. Estimates were examined by using random-effects meta-analysis.

RESULTS:

Fifty independent samples (89 894 participants) were included, and the mean effect size was significant (r = –0.14; 95% confidence interval [CI] = –0.19 to –0.10). Moderator analyses revealed stronger effect sizes in studies with cross-sectional versus longitudinal designs (k = 36, r = –0.17; 95% CI = –0.23 to –0.10 vs k = 14, r = –0.07; 95% CI = –0.10 to –0.04); using depression self-report versus interview (k = 46, r = –0.15; 95% CI = –0.20 to –0.10 vs k = 4, r = –0.05; 95% CI = –0.09 to –0.01); using validated versus nonvalidated physical activity measures (k = 29, r = –0.18; 95% CI = –0.26 to –0.09 vs k = 21, r = –0.08; 95% CI = –0.11 to –0.05); and using measures of frequency and intensity of physical activity versus intensity alone (k = 27, r = –0.17; 95% CI = –0.25 to –0.09 vs k = 7, r = –0.05; 95% CI = –0.09 to –0.01).

LIMITATIONS:

Limitations included a lack of standardized measures of physical activity; use of self-report of depression in majority of studies; and a small number of longitudinal studies.

CONCLUSIONS:

Physical activity is associated with decreased concurrent depressive symptoms; the association with future depressive symptoms is weak.

Research interest in the health and psychological benefits of exercise has grown exponentially over recent years. Evidence suggests that physical activity may ameliorate depressive symptoms, supporting the use of exercise as part of a comprehensive treatment plan for major depressive disorder (MDD).1,2 The reverse association has also been demonstrated: decreased physical activity (PA), as well as increased sedentary behaviors, confers vulnerability for developing depressive symptoms.3,5 To date, studies have investigated whether increased PA may also protect individuals against the development of MDD, and findings from observational studies are promising.3,6,8 However, the age range of participants in these studies has been wide, research has been conducted principally in adult populations, and results have been conflicting.9,11 Thus, using the current state of the literature for the purpose of clinical decision-making is challenging. A meta-analysis is warranted to resolve discrepancies in the literature and to examine the suggestion that the largest magnitude of protective effect may be found at younger ages,12 which would in turn provide support for a potential preventative role of physical activity in the development of depression.

Two recent systematic reviews13,14 have reported that increased PA is associated with fewer depressive symptoms. However, only 1 review focused on the child and adolescent age group,13 and neither review conducted a meta-analytic synthesis of the data, which can provide a powerful estimate of the mean effect size across studies. Compared with adult participants, in which the investigation of risk factors is confounded by years of the allostatic load of depression (exposure to depressive symptoms and their associated physiologic strain)15 and comorbid cardiometabolic disease,16 studies of children and adolescents enable the examination of the relationship between PA and depressive symptoms at their most nascent. To our knowledge, this is the first study to conduct a meta-analytic review of the protective effect of PA on depression and, as such, is the first to describe the magnitude of this association. Also, previous systematic reviews have not explored the potential moderating role of sex in the association between PA and MDD, although a stronger effect for females has been suggested in several independent studies.4,17,18 Understanding if the association between PA and MDD is sex-specific is relevant for the elucidation of potential underlying mechanisms of association.

The objective of this meta-analysis was to investigate the potential preventative effect of child and adolescent PA on depression. Several variables have been linked to differences in effects size; thus, we will examine whether between-study differences were observed for child age, sex, and social risk.19,21 We will also examine if heterogeneity in effect sizes can be explained by variation in study methodology (eg, methods of assessing physical activity and depression), as well as study quality (eg, longitudinal versus cross-sectional). Clarification on the role of these factors for understanding systematic differences in effect sizes are important for the design and implementation of targeted and effective public health prevention programs.

Published studies on PA and depression in children and adolescents were identified by searching Social Sciences Abstracts, International Bibliography of the Social Sciences, Scopus, SportDiscus, CBA Abstracts, Physical Education Index, Sociological Abstracts, and PsycINFO electronic databases for potential articles through October 2015. The search was limited to English language articles published between 2005 and 2015 using the keywords (“child*,” or “teen*,” or “adolesc*,” or “youth*,” or “infant,” or “infancy,” or “baby,” or “babies”) AND (“depress*”), AND (“sedentary behavio*” or “recreation” or “physical activity” or “leisure activity” or “exercise” or “fitness” or “sport*”). This search strategy yielded 3147 nonduplicate articles.

Titles and abstracts of the articles were reviewed to identify studies that met the inclusion criteria. Articles selected for the current study were based on the following criteria. (1) Cross-sectional study of PA and depression collected during childhood or adolescence (<18 years). (2) Longitudinal study of PA collected during childhood or adolescence (<18 years); (3) The constructs measured were PA (eg, energy expenditure) and depressive symptoms. Studies that measured broader, nonspecific constructs of either PA (eg, participation in extracurricular activities) or of depression (eg, psychological distress) were excluded. Because numerous standardized, validated and accessible measures of depression among youth are widely available, studies that assessed the outcome of depression by using a nonvalidated measure were excluded. Only 1 study22 needed to be excluded because it assessed depression by using a single self-report item with no demonstrated psychometric properties. In contrast to the depression literature, fewer standardized and validated measures exist for assessing physical activity. Thus, no validity criterion was applied to the measure of PA. However, a validated versus nonvalidated PA measure was examined as a moderator to determine if this measurement characteristic explained between study heterogeneity. (4) The study statistic could be transformed into an effect size (eg, correlations, odds ratios, means/SDs, and/or P values). (5) The full-text article was available and written in English. Studies in which PA was used as an intervention were not included in the current study.

Multiple results often emerge from a single dataset. If the same participants were used across multiple publications, only 1 study was included in the meta-analysis to ensure independence of effect sizes. A protocol was developed so that each sample of participants was only represented once in the meta-analysis. First, if a single dataset presented both cross-sectional and longitudinal analyses, we selected the study with longitudinal data because this study design was underrepresented in our analyses. Second, if multiple publications emerged from a single cross-sectional dataset, we selected the publication with the largest sample size and most comprehensive data extraction information.

Multiple samples or groups often exist within a particular study. For example, some studies present results separately for boys and girls within a sample. In such cases, effects sizes for both these nonoverlapping samples were calculated and entered into the meta-analysis separately.

All articles that met inclusion criteria were coded by using a standard coding form to collect information on study and sample characteristics. Several moderator variables were collected to explain effect size variability across studies. Moderator variables were divided into categorical moderators (sex, social risk [ie, low income, minority, or involved in child protective services], PA type, PA validated measure, depression measure type, study design, and country) and continuous moderators (age at PA/depression, time between assessments, and publication year). Some studies reported data stratified by level of PA. In such cases, data for the group with the greatest PA were used in the analysis. This was done to remain consistent with our primary objective. Data extraction was performed by 2 independent coders (DK and MC). Discrepancies were resolved through discussion, and consensus scores were entered into the final dataset.

Effect sizes were calculated and analyzed by using Comprehensive Meta-Analysis version 3.0 software.23 Effect sizes were calculated directly from information provided in each study. When provided, adjusted effect sizes were included. All effect sizes were transformed into correlations for the purpose of reporting mean effect sizes. Pooled effect size estimates were based on random effects model. We assessed for overall heterogeneity of the mean effect size using the Q statistic and by calculating the I2 statistic. The Q statistic is a test of the null hypothesis that all studies share a common effect size, and the I2 statistic examines the proportion of the variation across studies that is due to heterogeneity rather than chance, expressed as a percentage. General guidelines for the interpretation of the I2 are as follows: 25%, 50%, and 75% indicate low, moderate, and high heterogeneity, respectively.24 Categorical moderator analyses were conducted by using Q statistics,25,26 whereas the significance of each continuous moderator was assessed by using meta-regressions.27 Finally, we examined publication bias using funnel plots and Egger’s test.

To assess the quality of studies, a 7-point quality assessment tool was created based on those implemented in previous meta-analyses of observational studies.28,29 The tool evaluated the articles based on the following 7 criteria: (1) having a defined sample, (2) having a representative sample, (3) rater blinding, (4) report of relevant MDD and PA data, (5) adequate sample size, (6) statistical adjustment for covariates, and (7) a validated PA measure. Articles were given a score of 0 (“No”) or 1 (“Yes”) for each of the abovementioned criteria and summed to give a total score out of 7.

Our electronic search of 7 databases yielded 3147 nonduplicate articles. On review of the titles and abstracts, 87 articles met inclusion criteria and full articles were retrieved. A total of 40 studies with 50 independent samples (89 894 participants) met the inclusion criteria and were included in analyses. Figure 1 presents a flowchart of the review process.

FIGURE 1

PRISMA flow diagram of the literature search used to identify studies for analysis of physical activity and depression.

