Predicting heterogeneity of treatment response in a clinical sample of suicidal adolescents

Abstract

During the past two decades, several treatments specifically targeting youth suicide have been developed. Although most of these treatments have been shown to be more effective than treatment as usual, research has consistently indicated substantial variation in adolescents’ response to these treatments. The current study employed Growth Mixture Modeling (GMM) to categorize trajectories of symptoms over the course of a sixteen-week comparative efficacy trial for suicidal adolescents. One hundred and twenty adolescents (ages 12-18) were randomly assigned to receive either Attachment Based Family Therapy (ABFT) or Family-Enhanced Non-Directive Supportive Therapy (FE-NST). GMM identified three distinct classes of adolescents’ trajectories of depressive and suicidal symptoms: a) Non-Responders (15.8%), b) Good Responders (57.5%), and c) Slow Responders (26.7%). Well-established risk factors for suicide, as well as baseline sociodemographic variables were then tested as predictors of class membership. Results showed that baseline levels of suicide ideation, MDD diagnosis, pessimism, NSSI, and perceived burdensomeness distinguished class membership and predicted poorer response to treatment. These results point to baseline variables that predict lower likelihood of benefitting from treatment (ABFT and FE-NST) and add to the growing literature on individualized treatment planning.