Objectives. Depression is common after stroke and can impede rehabilitation. The aim of this study was to determine which factors predicted the severity of depression at the time of recruitment to a treatment study and depression 6 months later.

Design. A questionnaire-based longitudinal study.

Methods. A total of 123 depressed stroke patients who were between 1 and 6 months post-stroke and had mild to moderate disabilities were recruited to a randomized controlled trial of cognitive behavioural psychotherapy (CBT) to treat depression (Lincoln & Flannaghan, 2003). At recruitment and at 6 months follow-up, patients (N=112) were assessed on the Beck Depression Inventory (BDI), General Health Questionnaire-28 (GHQ-28), Wakefield Depression Inventory (WDI), Barthel index, Extended activities of daily living scale (EADL), Sheffield Screening Test for Acquired Language Disorders (SST) and Recovery locus of control scale (RLOC).

Results. Logistic regression showed that greater communication impairment on the SST at recruitment was predictive of severe depression on the BDI at recruitment (OR=0.72, p=.01). Patients with greater communication impairment (OR=0.69, p<.05) and a more external locus of control (OR=0.89, p<.05) at recruitment were more likely to remain depressed at follow-up. Patients who remained depressed at follow-up were more severely depressed at recruitment (p=.001).

Conclusions. Locus of control and communication impairment were related to depression. Communication impairment was the strongest predictor of depression severity and prognosis. Mild depression was more likely to resolve. The regression models only accounted for a small proportion of the variance in depression scores. Therefore, further studies of psychosocial factors in a more representative sample are required.