Current state of sex and gender data collection in nutrition science research

Data produced by nutrition surveillance programmes are essential for nutrition epidemiology and population-level analyses. Sex and gender data collection practices vary programme. In the United States, the National Health and Nutrition Examination Survey asks participants to report their ‘gender’ with response options of ‘male’, ‘female’ or ‘does not identify as either’. If the participant ‘cannot decide’, interviewers are instructed to probe with the question, ‘What would you tell a doctor?’ If the participant still ‘cannot decide’ or ‘refuses to answer’, interviewers are then instructed to ask for sex assigned at birth noted on the participant's original birth certificate^{(Reference Schier, Gunther and Landry2,6)} . This approach offers numerous opportunities for improvement. When querying gender, response options are limited to the female–male binary and do not adequately capture the identities of gender minorities. With the probe ‘What would you tell a doctor?’, participants may still be uncertain of whether to report their sex or gender; additionally, participants may feel uncomfortable or unsafe sharing sensitive information about their gender identity. Lastly, the switch to querying sex is a clear conflation of sex and gender data in data collection and reporting practices.

Next, the Behavioral Risk Factor Surveillance System collects state-level data about residents' health-related risk behaviours, chronic health conditions and use of preventive health services. The Behavioral Risk Factor Surveillance System began offering states an optional module in 2014 which includes sexual orientation and gender identity data; a single-question module on sex at birth was added in 2019. As of 2019, thirty-two states adopted the sexual orientation and gender identity module and seven states adopted the sex at birth module⁽⁷⁾. Although this approach offers separate questions for sex and gender, analyses are limited by whether individual states have adopted the modules. The Youth Risk Behavior Surveillance System, which tracks health behaviours among students in grades nine through twelve, conflates sex and gender in a different manner. Although the category of the question is labelled as ‘gender identity’, the question posed to respondents is, ‘What is your sex?’⁽⁸⁾ Similar to National Health and Nutrition Examination Survey data collection practices, this approach conflates sex and gender and offers the binary female–male response options.

In Ireland, the Irish Universities Nutrition Alliance collects data on the dietary intake and health status of the population from ages 1–90 with surveys administered throughout the lifespan. The National Adult Nutrition Survey reports data by gender with males or females⁽⁹⁾. The recent National Teens' Food Survey II and the National Children's Food Survey II report data by gender with boys or girls^(10,11) . The National Pre-School Nutrition Survey conflates the terms sex and gender throughout its reporting⁽¹²⁾. This approach could be strengthened by separately querying sex and gender, as well as offering a wider breadth of gender response options.

Lastly, throughout the United Kingdom, the National Diet and Nutrition Survey collects data on diet, nutrient intake and nutritional status of the population ages 1⋅5 and older. The National Diet and Nutrition Survey reports data on sex for men/boys and women/girls⁽¹³⁾. Similar to the Irish Universities Nutrition Alliance data collection practices, the National Diet and Nutrition Survey approach could be strengthened by providing a separate question for sex and gender with a breadth of gender response options.

Effects of conflation or omission of sex and gender data in nutrition science

Conflation or omission of sex and gender data obfuscates essential demographic data in nutrition studies conducted with human subjects. When sex and gender are conflated, such as a general question that asks participants to check a box for male or female, transgender and gender non-conforming patients may be unsure of whether to record their biological sex or gender identity, which will undermine the precision and generalisability of the study^{(Reference Schier, Gunther and Landry2)}.

In addition, limiting the response options for gender to male or female contributes to the erasure of transgender and gender non-conforming participants from a dataset. Portions of a country's population may be unrepresented or misrepresented as a result. In the United Kingdom, the Government Equalities Office estimated there are 200 000–500 000 transgender individuals, though acknowledged no robust data on the transgender population exist⁽¹⁴⁾. In Ireland, the Lesbian, Gay, Bisexual, Transgender and Intersex Ireland Report collected data on 2264 lesbian, gay, bisexual, transgender and intersex people, 12⋅3 % of whom identified as transgender^{(Reference Higgins, Doyle and Downes15)}. Based on a medical record review at a gender dysphoria clinic in Dublin, Ireland, scholars estimated that 0⋅0067 % of the population was impacted by gender dysphoria, a condition that describes unease or mismatch between one's sex and gender, by comparing the number of patients referred for hormone therapy (HT) with census data^{(Reference Judge, O'Donovan and Callaghan16)}. In the United States, scholars extrapolated data from the Behavioral Risk Factor Surveillance System and Youth Risk Behavior Surveillance System to estimate 1⋅4 % of youth ages 13–17, or 150 000 people, and 0⋅5 % of adults, or 1⋅3 million people identified as transgender^{(Reference Herman, Flores and O'Neill17)}. Precise estimates of the number of transgender individuals remain unknown throughout much of the world^{(Reference Coleman, Radix and Bouman5)}.