FIGURE 1

PRISMA flow diagram of the literature search used to identify studies for analysis of physical activity and depression.

Study Characteristics

As detailed in Table 1, 14 studies were longitudinal and 36 studies were cross-sectional. Sample sizes ranged from 55 to 14 594. Child age at the time of the assessment of PA ranged from 8 to 19 years. With respect to PA measures, 15 studies examined the frequency of activity only, 7 studies examined the intensity of the activity, and 27 examined a combination of frequency and intensity. With respect to the assessment of depression, 4 studies measured depressive symptoms by using interview methodology, whereas 46 studies used self-report questionnaires. The overall burden of depressive symptoms in studies that used a depression self-report measure was low (see Table 1). A clinical diagnosis of MDD was reported at follow-up for the 4 longitudinal samples that measured depressive symptoms by using a standardized interview. An MDD diagnosis was made in 5% to 13% of participants across these studies at follow-up.6,30,31 Although several studies specifically noted the absence of antidepressant medication use among participants, the large majority of studies did not include information regarding the use of medications.

TABLE 1

Independent Samples Included in the Meta-analysis of Physical Activity and Depression

StudyCountryN% BoysMean Agea (Range)PA Measure [F/I]Depression Measure [Q/INT] (Range)Depression Meanb (SD)QA Score (Out of 7)Time Lapse (T2 – T1)
Cross-sectional studies          
 Adeniyi et al32  Nigeria 1100 48.9 15.2 (12–17) PAQ-A [F/I] CDI [Q] (0–54) Boys: 8.8 (3.9); girls: 13.5 (6.7) — 
 Ammouri et al33  United States 186 — (10–19) Modified SAPAC [F] 6-item scale [Q] (0–18) 14.7 (3.8) — 
 Ammouri et al33  United States 98 100 — (10–19) Modified SAPAC [F] 6-item scale [Q] (0–18) 13.25 (4.08) — 
 Asare and Danquah34  Ghana 296 50.7 14.9 (13–18) PAQ-A [F/I] CDI [Q] (0–100) T scoresc: private school: 56.53 (13.9); public school: 44.07 (10.8) — 
 Babiss and Gangwisch35  United States 14 594 49 16 (11–21) No. d played sports in past wkd [F] CES-D [Q] — — 
 Cao et al36  China 5003 52 13.2 (11–16) Past-week PAd [F/I] DSRSC [Q] — — 
 Castillo et al37  United States 797 13.9 (11–18) MVPA past 2 wk (1-item) [F/I] 6-item screen [Q] (0–18) 11.3 (3.0) — 
 Castillo et al37  United States 711 100 13.9 (11–18) MVPA past 2 wk (1-item) [F/I] 6-item screen [Q] (0–18) 10.4 (2.9) — 
 Desha et al38  United States 371 15.3 (13–18) 24hr diaries, metabolic equivalents, total MVPAd [F/I] CDI-SF [Q] — — 
 Desha et al38  United States 356 100 15.3 (13–18) 24-h diaries, metabolic equivalents, total MVPAd [F/I] CDI-SF [Q] — — 
 Dockray et al39,e United States 55 10.5 (8–13) Parent-report of 6-items from CHIPd [F/I] CBCL – anx/dep [Q] 2.25 (2.1) — 
 Dockray et al39,e United States 56 100 11.4 (8–13) Parent-report of 6-items from CHIPd [F/I] CBCL – anx/dep [Q] 2.11 (2.2) — 
 Esmaeilzadeh40  Iran 265 100 9.7 (8–11) PAQ-C and physical fitness test [F/I] CDI [Q] — — 
 Fatiregun and Kumapayi41  Nigeria 1713 45 14 (10–19) Participation in sportsd,f PHQ-9 [Q] — — 
 Gray et al42  United States 95 46 12.8 (8–17) 1-item, amount of PA [F] CDI-SF [Q] (0–20) 13.6 (3.9) — 
 Hoare et al43  Australia 440 13.1 (11–14) ABAKQ [I] MFQ-SF [Q] (0–26) 13.1 (0.6) — 
 Hoare et al43  Australia 360 100 13.1 (11–14) ABAKQ [I] MFQ-SF [Q] (0–26) 13.1 (0.6) — 
 Hong et al44  China 1264 13.9 (–) min. of PA/weekday (end)d [F] CDI [Q] (0–54) 11.22 (7.3) — 
 Hong et al44  China 1180 100 13.9 (–) min. of PA/weekday (end)d [F] CDI [Q] [0–54] 12 (7.7) — 
 Jin et al45  Canada 3648 — — SHAPES PA, PASCQ, KKD [F/I] CES-D [Q] — — 
 Johnson et al46  United States 1397 12 (11–13) 3-d PA recall and accelerometry [F/I] CES-D [Q] (0–60) 14.7 (9.3) — 
 Kremer et al47  Australia 8029 48 11.5 (10–16) d and min/d participated in PAd [F] MFQ-SF [Q] (0–26) 6.4 (5.9) — 
 Maras et al48  Canada 2482 42 14.1 (11–20) GODIN [F/I] CDI [Q] (0–54) 7.8 (7.0) — 
 Mata et al49  United States 82 14 (10–16) PAQ-CA [F/I] CDI [Q] (0–54) 1.6 (2.0) — 
 Moljord et al50  Norway 560 15.6 (13–18) 1-item past mo PAd [F/I] ASQ (dep scale) [Q] (0–60) 2.5 (0.8) — 
 Moljord et al50  Norway 536 100 15.6 (13–18) 1-item past mo PAd [F/I] ASQ (dep scale) [Q] (0–60) 2.05 (0.7) — 
 Piko and Keresztes51  Hungary 1109 39 16.5 (14–21) No. times exercised past 3 mod [F] CDI-SF [Q] (0–20) Less active: 11.1 (2.7); regular active: 10.6 (2.4) — 
 Prasad et al52  United States 852 43 14.8 (–) 4-items from CDC YRBS [F/I] CDI-SF [Q] (0–20) 2.8 (0.1) — 
 Salah et al53  Egypt 546 53 — (12–20) 1-item PA measured [F] CES-DC [Q] (0–60) 16.4 (9.3) — 
 Shepherd et al54  New Zealand 148 — (16–18) NZPAQ-SF [F] DASS-42 (dep item) [Q] (0-42) Inactive: 2.4 (1.5); active: 1.7 (1.1); highly active: 1.6 (1.4) — 
 Sigfusdottir et al55  Iceland 7232 50 — 5-item PA scaled [F/I] 10-item scale [Q] (0–30) No family conflict: boys, 4.2 (4.8); girls, 6.7 (6.3). Family conflict: boys, 7.0 (6.0); girls, 11.4 (7.6) — 
 Soltanian et al56  Iran 735 54 — (15–19) IPAQ [F/I] GHQ (dep item) [Q] — — 
 Sun et al57  China 5453 — (8–18) YRBS [F/I] CDI [Q] (0–54) 12.2 (7.7) — 
 Sun et al57  China 3789 100 — (8–18) YRBS [F/I] CDI [Q] (0–54) 12.4 (6.8) — 
 Tao et al58  China 5141 52 15 (–) 2-questions and computed MVPAd [F/I] SCL-90 [Q] — — 
 Wiles et al59  United Kingdom 3298 47 13.8 (–) Accelerometry [F/I] MFQ [Q] — — 
Longitudinal studies          
 Birkeland et al60  Norway 912 55 13 No. times per wk did activities that made youth sweat/breathless [F/I] 7-item scale [Q] 1.89(0.96) 120 mo 
 Brunet et al61  Canada 860 47 12.7 MVPA, organized sport, IPAQ [F/I] MDI [Q] (0–50) 9.34 (7.5) 96 mo 
 Colman et al30  Canada 1137 50 16–17 Energy indexd [F/I] CIDI-SF [INT] — 168 mo 
 Hume et al62  Australia 155 40 14.5 Modified APARQd and accelerometry [I] CES-DC [Q] — 48 mo 
 Jerstad et al6  United States 496 13 Modified PYAS [F] SCID-IV [INT] — 72 mo 
 McKercher et al31  Australia 871 15 min/wk leisure PA [F] CIDI (self-admin.) [INT] — ∼240 mo 
 McKercher et al31  Australia 759 100 15 min/wk leisure PA [F] CIDI (self-admin.) [INT] — ∼240 mo 
 McPhie and Rawana63  United States 3676 49 15 Modified PA scale [F] CES-D (modified) [Q] (0–27) 6.04 (4.3) 156–168 y 
