SAN DIEGO Pay-for-performance programs must be carefully designed to avoid putting some of the most vulnerable patient populations at risk, officials with the American College of Physicians warned at the organization's annual meeting.

Although pay for performance has the potential to improve medical care, it could also endanger the physician-patient relationship, the financial stability of the health care system, and the elderly and the chronically ill, said Dr. Frederick E. Turton, chair of ACP's Ethics, Professionalism and Human Rights Committee.

To this end, ACP is preparing to publish a position paper on the issue of ethics in pay for performance. The paper, "Ethics Manifesto: Pay for Performance Principles that Ensure the Promotion of Patient Centered Care," focuses on what pay-for-performance programs should accomplish, what physicians should do if participating in these programs, and the potential unintended consequences of these incentive programs.

"We already have one system that is broken," Dr. Turton said during a panel session on the topic. "We don't want pay for performance to initiate yet another broken system."

Pay-for-performance programs should be designed to promote evidence-based care, encourage collaboration among providers, support patient autonomy, protect patient privacy, and include full disclosure of financial incentives. A well-designed program also should address the comprehensive needs of patients, not single-disease states, according to the ACP position paper.

For example, ACP officials are concerned about programs that base their incentives on meeting strict clinical targets, such as a specific hemoglobin A_1c level, because that might prompt physicians to select patients based on their ability to meet that target. Instead, programs that focus on improvement on a measure might be more appropriate, Dr. Turton said at a press briefing during the meeting.

For their part, physicians should be aware of the potential influences on their clinical judgment and strive to avoid discrimination. And physicians need to put medical considerations ahead of both their own and the payer's financial interests, Dr. Turton said.

Some of the unintended consequences highlighted by ACP in its upcoming ethics paper include the potential deselection of patients, gaming of the system by physicians, and an increase in unnecessary care and costs.

Pay-for-performance programs also have the potential to encourage physicians to perform to the measure, rather than thoughtfully evaluating the individual needs of the patients, Dr. Alan R. Nelson, a member of the Institute of Medicine's study committee on pay for performance. And quality measures may not lead to reductions in cost, he said. In the short term, in fact, pay for performance will probably increase utilization of services and cost, he said during the panel session.

Exploring the ethical implications of pay-for-performance programs is new territory, according to Dr. Matthew K. Wynia, director of the Institute for Ethics of the American Medical Association.

Limited data are available about pay-for- performance ethical concerns, in part because these programs are so new and researchers need more time to study their effects, he said. The programs are also variable, complex, and are often proprietary and confidential, making them hard to study. And pay for performance is generally not well understood by either patients or physicians at this point, Dr. Wynia said.

The limited information in the literature on pay-for-performance and public reporting programs has provided mixed results on the question of whether pay for performance will simply reward those who are already high performers.

For example, one study compared the performance of California physicians who were enrolled in a pay-for-performance program with the performance of physicians in the Pacific Northwest who were not enrolled. The study assessed outcomes on cervical cancer screening, mammography, and hemoglobin A_1c testing and found that the California physicians achieved greater quality improvement only in the area of cervical cancer screening. The researchers concluded that there was little gain in quality, and that the financial rewards were given mainly to those who had a higher performance at baseline (JAMA 2005;294:178893).

However, in another study, 207 hospitals involved in a Medicare-sponsored pay-for-performance demonstration showed greater improvement in a composite of 10 quality measures, compared with 406 hospitals involved in voluntary public reporting only. And among the pay-for-performance hospitals, those that had the worst baseline quality performance improved the most (16.1%), while those with the highest baseline quality improved the least (1.9%) across the measures (N. Engl. J. Med. 2007;356:48696).

There are data on both sides of this, Dr. Wynia said.

A recent study also calls into question how a pay-for-performance program under Medicare could reliably assign responsibility for a patient's care. For example, an analysis of Medicare claims from 2000 to 2002 among 1.79 million fee-for-service Medicare beneficiaries showed that, on average, beneficiaries saw two primary care physicians and five specialists across four practices. And about a third of Medicare patients also switched assigned physicians each year (N. Engl. J. Med. 2007;356:11309).

In light of these results, it could be difficult to assign rewards for care, Dr. Wynia said.

Pay for Performance Stirs Ethical Concerns