How Does Hospital Ownership of Physician Practices Affect Which Hospitals Patients Use?

Why This Study Is Important

Integration between hospitals and physician practices is becoming more widespread as hospitals buy out independent practices, with unknown impacts on these physicians’ patients. This study is the first to examine whether and how hospital ownership of physician practices affects the hospital choices made by Medicare patients treated by these practices. The work specifically examines whether these patients are more likely to choose the owning hospital when they need inpatient care and whether any such influences from their physician supersede considerations based on hospital costs, quality, or distance from home.

What This Study Found

• Depending on how the patient’s physician was defined, patients were anywhere from 16 to 33 percentage points more likely to choose to use a hospital owning their physician’s practice rather than opting to receive their care in other nearby hospitals.
• Without the influence of practice ownership, patients were more likely to choose lower cost, higher quality hospitals closer to their homes.
• Accounting for practice ownership, patients of hospital-owned practices were more likely to use the owning hospital even when its costs were higher or its quality was lower than offered by other local hospitals.
• Patients’ preferences for hospitals located closer to home were weakened but not overridden when the influence of physician practice ownership was taken into account.

What These Findings Mean

• When a hospital owns a physician’s practice, patients cared for by that practice are dramatically more likely to choose the owning hospital over other local options when they need inpatient care.
• Moreover, the hospitals chosen by these patients are more likely to be high cost and low quality, suggesting that adverse consequences can arise from the integration of hospitals and physician practices.

More About This Study

The study sample consists of Medicare patients admitted to a hospital in 2009. Information about the ownership of these patients’ physicians’ practices was drawn from SK&A files, and data on hospital characteristics came from the American Hospital Association. The intensity of each hospital’s resource use (its costs) was derived from the Dartmouth Atlas, whereas a composite quality measure was compiled from the Medicare Hospital Compare files.

These data were used to estimate conditional logit models of the probability that a patient chose a particular hospital as a function of hospital characteristics, practice ownership variables, and interactions between the two sets of variables. Hospital characteristics included measures of its costs, quality and distance from the patient, facility size, for-profit/non-profit ownership status, system membership, and teaching status. Ownership variables reflected whether the hospital owned any practices at all, whether it owned the patient’s physician practice specifically, and whether the practice was owned by any hospital. Several alternative analyses were conducted to assess the sensitivity of results to possible endogeneity issues and omitted variable bias.