 Neissaar and Raudsepp64  Estonia 181 11.4 3DPAR [F/I] CES-D [Q] (0–60) 20.4 (3.2) 24 mo 
 Rothon et al11  United Kingdom 863 11.5 PA question from HEAd [I] MFQ-SF [Q] — 24 mo 
 Rothon et al11  United Kingdom 812 100 11–12 PA question from HEAd [I] MFQ-SF [Q] — 24 mo 
 Stavrakakis et al65  Netherlands 2230 49 11.1 Amount of PA/wkd [F] Affective problems from YSR and CBCL [Q] 0.31 (0.29) 48–84 mo 
 Sund et al8  Norway 2360 49.5 13.7 4-questions about PA, vigorous exercise, nonvigorous PA, sedentary activity [I] MFQ [Q] (0–68) 10.6 (11.8) 12–24 mo 
 Toseeb et al66  United Kingdom 614 43 14.5 Heart rate and movement sensing [I] MFQ [Q] (0–68) 13.7 (10.6) 36 mo 
StudyCountryN% BoysMean Agea (Range)PA Measure [F/I]Depression Measure [Q/INT] (Range)Depression Meanb (SD)QA Score (Out of 7)Time Lapse (T2 – T1)
Cross-sectional studies          
 Adeniyi et al32  Nigeria 1100 48.9 15.2 (12–17) PAQ-A [F/I] CDI [Q] (0–54) Boys: 8.8 (3.9); girls: 13.5 (6.7) — 
 Ammouri et al33  United States 186 — (10–19) Modified SAPAC [F] 6-item scale [Q] (0–18) 14.7 (3.8) — 
 Ammouri et al33  United States 98 100 — (10–19) Modified SAPAC [F] 6-item scale [Q] (0–18) 13.25 (4.08) — 
 Asare and Danquah34  Ghana 296 50.7 14.9 (13–18) PAQ-A [F/I] CDI [Q] (0–100) T scoresc: private school: 56.53 (13.9); public school: 44.07 (10.8) — 
 Babiss and Gangwisch35  United States 14 594 49 16 (11–21) No. d played sports in past wkd [F] CES-D [Q] — — 
 Cao et al36  China 5003 52 13.2 (11–16) Past-week PAd [F/I] DSRSC [Q] — — 
 Castillo et al37  United States 797 13.9 (11–18) MVPA past 2 wk (1-item) [F/I] 6-item screen [Q] (0–18) 11.3 (3.0) — 
 Castillo et al37  United States 711 100 13.9 (11–18) MVPA past 2 wk (1-item) [F/I] 6-item screen [Q] (0–18) 10.4 (2.9) — 
 Desha et al38  United States 371 15.3 (13–18) 24hr diaries, metabolic equivalents, total MVPAd [F/I] CDI-SF [Q] — — 
 Desha et al38  United States 356 100 15.3 (13–18) 24-h diaries, metabolic equivalents, total MVPAd [F/I] CDI-SF [Q] — — 
 Dockray et al39,e United States 55 10.5 (8–13) Parent-report of 6-items from CHIPd [F/I] CBCL – anx/dep [Q] 2.25 (2.1) — 
 Dockray et al39,e United States 56 100 11.4 (8–13) Parent-report of 6-items from CHIPd [F/I] CBCL – anx/dep [Q] 2.11 (2.2) — 
 Esmaeilzadeh40  Iran 265 100 9.7 (8–11) PAQ-C and physical fitness test [F/I] CDI [Q] — — 
 Fatiregun and Kumapayi41  Nigeria 1713 45 14 (10–19) Participation in sportsd,f PHQ-9 [Q] — — 
 Gray et al42  United States 95 46 12.8 (8–17) 1-item, amount of PA [F] CDI-SF [Q] (0–20) 13.6 (3.9) — 
 Hoare et al43  Australia 440 13.1 (11–14) ABAKQ [I] MFQ-SF [Q] (0–26) 13.1 (0.6) — 
 Hoare et al43  Australia 360 100 13.1 (11–14) ABAKQ [I] MFQ-SF [Q] (0–26) 13.1 (0.6) — 
 Hong et al44  China 1264 13.9 (–) min. of PA/weekday (end)d [F] CDI [Q] (0–54) 11.22 (7.3) — 
 Hong et al44  China 1180 100 13.9 (–) min. of PA/weekday (end)d [F] CDI [Q] [0–54] 12 (7.7) — 
 Jin et al45  Canada 3648 — — SHAPES PA, PASCQ, KKD [F/I] CES-D [Q] — — 
 Johnson et al46  United States 1397 12 (11–13) 3-d PA recall and accelerometry [F/I] CES-D [Q] (0–60) 14.7 (9.3) — 
 Kremer et al47  Australia 8029 48 11.5 (10–16) d and min/d participated in PAd [F] MFQ-SF [Q] (0–26) 6.4 (5.9) — 
 Maras et al48  Canada 2482 42 14.1 (11–20) GODIN [F/I] CDI [Q] (0–54) 7.8 (7.0) — 
 Mata et al49  United States 82 14 (10–16) PAQ-CA [F/I] CDI [Q] (0–54) 1.6 (2.0) — 
 Moljord et al50  Norway 560 15.6 (13–18) 1-item past mo PAd [F/I] ASQ (dep scale) [Q] (0–60) 2.5 (0.8) — 
 Moljord et al50  Norway 536 100 15.6 (13–18) 1-item past mo PAd [F/I] ASQ (dep scale) [Q] (0–60) 2.05 (0.7) — 
 Piko and Keresztes51  Hungary 1109 39 16.5 (14–21) No. times exercised past 3 mod [F] CDI-SF [Q] (0–20) Less active: 11.1 (2.7); regular active: 10.6 (2.4) — 
 Prasad et al52  United States 852 43 14.8 (–) 4-items from CDC YRBS [F/I] CDI-SF [Q] (0–20) 2.8 (0.1) — 
 Salah et al53  Egypt 546 53 — (12–20) 1-item PA measured [F] CES-DC [Q] (0–60) 16.4 (9.3) — 
 Shepherd et al54  New Zealand 148 — (16–18) NZPAQ-SF [F] DASS-42 (dep item) [Q] (0-42) Inactive: 2.4 (1.5); active: 1.7 (1.1); highly active: 1.6 (1.4) — 
 Sigfusdottir et al55  Iceland 7232 50 — 5-item PA scaled [F/I] 10-item scale [Q] (0–30) No family conflict: boys, 4.2 (4.8); girls, 6.7 (6.3). Family conflict: boys, 7.0 (6.0); girls, 11.4 (7.6) — 
 Soltanian et al56  Iran 735 54 — (15–19) IPAQ [F/I] GHQ (dep item) [Q] — — 
 Sun et al57  China 5453 — (8–18) YRBS [F/I] CDI [Q] (0–54) 12.2 (7.7) — 
 Sun et al57  China 3789 100 — (8–18) YRBS [F/I] CDI [Q] (0–54) 12.4 (6.8) — 
 Tao et al58  China 5141 52 15 (–) 2-questions and computed MVPAd [F/I] SCL-90 [Q] — — 
 Wiles et al59  United Kingdom 3298 47 13.8 (–) Accelerometry [F/I] MFQ [Q] — — 
Longitudinal studies          
 Birkeland et al60  Norway 912 55 13 No. times per wk did activities that made youth sweat/breathless [F/I] 7-item scale [Q] 1.89(0.96) 120 mo 
 Brunet et al61  Canada 860 47 12.7 MVPA, organized sport, IPAQ [F/I] MDI [Q] (0–50) 9.34 (7.5) 96 mo 
 Colman et al30  Canada 1137 50 16–17 Energy indexd [F/I] CIDI-SF [INT] — 168 mo 
 Hume et al62  Australia 155 40 14.5 Modified APARQd and accelerometry [I] CES-DC [Q] — 48 mo 
 Jerstad et al6  United States 496 13 Modified PYAS [F] SCID-IV [INT] — 72 mo 
 McKercher et al31  Australia 871 15 min/wk leisure PA [F] CIDI (self-admin.) [INT] — ∼240 mo 
 McKercher et al31  Australia 759 100 15 min/wk leisure PA [F] CIDI (self-admin.) [INT] — ∼240 mo 
 McPhie and Rawana63  United States 3676 49 15 Modified PA scale [F] CES-D (modified) [Q] (0–27) 6.04 (4.3) 156–168 y 
 Neissaar and Raudsepp64  Estonia 181 11.4 3DPAR [F/I] CES-D [Q] (0–60) 20.4 (3.2) 24 mo 
 Rothon et al11  United Kingdom 863 11.5 PA question from HEAd [I] MFQ-SF [Q] — 24 mo 
 Rothon et al11  United Kingdom 812 100 11–12 PA question from HEAd [I] MFQ-SF [Q] — 24 mo 
 Stavrakakis et al65  Netherlands 2230 49 11.1 Amount of PA/wkd [F] Affective problems from YSR and CBCL [Q] 0.31 (0.29) 48–84 mo 
 Sund et al8  Norway 2360 49.5 13.7 4-questions about PA, vigorous exercise, nonvigorous PA, sedentary activity [I] MFQ [Q] (0–68) 10.6 (11.8) 12–24 mo 
 Toseeb et al66  United Kingdom 614 43 14.5 Heart rate and movement sensing [I] MFQ [Q] (0–68) 13.7 (10.6) 36 mo 

