Code for 'The impact of aspirin, statin therapy, and physical activity for the prevention of cardiovascular diseases in 12 African countries from 2027 to 2056: a mathematical modelling study'

Background: Cardiovascular disease prevention in African settings must balance expected health gains against uncertain long-term adherence, programme reach, safety, implementation costs, and limitations in surveillance data. We revised a multi-country cardiovascular disease model after peer review to make these assumptions explicit and to test them directly.

Methods: We used an annual first-event cohort model for 12 African countries over 2027-2056. The revised model uses bounded attenuated-logit risk-factor trajectories after 2030, restricts aspirin to secondary prevention in the base case, applies physical activity promotion to the inactive population, models a 5-year scale-up in coverage, and retains medication adherence as an explicit time-varying parameter. We ran 1,000 Monte Carlo draws per country and scenario, with convergence assessed against a 2,000-draw run. The reported results were regenerated with the included Python reference implementation using a fixed seed; companion R scripts implement the same model equations. Outcomes were CVD events, deaths, disability-adjusted life years (DALYs), healthcare treatment costs avoided, implementation-cost thresholds, and an event-equivalent aspirin harm threshold. GBD 2023 country estimates were used only as external epidemiological context and plausibility checks.

Results: Under the archived model scaling denominators, the statin, aspirin and physical activity strategy averted 1,388,760 (1,280,858 to 1,505,723) CVD events and 416,628 (384,257 to 451,717) deaths, equivalent to 6.0% of projected baseline CVD deaths. It averted 7.11 (6.55 to 7.72) million discounted DALYs and avoided 604.9 (557.3 to 657.0) million US dollars in CVD treatment costs before implementation costs. The corresponding break-even implementation cost was US$0.74 per reached person-year; at an illustrative cost of US$1.00 per reached person-year, implementation costs exceeded the healthcare treatment-cost offset. Statin monotherapy generated most of the health gain. For aspirin alone, an excess major-bleeding rate above approximately 0.38 per 1,000 reached person-years would erase the CVD-event benefit if one major bleed were weighted as equivalent in harm to one CVD event. This is a threshold, not an observed bleeding estimate. The 1,000-draw Monte Carlo run differed from the 2,000-draw reference by at most 0.17% in mean discounted DALYs and 3.68% in DALY uncertainty-interval width.

Conclusions: The revised model supports potential population-level benefits from integrated CVD prevention, particularly statin-based strategies, but it does not establish net cost savings or policy superiority. Direct threshold analyses show that implementation costs and aspirin-related harm can reverse the apparent benefit under adverse assumptions. Long-term adherence, programme effectiveness, safety, costing, and the quality and representativeness of country inputs remain important uncertainties. Country-specific implementation, safety, costing and empirical effectiveness studies are needed before national policy decisions are made.