Gender-Related Variables for Health Research

The Challenge

Medical evidence shows that both sex (biology) and gender (sociocultural behaviors and attitudes) interact to influence health and disease across the lifespan (Klein et al., 2015; Tannenbaum et al., 2019; Stefanick & Schiebinger, 2020; Case Study: Chronic Pain). The U.S. Institute of Medicine (IOM, now the Academies of Science) recognized that gender interacts with sex to influence health outcomes nearly 20 years ago (Wizemann & Pardue, 2001). Since then, both the Canadian Institute of Health Research and the European Commission have advocated the integration of both sex and gender into health research (CIHR, 2010; European Commission, 2013).

Recent studies have borne out the importance of gender for health. A 2007 study showed that men with higher “femininity” scores had lower risk of heart disease (Hunt et al., 2007). A 2016 study showed that young adults with high scores on a rating of “feminine gender-related characteristics” were more likely to experience a recurrence of acute coronary syndrome, whatever their biological sex (Pelletier et al., 2016; Pelletier et al., 2015).

These innovative studies demonstrate that gender is a critical determinant of health; they also demonstrate how challenging it is to measure. The assessment of gender in these studies depended in part on outdated gender identity constructs, such as the Bem Sex Role Inventory (Bem, 1974). Developed in 1974 using predominately white, higher socioeconomic US undergraduates, the Bem Sex Role Inventory uses now-outdated concepts of masculinity and femininity and is not included in state-of-the-art data collection protocols, such as those featured in the PhenX toolkit (see below). Historically, gender was conceptualized as a two-dimensional spectrum stretching from "masculine" to "feminine" understood as complementary opposites (Schiebinger, 1991). These antiquated concepts are too broad and imprecise to be useful in health research. We now know that gender is multidimensional and multifaceted, and that it changes over time (Connell, 2012). Collapsing different dimensions of gender into a single score risks replicating gender stereotypes and makes results less useful to medical researchers.

Method: Analyzing Gender

Gender is a capacious term: it can describe a society's norms and expectations; individual's beliefs, identities, and behaviors; and the ways those expectations and identities affect relations among people. Analyzing the influence of gender on health requires tools that disaggregate these different dimensions of gender and quantify them.

View General Method

 

Gendered Innovation 1: Revising Outdated Gender Measurements:

To address these limitations, an interdisciplinary team of researchers developed the Stanford Gender-Related Variables for Health Research (GVHR) tool. This case study is a summary of the paper published in the Biology of Sex Differences (Nielsen et al., 2021).

The first innovation was to operationalize the notion that gender is multidimensional. The goal was to capture relevant dimensions of gender for health research in a new instrument that offers more comprehensive and precise survey-based measures of gender in relation to health.

This decision has three major benefits. First, this tool enables more precision for researchers. A researcher may find that one gender-related trait, such as risk-taking or emotional intelligence, influences a health outcome but that other gender-related variables, such as work strain, do not. It may also allow researchers to see gender effects that would be masked by a single score. For example, a gender-related trait (such as caregiver strain) could negatively influence health outcomes, while another (such as social support) could positively influence health outcomes.

The second benefit is that conceptualizing gender as multidimensional and multifaceted may make the GHVR tool more relevant to trans*, gender-queer, and non-binary populations (Baker et al., 2021; National Academies, 2020). The GHVR team found, for example, that non-binary respondents reported many more experiences of discrimination than ciswomen or cismen. Those experiences cannot be understood as "masculine" or "feminine" and may have important impacts on health.

The final benefit is that, because gender norms, traits, and relations vary across and within cultures and change over time, these variables can be individually updated as needed, and culturally specific variables can be developed.

Gendered Innovation 2: Developing and Validating the Stanford Gender-related Variables for Health Research tool:

The GVHR survey measures 7 gender-related variables in 3 different categories:

• 1. Gender Norms
  
  • a. Caregiver strain captures perceived consequences of responsibility for unpaid, long-term caregiving to children, partners, friends, and elderly (excluding housework and caregiving occupations).
    b. Work strain captures job strain and emotional job demands.

• 2. Gender-Related Traits
  
  • a. Independence is a personality trait characterized by a focus on the person as an individual rather than part of a group and includes agency, self-confidence, and self-determination.
    b. Risk-Taking refers to a person's willingness to take physical, behavioral, and financial risks.
    c. Emotional Intelligence measures a person's ability to recognize and manage emotions.

• 3. Gender Relations
  • a. Social Support measures perceived satisfaction with the type and level of support available from friends, coworkers, relatives, partners, etc.
    b. Discrimination measures how often individuals felt discriminated against because of their gender in general, when getting hired, when at school, when receiving medical care, in other public settings, and in their family.
    

Each variable listed here provides specific information about how gender-related behaviors and attitudes contribute to health and disease processes, irrespective of—or in addition to—biological sex and self-reported gender identity, which are collected separately in the GVHR survey along with sexual orientation, age, ethnicity/race, relationship status, family status, income level, and educational background. Use of these gender-related variables in clinical and population research may also help health providers understand if gender factors play a role as treatment effect modifiers and thus need to be considered further in making treatment decisions.

GVHR questionnaire results capturing specific gender-related behaviors and attitudes. The figure displays the z-scores for the seven gender-related variables for respondents seeing themselves as men (green), women (orange) and non-binary or gender fluid (gray) in a sample of 1893 U.S. respondents. Click here for the Stanford GVHR survey.

To develop the GHVR survey, the team started with a systematic literature review that produced 74 gender-related questionnaires published in the English language literature between 1975 and 2015. From these, the researchers identified 11 gender constructs and developed 44 separate survey items. These items were assessed by experts and a diverse group of non-expert testers. The items were then validated in three diverse, adult, US populations, two internet-based (N = 2051; N = 2135) and one from a patient-research registry (N = 489). Exploratory and confirmatory factor analysis reduced these 11 variables to the final 7 listed above.

The gender-related items were also tested to see if they could predict self-reported physical and mental health. Caregiver strain and discrimination were associated with lower physical and mental health as well; social support was associated with higher mental health. Both caregiver strain and work strain were associated with smoking; risk-taking was associated with binge drinking, and discrimination was associated with vaping.