ABAKQ, Adolescent Behaviors, Attitudes, and Knowledge Questionnaire; anx, anxiety; APARQ, Adolescent Physical Activity Recall Questionnaire; ASQ, Adolescent Stress Questionnaire; CBCL, Child Behavior Checklist: Anxious/Depressed subscale; CDC YRBS, Centers for Disease Control and Prevention Youth Risk Behavior Survey; CDI/-SF, Children’s Depression Inventory (-Short Form); CESD/-C, Centre for Epidemiologic Studies Depression (-for Children); CHIP, Child Health and Illness Profile; CIDI/-SF, Composite International Diagnostic Interview (-Short Form); DASS-42, Depression, Anxiety, Stress Scale-42; dep, depression; 3DPAR, 3-day physical activity recall; DSRSC, Depression Self Rating Scale for Children; F, frequency; GHQ, General Health Questions; GODIN, Godin Leisure-Time Exercise Questionnaire; HEA, health education authority; I, intensity; INT, interview; IPAQ, International Physical Activity Questionnaire; KKD, kilocalories expanded per kilogram of body weight per day; L, longitudinal; MDI, Major Depression Inventory; MFQ/-SF, Mood and Feelings Questionnaire (-Short Form); MVPA, moderate to vigorous physical activity; NZPAQ-SF, New Zealand Physical Activity Questionnaire (-Short Form); PAQ/-A/-CA, Physical Activity Questionnaire (-for Adolescents/-for Older Children and Adolescents); PASCQ, Physical Activity Stages Questionnaire; PHQ-9, Patient Health Questionnaire; PYAS, Past Year Activity Scale; Q, questionnaire; QA, quality assessment; SAPAC, Self-Administered Physical Activity Checklist; SCID-IV = Structured Clinical Interview for DSM-IV; SCL-90, Symptom Checklist-90; self-admin, self-administered; SHAPES, School Health Action Planning and Evaluation System; T1, time 1; T2, time 2; XS, cross-sectional; YSR, youth self-report; —, insufficient information available in the article.

a

Age listed is at time 1 for longitudinal studies. Where available, the age range is reported for studies that do not report the mean age.

b

Mean depression score is at time 2 for longitudinal studies.

c

T scores >70 are recommended to determine clinical significance.

d

Nonvalidated measure.

e

Specifically noted that participants were not on medication that could interfere with hormone level or weight gain.

f

Not specified; asked about participation in sporting activities.

Study Quality

Validated measures of PA were used in 19 out of 36 (53%) cross-sectional studies and in 10 out of 14 (71%) longitudinal studies, as indicated in Table 1. The mean study quality score was 4.9 (SD = 0.9) out of 7. For cross-sectional studies, the mean percentage of participants with complete data were 96.6% (range: 68%–100%). For longitudinal studies, the mean rate of attrition between time points was 13.8% (range: 0.04%–30%). Additional detail regarding individual study- and item-level quality assessment scoring is summarized in Supplemental Table 6.

A significant mean effect size for the association between PA and depression was found: (r = –0.14; 95% confidence interval [CI] = –0.19 to –0.10) (Fig 2), suggesting that children’s PA is negatively associated with depressive symptoms. The funnel plot revealed asymmetry (Fig 3) and Egger’s test suggested that the asymmetry was significant (P < .01). Using the trim and fill analysis, the adjusted pooled effect size estimate was r = 0.06 (95% CI = –0.11 to –0.01). Statistically significant heterogeneity between the studies was found (Q = 1767.95; P < .0001; I2 = 95.23) and potential moderator analyses were explored, including demographic, measurement, and study design factors. The results of all moderator analyses are presented in Tables 2 and3, and significant moderators are discussed in detail below.

FIGURE 2

Forest plot of the overall mean effect size, as well as the effect size for each study included in the analysis. Observed effect sizes (r) and 95% CIs are indicated for each study included in the meta-analysis. The black diamond, located at the bottom of the forest plot, indicates the overall mean effect size. Inserting an average effect size across all stratified groups for studies that categorized PA into strata had no effect on the overall mean effect size (r = –0.14; 95% CI: 0.18 to 0.10).

FIGURE 2

Forest plot of the overall mean effect size, as well as the effect size for each study included in the analysis. Observed effect sizes (r) and 95% CIs are indicated for each study included in the meta-analysis. The black diamond, located at the bottom of the forest plot, indicates the overall mean effect size. Inserting an average effect size across all stratified groups for studies that categorized PA into strata had no effect on the overall mean effect size (r = –0.14; 95% CI: 0.18 to 0.10).

FIGURE 3

Funnel plot of the meta-analysis of included studies. The y-axis on the funnel plot represents the SE, and the x-axis is the effect size. Observed studies are indicated by open circles. The white diamond represents the observed mean effect size, and the black diamond represents the adjusted mean effect size.

FIGURE 3

Funnel plot of the meta-analysis of included studies. The y-axis on the funnel plot represents the SE, and the x-axis is the effect size. Observed studies are indicated by open circles. The white diamond represents the observed mean effect size, and the black diamond represents the adjusted mean effect size.

TABLE 2

Examination of Potential Effect Modifiers in the Association of Physical Activity and Depression: Categorical Variables

Categorical ModeratorskNR95% CIQContrast P
Sex     2.13 .15 
 Girls 15 13 864 –0.11** –0.15 to –0.06 
 Boys 11 8931 –0.07** –0.10 to –0.03 
Social Risk     9.62 .01 
 No 44 85 440 –0.15*** –0.21 to –0.10 
 Yes 4464 –0.05* –0.09 to –0.01 
Physical activity type     9.94 .01 
 Frequency only 15 35 283 –0.11*** –0.14 to –0.08 
 Intensity only 5604 –0.05* –0.09 to –0.01 
 Frequency/intensity 27 47 294 –0.17*** –0.25 to –0.09 
Physical activity measure     4.71 .03 
 Nonvalidated 21 53 589 –0.08*** –0.11 to –0.05 
 Validated 29 36 305 –0.18*** –0.26 to –0.09 
Depression measure type     1.38 .24 
 Interview 3263 –0.05* –0.09 to –0.01 
 Questionnaire 46 86 631 –0.15*** –0.20 to –0.10 
Study Design     7.32 .01 
 Cross-sectional 36 73 978 –0.17*** –0.23 to –0.10 
 Longitudinal 14 15 926 –0.07*** –0.10 to –0.04 
Continent     3.28 .51 
 North America 19 31 938 –0.08*** –0.11 to –0.04 
 Europe 20 707 –0.09*** –0.13 to –0.05 
 Australia 12 10 762 –0.09*** –0.12 to –0.05 
 Asia 22 830 –0.08*** –0.08 to –0.11 
 Africa 3655 –0.55* –0.84 to –0.01 
Categorical ModeratorskNR95% CIQContrast P
Sex     2.13 .15 
 Girls 15 13 864 –0.11** –0.15 to –0.06 
 Boys 11 8931 –0.07** –0.10 to –0.03 
Social Risk     9.62 .01 
 No 44 85 440 –0.15*** –0.21 to –0.10 
 Yes 4464 –0.05* –0.09 to –0.01 
Physical activity type     9.94 .01 
 Frequency only 15 35 283 –0.11*** –0.14 to –0.08 
 Intensity only 5604 –0.05* –0.09 to –0.01 
 Frequency/intensity 27 47 294 –0.17*** –0.25 to –0.09 
Physical activity measure     4.71 .03 
 Nonvalidated 21 53 589 –0.08*** –0.11 to –0.05 
 Validated 29 36 305 –0.18*** –0.26 to –0.09 
Depression measure type     1.38 .24 
 Interview 3263 –0.05* –0.09 to –0.01 
 Questionnaire 46 86 631 –0.15*** –0.20 to –0.10 
Study Design     7.32 .01 
 Cross-sectional 36 73 978 –0.17*** –0.23 to –0.10 
 Longitudinal 14 15 926 –0.07*** –0.10 to –0.04 
Continent     3.28 .51 
 North America 19 31 938 –0.08*** –0.11 to –0.04 
 Europe 20 707 –0.09*** –0.13 to –0.05 
 Australia 12 10 762 –0.09*** –0.12 to –0.05 
 Asia 22 830 –0.08*** –0.08 to –0.11 
 Africa 3655 –0.55* –0.84 to –0.01 
*

P < .05.

**

P < .01.

