Product tying and prices for physician services

Tuesday, June 14, 2016: 1:35 PM
B26 (Stiteler Hall)

Author(s): Laurence Baker; Kate Bundorf; Dan Kessler

Discussant: Richard C. Lindrooth

Background:  Multi-specialty physician practice has become more common.  This may facilitate improvements in care for patients, but may also be problematic.  One potential issue arises from the ability of physicians in different specialties that are part of the same organization to bargain jointly with insurance companies over prices. Multispecialty groups may be able to translate the market power of a one specialty into higher prices for other types of services for which physician markets are more competitive. “Product tying” has been studied in other contexts, but has not been well explored in physician services.  Some theory suggests that the opportunities for tying to create pricing advantages may be limited, but there are also reasons to believe it may be feasible.  We test the hypothesis that, for multispecialty groups, market power for specialist services can lead to higher prices for services provided by primary care physicians.

Data and Methods.  We create measures of physician practice characteristics using 2008-2011 Medicare claims data, identifying practices using tax ID numbers.  For each practice, we measure size, whether it is multi-specialty, and the specialty-specific HHIs.  HHIs are calculated using practice-specialty-specific market areas, using previously published techniques.  We measure prices (allowed amounts) paid to IM-internal medicine, FP-family practice, or GP-general practice physicians for office visits by private health plans using 2008-2011 claims data from the Health Care Cost Institute.  We use regressions to create ZIP-code-year level primary care log-price indices for office visits, holding fixed patient and health plan characteristics.  We associate the price indices with measures of practice characteristics rolled up to the ZIP-code-year level, and estimate regressions of prices on the characteristics of area practices, including size, HHIs of primary care physicians, share of primary care practices that also include other specialties, and the HHIs of other specialties, along with a range of other controls for area and doctor characteristics and time trends. We estimate models separately for FP/GP and IM physicians. Our test for the presence of tying is whether, after controlling the degree of concentration in the market for primary care services, primary care prices respond to the HHIs of other specialties in multi-specialty practices.

Results.  We find that primary care prices respond to HHIs of other specialties.  For FP/GP physicians, we have 82,496 ZIP-code-year price index measures.  After controlling for potential confounders, increases in the FP/GP HHI are associated with higher prices, as expected.  Importantly, increases in the HHIs of other medical and surgical specialties in the same practices as FP/GP physicians are also associated with higher FP/GP prices.  For example, a 1000 point increase in the mean HHI of other medical subspecialties in the practice is associated with a 3% increase in FP/GP prices (p<.001).  In ongoing analysis, these results are robust to the inclusion of ZIP-code fixed effects, though magnitudes are smaller. 

Conclusions.  Policy targeted to multi-specialty groups should take into account the potential for tying in multi-specialty physician practices to lead to higher prices.  Further work should examine the mechanisms through which tying operates.