The Erasure of Intersex, Transgender, Nonbinary, and Agender Experiences Through Misuse of Sex and Gender in Health Research

Abstract

Conflation of the terms and concepts of “sex” and “gender” continues to perpetuate the invisibility of sex and gender minorities and obscure information about the ways in which biological sex and gender affect health. The misuse of sex and gender terms and the sex and gender binaries can yield inaccurate results but also, more importantly, contributes to the erasure of intersex, transgender, nonbinary, and agender health experiences. In this article, we discuss ways in which public health researchers can use sex and gender terms correctly and center the health experiences of intersex, transgender, nonbinary, and agender individuals. This includes promoting sensitivity in approaching sex and gender minority communities, improving survey questions, and collaborating with GSM communities to improve research quality and participant experiences. Improving our standards for the quality of sex and gender term usage and centering sex and gender minorities in public health research are imperative to addressing the health inequalities faced by sex and gender minorities.

Abbreviations:

 
• GSM

  gender and sex minority(ies)

 
• HIV

  human immunodeficiency virus

INTRODUCTION TO SEX AND GENDER CONFLATION IN HEALTH RESEARCH

The conflation of sex and gender has haunted public health, epidemiology, and medical research since their creation, at the expense of gender and sex minorities (GSM). This article, brought forth by a gender-diverse group of students and practitioners, focuses on the erasure of GSM experiences in the United States, including the experiences of intersex, transgender, nonbinary, and agender people, but not on identities of sexual orientation (such as lesbian, gay, bisexual, asexual, etc.). Definitions of sex and gender terms are presented in Table 1. None of the experiences described by these terms are necessarily static across the life course.

The Centers for Disease Control and Prevention and the National Academy of Medicine recognize that GSM have not been adequately represented in national health surveys, as GSM are often coded as having missing data in sets of sex and gender binary questions or are imprecisely assessed with a single sex/gender question (1). Broadly, centering cisgender experience in health research causes systemic erasure of GSM experiences, where GSM patients encounter stigma and bias, refusal of care, and a lack of professional knowledge in the health-care system and in understanding determinants of GSM health (2). In this article, we aim to confront the public health field’s broad misuse of sex and gender, and their respective binaries, to draw attention to how behind we are in understanding GSM people’s health. By using sex and gender terms that permit centering GSM communities, we as public health researchers can move forward from a history of misrepresenting and erasing GSM experiences in health (2). For example, routine collection of homicide data blinds public health surveillance toward lethal violence directed at gender minorities (3, 4). This also holds true in active public health surveillance—for example, an overwhelming majority of states do not require reporting of transgender status or nonbinary or agender identities in coronavirus disease 2019 (COVID-19) case or exposure data (5).

An estimate of how far behind we are in public health in understanding the health of GSM people begins with a lack of consensus on our most basic measure, prevalence. We have poor statistical estimates of the size and health of the GSM community. For the diverse array of conditions labeled intersex, the prevalence is estimated to be between 0.02% and 1.7% of the US population, with 1–2 infants per 1,000 receiving surgery annually to normalize genital appearance (6, 7). For self-reported transgender identity, the prevalence is estimated to be between 0.4% and 0.6% of the population (8–10). For self-reported nonbinary and agender identity, there is no estimate of prevalence for the US general population. However, a meta-regression analysis of 12 national surveys using probability sampling estimates found that among transgender survey participants, 31% identify as nonbinary, 20% identify as gender-nonconforming, and 14% identify as agender (10). While we lack an estimate of gender minority prevalence in the United States, survey data from the Netherlands showed that 3.2%–3.6% of respondents reported “an ambivalent gender identity” (11, p. 381), and data from Israel showed that 35% reported feeling “to some extent as the `other' gender” (12, p. 291). This wide range in prevalence highlights the importance of cultural gender norms and survey question design in quantifying gender expression. Knowing these kinds of prevalences is important for estimating the denominators in essential public health measures such as prevalence and incidence rates.

