Introduction
Adolescents, in a heightened propensity stage for engaging in risk behaviors that can contribute to health risk behaviors (HRBs) aggregation or co-occurrence of multiple types of HRB [
1,
2], are susceptible and vulnerable to reckless and risky conduct [
3]. HRBs are defined as specific forms of conduct having potential threat to psychological or mental health in the present and future [
4], such as substance use, insufficient physical activity, poor dietary habits, unintentional or intentional injury, non-suicidal self-injury and so on, which can increase the susceptibility to some certain diseases or ill-health [
3,
5] and a series of psychological problems later [
6]. Globally, disability and death in adolescence are largely attributed to risky behaviours [
7]. Reportedly, 70% of premature deaths in adulthood are caused by HRBs occurring in adolescence (WHO, 2009) [
8]. Decreasing the prevalence of adolescents’ risky conduct has become a priority worldwide [
9].
Summarized data from the national Youth Risk Behavior Survey in America showed a marked increase in the prevalence of substance use (from 0.8% to 30.1%), the behaviors associated with an unhealthy diet (from 3.1% to 18.0%), inadequate physical activity (from 30.3% to 52.9%), long screen time (from 19.2% to 32.2), and suicide (from 0.7% to 12.5%) [
10]. A longitudinal birth cohort targeting adolescents aged 15–16 years in the UK, indicated that most adolescents engaged in physical inactivity (74%), and hazardous alcohol use (34%) [
11]. In early adolescence, the cumulative incidence rate of smoking and drinking each exceeded 21%, and separately more than 65% by the end of adolescence in urban South Africa [
12]. Investigation data based on 1 763 high school students aged 11–19 years from Ghana found 75.4% and 18.6% of students respectively reported insufficient physical activity and substance use [
13]. A cross-sectional survey among Chinese middle school students in six cities reported the prevalence of alcohol use (AU) and excessive screen time (ST) were both over 16%, nearly a third of students involved in non-suicidal self-injury (NSSI), and 15% of students had suicidal behavior (SB) [
14]. These survey data suggest that HRBs in adolescence are at a highly prevalent level and have become a major public concern in the health field worldwide.
Multiple risk behaviors commonly co-occur simultaneously with the way of clustering [
15]. In early literature on behavior clustering among adolescents, for example, MacArthur, et al., 30% of adolescents participated in three to five risky behaviors simultaneously, and 6.2% of adolescents with AU reported seven or more forms of HRB [
11]. Most of the students in Vietnam from cross-sectional research had a cluster of at least two HRBs and approximately half with three HRBs [
16]. Survey data from our research group, included 22628 middle students in China whose multiple behaviors were divided into four categories based on the latent class model [
14].
Additionally, the occurrence of risk behaviors, characterized by covariation among adolescents, can lead to a “risk behavior syndrome” which indicated engaging in one problem conduct increases the occurrence of other forms of problem behavior [
17]. For example, marijuana use obviously increased the behaviors related to suicide (suicidal ideation, suicidal plan, and repeated attempted suicide) [
18]. Adolescents with alcohol use had a higher risk of substance abuse and sexual risk behavior [
11]. Given the clustered and synergistic nature of adolescent risk behavior, clustering models that focus on multiple behaviors to classify them may play an effective role in promoting healthy adolescent development. The covariation trend of adolescent risk behavior, is ascribed to the shared social-ecological environment in which socially organized opportunities and normative expectations could be offered to learn and involve in different forms of risk behaviors [
19].
