Abstract
Using data from a clinical trial comparing cognitive-behavioral therapy (CBT-SAD) and light therapy (LT) for winter seasonal affective disorder (SAD; N = 177), we explored critical decision points, or treatment weeks, that predict likelihood of nonremission at post-treatment and depression recurrence following treatment. In receiver operator characteristic (ROC) curve analyses, we used weekly Structured Clinical Interview for the Hamilton Rating Scale for Depression-SAD Version (SIGH-SAD) scores during treatment to predict nonremission at post-treatment (Week 6) and recurrence one winter later (Winter 1), two winters later (Winter 2), and any recurrence. Although several C-statistics of ≥ .70 were found, only Week 4 SIGH-SAD scores in CBT-SAD for nonremission had enough predictive ability to inform clinical decision-making (C-statistic = .80; sensitivity = .91; specificity = .68). Week 4 of CBT-SAD may be a critical time point to identify likely nonremitters who need tailoring of intervention, based on SIGH-SAD cutpoint score ≥ 13. This study illustrates how clinical trial data can inform detecting optimal decision points in treatment for identifying patients unlikely to remit, a critical first step to developing adaptive treatment strategies using decision rules to operationalize when and for whom treatment should change to maximize clinical benefit.