Geographic and Hospital Variations by Payer Type in New Medical Technology Use: The Case of Drug Eluting Stents
Esra Eren Bayindir, Ph.D.1, Pinar Karaca-Mandic, Ph.D.2
1 Ipek University, Department of Economics; Email: ebayindir@ipek.edu.tr
2 University of Minnesota, School of Public Health, Division of Health Policy and Management, Minneapolis, MN; Email: pkmandic@umn.edu; Tel: (612)-624-8953
A large body of research documents significant variation in the timing, intensity, and appropriateness of the use of medical care across geographic regions. This variation suggests that there may be important under-utilization and/or over-utilization of medical care. Less is known about the geographic and hospital variations of the new medical technologies, and even less is known about how these variations differ by payer types such as private insurance, Medicare and Medicaid as well as the uninsured. While much of unexplained variations are attributed to differences in practice styles, it is unknown whether such variations, and thus practice styles vary by payer type in the context of new medical technology diffusion.
In this work, we study the geographic and hospital variations in the diffusion of drug eluting stents that came to market mid-2003, by focusing on how these variations differ by payer type: Medicare, Medicaid, privately insured and uninsured. Our primary data source is the 2004-2005 State Inpatient Databases (SID) for 11 states, which include discharge records of patients from 80 hospital referral regions and 451 hospitals. Our study sample includes all patients aged 18 and older with a discharge for percutaneous coronary intervention procedure that involved the use of least one cardiac stent. Final study sample consists of 495,167 inpatient records. We also use American Hospital Association Annual Survey data to construct measures of hospital characteristics. We estimate separate linear probability models by payer type at the admission level that predict the use of a drug eluting stent. We also conducted separate analyses to estimate the geographic and hospital variations. The specifications for the former included HRR fixed effects, and those for the latter included hospital fixed effects. All our estimates were adjusted controlling for patient characteristics, time-varying hospital characteristics, and indicators for the quarter of admission and state.
Our preliminary results show that variations among HRRs differed by payer type the most during the first year after the technology introduction, but converged the following year. In 2004, the variation among HRRs was smallest for privately insured patients and highest for uninsured patients. In 2005, while variations were still the smallest for privately insured patients, they were similar for Medicare, Medicaid and uninsured patients. Interestingly, variations across hospitals were stable across time. These results suggest that when a new technology becomes available, practice styles heavily depend on payer type and the differences diminish across regions over time however this is not the case for intra-region variations.