Abstract

STUDY OBJECTIVE--Health expectancy is an increasingly used indicator of population health status. It collapses both mortality and morbidity into a single indicator, and is therefore preferred to the total life expectancy index for populations with low mortality but high morbidity rates. Three methods of calculation exist: the Sullivan, double decrement, and multi-state methods. This report aims to describe their relative advantages and limitations when used to monitor changes in population health status over time. DESIGN--The differences between the three methods are explained. Using a dynamic model of heart disease, the effect of the introduction of thrombolytic treatment on the survival of patients with acute myocardial infarction is calculated. The resulting changes in health expectancy are calculated according to the Sullivan and multi-state methods. MAIN RESULTS--As opposed to the double decrement and the multi-state methods, the Sullivan method produces spurious trends in health expectancy in response to the change in survival. CONCLUSIONS--Estimates of health expectancy in a dynamic situation can be very misleading when based on the Sullivan method, with its attractively moderate data requirements. The multi-state method, which requires longitudinal studies of population health status, is often indispensable.