Predicting the burden of acute myocardial infarction in a country with limited resources: combining data from routine sources

BACKGROUND: Prior to implementation of a national surveillance system for cardiovascular disease in Barbados, a small island state with limited health resources, the national burden of acute myocardial infarction (MI) was unknown. METHODS: We retrospectively estimated national acute MI incidence rates (IRs) per 100,000 during the decade before registry implementation (1999-2008), using easily accessible routine data from different sources, assessing changes over time through Poisson regression. Future events (2009-2013) were estimated using simple sensitivity analysis to incorporate prediction uncertainty. Model predictions were compared with actual IRs from initial years of the registry. RESULTS: In 2000, crude IR was 85.5 (95% CI: 74.9-97.2), rising to 92.1 (81.2-103.9) in 2008. Accounting for population ageing, the model anticipated IR of 115.9 in 2010 (99.7-132.1), vs actual crude IR 129.7 (115.9-144.6). CONCLUSIONS: Despite no electronic medical record system in Barbados, data were simple to collect, and provided a rough baseline for acute MI burden. We show that, in countries with small populations, limited resources and in absence of surveillance, national mortality statistics and routine hospital data can be combined to adequately model national estimates of acute MI incidence. This cheap and simple, yet fairly accurate method could be a key tool for other low-resource countries with ageing populations and increasing cardiovascular disease levels.