It is also notable that we identified only additive effects on time-to-suicide attempt post-hospitalization. In contrast to the findings reported by Goldston et al. (1999), who found that some indices of adolescent functioning had differential predictive utility as a function of history of suicide attempt, we found no interactions between individual level variables. Specifically, we found no interactions between history of suicide attempt, depression severity, anxiety, alcohol/drug abuse, and hopelessness, in the prediction of suicide attempt. We also found no interactions between history of suicide attempt and parental history of mental health problems, or between indices of adolescent functioning and parental history of mental health problems. Rather, parental history of mental health problems, adolescent functional impairment, and adolescent suicidal thoughts were each predictors in our model. It is possible that this difference in findings is partially due to our use of time-varying covariates for variables of adolescent functioning rather than the baseline covariates used in the Goldston et al. study. It may also be due to the differences in the follow-up interval as the Goldston et al. study incorporated a
