Recommendations for the use of mathematical modelling to support decision‐making on integration of non‐communicable diseases into HIV care

Introduction Integrating services for non‐communicable diseases (NCDs) into existing primary care platforms such as HIV programmes has been recommended as a way of strengthening health systems, reducing redundancies and leveraging existing systems to rapidly scale‐up underdeveloped programmes. Mathematical modelling provides a powerful tool to address questions around priorities, optimization and implementation of such programmes. In this study, we examine the case for NCD‐HIV integration, use Kenya as a case‐study to highlight how modelling has supported wider policy formulation and decision‐making in healthcare and to collate stakeholders’ recommendations on use of models for NCD‐HIV integration decision‐making. Discussion Across Africa, NCDs are increasingly posing challenges for health systems, which historically focused on the care of acute and infectious conditions. Pilot programmes using integrated care services have generated advantages for both provider and user, been cost‐effective, practical and achieve rapid coverage scale‐up. The shared chronic nature of NCDs and HIV means that many operational approaches and infrastructure developed for HIV programmes apply to NCDs, suggesting this to be a cost‐effective and sustainable policy option for countries with large HIV programmes and small, un‐resourced NCD programmes. However, the vertical nature of current disease programmes, policy financing and operations operate as barriers to NCD‐HIV integration. Modelling has successfully been used to inform health decision‐making across a number of disease areas and in a number of ways. Examples from Kenya include (i) estimating current and future disease burden to set priorities for public health interventions, (ii) forecasting the requisite investments by government, (iii) comparing the impact of different integration approaches, (iv) performing cost‐benefit analysis for integration and (v) evaluating health system capacity needs. Conclusions Modelling can and should play an integral part in the decision‐making processes for health in general and NCD‐HIV integration specifically. It is especially useful where little data is available. The successful use of modelling to inform decision‐making will depend on several factors including policy makers’ comfort with and understanding of models and their uncertainties, modellers understanding of national priorities, funding opportunities and building local modelling capacity to ensure sustainability.