Jump to:Page Content
Testimony by Amitabh Chandra to the US Civil Rights Commission.
June 12, 2009
[Under Oath] Mr. Chairman and Commissioners, my name is Amitabh Chandra, and I am a professor at Harvard University’s Kennedy School of Government and a fellow with the Dartmouth Institute for Health Policy. Thank you for inviting me to the Commission to share my thoughts on how to improve healthcare for minority patients.
We are all aware of the stubborn persistence of racial disparities in treatment over time, even when patients are fully insured. Many believe that the clinical encounter is the most pernicious source of these disparities. My main point this morning is that we are unlikely to make great strides in improving minority health by prioritizing action on this channel for the importance of the clinical encounter is dominated by other shortcomings such as the lack of access to high-quality providers, which are far more injurious to minority health.
That racial disparities in healthcare emanate principally from the clinical encounter embodies the idea that a provider treats two identical patients, one white and the other black, differently. More precisely, treatment differences in the clinical encounter may occur because there is explicit discrimination where a provider consciously withholds valuable care from minority patients. This is the most malfeasant explanation for disparities in care, and is perhaps one reason for why there is so much interest in this mechanism. But disparities may also arise from implicit discrimination, where a harried provider operating in a time-sensitive environment makes unconscious mental decisions that are detrimental to minorities. Stereotyping, is the one manifestation of this indiscretion and it occurs when a provider uses a patients race to deduce information about the benefit of treatment. If African Americans patients are on average less likely to be compliant, then a physician may assume that her African American patient is less compliant. Such reasoning will worsen outcomes for that patient if he is different from the typical African American, and worsen outcomes for all African Americans if the stereotype about them is wrong. Such biases are compounded by poor communication between providers and their patients, which may create enormous psychological barriers to minority patients seeking care. Finally, some researchers have posited genetic or physiological differences between patients that affect the benefit of treatment by race while others have discounted such conclusions.
Have we conclusively established the role of the clinical encounter in affecting racial disparities in healthcare? Answering this simple question carefully, poses a formidable empirical challenge: we would need to observe the same provider treating two patients with the same economic and social resources, physiology, clinical history, severity, preferences, compliance, and future prognosis. These variables are routinely observed by providers treating patients but not by social scientists observing providers. The fact that multiple studies all note that minority patients get less care is often interpreted as evidence of pervasive bias in the clinical encounter, but could just as well be interpreted as evidence of pervasive shortcomings in observational studies that focus on the clinical encounter. Second, because of patterns of neighborhood segregation, the same provider is rarely observed treating both black and white patients. And so what we have been calling prejudice in the clinical encounter is really a difference in neighborhoods, referral patterns, and the resources of providers that serve in these neighborhoods. This is an unfortunate confusion because improving neighborhood schools or changing referral patterns is not the same thing as reforming provider behavior inside hospitals and offices.
Researchers have made some progress on this challenge by using patient-actors and Implicit Association Tests (IATs) where physicians and the researchers studying them observe the same information in a laboratory setting. This is an intriguing area of academic research, but its findings are still nascent for the purpose of informing policy and legislation. We do not know if the decisions made by self-selected physicians in these laboratory studies are representative of physicians who actually take care of minority populations. Nor do we know if the policy of expanding the number of minority physicians will reduce the bias measured in these studies. In fact, one prominent study found that racial differences in treatment were similar among patients treated by white and black physicians. Indeed, we could even damage minority health further if these interventions alienate or displace the least prejudicial physicians—those who have devoted their lives to treating minority patients.
My main point today is to elaborate on a new explanation for racial disparities in care: that they are partially the consequence of differences in where minorities and whites receive care. If different providers treat blacks and whites, then one reason for racial disparities in care is not only who you are—your race, but also where you live. Both sources of disparities are injurious to minority healthcare. The first type of variation, which I call "within provider variation,” is the role of the clinical encounter. The second, “between provider variation,” relates less to race per se and more to geographical variations in the quality of treatment patterns of all patients. It contributes to racial disparities in treatment because minorities are more likely to be cared for by lower performing providers (some large academic centers are an exception to this statement, but the link between such centers and quality is by no means automatic). Differences in where minorities are treated have to do with factors such as insurance and lower socio-economic status, but historical patterns of discrimination and neighborhood segregation surely exacerbate this variation. Confronted with these realities, we should be very cautious in concluding that malfeasance and nonfeasance are the sole purview of the medical profession.
So what is the evidence in favor of the role of geography as a determinant of racial disparities in healthcare? Dr. Peter Bach and his colleagues has demonstrated that blacks and whites have different providers, and those providers who treat minorities are often less clinically trained and have fewer resources. My collaborators and I have demonstrated that eighty-five percent of all black heart-attack patients are treated by only 1000 hospitals, and 60 percent of whites receive their care in hospitals that treat no black heart-attack patients. Within hospitals we found no racial disparities in effective care, but that patients admitted to hospitals that disproportionately served blacks were found had a risk-adjusted mortality rate that was 20 percent higher than that of non-minority serving hospitals. Others have noted similar findings for the performance of NICUs in minority serving hospitals. Forty years after the passage of Civil Rights Act, minority healthcare is de facto separate and unequal. Ironically, a close cousin of this embarrassment, segregated hospitals, was the original motivation for Title VI legislation.
The new focus on the geography of minority healthcare should not be viewed as taking attention away from reforming the clinical encounter. Rather, it notes that even if we could fully eliminate disparities in the clinical encounter, the healthcare of blacks would improve, but still lag behind that of whites because of differences in the quality of where the two groups receive care. For many of us, this is simply not good enough. Because a small group of providers treat minority patients, targeting quality improvements towards minority serving providers will dramatically reduce black-white disparities in care more generally. Such interventions would improve the health of both minority and white patients, but the gains would disproportionately accrue to minority patients whose care is concentrated in such providers. In the context of ambulatory care for diabetes, my collaborators at Dartmouth and I estimate that aggressively improving the performance of the 500 largest minority serving networks would improve minority healthcare more than the complete elimination of racial disparities within every provider in the US. Indeed, given the greater reliance on ambulatory care, one may want to think about expanding the reach of Title VI of Civil Rights legislation to go beyond the reach of hospital care and encompass care that is delivered in office visits and by managed care plans.
Finally, in closing let me make one simple point: the determinants of racial disparities in health are not the same as the determinants of disparities in healthcare. The principal determinants of health are genes, behaviors, education, neighborhoods, economic circumstance, and genetics. Health is secondarily affected by healthcare, but more likely to be influenced by prevention and the quality of ambulatory care, which can check the progression of diabetes, hypertension and chronic disease, and through this protection, the incidence of heart attacks and strokes. Of tertiary importance, at the very end of the causal chain, is the role of disparities in the clinical encounter; the quality of that encounter matters much more than the disparity within it. So the 6.5 year racial gap in life-expectancy for men and the 4.5 year racial gap for women, which are surely larger when one accounts for the condition of that life, are unlikely to be affected by the focus on the clinical encounter. The preoccupation with treatment disparities in the endgame misses the fact that minority patients find themselves confronting the endgame sooner than anyone else.