Data-driven course trajectories in primary care patients with major depressive disorder

Background: The chronic nature of MDD has been acknowledged as one of the key determinants of the burden associated with depression. Unfortunately, so far described prognostic factors have been inconsistent, possibly due to used course outcomes that are often based on arbitrary criteria/cut-offs. Therefore, the aim of the current study was to use data-driven trajectory groups based on closely spaced weekly severity ratings, as outcomes in prognostic research. Methods: The sample consisted of primary care patients with MDD (n = 153), who were followed up for a year with 52 consecutive weekly ratings of the nine DSM-IV MDD criterion symptoms. Growth Mixture Modeling (GMM) was used to reduce the interpersonal growth variation to an optimal set of clinically interpretable trajectory groups. Next, baseline course predictors were investigated and the prognostic (added) value of course-group membership was investigated for clinical outcomes after 1, 2, and 3 years. Results: GMM resulted in four trajectory groups: "early remission" (40.2%), "late remission" (9.8%), "remission and recurrence" (17.0%), and "chronic" (33.0%). Multivariate predictors of "chronic" group membership were a prior suicide attempt, comorbid dysthymia, and lower levels of somatic depressive symptoms. Group membership predicted differences in depression severity and/or quality of life after 1, 2, and 3 years. Conclusions: The used data-driven approach provided a parsimonious and clinically informative way to describe course variation across MDD patients. Using the trajectory groups to investigate prognostic factors of MDD provided insight in potentially useful prognostic factors. Importantly, trajectory-group membership was itself a strong predictor of future mental well-being