This study examined how childhood maltreatment is associated with the longitudinal trajectories of tobacco and alcohol use in a large sample of low-income, rural Chinese adolescents. The results yielded two novel contributions. First, there are unique effects of sexual abuse and neglect on subsequent risks of alcohol and tobacco use, independent of personal, family, and contextual characteristics. Second, the positive aspects of the school and neighborhood environment help to buffer the risks of alcohol use associated with childhood physical abuse experiences. This finding has implications for designing and implementing targeted school and neighborhood-based substance use preventions and interventions.

Different forms of childhood maltreatment were independently associated with baseline adolescent substance use. Sexual abuse and neglect were more strongly associated with substance use than was physical or emotional abuse. The finding that childhood sexual abuse was associated with increased risk of past-month drinking, binge drinking, and smoking is consistent with prior research documenting the negative consequences of sexual abuse [4]. Childhood sexual abuse can impair stress response systems, leading to maladaptive coping behaviors, poor decision-making and low self-regulation [9]. Victims of sexual abuse may be more likely to use alcohol or tobacco as a coping mechanism [33, 34]. We also found sexual abuse was more strongly associated with smoking among youth in middle school than those in elementary school. While the effects of sexual abuse on some aspects of adjustment (e.g., internalizing problems) manifest early, its effect on other domains such as substance use may emerge later in adolescence, possibly through a cascade of developmental challenges involving sexual risk behaviors, delinquent behaviors, and involvement with deviant peers [4].

Neglect uniquely predicted adolescent past-month drinking, binge drinking, and smoking, above and beyond the effects of physical, sexual, and emotional abuse. The effect size of neglect was similar to that of sexual abuse for drinking, and even stronger for smoking. This finding is consistent with past research that the negative effect of neglect on adolescent health and well-being was similar to or even greater than abuse [14, 35, 36]. Neglect is a severe form of early deprivation and may lead to substance use via insecure attachment, which has been associated with emotional dysregulation, externalizing behaviors, and psychosocial adjustment difficulties [37]. Neglect was the most common form of maltreatment in this sample, a result consistent with other research [38]. The high rate of neglect in our sample may reflect the fact that many parents migrated to cities, leaving their children to a single parent, a grandparent, or another adult. Considering that recent research has found distinct components of neglect (e.g., not providing emotional support and affect, exposing youth to risky situations, or inadequate monitoring) that independently influence the likelihood to exhibit health-damaging behaviors [36], it is important for future research to examine the specific component of neglect that influences the well-being and development of children and adolescents.

In prior studies it has also been found that adolescents who had experienced emotional abuse were more likely to initiate alcohol and tobacco early and to use these substances more frequently [12, 33]. Surprisingly, we found emotional abuse was not significantly associated with drinking, binge drinking, or smoking. Physical abuse predicted binge drinking and smoking; however, it was only associated with drinking for boys, but not for girls. Previous findings are mixed regarding sex differences in the effects of physical abuse [39, 40]. Future research is needed to explore the effects of abuse with consideration of multiple factors including cultural values and norms, the developmental stage of participants, and features of physical and emotional abuse such as severity and chronicity.

Consistent with the ecological systems theory, school and neighborhood connections moderated the developmental pathways from early physical abuse to adolescent drinking behavior [16]. Although it was previously found that individuals with a history of childhood maltreatment have heightened vulnerability to adverse environment influences [41], there has been less empirical support for the protective effects of positive environments. For adolescents who experienced childhood physical abuse, feelings of connection to peers and adults in school and neighborhood were associated with a reduced risk for alcohol drinking in adolescence. One plausible explanation is physical abuse is more common for children who exhibit a difficult temperament or other personality or biological traits. As posited by the differential susceptibility theory, these traits may confer a heightened susceptibility to environmental influences, for better and for worse [42]. It is also likely that early physical abuse alters the physiological systems involved in stress responsivity, subsequently enhancing youth sensitivity to their environments [19]. It will be valuable for future research to further examine possible mechanisms of environmental susceptibility among youth who experienced physical abuse. Nonetheless, our findings underscore the important role of a positive context in cultivating resilience among physically abused youth. Policymakers and school administrators may consider interventions geared towards improving overall school and neighborhood connectedness given its potential to effectively mitigate the negative developmental sequalae affecting youth who experienced early-life physical abuse. It is also important to note that the moderating effects of school and neighborhood connections appear limited to physical abuse; thus, caution is needed in interpreting these findings. The potential interaction effects between abuse or neglect experiences and school and neighborhood contexts in predicting adolescent substance use and other developmental outcomes warrant further exploration.

