Health
The Problems with Discrimination Research in Medicine
The narrative that has emerged from the conclusions of these limited studies could inadvertently cause some populations to avoid medical follow-up and form an inaccurate view of healthcare practices.
Like many professions in Western society, medicine is examining itself for the presence of racial inequities and strategies that can ameliorate these differences. Many publications have focused on the disproportionately poor outcomes of minorities in our healthcare system with an emphasis on systemic and structural forces that shape such inequities. As I concluded in my last article for Quillette, these discussions should proceed with the utmost scientific caution, as the answers and implications stand to affect the most vulnerable populations.
With this in mind, there are limitations in the current literature on alleged medical discrimination and the associated health outcomes. In addition, much of the literature on this topic relies heavily on surveys and patient self-reports to assess bias and discrimination while downplaying or ignoring alternative hypotheses. The narrative that has emerged from the conclusions of these limited studies could inadvertently cause some populations to avoid medical follow-up and form an inaccurate view of healthcare practices.
As background, the relevant literature appears to show that black and Hispanic patients tend to view medicine as more discriminatory than white patients do, and black patients often report poorer quality of care overall.1, 2 On top of this, even with greater levels of high school graduation, black patients reported feeling more discriminated against and having more unmet health needs than the Hispanic cohort. In the context of the coronavirus pandemic, a recent article discussed data demonstrating that not only were certain minority groups being affected disproportionally by the virus, but they also experienced more episodes of racist comments due to a perceived connection between their ethnicity and the pandemic. The belief that physicians of a different race act in a discriminatory manner appears to increase one’s preference for a physician of the same race,3 which makes it more notable that black, Hispanic, and Native American physicians are typically in short supply.4
The first thing to be aware of when evaluating these studies is that, although the affront caused by perceiving discrimination or racism may be real, the perception itself might still be false. Operating under the assumption that correlation of perceived discrimination with negative health outcomes implies an overt discriminatory cause not only perpetuates negative attitudes towards the medical profession, it also perpetuates research habits that can open the door to ideological speculation masquerading as a robust explanation. Dr. Danielle B. Moody has evaluated perceptions of discrimination in relation to health outcomes in multiple articles now, but even when confronted with data that conflicts with the hypothesis that discrimination is a main driver of poor health, researchers do not readily accept such findings.
For example, when studying whether or not an abnormality—in this case, something called “white matter lesion volume” (WMLV)—found on brain imaging studies correlated with self-reported discrimination in African Americans, the team found this abnormality increased with decreased ratings of discrimination in black youth. And yet the conclusions read:
Among older AA [African Americans], as lifetime discrimination burden and racial discrimination increased, so did WMLV. However, in younger AA, decreases in racial discrimination were associated with increased WMLV. Elucidation of complex mechanistic underpinnings, including potentially differential impacts of the acknowledgment versus suppression or underreporting of discriminatory experiences, among AA of different age cohorts, is critical to understanding the present pattern of findings.5
Asserting that younger subjects are merely engaging in “suppression” or “underreporting” as the only explanatory factors for this anomalous result can lead to downplaying other relevant factors for assessing and treating poor outcomes among specific groups—such as African Americans merely having a higher baseline prevalence of WMLV on average.6 Furthermore, the brain images were obtained around five years after the original self-reported discrimination ratings were completed—something the team acknowledged, as multiple aspects of this group were examined in stages years apart. Such a time lapse between the original discrimination ratings and the eventual brain images makes it difficult to establish any firm association between the two factors. This is not to disparage the labor of the researchers, which I am sure was immense. I simply wish to highlight that even outside of the narrative of discrimination, methodology can confound otherwise promising results.
A similar publication by Moody et al studied whether or not a marker of cell aging and overall lifespan was negatively influenced by perceived discrimination in subjects. The study concluded there was a significant correlation, but only for African American women and white women at higher socioeconomic status.7 The research team asserts this is explained by African American women with higher incomes being exposed to more white communities and therefore more “chronic, interpersonal discrimination.” When faced with the same finding in white women, the following rationale is offered: “perceptions of progress among racial minorities have been found to stoke concerns of destabilization of the traditional social hierarchy among non-racial minorities.”
This is a dubious conclusion given the same finding was not seen for the white male cohort, but even when reviewing the citation used to support the above claim for white women, it reads as follows: “the effects on whites’ expectations of increased anti-white discrimination found in the present work were not primarily driven by increased concern about whites’ group status in a future ‘majority-minority.’”8 With this in mind, one cannot help but feel a more meretricious explanation was chosen to better support perceived discrimination as the explanation for the team’s findings. The results might be partly explained by average gender differences in levels of distrust during clinical trials9 but this does not support the narrative of discrimination as a health determinant, and so it does not get a thorough discussion.
