Association for Public Policy Analysis & Management

Lately, policymakers have been particularly interested in increasing the use of preventive health care services to improve Americans’ health and lower health care costs. Combined with the well-documented geographic variation in health outcomes, utilization and spending, the diffusion of novel, inexpensive, and efficacious medical technologies is relevant for shaping future policy. I focus on the diffusion of Zostavax, the vaccine used to prevent shingles in the elderly, and evaluate which factors are most strongly related to its adoption. Do socio-demographics or social influences matter? I construct a proxy for physician networks using the number of patients shared by pairs of physicians. I use dynamic panel data models to correct for state-dependence and potential bias arising from the endogeneity of social network formation. I find that treatment patterns in small areas ultimately drive take-up. Observed factors do not explain the large differences in adoption, and social networks have minimal effect on Zostavax diffusion. Patients’ behavior is largely prevalence-inelastic. Heavily subsidized, sicker patients are far more price sensitive than the general elderly population, despite facing very low prices comparatively. Estimates from OLS models with fixed effects and lagged dependent variables are considerably biased compared with those carefully utilizing dynamic GMM estimators.