Soc Psychiatry Psychiatr Epidemiol (2011) 46:431441 DOI 10.

1007/s00127-010-0208-0

ORIGINAL PAPER

Role of physical and sedentary activities in the development of depressive symptoms in early adolescence
Anne Mari Sund Bo Larsson Lars Wichstrm

Received: 22 June 2009 / Accepted: 2 March 2010 / Published online: 1 April 2010 Springer-Verlag 2010

Abstract Objective To examine whether levels of physical activity or sedentary activity are risk factors for the development of depressive symptoms in early adolescence. Methods A representative sample of 2,464 12- to 15year-old adolescents living in the middle of Norway was assessed twice, during the years 1998 (T1) and 1999/2000 (T2). The attrition rate was 4.3%. We assessed depressive symptoms (using the Mood and Feelings Questionnaire, MFQ) and levels of physical and sedentary activities at baseline and follow-up. Various potentially confounding factors, including demographic factors, were assessed at baseline. Results The MFQ scores at T1 were cross-sectionally associated with low levels of vigorous exercise. A possible buffering effect of vigorous exercise on the relationship between stressful life events and depression was demonstrated. In longitudinal analysis low levels of vigorous exercise and high levels of sedentary activities (boys only) predicted a high score (MFQ C 25) at T2.

Conclusions Low levels of vigorous exercise and high levels of sedentary activities (boys only) constituted independent risk factors for the development of a high level of depressive symptoms in a 1-year study of young adolescents. This knowledge should be considered by policy makers, preventative services, and health-care professionals. Keywords Vigorous Sedentary Physical activity Depression Longitudinal Youth

Introduction According to the World Health Organization, depression is the second most important disease in terms of the burden of disease (disability and mortality) among 15- to 44-year-olds worldwide [57]. During the few years from early to middle adolescence, levels of depressive symptoms rise sharply, particularly among girls [4]. Depressive symptoms in adolescence are often forerunners of adult depression [45]. Levels of physical activity in early childhood are high and decrease gradually in childhood and further in adolescence [1, 5, 38]. The results of most cross-sectional studies of adolescents have shown associations between high levels of vigorous physical activity and low levels of depressive symptoms [47]. However, other studies have reported no such relationships [2], or have been restricted to social functioning [2], to girls [16, 26], to low to moderate levels of activity only among girls [53], or to moderate to vigorous exercising only among boys [19]. Because depressive symptoms can obviously lead to reduced activity levels, cross-sectional studies can tell us little about the temporal relationship between depression and physical activity. Only a few studies have addressed such relationships longitudinally. In a 2-year follow-up

A. M. Sund B. Larsson Department of Neuroscience, Faculty of Medicine, Norwegian University of Science and Technology, 7489 Trondheim, Norway A. M. Sund St.Olavs University Hospital, Trondheim, Norway L. Wichstrm Department of Psychology, Norwegian University of Science and Technology, 7491 Trondheim, Norway A. M. Sund (&) Department of Neuroscience, Medical Faculty, NTNU, Klostergata 46/48, 7489 Trondheim, Norway e-mail: Anne.M.Sund@ntnu.no

123

432

Soc Psychiatry Psychiatr Epidemiol (2011) 46:431441

study, Motl and collaborators [37] found that changes in physical activity during leisure time were inversely related to changes in levels of depressive symptoms, but no sexbased differences emerged. In a study of 15-year-olds in the USA, participating in a sporting club was inversely related to the development of depressive symptoms, but only among boys [16]. In a recent large total-population study, Wiles [59] found that early adolescents who undertook the recommended levels of physical activity, i.e. at least 1 h every day, had less emotional distress 1 year later than those who undertook less physical activity. In Norway, Sagatun [46] showed that the hours spent on physical activity among 15- to 16-year-olds were negatively associated with emotional symptoms and peer problems 3 years later, but only among boys. Those who exercised at a moderate level (57 h a week) reported the lowest emotional stress levels. In two large cross-sectional school-based studies of young adolescents, high levels of sedentary behaviour correlated positively with levels of depressive symptoms [48] and emotional stress [55]. There are probably multiple mechanisms by which sedentary and exercise behaviours inuence depressive symptoms including both biological pathways and social and psychological factors. There is solid evidence supporting the roles of negative life events and stress in the aetiology of depression [51, 52]. Several constitutional and potentially hard-to-change factors, such as coping styles [31] and genetics [9], may buffer the negative effects of stress. It is also possible that lifestyle factors such as physical activity act as resilience factors mitigating depression. Supporting evidence was found in a study of inpatient adolescent girls aged 14 years [8] and in a general population sample [40]. In both studies, the negative impact of stressful life events on emotional health declined as exercise levels increased. However, these studies included small and non-representative samples of adolescents. In a recent study of young mothers, frequency of leisure time physical activity was acting as a buffer of the association between life stress and depressive symptoms [14]. To the best of our knowledge, this hypothesis has not yet been tested by specically addressing depression in a large representative sample. While a few longitudinal studies that have investigated the inuence of physical activity on depression in adolescents, the evidence is conicting [36, 47, 55]. To date, few studies have included measures of vigorous, non-vigorous and sedentary activities, and stressful events together with a wide range of potentially confounding variables in the assessment of risk factors for the development of high levels of depressive symptoms among adolescents. A review of the literature suggests that the following variables might be confounding factors: age [43, 60], sex [47], family structure [34], socio-economic status (SES) [28,

60], ethnicity [43, 47, 51], somatic disease [23, 32, 33], pubertal development [21], body mass index (BMI) [18], obesity [20] and physical impairment [33]. The aims of the present study were to examine 1. the cross-sectional and 1-year longitudinal relationships between vigorous exercise, non-vigorous physical and sedentary activities versus depressive symptoms, while controlling for possible confounding variables, including demographic factors, among 12- to 15-yearold Norwegian girls and boys; and the hypothesis that vigorous exercise moderates the effects of high levels of stressful life events on levels of depressive symptoms.

