Assessment of Health Disparities and Sexual Orientation Response Choices Used in Two US National Population-Based Health Surveys, 2020‒2021

Abstract

Objectives. To (1) compare responses to 2 survey questions designed to measure sexual orientation and (2) understand how variation in responses is associated with mental health.

Methods. Data were from the National Longitudinal Study of Adolescent to Adult Health (Add Health) Sexual Orientation/Gender Identity, Socioeconomic Status, and Health Across the Life Course (SOGI-SES) study (2020–2021) in the United States. We used the adjusted Wald test to compare proportions of respondents who were (1) categorized as heterosexual or straight and sexual minorities using the sexual orientation questions designed for the Add Health study and the National Health Interview Survey (NHIS) and (2) diagnosed with depression or anxiety or panic disorder.

Results. The Add Health question detected more than twice as many sexual minority respondents as the NHIS question. Those who responded as sexual minorities to the Add Health question but as heterosexual or straight to the NHIS question, primarily “mostly heterosexuals,” had mental health outcomes that were more like those who were consistently classified as sexual minorities versus those consistently classified as heterosexual or straight.

Conclusions. Current measures of sexual orientation in national-level surveys may underestimate the sexual minority population and sexual orientation‒related health disparities.

Public Health Implications. Results illustrate the need for further research to expand measurement of sexual orientation on population-based health surveys. (Am J Public Health. 2024;114(12):1375–1383. https://doi.org/10.2105/AJPH.2024.307839)

The lesbian, gay, bisexual, and transgender (LGBT) population continues to grow, with recent estimates suggesting that LGBT people represent 7.2% of the US adult population.¹ This proportion is even greater among younger generations, with 19.7% of Generation Z (born 1997–2004) and 11.2% of millennials (born 1981–1996) identifying as LGBT compared with 3.3% among Generation X (born 1965–1980).¹ However, these figures vary across surveys,² reflecting, at least in part, differences in measurement approaches and specific measures used to identify LGBT people (e.g., measuring both sexual orientation and gender identity with 1 question, different response options).³

Sexual orientation is a multidimensional construct that includes aspects of sexual identity, attraction, and behavior, making it particularly difficult to measure comprehensively.³ Most measurement efforts, including those recommended by the recent report from the National Academies of Sciences, Engineering, and Medicine (NASEM), have focused on the identity dimension of sexual orientation, which is often considered the most relevant for understanding and responding to health disparities and inequities.⁴ This is important because lesbian, gay, and bisexual (LGB) populations have been shown to experience significantly poorer health outcomes relative to non-LGB people, particularly related to mental health, which are often attributed to minority stress resulting from stigma, discrimination, and inequitable treatment.^3,5–8

However, research also shows that some people who report same-sex attraction or behavior do not identify as lesbian, gay, bisexual, or another minority sexual orientation when asked about sexual orientation identity.^9,10 These studies further indicate that those who do not identify with traditional labels such as gay, lesbian, or bisexual may represent an “invisible” sexual minority population that experiences health outcomes that are more similar to other sexual minorities yet are missed by programs and policies intended to address these disparities.^{1,3,5,11–13} For example, one study using data from the National Longitudinal Study of Adolescent to Adult Health (Add Health) found that those who characterized their sexual orientation as “mostly heterosexual” comprised the largest sexual minority group among both males and females.¹⁴ Furthermore, levels of perceived stress and depressive symptoms of mostly heterosexual respondents were more similar to those of other sexual minority groups than they were to those of 100% heterosexual respondents.¹⁴ As a result, more research focused on understanding who is detected through various measures of sexual orientation may help not only to identify those who could benefit from targeted services and supports but also to inform efforts to measure sexual orientation in population health surveys.

Thus, based on the NASEM report’s recommendation to further improve the quality and inclusivity of current sexual orientation identity measures, the purpose of this study was to (1) describe and compare responses from respondents asked to complete 2 different survey questions designed to measure sexual orientation and (2) understand how variation in responses to these questions are associated with mental health outcomes. We hypothesized that (1) the sexual orientation question that measured sexuality by approximating a continuum would detect more sexual minority respondents compared with the item that uses a narrower set of specific sexual orientation identity labels, and (2) sexual minorities who were undetected by either question (i.e., only identify as heterosexual or straight on 1 of the 2 questions) would exhibit similar mental health outcomes compared with those who were categorized as sexual minorities on both measures.

