Allostatic Load Among Black Sexual Minority Women

Abstract

Background: Allostatic load is the physiological deterioration that accumulates as the body responds to stress, resulting in disparities in chronic disease. Although perceptions of stress vary, marginalization and social disadvantage are associated with elevated allostatic load. Allostatic load is understudied in the multiply marginalized populations of sexual minority Black women.

Methods: We used data from six waves of the National Health and Nutrition Examination Survey (2001–2010, 2015–2016) to identify heterosexual (n = 78), lesbian (n = 21), and bisexual (n = 57) Black women. We quantified allostatic load using nine biomarkers, and compared mean allostatic load across the three groups, adjusting for age, educational attainment, income, and country of birth to account for possible confounding.

Results: We found no significant differences in allostatic load among heterosexual, lesbian, and bisexual Black women.

Conclusions: These findings suggest that sexual orientation may not contribute to within-group differences in allostatic load among Black women, a group previously noted to have elevated allostatic load.

Introduction

Allostatic load is the cumulative physiological depreciation that occurs as the body continuously responds to stress through allostasis.¹ Although allostasis is an essential process that allows the body to adapt when confronted with internal and external stress, repeated activation of this process can have adverse effects on an individual's health.^2,3 As various systems in the body become dysregulated under allostatic load, the likelihood of unfavorable clinical outcomes increases.⁴ For instance, dysregulation of the sympathetic nervous system and the hypothalamic–pituitary–adrenal axis due to overactivation of allostasis is associated with declines in cognitive and physical performance.⁴

Previous studies have demonstrated a strong association between allostatic load and social disadvantage. Increased neighborhood poverty^5,6 and lower educational attainment and income⁷ have been associated with elevated allostatic load among Blacks. Racial differences in allostatic load are likely associated with these social determinants, as Blacks have exhibited elevated allostatic load compared with Whites.⁸ Moreover, Black women were shown to have higher allostatic load compared with Black men, White men, and White women across all age groups.⁹ These data are consistent with the weathering hypothesis, which proposes that Black women experience a deterioration in health by virtue of repeated cumulative exposure to social, economic, and political disadvantage.¹⁰

Among sexual minorities (individuals who do not identify as heterosexual), social disadvantage may also be associated with a decline in health. The sexual minority stress model suggests that sexual minorities are continuously confronted with stress due to heightened exposure to discrimination, prejudice, and stigma.¹¹ This chronic stress may manifest distally, from stressors such as discrimination, or proximally, from stressors such as internalized homophobia, and may lead to unfavorable health outcomes.¹¹ More recently, a number of studies have explored possible relationships between allostatic load and sexual orientation.^12–14 Among men of varying racial and ethnic identities, bisexual men have demonstrated elevated allostatic load compared with heterosexual men.¹² Conversely, allostatic load does not seem to differ by sexual orientation among women of varying racial and ethnic identities.^12,14

Although research regarding allostatic load and its possible associations with sexual orientation and racial/ethnic identification continues to grow, there remains a paucity of published work on allostatic load and sexual orientation among Black women. Although Black women of all sexual orientations are confronted with the effects of weathering, intersectionality suggests that social identities are interconnected and interdependent, generating distinct experiences of disadvantage and privilege.¹⁵ Therefore, as a consequence of living with multiply marginalized identities, namely identifying as Black, a woman, and a sexual minority, Black sexual minority women face unique combinations of stressors that could have profound impacts on their health status.¹⁶ Using data from the National Health and Nutrition Examination Survey (NHANES), this study examined differences in allostatic load among Black women who self-identify as heterosexual, lesbian, or bisexual.

Methods

Data set

NHANES is a cross-sectional survey conducted by the National Center for Health Statistics. NHANES collects data from a representative sample of noninstitutionalized civilians. The data include interview questionnaires, physical examinations, and laboratory analyses of blood and urine samples. NHANES releases these data in cycles every 2 years. Because the data in NHANES are publicly available and the participants have been deidentified, a review of this study by an institutional review board was not required. We included data from 2001 (the first wave that sexual orientation data were included) to 2016 (the latest wave that data were available from). In the 2011/2012 and 2013/2014 waves, NHANES did not report data on C-reactive protein, a key component of allostatic load measure in this study. Therefore, data from these waves were not included.

