A pragmatic randomised controlled trial to compare antidepressants with a community-based psychosocial intervention for the treatment of women with postnatal depression: the RESPOND trial.

摘要

OBJECTIVES: To evaluate clinical effectiveness at 4 weeks of antidepressant therapy for mothers with postnatal depression (PND) compared with general supportive care; to compare outcome at 18 weeks of those randomised to antidepressant therapy with those randomised to listening visits as the first intervention (both groups were to be allowed to receive the alternative intervention after 4 weeks if the woman or her doctor so decided); and to assess acceptability of antidepressants and listening visits to users and health professionals. DESIGN: A pragmatic two-arm individually randomised controlled trial. SETTING: Participants were recruited from 77 general practices: 21 in Bristol, 21 in south London and 35 in Manchester. PARTICIPANTS: A total of 254 women who fulfilled International Classification of Diseases version 10 criteria for major depression in the first 6 postnatal months were recruited and randomised. INTERVENTIONS: Women were randomised to receive either an antidepressant, usually a selective serotonin reuptake inhibitor prescribed by their general practitioner (GP), or non-directive counselling (listening visits) from a specially trained research health visitor (HV). The trial was designed to compare antidepressants with general supportive care for the first 4 weeks, after which women allocated to listening visits commenced their sessions. It allowed for women to receive the alternative intervention if they had not responded to their allocated intervention or wished to change to, or add in, the alternative intervention at any time after 4 weeks. MAIN OUTCOME MEASURES: The duration of the trial was 18 weeks. Primary outcome, measured at 4 weeks and 18 weeks post randomisation, was the proportion of women improved on the Edinburgh Postnatal Depression Scale (EPDS), that is scoring < 13. Secondary outcomes were the EPDS measured as a continuous variable at 4 and 18 weeks, and scores on various other questionnaires. RESULTS: At 4 weeks, women were more than twice as likely to have improved if they had been randomised to antidepressants compared with listening visits, which started after the 4-week follow-up, i.e. after 4 weeks of general supportive care [primary intention-to-treat (ITT), 45% versus 20%; odds ratio (OR) 3.4, 95% confidence interval (CI) 1.8 to 6.5, p < 0.001]. Explanatory analyses emphasised these findings. At 18 weeks, ITT analysis revealed that the proportion of women improving was 11% greater in the antidepressant group, but logistic regression analysis showed no clear benefit for one group over the other [62% versus 51%, OR 1.5 (95% CI 0.8 to 2.6), p = 0.19]. Overall, there was a difference between the groups in favour of the antidepressant group of about 25 percentage points at 4 weeks, which reduced at 18 weeks. No statistical support existed for a benefit of antidepressants at 18 weeks, but 95% CIs could not rule out a clinically important benefit. It was difficult for GPs not to prescribe antidepressants to women randomised to listening visits after the initial 4 weeks, so many women received both interventions in both groups by 18 weeks and consequently power was reduced. Qualitative interviews with women revealed a preference for listening visits but an acceptance that antidepressants might be necessary. They wished to be reassured that their GP and HV were offering continuity of care focusing on their particular set of circumstances. Interviews with GPs and HVs revealed lack of collaboration in managing care for women with PND; neither professional group was willing to assume responsibility. CONCLUSIONS: At 4 weeks, antidepressants were significantly superior to general supportive care. Trial design meant that by 18 weeks many of the women initially randomised to listening visits were also receiving antidepressants, and more vice versa. The lack of evidence for differences at 18 weeks is likely to reflect a combination of reduced power and the considerable degree of switching across the two i