A collaborative care package for depression comorbid with chronic physical conditions in South Africa.

INTRODUCTION: A task-sharing collaborative care model for integrated depression care for South Africa's burgeoning primary health care population with chronic conditions was developed and tested through two pragmatic cluster randomized controlled trials. One trial focused on patients with hypertension and was located in one district where a collaborative care model was co-designed with district stakeholders. The other trial, focused on patients on antiretroviral treatment, was located in the same district site, with the addition of a second neighbouring district, without adaptation of the original model. This paper describes the package used to implement this model, and implementation outcomes across the two sites, and summarises lessons and challenges.

METHODS: The Template for Intervention Description and Replication (TIDieR) framework, adapted for complex health systems interventions, was used to describe components of the package. Additional elements of 'modifications made' and 'actual implementation' introduced in the 'Getting messier with TIDieR' framework, were used to describe implementation outcomes in terms of reach, adoption and implementation across the two trial districts.

RESULTS: In the absence of a co-design process to adapt the model to the context of the second site, there was less system level support for the model. Consequently, more project employed human resources were deployed to support training of primary care nurses in identification and referral of patients with depression; and supervise co-located lay counsellors. Referrals to co-located lay counselling services were more than double in the second site. However, uptake of counselling sessions was greater in the first site. This was attributed to greater in-vivo supervision and support from existing mental health specialists in the system. There was greater reliance on online supervision and support in the second site where geographical distances between clinics were larger.

CONCLUSION: The need for in-country co-designed collaborative care models, and 'implementation heavy' implementation research to understand adaptations required to accommodate varying in-country health system contexts is highlighted.