Study participants

A total of 7,258 non-Hispanic Black women were included in NHANES during the six waves of data we were able to analyze between 2001 and 2016. Among these, 1,997 completed the sexual behavior portion of the interview and had a valid response to the sexual orientation question, specifically identification as either heterosexual, lesbian, or bisexual. We excluded 132 pregnant women from the study due to biological changes during pregnancy that can interfere with the biomarkers used to index allostatic load.^17,18 We also excluded 545 participants who were missing at least one of the biomarkers used to index allostatic load. Therefore, we identified 1,320 eligible, non-Hispanic Black (hereafter “Black”) participants, of which 1,242 identified as heterosexual, 21 identified as lesbian, and 57 identified as bisexual. To account for the notable differences in sample size across the three sexual orientation groups, we implemented an individual age matching algorithm that matched heterosexual participants to bisexual and lesbian participants based on age. We then eliminated any heterosexual participant who did not match in age to a bisexual or lesbian participant. Our final analytic sample thus became 156 Black women of whom 78 identified as heterosexual, 21 identified as lesbian, and 57 identified as bisexual.

Sexual orientation

In 2001, NHANES began collecting data on the sexual orientation of participants aged 14 years and older. The sexual behavior questionnaire includes data on lifetime sexual history, sexual history during the 12 months before the survey, and sexual orientation. With slight variations from cycle to cycle, female participants are asked to describe their sexual orientation by selecting one of the following options: homosexual/lesbian/gay, heterosexual/straight, bisexual, something else, don't know, not sure, and refused.

Allostatic load

To calculate allostatic load, we used nine biomarkers previously used in studies measuring allostatic load with NHANES data.^7,9,12,19 These biomarkers include cardiovascular (systolic blood pressure, diastolic blood pressure, and resting heart rate), immune (serum albumin and C-reactive protein), and metabolic markers (glycosylated hemoglobin, body mass index (BMI), total cholesterol, and high-density lipoprotein (HDL) cholesterol). As in prior research,^12,19 we used the following clinical ranges and cutoffs to classify each biomarker: BMI ≥30 kg/m², C-reactive protein >0.3 mg/dL, glycosylated hemoglobin ≥6.4%, HDL <40 mg/dL, diastolic blood pressure ≥90 mm, systolic blood pressure ≥140 mm, resting heart rate ≥90 beats/min, total cholesterol ≥240 mg/dL, and serum albumin <3.8 g/dL. An individual received a “1” if the biomarker was less favorable than the clinical cutoff and a “0” if the biomarker was at or more favorable than the clinical cutoff. Total allostatic load was calculated by adding all counts, resulting in a possible range of 0 to 9 for each individual.

NHANES includes data on participants' medications used to treat hypertension, hyperlipidemia, or diabetes at the time of interview. Regardless of laboratory values, all individuals who reported taking medication for hyperlipidemia received a “1” for the total cholesterol biomarker. Similarly, individuals who reported taking medication for hypertension or diabetes received a “1” for systolic and diastolic blood pressure, or glycosylated hemoglobin, respectively.

Covariates

We identified age, educational attainment, income, and country of birth as covariates based on prior literature^20,21 and the plausibility of not being an intermediary factor between sexual orientation and allostatic load. Age was defined as years of age at the time of the baseline NHANES examination. Educational attainment was self-reported and measured as the highest grade, level of school, or degree completed at the time of interview. Income was self-reported as a range in dollars. NHANES has amended its measurement of country of birth across waves. These means of measurement have included identifying participants from the United States, Mexico, other Spanish-speaking countries, and non-Spanish-speaking countries. To maintain uniformity of this variable among the different cycles, we transformed country of birth into a dichotomous variable of “born in the United States” and “born outside of the United States.”

