Protocols and individualized treatments: Medical practice variation among Medicare beneficiaries
Date of Completion
Economics, General|Sociology, Public and Social Welfare|Health Sciences, Health Care Management
In this dissertation I study the effect of physician behavior on medical practice variation among regions of the United States. The objectives of this study are to identify non-clinical influences on medical practice variation, and to provide a model of physician behavior that improves the predictive ability of regional variation models. In the theoretical “protocol model,” practice style is reflected in the physician's decision to treat a given condition with a well-defined protocol or a more costly individualized treatment. ^ Empirically, I combine risk adjustment methods and multilevel modeling to isolate and measure the effects of health status and practice style on medical practice variation. Medicare claims data for total annual medical expenditures of sample beneficiaries, and for a specific category of expenditures (vascular disease), are used to test various hypotheses regarding physician practice style. The expenditure analysis demonstrates that a multilevel model provides a more efficient assignment of variation by accounting for correlation or “clustering” at the Health Service Area level, and provides limited support for the protocol model. ^ Additional support for hypotheses of the protocol model is provided by the analysis of vascular disease treatment patterns. Both empirical analyses indicate that physician treatment decisions are influenced by regional health patterns, even after accounting for individual level health status. ^
Olmsted, Eric Arthur, "Protocols and individualized treatments: Medical practice variation among Medicare beneficiaries" (2004). Doctoral Dissertations. AAI3134795.