Lack of data on transgender and gender non-conforming populations perpetuates a research gap on the distinct nutrition needs and health disparities impacting the population, and therefore insufficient evidence to inform nutrition policy and practice. The pipeline of data to policy and practice is cut short by omission of basic demographic data.

Food and nutrition considerations for transgender and gender non-conforming populations

Emerging research has identified prominent food and nutrition-related considerations among transgender and gender non-conforming populations (Fig. 1). The nutrition-related considerations are broadly classified as clinical or psychosocial in nature^{(Reference Rahman and Linsenmeyer18)}. Importantly, these studies have relied on smaller samples relative to the large samples supported by national nutrition surveillance programmes.

Fig. 1. Clinical and psychosocial nutrition considerations with transgender and gender non-conforming populations.

A sex- and gender-informed approach to nutrition research design, analysis and reporting: practical strategies for nutrition researchers

A sex- and gender-informed approach can improve the accuracy and inclusion of nutrition research design, analysis and reporting. The Sex and Gender Equity in Research guidelines offer researchers a starting point for reporting sex and gender data throughout the research process but is not inclusive of gender minorities^{(Reference Heidari, Babor and De Castro35)}. This section, based on the published literature, details practical strategies for researchers to apply when designing studies with human subjects, especially transgender participants (Fig. 2).

1. (1) Treat sex and gender as separate, relevant demographic data. When designing the demographic data section of a study with human subjects, researchers can collect sex and gender as separate, relevant data. Sex and gender should be treated as essential data on par with other demographic data such as age, race and ethnicity^{(Reference Heidari, Babor and De Castro35,36)} . The ‘two-step’ approach requires a separate question to query sex and gender. Table 1 shows two approaches to framing questions on these constructs. Given that culturally appropriate language is constantly evolving, the ideal wording may vary by country, culture or time period^{(Reference Coleman, Radix and Bouman5)}. For example, although recommendations in the United States use the culturally appropriate language of ‘two-spirit’ for American Indian and Alaskan Native participants, this may not apply for populations in other countries.

2. (2) Appreciate gender as a fluid construct. Researchers can appreciate gender as a fluid construct, rather than a female–male binary. This approach can improve the inclusivity of nutrition research by ensuring that gender minorities are captured in a dataset. Researchers can include response options that capture gender fluidity, such as transgender, genderqueer or gender non-conforming. Providing a free-text response options allows research participants to provide their own gender identity that may not be captured in the response options listed (i.e. non-binary, genderfluid, agender), thus honouring a participant's autonomy in naming their own identity⁽³⁾.

3. (3) Approach data collection on gender minorities with sensitivity to privacy and confidentiality. Although privacy and confidentiality are requisite in all human subjects research, collection of sex and gender data is especially sensitive for gender minorities who face gender-based stigma or discrimination^{(Reference James, Herman and Rankin29,36)} . Researchers can ensure they are only collecting the minimum amount of data necessary to meet a study's purpose. For example, a study that measures metabolic rate may require collecting data on sex, but not gender identity. Lastly, researchers can appreciate that participants may not feel safe or comfortable their gender identity, especially if the study sample is small and they would be more likely to be identified. Offering a response option such as ‘decline to state’ or ‘prefer not to answer’ ensures that participants are not forced to share this data if they do not feel comfortable doing so (Table 1).

Fig. 2. Sex- and gender-informed approach to nutrition research.

Table 1. Two proposed wordings of a two-step method to query sex and gender data