Analysis

To account for unequal sampling probabilities and the hierarchical sampling structure of NHANES data, we weighted and clustered all analyses using standard procedures described in the NHANES documentation.²² We conducted a one-way analysis of variance (ANOVA) to compare mean allostatic load (composite score and individual biomarker components) across the three sexual orientation groups. Linear modeling was used to measure the association between allostatic load and sexual orientation, after adjusting for age as a linear variable. We also conducted a sensitivity analysis that included covariates that are potentially part of the causal path between sexual orientation and cross-sectional allostatic load: educational attainment and income. Owing to the relative homogeneity of the participants' country of birth (<10% of the participants were born outside the United States), we did not include country of birth as an adjustment variable. Although some previous studies have combined groups of sexual minority women for analysis, we sought to examine distinctions between lesbian and bisexual women; moreover, we were adequately powered to perform such analyses. All analyses were performed using Stata 15.

Results

Participant characteristics

Table 1 presents descriptive statistics of the respondents. As a result of age matching, heterosexual women (31.5 years) had similar age to lesbian (31.3 years) and bisexual women (31.7 years). Bisexual- and lesbian-identified women were less likely to have at least completed high school (75% and 76%) than heterosexual women (82%). The income distributions for Black women were similar across sexual orientation, and fewer than 10% of all sexual orientation groups were born outside the United States.

Table 1.

Demographic Characteristics of Black Women, Ages 20–59 Years, by Sexual Orientation with Age Matching

	Heterosexual n = 78 M (SD) or %	Lesbian n = 21 M (SD) or %	Bisexual n = 57 M (SD) or %
Age	31.5 (8.9)	31.3 (8.6)	31.7 (9.3)
Educational attainment
Less than high school	18.5	24.1	25.4
High school graduate or GED	17.1	13.3	27.9
Some college or associate degree	40.5	39.7	35.1
College graduate or above	23.9	22.9	11.6
Family income
Less than $20,000	29.4	45.4	37.6
$20,000 to $74,999	59.8	46.3	54.6
$75,000 or more	10.8	8.3	7.9
Country of birth
Outside United States	6.4	4.4	5.2

GED, general educational development; M, mean; SD, standard deviation.

Allostatic load

Table 2 gives the weighted mean allostatic load by sexual orientation and the weighted proportion of participants who had unfavorable measures for each biomarker. Among Black women, heterosexuals had a slightly higher mean allostatic load (1.9) than both bisexual (1.6) and lesbian women (1.5). Notably, lesbian women had higher diastolic blood pressure and more unfavorable HDL cholesterol than both heterosexual and bisexual women. Heterosexual women had higher heart rate, metabolic markers (BMI and total cholesterol), and immune markers (serum albumin and C-reactive protein) than both groups of sexual minority women.

Table 2.

Allostatic Load and Frequency of Unfavorable Biomarkers Among Black Women, Ages 20–59 Years, by Sexual Orientation with Age Matching

	Heterosexual n = 78 M (SD) or %	Lesbian n = 21 M (SD) or %	Bisexual n = 57 M (SD) or %
Allostatic load	1.9 (1.6)	1.6 (1.6)	1.5 (1.6)
Systolic blood pressure ≥140 mm Hg and/or being on an antihypertensive medication	7.8	6.5	8.7
Diastolic blood pressure ≥90 mm Hg and/or being on an antihypertensive medication	7.8	15.7	7.0
Heart rate ≥90 beats/min	10.5	5.1	8.6
Glycosylated hemoglobin ≥6.4% and/or on an antidiabetic medication	20.6	20.4	12.3
BMI ≥30 kg/m2	51.4	43.6	40.7
Total cholesterol ≥240 mg/dL and/or on medication for hypercholesterolemia	10.8	4.6	8.0
HDL cholesterol <40 mg/dL	13.2	19.3	6.2
Serum albumin <3.8 g/dL	19.7	8.5	12.1
C-reactive protein >0.3 mg/dL	50.8	39.2	41.5

BMI, body mass index; HDL, high density lipoprotein.

Linear models

Table 3 presents the results from the linear models. In unadjusted analyses of the associations between allostatic load and sexual orientation (Supplementary Fig. S1), allostatic load was not significantly associated with sexual orientation. Upon adjustment for age, the relationship between allostatic load and sexual orientation remained statistically nonsignificant, though age was significantly associated with allostatic load (Table 3). Further adjustments for educational attainment and income did not alter the relationship between allostatic load and sexual orientation.

