Effectiveness of a task-sharing collaborative care model for identification and management of depressive symptoms in patients with hypertension attending public sector primary care clinics in South Africa: pragmatic parallel cluster randomised controlled trial.

Background: We tested the real-world effectiveness of a collaborative task-sharing model on depressive symptom reduction in hypertensive Primary Health Care (PHC) patients in South Africa. Method: A pragmatic parallel cluster randomised trial in 20 clinics in the Dr Kenneth Kaunda district, North West province. PHC clinics were stratified by sub-district and randomised in a 1:1 ratio. Control clinics received care as usual (CAU), involving referral to PHC doctors and/or mental health specialists. Intervention clinics received CAU plus enhanced mental health training and a lay counselling referral service. Participant inclusion criteria were ≥ 18 years old, Patient Health Questionnaire-9 (PHQ-9) score ≥ 9 and receiving hypertension medication. Primary superiority outcome was ≥ 50% reduction in PHQ-9 score at 6 months. Statistical analyses comprised mixed effects regression models and a non-inferiority analysis. Trial registration number: NCT 02425124. Results: Between April 2015 and October 2015, 1043 participants were enrolled (504 intervention and 539 control); 82% were women; half were ≥ 55 years. At 6 and 12 months follow-up, 91% and 89% of participants were interviewed respectively. One control group participant committed suicide. There was no significant difference in the primary outcome between intervention (N=256/456) and control (N=232/492) groups (55.9% versus 50.9%; adjusted risk difference = -0.04 ([95% CI = -0.19; 0.11], p = 0.6). The difference in PHQ-9 scores was within the defined equivalence limits at 6 and 12 months for the non-inferiority analysis. Limitations: The trial was limited by low exposure to depression treatment by trial participants and by observed co-intervention in control clinics Conclusions: Incorporating lay counselling services within collaborative care models does not produce superior nor inferior outcomes to models with specialist only counselling services. Funding: This work was supported by the UK Department for International Development [201446] as well as the National Institute of Mental Health, United States of America, grant number 1R01MH100470-01. Graham Thornicroft is supported by the National Institute for Health Research (NIHR) Applied Research Collaboration (ARC) South London at King's College London and King's College Hospital NHS Foundation Trust.