Quality-adjusted HMO cost functions.

摘要

This dissertation investigated the effects of output quality on the cost of producing comprehensive medical care for persons insured by Health Maintenance Organizations (HMOs). Higher quality was hypothesized to increase the quantity of resources required to provide care and hence to increase the cost of producing HMO coverage.; An essential part of this work was the development of a validated firm-level measure of HMO output quality. The source data for assembling a composite quality measure consisted of four years of discrete firm-level output quality measurements contained in the Health Plan Employers Data Information Set (HEDIS). Prediction regressions were used to impute missing HEDIS scores so that the discrete quality data could then be factor-analyzed. A first stage of factor analysis led to seven type-of-care scales and a second stage of factor analysis performed on the type-of-care scales led to a plausible firm-level output quality scale which exhibited strong evidence of internal consistency (Alpha = 0.90). Further composite scales were also constructed, one that included a stochastic dimension and another that relied exclusively on actual rather than predicted HEDIS scores. All three composite quality measures correlated positively and significantly with scales constructed from consumer satisfaction survey results.; A composite measure of quality was then included in two multiproduct and two single-product translog HMO variable cost functions. In addition to quality and output variables, other variables in the specification of the cost functions included input prices (for hospital days, physician visits, and administrative hours), other expenses, a control variable for community-level service-intensity, and time trend. Financial variables needed to estimate cost functions was derived from firm-level data collected regularly by Interstudy from HMO filings with state regulatory agencies. The parameters of each cost function were estimated using a generalized method of moments system estimator on data from which firm-level fixed effects had first been removed.; Estimation results indicated that quality coefficients were significant in all models. Simulation modeling indicated that the cost of improving HMO output quality is small and in some cases may even be negative. Variation in quality is also associated with changes in relative demand for different inputs.