Geographic Variation in Part B Reimbursement and Physician Offsetting Behavior: A Physician Matching Approach

Historically, insurers and researchers have expressed extensive interest in how physicians respond to reductions in reimbursement. The McGuire-Pauly synthesis of Supplier-Induced Demand (SID) implies two possible reactions when a health insurer such as Medicare cuts fees. With lower levels of reimbursement, a utility maximizing physician will have an incentive to reduce service provision from those covered by the insurer, substituting to more profitable patients or toward leisure. Given the complexity of medicine, however, the physician acting as an agent for the patient also has the incentive to induce additional services to "offset" lost income from the fee reduction. The relative dominance of these two incentives determines whether a fee reduction results in a reduction in the quantity of provided services (i.e. when the impact of the fee reduction on the physician's overall income is relatively small) or an increase in the quantity of provided services (i.e. when the impact of the fee reduction on the physician's overall income is relatively large).

One insurer that has historically assumed a SID response is Medicare. When setting its annual updates for the Medicare Part B fee schedule, actuaries for the Centers for Medicare and Medicaid Services have historically operated under the premise that physicians faced with fee reductions, increase their volume and intensity of services to ``offset’’ lost income, to the tune of 30 to 50% of the fee reduction. These values originated from CMS’ own calculations and were validated by a number of empirical studies. However, recent empirical work examining fee reductions has found either small offsetting effects or evidence of reductions in the quantity supplied.

This study contributes to an ongoing debate about physician responses to reimbursement using a distance matching identification strategy that has never been applied in the analysis of SID. It uses the population of physicians providing non-facility services under Medicare Part B between 2013 and 2015 and exploits systematic variation in reimbursement across both Medicare localities and across time due to fee schedule updates to examine the relative response to reimbursement.

Using geocoded practice location data, practices are distance matched to nearby practices (within 30 miles) with identical specialty, gender practitioner composition, and number of practitioners, but who are subject to distinct reimbursement rates. Using inverse distance weighted least squares and controlling for practice level differences in the number of patients, patient health status, demographics, year of observation, the underlying measured differences in county practice costs, and county pair fixed effects, this study examines the effects of relative reimbursement between matched practices on relative annual RVUs, volume of services, and intensity.

Examining all services collectively, this study finds that physicians who are compensated less generously provide more annual RVUs, increased volume of services, and more intense services than nearby physicians and that these results are primary driven by evaluation and management services. It finds little to no evidence of any response to elective and non-elective services and a traditional volume response for imaging and testing services.