Applying a General and Substance-Specific Risk Factor Model to Explain Problem Behavior in Young Adulthood: Replication and Extension in a Longitudinal Study

Introduction: There is an emerging research focus on gene-environment (G-E) interplay in the prediction of problem behaviors, including substance use, crime, and HIV sexual risk behavior. Correct specification of phenotypes (outcomes) and characterization of the environment are essential for G-E studies. Longitudinal samples adding genetic data often have advantages over cross-sectional G-E studies in the depth, breadth, and frequency of measurement of both environments and phenotypes, but  capturing this array of developmental, environmental and outcome complexity is challenging. Bailey, et al. (2011) proposed and tested an organizing framework that groups risk and protective factors for problem behavior into two sets: those which are descriptive of the general social environment and those which are descriptive of behavior-specific aspects of the environment. For example, family management, conflict, and bonding were grouped into a latent factor describing general family environment. Parent smoking, sibling smoking, and parent attitudes about smoking were grouped into a latent factor describing smoking-specific family environment. Outcomes were similarly divided into general and specific components. The present study applies the general/specific approach to the Raising Healthy Children (RHC) longitudinal sample in order to test generalizability of the approach. This study extends the previous study by considering general and substance-specific peer influences in addition to family influences.

Methods: Data are drawn from the RHC study, a longitudinal study of the etiology of prosocial and antisocial behavior. It followed 1040 youth from age 6/7 to age 24/25. SEM is used to partition variance in age 24 daily smoking, alcohol abuse/dependence, illicit drug abuse/dependence, crime, and HIV sexual risk behavior into general problem behavior and behavior-specific variance. Measures of adolescent general, drinking-specific, and smoking-specific family environment were constructed.

Results: The general/specific approach was supported in this sample: general family climate during adolescence predicted general problem behavior and family smoking environment predicted unique variance in daily smoking at age 24. Unlike in the Bailey study, however, adolescent family smoking- and drinking-specific environments also were related to general problem behavior and adolescent family smoking-specific environment predicted unique variance in alcohol problems at age 24.

Conclusions: The current results and results from Bailey et al. suggest that prevention programs should focus both on general and substance-specific risk and protective factors.