Résumé
Study objective - To compare health expectancies calculated by Sullivan's method and the multistate life table method in order to identify the magnitude of the bias in Sullivan's method and assess how seriously this limits its use for monitoring population health expectancies. Design - A simulation model was used to compare health expectancies calculated using Sullivan's method and the multistate life table method under various scenarios for the evolution of disability over time in populations. The simulation model was based on abridged cohort life tables using data on French mortality from 1825-90 and disability prevalence data from the 1982 French health survey. Main results - The Sullivan method could not detect a sudden change in disability transition rates, but the simulations suggested that it provides a good estimate of the true multistate value if there are smooth and relatively regular changes in disability prevalence over the longer term. When disability incidence rates are increasing or decreasing smoothly over time, the absolute bias in the Sullivan estimate of disability free life expectancy is relatively constant with age. The relative bias thus increases at older ages as disability free life expectancy decreases. Conclusions - The difference between the estimates produced by the two methods was small for realistic scenarios for the evolution of population health and Sullivan's method is thus generally acceptable for monitoring relatively smooth long term trends in health expectancies for populations.