The buffering effect of physical activity on adolescent psychological stress: a cross-sectional survey and a longitudinal follow-up of Chinese adolescents | BMC Public Health

Sampling method and survey subjects

From March to April 2022, in the three economic belts in eastern, western, and central China, one city with a high GDP level, one city with a medium GDP level, and one city with low GPD level were selected from each economic belt (Eastern: Nantong, Jiangsu Province, Lishui, Zhejiang Province, Fuxin, Liaoning Province; Western: Chongqing, Yibin, Sichuan Province, Yili, Xinjiang Uygur Autonomous Region; Central: Taiyuan, Shanxi Province, Puyang, Henan Province, Xiangxi Tujia and Miao Autonomous Prefecture, Hunan Province). Among the 9 cities, 1 junior high school and 1 senior high school were selected, and stratified random cluster sampling was adopted to entrust local teachers to distribute 100 paper questionnaires to each grade, totaling 5,400, 5,216 questionnaires were recovered, 221 invalid questionnaires were excluded, and 4,995 valid questionnaires were obtained, with an effective recovery rate of 95.7%. The subjects included 2,509 boys and 2,486 girls, and no gender imbalance. 826 students in Grade 1 of junior high school, 913 students in Grade 2 of junior high school, 681 students in Grade 3 of junior high school, 876 students in Grade 1 of senior high school, 847 students in Grade 2 of senior high school, 852 students in Grade 3 of senior high school, with no significant distribution difference among different grades (X² = 9.033, p = 0.108).

After 8 weeks, the same subjects were surveyed again, among which 128 were not contacted and 73 were refused to investigate for some reason. Therefore, 4,794 questionnaires were distributed, 4,613 questionnaires were returned, 287 invalid questionnaires were excluded, and 4,326 valid questionnaires in the end, with an effective recovery rate of 93.8%. Among them, there were 2,158 boys, and 2,168 girls, with a balanced female ratio. 750 students in Grade 1 of junior high school, 870 students in Grade 2 of junior high school, 522 students in Grade 3 of junior high school, 780 students in Grade 1 of senior high school, 758 students in Grade 2 of senior high school, 646 students in Grade 3 of senior high school, with no significant distribution difference among different grades (X² = 9.253, p = 0.099).

Survey tools

Adolescent Self-rating Life Events Checklist (ASLEC)

Adolescent Self-rating Life Events Checklist (ASLEC) revised by Xin Xiuhong et al. [32] was used to assess the recent stress events in adolescents. There are 26 items on the checklist, including 5 dimensions (4 interpersonal stress items, 4 learning stress items, 7 punishment items, 6 loss items, and 5 adaptation items). The checklist adopts a 6-point scoring method, “0” means “never happened”, “1” means “happened but had little impact”, and “5” means “a great impact”. The higher the score is, the more seriously affected by life events. In this study, the checklist’s Cronbach’α = 0.887 (T1).

Calgary Symptoms of Stress Inventory (C-SOSI)

Calgary Symptoms of Stress Inventory (C-SOSI) compiled by Carlson [33] was used to measure adolescents’ stress symptoms. Two psychology graduate students were invited to translate the questionnaire, and then a native English student was invited to translate the Chinese questionnaire back, afterwards, two psychology professors were invited to proofread and modify the Chinese and English questionnaires to form the Chinese version of the stress symptoms questionnaire. The questionnaire includes 56 items, contains eight dimensions (8 depression items, 7 anger items, 9 cross-nerve arousal items, 6 nerve items, 8 muscle tension items, 6 cardiopulmonary items, 6 cognitive confusion items, 6 upper respiratory tract items), using a 5-point scoring method, “0” indicates “never”, “4” indicates “very frequent”, the higher the score, the more serious stress symptoms. In this study, the questionnaire’s Cronbach’α = 0.934 (T1) and 0.963 (T2). A confirmatory factor analysis of the C-SOSI was performed, which showed that the results of the fitting of the model were x²/df = 1.483, CFI = 0.989, TLI = 0.985, GFI = 0.981, IFI = 0.989, RMSEA = 0.058, and SRMR = 0.016, which were in line with the criteria, indicating that the fitting of the model fit was good.

Amount of physical activity

Using the Physical Activity Level Scale [34] to measure adolescents’ physical activity level. The scale includes three items, namely the frequency of weekly physical activity (1–5), the duration of each time physical activity (0–4), usually the intensity of physical activity (1–5), the total score of 0–100, the higher the score, the greater the youth amount of physical activity. The scale’s retest reliability r = 0.82.

Statistical method

SPSS 25.0 was used to test the reliability and validity of the recovered data, Pearson Correlation was used to analyze the relationship between variables, Linear Regression was conducted to analyze the effect of stress events and amount of physical activity on stress symptoms of adolescents, and “Model 1” of Process 4.0 was used to test the moderating effect of amount of physical activity on the relationship between stress events and stress symptoms. AMOS 24.0 was used to perform cross-lagged regression analysis on the data of the two investigations (T1 and T2), and “p < 0.05” was taken as the standard of statistical significance in all the above operations.

Common method bias test

The data were tested for common method bias using the Harman single-factor test. The results show that 11 factors with an eigenvalue greater than 1 explained 45.024% of the total variance, and the first common factor explained the total variance of 24.78%, 40% below the critical value, indicating that there is no serious common method bias problem.