Effect of collaborative care between traditional and faith healers and primary health-care workers on psychosis outcomes in Nigeria and Ghana (COSIMPO): a cluster randomised controlled trial.

Background: Traditional and faith healers (TFH) provide care to a large number of people with psychosis in many sub-Saharan African countries but they practise outside the formal mental health system. We aimed to assess the effectiveness and cost-effectiveness of a collaborative shared care model for psychosis delivered by TFH and primary health-care providers (PHCW).

Methods: In this cluster-randomised trial in Kumasi, Ghana and Ibadan, Nigeria, we randomly allocated clusters (a primary care clinic and neighbouring TFH facilities) 1:1, stratified by size and country, to an intervention group or enhanced care as usual. The intervention included a manualised collaborative shared care delivered by trained TFH and PHCW. Eligible participants were adults (aged ≥18 years) newly admitted to TFH facilities with active psychotic symptoms (positive and negative syndrome scale [PANSS] score ≥60). The primary outcome, by masked assessments at 6 months, was the difference in psychotic symptom improvement as measured with the PANSS in patients in follow-up at 3 and 6 months. Patients exposure to harmful treatment practices, such as shackling, were also assessed at 3 and 6 months. Care costs were assessed at baseline, 3-month and 6-month follow-up, and for the entire 6 months of follow-up. This trial was registered with the National Institutes of Health Clinical Trial registry, NCT02895269.

Findings: Between Sept 1, 2016, and May 3, 2017, 51 clusters were randomly allocated (26 intervention, 25 control) with 307 patients enrolled (166 [54%] in the intervention group and 141 [46%] in the control group). 190 (62%) of participants were men. Baseline mean PANSS score was 107·3 (SD 17·5) for the intervention group and 108·9 (18·3) for the control group. 286 (93%) completed the 6-month follow-up at which the mean total PANSS score for intervention group was 53·4 (19·9) compared with 67·6 (23·3) for the control group (adjusted mean difference -15·01 (95% CI -21·17 to -8·84; 0·0001). Harmful practices decreased from 94 (57%) of 166 patients at baseline to 13 (9%) of 152 at 6 months in the intervention group (-0·48 [-0·60 to -0·37] p<0·001) and from 59 (42%) of 141 patients to 13 (10%) of 134 in the control group (-0·33 [-0·45 to -0·21] p<0·001), with no significant difference between the two groups. Greater reductions in overall care costs were seen in the intervention group than in the control group. At the 6 month assessment, greater reductions in total health service and time costs were seen in the intervention group; however, cumulative costs over this period were higher (US $627 per patient vs $526 in the control group). Five patients in the intervention group had mild extrapyramidal side effects.

Interpretation: A collaborative shared care delivered by TFH and conventional health-care providers for people with psychosis was effective and cost-effective. The model of care offers the prospect of scaling up improved care to this vulnerable population in settings with low resources.

Funding: US National Institute of Mental Health.