BACKGROUND ON THE SEX AND GENDER BINARIES

Sex, as a binary measure of male and female categories used to justify the social concept of a gender binary, was a relatively novel concept created in the 18th century by physicians based on typed reproductive organs (13). The binary phenotypic sex assigned at birth and inscribed in one’s medical record is often what is referred to as “sex”; however, this label represents a host of biological complexities. Binary sex collapses many phenomena into a single category in research, including self-reported sex, legal sex (passports, driver’s licenses, etc.), external genitalia, gonads, sex assigned at birth, hormonal sex, and hormone receptors. The correspondence between these phenomena may vary between people and within a person over time, presenting conceptual difficulties when utilizing a static unidimensional binary measure to describe biological sex (13). By contrast, understanding sex as dynamic, bimodal, and multidimensional—for example, by inventorying organs, organ systems, and physiologies—permits representation of the diversity of experiences related to human reproductive biology and changes in those experiences across the life course (13). Phenotypic sex has a role in research and practice for predicting medication response, discerning immunological function, and understanding differences in the etiology and distribution of diseases such as cardiovascular disease, as well as demarcating patterns of reproductive experience (14, 15). However, we are far from knowing why “sex differences” occur, because sex is conflated with gender and a gendered environment, and current methods have failed to discern how the social differences inform the biological ones (14, 15).

Genotypic sex and phenotypic sex overlap but are not equivalent. Not only can people have multiple combinations and duplications of X and Y genes, but imprinting and exogenous and endogenous changes to gene expression also complicate the use of sex as a binary measure (14, 16). Hormonal sex also does not fit well into a binary categorization. All humans have testosterone, estrogens, and progesterone, which have stereotyped patterns relating to genotype, with notable changes resulting from patterns across one’s life span (e.g., puberty, aging, menopause, pregnancy), use of hormone replacement or suppression therapy, adverse effects from medications, and hormonal effects of disease courses (e.g., polycystic ovarian syndrome) (14). Organ inventories based upon self-report may describe sex, although there is a diversity of sexual organs possible for any person related to genetically driven differences in what internal and external organs people have, which they may or may not be aware of without having had diagnostic imaging, biopsy, or endoscopy. Additionally, sex organs can be altered by surgical procedures for the purposes of gender affirmation, tumor removal, or cancer risk reduction, or at birth (for intersex babies) to ascribe phenotypic sex.

The gender binary is built on the foundation of the medicalized sex binary, where cultural roles, social hierarchies, and political power were drawn from observed differences in reproductive physiology (13). Gender has long since been recognized as something that can change over one’s lifetime (13, 17). Gender expression is not linear but a continuum wherein individuals may express themselves as more than 1 gender or as androgynous (17). An individual’s gender identity may not align with their gender assigned at birth; otherwise these individuals are characterized as “cisgender.” However, gender continues to be used socially and in research as a static binary measure, even when it is not conflated with binary sex. Gender minorities are often categorized as “other”—sometimes literally, as in questions which ask whether one identifies as a man, as a woman, or other, an enactment of “othering” which also occurs for sex minorities who do not fit neatly within binary sex categories of male or female—preceded by little to no information on what “other” refers to and yielding unusable results due to small sample sizes and a lack of specificity. It is also important to consider that regardless of whether the data are collected in the clinic or outside the clinic, people with gender minority experiences may fear disclosure of their sex or gender identity. These concerns include a fear that disclosure could incite bias or discrimination, personal discomfort, or further disclosure to others without their consent—concerns that become frustrating when it is not apparent how these disclosures are relevant to the physician or researcher (18). Therefore, explicit contextualization of questions is important. An example of such contextualization would be: “We ask everyone about their gender and transgender status to ensure everyone receives the highest quality of care. These questions help us identify and address disparities in our services.”

CONSEQUENCES OF SEX AND GENDER TERM MISUSE

There are several consequences for public health when sex and gender terms are misused. First, when sex and gender are conflated or falsely combined into sex/gender, our research lacks validity in describing health experiences. Sex, described phenotypically or genotypically, is generally more appropriate for questions of physiology, while the social constructs of gender and transgender are generally more appropriate for questions implicating social structures. These terms are related because both are inhabited at once by every person, but the terms are not interchangeable, and conceptually and methodologically both, neither, or either may be relevant to a specific health question (19).