2.

Methods Procedure This study was conducted at the schools as part of The Youth and Mental Health Study, a longitudinal study of depressive symptoms and disorders in 12- to 15-year-old school adolescents in two counties in the middle of Norway, South and North Trndelag, during the years 1998 (T1) and 1999/2000 (T2). These areas had a population of 390,000 inhabitants and included one large city, Trondheim, with 150,000 inhabitants. The total population of 12 to 15-year-olds attending 8th and 9th grades in private or public schools during the autumn of 1998 in South and North Trndelag was 9,292, 98.5% of whom attended public schools. The schools are highly integrated, i.e. also including mildly retarded and physically handicapped adolescents. Twenty-nine pupils who attended special schools were not included in the study. In addition, 534 pupils at the smallest schools, e.g. 5.7% of the total population were not included mostly because of practical reasons. A cluster sampling method was chosen using the schools as sampling units. This strategy was chosen to reduce costs (e.g. travelling costs) and to achieve a high response rate. The schools were drawn with a probability according to size (proportional allocation) within each stratum, i.e. for each individual drawn from the total population sample, all pupils in 8th and 9th grades at the same schools were selected. This resulted in 22 schools and 2,813 pupils. Twenty-one pupils (0.7%) were not eligible for the following reasons: admitted to hospital, being temporarily abroad or lacking sufcient knowledge in Norwegian language (recently arrived to Norway). Thus, a total of 2,792 adolescents were eligible for the study. At each school, a member of the staff was trained and was responsible for the collection of the written consents both from the adolescents and the parents and for the data

123

Soc Psychiatry Psychiatr Epidemiol (2011) 46:431441

433

collection. Of the 2,792 eligible adolescents, 2,464 (88.3%) participated in the study at T1 (50.8% girls), stratied according to urbanity and geography: 1: City of Trondheim (N = 484, 19.5%. 2. Suburbs of Trondheim (N = 432, 17.5%) 3. Coast (N = 405, 16.4%) 4. Inland (N = 1,143, 46.4%). The Regional Ethics Committee for Medical Research and the Norwegian Data Inspectorate, the local school authorities in the two counties, and the school boards approved the study. Study sample At T1, 2,464 adolescents participated (50.5% girls) with a mean age of 13.7 years (range 12.515.7, SD 0.58) and at T2 14.9 years (range 13.717.0, SD 0.6). The attrition rate was 4.2%. Only those adolescents who had participated at both time points (N = 2,360) were included in the longitudinal analyses of the present study. The average time between T1 and T2 was 1.1 year (range 0.941.36, SD 0.12) and depended upon follow-up interviews. Measures at T1 Depressive symptoms The Mood and Feelings Questionnaire (MFQ) [3]. This is a 34-item questionnaire designed for children and adolescents of 818 years to report depressive symptoms as specied by the DSM-III Revised criteria [6]. The MFQ covers affective, melancholic, vegetative, cognitive and suicidal aspects of depression. The individual is asked to report his or her feelings during the preceding 2 weeks on a three-point scale (0 = not true, 1 = sometimes true, and 2 = true) and the total summed scores range between 0 and 68. The psychometric properties of the MFQ have previously been shown to be satisfactory [13], as they were for the present sample [50]. The MFQ has been used to screen for depressive symptoms among adolescents in clinical samples [15] and in the general population [11]. In the present sample, 18 subjects (0.8%) who had omitted more than 10% of the items were not included in the analysis, and the missing values for those with a lower proportion of omitted items were replaced with the mean MFQ item score. The total summed scores were used in the bivariate and linear regression analyses. Physical and sedentary activities Short self-reported measures of vigorous physical activity among young people in mid-adolescence have shown acceptable validity and reliability [7]. Four questions about physical activity, modied according to Paffenbarger [41],

were used. Vigorous exercise was assessed by asking the adolescents about the number of hours they exercised vigorously per week, using ve response categories ranging from less than 3 h to more than 14 h (including gym in school, which represented about 2 h every week). Nonvigorous physical activities (e.g., walking, bicycling, playing, skateboarding) performed each day were assessed in four response categories, ranging from less than one hour to more than four hours. Time spent on sedentary activities everyday outside school (e.g. homework, reading, watching TV, games) was assessed in four response categories ranging from less than three hours to more than six hours. The variables were used as grouping variables in the bivariate analyses and ordinal scales in the regression analyses. In addition, the question What do you like to do? included ve response alternatives ranging from considering oneself being physically sedentary to physically very active. The response on this variable at T1 was used to validate the other physical activity variables, showing moderate high correlation with vigorous activity (r = 0.6, P \ 0.01), but low correlation with non-vigorous activity (r = 0.16, P \ 0.01) and low negative correlation with the sedentary variable (r = -0.15, P \ 0.01). Stressful life events The adolescents answered a 33-item questionnaire constructed for the purpose of the present study, the Early Adolescence Stress Questionnaire (EASQ) [51]. In the EASQ, the sum of stressful life events and chronic stressors salient to young adolescents, in the school, family, and social network domains during the previous 12 months are recorded. The mean number of reported stressful events for the previous year at T1 was 4.5 (SD 3.6) for the whole sample, and no signicant sex difference was found [51]. In the bivariate and regression analyses, the total sum of stressful events was used. A dichotomized variable was used when visualizing the interaction analyses, i.e. a cutoff was set at ve stressful events. To avoid confounding between stress and depressive symptoms at T2, only events assessed at T1 were used in the longitudinal analysis. Pubertal stage The adolescents completed the Pubertal development scale [44] (range 14). Most were in the initial (37.5%) or middle (34.4%) phase of puberty. Girls were signicantly more advanced in their pubertal development than boys (Z2,226 = -12.5, P \ 0.001). The variable was treated as a grouping variable in the bivariate analyses and as a continuous variable in the regression analyses.

