The Statistics of Health and Longevity: A Dynamic Analysis of Prevalence Data

Virginia Zarulli, Max Planck Odense Center

Increases in human longevity have made it critical to distinguish healthy longevity from longevity without regard to health. We present a new method for calculating the statistics of healthy longevity which extends, in several directions, current calculations of health expectancy (HE) and disability-adjusted life years (DALYs), from data on prevalence of health conditions. Current methods focus on binary conditions (e.g., disabled or not disabled) or on categorical classifications (e.g. in good, poor, or very bad health) and report only expectations. Our method, based on Markov chain theory, applies to both binary and continuous measures, and provides not only the expectation but also the variance, coefficient of variation, skewness and other distributional properties of longevity. We apply the method to 9 European countries using the SHARE survey data on disability and on grip strength. The method can extend DALY calculations to produce information on variance as well as expectations.