Relationship between chronotype and mental behavioural health among adolescents: a cross-sectional study based on the social ecological system

This was a cross-sectional study using an extensive survey procedure to assess the sociodemographic and psychosocial environmental factors that contribute to health and HRBs of the general population [25, 26]. The study was designed and reported according to the Strengthening the Reporting of Observational Studies in Epidemiology checklist.

We used data from a collaborative cross-province survey of adolescent health and well-being that examine health and well-being of adolescents in their social context every 2 years in three-to-four cites in China [25, 26]. Data were collected in all participating cities through self-reported school-based surveys using a standard methodology described in detail in the study protocol [25, 26]. A cross-sectional study design was used to conduct nationwide sample surveys across China. The participants were high-school and junior high-school students in China, who were surveyed between October, 2020 and June, 2021 using multistage stratified cluster random sampling. The design and data collection procedures were approved by the Ethics Committee of the Anhui Medical University (reference no: 20200965). Written informed consent was obtained from the parents or guardians of all students. All methods used in this study were in accordance with the relevant guidelines and regulations. Detailed methods are shown in the Supplementary Methods (page 1).

The inclusion criteria were as follows: 1) informed consent of the participant and his or her guardian was obtained; 2) the participant was a junior high or hig- school student (aged 12–20 years); 3) the participant had no history of mental illness; and 4) the participant was a student attending a chosen school.

The exclusion criteria were as follows: 1) failure to obtain informed consent for participation in the study; 2) participant was a college student; 3) failure to submit the questionnaire; and 4) the participant had a congenital or acquired immunodeficiency. The flow chart of participant recruitment in this study is shown in Fig. 1.

Age, sex, parental education level, residential area, whether the participant was an only child, family economic status, number of friends, and academic record were included as covariates.

Results for categorical variables were described using frequency and percentage and analyzed using T-test (dichotomous) and Chi-square test (multivariate). Continuous variables were described using mean and standard deviation and analyzed using one-way analysis of variance. In order to identify clusters of HRBs, latent class analysis (LCA) was used to identify homogeneous, mutually exclusive "patterns" of 15 HRBs using Mplus 7.4, according to the method described in a previous study [25]. Multivariable logistic regression was used to examin the relationships between SERFs, chronotype and mental health, HRBs and presented as adjusted odds ratios (aORs) with 95% confidence intervals (CIs). Mediation moderate analysis was used according to the PROCESS method to explore the relationship between SERFs, chronotype, mental health and HRBs. All analyses were conducted using the Statistical Package for the Social Sciences (SPSS) software, version 23.0. After preliminary data sorting, missing data was processed in SPSS 23.0. In general, the data missing rate of this study was very low, and the missing rate of each item was less than 1%. Therefore, the multiple imputation method was adopted in SPSS 23.0 to process missing data at the project level. Our sensitivity analysis method is shown in the Supplement Methods.

Supplement Table 2 (page 7) depicts the fit statistics for the one- to nine-class models. The three-class solution was chosen as the final, best-fitting model based on its lower AIC (akaike information criterion), as well as lowest BIC (bayesian information criterion) and aBIC (bayesian information criterion) values. Moreover, we listed the trend of aBIC values and found that the three-class solution had the most significant decrease in aBIC values. With regard to model 3, bootstrap validation procedures also demonstrated a good fit (p < 0.001). All remaining results were reported specifically to the three-class solution.

Figure 2 shows the three-class model of HRBs and item-response probabilities for the 15 HRBs for each class. Class 1 was characterized by a high probability of exposure to each of the 15 HRBs; therefore, we labeled this latent class “suicide and NSSI” (21.3%). Class 2 comprised individuals with medium-level HRBs and was labeled as “fast food, SSB, and takeout (dietary behaviors)” (27.9%). Class 3 was characterized by a low probability of exposure to each of the 15 HRBs and was labeled as “low risk” (50.8%).

In total, 17,800 junior and senior middle school students aged 10 to 20 years were recruited in this study using a health survey of adolescents in junior and senior middle schools (grades 7–12). We checked the quality of all questionnaires and excluded those with regular answers, contradictory answers (obvious logical errors), or more than five consecutive unanswered questions. After excluding 947 (5.32%) questionnaires, 16,853 questionnaires were included in the analysis, and the efficiency rate was 94.68%. The mean participant age was 15.33 ± 1.08 years.

Sample characteristics for covariates, SERFs, and chronotypes are described in Table 1 for each of the three HRB groups. Our results showed that higher total SERF and eveningness were associated with higher suicide and NSSI scores (total SERF: χ² = 1,737.84 and eveningness: χ² = 751.52). Girls were more likely to have higher suicide and NSSI scores (χ² = 86.54) than boys. Lower parental education levels were also found to be correlated with higher suicide and NSSI scores (father: χ² = 201.65 and mother: χ² = 203.95).

