Published June 9, 2022
Something strange is happening at Boston’s Brigham and Women’s hospital. In late January, protesters picketed the facility to denounce efforts by Drs. Michelle Morse and Bram Wispelwey to, as the protesters put it, establish “preferential health care treatment policies for non-White patients.”
The small and relatively benign demonstration attracted a surprising amount of attention. This in part explained by the prestige of Brigham and Women’s, a teaching hospital affiliated with Harvard Medical School. Morse and Wispelwey both hold prominent positions there; among other things, Morse is also the first-ever chief medical officer for New York City’s health department.
Still, the reaction seemed outsized: Over 800 doctors and dozens of medical groups signed an open letter expressing solidarity with Morse and Wispelwey, and the New England Journal of Medicine (NEJM) published an opinion piece by three Chicago-based researchers that condemned the picketers—at least one of whom was traced to a white-supremacist group—while praising Morse and Wispelwey’s push for “race-explicit programs [to] directly address the underlying causes of racial health inequities and mitigate the current impact of racism in medicine.”
What are these race-explicit programs being championed by Morse and Wispelwey, both of whom hold prestigious positions at a Harvard-affiliated teaching hospital? And why did a handful of low-key demonstrators—there appeared to be fewer than two dozen in all—inspire so much outrage?
Brigham and Women’s website features a pilot initiative to address racial “disparities in heart failure admissions.” The project, which is part of a broader United Against Racism campaign being promoted by the Massachusetts General hospital system, does not appear especially dramatic. (Bear with us; the program itself is a bit boring.) When doctors at Brigham and Women’s refer black or Latino heart-failure patients to the hospital’s general medical service, rather than to the hospital’s specialty cardiology wing, they will be automatically prompted to reconsider that decision. According to the website, “The prompt will educate the clinician about the fact that, historically, this population has had inequitable access to specialized cardiology care and offer a recommendation to consider changing admission to the cardiology service. The notice does not restrict clinicians’ individual judgment and decision-making in consideration of the best interests of the patient.”
While a scheme to automate prompts based on the racial identity of patients might seem offensive or simply annoying, it may not be especially harmful in itself. Physicians who are required to use electronic health records in their daily work receive countless distracting notifications. The faulty research being used to justify the program is worth scrutinizing, however. It was conducted by Morse and Wispelwey and has been widely celebrated but, it seems, never closely examined. It is important to understand the reasons why.
What Morse and Wispelwey Got Wrong
Brigham and Women launched this pilot program in response to a 2019 study by Morse and Wispelwey, which appeared in the journal Circulation: Heart Failure and which claimed to demonstrate “the presence of structural racism in admission service for heart failure patients.” Their paper now drives much of the “anti-racism” work being pursued at medical institutions nationwide.
In the study, Morse and Wispelwey sought to compare the historical frequency with which black, Latino, and white heart-failure patients admitted to Brigham and Women’s emergency room have been referred to the hospital’s specialty cardiology service, rather than to the general medicine service. Using retrospective data from September 2008 to November 2017, Morse and Wispelwey found that 67% of white heart-failure patients had been admitted to the cardiology unit as compared to 53% of both black and Latino heart-failure patients. They cite this disparity as it if it constitutes an obvious problem in itself.
Their own study’s findings do not indicate a problem, certainly not an obvious one. But because other studies had indicated that, in general, treatment in the cardiology unit was associated with lower re-admission rates to the hospital within 30 days, Morse and Wispelwey concluded that black and Latino patients at Brigham and Women’s had been treated poorly in comparison with white patients. They linked this conclusion to their explicitly stated belief that black and Latino patients are victims of structural racism at Brigham and Women’s—just one example of a nationwide problem that, they contend, is “a major impediment to achieving health equity.”
The study suffers from several fundamental flaws. The first is structural. Retrospective studies are inherently less rigorous than prospective studies; when researchers look to the past to try to organize raw data that has not been collected for the purposes of a carefully designed study, they are much more likely to get something wrong.
This is because retrospective studies involve information that has been roughly accumulated over the course of daily life and later scrutinized for clues; it is not gathered under controlled conditions with specific goals in mind. Retrospective researchers get to choose the data they examine in a way that prospective researchers cannot. Retrospective researchers also have the opportunity to try out various statistical techniques in an attempt to arrive at a hoped-for result.
Prospective studies, by contrast, are carefully designed from the outset to prevent unconscious biases by researchers from distorting the results. The best studies pose well-defined, limited, exploratory hypotheses, which are then tested within tightly controlled frameworks and with statistical techniques that are specified in advance. When information is gathered in this way, researchers are far less likely to allow their personal prejudices to cloud their findings, and so they can have more confidence in their conclusions.
