Recursive Partitioning Analysis of Lifetime Suicidal Behaviors in Asian Americans

Previous studies have shown that multiple factors predict suicidal behavior in Asian Americans but have provided little guidance on how to rate the relative impact of these risks. To increase our understanding of suicide in Asian Americans, recursive partitioning analysis was used to identify risk factors most relevant to predicting suicidal behaviors and enhance the decision-making process of clinicians faced with clients presenting with multiple risk factors. Using a classification and regression tree (CART) and random forest analyses, we examined the role of previously identified correlates of suicidal ideation and suicide attempts in 2,095 Asian Americans from the National Latino and Asian American Study (Alegria et al., 2008). Random forest analysis showed that depressive and anxiety disorder were the best predictors of lifetime suicidal ideation, followed by family conflict and family cohesion. The model correctly classified lifetime suicidal ideation with moderate sensitivity (72%) and specificity (76%). CART analysis found that family conflict predicted suicide attempts in more than half of the individuals with suicidal ideation. Of those who reported low family conflict, 73% attempted suicide if they had low family support. Perceived discrimination was the next best predictor of suicide attempts (63%) when family issues were not a problem. Cross-validation of this model had moderate sensitivity (75%) but low specificity (39%). These findings highlight the importance of considering family relationship when assessing suicide risk in Asian Americans.