Sample characteristics for covariates (sex, age, parental education level, family residence, only-child status, family economic status, number of friends, and academic performance), SERFs, and chronotypes are described in Supplement Table 3 (page 7) for each of the two mental health groups. Higher SERFs and eveningness were associated with higher negative mental health (total SERF: χ² = 275.79 and eveningness: χ² = 365.29). Boys were more likely to have mental health issues (χ² = 26.41) than girls.

As shown in Table 2, a high SERF level, intermediate chronotype, and eveningness were associated with higher suicide and NSSI scores (total SERF OR = 8.80, 95% CI: 7.81, 9.90; eveningness OR = 5.21, 95% CI: 4.56, 5.95; and intermediate chronotype OR = 1.75, 95% CI: 1.57, 1.96). After adjusting for covariates, these results were identical (all p-values < 0.01).

As shown in Supplement Table 4 (pages 10–11), higher levels of SERF, intermediate chronotype, and eveningness were associated with higher cooccurrence of HRBs (total SERF OR = 7.36, 95% CI: 4.50, 12.05; eveningness OR = 12.34, 95% CI: 5.91, 25.77; and intermediate chronotype OR = 4.36, 95% CI: 3.08, 6.15). After adjusting for covariates, these results were identical (all p-values < 0.01).

As shown in Supplement Table 5 (page 12), we also explored the interactions between chronotype, SERF, and clustering of HRBs and mental health. Higher SERF and eveningness were associated with higher suicide and NSSI scores (OR = 27.84, 95% CI: 22.03, 35.19) and a greater number of mental health issues (OR = 18.46, 95% CI: 13.16, 25.88).

We performed moderated mediation analyses by controlling the covariates noted previously. The results demonstrated that the effect of SERFs on the clustering of HRBs was significantly moderated by chronotype. In particular, the mediation of mental health on HRBs was significantly moderated by chronotype (Supplement Table 6, page 12). Furthermore, morningness predicted more severe mental health issues. However, while morningness enhanced the effect of mental health, it weakened the effect of SERFs (Supplement Table 7 and Supplement Fig. 1, page 13).

We also performed moderated mediation analyses by controlling the covariates noted previously for HRBs cooccurrence. The results demonstrated that the effect of SERFs on mental health was significantly moderated by chronotype (SERF*chronotype: β = 0.071, p < 0.01). Specifically, the mediation of mental health on HRBs was significantly moderated by chronotype (mental health*chronotype: β = ˗0.001, p < 0.01; Supplement Table 8, page 13). Similar results were shown in supplement Table 9 and supplement Fig. 2 (pages 14–15).

This study examined the correlation between chronotypes and SERFs and their respective associations with mental health and HRBs in a school-based sample of adolescents. This study used LCA to deduce adolescent HRB patterns from pre-specified behavioral variables. Multiple indicators of HRBs were examined using the LCA method within a theoretically derived model of SES influence on class membership. In our study, the prevalence of clustering of HRBs was 21.3%. Our study also explored the relationship in different dimensions of SERFs, including spatial and chronosystem. Moreover, an interaction between chronotypes and SERFs and HRBs and mental health was observed in this study. We examined the mediating effect of mental health and the moderating role of the chronotype on the relationship between HRBs and SERFs. Our results strongly supported the hypotheses regarding different factors and potential mechanisms associated with adolescent HRBs.

SERFs refer to risk factors and events that are not conducive to individual development and adaptation when an individual's living environment includes a variety of backgrounds, and these factors and events may increase the risk of negative outcomes [30]. Although these frameworks may vary, most of them reflect the principles of the social ecological model, a simple but far-reaching framework for public health [31, 32]. Based on the research on SERFs, some studies have reported that the high social-ecological risk group of adolescents is at higher risk of psychological, emotional, and behavioral problems [23, 33]. Individuals are often faced with multiple risks at the same time, if only focus on a single area is inconsistent with the reality of life. In other words, studies on the risk factors that lead to adolescent health risk behaviors are mostly explored from a single or minority level. However, adolescents will be affected by multiple risks such as family, community, school and peers during their growth, and their psychological problems are rarely caused by a single risk factor. Sometimes different risk factors do not act independently. More often, multiple risk factors at the same time [33, 34]. Our results demonstrated that cumulative SERFs were positively correlated with clustering of HRBs and HRB cooccurrence, consistent with ecological system theory. The concept of cumulative risk has also gained support across social and cognitive domains following the pioneering research on ecological risk [13]. Cumulative risk models suggest that it is better to predict individual outcomes by exploring the accumulation of risk factors than by focusing on adverse outcomes of a single indicator.