Different scholars, from different perspectives, have proposed different theoretical models of risky behaviors in adolescents. For example, the problem behavior theory indicated the interaction of the social environment system, individual characteristics system, and behavior system, which jointly act on adolescents’ behavioral development and lead to unhealthy outcomes [
17]. In the theory of planned behavior (TPB), attitude, subjective norm, and perceived behavioral control were thought to be determinant factors of intent that were regarded as a major determinant of behavior [
20]. Viewed from double jeopardy, multiple socioeconomic stressors experienced by adolescents may be harmful [
21,
22], and resulted in the occurrence of problem behaviors [
23,
24]. The theory of compensatory cognition, based on compensatory health faith, pointed out that certain unhealthy behaviors could be compensated by engaging in other behaviors [
25]. Based on the perspective of social ecology, an individual was nested in a series of interacting environment systems (individual, family, school, community, policy, culture, time) and whose conduct was constrained by multiple factors which affect jointly the development of adolescents [
26]. Risk factors from social ecology show a strong co-occurrence. Exploring risk factors only from a single dimension could not reveal the real environment which may bring about excessive inference [
27]. Additionally, an individual tends to be more sensitive to more risk factors which contributes a greatly negative impact on the development and adaptability in adolescence [
28]. Therefore, on the basis of the ecological systems theory, to examine the risk factors for adolescent-initiated HRBs within micro- and meso- (natural environment support from individual, family, and school, and interpersonal environment support from school), exo- (community environment and safety), macro- (policy and culture), and chronosystem (change in family structure), is probably more in line with adolescents in the true adversity. In addition, there is a need to explore the social-ecological risks from multiple dimensions and HRBs.
Previous researches, mainly concentrated on a single or few dimensions of risk factors to explore the association between social ecological risk and behavior [
29‐
31]. The influences of risk and protective factors at different levels (e.g., family, school, company, community) on the multiple behavioral problems have been confirmed [
32,
33]. Concerning the co-occurrence of risk factors, based on previous literature, the current study started from a cumulative risk perspective to capture exposure to risk factors and reveal real adversity the in adolescents. We aimed to assess the association between cumulative social-ecological risk from seven dimensions and multiple forms of risky behavior. In addition, the present study measures the latent class clustering of HRBs based on the properties of the aggregate of multiple forms of HRBs. We further examine the relationship between cumulative social-ecological risk and latent category clustering of HRBs, and gender difference.
Methods
Participants and study design
From October to December 2021, based on multistage cluster sampling, we conducted a sample investigation at eight first-tier and second-tier cities (Shenzhen, Chongqing, Zhengzhou, Shenyang, Kunming, Xuzhou, Taiyuan and Nanchang) in China that had good cooperation intention with our research team. In each selected city, two schools including one junior and senior high school apart from village and city were taken into consideration. In each school, more than 200 students from each grade were chosen. A total of 22 868 adolescents aged 14.64 were recruited to carry out questionnaire survey.
The data collection and design in present study were approved by the Ethics Committee of Anhui Medical University (Ethical NO.20200965). All participants and their parents or guardian signed the informed consent for inclusion before the survey administration.
Measurement of sociodemographic characteristics, social-ecological risk factors and HRBs
Social-demographic characteristics
Demographic variables containing age, grade, gender (boys/girls), registered residence (rural/urban), paternal/maternal education level (primary school degree or less, junior high or senior high school or more) and self-reported family economy (worse, general or better), regional economic level (first-tier/second-tier cities).
Social-ecological risk factors
The exposure of social-ecological risk was assessed with the simplified rating questionnaire which simplified by the Social Ecological Risk Assessment Questionnaire of adolescents involving seven dimensions (individual, family, school, community, culture, policy and time dimension) [
34]. The risk factors from individual dimension, were respectively measured by 3-item, e.g., “You are confident to effectively deal with any emergency”. Family risk factors in the past six-month were assessed with 3 items, e.g., “Your parent often blame you owing to something”. The exposure evaluation of school and community risk were separate performed by 4 items, e.g., “You can fit in school life well”, “You live near the road including where the noise disturbs your rest”. Policy factors were measured by 5-item, e.g., “Penalties are laid down used to administrate the behavior of damaging the public property in your school”. The questionnaire comprised 3-item respectively related to time and culture, e.g., “Your parents divorced in the past three years”, “You would buy anything expensive as long as you like.” The answer to each item consisted of five choices, 1 = “completely consistent”, 2 = “basically consistent”, 3 = “somewhat consistent”, 4 = “basically inconsistent”, 5 = “completely inconsistent”. Score of each item was summed to create a total score. The total score, ranged from 25 to 75, had a good internal consistency with a significant Cronbach’s α coefficient of 0.79 in the present research, and was dichotomized at the 50th percentile to evaluate the exposure level of social-ecological risk. As a sensitivity analysis, the social-ecological risk was also dichotomized at the 33th percentile to determine variation in the results as a consequence of binary decision.
