Abstract
BACKGROUND: Current guidelines for the management of depression suggest the use of guided self-help for patients with mild to moderate disorders. However, there is little consensus concerning the optimal form and delivery of this intervention. To develop acceptable and effective interventions, a phased process has been proposed, using a modelling phase to examine and develop an intervention prior to preliminary testing in an exploratory trial. This paper (a) describes the modelling phase used to develop a guided self-help intervention for depression in primary care and (b) reports data from an exploratory randomised trial of the intervention. METHODS: A guided self-help intervention was developed following a modelling phase which involved a systematic review, meta synthesis and a consensus process. The intervention was then tested in an exploratory randomised controlled trial by examining (a) fidelity using analysis of taped guided self-help sessions (b) acceptability to patients and professionals through qualitative interviews (c) effectiveness through estimation of the intervention effect size. RESULTS: Fifty eight patients were recruited to the exploratory trial. Seven professionals and nine patients were interviewed, and 22 tapes of sessions analysed for fidelity. Generally, fidelity to the intervention protocol was high, and the professionals delivered the majority of the specific components (with the exception of the use of feedback). Acceptability to both professionals and patients was also high. The effect size of the intervention on outcomes was small, and in line with previous analyses showing the modest effect of guided self-help in primary care. However, the sample size was small and confidence intervals around the effectiveness estimate were wide. CONCLUSION: The general principles of the modelling phase adopted in this study are designed to draw on a range of evidence, potentially providing an intervention that is evidence-based, patient-centred and acceptable to professionals. However, the pilot outcome data did not suggest that the intervention developed was particularly effective. The advantages and disadvantages of the general methods used in the modelling phase are discussed, and possible reasons for the failure to demonstrate a larger effect in this particular case are outlined.