123

434

Soc Psychiatry Psychiatr Epidemiol (2011) 46:431441

BMI Self-reported weight and height data were converted to Quetlets body mass index [56]. This score was subdivided into three categories:\10th percentile (BMI \ 16.2 kg/m2), between the 10th and 90th percentiles (BMI = 16 22.5 kg/m2) and [90th percentile (BMI [ 22.5 kg/m2). The mean BMI for the whole sample was 19.2 kg/m2 (range 11.838.8, SD = 2.7). The variable was treated as a grouping variable in the bivariate analyses, and as a dummy variable or a continuous variable in the regression analyses. Somatic health The adolescents were asked about their current somatic health with the open-ended question Do you have any disease? The answers were coded according to the International Classication of Primary Care criteria [24] and grouped into none, one, two and three or more diseases. The variable was treated as a grouping variable in the bivariate analysis and an ordinal scale in the regression analysis. The reported diseases were predominantly allergy (24.1%), hay fever (28.1%) and asthma (21.4%). Physical impairment The adolescents were asked whether they had reduced (no/ yes) their leisure/sport activities/seeing friends because of disease/injury or pain during the previous 12 months. Demographic variables Age, sex, parental SES, ethnicity and living situation were included in the analyses. The parental SES was measured by classifying the mothers and fathers occupations according to the international classication of occupation, ISCO-88 [25]. A distinction was made between those subjects with one or two parents of Norwegian background and those with a non-Norwegian background. Subjects who lived with both parents or shared their time equally between their parents were contrasted with all other domestic situations. Measures at T2 Depressive symptoms (MFQ) At T2, the MFQ was readministered to the adolescents and the total scores were dichotomized at the 90th percentile to subdivide the subjects into low and high scorers in the logistic regression analysis.

Physical and sedentary activities Questions identical to those asked at T1 regarding selfreported perception of physical activity, vigorous exercise, non-vigorous physical and sedentary activities and on impairment the previous year because of somatic illness or injury. Statistical analysis The data were analysed with the statistical package for the social sciences (SPSS), version 15. Correlations between the continuous data were assessed by the Pearson productmoment coefcient. Student t test or ANOVA/ANCOVA was used for parametric data and the MannWhitney test was used for non-parametric data to assess the differences between groups. Bonferroni corrections were used to adjust for multiple comparisons in post hoc analyses. Paired t tests were used for parametric analyses and Wilcoxons test for paired non-parametric analyses. All tests were two-tailed and an a level of P \ 0.05 indicated statistical signicance. The assessed variables that showed bivariate relationships with depressive symptom levels at T1, in addition to demographic variables, were entered into a subsequent stepwise hierarchical multiple linear regression analysis to predict depressive symptom scores cross-sectionally. Hierarchical logistic regression analyses were performed to investigate whether vigorous exercise, non-vigorous activities, and sedentary activities, in addition to potentially confounding factors, predicted levels of depressive symptoms longitudinally. In the regression analysis, vigorous exercise, non-vigorous activities, sedentary variables, pubertal groups and somatic health variables were treated as ordinal variables and were transformed into Z scores, and the scores of the vigorous exercise variable reversed. The following variables were transformed into dummy variables: sex, SES, living situation, ethnic group, physical impairment, and BMI categories. The variables were entered in three steps with a backward elimination procedure at each step, and variables with a P value of B0.10 were retained in the analysis. Step 1 contained variables found to predict depression in an earlier analysis of the present sample [51]. Step 2 contained all potentially confounding variables, and step 3 included all variables of major interest. In a fourth step, potential interaction variables were introduced individually. There were no problems with multicollinearity between the explaining factors. In the hierarchical logistic regression analysis, the HosmerLemeshow test for goodness of t [22] was used to predict the likelihood that an individual would be a low or high scorer on the MFQ at T2.

123

Soc Psychiatry Psychiatr Epidemiol (2011) 46:431441 Table 1 Levels of depressive symptoms on the MFQ, vigorous exercise, non-vigorous physical and sedentary activities for the whole sample and by sex over 1 year in 2,360 Norwegian adolescents and the results of between-group differences Z989 = -5.1***d

435

t1,182 = -3.7***c

Results Gender differences and changes over the 1-year period Using identical measures to those used at T1, the adolescents were reassessed 1 year later (T2) with an attrition rate of 4.3% (N = 105). The adolescents who did not participate at T2 were characterized by having signicantly higher mean total scores on the MFQ at T1 (17.3 vs. 10.4 for the continuing subjects) (t2,442 = 7.13, P \ 0.001) [52], more frequent non-Norwegian backgrounds [52] and lower levels of vigorous exercise at T1 (t2,405 = 3.1, P \ 0.01). No sex, grade, or SES differences between the two groups were observed. Girls had signicantly higher mean total levels of depressive symptoms at both T1 and T2 than those of boys. Although the mean total MFQ score for the whole sample did not increase from T1 to T2, a signicant difference between the sexes was observed (P \ 0.001). The girls showed an increase in depressive symptom levels between the two time points, whereas a small nonsignicant reduction in symptom levels was observed among boys (Table 1). Boys displayed signicantly higher levels of vigorous exercise and non-vigorous physical activities than did girls at both time points. However, no sex-based difference was observed for levels of sedentary activities. The whole sample showed a decline in vigorous exercise and nonvigorous physical activity from T1 to T2. However, the decline in vigorous exercise among boys was nonsignicant. An increase in level of sedentary activities was observed for the whole sample (Table 1). Cross-sectional analysis at T1 Bivariate analyses The results of variables were examined in successive ANOVA/ANCOVAs. Interaction effects between sex and depressive symptom levels and all the other study variables were examined. Our results showed that depressive symptom levels at T1 increased signicantly with lower levels of vigorous exercise and higher levels of sedentary

t2,347 = 3.6***a n.s

2.56 (1.1) Z2,280 = 5.2***b

2.15 (1.0) Z2,315 = 7.6***b 1.93 (0.9) 2.26 (1.0) Z2,267 = 4.4***b 2.04 (0.9)

Sex difference

10.6 (11.8) 13.4 (11.2) 7.8 (8.5)

Girls mean (SD)

2.20 (0.8)

T2

t2,330 = 11.3***

2.60 (1.0) Z2,310 = 9.1***b

Sex difference

2.37 (1.0)

1.99 (0.8)

All mean (SD)

1.98 (0.8)

2.0 (0.9)

Boys mean (SD)

n.s.