The prevalence rates of 10.2% for past-month drinking and 4.9% for past-month smoking are similar to those found in prior studies of Chinese adolescents [43, 44]. For example, a nationally representative survey of Chinese adolescents yielded rates of past-month smoking of 3.9% for middle school students [43]. In another national survey to middle and high school students in eight Chinese provinces, 11.9% of adolescents reported using alcohol in the past month [44]. We also found that rates of alcohol and tobacco use were higher in middle school than in primary school and higher among boys than girls, patterns consistent with results from other studies [43, 44]. We observed that the proportion of youth in our sample who endorsed past-month alcohol use increased, while it remained unchanged for tobacco use, suggesting that exposures to alcohol and tobacco lasted or even increased over the study period. None of childhood maltreatment forms predicted the rate of change of drinking or smoking behaviors, suggesting that sexual and physical abuse and neglect are associated with higher alcohol and tobacco use at baseline, but this effect may stabilize over time. There are two plausible explanations for the lack of association for substance use change. First, participants who are lost to attribution have higher baseline rates of substance use; thus, our analysis does not capture high-risk behaviors that potentially persist or increase over time and likely underestimates the true impact of childhood maltreatment on the substance use trajectory. Second, there may exist distinct trajectory patterns of substance use within maltreated youth (e.g., some youths have increasing or persistent use, whereas others have decreasing use). Further research to explore developmental patterns and/or risk profiles of substance use (e.g., lifetime exposure, exposure amount, cumulative amount, age of initiation, duration of use) among maltreated samples is warranted.

There are several limitations of this research that suggest the need for future research. First, the study relied on self-reported measurement of the key variables. For example, childhood maltreatment was retrospectively reported by youth, which may be subject to recall bias (e.g., current life situations could influence the recall of past experiences) and could be underestimated given social stigma and victim-blaming faced by victims [45]. Future research may consider using multiple data sources including official records, self-and parent-reported maltreatment data, and parental, teacher, and peer perceptions of youth substance use. Also, the self-report measure of childhood maltreatment did not allow an examination of the age of onset, perpetrator, and chronicity of the maltreatment exposure, which should be considered in future research on childhood maltreatment. Given that abuse and neglect often co-occur, it would be useful to examine the complexity of maltreatment experiences using person-centered analysis such as the latent profile analysis and evaluate how complex patterns of maltreatment experiences relate to adolescent substance use trajectories.

Second, although the validity of the CTQ scale has been supported in prior research [23, 24], the internal consistency of the physical neglect subscale was only moderate. The lower internal consistency of this subscale was not specific to our study but rather common among community samples of adolescents across countries [24,25,26]. We therefore used a combined neglect measure because it provided an acceptable internal consistency and was consistent with the broad conceptualization of neglect used in previous research [27, 35, 46]. In future research it would be helpful to distinguish physical and emotional forms of neglect, using case records, diagnostic interview measures, or other validated multidimensional measures of neglect.

Third, this study assessed alcohol and tobacco trajectory over the period of one year, which may not fully capture the lasting effect of childhood maltreatment. Although we found abuse or neglect did not predict increases in the use of these substances, it is possible that the effects of early maltreatment experiences extend beyond a one-year period and persist into late adolescence and adulthood [47]. Future studies with more years of follow-up data are needed to evaluate the long-term effects of childhood maltreatment on substance use behaviors.

Fourth, given that the protective effects of school and neighborhood connection were only observed in the link between physical abuse and alcohol use, it is important to interpret the school and neighborhood effect with caution. We encourage future research to assess if context matters via various school and neighborhood processes (i.e., perceived safety, access to resources, built environment) and across different developmental outcomes and samples from diverse backgrounds among maltreated youth.

Finally, this study was conducted in a low-income rural area of Eastern China and the generalizability of the reported findings to other populations in China and other regions of the world should be interpreted cautiously. For example, there are considerable differences in economic development and societal norms pertaining to tobacco and alcohol use across China. Results from a recent meta-analysis showed that tobacco use is more prevalent in Eastern China than is in either Central China or Western China [48]. Future research should replicate current findings in different parts of China and in other societies. Nonetheless, our use of a low-income rural Chinese youth sample contributes to the understanding of the impact of childhood maltreatment in underserved populations.