A0421
Title: Obtaining population-based estimates for survey data using Bayesian hierarchical models with poststratification
Authors: Emma Zang - Yale University (United States) [presenting]
Abstract: For large-scale surveys such as the National Health and Aging Trends Study (NHATS), investigators may wish to combine data from two (or more) cohorts in a single analysis to obtain larger sample sizes. Unfortunately, it is not possible to combine the 2011 and 2015 NHATS cohorts while retaining the sample weights. Bayesian hierarchical models are applied with poststratification as an alternative strategy for obtaining population-based estimates from NHATS. As proof of principle, prevalence estimates of frailty obtained from the Bayesian approach are compared with those obtained from the 2011 and 2015 cohorts using the NHATS sample weights. Once validated, the strategy is applied to combine the cohorts into a single analytical dataset without overlapping participants and generate Bayesian estimates of frailty for the combined cohort. The Bayesian models were validated within each cohort, producing nearly identical results to those using NHATS sample weights. The Bayesian estimates for the combined cohort were similar to cohort-specific estimates but were more precise. The ability to combine cohorts while generating population-based estimates will permit investigators to not only produce more precise estimates but also address questions that require larger sample sizes and, in turn, increase the value of NHATS to the scientific community.