METHODS

The National Longitudinal Study of Adolescent to Adult Health (Add Health) is a large, nationally representative sample of more than 20 745 in-school adolescents who were in the 7th through 12th grades during the 1994–1995 school year (wave I).¹⁵ The cohort has been followed longitudinally, with wave V completed in 2016–2019 (n = 12 300, ages 33–44 years), and wave VI currently in the field. The Sexual Orientation/Gender Identity, Socioeconomic Status, and Health Across the Life Course (SOGI-SES) study is an ancillary, online survey fielded in the fall of 2020 and spring of 2021.¹⁶ The sampling frame included all living Add Health wave V respondents. Those who identified as mostly heterosexual, bisexual, mostly homosexual, or homosexual; reported same-sex partners at waves III, IV, or V (see Measures); or were discordant on sex assigned at birth and gender expression (e.g., androgynous or gender nonconforming) at wave V were all solicited for participation in SOGI-SES. We refer to these respondents as sexual and gender minorities (SGMs).

The study also included a comparison sample of 1500 non-SGM Add Health respondents who identified as completely heterosexual; did not report same-sex partners at waves III, IV, or V; and were conforming in their gender expression. SGMs were selected with certainty. To obtain the comparison sample, remaining non-SGMs were selected across 16 strata defined by sex assigned at birth, race/ethnicity, and percentage of the federal poverty level (according to the US Census) at wave V to achieve a comparison sample that was diverse across race/ethnicity and economic status and was comparable to the SGM population on proportion female sex assigned at birth.¹⁶ In total, 4661 wave V respondents were selected to participate in the SOGI-SES survey. The final sample included 2614 respondents aged 37 to 46 years—56.5% of the selected sample. We conducted nonbias analyses, and determined relative bias to be moderately small.¹³ Thus, weighted estimates from the SOGI-SES study are representative of wave V respondents who were in school and in the 7th through 12th grades in 1994 to 1995.

Measures

Sexual orientation

As a part of the SOGI-SES survey, all respondents were asked 2 survey questions that measured sexual orientation.

One of these items was the Add Health question, which has been asked of all Add Health respondents from wave III in 2001 on and defines sexuality approximating a continuum:

“Please choose the description that best fits how you think about yourself.”

• •100% heterosexual (straight)
• •Mostly heterosexual (straight), but somewhat attracted to people of your own sex
• •Bisexual, that is, attracted to men and women equally
• •Mostly homosexual (gay), but somewhat attracted to people of the opposite sex
• •100% homosexual (gay)
• •Not sexually attracted to either males or females

The other item was the National Health Interview Survey (NHIS) question, which was developed at the National Center for Health Statistics¹⁷ to measure sexual orientation identity and has been used on the NHIS since 2013.¹⁸ This question was modified slightly by the SOGI-SES investigators, including adding the term bisexual to define what is not straight and adding the last 2 response options to the question.

“Which of the following best represents how you think of yourself?”

• •Gay or lesbian
• •Straight, that is, not gay, lesbian, or bisexual
• •Bisexual
• •Something else
• •I am not sure of my sexuality (I am “questioning” my sexuality)
• •I am not sure what this question is asking

Each respondent received both questions sequentially, but the order in which these 2 items were presented was randomized so that half of respondents received the Add Health question first and vice versa.

Two additional variables, “Undetected by Add Health” and “Undetected by NHIS,” were created to categorize those respondents who may have self-identified as a sexual minority on only 1 of the items (see Figure 1 for visual representation of these categories). For the purposes of this article, those who selected “I am not sure what this question is asking” to the NHIS question (n < 20) were excluded from the constructed variable and related analyses.

FIGURE 1—

Categorization of Two Sexual Orientation Question Responses From the Sexual Orientation/Gender Identity, Socioeconomic Status, and Health Across the Life Course (SOGI-SES) Study: United States, 2020–2021

Note. Add Health = National Longitudinal Study of Adolescent to Adult Health; NHIS = National Health Interview Study; SM = sexual minority.

