85,509 out of 96,218 (88.9%) participants completed the primary assessment. The baseline characteristics of participants engaged in various exercise types are summarized in Table 1. Among these, 79,011(92.4%) participants reported being involved in physical activity such as single anaerobic (16,376 [19.2%]), low-intensity aerobics (14,560 [17.0%]), medium-intensity aerobics (13,059 [15.3%]), two-player swing-type exercises (double swing) (20,000[23.4%]) or team ball games (15,016 [17.6%]). An additional 6,498 participants (7.6%) reported pursuing physical activities not specified in our list. Table S2 outlines exercise frequency, duration, and various mental health outcomes concerning different exercise types. For participants engaged in double swing exercises, the majority (10,346 [51.7%]) exercised 0–1 times a week for 0.5-1 hour (7,561 [37.8%]). Most low-intensity aerobic exercise participants (6,266 [43%]) exercised 0–1 times a week for 0.5-1 hour (5,953 [40.9%]). The majority of moderate-intensity aerobic exercise participants (5,153 [39.5%]) exercised 2–3 times a week, for 0.5-1 hour each time (5,396 [41.3%]). Most single anaerobic exercise participants (6,314 [38.6%]) exercised 2–3 times a week for 0.5-1 hour each time (6,559 [40.1%]). Regarding group ball games, most participants (6,321 [42.1%]) exercised 2–3 times a week, spending 1-1.5 hours on each session (4,417 [29.4%]). Finally, most participants who reported engaging in physical activities other than those mentioned exercised 0–1 times a week (3,151 [48.5%]), with an exercise duration of 0-0.5 hours per session (2,889 [44.5%]). To ensure that the results remained interpretable, we removed the 6,498 participants who chose “other” types of exercise from the final logistic regression. Therefore, our cohort for the regression analysis consisted of 79,011 participants (see Table 2&3).
Mental health outcomes varied as a function of exercise type when controlling for all sociodemographic variables (Table 2, appendix Tables 3–7). Relative to team ball sports, other types of sports significantly predicted negative mental health outcomes (Fig. 2A, all Odds Ratios > 1). Other types of exercise (except team ball games) more significantly predicted negative mental health outcomes in comparison to single-player anaerobic exercise (Fig. 2B, Odds Ratios > 1). Other types of sports (except team ball sports and single-player anaerobic sports) more significantly predicted negative mental health outcomes compared to two-player swing sports (Fig. 2C, Odds Ratios > 1). All other types of exercise (except low-intensity aerobic exercise) were significant protective factors for negative mental health outcomes when compared to moderate-intensity single-player aerobic exercise (Fig. 2D, Odds Ratios < 1). All other types of exercise were protective factors for negative mental health outcomes compared to low-intensity single-player aerobic exercise (Fig. 2E, Odds Ratios < 1). Hence, based on the protective importance for mental health, team ball sports > single anaerobic sports > double swing sports > single moderate-intensity aerobic sports > single low-intensity aerobic sports.
We found significant nonlinear associations between exercise frequency and mental health outcomes such as anxiety, depression, PTSD, suicidal ideation, and non-suicidal self-injury, controlling for sociodemographic variables. However, the frequency of low-intensity and medium-intensity aerobics were not significantly associated with suicidal ideation and non-suicidal self-injury as seen in Table 3 and Appendix Table 8. As presented in Fig. 2, individual's increases in team ball exercise were beneficially and monotonically associated with odds reductions in all mental health outcomes. Moreover, we observed U-shaped relationships between exercise frequency, anxiety, and depression for double swing, low, and medium-intensity aerobics exercises. Specifically, individuals with exercise frequency between three and four times a week had a lower probability of anxiety and depression than those who exercised fewer than three or more than four times, as illustrated in Fig. 2D & E, appendix S Figs. 1 & 2. Additionally, for double swing, medium-intensity aerobics, and single anaerobic exercise frequency, we found a U-shaped relationship with suicidal ideation. Individuals performing these exercises three to four times per week had a lower probability of suicidal ideation than those exercising fewer than three or more than four times, as seen in Fig. 2C and Appendix S Fig. 5.
Similarly, we observed significant nonlinear relationships between all mental health outcomes and exercise duration while adjusting for sociodemographic variables, as presented in Table 3 and Appendix Table 9. Figure 3 depicts that exercise for 150 minutes per session of team ball was associated with the lowest probability of all mental health outcomes. Additionally, exercise durations ranging from 90 to 120 minutes of single anaerobic, low-intensity aerobics, medium-intensity aerobics, and double swing exercises were associated with the lowest probability of anxiety and depression, with the peak at around 120 minutes, as shown in Fig. 3D & 3E and appendix S Fig. 7& 8. Furthermore, the exercise duration of 180 minutes or more was associated with the lowest probability of NSSI and SI for most exercise types, as presented in Fig. 3A & 3B and Appendix S Figs. 9 & 12. For double swing, single anaerobic, and medium-intensity aerobics, exercising for 180 minutes or more was associated with a decrease in the probability of PTSD, while low-intensity aerobics for 120 minutes was associated with the lowest probability of PTSD, as seen in Fig. 3F, and Appendix S Fig. 10. Finally, exercising for 60 to 90 minutes for single anaerobic, low-intensity aerobics, and double swing was associated with the lowest probability of SA, whereas exercising for 120 minutes of medium-intensity aerobics was linked with the lowest probability of SA, as depicted in Fig. 3C and Appendix S Fig. 11.
According to our sensitivity analysis, we limited the analytical sample to only children in their families (N = 37,668). We discovered similar results to the entire sample, which showed that after accounting for sociodemographic variables, mental health outcomes fluctuated based on exercise type. Team ball exercise offered the most protection, trailed by single anaerobic exercise, double swing exercise, single moderate-intensity aerobic exercise, and single low-intensity aerobic exercise (refer to Appendix S Fig. 13, and S Table 10). We also noticed significant nonlinear associations between exercise frequency and duration with mental health outcomes (refer to Appendix S Figs. 14 & 15, and S Table 11).