Dehlendorf et al ask the important question: does a patient’s race/ethnicity and socioeconomic status (SES) affect physician recommendations for intrauterine contraception (IUC)? IUC is a superb method of long-acting reversible contraception. Because unintended pregnancy occurs at higher rates in poor women and women of color than in affluent white women, it would be useful to know if clinician recommendations contribute to disparities in rates of unintended pregnancy. To answer this question, the authors used standardized patient videos, and thus were able to control for clinically relevant patient variables. However, even though this study’s overall methodological choices were solid, its findings, in the authors’ own words, “do not lend themselves to easy interpretation.”

Indeed, the authors’ stratified multivariate analysis showed that patient race and SES do shape physician IUC recommendations, but in inconsistent ways: providers were more likely to recommend IUC to women of color than to white women, but only when women were of low SES. Providers were more likely to recommend IUC to high SES women than to low SES woman, but only when they were white. These findings leave us with 2 vexing questions: (1) how do we explain the specific, seemingly contradictory finding that highest odds of physician IUC recommendation were in low SES women of color and high SES whites; and (2) why is race/ethnicity and SES a factor at all in physician recommendations when patients differ on no other clinically relevant variables? Is statistical discrimination an adequate explanation? Is the difference due to race- or class-based bias or prejudice of individual physicians? The authors appropriately consider both possibilities. However, powerful answers to these questions can also be gained by turning to the humanities to enhance our understandings of race, ethnicity, and class. These understandings focus on the well-known history of race- and class-based reproductive health policy; the ways in which ideas of race and class get built into medical diagnoses themselves; and the complex and intersectional nature of identity.

But first let us highlight the good news in the study of Dehlendorf et al: they found that overall physicians recommended IUC 42-86% of the time. Given that in 2002, 89% of obstetrician-gynecologists inserted ≤10 IUCs, and 20% inserted none, the findings of Dehlendorf et al may represent a new or growing level of provider comfort with IUC. Nevertheless, we might still ask why IUC was not recommended at even higher rates overall, and we may wonder why physicians’ recommendations differed in patients who were nulliparous or had a history of pelvic inflammatory disease, since contemporary evidence argues that these should not be considered contraindications for IUC use. To be sure, the overall underuse of IUC in the United States (only about 1% of American women use IUC) deserves attention; however, this was not the focus of Dehlendorf et al’s study.

We can speculate on possible meanings for the seemingly contradictory findings. The positive, half-full glass interpretation of these results is: doctors (appropriately) do not associate IUC with infertility, believe women who report no history of sexually transmitted infection (STI), and realize that low SES and sometimes minority status is a barrier to regular health care. Therefore women in these categories deserve immediate access to the best contraception available, just like their high SES counterparts. Unfortunately, this interpretation cannot account for the discrepancy in IUC recommendation between low and high SES whites. However, all findings can be explained by a more negative, half-empty glass interpretation: providers believe IUC causes infertility; they believe low SES women are at greater risk for STI than affluent women and are therefore at greater risk of infertility in the setting of IUC. This risk of infertility is disregarded for women of color, and taken seriously for white women. This of course is a profoundly dangerous interpretation, for it means, among other things, that some physicians do not value the fertility of white women and women of color equally. It is a particularly problematic interpretation today, because ideas such as this are being deployed to advance antiabortion political agendas.

Dangerous as this interpretation is, to dismiss it outright would be to neglect the fact that some populations’ fertility has been valued more than others, a phenomenon anthropologists call “stratified reproduction.” The social value of fertility has long depended upon one’s relationship to power and privilege in society. As European settlers moved westward in the 19th century, the homestead system of land grants valued and subsidized white fertility, while the visibly declining population of American Indians was hailed as divinely ordained “manifest destiny” and enforced by federal government policies. Racial slavery, too, institutionalized differential fertility. White slave owners interpreted black fertility as a liability until Congress outlawed overseas slave trade in 1807, when births to enslaved women became the only way to increase slave populations. In the late 1800s through 1940s an active US eugenics movement sought to increase births among the “fittest” and limit births among the “unfit” through coercive sterilization of poor people, immigrants, prisoners, and men and women hospitalized in psychiatric facilities. During this time the high fertility rates of new immigrants and declining fertility of native-born whites caused President Theodore Roosevelt to fear white “race suicide.” White fertility control through contraception or abortion was stigmatized as a “crime against the citizenry.” More recently, pernicious stereotypes of women of color’s hyperfertility led to a variety of forms of medical discrimination, including compelled sterilization and coercive use of contraception–for example, making receipt of welfare benefits conditional on having contraceptive implant insertion.

