Life Expectancy Estimation with Breast Cancer: Bias of the Declining Exponential Function and an Alternative to Its Use

Publisher Website
Google Scholar
Abstract
Background. Life expectancy gain (LEG) is an outcome measure commonly estimated with a declining exponential function in a Markov model. The accuracy of such estimates has not been objectively evaluated. Purpose. To compare LEGs from declining exponential function estimates with those calculated from population data, using published screening mammography studies as examples. Method. SEER-based population data are used to compare LEG calculation with declining exponential function estimation and empinc population data in a new model, the "nested" Markov. Results. Analyses of the LEG of mammographic screening based on the declining exponential function significantly overestimate LEGs for younger women and underestimate them for older women. Because of offsetting errors, all-age analyses paradoxically appear accurate. Conclusion. Declining exponential function estimates of LEGs for chronic diseases with low mortality rates and long time horizons are liable to significant bias, especially with limited age cohorts. Key words: life expectancy; survival; mortality; DEALE; Markov model; misestimation bias; misspecification bias. (Med Decis Making 1999; 19:385-393)