The study on perceived discrimination as well as Moody’s latter article do concede that the bias of perceived discrimination studies can make the results questionable. Subjects can only report what they recall, and the very methods used may over-estimate the results and therefore the alleged effect of perceived discrimination. But these caveats did not stop the New England Journal of Medicine (NEJM) from citing the above literature and making the bold statement: “Discrimination and racism as social determinants of health act through biologic transduction pathways to promote subclinical cerebrovascular disease, accelerate aging, and impede vascular and renal function, producing disproportionate burdens of disease on black Americans and other minority populations.”10 Unfortunately, in the wake of George Floyd’s death, the NEJM and the Lancet have doubled down on publishing articles with increasingly hyperbolic rhetoric that sometimes stretch the boundaries of scientific journalism into platforms for radical thought-pieces.
Outside of these issues, medical literature can be communicated to providers in a way that reinforces discrimination as a major problem even if a study conveys contrary findings. This might have happened in a recent publication regarding hypertension (HTN) and its relation to lifetime discrimination burden. The summary for providers simply concludes in the highlights section, “A study of African Americans in Mississippi shows an association between experiencing discrimination over a lifetime and developing hypertension.” When reviewing the data in full: smoking status, having no high school diploma, lower physical activity, and higher age were significantly correlated with hypertension.11 In contrast, coping methods for discrimination, rating of lifetime discrimination, and factors patients attributed to lifetime discrimination, had no demonstrable link with incidence HTN. Interestingly, the stress attributed to lifetime discrimination burden was significantly correlated with HTN incidence, but the authors note: “The associations with stress from discrimination were weakened and no longer statistically significant after adjustment for hypertension risk factors.” I view this as overwhelmingly good news and I wish it were emphasized more, as we can modify certain risk factors to varying degrees without conveying to patients that perceived discrimination limits their agency in medical care.
The language used in perception-based studies can also dramatically alter results and over-state findings. The word “discrimination” can lead participants to believe their group is being discriminated against more than usual, and this effect is likely higher for minority participants. A study performed on university students demonstrated that manipulating the term “discrimination” in terms of the severity and frequency of events could modify both personal and group judgments, resulting in a disparity between claims of group discrimination in relation to individual discrimination.12 A separate study of this phenomenon demonstrated that using stereotyped examples of respective groups resulted in higher ratings of group discrimination than the personal variety, and each group was also quicker to perceive group rather than individual discrimination.13 This could be worsened when surveys are broad to the point where affirmative answers are almost inevitable. Such surveys include questions like “Do you think you have ever been unfairly fired or denied promotion?” and yes/no answers to having ever been treated with “less respect than others” or do you ever perceive that people think “they are better than you.” I find it unlikely these map on to most humans’ concept of discrimination.
Tim Rogers
The above finding is called the personal/group discrimination discrepancy: heightened claims of group discrimination despite lower experiences on an individual level. Although originally thought to be a mechanism by which groups suppressed personal experiences of discrimination, subsequent research has demonstrated that even when subjects reported experiencing clear prejudice, they still reported higher perceived levels of discrimination for their identity group in comparison to their own individual rating.14 The authors note that a possible explanation for higher claims of group discrimination are “claims of discrimination can be used as a basis for promoting social change designed to improve the status of the minority group,” and also noted there was no direct evidence for groups exaggerating claimed discrimination in this study, but there also did not appear to be evidence that ratings of discrimination are “reality-based.”15
Studies that rely upon reports of perceived discrimination can be flawed by bias in both the subjects and experimenters, but these examples demonstrate that perceptions vary greatly within groups and are more complex than explanations reliant on majority-minority discrimination. In one of the aformentioned articles regarding preferences for a same-race provider, we see that this preference diminishes when respondents perceived that discrimination was not a matter of race, but rather from healthcare in general or from positions of authority regardless of race. Additionally, although much of the contemporary discussion revolves around addressing systemic forces that result in unequal outcomes, polling data from a national survey demonstrated only a quarter of African Americans felt that the bigger problem regarding discrimination was based in law or government policy compared to nearly half of respondents feeling most discrimination was merely episodes of prejudice from individuals. This same poll demonstrated that personal experiences of discrimination in terms of “slurs,” “negative assumptions,” and feeling “people acting afraid of them” were much lower for African Americans earning less than $25,000 compared to those earning more than $75,000, which could indicate the perception of discrimination might be more of a luxury belief than one ailing the less fortunate. This might also convey that researchers strictly focusing on systemic factors have incongruent beliefs with the population of intended benefit.