Health risk behavior
Ten types of HRBs including breakfast intaking not daily, AU, smoking, insufficient physical exercise, prolonged ST in weekday or weekend, suicidal ideation, suicidal plan, suicidal attempt, and NSSI were estimated. Breakfast intaking was evaluated by asking “During the past a week, how many days did you eat breakfast?” with eight response options from 1 = 0 day to 8 = 7 days. For the item, choosing 8 was coded as no (everyday) and other options were coded as yes (not daily). AU and smoking were assessed respectively by asking “During the past a moth, how many days did you drink at least one glass of wine?”, and “During the past a moth, how many days did you smoke cigarettes?” with response options of 1 = 0 day, 2 = 1-2 days,3 = 3–5 days, 4 = 6-9 days, 5 = 10 = 19 days, 6 = 20–29 days and 7 = 30 days”, choosing 1 was separate coded as no and other option were separately coded as yes. The validity of adolescent self-reported data related to breakfast intaking, AU and smoking has been evaluated [
19,
35]. Inadequate physical exercise, was assessed by asking “During the past a month, how many days did you engage in exercise, such as sit-ups, or weigh-lifting?” with the responses ranging from 0 to 7. Inadequate physical exercise was defined as less than three-day per week, with assessed validity [
36]. ST on weekday or weekend was respectively measured by asking “During the past 30 days, how much time did you spend watching video, computer games or engaging in something not related to studying with electronics (such as TV, smart phone, iPad).” with responses from 1 = little to 6 = more than four hours. Two hours or more were defined as prolonged ST in weekday or weekend [
37]. Suicidal ideation, suicidal plan and suicidal attempt was separately defined as yes in response to three questions “During the past a year, had you ever considered killing yourself?”, “During the past a year, had you ever made plans of suicide?”, “During the past a year, had you ever engaged in suicidal behavior?” NSSI, assessed with 12-item [
38], e.g., “During the past 12 months, have you ever intentionally bitten yourself?”, was dichotomized (frequency of any NSSI of 0 versus at least 1 as no or yes). In the current study, the internal consistency of NSSI and all ten risk behaviors was respectively 0.83 and 0.77.
Statistical analysis
All statistical analyses in the present study were conducted with SPSS version 23.0 (SPSS, Chicago, IL, USA) and Mplus version 7.0. Sociodemographic data, social-ecological risk factors and HRBs were separately describe in the total sample, girls and boys. Gender differences were compared with Chi-square test for categorical variables and ANOVA analyze for continuous variables. The categories of clustered behaviors were identified with the latent class analysis. Logistic regression analysis was used to examine the relationships between the level of social-ecological risk and different HRBs or clustered HRBs categories. In the adjustment model, we adjusted for age, grade, registered residence, parents’ educational level, self-reported family economy to examine the correlations of social ecological risk and HRBs. Furthermore, we also dichotomized the total score of social ecological risk at 33th percentile as a sensitivity analysis.
Gender differences in the relationships were examined with ratio of two odds ratios (ROR). All p-values were two-side, and p-value < 0.05 indicated significant statistical significance.
Discussions
In this school-based, large-scale study, the relationship between cumulative ecological risk and multiple forms of risky behavior and clustering of HRBs is investigated. The present survey revealed engaging in breakfast intaking not daily, AU, smoking, physical inactivity, prolonged ST on weekday or weekend, SB, and NSSI was each related to the level of cumulative ecological risk among the total sample, boys and girls, and a stronger association was showed in girls than boys, with the exception of breakfast intaking behavior. In addition, adolescents who encounter high levels of cumulative ecological risk are more likely to engage in cluster HRBs, and girls are more susceptible to experiencing cluster HRBs than boys.