Respondents were included in the “heterosexual or straight” category for each question if they selected “100% heterosexual (straight)” for the Add Health question and “Straight, that is, not gay, lesbian, or bisexual” for the NHIS question. Respondents were included in the “Sexual minority” category if they selected any other response besides heterosexual or straight for both questions. For the Add Health question, the “Undetected sexual minority” category included those who selected “100% heterosexual (straight)” for the Add Health question but a sexual minority response option for the NHIS question. Similarly, the “Undetected sexual minority” category for the NHIS question included those who selected “Straight, that is, not gay, lesbian, or bisexual” for the NHIS question but a sexual minority response option for the Add Health question.

Mental health

Respondents were asked, “Has a doctor, nurse, or other health care provider ever told you that you have or had any of the following?” and given a series of health conditions from which they could select “Yes” or “No.” This analysis included lifetime diagnoses of depression and anxiety or panic disorder.

Covariates

Given significant differences in the proportion of men and women who identify as LGB,¹ analyses were completed separately by gender. Sex assigned at birth was measured using an item that asked, “What sex were you assigned at birth, on your original birth certificate?” Response choices were “male” and “female.” Gender identity was measured using the following item:

“What is your current gender identity? Select all that apply.”

• •Male
• •Female
• •Transgender
• •Gender nonbinary/genderqueer
• •I am not sure of my gender identity (I am “questioning” my gender identity)
• •I do not know what this question is asking [coded as missing]

These 2 items were used to construct the gender variable. We included respondents in the “cisgender male” category if they selected “male” for sex assigned at birth and selected only “male” for gender identity. Similarly, we included respondents in the “cisgender female” category if they selected “female” for sex assigned at birth and selected only “female” for gender identity. We excluded those whose sex assigned at birth did not align with their gender identity or selected more than 1 gender identity from this analysis (n < 35) because there were too few transgender respondents to provide group-specific results about sexual orientation response patterns.

Analyses

We used descriptive statistics (frequencies and percentages) to present responses to the Add Health and NHIS questions and the lifetime mental health diagnosis variables. Next, we used the adjusted Wald test to compare the proportions of respondents who (1) were categorized as undetected sexual minorities and (2) had ever been diagnosed with depression or anxiety or panic disorder by these sexual orientation categories and by gender. We used the Bonferroni correction to adjust for multiple comparisons, with the P level set to .05.

All SOGI-SES respondents have valid sampling weights. We restricted statistical analyses to those respondents who had complete data on all variables of interest (< 5% missing). For all analyses, we used sampling weights and adjusted variance estimates for the Add Health complex survey design to yield population representative estimates and completed them with Stata version 17.0 (StataCorp, College Station, TX).

RESULTS

Of 2576 respondents with complete data on the sexual orientation, sex, and gender items, 50.2% were cisgender males, and 49.8% were cisgender females. The mean age was 40.8 years (SD = 1.9; range = 37–46). A larger proportion of cisgender male (93.9%) and cisgender female respondents (91.2%) identified as “straight” in response to the NHIS question compared with cisgender males (92.2%) and cisgender females (79.1%) who identified as “100% heterosexual” in response to the Add Health question. For both questions, a larger proportion of cisgender females identified as sexual minorities compared with cisgender males. Comparisons are illustrated in Figure 2 for cisgender males and cisgender females. The vertical bars show how a given Add Health answer maps onto answers to the NHIS question. For example, of the 3.2% of cisgender males who selected “gay/lesbian” for the NHIS question, 2.7% selected “bisexual” in response to the Add Health question, 13.6% selected “mostly homosexual,” and 83.7% selected “100% homosexual.”

FIGURE 2—

Sexual Orientation/Gender Identity, Socioeconomic Status, and Health Across the Life Course (SOGI-SES) Study Responses to Two Different Sexual Orientation Questions by (a) Cisgender Male and (b) Cisgender Female: United States, 2020‒2021

Note. Add Health = National Longitudinal Study of Adolescent to Adult Health; NHIS = National Health Interview Study. Percentages are weighted to yield population-representative estimates and may not sum to 100% because of rounding.

