Pay for Performance Not Ready for Prime TimeBy M.L. Baker
The premise of pay for performance is hard to disagree with: Doctors and institutions that provide better care should earn more money. This premise, however, is hampered by two facts: No one knows the best way to implement such a program, and if different health care payers implement different programs, the effects will be diluted.
Over 100 pay-for-performance efforts have been launched so far in the United States, according to congressional testimony given last week by Meredith Rosenthal, a professor of health economics and policy at the Harvard School of Public Health.
Most programs use simple measures of performance: Providers are financially rewarded for achieving goals in, say, childhood immunization rates, the percent of patients receiving aspirin after undergoing coronary bypass surgery, or the length of time required to administer antibiotics to hospitalized pneumonia patients. Last year, CMS (Centers for Medicare and Medicaid Services) launched a program that will reimburse hospitals up to 2 percent more than standard rates, according to their performance of 34 such measures. Early results from a study by CMS show that hospitals are improving in these areas.
"Quality has crossed over into the CFO's office," said Wes Champion, vice president of Capgemini Health's business strategy and transformation practice. "There is a good creative tension between the payer and provider because at the heart of this pay for performance is a better quality of care delivered to the patient." Champion thinks pay for performance, if implemented properly, could be a win-win-win situation for patients, payers and providers.
But if programs don't measure what matters, they could actually squeeze out other safety practices, said Richard Ward, CEO of Reward Health Sciences, a consultancy specializing in IT and care management. Ward told me that he likes that concept of measuring improvement, but that he's seen health care providers abandon more worthwhile goals to boost rates of less important services: "The more you standardize, the more people will pursue superficial means [to hit measurement targets]."
Worse, providers might actually shun patients who are less healthy or less likely to comply with doctors' orders.
Rosenthal pointed out that little evidence shows that pay-for-performance programs actually improve care. In fact, one study showing improved childhood vaccination noted that the improvement was not in vaccination rates but rather documentation of them. And if implementing pay-for-performance measures costs more than bonuses, she said, they won't work. Of course, that doesn't mean pay-for-performance programs can't work, just that they haven't been studied well or long enough to conclusively prove themselves.
Another problem is that physicians tend to be reimbursed by an array of entities, who may assess performance in different ways. Cap Gemini's Champion said that EMR (electronic medical records) could improve this situation: "If quality measures can be pulled from an EMR, it will enhance for that consistency." In fact, pay-for-performance programs could help build a business case for EMR, he said.
EMRs could also, eventually, be used to develop more accurate measures of what clinicians can do to keep costs down and improve patients' health. That, in turn, will make pay-for-performance plans more effective because health care payers will be willing to make them more rewarding.
However, pay-for-performance programs are not going to perform the magic trick so many hope for. The goals of patients, providers, those who pay for health care and those who pay for health insurance are fundamentally misaligned in our health care system.
It will take more than complicated new ways of reimbursing doctors and hospitals to make everyone pull together, but perhaps pay-for-performance programs, and the ability to monitor them, will inspire something akin to real magic.
What do you think? Can health IT inspire a better payment system?
M.L. Bakeris health IT and biotechnology editor for Ziff Davis Internet's Enterprise Edit group.