Key points are not available for this paper at this time.
BACKGROUND: The control of tuberculosis (TB) in India is complicated by the presence of a large, disorganised private sector where most patients first seek care. Following pilots in Mumbai and Patna (two major cities in India), an initiative known as the 'Public-Private Interface Agency' (PPIA) is now being expanded across the country. We aimed to estimate the cost-effectiveness of scaling up PPIA operations, in line with India's National Strategic Plan for TB control. METHODS: Focusing on Mumbai and Patna, we collected cost data from implementing organisations in both cities and combined this data with models of TB transmission dynamics. Estimating the cost per disability adjusted life years (DALY) averted between 2014 (the start of PPIA scale-up) and 2025, we assessed cost-effectiveness using two willingness-to-pay approaches: a WHO-CHOICE threshold based on per-capita economic productivity, and a more stringent threshold incorporating opportunity costs in the health system. FINDINGS: A PPIA scaled up to ultimately reach 50% of privately treated TB patients in Mumbai and Patna would cost, respectively, US228 (95% uncertainty interval (UI): 159 to 320) per DALY averted and US564 (95% uncertainty interval (UI): 409 to 775) per DALY averted. In Mumbai, the PPIA would be cost-effective relative to all thresholds considered. In Patna, if focusing on adherence support, rather than on improved diagnosis, the PPIA would be cost-effective relative to all thresholds considered. These differences between sites arise from variations in the burden of drug resistance: among the services of a PPIA, improved diagnosis (including rapid tests with genotypic drug sensitivity testing) has greatest value in settings such as Mumbai, with a high burden of drug-resistant TB. CONCLUSIONS: To accelerate decline in TB incidence, it is critical first to engage effectively with the private sector in India. Mechanisms such as the PPIA offer cost-effective ways of doing so, particularly when tailored to local settings.
Arinaminpathy et al. (Fri,) studied this question.