When donor funding leaves: an interrupted time-series analysis of the impact of integrating direct HIV care and treatment into public health services in a region of Johannesburg

BACKGROUND: Early in South Africa’s HIV response, donor-funded organizations directly provided HIV treatment through Comprehensive HIV Care, Management and Treatment sites (CCMTs), using their own and government staff. From 2012 to 2014 the donor-funded CCMT model was phased out, leaving nurses in South Africa’s public clinics responsible for delivery of antiretroviral treatment (ART) services. We aimed to examine the impact on resources, staff workloads, and service delivery throughout this period of integration of HIV treatment into primary health clinics. METHODS: We conducted an Interrupted Time-Series Analysis (ITSA) using data from three public clinics, including one former CCMT site, in one administrative region of Johannesburg. The ITSA was complemented by visual inspection of the data in Excel. We compared trends in expenditure, clinical staffing levels, patient headcounts, and services rendered at the clinics during four periods: pre-CCMT (2004–2007), CCMT operational (2007–2012), CCMT closure (2012–2014), and post-CCMT (2014–2016). Data were drawn from the country’s District Health Information System, a national HIV treatment database, local budget and expenditure reports, National Health Laboratory Service charge records, and staff records. RESULTS: Closure of the CCMT differentially impacted the study clinics. As expected, ART services decreased at Clinic 1, where the CCMT was co-located, and increased at Clinics 2 and 3 possibly reflecting redistribution of patients. Despite a reduction in patient headcounts post-CCMT, Clinic 1 experienced a decrease in staff and a large increase in patients seen per clinical staff member per month. In contrast, Clinics 2 and 3 increased or maintained stable workforces, and staff workloads post closure were similar to pre-closure levels. Other primary care services—contraception and immunisations—seemed largely unaffected at Clinics 1 and 2. At Clinic 3, service delivery reduced, but this was accompanied by lowered patient headcounts generally, likely due to clinic renovations. CONCLUSIONS: In this study, integration of HIV treatment into primary healthcare services did not result in large-scale reductions in overall service delivery. One facility did experience increased staff workloads, but we were unable to assess service quality. To mitigate potential problems, monitoring systems should be introduced in advance and acknowledge the disparate and decentralised management of various data sources.