Table 1 shows the proportions of respondents in the combined sexual orientation category variables. For both questions, 85.6% were identified as heterosexual or straight. The Add Health question detected more than twice as many sexual minorities as the NHIS question (14.4% vs 6.8%). Less than 0.1% of respondents were undetected by the Add Health question, while 7.6% of respondents were undetected by the NHIS question. These proportions did not differ by the order in which the questions were asked (Add Health first vs NHIS first; NHIS: χ² = 0.34; P = .66; Add Health: χ² = 0.18; P = .77). Nearly all (97.8%) of those undetected by NHIS had responded as “mostly heterosexual” to the Add Health question. Cisgender female respondents were significantly more likely than cisgender male respondents to be undetected by NHIS (12.8% vs 2.6%; F = 38.73; P < .001). There were no statistically significant differences between cisgender males and cisgender females for the Add Health question (< 0.1% vs < 0.1%; F = 0.16; P = .69). Given the extremely small number of undetected sexual minorities by the Add Health question, additional analyses could only be completed for the NHIS question.

TABLE 1—

Sexual Orientation Categories by Question and Gender in the Sexual Orientation/Gender Identity, Socioeconomic Status, and Health Across the Life Course (SOGI-SES) Study: United States, 2020‒2021

	Add Health Question			NHIS Question
	Cisgender Male, %	Cisgender Female, %	Total, %	Cisgender Male, %	Cisgender Female, %	Total, %
Heterosexual or straight	92.2	79.0	85.6	92.2	79.0	85.6
Sexual minority	7.8	20.9	14.4	5.3	8.3	6.8
Undetected sexual minority	< 0.1	< 0.1	< 0.1	2.6	12.8	7.6

Note. Add Health = National Longitudinal Study of Adolescent to Adult Health; NHIS = National Health Interview Study. Percentages are weighted to yield population-representative estimates and may not sum to 100% because of rounding. The sample size was n = 2558.

Results of the mental health diagnosis analyses for the undetected by NHIS variable are shown in Table 2. Overall, 30.2% of heterosexual or straight respondents, 48.4% of sexual minorities, and 51.5% of undetected sexual minorities endorsed a lifetime depression diagnosis. Adjusted Wald tests comparing these proportions indicated that both sexual minorities (F = 18.49; P < .001) and undetected sexual minorities (F = 13.91; P < .001) were significantly more likely than heterosexual or straight respondents to endorse a depression diagnosis, and the difference between sexual minorities and undetected sexual minorities was not significantly different (F = 0.24; P = 1.00). After stratifying by gender, cisgender female sexual minorities (52.2%; F = 10.42; P < .01) were significantly more likely than cisgender female heterosexual or straight respondents (35.8%) to have ever been diagnosed with depression. Cisgender females who were undetected sexual minorities (52.3%) were not significantly different from cisgender female heterosexual or straight respondents after the Bonferroni correction for multiple comparisons (F = 5.73; P = .05) or cisgender female sexual minorities (F = 0.00; P = 1.00). Among cisgender male respondents, sexual minorities were significantly more likely than heterosexual or straight respondents (42.8% vs 25.7%; F = 6.71; P = .03) to have received a depression diagnosis. No such statistically significant differences in depression diagnosis emerged when comparing cisgender males who were undetected sexual minorities (47.6%) to cisgender male heterosexual or straight respondents (F = 3.63; P = .18) or to cisgender male sexual minorities (F = 0.15; P = 1.00).

TABLE 2—

Adjusted Wald Tests Comparing Proportions Who Endorsed Ever Receiving Mental Health Diagnoses in the Sexual Orientation/Gender Identity, Socioeconomic Status, and Health Across the Life Course (SOGI-SES) Study, by Sexual Orientation Category and Gender: United States, 2020‒2021

	Heterosexual or Straight, % (95% CI)	Sexual Minority, % (95% CI)	Undetected by the NHIS Question, % (95% CI)
Depression
Cisgender male	25.7 (18.7, 34.1)a	42.8 (31.7, 54.6)b	47.6 (26.5, 69.6)
Cisgender female	35.8 (29.3, 42.8)a	52.2 (43.7, 60.6)b	52.3 (41.2, 63.3)
Overall	30.2 (25.0, 36.1)a,c	48.4 (41.3, 55.6)b	51.5 (41.3, 61.6)b
Anxiety or panic disorder
Cisgender male	30.1 (23.1, 38.1)a	46.3 (36.6, 56.3)b	42.9 (22.2, 66.5)
Cisgender female	37.1 (30.5, 44.2)a	54.7 (46.1, 63.0)b	47.6 (37.3, 58.0)
Overall	33.3 (28.2, 38.7)a	51.3 (44.8, 57.7)b	46.8 (37.3, 56.4)

Note. CI = confidence interval; NHIS=National Health Interview Survey. Gender diverse identity category was excluded because of small cell sizes. Percentages are weighted to yield population representative estimates and may not sum to 100% because of rounding. The sample size was n = 2472.