The survey found that there was clustering of breakfast intaking not daily, AU, smoking, physical inactivity, prolonged ST on weekday or weekend, SB, and NSSI among adolescents, and two significant latent classes of HRBs were identified. The prominent disparities between two class groups were unveiled, particularly in areas such as breakfast consumption, exercise patterns, suicidal attempts, and NSSI. In the sample, the high-risk behavior group accounted for 22.2%, in which 59.3% adolescents had unhealthy breakfast consumption, 65.2% adolescents engaged in inadequate physical exercise, 25.6% adolescents had suicidal attempt and 61.7% adolescents engaged in NSSI. The incidence of various HRBs was respectively higher than the low-risk behavior group. We proposed that unhealthy breakfast consumption, insufficient physical activity, and SB may occur simultaneously, but data to elucidate this were not collected unfortunately, that need to be further explored with more research.
Influences, within the family, school, and company environment, as a part of joint efforts to account for the underlying social needs of healthy development among adolescents [
42,
43], have been investigated widely in plenty of exiting studies. For example, A school-based investigation of 2 999 students in British Columbia indicated that good school connectedness reduced the prevalence of excessive drinking, and smoking [
29]. Community factors, such as neighborhood (the social network, neighborhood disadvantage) [
44], culture, and norms [
30], greatly influenced the reckless behavior development related to adolescents. The current study extends the earlier literature by examining the impact of cumulative ecological risk from seven dimensions on multiple forms of risky behavior and clustered HRBs, and by addressing the gender differences in the association. Similar to the result of an early survey, we found a positive relationship between social-ecological risk factors clustering and multiple behaviors [
45].
Significant gender differences were observed in the relationship between cumulative ecological risk and multiple behaviors. Data indicated girls were more likely to have suicidal ideation and plan than boys which was similar to the other surveys with the exception of suicidal attempts [
46,
47], which might attribute to the phenomenon that girls tend to accumulate negative emotions and internalize problems due to the lack of emotional management and attention to the harmonious interpersonal relationship when faced with adverse conditions [
48]. However, the stronger association between cumulative ecological risk and other externalizing behavior problems and clustering HRBs were more obvious among girls than boys. Sociocultural theory suggests that the differences in the social division of labour lead to differences in the development of sex in individuals. Traditional social divisions of labor and gender stereotypes tend to assume that males are more powerful, which requires males to undertake more challenging tasks. Compared with males, females are often thought to be more susceptible to the environment and have poorer adaption to adversity which resulted in the occurrence of risk behaviors [
49]. Furthermore, the view has been verified in early empirical research, which indicated girls were more vulnerable to HRBs than boys [
50]. The proportion of children intake breakfast daily is at an extremely high level, reducing somewhat in adolescence [
51]. However, our study suggests that there is no gender difference in the association between cumulative ecological risk and breakfast intake, which indicating a significant correlation for both girls and boys, respectively. Further research is needed to specifically classify gender differences in the relationship between cumulative ecological risk from multiple dimensions and risky behavior.
The study sample used in the present study was favorable for a large-scale adolescents base, covering a wide area involving both rural and urban residences in eight cities in China. This survey examined the cumulative ecological risk associated with multiple forms of risky behavior and aggregated HBRs, and further examined gender differences in the relations further, based on the perspective of co-occurrence of risk factors from seven dimensions including individual, family, school, community, policy, culture and time, and multiples behaviors, respectively. However, several weaknesses in the present investigation had to be considered. First, due to the cross-sectional study design and the lack of longitudinal data, causal correlations could not be established. Second, survey data collected by self-reporting was prone to recall bias. Third, data related to unhealthy dietary diets (vegetable and fruit intake) was excluded from the study due to a large number of missing and other reasons. Fourth, the choice of the best-fitting latent class model on HRBs mainly took into account entropy, regardless of other metrics, but it may be appropriate to cluster HRBs into two latent classes, given that the large difference between the two and the number of classes.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.