Physicians and teachers are the focus of a major societal effort to reward excellence, encourage improvement, and weed out poor performers. The intent of these efforts is noble and good, but they also entail major risks.
Since biblical times, many health professionals have been drawn to caring for disadvantaged, vulnerable populations. I see this in applicants for the primary care residency program for which I am part of the faculty. These young physicians are among the best, the brightest, and the most idealistic.
But a recent article in JAMA - "Relationship Between Patient Panel Characteristics and Primary Care Physician Clinical Performance Rankings" - suggests that they may be heading for trouble!
A research team studied 162 primary care physicians in the Massachusetts General Hospital system. All were hired and credentialed by similar criteria. They shared the same compensation plan, had similar staffing resources and the same advanced electronic medical record system. This was an experienced group, averaging almost 19 years from medical school graduation.
The researchers created a composite quality score based on 9 HEDIS measures commonly used for quality assessment. On the basis of these measures, they grouped the physicians into thirds. Then they adjusted their results for patient variables - age, sex, number of comorbidities, race/ethnicity, primary language spoken, and insurance status. When these adjustments were made, 6 in 10 physicians changed more than 5 percentiles and one third change more than 10. 14.3% of the bottom third increased in ranking to the middle third. 25% of the middle third moved into the top or bottom category. The 34 primary care physicians whose quality rankings increased by more than 10 percentiles were more likely to be practicing at community centers, with larger panels, a higher proportion of minority, non-English speakers, and more who were uninsured or insured through Medicaid.
A quality ranking system that did not adjust for patient variables would have penalized PCPs who work with a poor, vulnerable population. If quality rankings drove differences of income, the system could worsen health disparities by diverting resources away from patients with greatest need and rewarding physicians for avoiding these patients.
In evaluating teachers, systems have been developed that take student vulnerability into account. If a district or state looks at all students at the same level of vulnerability, it can compare teachers in relation to the same student "inputs." That's the meaningful comparison, rather than comparing teachers in the inner city or poor rural areas to teachers in the wealthiest suburban systems.
It's vital to measure our performance in health care and work to do the best that can be done. But if we apply physician ratings without careful attention to patient panel characteristics, we'll prove once again that no good deed will go unpunished!