Abstract:
|
The MEPS-HC is a large household survey sponsored by AHRQ. The survey's purpose is to collect information from households on use of health care, costs of these services, who pays for the cost, and characteristics of the household that relate to health care use, such as demographics, insurance, and employment status. Because the MEPS-HC has a stratified cluster sample design, and strata are not limited to individual states, direct state estimates are not made with the sample data.
The paper gives a method being tested to make state estimates for types of uses and their costs, such as, dental or doctor visits. The method uses models developed from the sample and population values to make estimates. This is done by predicting values for types of persons by location and using population totals to calculate averages and totals for geographic areas. Mean squared errors of the estimates are made using variance estimates plus bias estimates made by comparing model and sample estimates in similar areas. Errors for State estimates are made using the variance of the State estimates and extrapolation of bias estimates from areas where both model and sample estimates can be compared.
|