To identify when and how risk (and protective) factors interact to influence the likelihood of suicidal behaviors and substance use, longitudinal studies are needed to begin to understand causal relationships. Well-designed epidemiological and longitudinal studies will help to reduce bias that can complicate cross-sectional study from highly selected or clinical samples (Wilcox and Anthony, 2004). Prospective studies of samples recruited from clinical settings have the advantage of examining the predictors and correlates of suicidal behavior found in treatment populations of substance abusers (Wines et al. 2004) and previously hospitalized adolescents (Goldston et al. 2006). Prospective longitudinal designs also allow for study of developmental timing, duration, and severity of risk factors, as well as examination of etiologic mechanisms. For example, genetically informative prospective designs, such as children-of-twins studies, can be used to examine gene-environment interactions and correlations (Glowinski et al. 2004). Furthermore, case-crossover designs can be used to test hypotheses linking triggering mechanisms to recent substance use and suicidal behavior (Wilcox et al. 2004; Wu and Anthony, 2000).
