The accuracy of hospital records and death certificates for acute myocardial infarction

BACKGROUND: In Australia information on the incidence of acute myocardial infarction (AMI) is available from routinely collected morbidity and mortality data. Given that these data are used for monitoring AMI it is important to assess their quality. AIMS: This paper examines the accuracy of the hospital records and death certificates. METHODS: Morbidity and mortality data were compared with the Newcastle heart disease register which is part of the WHO MONICA Project for 1986-1991 and sensitivity and positive predictive values calculated. RESULTS: For non-fatal AMI the hospital diagnostic coding had sensitivity of 78.9% (95% confidence interval CI 77.1%-80.7%), and positive predictive value of 65.6% (95% CI 63.7%-67.4%). Sensitivity was higher for men than for women and decreased with increasing age. Sensitivity was higher in those with no history of either AMI or other ischaemic heart disease (IHD), higher in current smokers than ex-smokers or never smokers, and lower in those with a self-reported history of high blood pressure. Sensitivity also varied among hospitals. Positive predictive value varied only with hospital. Both sensitivity and positive predictive value were high for death certificate data--89.9% (95% CI 88.4%-91.3%) and 96.0% (95% CI 95.1%-97.0%), respectively. CONCLUSIONS: Although the mortality data appear to be quite accurate, the hospital data alone are not accurate enough to be used to estimate rates or trends of heart attacks. Additional data are required in order to determine numbers of non-fatal AMIs accurately.