Z2,249 = -7.1***d Z1,140 = -4.9**d

T1T2 differences

Girls

Z1,145 = -2.1*d Z2,270 = 2.13*d

There were few missing data for most of the variables at T1, ranging from 1 to 3%. The levels of missing data were higher for physical impairment (11.1%) and the BMI (13.1%) variables. In the nal analyses, 19.7% of the data had missing values in the cross-sectional regression analysis, and 7.8% in the longitudinal logistic regression analysis.

Z2,198 = 5.1***d

All

Z1,105 = -4.5***d Z1,093 = -2.9**d

Missing data

Boys

n.s.

n.s.

Non-vigorous activities 2.16 (1.0) 2.07 (0.9) (14)

Student t test, bMannWhitney, cPaired t test, dWilcoxon paired test

a

10.3 (9.2) 12.4 (10.1) 8.2 (7.4)

2.42 (0.9) 2.24 (0.8)

1.84 (0.8) 1.85 (0.8)

1.83 (0.9) n.s.

Boys mean (SD)

* P \ 0.05 ** P \ 0.01 *** P \ 0.001

Variable

MFQ the mood and feelings questionnaire

All mean (SD)

T1

Girls mean (SD)

Sedentary activities (14)

Vigorous exercise (15)

MFQ (064)

123

436 Table 2 Mean total MFQ scores with standard deviations within parenthesis, the results of ANOVAS/ANCOVAs using levels of vigorous exercise, non- vigorous physical activities, sedentary Variable Categories

Soc Psychiatry Psychiatr Epidemiol (2011) 46:431441 activities, pubertal status, BMI group, number of somatic diseases and reduced activity levels because of disease/injury as grouping factors at T1 (N = 2,464) N Mean total MFQ Main effects Score (SD) 288 13.7 (10.4) 1,168 11.2 (10.0) 683 9.2 (8.5) 174 8.8 (8.4) 79 8.6 (7.1) 655 11.5 (9.6) 933 10.5 (9.6) 498 10.0 (9.4) 264 10.1 (9.4) 922 10.4 (9.4) 1,015 10.5 (9.4) 327 11.0 (10.0) 116 13.4 (9.5) 62 10.4 (9.9) 417 11.0 (9.0) 528 13.4 (10.4) 198 16.0 (10.4) 228 7.3 (7.0) 487 8.6 (7.9) 323 8.6 (8.1) 98 8.8. (7.9) 127 11.7 (10.0) 830 12.1 (9.7) 115 17.4 (11.9) 100 8.7 (8.7) 836 8.2 (7.8) 108 8.4 (7.8) 1,631 10.5 (9.7) 653 10.4 (8.8) 136 13.0 (10.1) 24 11.7 (11.4) 2,240 10.4 (9.3) 204 13.7 (11.6) F1,2441 = 24.5*** 2 [ 1*** n.s n.s. F2,1069 = 14.43*** 3 [ 1, 2*** n.s. F3,1201 = 12.41*** 4 [ 1, 2** 4 [ 3* 3 [ 1, 2* F3,2375 = 5.55** 4 [ 1, 2*** 4 [ 3* n.s. F4,2386 = 10.3*** Bonferroni posthoc comparisons 1 [ 2, 3, 4*** 2 [ 3* 1 [ 5*

Vigorous exercise per week (including gym at school, h) 1. \3 2. 36 3. 710 4. 1114 5. [14 Non- vigorous physical activity per day (h) 1. \1 2. 12 3. 34 4. [4 Sedentary activities per day (h) 1. \3 2. 34 3. 56 4. [6 Pubertal group girls 1. Prepub 2. Beginning- pub 3. Mid-pub 4. Advanced pub Pubertal group boys 1. Prepub 2.Beginning-pub 3. Mid- pub 4. Advanced pub BMI group (kg/m ) Girls
2

1. \16.2 2. 16.222.5 3. [22.5

BMI group (kg/m2) boys

1. \16.2 2. 16.222.5 3. [22.5

Number of medical diseases

1. None 2. One 3. Two 4. Three or [

Physical impairment

1. No 2. Yes

* P \ 0.05 ** P \ 0.01 *** P \ 0.001

activities, with increased pubertal development (only for girls), higher BMI category (only for girls), and with the presence of physical impairment some times during the last year (for details, see Table 2). Moderate correlations were observed between the number of stressful events and depressive symptom levels [(r = 0.50, P \ 0.01) for girls and (r = 0.52, P \ 0.01) for boys], whereas a nonsignicant relationship with levels of vigorous exercise was evident for both sexes.