Given this history, Dehlendorf et al understandably worry that observed differences in IUC recommendations reflect conscious or unconscious racism of individual doctors. However, race and class work in US culture and medicine in many ways other than at the level of the individual. Race and class organize power in society, and are embedded deeply and often invisibly in the structures of US society, including medicine. Recently, psychiatrist and cultural historian Jonathan Metzl showed how schizophrenia moved from a diagnosis made primarily in white women, describing behaviors of docility and inability to function as a wife and mother, to one made in African American men, describing behaviors of hostility, rage, and aggression. This diagnostic shift coincided with the growth of the civil rights movement in the 1960s. In other words, anxieties about changing racial hierarchies became encoded in psychiatric diagnostic criteria. Ultimately Metzl conclude that in any patient encounter there are 3 races present: the race of the physician, the race of the patient, and the race of the diagnosis. In his words, “racial tensions are structured into clinical interactions long before doctors or patients enter examination rooms.” Thus, when we interpret the findings of Dehlendorf et al we must consider that terms such as “sexually transmitted infection,” “fertility,” and “infertility” all have ideas of race and class built into them.

This understanding of medicine complicates the idea of statistical discrimination. Statistical discrimination occurs when clinicians erroneously use epidemiological data to assess individual risk and guide individual treatment decisions for patients in particular sociodemographic groups. In this study, physicians perceived that low SES patients had a higher risk of STI than high SES patients. They appear to have used epidemiological evidence of higher prevalence of STIs in low SES women to guide IUC recommendations, rather than individualized data reported by the patients themselves. However, consider in contrast a recent study on infertility by Ceballo et al. These authors showed that only 16% of physicians surveyed correctly identified African American as the racial group most at risk for infertility, and only 13% correctly identified women without a high school degree as being most at risk of infertility. Here, physician beliefs ran in direct opposition to epidemiological data. This may represent simple lack of knowledge, but may also mean that the diagnosis “infertility” itself has ideas of whiteness and educational achievement built into it. In the study of Dehlendorf et al it may not simply be the case that physicians thought low SES women had a higher rate of STI because epidemiologically it was true. It may be the case that “STI” itself has ideas of low SES built into it. If that were the case, and if “infertility” has ideas of whiteness built into it, we have a potential explanation for higher IUC recommendations in low SES women of color than in low SES white women: infertility as an IUC complication was simply not on the radar for doctors who viewed the video histories portrayed by women of color.

The half-full and half-empty glass interpretations of the data of Dehlendorf et al reflect the outer boundaries of a range of interpretations. What the findings really mean cannot be ascertained because the authors measured 2 things about their physician respondents, although they only collected data on 1: (1) respondent ideas about patient identity–race, ethnicity, and SES; and (2) respondent ideas about IUC. Since the authors did not perform an IUC knowledge or attitudinal questionnaire, we do not know what stereotypes physician respondents had about the contraceptive device itself–whether they believed it was the best contraception available or a significant risk factor for infertility. Without knowing clinician attitudes toward IUC, we cannot draw valid conclusions about benign or invidious intent of subsequent recommendations.

Perhaps the most important conclusion from this study is that race and SES matter, and they matter together . Neither patient race nor SES alone predicted provider recommendations. Dehlendorf et al are correct to highlight that looking at race/ethnicity and SES in combination is important in the study of health care disparities. As challenging as this is for data analysis, it is essential if health disparities research is to reflect real-life contexts. No person can experience him- or herself as only “man” or only “white.” Instead, real-life identity is formed at the intersection of multiple categories, including sex, race, ethnicity, and class, which mutually shape one another, and shape experience. This idea is central in contemporary humanities scholarship. Stripping away one aspect of identity–or controlling for it in statistical terms–may seem to make sense in the context of a research study, but will always be a poor surrogate for analysis of multiple interactions, and lived experience of identity.

Since US history cannot be undone, since medical diagnoses inevitably encode social tensions, and since the complexity of identity will always exceed our ability to measure it, the important study of Dehlendorf et al should remind us that the best clinical tool we have at our disposal now is evidence-based, patient-centered care. The best we can offer is to understand the very strong evidence in support of IUC use in nearly all women, and to try to understand a patient in the multiple contexts of her own life and her own reproductive wishes.