Table 3.

Adjusted Linear Models with Age Matching for Mean Difference in Allostatic Load by Sexual Orientation Among Black Women, Ages 20–59 Years

	Age adjustment Adj β (95% CI)	Socioeconomic adjustment Adj β (95% CI)
Sexual orientation
Heterosexual	Reference	Reference
Lesbian	−0.27 (−0.96 to 0.42)	−0.27 (−0.95 to 0.41)
Bisexual	−0.48 (−1.03 to 0.07)	−0.48 (−1.02 to 0.07)
Covariates
Age	0.07 (0.05 to 0.09)	0.07 (0.05 to 0.09)
Educational attainment		−0.01 (−0.24 to 0.22)
Income		−0.11 (−0.49 to 0.28)

CI, confidence interval.

Sensitivity analysis

We ran the primary analyses on the full non-age-matched sample for comparison. We did not find a significant difference in the results (Supplementary Tables S1–S4).

Discussion

This study examined differences in allostatic load by sexual orientation using a population-based sample of Black women who self-identified as heterosexual, lesbian, or bisexual. We found no significant differences in allostatic load among these three groups of women. We believe that the results demonstrate that the effects of stress associated with multiply marginalized identities may be more nuanced and unique than an additive approach would imply.

The similarity in allostatic load among Black women of varying sexual orientations is consistent with previous findings in which variation in allostatic load was detected among men but not among women.^12,14 Using data from NHANES, Mays et al. identified no differences in allostatic load among women across sexual orientation. Similarly, Juster et al. found no difference in allostatic load by sexual orientation in a sample of women in the Montréal region.

In a different study, Juster et al. assessed the relationship between allostatic load and gender roles, detecting within-sex differences in allostatic load among men but not among women.²³ Juster et al. found that the procurement of masculine and feminine gender roles may have a protective effect for men but may not be as beneficial for women.²³ Many women have acquired traditionally masculine gender roles to gain socioeconomic advancement through education and employment. However, despite this supposed advancement, women in the United States still face substantial socioeconomic disadvantage compared with men.^24–26 These realities suggest that the social inequalities and pervasive marginalization that women face could impact their stress experiences and, thus, their allostatic load, irrespective of other aspects of their identities such as their gender roles or sexual orientation.

In addition to sexism and gender inequality, Black women are confronted with racism and antiblackness that may have profound impacts on their health.^27,28 The stress that Black women face due to multiple forms of discrimination based on their gender and racial identification may influence allostatic load, outweighing the possible influence of sexual orientation-based stress. When compared with Black men, White women, and White men, Black women have demonstrated higher allostatic load, irrespective of sexual orientation.⁹ Furthermore, despite experiencing multiple forms of discrimination, Black women have been shown to have high levels of resilience.¹⁶ These high levels of resilience could be protective against some stress-induced health outcomes. Although this resilience may not entirely avoid allostatic load, the impact may be a reduction in perceivable within-group differences in allostatic load.

Strengths and limitations

This study is based on the largest population-based sample available to address our research question of interest, combining six waves of data from NHANES between 2001 and 2016. Although the sample of Black bisexual (57) and lesbian (21) women identifiable in this data set is not large, it has the advantage of being based on a population-based sample and, thus, is representative of Black women living in the United States during this period. This study also intentionally focused on a more nuanced and intersectional analysis of Black women's sexuality than is available in many similar studies. Rather than eliding lesbian and bisexual Black women together, or even Black gay and bisexual men with Black lesbian and bisexual women together to achieve a more robust sample size (and thus resulting in overly averaged effect estimates), we restricted our focus to this population of conceptual interest. We hope that this focus brings attention to the need for greater data resources to understand these (and other) intersectional populations at appropriate levels of detail.

