Are Physical Activity, Screen Time, and Mental Health Related During Childhood, Preadolescence, and Adolescence?

11-Year Results From the German Motorik-Modul Longitudinal Study

Claudio R. Nigg; Kathrin Wunsch; Carina Nigg; Claudia Niessner; Darko Jekauc; Steffen C. E. Schmidt; Alexander Woll


Am J Epidemiol. 2021;190(2):220-229. 

In This Article



The study protocol was approved by the Ethics Committee of the Karlsruhe Institute of Technology (Karlsruhe, Germany). Participants volunteered and provided informed assent and (parental) consent. Data were derived from both the Motorik-Modul (MoMo) Longitudinal Study,[28,29] an in-depth study of the German Health Interview and Examination Survey for Children and Adolescents (KiGGS), and KiGGS itself.[30–32] Baseline (time 1 (T1)) data were gathered between 2003 and 2006, wave 1 (time 2 (T2)) data between 2009 and 2012, and wave 2 (time 3 (T3)) data between 2014 and 2017—corresponding to a span of 11 years for each participant.


Persons aged 18 years or younger at wave 2 (T3) who participated in all 3 measurement phases of MoMo and KiGGS were included in analyses, since MH was only assessed up to 18 years of age. Data from children below 11 years of age were obtained from parents. Thus, a sample of 686 participants (55.2% female) was used for this study. The mean age at baseline (T1) was 5.57 (standard deviation (SD), 1.00) years; the mean age at wave 1 (T2) was 11.85 (SD, 1.03) years; and the mean age at wave 2 (T3) was 16.86 (SD, 1.04) years.

Regarding participants' socioeconomic backgrounds, 12.6% of mothers (23.3% of fathers) had a lower secondary-level education ("Hauptschulabschluss"), 36.4% of mothers (22.8% of fathers) had a "Realschulabschluss"-level education (equivalent to a high school diploma), and 26.7% of mothers (25.4% of fathers) had an "Abitur"-level education (the highest level of German secondary educational achievement).

Participants in the full longitudinal cohort study (n = 1,407) and the participants analyzed in our study did not differ with regard to variables that are indicative of representativeness (sex, region, migration background, and educational level; all P values > 0.05).


All data were derived via self-report questionnaires, except for children under 11 years of age, for whom parent proxy reports were collected. At T1, demographic information on the young person's age, sex, and socioeconomic status (which included parents' education, job qualification and level, and monthly household income, according to the method of Lampert et al.[33]) was assessed. PA, ST, and MH were measured at all 3 time points.

Physical Activity. PA was measured using a single item derived from the MoMo activity questionnaire, asking about the frequency of days with at least 60 minutes of moderate-to-vigorous PA during an ordinary week (ranging from 0–7). Acceptable validity with the use of an accelerometer and 1-week test-retest reliability have been reported.[34]

Screen Time. Because recent studies have shown that self-reported TV viewing time is a reliable and valid indicator of sedentary behavior[35] and that time spent on computers or the Internet has dramatically increased in recent years,[14] participants were asked to report the amount of time they spent watching TV or videos per day (hereafter called "TV/video time") and the amount of time they spent using the computer or surfing the Internet per day (hereafter called "PC/Internet time") at T1 and T2. For T3, the latter item was changed slightly to the amount of daily computer and Internet use outside of computer games.[36]

Self-reported MH. To examine MH status, the German version of the SDQ was used, measuring self-reported MH on 5 subscales: emotional symptoms (Cronbach's α = 0.54), conduct problems (α = 0.49), hyperactivity/inattention (α = 0.77), peer relationship problems (α = 0.59), and prosocial behavior (α = 0.66). Overall SDQ score was also calculated (α = 0.75). All subscales consisted of 5 items, and the overall SDQ scale consisted of 20 items (excluding the prosocial behavior scale); scores for each subscale and the overall SDQ were summed.[37] Response options included "0: does not apply," "1: applies partly," and "2: does apply." A higher score indicates lower MH, except for the prosocial behavior scale, where a higher score indicates higher MH. Validity with regard to other measures and methods has been reported.[37]

Data Analysis

Because of the observed sex differences, we conducted analyses for females and males separately. Path panel prediction models using structural equation modeling software (AMOS; SPSS Inc., Chicago, Illinois) were set up with T1, T2, and T3 indicators of PA, ST, and MH. Missing data were accounted for through the full information maximum likelihood estimation method used in AMOS. Model fit was evaluated with the comparative fit index and the root mean square error of approximation. Because of overparameterization (not enough degrees of freedom to estimate the model), none of the full models converged. Hence, study variable correlations were calculated for each sex. The significant correlations from between the time points were then modeled.