Multivariate analysis Hierarchial multiple linear regression was rst employed to determine if vigorous exercise improved prediction of depressive symptom levels beyond the main predictors found in earlier cross-sectional analysis of the present sample [51] while controlling for potential confounders (see Table 3). In step 1, the model was signicant after including gender, ethnicity and sum of stressful life events

123

Soc Psychiatry Psychiatr Epidemiol (2011) 46:431441

437

Table 3 The results of hierarchical multiple regression analyses using depressive symptom (MFQ) scores at T1 as an outcome measure (N = 1,979) Predictor Step 1 Gendera Ethnic groupb Stressful life events Step 2 Pubertal group BMI [ 22.5 kg/m2 Step 3 Vigorous exercise
c

Unstandardizedcoefcient (B)

Standard error (SE)

t value

sr2 change (unique) (%)

Total R2 (%) 29.8

4.14 3.56 1.25 0.42 2.12 0.81 1.43 4.22 0.14

0.33 1.03 0.04 0.57 0.18 0.18 0.36 1.14 0.50

12.63*** 3.44** 28.35*** 2.29* 3.75*** 4.47*** 3.95*** 3.71*** 2.83**

4.4 0.3 23.9 30.7 0.2 0.5 31.4 0.7 0.5 0.5 0.3

Two- way interactions Pubertal group 9 gender BMI [ 22.5 kg/m2 9 gender Vigorous exercise 9 stressful events
a b c

Reference group: girls At least one parent Norwegian versus all other non-Norwegian parental groups

The vigorous exercise variable is reversed * P \ 0.05, ** P \ 0.01, *** P \ 0.001 Before interactions: unique variability = 30.0%, shared variability = 1.2%

in the equation, R2 = 0.30, F3,1974 = 280.76, P \ 0.001. The model improved signicantly after including pubertal group and BMI group in step 2, R2 = 0.307, F2,1972 = 11.27, P \ 0.001 and vigorous exercise in step 3, R2 = 0.314, F1,1971 = 19.95, P \ 0.001. (see Table 3 including only signicant variables). No effect of nonvigorous physical activities was observed. Analysis of two-way interactional effects between pubertal group by sex and BMI group by sex on depressive symptom levels showed that they remained signicant in the multivariate analysis. For both sexes, an interactional effect was observed between levels of vigorous exercise and stressful events, step: R2 = 0.32, F1,1970 = 8.0, P \ 0.01 (see Table 3). The nature of the interaction showed that higher levels of vigorous activities were associated with a stronger reduction in depressive symptoms when the levels of stressful events were high. The dichotomized stressful event variable was used to visualize this interaction (Fig. 1). Longitudinal analyses To predict group membership, i.e. low versus high scorers on the MFQ at T2, a hierarchical logistic regression analysis was performed using the dichotomized MFQ variable at T2 as the dependent variable, while controlling for the depressive symptom level at T1 and other potentially confounding variables. The impairment variable at T2 was

Low stress (<=5last year)

20

High stress (>5 last year)

18 16 14 12

MFQ

10 8 6 4 2 0 <3 3-6 7-10 11-14 > 14

hours

hours

hours

hours

hours

Vigorous exercise per week

Fig. 1 Interaction (P \ 0.01) at T1 showing mean total sum scores of depressive symptoms (MFQ) as a function of levels of stressful events (15 vs.[5) last year and hours of vigorous exercise per week among a representative sample of Norwegian adolescents (N = 1,979)

also included in the rst step of the analysis, but emerged as nonsignicant. Both high levels of sedentary activities and low levels of vigorous exercise at T1 signicantly

123

438

Soc Psychiatry Psychiatr Epidemiol (2011) 46:431441

Table 4 The results of hierarchical logistic regression analysis using depressive symptoms, sex, stressful life events, ethnicity, sedentary activities and vigorous exercise at T1 (N = 2,175) as predictors of severe depressive symptoms (MFQ C 25) 1 year later Predictor Step 1 MFQ T1 Sex Stressful life events T1 Step 2 Ethnicity Step 3 Sedentary activities Vigorous exercisea Interaction Sedentary activities 9 sex
a

Unadj OR Exp(B)

95% CI

Adj OR Exp(B)

95% CI

14.49*** 3.09*** 1.18*** 3.75*** 1.20* 1.42*** 1.46*

10.4520.11 2.284.19 1.141.22 2.076.82 1.021.41 1.201.68 1.032.06

8.58*** 2.42*** 1.09*** 3.66*** 1.22* 1.23* 1.45*

5.8212.64 1.703.44 1.051.14 1.817.42 1.021.47 1.011.49 1.002.11

Unadj OR unadjusted odds ratio, Adj OR adjusted odds ratio, CI condence interval The scores of vigorous exercise reversed * P \ 0.05, ** P \ 0.01, *** P \ 0.001

predicted the likelihood of being a high scorer (MFQ C 25) 1 year later. The increment in the model t attributed to sedentary activities and levels of vigorous exercise was signicant (v2 = 9.22 df = 1, P \ 0.05). The result of the HosmerLemeshow test was nonsignicant, indicating a good model t (P = 0.98). A signicant sex by sedentary activities interaction effect was also found in that sedentary activities was signicant only for boys in predicting high scorers (adjusted odds ratio = 1.53; 95% condence interval: 1.152.03; Table 4).

Discussion The overall aim of this study was to examine the relationships between physical and sedentary activities in regard to depressive symptom levels in cross-sectional and 1-year longitudinal studies of a large, representative sample of early adolescents in Norway. After we controlled for possible confounders, low levels of vigorous exercise remained an independent predictor of depressive symptom levels in the cross-sectional analysis, a nding consistent with previous reports [28, 47]. Low levels of vigorous exercise for the whole sample and high levels of time spent on sedentary activities among boys were predictors of high levels of depressive symptoms 1 year later. This protective effect of vigorous exercise against depressive symptoms 1 year later, while controlling for various confounding variables, has seldom been reported. To the best of our knowledge, the nding that sedentary activities predict high levels of depressive symptoms among young adolescent boys has not been shown in earlier research using a longitudinal representative design, and

