Health Service Utilization and Insurance Coverage: A Multivariate Probit Analysis

A computationally practical form of probit analysis for multiple response variables was applied to data on health service utilization as it relates to type of supplemental insurance coverage and other demographic covariates in a sample of 4,658 individuals over age 65 on Medicare. The quantal response variables comprised use of five health services: Medical provider visit, hospital outpatient visit, emergency room visit, hospital inpatient stay, and home health care visit. An individual could use any or all of these services. Supplemental insurance coverage was categorized as Medicaid, employer-related private insurance, and privately purchased insurance, and any combination of these insurance coverages was possible. Covariates included health status, education, urban location, race, sex, marital status, employment status, age, and income. The multivariate probit model estimates association between insurance coverage variables and health service utilization patterns jointly for the five health service variables while controlling for the effects of the covariates and correlation among the health service utilization variables. The estimated correlation among the five health service utilization variables is then taken into account when testing the hypothesis of no insurance coverage effects, providing an advance over traditional approaches such as univariate logistic regression. Data from all five service utilization variables are used to perform a single test of insurance coverage effects taking into account their intercorrelation and effects of covariates—the major contribution of this article. In addition, the multivariate probit model provides probability estimates for specific patterns of service use conditional on type of insurance coverage and demographic factors.