This is important because when scientists undertake any sort of study, they are trying—and hoping—to prove something. There is an entire discipline devoted to illuminating and reducing the influence of unconscious biases in research findings. In retrospective studies, there is a far greater likelihood that researchers will “see” their sought-after results in the data they’ve chosen to evaluate. For all these reasons, retrospective studies are virtually always considered inferior to prospective studies. They are most valuable as a means to generate hypotheses, which can then be tested in controlled clinical trials.
Now let’s turn to the findings. To their credit, Morse and Wispelwey acknowledge some of their paper’s shortcomings, noting that it “is limited by its observational nature, and [that] unmeasured confounders are likely present.” Nevertheless, they claim that their “study is one of the first to demonstrate that admission service for patients admitted with [heart failure] is an intrahospital driver of racial inequities in [heart-failure] outcomes.” (Emphasis added.) They further insist that “ongoing institutional insistence on self-critique and recognition of the pervasiveness of structural racism and bias will increase the likelihood of success in achieving health equity at all US institutions.”
None of this is borne out by their work. All that Morse and Wispelwey successfully prove in their research is that doctors directed certain black and Latino heart-failure patients to the general-medicine service, rather than to the specialty cardiology unit, at a higher rate than certain white heart-failure patients. (We do not know how they selected the groups they studied, so we cannot be sure the sample is representative of all heart-failure patients at Brigham and Women’s.) They do not even attempt to demonstrate that any of these decisions were wrong. More to the point, they fail to demonstrate that these decisions led to worse health outcomes for black and Latino patients. In fact, the study shows that the mortality rate among black patients was lower than that of white patients by a statistically significant percentage.
This might suggest that the black patients in the study were not as sick to begin with and that the physicians who referred them to the general medicine service were correct in their assessment that they did not require specialty cardiology care. (This idea is further bolstered by the fact that there were no statistically significant differences in re-admission rates within 30 days between patients who were admitted to the general medicine service and those who were admitted to the cardiology ward, among the patients studied by Morse and Wispelwey.) But all this is speculative; we do not know why more white patients died than black patients.
The Real Lessons of Morse and Wispelwey’s Research
What we do know from the information that they supplied about the two cohorts is that there were important clinical differences in the two patient populations that likely determined where they were sent for care in the hospital. For example, significantly more of the black patients had chronic kidney disease and were on regular dialysis treatments. When such patients develop the symptoms of congestive heart failure, the mechanism of their clinical disorder is generally fluid accumulation in their lungs, since kidney failure inhibits their ability to excrete excess ingested fluid. When this occurs, dialysis patients require even more aggressive dialysis treatments than usual. In many hospitals, such treatments are more efficiently performed in the general medical services than in a specialized cardiology service. Hence, clinical need likely drove some if not most of the clinical decisions to admit more black patients to the general medical service.
White patients, according to the study’s data tables, also suffered from a preponderance of valvular disorders, which are much better cared for in specialty cardiology services.
Some other notable differences include a significant age gap between the white and black patients; the average white patient in the study was 77 while the average black patient was 66. Perhaps most strikingly, Morse and Wispelwey also do not seem to have accounted for the fact that a much higher proportion of the white patient group did not live in the Boston metro area; 21% of white patients lived outside of Boston, compared to only five percent of black patients. In other words, a large number of emergency patients chose to pursue specialty care at a hospital far away from where they actually live—or perhaps were instructed to do so by their doctor.
This fact hints at unusual circumstances among the white patients. Some might have required complex care, had a pre-existing relationship with a cardiologist at Brigham and Women’s, or were perhaps drawn by the reputation of the Brigham and Women’s specialty cardiology unit, which provides an array of rare services. In any case, it indicates a particular sort of patient seeking a particular sort of care.
These are just a few of the confounding elements in the study, none of which stopped Morse and Wispelwey from publishing “An Antiracist Agenda for Medicine,” in the March 2021 issue of the Boston Review. It is a manifesto of sorts, in which they draw on their highly suppositional research to promote a sweeping “case for medical restitution.” The pilot initiative at Brigham and Women’s is but a small part of their mission. The “Agenda” has been widely cited as a call to arms for the U.S. medical system.
The details of their vision are important and will be addressed in upcoming posts. But the uncritical embrace by leading medical institutions of Morse and Wispelwey’s conjectural research is more significant. Considering how flimsy their actual findings are, their conclusions are absurdly bold. While the mild protest against their work was met with extreme outrage, the endorsement of their research by hundreds of physicians and dozens of medical establishments should be cause for serious concern.