Second, by accepting binary gender or sex methodologies, we erase GSM experiences. Relying on the current gender binary discourages individuals who do not identify as cisgender, potentially causing them to feel ostracized, and creates barriers for them to seek medical attention or other services due to the lack of inclusivity (2, 18). We have little or no idea what the top causes of mortality among GSM communities versus the general public are, and even when we identify a health issue that affects GSM communities, we characterize sex and gender poorly. For example, in the human immunodeficiency virus (HIV)/acquired immunodeficiency syndrome (AIDS) epidemic, the term “men who have sex with men” could be understood to exclude transgender women, limiting health research, advocacy, and community health efforts for vulnerable transgender women, who have similar rates of HIV infection as men who have sex with men and experience both transphobia and homophobia (20). The term “men who have sex with men” also occludes the commonalities and differences in the health needs of transgender men who have sex with cisgender or transgender men. For example, as of February 2021, less than 2% of the academic citations reported for “sex with” and “men who have sex with men” on Google Scholar (Google LLC, Mountain View, California) included “transgender women.”

The lack of recognition of GSM experiences in research avoids health inequities and furthers personal and institutional biases in health care. The US Transgender Survey highlights that out of 27,715 transgender individuals, overall 14% are still uninsured, 33% have had a negative experience with a health-care provider in the last year, and 7% have attempted suicide in the last year (21). Additionally, transgender people are 8 times more likely to experience serious psychological distress and more than twice as likely to have a disability than the general population (21). Through the recent inclusion of a question about transgender and gender-nonconforming identity in the Behavioral Risk Factor Surveillance System survey, we know that GSM are less likely to have up-to-date cancer screenings, probably because of health-care discrimination and a lack of clinical guidelines for GSM (22). We have also learned from the Behavioral Risk Factor Surveillance System that GSM people have higher odds of depression, cardiovascular issues, and arthritis and lower odds of accessing primary care (23). Given the increased burden of health issues experienced by GSM, the moral imperative for studying GSM in public health research is clear.

RECOMMENDATIONS FOR IMPROVING RESEARCH ON GSM

Having discussed the implications of sex and gender conflation, we will now review how to work toward improving GSM health equity and representation in public health. The following recommendations are supported by personal experiences of living while transgender, transsexual, and androgynous and professional experiences of collaborating with GSM community members and GSM-serving professionals to create epidemiologic methods to improve GSM representation in public health. First, as a field that relies on rigorous methods, we need to focus on correcting and standardizing our use of sex and gender terms. Meanwhile, we need to focus on filling in the gaps of public health research on GSM people.

The Sex and Gender Equity in Research guidelines were created by the European Association of Science Editors in 2016 to standardize the ways in which sex and gender are reported in scientific publications (24). Among other recommendations, they assert that authors should report and justify their use of sex and gender in the Methods section of their papers and then discuss the implications of how sex and gender were used in the Discussion section (24). The new guidelines from the Federal Committee on Statistical Methodology provide practical advice for researchers on using appropriate terminology in surveys for sexual orientation and GSM identities (25).

Planning for the representation of GSM should occur during the research design phase, not after the data are already being collected. Methods for representing and including minority groups depend on the research question and funding. Questions should be designed with forethought about how the data will be analyzed. Sample sizes needed to power the study based on estimated prevalence and the P value cutoff should be calculated for all minority groups researchers claim to represent. Oversampling of minority populations and subsequent weighting can help reduce the population size and the amount of funds needed to answer a research question (see Kalsbeek (26) for statistical methods for sampling minority groups). Respondent-driven sampling methods have been developed specifically to capture representative samples of occult or small minority populations without the cost of probability samples, including in lesbian/gay/bisexual/transgender research (27–29). Additionally, with the relative dearth of health research on GSM, researchers should consider centering their study on GSM by including only people who identify as such, or should at least suggest doing so in a follow-up study (2). This also allows for more possibilities in analyzing intersectional experiences between sex, gender, race, class, and disability status.

Many data sources will bear small numbers of persons in differing GSM categories, and researchers may consider aggregating disparate minority gender and sex experiences into a single “is GSM” category. We see 2 concerns here. First, comparing a single “is GSM” category with a single “is not GSM” category might give some insights into health inequity across a gender minority/majority social divide, while paving over the specifics of gender identity, gender modality, sex, and their intersections (including with race and ethnicity, class, and other social categories). Second, comparing a collapsed GSM category with “cisgender women” and “cisgender men” categories or with “female” and “male” categories amounts to an “othering” practice, albeit using a different label than “other,” and reifies sex and gender binaries.