^a Statistically significant difference (P < .05) from “sexual minority” after the Bonferroni correction.

^b Statistically significant difference (P < .05) from “heterosexual or straight” after the Bonferroni correction.

^c Statistically significant difference (P < .05) from "undetected by the NHIS question” after the Bonferroni correction.

The overall pattern was similar for anxiety or panic disorder. A total of 33.3% of heterosexual or straight respondents, 51.3% of sexual minorities, and 46.8% of undetected sexual minorities endorsed a lifetime anxiety or panic disorder diagnosis. Adjusted Wald tests showed that sexual minorities were significantly more likely than heterosexual or straight respondents (F = 18.66; P < .001) to have ever been diagnosed with anxiety or panic disorder. Undetected sexual minorities were not significantly different from heterosexual or straight respondents after the Bonferroni correction (F = 5.87; P = .05) or from sexual minorities (F = 0.69; P = 1.00). Similarly, after stratifying by gender, the only statistically significant differences in the proportions endorsing an anxiety or panic disorder diagnosis were between sexual minorities and heterosexual or straight respondents among both cisgender male (46.3% vs 30.1%; F = 6.99; P = .03) and cisgender female respondents (54.7% vs 37.1%; F = 10.51; P < .01).

DISCUSSION

As hypothesized, we found that the NHIS survey question detected significantly fewer sexual minority respondents compared with the Add Health survey question. In addition, the mental health outcomes of those who reported being sexual minorities on one question but not the other (e.g., heterosexual or straight to the NHIS question and sexual minority to the Add Health question) were more similar to those who reported being sexual minorities to both questions than to those who reported being heterosexual or straight to both questions. Such findings suggest that a large group of sexual minorities are undetected by measures that offer no options in between heterosexual and bisexual, which may lead to an underestimation of the scope and prevalence of health disparities. Importantly, this research further illustrates the critical need to expand measurement of sexual orientation.^4,19 A singular focus on identity may exclude those who experience same-sex attraction or behaviors who are at similar or greater risk for negative health outcomes, even though they do not identify with a sexual minority label. Therefore, future work in this area should consider (1) testing response options that allow for identification in between straight and bisexual and (2) also measuring current sexual attraction and adult lifetime behavior, in addition to identity, when collecting information about sexuality on health surveys.

Limitations

Study findings should be interpreted within the context of the following limitations. First, sample size limited our ability to detect differences in smaller sexual minority subgroups. Similarly, sample size prevented us from understanding how these patterns may vary by other demographic characteristics such as race and ethnicity. This is particularly important given the large body of research showing significant health inequities experienced by people of color who also identify as lesbian, gay, bisexual, or any other minority sexual orientation.^20,21

It is also important to acknowledge the limitations of the Add Health sexual orientation question. Importantly, this question asks about both identity and attraction in the same item, which may be interpreted differently and thus lead to less-precise estimates.^3,11 In addition, certain terms, such as “heterosexual” and “homosexual,” may be less effective in detecting sexual minority populations as they do not reflect the terms used by more contemporary cohorts.¹¹ A further limitation is that SOGI-SES respondents have responded to the Add Health question previously, possibly as many as 3 previous times. Present findings may not generalize to samples who are answering a survey question for the first time. Finally, the SOGI-SES study was conducted only in English, so these results do not necessarily represent the experiences of those who are monolingual in other languages. Despite these limitations, the Add Health question did detect a larger sexual minority group that would have otherwise been uncounted, suggesting the importance of including other dimensions of sexual orientation (i.e., attraction, behavior) and approximating a continuum when estimating the size of the sexual minority population.

Public Health Implications

The results of this study show how current 1-dimensional measures of sexual orientation in national-level surveys may lead to underestimates of the sexual minority population and, therefore, further underestimate the significant health disparities they experience. Thus, further research on other measures that consider response options that reflect a broader continuum of sexuality are needed to inform health policy and service planning to meet the needs of the sexual minority population.