Quality Rounds
Performance and Patients
Patient characteristics such as poverty, age, and ethnicity are not stopping some clinics from achieving good outcomes.
By Scott D. Smith
One of the often-used arguments against report cards and the pay-for-performance initiatives based on them is that they favor physicians who serve certain types of patients—the assumption being that younger, more affluent, and white patients will have better outcomes than older, poorer, and minority ones.
But as the reports on performance come in, some long-held assumptions about who’s a tough patient and who isn’t are coming into question.
Barry Bershow, M.D., director of quality and informatics for Fairview Health Services, offers one clinic in that system as a case in point. He says a number of Fairview doctors had suggested that a heavy case load of elderly patients would hurt their diabetes scores and incentive eligibility. When Fairview looked at the numbers, it found that the clinic’s patients on Medicare actually were doing better than its younger diabetics. “It blew that theory right out of the water,” Bershow says.
It turns out that the Bloomington Oxboro Clinic, which serves a large Medicare population, was the best in the Fairview system at managing diabetes during 2005. Thirty percent of the clinic’s diabetic patients had all of the following: an A1c below 7 (measured at least every six months), LDL cholesterol under 100 during the previous 12 months, and a blood pressure reading under 130/80 at the last visit. Bershow says the key to the results was the fact that the clinic had a core group of providers who were committed to doing well on these measures.
Bershow says Fairview also found that its clinics that serve a higher percentage of poor, urban patients are not necessarily losing the quality race to clinics in more affluent suburbs.
In fact, in some cases, the opposite is happening. Two Minneapolis clinics, Fairview’s Riverside Women’s Clinic and Staub Pediatrics (now Fairview Children’s Clinic), which both serve large Somali populations, did better than clinics in many of the affluent suburbs Fairview serves at treating asthma and screening for Chlamydia.
“There does not seem to be a population that is tougher to treat,” Bershow says. “M.D. buy-in to the initiative and patterns of care may play a bigger role.”
If that is the case, it would contradict a preponderance of studies that show poor and minority populations do have significantly worse health outcomes, according to David Satin, M.D., a family physician at Smiley’s Clinic in Minneapolis. Satin points out that the National Institutes of Health has made health care disparities a research priority and that the public health community has for decades accepted a correlation between poverty and poor health.
Adjusting Measures?
Many physicians continue to believe that determining which patients are most difficult to treat is important because it will ultimately affect their compensation and reputation. For performance measures to be fair, they say, the yardstick needs to be risk-adjusted or handicapped based on the difficulty of a physician’s case load.
An example that vividly illustrates what can happen when measures are not risk-adjusted occurred in 1986, when the Health Care Financing Administration released a hospital mortality report that included one hospital with a death rate of about 88 percent. It turned out that the hospital was a hospice full of terminally ill patients.
Proponents of adjusting measures based on patient population say it can be done and often point to the state of New York, whose health department considers the age and health status of the patient in its reporting of outcomes of cardiac bypass and valve surgeries, as an example of how it can work.
New York uses about 40 clinical and demographic factors to create a cardiac-risk profile for each patient. The system accounts for the fact that an 80-year-old patient who had a heart attack within the past six hours has a risk profile that’s very different from that of a 40-year-old who has never suffered a heart attack. It then compares a surgeon’s actual mortality rate to a risk-adjusted mortality rate, and arrives at an estimate of what the provider’s mortality rate would have been if the provider had a mix of patients similar to the statewide mix.
Satin, who has written on the topic of pay for performance, says report cards that don’t account for the socioeconomic status of patients are asking doctors who work with the poor to do more than doctors who work with middle- and upper-income patients.
He cites the example of his own care of diabetic homeless patients, noting that a key step toward getting them healthy is finding them a place to live, which makes it easier for them to check their blood sugar and eat right.
“The bottom line is that it is simply easier to care for the patient if they already have a home. So both on a level of fairness and in regard to results, the doctor who takes care of the poor patient is starting 20 yards behind the start line.”
Satin suggests that adjusting for patient enrollment in government-funded insurance programs could be one way of leveling the playing field when using outcomes as a measure in pay-for-performance programs.
And he holds up both Britain and New Zealand as examples to emulate. Britain adjusts performance goals based on the economic status of the clinic’s postal code, and New Zealand has lower goals for practices with higher proportions of aboriginal patients.
Satin is especially concerned that pay-for-performance programs based on unadjusted measures will result in a reverse Robin Hood scenario in which programs will reward clinics that treat middle- and upper-income patients and not those that treat the poor.
But right now, those who are leading measurement programs in Minnesota do not believe risk adjustment would have much impact on medical groups’ results.
Jim Chase, executive director of MN Community Measurement, a nonprofit that issues annual performance reports on medical groups, says the percentage of a clinic’s patients enrolled in public programs is not a strong predictor of whether that clinic will have good or bad results. For most measures, the difference between patients with private insurance and those covered through public programs is small compared with the differences between the medical groups with the best and worst scores (see figures).
Consider the percent of diabetics older than 40 who take aspirin daily (Figure 1). About 56 percent of Medicaid patients met this goal compared with about 60 percent of commercially insured patients, according to 2005 MN Community Measurement data. That 4 percentage point gap pales in comparison with the 64 percentage point gap between Columbia Park Medical Group, the best-performing group with an 89 percent compliance rate, and Avera Health/Tri-State Health, the worst-performing group, which had 25 percent compliance.
In general, the commercially insured populations do better than the Medicaid populations in MN Community Measurement reports, although Medicaid patients did as well or better than those with private insurance when it came to blood pressure control and Chlamydia screenings.
But those advocating for risk adjustment could say the 2005 MN Community Measurement data bolsters their case because it showed that patients with private insurance performed better than Medicaid patients on four of the five measures for diabetes care (see Figure 1), and that patients on Medicaid had much worse results for medication maintenance for depression, well-child visits, and mammography screening (see Figure 2).
Equal-Opportunity Problem
Bershow acknowledges the fact that some studies show disparities in care, but he says the problem with basing risk-adjustment schemes on those studies is that the medical profession still doesn’t know why the disparities exist.
And he and others say they’re less worried that difficult patients will blow a clinic’s performance score and more concerned that scores in general are too low. “Physicians as a whole are so far off our diabetes targets that we don’t have to worry about risk adjustment at this time,” Bershow says. “We also don’t want to buy into the concept that poor patients should be satisfied with a lower level of care.”
A RAND Corporation study published in the New England Journal of Medicine in March that evaluated the relationship between sociodemographics and quality of care supports the notion that physicians are missing their care targets in general.
The study found that although disparities in care between different groups of patients exist, those gaps are small compared with the gap between the amount of preventive care all patients need and what they actually receive. The study, which measured preventive services and care for 30 acute and chronic conditions, found that patients only received recommended care about 55 percent of the time.
“This tells us that the U.S. health care system is unreliable and cannot guarantee that patients—rich or poor, white or black, insured or uninsured—will receive the right care at the right time,” lead researcher Steven Asch, M.D., said in a press release.
Going forward, quality leaders will likely need to factor in patient characteristics as they attempt to measure—and pay for—performance, Bershow says.
But right now, more work needs to be done to identify demographic or other factors that can predict patient results. “We don’t, at this stage of the science of difficult patients, really know who they are,” Bershow says. MM
Scott Smith is a staff writer for Minnesota Medicine.