extends the ndings of a previous cross-sectional studies that showed a relationship between low levels of sedentary activities and psychological well being among adolescents [55] and with depression and anxiety levels among children [42]. Although the sizes of the relationships in the present study were modest, they still indicate how low levels of vigorous exercise and high levels of sedentary activities might signal future depressive problems. A difculty encountered when interpreting the results of the cross-sectional analysis is the fact that slower psychomotor activity and lower levels of physical activity among adolescents might also be an inherent characteristic of a depressive state. The ndings of the present longitudinal analysis might be explained by a common third factor related to both activity levels and the emergence of depressive symptoms. The mechanisms by which sedentary and exercise behaviours inuence depressive symptoms include both biological pathways, social and psychological factors. Levels of monoamines (noradrenalin and serotonin) are reduced in depression. Possible mechanisms are that the neurotransmitters increase and that endorphins increase during hard physical exercise [35]. Theoretically, physical inactivity might inuence the body in the opposite way, by reducing the levels of these agents and thus precipitating depressive feelings. A third hypothesis emphasizes the role played by exercise in regulating stressful signals to the limbic system and preventing the hypersecretion of cortisol [54]. A positive effect of exercise on sleep among children has been reported [17, 49], and might be relevant to the frequent occurrence of sleep problems during depression. Psychosocial factors might be related to feelings of loneliness elicited by sedentary activities, which often take

123

Soc Psychiatry Psychiatr Epidemiol (2011) 46:431441

439

place in solitude (e.g., playing computer games). The opposite effect might also occur during vigorous exercise, which is often undertaken with peers in this age group [58]. Spending time on sedentary activities might take time from physical activities or time spent with family or friends. Rumination is a coping style that increases from childhood to adolescence [10] and is associated with the duration of depressive symptoms [39] and the onset of depressive episodes among young adults [27]. Being sedentary might provoke rumination to a greater extent than does physical activity. Conversely, exercising might elicit a sense of mastering the body, an increase in self-esteem and positive cognitions that counteract the development of depression. Sex differences The nding that only girls who were more advanced in their pubertal development (in the upper ten BMI percentile) having higher levels of depressive symptoms also disappeared in the longitudinal analysis, showed that these factors had no lasting effect. No sex-based differences in the relationships between vigorous activity and depressive symptom levels were found in the cross-sectional or longitudinal analyses. However, sedentary activities seem to be more harmful for the mental health of early adolescent boys than girls in the longitudinal 1-year analysis, which is consistent with the report by Johnson et al. [26], who noted that high sedentary activity among girls was related to the lowest levels of depressive symptoms. Stressful events In this study, interactive effects between stressful life events and vigorous exercise on depressive symptom levels were identied. Vigorus exercise was associated with a greater reduction in depressive symptoms when levels of stressful events were high. Thus, vigorous exercise had a possible buffering effect on stress. These results provide support for the stress-buffering effects of excercising; ndings in line with recent results of a study of young mothers [14].Our ndings are also consistent with earlier research in small clinical [8] and general population samples [40]. In the context of high levels of stress, vigorous exercising might represent a positive coping strategy counteracting depression. Thus, exercise could have a stealing effect, i.e. distracting the adolescent from a stressful burden. Second, by strengthening the individual physically, this might have positive effects on thought processes and concentration, possibly inducing a more problem-solving strategy towards stressful events. However, these possible interpretations remain to be tested empirically. The nding in the present study supports the theory that exercise activities have a buffering effect on the

relationship between stressful life events and depression, but the cross-sectional nature of the ndings is open to alternate interpretations. Strengths and limitations All the adolescents participated in compulsory physical exercise lessons at their schools. The effects of exercising at some level, compared with no exercise at all, might therefore have been concealed in the study. Possible confounding factors were measured only at T1 and other confounding factors may have inuenced the results. However, the fact that non-vigorous physical activities had no effect suggests that vigorous exercise exerts stronger protective effects than does being generally active. Because the dropouts at T2 were likely to be less physically active than continuing subjects and had higher MFQ scores at T1, the ndings of the longitudinal study might be biased towards weakened relationships. Whereas no inferences can be drawn about causality from the present study, risk factors might be assessed based on the longitudinal design [29]. Depressive symptoms were measured at only two time points, 1 year apart. While it is possible that these ndings only mirror mood swings among adolescents, half of the high scorers at T1 were still high scorers at T2 [52]. We used a self-reported measure of depressive symptoms, the MFQ, and these ndings are not readily applicable to depressive disorders. Measurement of physical activity among young people is complicated by the cognitive, physiological and other changes that occur during adolescent development, in addition to more intermittent patterns of habitual physical activity in youth [12]. Whereas self-reported measures of exercise and physical activity may have limited validity in children, it has been shown that the quality of self-reported information increases with age [7]. Therefore, it is likely to be valid for our participants, who had reached junior high school. A more general question about self-perception of levels of physical activity also concurred with the quantitative self reports, supporting to some extent the validity of the measurements in the present study. Still, because of some ambiguity regarding the questions posed, there might be some risk for misclassication. Today, importance is laid on measuring intensity, frequency and duration during a week including weekends [12]. In the present study, we asked about total sum of hours spent on exercising, on nonvigorous physical activity and sedentary activities per week. However, more-objective measures, such as monitoring physical activities based on systematic observations, accelerometry, pedometri, oxygen uptake or diaries, might have yielded more valid and reliable information. The strengths of the present study are the large representative

