Health care spending in the U.S. reached $3.2 trillion in 2013, which accounted for 17% of U.S. GDP. This is almost twice as much as the OECD average of 9%, yet health outcomes in the U.S. are not twice as good as in these other countries. Many studies have documented enormous geographic variation in spending within the U.S., while finding no clear relationship with quality of care and health outcomes. Yet surprisingly few studies have attempted to analyze how health care spending patterns vary across individual doctors, and more important, whether the practice patterns of individual doctors relate to their patients’ outcomes. A new study of Medicare hospitalizations in JAMA Internal Medicine found that individual physicians vary substantially in their health care spending, even within the same hospital, and that greater spending does not lead to improvement in patient outcomes.
Health care spending in the United States reached $3.2 trillion in 2013, which accounted for 17% of U.S. GDP. This is almost twice as much as the OECD average of 9%, yet health outcomes in the U.S. are not twice as good as in these other countries. In fact, many outcomes are worse. For example, life expectancy at birth in the U.S. is 78.8 years, which falls short of the OECD average of 80.5 years.
Health care spending also varies substantially within the United States. Many studies have documented enormous geographic variation in spending, finding no clear relationship with quality of care and health outcomes. While some differences in spending and patient outcomes are due to factors outside the health care system, this evidence suggests that there is considerable waste in U.S. health care spending. Many have concluded that at least 20% of spending could be reduced without harming patients.
Geographic regions, however, do not make health care treatment decisions; hospitals, doctors, and patients do. Yet surprisingly few studies have attempted to analyze how health care spending patterns vary for individual doctors, and more important, whether the practice patterns of doctors relate to their patients’ outcomes. Understanding how practice patterns differ among doctors and whether higher-spending doctors have better outcomes is critically important for finding ways to reduce health care costs and improve efficiency of care without harming patients.
In a study recently published in JAMA Internal Medicine, we investigated how spending varies among individual doctors and how spending relates to patient outcomes. We found that individual physicians vary substantially in their health care spending, even within the same hospital, and that greater spending does not lead to improvement in patient outcomes.
We used a 20% random sample of nationally representative data on Medicare patients who were hospitalized with a general medical condition and treated by a hospitalist physician, a general internist who specializes in the care of hospitalized patients, at some point between 2011 and 2014. We later studied patients treated by general internists as well.
We focused on hospitalists in our main analysis for a specific reason. Like emergency room doctors, hospitalists typically work in scheduled shifts, so they are unlikely to select their patients, and the patients are unlikely to select them. This allowed us to circumvent the problem of “selection bias,” namely that physicians who spend more have worse patient outcomes simply because they treat sicker patients, who are more expensive to treat. Because the patients in our study were, in effect, randomly assigned to hospitalists with varying spending patterns, we were able to determine that some hospitalists have a tendency to order more procedures, tests, and imaging, thereby spending more than others.
Our final sample consisted of approximately 500,000 hospitalizations, treated by 20,000 hospitalists, in 3,000 hospitals, across the country. We found that even when similar patients were treated, hospitals varied dramatically in their health care spending, but doctors within those hospitals varied even more. The differences in health care spending between individual hospitalists practicing within the same hospital were larger than the differences observed across hospitals.
Specifically, 8.4% of the variation in patient health care spending could be explained by differences between individual doctors, whereas 7.0% could be explained by differences in hospitals. (Of course, most of the variation in spending was explained by patient characteristics, but even after adjusting for those, the variation in spending exists due to the differences in practice patterns between physicians.) Within the same hospital, the highest-spending doctors (those in the top quartile) spent 40% more than the lowest-spending doctors (those in the bottom quartile) for similar patients.
We then examined whether higher-spending physicians have better patient outcomes than lower-spending physicians within the same hospital. Although our study design relied on the quasi-randomization of patients to hospitalist physicians, we adjusted for patient characteristics and physician characteristics that could influence spending and our two patient outcomes, mortality and readmissions. Patient characteristics included patient age, sex, race or ethnicity, primary diagnosis, nearly 30 coexisting chronic conditions, median household income, and whether patients were covered by Medicaid. Physician characteristics included age, sex, medical school, and hospital.
We found that higher-spending physicians did not have lower patient mortality or readmission rates than lower-spending physicians. When we extended our analysis to general internists, we found the same results.
What Do These Findings Mean?
Within the same hospital, physicians vary substantially with respect to how much they spend on patients, and higher spending doesn’t lead to better patient outcomes. These results suggest that policies to improve the efficiency of health care should not only focus on hospitals, as has historically been the case; they should also take doctors into account.
For example, the Hospital Value-Based Purchasing program, at the Centers for Medicare & Medicaid Services, which compensates hospitals on the basis of achieving certain quality and resource use targets, may be more effective if individual physicians are targeted as well. A large step forward will be the Medicare Access and CHIP Reauthorization Act, which went into effect in 2017, requiring most physicians to be measured and reimbursed on the basis of the quality and costs of care that they provide.
It is important to note that we could not assess why some physicians spent more than their colleagues in the same hospital. So it is too soon to conclude that physician-focused health care reforms can improve the efficiency of health care spending without compromising patients’ health. Understanding the reasons behind greater spending is incredibly important.
If some physicians spend more to compensate for lower clinical skill or lower comfort with the uncertainty of clinical decision making, then incentivizing these physicians to utilize fewer resources could worsen patient outcomes. In contrast, if higher-spending physicians simply spend more because they do not directly face the costs of ordering an additional test or procedure, it is possible that they could spend less without adversely affecting patients.
Ultimately, we need to experiment with physician-focused strategies for improving health care efficiency. Further studies are warranted to understand why some physicians spend more health care dollars than others, and to reduce health care services that are not creating value.