Mending Medicare

CMS' physician reimbursement system is ailing, but many fear the cure could hurt patients.

Medicare's formula for paying health care providers is in trouble. Costs are up, but doctors' pay threatens to go down. Many in Congress and the Bush administration say the cure is pay for performance, which would hold doctors accountable for the efficiency and effectiveness of their care. But doctors are holding out for a second opinion.

Each year, the Health and Human Services Department's Centers for Medicare and Medi- caid Services processes more than a billion claims for patient services from a million doctors, hospitals and nursing homes. Medical costs keep going up, but that doesn't mean "the quality is keeping up with that," John Dyer, CMS' chief operating officer, told an industry audience in November.

Medicare doctors' expenses will rise by 15 percent over the next six years, according to the Medicare trustees' latest annual report. But the formula that determines physicians' pay is set to reduce fee reimbursements by about 5 percent a year through 2011, unless Congress overrides it.

No clear connection exists between dollars spent on treating an illness and the ultimate outcome, says CMS Administrator Mark McClellan. In the past, when Medicare cut reimbursements, medi-cal practices often compensated by running up expenses, according to CMS officials. Today's payment methodology can even discourage good performance because Medicare reimbursements often are higher "when a serious illness or injury occurs or recurs," states a March 2005 report from the independent Medicare Payment Advisory Commission, which endorses pay for performance. Tying medical efficiency to doctors' bottom line would improve care and control spiraling costs, supporters say.

Bills in the House and Senate would stop the practice of paying all Medicare care providers equally, regardless of which treatments they use. With administration and congressional support for pay for performance, CMS already is laying the groundwork. In October, the agency announced a pilot program for physicians willing to participate in a pay-for-performance data collection effort. By January, the agency will begin to collect data on 36 clinical measures.

Dr. Richard O. Dolinar, a clinical endocrinologist in Phoenix, will not be among the volunteers. Pay for performance would have serious consequences for doctors and patients, he says. As a senior fellow for health care policy at the Heartland Institute, a Chicago-based think tank, Dolinar is no fan of Medicare's current system, which he says is based on price controls and central planning. But he's worried that pay for performance would result in doctors playing hot potato with particularly risky patients. A case "that's going to wreck my statistic-I'm going to ship this patient downstream," is how Dolinar predicts doctors would react. In fact, a study published in the July 25, 1996, New England Journal of Medicine found a majority of Pennsylvania cardiologists who were graded annually in a public report card became less willing to operate on severely ill patients.

CMS' Dyer says pay for performance won't harm patient care. The point is to encourage effective therapies, such as an arteriovenous fistula for dialysis as opposed to a venous catheter-therapies that might cost more initially but are better in the long term. "Let's see if we can't move to a financing reimbursement system where we pay a little more, but the other things that aren't as effective, aren't as efficient, we pay a lot less for," Dyer says.

How CMS plans to measure physician performance is key, says Dr. John Armstrong, a Miami surgeon and an American Medical Association trustee. The pilot metrics aren't a good sign, according to Armstrong. In a letter to McClellan in November, the association charged CMS with bypassing its standard collaborative process for establishing clinical measurements.

But Dolinar rejects the possibility that any set of measurements could accurately capture physician performance. Human biology is complex and curing people cannot be reduced to a set of preordained steps, he warns. For example, among the 36 CMS metrics is the level of hemoglobin A1c in diabetic patients, a measure of blood glucose. Generally, the lower the level the better. But sometimes Type 1 diabetics should err on the side of keeping their glucose a little high, Dolinar says. "You can have artificially low hemoglobin A1c . . . [but] in reality the patient is worse off because they're having hypoglycemic reactions," he says.

Armstrong also finds fault with the metrics. One measures whether doctors administer beta blockers to heart attack patients in order to reduce the heart rate and muscle contractions. But other pre-existing conditions such as chronic obstructive pulmonary disease could be aggravated by beta blockers. "There's nothing in guidelines and performance measures that's going to account for this," he says.

Besides, the infrastructure to collect performance metrics doesn't yet exist, skeptics say. CMS will use the existing administrative claims system to start tracking performance data, adding another layer of complexity to the billing process. Adoption of electronic health records would facilitate performance reporting, McClellan says.

Electronic records aren't just preferable but necessary, Armstrong says. Even if CMS institutes a clinically valid set of measurements, the data must come from medical charts, not billing documents, he says. Extracting data manually from medical charts is possible but not tenable with more than five or 10 metrics, he adds. The AMA letter to McClellan charges that the administrative burden outweighs any intended improvement in care.

"There are voices in Washington that want to launch pay for performance very soon, too soon," Armstrong says. "They are forgetting about the consequences of this for the patients and for the physicians who serve them."