***

P < .001.

Effect sizes were stronger in samples using cross-sectional designs (k = 36, r = –0.17; 95% CI = –0.23 to –0.10) compared with those using longitudinal designs (k = 14, r = –0.07; 95% CI = –0.10 to –0.04), in which a weak inverse relationship between physical activity and future depressive symptoms was found. Similarly, studies that used interview-based MDD measures demonstrated weaker effect sizes compared with those that used questionnaires (k = 4, r = –0.05; 95% CI = –0.09 to –0.01 vs k = 46, r = –0.15; 95% CI = –0.20 to –0.10). Stronger effect sizes were also observed in samples with no known risks (k = 44; r = –0.15; 95% CI = –0.21 to –0.10) compared with samples with social risk (eg, low income) (k = 6; r = –0.05; 95% CI = –0.09 to –0.01). Effect sizes were stronger in samples examining a combination of PA frequency and intensity (k = 27; r = –0.17; 95% CI = –0.25 to –0.09) compared with intensity alone (k = 7; r = –0.05; 95% CI = –0.09 to –0.01). Finally, stronger effect sizes were found in studies that used validated (k = 29, r = –0.18; 95% CI = –0.26 to –0.09) versus nonvalidated PA measures (k = 21, r = –0.08; 95% CI = –0.11 to –0.05).

Because there were significant differences in effect sizes between cross-sectional and longitudinal studies, and because longitudinal associations may provide insight into the directionality of associations, we performed a set of subanalyses with longitudinal studies only to more explicitly examine the magnitude of the association, as well as the between-study variability, for studies assessing a baseline metric of physical activity and its association with later depressive symptoms. There were 14 studies involving 15 926 participants that reported on longitudinal associations between PA and depression. Five studies6,8,30,64,66 reported on depression-related covariates, including baseline depressive symptoms, number of weeks depressed during the preceding year, body dissatisfaction, social support, self-efficacy, history of childhood trauma or stressful life events, and medication status (Table 1).

The mean effect size for the longitudinal association between PA and depression was r = –0.07 (95% CI = –0.10 to –0.04). Statistically significant heterogeneity between studies was found (Q = 59.25; P < .0001; I2 = 77.52) and potential moderator analyses were explored (Tables 4 and5). However, because the number of studies for several subgroups was small (eg, there were only 2 studies with social risk), the results of these moderator analyses should be interpreted with caution (Table 5).

TABLE 3

Examination of Potential Effect Modifiers in the Association of Physical Activity and Depression: Continuous Variables

Continuous ModeratorskNSlopeSEz ScoreP
Age at PA assessment 50 89 894 –0.009 0.014 −0.65 .52 
Age at depression assessment 50 89 894 0.005 0.005 1.03 .30 
Time between assessments 50 89 894 0.008 0.003 1.46 .15 
Publication year 50 89 894 0.002 0.009 0.17 .87 
Continuous ModeratorskNSlopeSEz ScoreP
Age at PA assessment 50 89 894 –0.009 0.014 −0.65 .52 
Age at depression assessment 50 89 894 0.005 0.005 1.03 .30 
Time between assessments 50 89 894 0.008 0.003 1.46 .15 
Publication year 50 89 894 0.002 0.009 0.17 .87 
TABLE 4

Examination of Potential Categorical Effect Modifiers in Studies With Longitudinal Associations Between Physical Activity and Depression

Categorical ModeratorskNR95% CIQContrast P
Sex     1.16 .28 
 Girls 812 –0.10* –0.20 to –0.00 
 Boys 2411 –0.03 –0.11 to –0.04 
Social Risk     12.13 .0001 
 No 12 14 251 –0.09*** –0.12 to –0.05 
 Yes 1675 –0.01 –0.03 to 0.01 
Physical activity type     2.44 .30 
 Frequency only 8032 –0.06** –0.08 to –0.03 
 Intensity only 3090 –0.04+ –0.09 to 0.01 
 Frequency/intensity 4804 –0.14* –0.25 to –0.02 
Physical activity measure     11.66 .01 
 Nonvalidated 5042 –0.01+ –0.03 to 0.00 
 Validated 10 10 884 –0.10*** –0.14 to –0.05 
Depression measure type     1.12 .29 
 Interview 3263 –0.05* –0.09 to –0.01 
 Questionnaire 10 12 663 –0.08*** –0.12 to –0.04 
Continent     1.31 .52 
 North America 6169 –0.06*** –0.08 to –0.03 
 Europe 7972 –0.09** –0.15 to –0.04 
 Australia 1785 –0.05* –0.10 to –0.01 
Categorical ModeratorskNR95% CIQContrast P
Sex     1.16 .28 
 Girls 812 –0.10* –0.20 to –0.00 
 Boys 2411 –0.03 –0.11 to –0.04 
Social Risk     12.13 .0001 
 No 12 14 251 –0.09*** –0.12 to –0.05 
 Yes 1675 –0.01 –0.03 to 0.01 
Physical activity type     2.44 .30 
 Frequency only 8032 –0.06** –0.08 to –0.03 
 Intensity only 3090 –0.04+ –0.09 to 0.01 
 Frequency/intensity 4804 –0.14* –0.25 to –0.02 
Physical activity measure     11.66 .01 
 Nonvalidated 5042 –0.01+ –0.03 to 0.00 
 Validated 10 10 884 –0.10*** –0.14 to –0.05 
Depression measure type     1.12 .29 
 Interview 3263 –0.05* –0.09 to –0.01 
 Questionnaire 10 12 663 –0.08*** –0.12 to –0.04 
Continent     1.31 .52 
 North America 6169 –0.06*** –0.08 to –0.03 
 Europe 7972 –0.09** –0.15 to –0.04 
 Australia 1785 –0.05* –0.10 to –0.01 
+

P < .10.

*

P < .05.

**

P < .01.

***

P < .001.

TABLE 5

Examination of Continuous Moderators in Studies With Longitudinal Associations Between Physical Activity and Depression

Continuous ModeratorskNSlopeSEz ScoreP
Age at physical activity 14 15 926 0.004 0.009 0.45 .66 
Age at depression 14 15 926 0.001 0.002 0.47 .64 
Time between assessments 14 15 926 0.001 0.003 0.45 .65 
Publication year 14 15 926 0.009 0.009 1.03 .30 
Continuous ModeratorskNSlopeSEz ScoreP
Age at physical activity 14 15 926 0.004 0.009 0.45 .66 
Age at depression 14 15 926 0.001 0.002 0.47 .64 
Time between assessments 14 15 926 0.001 0.003 0.45 .65 
Publication year 14 15 926 0.009 0.009 1.03 .30 

This systematic review and meta-analysis of 50 samples involving 89 894 participants found that a greater PA level was associated with fewer depressive symptoms, although not with decreased diagnoses of MDD. This association was stronger for cross-sectional studies than for longitudinal studies, in which the mean effect size was significant, but weak. The nature of the PA was also associated with the presence of depressive symptoms, in that PA of increased frequency and intensity was more strongly associated with decreased depressive symptoms compared with PA that was defined by intensity of activity alone.

Significant effect sizes were observed for studies that examined depressive symptomatology by using questionnaire measures and were considerably stronger than those of studies assessing MDD by using interview measures. Indeed, the majority of studies in this meta-analysis employed self-report inventories to assess depressive symptoms (n = 46) rather than diagnostic interviews (n = 4), which are considered to be the gold-standard measure for MDD. Self-report measures are frequently used in research studies due to their ease of administration, low cost, minimal time requirement, and low patient response burden. These measures are useful screening tools; however, self-report instruments are limited by their inability to confirm the presence or absence of an MDD diagnosis. That increased PA was more highly associated with decreased depressive symptoms in this meta-analysis, as compared with an MDD diagnosis, is a critical finding. This finding suggests that individuals who are at risk for more severe, syndromal-level symptom burden, impairment, and associated poor health outcomes may not respond to the potential preventative effects of PA. Although it is possible that these results may also reflect the relative methodological limitations associated with the examination of a dichotomous versus a continuous variable, our findings are consistent with previous data reporting that MDD severity is distinguished from subsyndromal depressive symptoms by its decreased sensitivity to prevention strategies, greater association with cardiovascular risk factors and health outcomes, and greater treatment resistance.67,70 

Increased PA was more strongly associated with decreased depressive symptoms in cross-sectional studies compared with longitudinal studies, where the effect size was small. Cross-sectional studies are limited in their ability to probe causality, because the temporal relationship between variables cannot be determined. Thus, it is possible that the cross-sectional studies included in this meta-analysis are actually indicative of the reverse association of PA and depression: that children and adolescents with increased depressive symptoms are less likely to participate in PA. Indeed, amotivation, pessimism, and anhedonia associated with the depressed state have been reported to lead to decreased PA among adult populations.71 In contrast, longitudinal studies provide insight into the direction of the association and, in the present meta-analysis, demonstrated a weak inverse relationship between PA and future depressive symptoms measured 2 to 17 years later, suggesting that PA has a weak but positive association with future mood.

