Is magnitude of co-payment effect related to income? Using census data for health services research

This study: (1) describes and evaluates a method for estimating household income using home addresses in conjunction with census data, and (2) uses this method in a study designed to determine if the introduction of copayments caused primary care office visit rates to decrease more for lower income enrollees than for higher income enrollees of a large health maintenance organization (HMO). Each step in the process of linking income estimates for small census units (block groups) to specific home addresses is described and the validity and precision of these estimates is assessed through comparisons of estimated income with self-reported income for residents of the Puget Sound area of Washington state. Although subject to considerable measurement error, this approach provided valid income estimates. This method was then used to estimate the incomes of over 20,000 households of HMO enrollees in a controlled study of the relationship between copayment effect on visit rates and enrollee income. Copayments were found to have a similar effect on higher and lower income enrollees.