Medicare pay-for-performance dilemma: Who gets the bonus?

Washington -- Medicare's march toward implementing physician pay-for-performance may run into a brick wall when federal officials get down to the job of determining which doctors should be rewarded when things go well, according to a recent study.

The leading pay-for-performance model involves looking back at medical claims data to identify which physician is primarily responsible for a patient's care and then to measure the physician's performance based on predetermined quality measures, said Hoangmai H. Pham, MD, MPH, a senior health researcher at the Center for Studying Health System Change, a Washington, D.C., policy research organization.

In an attempt to determine how readily Medicare could handle this task, Dr. Pham and several colleagues selected about 8,600 physicians from one of the center's past surveys and analyzed their Medicare claims from 2000 to 2002 for roughly 1.8 million beneficiaries to "assign" each patient to individual doctors.

The study, which appeared in the March 15 New England Journal of Medicine, concluded that this process would not be easy for Medicare. The typical beneficiary during the course of a year saw two primary care physicians and five specialists working in four different practices. About one-third of the seniors changed their main doctor from one year to another.

Seniors with certain chronic diseases or multiple conditions had longer lists of physicians, with typical beneficiaries in some categories seeing 10 or more doctors in a given year.

Judgment based on fewer patients

Such "care dispersion" means that even primary care physicians would be held responsible for the care of less than 40% of their Medicare patients if the program retrospectively assigns each beneficiary to a doctor for pay-for-performance. The typical specialist provides more costly care but would be judged based on only about 12% of their Medicare patients.

Even if Medicare could implement a system that used claims data to connect patients with doctors for performance measurement, questions remain about how meaningful the resulting quality reports would be, Dr. Pham said. Just because a patient sees one doctor more than any other, for instance, doesn't automatically mean that physician is the one chiefly responsible for the patient's care.

Moreover, she said, targeted pay-for-performance incentives might be of limited use if physicians can't find out ahead of time which limited group of patients will form the basis of their evaluations. Doctors will not know for sure which patients Medicare eventually will deem the best candidates for care improvements.

"These retrospective methods of attribution are all about slicing and dicing Medicare claims data after the fact," she said. "We've shown that there is an arithmetically inherent reason why care dispersion will make this difficult."

New data for an old concern

The potential problem of deciding where credit is due when a patient receives high-quality care -- and who is held responsible when care does not meet set standards -- is not new for physicians and other stakeholders. Organizations such as the American Medical Association have said that performance measurement by Medicare or other health care payers must be accurate and fair to participating physicians. But the investigation provides new supporting evidence of the scope of the difficulty.

"The AMA is committed to always improving the quality of health care for patients, and, as we examine ways to improve, this study adds to the body of knowledge on the challenges of pay-for-performance implementation in Medicare," said AMA Board of Trustees Chair Cecil B. Wilson, MD.

If Medicare were to adopt a pay-for-performance plan that relied on retrospective assignment of beneficiaries to physicians, some doctors would be starting out with a significant disadvantage, said Richard O. Dolinar, MD, an endocrinologist in Phoenix. He predicted that the vast majority of his diabetic Medicare patients would be assigned to their primary care physician for performance measurement even though he directs a major element of their care.

"If that's how they're going to do it, the chances of my getting a bonus are pretty darn slim," he said.

In addition, by selecting a relatively small number of patients upon which to judge a physician's entire performance, Medicare would create an incentive for some doctors to identify potential problem patients who could skew their quality score results and to refer them to other physicians, Dr. Dolinar said.

"One of the questions the doc is going to be asking himself is, 'Will treating this patient help my statistics or hurt my statistics?' If they're going to hurt his statistics, he's going to move this patient out of his practice," he said.

Looking forward, not backward

Medicare pay-for-performance is a promising concept, but accomplishing it through a review of past claims data assumes that the typical beneficiary has a primary physician who is coordinating care with the patient's other doctors, Dr. Pham said. The wide dispersion of care found in the study and surveys of patients find that this level of coordination does not exist.

"Putting pay-for-performance into a world like that is a little bit like putting the cart before the horse," she said.

Bolstering system-wide care coordination levels before making the leap into reimbursement based on quality is a process that resonates with other health policy experts as well.

"The challenge is not just to design a pay-for-performance system that can rationally assign care to the various physicians involved in treating Medicare beneficiaries," said Karen Davis, PhD, president of the Commonwealth Fund, a private New York health care foundation. "It is also to change current practices to ensure a stable and responsible primary source of care and to improve care coordination."

One way that Medicare could associate performance measurement with care coordination is to assign patients prospectively rather than retrospectively, Drs. Pham and Davis said. By creating incentives for a beneficiary to choose a single physician or practice ahead of time to be his or her primary source of care, Medicare could better enable these doctors to coordinate care, while holding the chosen physicians accountable for the quality of care given to a predetermined set of patients.

Still, Dr. Dolinar said, many physicians already are working with each other to coordinate their Medicare patients' care. Simply looking through Medicare claims will not be enough to determine how widely this is occurring within the system.

In addition, prospectively assigning beneficiaries to physicians will limit the freedom of patients to vote with their feet on quality of care by switching to other doctors that will better serve their needs, he said.