Studies that included a measure of both increased PA frequency and intensity demonstrated stronger associations with depressive symptoms than those that used measures of intensity alone. This finding is consistent with other systematic reviews examining the role of PA as an intervention for depressed adults.1 Currently, some clinical guidelines recommend the inclusion of 45 minutes of moderately intense exercise at least 3 days per week in the treatment of MDD among adults.72 In contrast, guidelines for general health promotion by the Canadian Pediatric Society73 and American Academy of Pediatrics74 recommend that children and adolescents get at least 60 minutes of moderate to vigorous PA daily to maintain general health. As such, the findings from the current study support the inclusion of both the PA frequency and intensity components in the Canadian Pediatric Society and American Academy of Pediatrics recommendations with respect to the benefit to depressed mood. Many hypotheses regarding the mechanism by which PA may lead to improved mood have been theorized, including via antiinflammatory effects, increased growth factors leading to neural plasticity, neuroendocrine effects on the hypothalamic-pituitary-adrenal axis and insulin sensitivity, and improvements in self-efficacy.75,77 However, neither the pathophysiological pathways themselves nor whether they are specific to mood state are known. These factors are important for determining rational prevention versus treatment strategies, gaining insight into the etiology of depression, and for research into novel treatments for depression for medically ill populations and those unable to participate in PA.

Studies that examined the association of PA with depression in samples of higher social risk (eg, low income, minority, or involved in child protective services) reported weaker effect sizes than those of lower-risk groups. Socioeconomic status and its associated risk factors (eg, disadvantaged neighborhoods) explain a significant proportion of the variance in childhood psychopathology, including depression.78 Because children in high–social risk environments may be exposed to many more risk factors for depression, including lower socioeconomic status,79 increased PA may have relatively less influence with respect to the proportion of the variance in depression it explains when compared with children of lower social risk.80 Also, because measures of depression and PA have traditionally been developed in samples of low social risk, they may be less well calibrated to capture the variation in depression or PA seen in high–social risk children.81,82 These results should be interpreted with caution, however, because few studies have examined the association of PA with depression in high–social risk samples. Given the increased prevalence of both depression and obesity in populations of high social risk, however, additional research examining potential targets for prevention among this vulnerable group of children is needed.

As the first study to conduct a meta-analytic review of the potential protective association of childhood PA with depression, this study has many strengths, including the analysis of a large number of studies to increase the precision of effect size estimates, subanalysis of cross-sectional versus longitudinal associations, and examination of PA frequency and intensity as potentially contributing effect modifiers. However, our findings must be interpreted within the context of the limitations of this study. The measurement of PA in the majority of studies relied on self-report measures of frequency, intensity, and type of activity, which were not correlated with objective measures of activity (eg, accelerometry). This also reflects a limitation of the PA literature more broadly, in that the use of standardized instruments that have demonstrated reliability and validity was not consistent across studies. The current meta-analysis demonstrated that studies with validated measures of PA had stronger effect sizes than those that used nonvalidated measures. Thus, future PA research should focus on the methodology for PA measurement in children and adolescents to increase confidence in the study results. In addition, the majority of the literature relies on the self-report of depressive symptoms, with few studies able to confirm a diagnosis of depression, leading to wide precision estimates of the magnitude of the effect of PA on clinical depression. Finally, we only included studies that were published in English, and this inclusion criterion may limit the generalizability of our findings to predominantly English-speaking countries.

This systematic review and meta-analysis finds that increased PA in childhood and adolescence is associated with decreased depressive symptoms. Substantive moderators of this association include (1) study design, with the strongest association found in cross-sectional studies; (2) type of PA, with a combination of PA frequency and intensity resulting in the greatest effect on depressive symptoms; and (3) depression measure, with a stronger protective effect of increased PA for depressive symptoms than for a clinical diagnosis of MDD. Taken together, this study suggests that PA in childhood and adolescence is associated with improved concurrent symptoms of depression, particularly when undertaken regularly and with vigor, and has weak but significant effects on future depressive symptoms. Future research is needed to advance the knowledge of PA measurement, elucidate the mechanism of association between PA and depression, and examine the longitudinal relationships between PA, depression, and health outcomes to determine the critical periods in which preventative efforts may be most effective.

     
  • CI

    confidence interval

  •  
  • MDD

    major depressive disorder

  •  
  • PA

    physical activity

Dr Korczak conceptualized and designed the study, assisted in data collection, and drafted the initial manuscript; Dr Madigan assisted in data collection, carried out the initial analyses, contributed to, and reviewed and revised the manuscript; Ms Colasanto coordinated and assisted in data collection and contributed to and critically reviewed the manuscript; and all authors approved the final manuscript as submitted.

FUNDING: Research support was provided to Dr Madigan by the Alberta Children’s Hospital Foundation and the Canada Research Chairs program.

We thank Ms Qi Fang (University of Toronto) for assistance in the literature search.

