Embargoed Until:
April 10, 2007
12:01 a.m. Eastern Time



Christopher Fleming

Pay-For Performance And Public Quality Reporting:
 A Boon For Most, But A Potential Burden For The Most Vulnerable Patients

In Health Affairs Article, Researchers Examine How Quality Incentives For Physicians Could Hurt The Poor And Minorities,And What To Do About It

Bethesda, MD -- Pay-for-performance (P4P) and public quality-reporting programs offer the potential to increase the quality of health care overall, but they threaten to actually decrease quality for minority and low-income patients in the process. In an article published today on the Health Affairs Web site, Larry Casalino and Arthur Elster explain, first, how current incarnations of P4P and reporting programs for physicians could worsen health care quality disparities and, second, how these rapidly proliferating “external incentives” might be revamped to avoid this result.

“Programs designed to minimize the unintended consequences of increasing disparities will be more costly and time-consuming to implement, maintain, and modify over time than those that ignore the possibility of such consequences. But they should have more staying power and should be less likely to generate a backlash from physicians and patients,” say Casalino, an assistant professor of health studies at the University of Chicago, and Elster, director of the Division of Medicine and Public Health at the American Medical Association. Their article focuses on P4P and public reporting programs for physicians, but they say that similar points apply to external incentives aimed at other types of health care providers, such as hospitals.

Casalino and Elster offer several ways in which P4P and public reporting could actually worsen care for the poor and minorities. Physicians might avoid poor and minority patients if they perceive them, rightly or wrongly, as less likely to have good outcomes from treatment, or less likely to adhere to treatment recommendations. Physicians might also “teach to the test” more with poor and minority patients than with more affluent and nonminority patients. “For example, with a relatively uneducated diabetic patient who speaks poor English, the physician might focus on making sure the patient has a hemoglobin A1c test (because this is measured) but not on the time-consuming” -- and unmeasured -- “task of explaining to the patient how to control his or her diabetes and blood pressure,” Casalino and Elster write.

Residents of poor and minority neighborhoods could also end up paying more for medical care if health plans charge higher copays to visit “poor-quality” physicians. Physicians in these neighborhoods are “doubly disadvantaged” in achieving high quality scores, Casalino and Elster explain. Their many Medicaid and uninsured patients leave them little revenue to invest in quality improvement, and their patients may be less likely to obtain recommended follow-up treatment and preventive care because of child care or transportation problems or because of failure to understand the recommendations.

Other physician responses to P4P and public reporting could increase quality for affluent and nonminority patients, while leaving care for the poor and minorities unchanged or improved to a significantly lesser extent. For instance, poor and minority patients might be less likely to participate in quality improvement initiatives such as disease management programs if they are presented at literacy levels or in cultural styles that these patients find confusing or insensitive. Affluent and educated patients might also be better able to understand quality report cards and better able to act on them, either because they live in areas with high quality providers or because they have the money and work-schedule flexibility to travel to more distant physicians.

Design Elements For External Quality Incentives
That Will Reduce Disparities, Not Increase Them

Casalino and Elster offer several strategies for increasing the likelihood that P4P and public reporting programs improve quality for poor and minority patients:

Risk-adjust quality scores for health status, and for race/ethnicity and/or socioeconomic status. Contrary to what many argue, risk adjustment must be used for both outcome and process measures. The authors write, “It should be easier for physicians who practice in affluent areas such as Marin County, California, than for those who practice in poor areas of Oakland to achieve higher rates of screening mammography. Their patients are more likely to be wealthy, well-educated, well-insured women who are aware of the benefits of mammograms, are more likely to request them, and are more able to travel to obtain them.”

Use stratified analysis. “Stratified analysis” could be used to compare physicians’ performance against that of physicians treating similar patients. “For example, physicians practicing in poor minority areas could be compared -- for purposes of P4P or public reporting, or both -- with other physicians practicing in similar areas.”

While risk adjustment only reduces the incentive for physicians to avoid perceived high-risk patients, stratified analysis can provide positive incentives to treat such patients well, and it might provide useful information for future quality improvement efforts directed at specific patient groups. However, “unlike risk adjustment, which can be done for each individual patient, reliable and valid stratified analyses of quality require that a physician or a medical group have a large number of, for example, African American patients with diabetes. Most do not, so stratified analyses would be reserved for large medical groups, hospitals, and health plans -- and even some of these organizations might not have enough minority patients to permit stratified analyses.”

Casalino and Elster point out that both risk-adjustment and stratified analysis can be “two-edged swords”: They give physicians more incentive to treat patients who might otherwise be perceived as likely to lower their quality scores, but – by rewarding physicians even when their quality scores for minority patients were lower than for other patients – they also “risk rewarding them for continuing to provide mediocre care.” To combat this, Casalino and Elster suggest scoring physicians both on overall quality and on improvement over time; improvement would be measured for an organization’s overall patient population and -- when patient numbers permit reliable measurements -- for minority and poor patients alone.

Combat “teaching to the test.” Rotating the quality measures used, adding new measures frequently, and/or including data from patient satisfaction surveys “could counterbalance incentives to focus on narrow quality measures at the expense of communicating with patients and coordinating patient care,” Casalino and Elster write. More ambitiously, physicians or medical groups that score highly on announced quality measures could be evaluated on a second group of measures that would not be announced in advance; the rewards these physicians received for scoring highly on the initial measures would then be increased or decreased, depending on how well they did on the second round of measures.

You can read Casalino and Elster’s article at


Health Affairs, published by Project HOPE, is the leading journal of health policy. The peer-reviewed journal appears bimonthly in print with additional online-only papers published weekly as Health Affairs Web Exclusives at www.healthaffairs.org.


©2007 Project HOPE–The People-to-People Health Foundation, Inc.