Variance Estimation for a Complex Life Table Quantity: Disease-free Life Expectancy

摘要: Background: In the last decade, adult mortality in United States has continued its long-run decline, while diabetes prevalence increased. It is unknown whether additional person-years lived population have mostly been spent a diseased or disease-free state. Furthermore, although illness and death are stochastic processes, little known about variance diabetes-free life expectancy (DFLE) when compared across ages. More generally, methods of obtaining complex table quantities under-explored. Objective: Estimate DFLE 2000 2010. Methods: Data on for ages 20+ come from National Health Nutrition Examination Surveys (NHANES), 1999-2000 (n=4,205) 2009-2010 (n=5,752). Diabetes was defined as HbA1c at least 6.5% taking medication. Deaths counts by age sex Human Mortality Database, covering entire U.S. population. estimated using Sullivan’s method. Three estimating were explained compared: delta method, Monte Carlo simulation, bootstrapping. Results: Although 20 rose approximately 3 years both males females between 2010, did not change during this decade. At 70, 2.5 2.7 females, but only 0.7 0.8 females. For all methods, sexes years, larger younger (males, 2000, 20, method: 0.020) than older 0.012). any given age/sex/year, method produced smallest estimates DFLE, followed Carlo. Bootstrapping that far largest, often ten times variances. Differences accounted most differences DFLE. Conclusions: The vast majority gained 2010 with diabetes. Variance arises disease prevalence. life-table can be obtained multiple appropriate research problem will vary.