1
Cooney
GM
,
Dwan
K
,
Greig
CA
, et al
.
Exercise for depression.
Cochrane Database Syst Rev
.
2013
;(
9
):
CD004366
[PubMed]
2
Nyström
MB
,
Neely
G
,
Hassmén
P
,
Carlbring
P
.
Treating major depression with physical activity: a systematic overview with recommendations.
Cogn Behav Ther
.
2015
;
44
(
4
):
341
352
[PubMed]
3
Åberg
MA
,
Waern
M
,
Nyberg
J
, et al
.
Cardiovascular fitness in males at age 18 and risk of serious depression in adulthood: Swedish prospective population-based study.
Br J Psychiatry
.
2012
;
201
(
5
):
352
359
[PubMed]
4
Carroll
DD
,
Blanck
HM
,
Serdula
MK
,
Brown
DR
.
Obesity, physical activity, and depressive symptoms in a cohort of adults aged 51 to 61.
J Aging Health
.
2010
;
22
(
3
):
384
398
[PubMed]
5
Hamer
M
,
Molloy
GJ
,
de Oliveira
C
,
Demakakos
P
.
Leisure time physical activity, risk of depressive symptoms, and inflammatory mediators: the English Longitudinal Study of Ageing.
Psychoneuroendocrinology
.
2009
;
34
(
7
):
1050
1055
[PubMed]
6
Jerstad
SJ
,
Boutelle
KN
,
Ness
KK
,
Stice
E
.
Prospective reciprocal relations between physical activity and depression in female adolescents.
J Consult Clin Psychol
.
2010
;
78
(
2
):
268
272
[PubMed]
7
Strawbridge
WJ
,
Deleger
S
,
Roberts
RE
,
Kaplan
GA
.
Physical activity reduces the risk of subsequent depression for older adults.
Am J Epidemiol
.
2002
;
156
(
4
):
328
334
[PubMed]
8
Sund
AM
,
Larsson
B
,
Wichstrøm
L
.
Role of physical and sedentary activities in the development of depressive symptoms in early adolescence.
Soc Psychiatry Psychiatr Epidemiol
.
2011
;
46
(
5
):
431
441
[PubMed]
9
Cooper-Patrick
L
,
Ford
DE
,
Mead
LA
,
Chang
PP
,
Klag
MJ
.
Exercise and depression in midlife: a prospective study.
Am J Public Health
.
1997
;
87
(
4
):
670
673
[PubMed]
10
Kritz-Silverstein
D
,
Barrett-Connor
E
,
Corbeau
C
.
Cross-sectional and prospective study of exercise and depressed mood in the elderly: the Rancho Bernardo study.
Am J Epidemiol
.
2001
;
153
(
6
):
596
603
[PubMed]
11
Rothon
C
,
Edwards
P
,
Bhui
K
,
Viner
RM
,
Taylor
S
,
Stansfeld
SA
.
Physical activity and depressive symptoms in adolescents: a prospective study.
BMC Med
.
2010
;
8
:
32
[PubMed]
12
Pinto Pereira
SM
,
Geoffroy
MC
,
Power
C
.
Depressive symptoms and physical activity during 3 decades in adult life: bidirectional associations in a prospective cohort study.
JAMA Psychiatry
.
2014
;
71
(
12
):
1373
1380
[PubMed]
13
Bursnall
P
.
The relationship between physical activity and depressive symptoms in adolescents: a systematic review.
Worldviews Evid Based Nurs
.
2014
;
11
(
6
):
376
382
[PubMed]
14
Mammen
G
,
Faulkner
G
.
Physical activity and the prevention of depression: a systematic review of prospective studies.
Am J Prev Med
.
2013
;
45
(
5
):
649
657
[PubMed]
15
Kobrosly
RW
,
van Wijngaarden
E
,
Seplaki
CL
,
Cory-Slechta
DA
,
Moynihan
J
.
Depressive symptoms are associated with allostatic load among community-dwelling older adults.
Physiol Behav
.
2014
;
123
:
223
230
[PubMed]
16
Vaccarino
V
,
McClure
C
,
Johnson
BD
, et al
.
Depression, the metabolic syndrome and cardiovascular risk.
Psychosom Med
.
2008
;
70
(
1
):
40
48
[PubMed]
17
Mikkelsen
SS
,
Tolstrup
JS
,
Flachs
EM
,
Mortensen
EL
,
Schnohr
P
,
Flensborg-Madsen
T
.
A cohort study of leisure time physical activity and depression.
Prev Med
.
2010
;
51
(
6
):
471
475
[PubMed]
18
Wang
F
,
DesMeules
M
,
Luo
W
,
Dai
S
,
Lagace
C
,
Morrison
H
.
Leisure-time physical activity and marital status in relation to depression between men and women: A prospective study.
Health Psychol
.
2011
;
30
(
2
):
204
211
[PubMed]
19
Matsudo
VK
,
Ferrari
GL
,
Araújo
TL
, et al
.
Socioeconomic status indicators, physical activity, and overweight/obesity in Brazilian children [in Portuguese]
.
Rev Paul Pediatr
.
2016
;
34
(
2
):
162
170
20
Ritterman Weintraub
ML
,
Fernald
LC
,
Goodman
E
,
Guendelman
S
,
Adler
NE
.
Obesity-related behaviors among poor adolescents and young adults: is social position associated with risk behaviors?
Front Public Health
.
2015
;
3
:
224
[PubMed]
21
Villagran Perez
S
,
Novalbos-Ruiz
JP
,
Rodriguez-Martin
A
,
Martinez-Nieto
M
,
Lechuga-Sancho
AM
.
Implications of family socioeconomic level on risk behaviors in child-youth obesity.
Nutr Hosp
.
2013
;
28
(
6
):
1951
1960
22
Eassa
S
,
Hagag
SA
,
Seliem
HAEW
,
Amar
HA
.
Assessment of sport practice among adolescent school students and its effect on perceived health in Sharkia Governorate–Egypt.
J Am Sci
.
2011
;
7
(
3
):
544
551
23
Borenstein
M
,
Hedges
LV
,
Higgins
JP
,
Rothstein
HR
. Comprehensive Meta-analysis [computer program]. Version 3.
Englewood, NJ
: Biostat;
2014
24
Higgins
JP
,
Thompson
SG
,
Deeks
JJ
,
Altman
DG
.
Measuring inconsistency in meta-analyses.
BMJ
.
2003
;
327
(
7414
):
557
560
[PubMed]
25
Borenstein
M
,
Hedges
LV
,
Higgins
JPT
,
Rothstein
HR
.
Introduction to Meta-Analysis
.
West Sussex, United Kingdom
:
John Wiley & Sons
;
2009
26
Rosenthal
R
.
Writing meta-analytic reviews.
Psychol Bull
.
1995
;
118
(
2
):
183
192
27
Thompson
SG
,
Higgins
JPT
.
How should meta-regression analyses be undertaken and interpreted?
Stat Med
.
2002
;
21
(
11
):
1559
1573
[PubMed]
28
Sanderson
S
,
Tatt
ID
,
Higgins
JP
.
Tools for assessing quality and susceptibility to bias in observational studies in epidemiology: a systematic review and annotated bibliography.
Int J Epidemiol
.
2007
;
36
(
3
):
666
676
[PubMed]
29
Tsang
TW
,
Lucas
BR
,
Carmichael Olson
H
,
Pinto
RZ
,
Elliott
EJ
.
Prenatal alcohol exposure, FASD, and child behavior: a meta-analysis.
Pediatrics
.
2016
;
137
(
3
):
e20152542
[PubMed]
30
Colman
I
,
Zeng
Y
,
McMartin
SE
, et al
.
Protective factors against depression during the transition from adolescence to adulthood: findings from a national Canadian cohort.
Prev Med
.
2014
;
65
:
28
32
[PubMed]
31
McKercher
C
,
Sanderson
K
,
Schmidt
MD
, et al
.
Physical activity patterns and risk of depression in young adulthood: a 20-year cohort study since childhood.
Soc Psychiatry Psychiatr Epidemiol
.
2014
;
49
(
11
):
1823
1834
[PubMed]
32
Adeniyi
AF
,
Okafor
NC
,
Adeniyi
CY
.
Depression and physical activity in a sample of Nigerian adolescents: levels, relationships and predictors.
Child Adolesc Psychiatry Ment Health
.
2011
;
5
:
16
[PubMed]
33
Ammouri
AA
,
Kaur
H
,
Neuberger
GB
,
Gajewski
B
,
Choi
WS
.
Correlates of exercise participation in adolescents.
Public Health Nurs
.
2007
;
24
(
2
):
111
120
[PubMed]
34
Asare
M
,
Danquah
SA
.
The relationship between physical activity, sedentary behaviour and mental health in Ghanaian adolescents.
Child Adolesc Psychiatry Ment Health
.
2015
;
9
:
11
[PubMed]
35
Babiss
LA
,
Gangwisch
JE
.
Sports participation as a protective factor against depression and suicidal ideation in adolescents as mediated by self-esteem and social support.
J Dev Behav Pediatr
.
2009
;
30
(
5
):
376
384
[PubMed]
36
Cao
H
,
Qian
Q
,
Weng
T
, et al
.
Screen time, physical activity and mental health among urban adolescents in China.
Prev Med
.
2011
;
53
(
4–5
):
316
320
[PubMed]
37
Castillo
F
,
Francis
L
,
Wylie-Rosett
J
,
Isasi
CR
.
Depressive symptoms are associated with excess weight and unhealthier lifestyle behaviors in urban adolescents.
Child Obes
.
2014
;
10
(
5
):
400
407
[PubMed]
38
Desha
LN
,
Ziviani
JM
,
Nicholson
JM
,
Martin
G
,
Darnell
RE
.
Physical activity and depressive symptoms in American adolescents.
J Sport Exerc Psychol
.
2007
;
29
(
4
):
534
543
[PubMed]
39
Dockray
S
,
Susman
EJ
,
Dorn
LD
.
Depression, cortisol reactivity, and obesity in childhood and adolescence.
J Adolesc Health
.
2009
;
45
(
4
):
344
350
[PubMed]
40
Esmaeilzadeh
S
.
Relationship between depressive symptoms with physical activity and physical fitness among children.
Mental Health & Prevention
.
2014
;
2
(
1–2
):
11
17
41
Fatiregun
AA
,
Kumapayi
TE
.
Prevalence and correlates of depressive symptoms among in-school adolescents in a rural district in southwest Nigeria.
J Adolesc
.
2014
;
37
(
2
):
197
203
[PubMed]
42
Gray
WN
,
Janicke
DM
,
Ingerski
LM
,
Silverstein
JH
.
The impact of peer victimization, parent distress and child depression on barrier formation and physical activity in overweight youth.
J Dev Behav Pediatr
.
2008
;
29
(
1
):
26
33
[PubMed]
43
Hoare
E
,
Millar
L
,
