The effects of benefits plan design on the utilization of preventive medical services

This dissertation presents a multivariate regression analysis of the demand for key preventive health services. Utilization of routine physical examinations and various early detection diagnostic procedures are empirically modeled in linear and logistic regression specifications. This dissertation is unique in that it is the first study we know of that estimates cost share responses and other demand coefficients for preventive services using actual claim data from a major managed care organization (MCO). It is also unique in that it estimates these demand equations for a number of managed care products and for a broader array of preventive services than analyzed in other studies.^ The study focuses on the utilization of various key screening services that are considered secondary prevention, as they are prescribed to screen asymptomatic individuals for possible diseases. There are six groups of services in this study: routine physical examinations, diagnostic testing for four cancers (breast, cervical, colorectal and prostate) and cholesterol screening. ^ The rich data in this MCO\u27s claim and enrollment systems enabled me to include a number of independent variables that are typically not available in survey-based empirical studies. Independent variables include the terms of insurance, the managed care product, and a number of patient attributes including age, gender, active/retired status, employee/dependent classification and health status.^ The regression equations are specified as binary choice models: The individual either utilized the selected preventive service in 2002 or did not. Results are generally as expected and cost-share sensitivities are consistent with other empirical studies. Age is a strong predictor of utilization, and shows a positive coefficient for all of the groups, except PAP (cervical cancer screening services). Other broad conclusions are that utilization of these service types is cost-share sensitive, employees are more likely to consume most of these services than dependents, actives more so than retirees and females more so than male members.^ The results in this dissertation may provide employers and plan design decision makers with information regarding consumer behavior in utilizing preventive medical care. Enriching the datasets with individual level predictive risk scores and health risk surveys may improve the predictive power of these models.