Background. A randomized controlled clinical trial compared three treatments for major depression in primary care. Ninety-one patients were randomized to receive either problem-solving, or amitriptyline with standard clinical management, or drug placebo with standard clinical management. This paper examines the possible predictive value of a range of demographic clinical and personality variables in determining outcome.

Methods. To examine the interaction between treatment group and predictor variables, analyses of co-variants were carried out for two outcome measures: the Hamilton Rating Scale for Depression and the Beck Depression Inventory. Predictor variables were also entered into a stepwise forward logistic regression in which patients were categorized as recovered or not recovered.

Results. None of the predictor variables examined interacted with treatment to predict outcome. Patients with a longer duration of illness, chronic social difficulties, or definite, as opposed to probable, major depression had a poorer outcome at the end of treatment, irrespective of treatment group.

Conclusions. The findings in this study do not enable clear guidelines to be given as to which patients with major depression in primary care might benefit from either a psychological or pharmacological treatment.