Cases of Group Harm in Biorepository Research
An overview of notable cases of group harm in biorepository research
Written by the CHIRON Project Team
Published on May 2, 2024
An overview of notable cases of group harm in biorepository research
Written by the CHIRON Project Team
Published on May 2, 2024
As the (re)production of group harm from biorepository researchers often has multiple contributing factors, the CHIRON toolkit intervenes at several points in the research lifecycle. While group harm is rarely reducible to one particular touchpoint in a research study’s trajectory, this piece will look at several instances of harmful research from different perspectives to illustrate potential points of intervention.
When researchers are planning their approach to working with biorepository data, at times they choose research questions that are of low priority to the population group being studied. Kim et al., 20121 used data from the multiethnic cohort study (MEC), which is a repository of samples from Japanese Americans, Native Hawaiians, African Americans, Latinos, and whites, broadly collected for cancer research. Kim and colleagues merged a subset of MEC data with a subset of Human Genome Diversity Project (HGDP) panel data, another major sample repository which will be discussed later in this piece. Kim and colleagues used this aggregated data to study the “genetic origins” of Native Hawaiians, making claims about population movement and contact with other populations. This line of research inquiry is not unusual; indeed, genetic origin research is a widely published field. However, Native Hawaiian researchers report that these research questions are of low priority for Native Hawaiian participants, who prefer that biorepository samples are used to further research on diseases that disproportionately affect Native Hawaiians, like heart disease and diabetes. In order to address the problem of low-priority research, CHIRON tool 3 asks researchers to look into priority-setting literature on the population in their dataset(s).
Given the stated reasons for collection of MEC and HGDP data, one may wonder whether genetic origin studies actually stretch the boundaries of research types consented to by participants. Bird and Carlson, 20242 raise this question in their extensive analysis of how “racial hereditarian researchers” use biomedical datasets to publish ideologically-motivated papers on intelligence and educational attainment in black and white Americans. These papers provide fodder for far-right social forums, as Carlson’s work has shown. Although informed consent for biorepositories is characteristically broad, “study participants were recruited with the understanding that results would be used for medical research,”2 calling into question the acceptability of some stigmatizing types of research questions. The CHIRON toolkit has questions that address participant consent, and the project team also recommends further innovation in this area by data oversight boards.
The way in which biorepository data is curated and analyzed shapes the conclusions that can be drawn and the meanings that researchers, clinicians, and patients can make. At times, researchers choose labeling and categorization schemas that are unscientific, stigmatizing, or just inappropriate. Rosenberg et al., 20023 presents the first major characterization of genetic variation in the HGDP panel dataset.4 This paper’s impact transcends what most scientists experience in their career; “it was the most-cited paper in Science in its year and continues to feature in scientific, philosophical, and cultural debates” long after its publication.5 Rosenberg and colleagues identified major genetic clusters that correspond with continental regions, which has been interpreted by readers to reify a concept of biological race.4,5 While Rosenberg and colleagues later argued that their work should not be used to support a concept of biological race,6 critical scholarship has argued that this is far from a case of runaway interpretation.5,7 The paper presents the sample in two to six population clusters, which Rosenberg’s team selected using the program structure.3 At five population clusters, Rosenberg’s team produced a result corresponding to the regions of Africa, Eurasia, East Asia, Oceania, and America, which created the legacy associated with this paper.7 Indeed, the popular science writer Nicholas Wade used this finding as foundational evidence in his articles and book that promulgate arguments rooted in genetic essentialism.5,7 While Rosenberg, his colleagues, and many other geneticists have disavowed Wade’s work,5 renowned scholar of scientific racism Dorothy Roberts argues that Rosenberg and team’s choice of clustering schema gave rise to this popular interpretation: “There is nothing in the team’s findings to suggest that six clusters represent human population structure better than ten, or fifteen, or twenty.”7
To expand the scope beyond the specifics of Rosenberg et al., 2002, the choices that researchers make in their curation and analysis of biorepository data are not neutral and are not without consequence. The CHIRON toolkit includes a number of questions that address the labeling, lumping, splitting, and “other”ing of population groups, as well as supplementary reading and guides for these processes.
As researchers write their discussion, submit their paper, and publicize their work, at times they fail to contextualize their findings within social and cultural structures. At a 2006 academic conference in Australia, a group of researchers [cited as Hall et al.8 as part of a research team containing Rod Lea and Geoffrey Chambers,9,10 although finding an abstract for this presentation is challenging due to link degradation] presented their findings that their sample of Māori people, a group indigenous to New Zealand, had higher levels of a gene which codes for monoamine oxidase, an enzyme linked to tobacco and alcohol dependency.8,9 The term “warrior gene” appears to have been applied to this particular genetic sequence prior to its usage by this group of researchers,10 although this language could also be problematized as it conflates a gene with a phenotype or displayed set of behaviors.9 In popular press, members of the research team made sense of their findings in that “historically Maori were fearless warriors”10 and linking it to alleged criminality, aggression, and violence in Māori people—claims they later walked back following controversy.8 Claims like these—that carrying a given gene “predisposes people to be more likely to be criminals” in the words of Rod Lea9—reveal a genetic essentialist bias in the research team. They provide a genetic “explanation” for socially situated constructs such as criminality, which are influenced by definitions of crime, policing and biased application therein, and social inequality. In failing to contextualize their findings with histories of colonialism and discrimination that are highly relevant to research involving Māori people, Lea and Chambers and their colleagues actually present an incomplete picture. The CHIRON toolkit asks researchers to contextualize their findings with historical, environmental, social, and cultural factors that may influence researchers’ interpretation of their results, see especially CHIRON tools 7 and 10.
In another case of lacking context, the work of Dhejne et al., 2011,11 to investigate mortality and criminality in transgender people in Sweden has provided extensive fodder to anti-trans activists that seek to reduce access to gender affirming care. Journalist Erin Reed details how this study, which found that transgender people in Sweden who received gender affirming care had higher risks of mortality and suicidal behavior as compared to cisgender people, has been used in congressional hearings, lawsuits, and by right-wing press to argue against providing gender affirming care. As Reed notes, Dhejne and colleagues do not provide a point of comparison with trans people who have not received gender affirming care, only the cisgender population writ large. Given the virulence of transphobia, it is not surprising, then, that trans people even after gender affirming surgery have higher rates of suicidal behavior and psychiatric illness than cis people. Dhejne has expressed frustration over what she describes as misrepresentations of her work, refuting some of the most egregious anti-trans claims that have been made using her study as “evidence.” While Dhejne’s endeavors to clarify bad faith interpretations of her work are appropriate, one wonders if the authors could have done more in their paper to compare their findings with published work or other elements of their dataset on the effect of obtaining gender affirming surgery versus lack thereof to the health and wellbeing of trans people. Given other work exhibiting how mental health improves in transgender youth following gender affirming care, perhaps additional context to this end could have staved off the 2011 study’s worst interpretations. The CHIRON toolkit prompts researchers early on to plan for contextualizing their findings, especially in a sensitive political climate, see especially CHIRON tool 4.
Roberts D. Fatal Invention: How Science, Politics, and Big Business Re-Create Race in the Twenty-First Century. New Press; 2011.