Fuller-Tyszkiewicz
M
, et al
.
Associations between obesogenic risk and depressive symptomatology in Australian adolescents: a cross-sectional study.
J Epidemiol Community Health
.
2014
;
68
(
8
):
767
772
[PubMed]
44
Hong
X
,
Li
J
,
Xu
F
, et al
.
Physical activity inversely associated with the presence of depression among urban adolescents in regional China.
BMC Public Health
.
2009
;
9
:
148
[PubMed]
45
Jin
S
,
Muhajarine
N
,
Cushon
J
,
Lim
HJ
.
Factors associated with childhood depression in Saskatoon students: a multilevel analysis.
Can J Commun Ment Health
.
2013
;
32
(
1
):
29
42
46
Johnson
CC
,
Murray
DM
,
Elder
JP
, et al
.
Depressive symptoms and physical activity in adolescent girls.
Med Sci Sports Exerc
.
2008
;
40
(
5
):
818
826
[PubMed]
47
Kremer
P
,
Elshaug
C
,
Leslie
E
,
Toumbourou
JW
,
Patton
GC
,
Williams
J
.
Physical activity, leisure-time screen use and depression among children and young adolescents.
J Sci Med Sport
.
2014
;
17
(
2
):
183
187
[PubMed]
48
Maras
D
,
Flament
MF
,
Murray
M
, et al
.
Screen time is associated with depression and anxiety in Canadian youth.
Prev Med
.
2015
;
73
:
133
138
[PubMed]
49
Mata
J
,
Thompson
RJ
,
Gotlib
IH
.
BDNF genotype moderates the relation between physical activity and depressive symptoms.
Health Psychol
.
2010
;
29
(
2
):
130
133
[PubMed]
50
Moljord
IEO
,
Moksnes
UK
,
Espnes
GA
,
Hjemdal
O
,
Eriksen
L
.
Physical activity, resilience, and depressive symptoms in adolescence.
Ment Health Phys Act
.
2014
;
7
(
2
):
79
85
51
Piko
BF
,
Keresztes
N
.
Physical activity, psychosocial health, and life goals among youth.
J Community Health
.
2006
;
31
(
2
):
136
145
[PubMed]
52
Prasad
A
,
St-Hilaire
S
,
Wong
MM
,
Peterson
T
,
Loftin
J
.
Physical activity and depressive symptoms in rural adolescents.
N Am J Psychol
.
2009
;
11
(
1
):
173
188
53
Salah
EM
,
Yamamah
GA
,
Megahed
HS
,
Salem
SE
,
El-din
S
,
Khalifa
AG
.
Screening for depressive symptoms and their associated risk factors in adolescent students in South Sinai, Egypt.
Life Sci J
.
2013
;
10
(
3
):
433
443
54
Shepherd
D
,
Krägeloh
C
,
Ryan
C
,
Schofield
G
.
Psychological well-being, self-reported physical activity levels, and attitudes to physical activity in a sample of New Zealand adolescent females.
Psychology (Irvine)
.
2012
;
3
(
6
):
447
453
55
Sigfusdottir
ID
,
Asgeirsdottir
BB
,
Sigurdsson
JF
,
Gudjonsson
GH
.
Physical activity buffers the effects of family conflict on depressed mood: a study on adolescent girls and boys.
J Adolesc
.
2011
;
34
(
5
):
895
902
[PubMed]
56
Soltanian
AR
,
Nabipour
I
,
Akhondzadeh
S
, et al
.
Association between physical activity and mental health among high-school adolescents in Boushehr province: A population based study.
Iran J Psychiatry
.
2011
;
6
(
3
):
112
116
[PubMed]
57
Sun
Y
,
An
J
,
Wang
X
,
Zu
P
,
Tao
FB
.
Gender- and puberty-dependent association between physical activity and depressive symptoms: national survey among Chinese adolescents.
J Phys Act Health
.
2014
;
11
(
7
):
1430
1437
[PubMed]
58
Tao
FB
,
Xu
ML
,
Kim
SD
,
Sun
Y
,
Su
PY
,
Huang
K
.
Physical activity might not be the protective factor for health risk behaviours and psychopathological symptoms in adolescents.
J Paediatr Child Health
.
2007
;
43
(
11
):
762
767
[PubMed]
59
Wiles
NJ
,
Haase
AM
,
Lawlor
DA
,
Ness
A
,
Lewis
G
.
Physical activity and depression in adolescents: cross-sectional findings from the ALSPAC cohort.
Soc Psychiatry Psychiatr Epidemiol
.
2012
;
47
(
7
):
1023
1033
[PubMed]
60
Birkeland
MS
,
Torsheim
TR
,
Wold
B
.
A longitudinal study of the relationship between leisure-time physical activity and depressed mood among adolescents.
Psychol Sport Exerc
.
2009
;
10
(
1
):
25
34
61
Brunet
J
,
Sabiston
CM
,
Chaiton
M
, et al
.
The association between past and current physical activity and depressive symptoms in young adults: a 10-year prospective study.
Ann Epidemiol
.
2013
;
23
(
1
):
25
30
[PubMed]
62
Hume
C
,
Timperio
A
,
Veitch
J
,
Salmon
J
,
Crawford
D
,
Ball
K
.
Physical activity, sedentary behavior, and depressive symptoms among adolescents.
J Phys Act Health
.
2011
;
8
(
2
):
152
156
[PubMed]
63
McPhie
ML
,
Rawana
JS
.
The effect of physical activity on depression in adolescence and emerging adulthood: a growth-curve analysis.
J Adolesc
.
2015
;
40
:
83
92
[PubMed]
64
Neissaar
I
,
Raudsepp
L
.
Changes in physical activity, self-efficacy and depressive symptoms in adolescent girls.
Pediatr Exerc Sci
.
2011
;
23
(
3
):
331
343
[PubMed]
65
Stavrakakis
N
,
de Jonge
P
,
Ormel
J
,
Oldehinkel
AJ
.
Bidirectional prospective associations between physical activity and depressive symptoms. The TRAILS Study.
J Adolesc Health
.
2012
;
50
(
5
):
503
508
[PubMed]
66
Toseeb
U
,
Brage
S
,
Corder
K
, et al
.
Exercise and depressive symptoms in adolescents: a longitudinal cohort study.
JAMA Pediatr
.
2014
;
168
(
12
):
1093
1100
[PubMed]
67
Almas
A
,
Forsell
Y
,
Iqbal
R
,
Janszky
I
,
Moller
J
.
Severity of depression, anxious distress and the risk of cardiovascular disease in a Swedish population-based cohort.
PLoS One
.
2015
;
10
(
10
):
e0140742
[PubMed]
68
Balestri
M
,
Calati
R
,
Souery
D
, et al
.
Socio-demographic and clinical predictors of treatment resistant depression: a prospective European multicenter study.
J Affect Disord
.
2016
;
189
:
224
232
[PubMed]
69
Lasserre
AM
,
Marti-Soler
H
,
Strippoli
MP
, et al
.
Clinical and course characteristics of depression and all-cause mortality: a prospective population-based study.
J Affect Disord
.
2016
;
189
:
17
24
[PubMed]
70
Trivedi
MH
,
Morris
DW
,
Wisniewski
SR
, et al
.
Increase in work productivity of depressed individuals with improvement in depressive symptom severity.
Am J Psychiatry
.
2013
;
170
(
6
):
633
641
[PubMed]
71
Roshanaei-Moghaddam
B
,
Katon
WJ
,
Russo
J
.
The longitudinal effects of depression on physical activity.
Gen Hosp Psychiatry
.
2009
;
31
(
4
):
306
315
[PubMed]
72
National Collaborating Centre for Mental Health
.
The Treatment and Management of Depression in Adults (Updated Edition): National Clinical Practice Guideline 90
.
London, United Kingdom
:
National Institute for Health & Clinical Excellence
;
2010
73
Canadian Paediatric Society & Healthy Active Living Committee
.
Healthy active living for children and youth.
Paediatr Child Health
.
2002
;
7
(
5
):
339
358
[PubMed]
74
Council on Sports Medicine and Fitness
;
Council on School Health
.
Active healthy living: prevention of childhood obesity through increased physical activity.
Pediatrics
.
2006
;
117
(
5
):
1834
1842
[PubMed]
75
Cotman
CW
,
Berchtold
NC
,
Christie
LA
.
Exercise builds brain health: key roles of growth factor cascades and inflammation.
Trends Neurosci
.
2007
;
30
(
9
):
464
472
[PubMed]
76
Silverman
MN
,
Deuster
PA
.
Biological mechanisms underlying the role of physical fitness in health and resilience.
Interface Focus
.
2014
;
4
(
5
):
20140040
[PubMed]
77
Tsatsoulis
A
,
Fountoulakis
S
.
The protective role of exercise on stress system dysregulation and comorbidities.
Ann N Y Acad Sci
.
2006
;
1083
:
196
213
[PubMed]
78
Zahn-Waxler
C
,
Shirtcliff
EA
,
Marceau
K
.
Disorders of childhood and adolescence: gender and psychopathology.
Annu Rev Clin Psychol
.
2008
;
4
:
275
303
[PubMed]
79
Colman
I
,
Jones
PB
,
Kuh
D
, et al
.
Early development, stress and depression across the life course: pathways to depression in a national British birth cohort.
Psychol Med
.
2014
;
44
(
13
):
2845
2854
[PubMed]
80
Madigan
S
,
Brumariu
LE
,
Villani
V
,
Atkinson
L
,
Lyons-Ruth
K
.
Representational and questionnaire measures of attachment: A meta-analysis of relations to child internalizing and externalizing problems.
Psychol Bull
.
2016
;
142
(
4
):
367
399
[PubMed]
81
Brooks
SJ
,
Kutcher
S
.
Diagnosis and measurement of adolescent depression: a review of commonly utilized instruments.
J Child Adolesc Psychopharmacol
.
2001
;
11
(
4
):
341
376
[PubMed]
82
Ekelund
U
,
Tomkinson
G
,
Armstrong
N
.
What proportion of youth are physically active? Measurement issues, levels and recent time trends.
Br J Sports Med
.
2011
;
45
(
11
):
859
865
[PubMed]

Competing Interests

POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.

FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.

Supplementary data