Being able to predict a chronic course in depressive disorders is highly relevant for clinicians. Naturalistic studies on clinical predictors at symptom level are scarce.

To investigate the predictive value of individual depressive symptoms measured with the widely used Beck Depression Inventory self-report scale (BDI-II) on naturalistic outcome (remission/response; measured on the observer-rated Montgomery Äsberg Depression Rating Scale [MADRS]) in a large cohort of depressive outpatients in a psychiatry outpatient setting.

We used a cohort of 1489 adult patients aged 18–65 years with MDD or dysthymic disorder established with the MINI-Plus diagnostic interview. All patients had a Routine Outcome Monitoring (ROM) baseline measurement in 2004–2009, with a maximum of two years of follow-up. We used multivariable Cox regression models to predict remission and response measured on the MADRS, and adjusted for clinical and demographic characteristics that had been identified as correlates of outcome in earlier studies.

Of the 21 BDI-II items, the symptoms pessimism and loss of energy independently predicted non-remission and non-response. For pessimism, the HRs for remission and response were 0.81 (95% CI: 0.73–0.89, P< 0.001) and 0.86 (0.78–0.94, P = 0.001), respectively. For loss of energy, the HRs for remission and response were 0.81 (0.72–0.92, P = 0.001) and 0.84 (0.75–0.95, P = 0.003), respectively.

These findings in a naturalistic treatment setting may help clinicians to identify depressive patients at risk for an unfavourable outcome.