123

440

Soc Psychiatry Psychiatr Epidemiol (2011) 46:431441 behaviour in schoolchildren (HBSC) survey: a population study. Br J Sports Med 35:263267 Brown JD, Lawton M (1986) Stress and well-being in adolescence: the moderating role of physical exercise. J Human Stress 12:125131 Caspi A, Sugden K, Moftt TE, Taylor A, Craig IW, Harrington H, McClay J, Mill J, Martin J, Braithwaite A, Poulton R (2003) Inuence of life stress on depression: moderation by a polymorphism in the 5-HTT gene. Science 301:386389 Compas BE, Orosan PG, Grant KE (1993) Adolescent stress and coping: implications for psychopathology during adolescence. J Adolesc 16:331349 Cooper PJ, Goodyer I (1993) A community study of depression in adolescent girls. I: Estimates of symptom and syndrome prevalence. Br J Psychiatry 163(369374):379380 Corder K, Ekelund U, Steele RM, Wareham NJ, Brage S (2008) Assessment of physical activity in youth. J Appl Physiol 105:977987 Costello EJ, Angold A (1988) Scales to assess child and adolescent depression: checklists, screens, and nets. J Am Acad Child Adolesc Psychiatry 27:726737 Craike MJ, Coleman D, Macmahon C Direct and buffering effects of physical activity on stress-related depression in mothers of infants. J Sport Exerc Psychol 32:23-38 Daviss WB, Birmaher B, Melhem NA, Axelson DA, Michaels SM, Brent DA (2006) Criterion validity of the Mood and Feelings Questionnaire for depressive episodes in clinic and non-clinic subjects. J Child Psychol Psychiatry 47:927934 Desha LN, Ziviani JM, Nicholson JM, Martin G, Darnell RE (2007) Physical activity and depressive symptoms in American adolescents. J Sport Exerc Psychol 29:534543 Dworak M, Wiater A, Alfer D, Stephan E, Hollmann W, Struder HK (2008) Increased slow wave sleep and reduced stage 2 sleep in children depending on exercise intensity. Sleep Med 9:266 272 Erickson SJ, Robinson TN, Haydel KF, Killen JD (2000) Are overweight children unhappy?: Body mass index, depressive symptoms, and overweight concerns in elementary school children. Arch Pediatr Adolesc Med 154:931935 Fulkerson JA, Sherwood NE, Perry CL, Neumark-Sztainer D, Story M (2004) Depressive symptoms and adolescent eating and health behaviors: a multifaceted view in a population-based sample. Prev Med 38:865875 Goodman E, Whitaker RC (2002) A prospective study of the role of depression in the development and persistence of adolescent obesity. Pediatrics 110:497504 Hayward C, Sanborn K (2002) Puberty and the emergence of gender differences in psychopathology. J Adolesc Health 30:49 58 Hosmer DW, S L (1989) Applied Logistic Regression. John Wiley, Chichester Huurre TM, Aro HM (2002) Long-term psychosocial effects of persistent chronic illness. A follow-up study of Finnish adolescents aged 16 to 32 years. Eur Child Adolesc Psychiatry 11:8591 International Classication of Primary care (ICPC) WONCA (1990) Tano, Oslo International Labour Ofce (1990) ISCO-1988. International Standard Classication of Occupation International Ofce, Geneva Johnson CC, Murray DM, Elder JP, Jobe JB, Dunn AL, Kubik M, Voorhees C, Schachter K (2008) Depressive symptoms and physical activity in adolescent girls. Med Sci Sports Exerc 40:818826 Just N, Alloy LB (1997) The response styles theory of depression: tests and an extension of the theory. J Abnorm Psychol 106:221 229

sample, the high response rate, the small attrition rate at T2, and that both physical and sedentary activities were measured.

8.

9.

Conclusion Low levels of vigorous exercise for both sexes and high levels of sedentary activities for boys emerged as risk factors for high levels of depressive symptoms at the 1-year follow-up. Conversely, vigorous exercise possibly moderates the effects of high levels of stressful events on depressive symptom levels. Increasing vigorous exercise and reducing sedentary activities might improve the mental health of adolescents in the general population. This knowledge should be considered by policy makers, preventative services and health-care professionals. In a treatment outcome study of adults, exercise was suggested to be an alternative or adjunct to both psychotherapy and medication for mild to moderate depression [36]. Very limited research exists on the treatment of adolescents, with no conclusive evidence for positive effects of physical activity on depression [30]. Low risk and readily accepted therapies are required for this age group. Whether increased levels of vigorous exercise and a reduction in sedentary activities also reduce depressive symptom levels in adolescents should be evaluated in treatment outcome studies.
Acknowledgments This study was supported by grants from the Research Council of Norway and the National Council for Mental Health, Norway. 19. 10.

11.

12.

13.

14.

15.

16.

17.

18.

References
20. 1. Allison KR, Adlaf EM, Dwyer JJ, Lysy DC, Irving HM (2007) The decline in physical activity among adolescent students: a cross-national comparison. Can J Public Health 98:97100 2. Allison KR, Adlaf EM, Irving HM, Hatch JL, Smith TF, Dwyer JJ, Goodman J (2005) Relationship of vigorous physical activity to psychologic distress among adolescents. J Adolesc Health 37:164166 3. Angold A, Costello EJ, Pickles EJ, Winder F (1987) The development of a questionnaire for use in epidemiological studies of depression in children and adolescents. Medical Research Council, London 4. Angold A, Erkanli A, Silberg J, Eaves L, Costello EJ (2002) Depression scale scores in 817-year-olds: effects of age and gender. J Child Psychol Psychiatry 43:10521063 5. Armstrong, Mechelen V (1998) Are young people young and active? In: Biddle S (ed) Young and Active. Health Education Authority, London, p 69-97 6. Association AP (1987) Diagnostic and Statistical manual of Mental Disorders DSM-III.R (third edtion revised). American Psychiatric Association, Washington DC 7. Booth ML, Okely AD, Chey T, Bauman A (2001) The reliability and validity of the physical activity questions in the WHO health

21.

22. 23.

24. 25.

26.

27.