Our study further exaimed the association between chronotypes, SERFs, mental health and HRBs. First, eveningness was found to be associated with higher HRBs and higher rates of mental health issues, which is consistent with previous findings in college students [35] and applies these to adolescents in this study. Second, previous studies found that certain social factors (such as parental or peer behavior) and individual factors (such as low self-esteem) were also associated with higher HRBs [36]; in addition, Williamson et al. found cumulative socio-ecological risk indexes were most associated with increased middle childhood sleep problems, this study systematically evaluated the interaction of socio-ecological risks, sleep problems, and individual adolescent factors, as well as the direct and indirect effects of maternal and paternal risk factors [37]. We performed a latent class analysis of HRBs based on the previous literature [25]. Our analysis was informed by hypotheses derived from the social development model between SERFs and HRBs [38]. Moreover, our results further provided a theoretical basis for the further study of moderated mediation analyses and their role in the significance of mental health, SERFs, chronotypes and HRBs. Third, there was an interaction between higher-risk SERFs, eveningness and higher HRBs, negative mental health symptoms. This means that adolescents who experience high-risk psychosocial environments where life and rest are not in harmony, such as circadian rhythm disturbances, may have a higher incidence of physical and mental health problems [14]. These results are similar to those of previous studies, the study of social ecological psychology aims to explore how specific features of social ecology lead to a state of mind, which can also affect people's emotions or behaviors [39]. Possible mechanisms underlying this have been explained by the diathesis-stress model [40] or comprehensive social-ecological diathesis-stress model [41].

The interaction and correlation results also suggested that a positive social environment can counteract the physical and mental health effects caused by an individual’s (potentially genetically) disturbed circadian rhythm [42]. Our study addressed this question from the perspective of chronotype, a trait related to circadian rhythms [43]. Wills et al. developed the moderator model to explore the relationship between chronotype score (an independent variable), depression (outcome), and social support (the moderator and mediator) and found that chronotype was associated with depression in individuals with low social support from friends and teammates [44]. This result further verifies that there is an association between SERFs, chronotype and HRBs. We also explored the moderating effects of chronotype by dividing the relationship between mental health, chronotype and HRBs in three aspects from a social ecological system perspective. In our study, we found that high SERF scores were associated with higher HRBs and mental health issues. Possible reasons for this may be that evening chronotypes may be predisposed to mental health disturbances, potentially through a behavioral inhibition pathway [39]. Similarly, adolescents who experienced higher SERFs combined with eveningess, experienced more negative mental health effects and higher levels of HRBs. Eveningness may influence the HRB problems caused by SERFs. Moreover, eveningness may be a part of a broader temperamental [45] and personality [46] profile of adolescents that may predispose to mental health problems. Future research should focus on consideration of individual differences in temperament or personality.

Based on the mediate moderation analysis, it has been found that an individual's chronotype can interfere with the timing of working hours, school, or social schedules. One study discussed the association between adverse social events, such as adverse childhood experiences, chronotypes, and HRBs, such as smoking and alcohol consumption [47]. Potential SERFs and the resulting HRBs that affect mental health may be blocked or inhibited by morning type effects, that is, reasonable sleep habits can offset the adverse psychological outcomes brought by multi-level environment and unfavorable HRBs. These results confirm our previous findings exploring the mediating role of chronotype in mental health and the moderating role of chronotype in our use of multiple social contexts to measure the actual social reality of adolescents and their circadian habits (such as eveningness). This study also provided some new perspectives other than just life events to further investigate the association between circadian rhythm and SERFs, HRBs and mental health. A previous study indicated that the relationship between chronotype and experiences of negative life events may be mediated either by personality traits (neuroticism, risk-taking behavior, and externalizing behavior) or poor sleep quality [45, 48]. This concept offered a theoretical basis for our research, Through social ecological models, the description of mental and behavioral health can be understood from the perspective of holism, in which the individual is understood to be connected to historical and cultural backgrounds, family, community, land and even future generations in a highly relevant context [49]. Not only SERFs have an important effect on adolescent HRBs, but this association may be mediated by mental health, mediated by chronotype. Therefore, according to the results of this study, we should advocate ameliorating adolescents' living habits in the evening as much as possible, and provide them with more support at home, at school or in the community.

The strengths of this study included the followings: this was multicenter and multilevel design and the large number of study participants. Based on the previous concept of chronosystem, this study innovatively proposed factors that may affect adolescent HRBs, supplemented the original concept of ecological theory, and provided theoretical and practical basis for the subsequent expansion of the application of chronosystem.

However, some limitations should be noted. Because of the cross-sectional design, a casual association between SERFs and HRBs could not be established. Future longitudinal studies are needed to elucidate the association between the variables. Fifteen types of adolescent HRBs were examined in this study, although these did not include addiction, intentional injury, and unsafe behaviors, which are variables that have been included in other studies. However, we will consider these behaviors in a future study involving college students. All of the data used in this study were obtained from self-reported questionnaires; therefore, there were issues with subjectivity, validity, and reliability of the answers. This study only analyzed results from three cities; therefore, it is not clear how representative this sample is of the overall population in China. A follow-up survey will be conducted on population samples from different regions and cultures across China. In this study, the cumulative risk index was adopted to explore SERFs, which limited the universality of this study. Although this study focused on the association between SERFs and HRBs in certain fields, the possible application of this correlation in other fields still needs further investigation. Future research should explore the association between SERFs and HRBs in different fields.