Major depressive disorder is highly recurrent over the lifespan, even following successful pharmacological and/or psychological intervention. Here we investigate whether predictive modeling can be used to optimize treatment recommendations, when choosing between continuing maintenance antidepressant medication (mADM) treatment or switching to Mindfulness-Based Cognitive Therapy and tapering their ADM (MBCT) as preventive interventions for recurrent depression. Using data (N=424) from the PREVENT trial (Kuyken et al, 2015), we applied elastic net regularized regression for variable selection and built models for predicting relapse over 24-month follow-up in the mADM and MBCT groups compared in the trial. Only the mADM model, including a combination of demographic, clinical and psychological factors, showed better predictive utility than chance. Individuals who were predicted to have a poor prognosis if staying on mADM had a predicted 40% lower risk of relapse across the follow-up period if switching to MBCT. For those with moderate-to-good mADM prognosis, both treatments offered comparable predicted risk. The results suggest that predictive modeling has the potential to guide therapeutic choice around relapse prevention in clinical settings.
Center for Open Science