123

Soc Psychiatry Psychiatr Epidemiol (2011) 46:431441 28. Kantomaa MT, Tammelin TH, Ebeling HE, Taanila AM (2008) Emotional and behavioral problems in relation to physical activity in youth. Med Sci Sports Exerc 40:17491756 29. Kraemer HC, Kazdin AE, Offord DR, Kessler RC, Jensen PS, Kupfer DJ (1997) Coming to terms with the terms of risk. Arch Gen Psychiatry 54:337343 30. Larun L, Nordheim LV, Ekeland E, Hagen KB, Heian F (2006) Exercise in prevention and treatment of anxiety and depression among children and young people. Cochrane Database Syst Rev 3:CD004691 31. Lee A, Hankin BL (2009) Insecure attachment, dysfunctional attitudes, and low self-esteem predicting prospective symptoms of depression and anxiety during adolescence. J Clin Child Adolesc Psychol 38:219231 32. Lewinsohn PM, Roberts RE, Seeley JR, Rohde P, Gotlib IH, Hops H (1994) Adolescent psychopathology: II. Psychosocial risk factors for depression. J Abnorm Psychol 103:302315 33. Lewinsohn PM, Seeley JR, Hibbard J, Rohde P, Sack WH (1996) Cross-sectional and prospective relationships between physical morbidity and depression in older adolescents. J Am Acad Child Adolesc Psychiatry 35:11201129 34. Lindquist CH, Reynolds KD, Goran MI (1999) Sociocultural determinants of physical activity among children. Prev Med 29:305312 35. Martinsen EW (1994) Physical activity and depression: clinical experience. Acta Psychiatr Scand Suppl 377:2327 36. Martinsen EW (2008) Physical activity in the prevention and treatment of anxiety and depression. Nord J Psychiatry 62(Suppl 47):2529 37. Motl RW, Birnbaum AS, Kubik MY, Dishman RK (2004) Naturally occurring changes in physical activity are inversely related to depressive symptoms during early adolescence. Psychosom Med 66:336342 38. Nader PR, Bradley RH, Houts RM, McRitchie SL, OBrien M (2008) Moderate-to-vigorous physical activity from ages 9 to 15 years. Jama 300:295305 39. Nolen-Hoeksema S, Morrow J, Fredrickson BL (1993) Response styles and the duration of episodes of depressed mood. J Abnorm Psychol 102:2028 40. Norris R, Carroll D, Cochrane R (1992) The effects of physical activity and exercise training on psychological stress and wellbeing in an adolescent population. J Psychosom Res 36:5565 41. Paffenbarger RS Jr, Blair SN, Lee IM, Hyde RT (1993) Measurement of physical activity to assess health effects in free-living populations. Med Sci Sports Exerc 25:6070 42. Partt G, Pavey T, Rowlands AV (2009) Childrens physical activity and psychological health: the relevance of intensity. Acta Paediatr 98:10371043 43. Pate RR, Heath GW, Dowda M, Trost SG (1996) Associations between physical activity and other health behaviors in a representative sample of US adolescents. Am J Public Health 86:15771581 44. Petersen AC, Crockett L, Richards MAB (1988) A self-report measure of pubertal status: reliability, validity and initial norms. J Youth Adolescence 17:117133

441 45. Pine DS, Cohen E, Cohen P, Brook J (1999) Adolescent depressive symptoms as predictors of adult depression: moodiness or mood disorder? Am J Psychiatry 156:133135 46. Sagatun A, Sogaard AJ, Bjertness E, Selmer R, Heyerdahl S (2007) The association between weekly hours of physical activity and mental health: a three-year follow-up study of 1516-yearold students in the city of Oslo, Norway. BMC Public Health 7:155 47. Sallis JF, Prochaska JJ, Taylor WC (2000) A review of correlates of physical activity of children and adolescents. Med Sci Sports Exerc 32:963975 48. Schmitz KH, Lytle LA, Phillips GA, Murray DM, Birnbaum AS, Kubik MY (2002) Psychosocial correlates of physical activity and sedentary leisure habits in young adolescents: the Teens Eating for Energy and Nutrition at School study. Prev Med 34:266278 49. Stathopoulou G, Powers M (2006) Exercise intervention for mental health: A quantitative and qualitative review. Clinical Psychology: Science and Practice 13:179193 50. Sund AM, Larsson B, Wichstrom L (2001) Depressive symptoms among young Norwegian adolescents as measured by the Mood and Feelings Questionnaire (MFQ). Eur Child Adolesc Psychiatry 10:222229 51. Sund AM, Larsson B, Wichstrom L (2003) Psychosocial correlates of depressive symptoms among 1214-year-old Norwegian adolescents. J Child Psychol Psychiatry 44:588597 52. Sund AM, Wichstrom L (2002) Insecure attachment as a risk factor for future depressive symptoms in early adolescence. J Am Acad Child Adolesc Psychiatry 41:1478-1485. Erratum: Jam Acad Child Adolesc Psychiatry. June 2003: 42:744 53. Tao FB, Xu ML, Kim SD, Sun Y, Su PY, Huang K (2007) Physical activity might not be the protective factor for health risk behaviours and psychopathological symptoms in adolescents. J Paediatr Child Health 43:762767 54. Tsang HW, Fung KM (2008) A review on neurobiological and psychological mechanisms underlying the anti-depressive effect of qigong exercise. J Health Psychol 13:857863 55. Ussher MH, Owen CG, Cook DG, Whincup PH (2007) The relationship between physical activity, sedentary behaviour and psychological wellbeing among adolescents. Soc Psychiatry Psychiatr Epidemiol 42:851856 56. Van Ittalie TB (1985) Health implications of overweight and obesity in the United States. Ann.Intern.Med. 103:983988 57. WHO (2001) The world health report 2001- Mental Health: New Understanding. New Hope, In, Geneva 58. Wichstrom L (1995) Who? What? Where? A national survey on youth and sport participation. In:Ungforsk. The Norwegian research Council, Oslo 59. Wiles NJ, Jones GT, Haase AM, Lawlor DA, Macfarlane GJ, Lewis G (2008) Physical activity and emotional problems amongst adolescents : A longitudinal study. Soc Psychiatry Psychiatr Epidemiol 43:765772 60. Wold B, L H (1998) Social and Environmental factors associated with physical activity in young people. In: Biddle S, Sallis J, N C (eds) Young and active?, London, p 119-132

123