However, this relatively small sample of Black sexual minority women means that only large differences in mean allostatic load would be identified as significant in a statistical sense. Although allostatic load levels across sexual orientation showed no significant differences after adjusting for age among the Black women in this sample, it remains plausible that clinically relevant differences too modest to be detected in this sample would be identifiable in a larger sample, given that lesbian and bisexual women have reported and evidenced worse physical health than heterosexual women.^29,30 Furthermore, the cross-sectional nature of NHANES data means that it is challenging to establish a causal linkage between sexual orientation and allostatic load in this population. However, identification as a sexual minority is usually established in adolescence or early adulthood,³¹ younger than most adults who were enrolled in these NHANES cohorts, and reverse causation is conceptually implausible. We cannot entirely rule out differential survival in this prevalence sample, although poor survival among Black sexual minority women with higher allostatic load would have to be quite dramatic, given the younger age distribution of these women, to explain these findings.

Response rates to NHANES are fairly high for such an intense study,³² with over half of potential respondents agreeing to participate. Furthermore, data are weighted to account for nonresponse biases due to the age and sex distribution of respondents. However, if bisexual and lesbian Black women were differentially less likely to participate, this would not be reflected in the weighting scheme. Furthermore, if Black sexual minority women with high allostatic load were differentially less likely to agree to participate in NHANES due to past experiences of discrimination and/or mistrust of medical research,^33,34 this could potentially explain the similar mean allostatic loads we found.

Sexual minority Black women may, for reasons related to a lack of trust in medicine as already cited, decline to share their sexual minority status as lesbian or bisexual with a government-sponsored research effort like NHANES. If this was particularly true for lesbian and bisexual Black women with prior experiences of discrimination resulting in higher allostatic loads, these women may have been counted among the heterosexual population, resulting in more similar allostatic load levels than would have been the case had all women felt equally comfortable sharing their sexual orientation with researchers. Similarly, those Black women who feel comfortable acknowledging being lesbian or bisexual in the setting of NHANES, where data collection often takes place in the presence of family members, may be particularly well adapted to respond to discrimination in ways that minimize long-term impacts of stressful responses.

In addition, sexual orientation is not a fixed construct, and as a result, Black women may interpret their experiences and desires differently across the lifespan.³⁵ Because NHANES is a set of cross-sectional samples, it was not possible for us to examine the potential influence of evolving sexual orientation identity labels on allostatic load levels. Similarly, allostatic load is variable across the lifespan. Although generally rising with age, we were unable to establish temporally when during these women's lives they experienced increases and/or stabilization in allostatic load levels.

Furthermore, the method used to calculate allostatic load may not have been adequate for assessing within-group differences, particularly among Black women. Of the 21 different studies from 1988 to 2010 that have used NHANES data to calculate allostatic load, 26 biomarkers and 18 different methods have been used, none of which have included population-specific calculations.³⁶ Although biomarkers are sufficient summary measures of allostatic load in population samples, the biological pathways that lead to allostatic load have shown some variation by race/ethnicity.³⁷ Moreover, calculations of allostatic load using NHANES data are limited by the data NHANES collects and makes available, so calculations using endocrine markers such as cortisol are currently not possible.

Conclusions

Black sexual minority women had similar mean allostatic load as Black heterosexual women in this study, after adjustment for age differences. Although the number of Black bisexual and lesbian women was small, even after combining six waves of NHANES data, this suggests that sexual orientation plays a more nuanced role than we were able to detect in adding to the burden of lifetime stress among this population of Black women, many of whom have been subjected to a lifetime of microaggressions, prejudice, and discrimination. Although we did not find dramatic differences in this cross-sectional study, we hope that our focus on the intersectional identities of this highly informative group of Black women will spur greater interest in collecting and analyzing relevant data from a life course perspective.

Authors' Contributions

T.W. conceived of the presented study and performed the methods. B.M.J. and S.N.F. aided in developing the theory associated with the study. S.N.F. aided in performing analyses. All authors interpreted the data. T.W. led the writing of the article with input from all authors. All authors have approved the article and take accountability for the accuracy and integrity of the study.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

This research was funded by the National Institute of Health (NIH) Postbaccalaureate Research Education Program (PREP) Grant (R25GM121220).

Supplementary Material

Supplementary Table S1

Supplementary Table S2

Supplementary Table S3

Supplementary Table S4

Supplementary Figure S1