Asking questions about GSM status requires sensitivity to the collective and individual traumas experienced by minorities within health care and research. When possible, GSM communities should be involved in research to identify research priorities, assure sensitivity of research methods, and endorse survey language (2). Before initiating questions about sex and gender, the researcher should state why the questions are being asked and reiterate the right to decline participation. For surveys with any kind of verbal or written follow-up, questions on gender and sex should be preceded by a space for providing one’s preferred name and pronouns to be used in all correspondence. Additionally, for all questions, participants should have an option for providing a “do not know” or “do not want to answer” response; and inclusion of a “questioning” option explicitly recognizes the possibility of the contemplation, exploration, and discovery that often occur for individuals transitioning in aspects of sex, gender identity, and transgender status (3).

The appropriate selection of questions depends on the project’s purpose, methods, and resources. Multiple-choice questions are shown in the Web Appendix (available at https://doi.org/10.1093/aje/kwab221) and are suitable bases for forming the denominators of population health and health disparity measures. However, open-ended questions can be analyzed qualitatively and may be useful in building rapport, as when a question like “What name would you like us to use, and how would you like to be addressed?” precedes more structured questions like those above. Many of the recommended questions will require up-front transparency about why these questions are being asked, and what the data will be used for. For example, asking about transgender status in a clinical context may be important to know in order to provide transgender-specific care, to track transgender/ cisgender inequities in services as part of routine health equity work, and to differentiate insight into transgender-specific versus cisgender-specific health concerns in the clinic's community. However, asking a question about genitalia as part of routine practice is unlikely to serve those purposes.

A common concern is that asking questions about gender identity (and sexual orientation) will upset older cisgender heterosexual patients; however, some survey research from this population showed that only 3% of 491 participants reported being distressed, upset, or offended by questions on sexual orientation and gender identity (30). Another common concern for researchers is how to analyze multiple identities held by participants. This is an active area of research, and tools created thus far have included hierarchical classes analysis for intersectional identities (31) and use of the validated Genderqueer Identity Scale for gender identity, awareness, and fluidity (32).

Questions such as those we advocate must be integrated into routine research practices, clinical practices, and health-care systems practices to invite, include, and honor gender and sex diversity. In our experiences, there will be different kinds of resistance to answering such questions. For example, there are persons committed to sex and gender binaries who do not want to answer questions about gender identity. For another example, there are persons who identify as GSM but want to remain anonymous as such. The “I do not want to answer” option is therefore important to avoid the biases in representation which arise through sex/gender binary default answers. As with once out-of-bounds questions about sexual behavior and sexual orientation, this kind of resistance will lessen over time. We have also learned that response matrices can help interviewers navigate the social complexity of routinely asking questions which transgress the sex/gender binary by anticipating specific language around emotional activation, specific questions about motivation, etc., and providing scripted responses to direct attention to the ways in which inclusion and equity in the research, clinic, or other context are served.

CONCLUSION

The goal of this article was to address the misuse of sex and gender terms and their binaries and discuss how this affects communities of GSM people. Not only is the conflation and reduction of sex and gender terms harmful to the validity of public health research, it also erases GSM experiences of medical discrimination and increased prevalence of physical and mental health conditions. To adequately serve the GSM community, public health researchers need to conduct research with methodological integrity. This includes careful consideration, planning, and defense of using appropriate sex and gender lenses for medical and social public health research, as well as using the right survey tools to accurately and sensitively assess GSM status. For these changes to occur, we need to adequately train health-care professionals and researchers to use these tools, agree on research standards for methodologically and conceptually separating sex and gender, and involve the GSM community as knowledgeable partners in this change.

ACKNOWLEDGMENTS

Author affiliations: School of Medicine, Oregon Health and Science University, Portland, Oregon, United States (Tessalyn Morrison); and Department of Epidemiology, OHSU-PSU School of Public Health, Portland, Oregon, United States (Alexis Dinno, Taurica Salmon).

No new data were generated or analyzed in support of this research.

We thank Grace Walker-Stevenson, Kalera Stratton, and Lela Brown for their support in the ideation and initial planning of this work.

This work was presented at the OHSU-PSU School of Public Health Annual Public Health Conference, held virtually on April 8, 2021.

Conflict of interest: none declared.

REFERENCES