Objective: To determine the relative cost-effectiveness of three classes of antidepressants: tricyclic antidepressants (TCAs), selective serotonin reuptake inhibitors (SSRIs), and the modified TCA lofepramine, as first choice treatments for depression in primary care.
Design: Open, pragmatic, controlled trial with three randomised arms and one preference arm. Patients were followed up for 12 months.
Setting: UK primary care: 73 practices in urban and rural areas in England.
Participants: Patients with a new episode of depressive illness according to GP diagnosis.
Interventions: Patients were randomised to receive a TCA (amitriptyline, dothiepin or imipramine), an SSRI (fluoxetine, sertraline or paroxetine) or lofepramine. Patients or GPs were able to choose an alternative treatment if preferred.
Main outcome measures: At baseline the Clinical Interview Schedule, Revised (CIS-R PROQSY computerised version) was administered to establish symptom profiles. Outcome measures over the 12-month follow-up included the Hospital Anxiety and Depression Scale self-rating of depression (HAD-D), CIS-R, EuroQol (EQ-5D) for quality of life, Short Form (SF-36) for generic health status, and patient and practice records of use of health and social services. The primary effectiveness outcome was the number of depression-free weeks (HAD-D less than 8, with interpolation of intervening values) and the primary cost outcome total direct NHS costs. Quality-adjusted life-years (QALYs) were used as the outcome measure in a secondary analysis. Incremental cost-effectiveness ratios and cost-effectiveness acceptability curves were computed. Estimates were bootstrapped with 5000 replications.
Results: In total, 327 patients were randomised. Follow-up rates were 68% at 3 months and 52% at 1 year. Linear regression analysis revealed no significant differences between groups in number of depression-free weeks when adjusted for baseline HAD-D. A higher proportion of patients randomised to TCAs entered the preference arm than those allocated to the other choices. Switching to another class of antidepressant in the first few weeks of treatment occurred significantly more often in the lofepramine arm and less in the preference arm. There were no significant differences between arms in mean cost per depression-free week. For values placed on an additional QALY of over 5000 pounds, treatment with SSRIs was likely to be the most cost-effective strategy. TCAs were the least likely to be cost-effective as first choice of antidepressant for most values of a depression-free week or QALY respectively, but these differences were relatively modest.
Conclusions: When comparing the different treatment options, no significant differences were found in outcomes or costs within the sample, but when outcomes and costs were analysed together, the resulting cost-effectiveness acceptability curves suggested that SSRIs were likely to be the most cost-effective option, although the probability of this did not rise above 0.6. Choosing lofepramine is likely to lead to a greater proportion of patients switching treatment in the first few weeks. Further research is still needed on the management of depressive illness in primary care. This should address areas such as the optimum severity threshold at which medication should be used; the feasibility and effectiveness of adopting structured depression management programmes in the UK context; the importance of factors such as physical co-morbidity and recent life events in GPs' prescribing decisions; alternative ways of collecting data; and the factors that give rise to many patients being reluctant to accept medication and discontinue treatment early.