Abstract
Background
Associations between multiple forms of discrimination and blood pressure control in older populations remain unestablished.
Methods
Participants were 14 582 noninstitutionalized individuals (59% women) in the Health and Retirement Study aged at least 51 years (76% non-Hispanic White, 15% non-Hispanic Black, 9% Hispanic/Latino). Primary exposures included the mean frequency of discrimination in everyday life, intersectional discrimination (defined as marginalization ascribed to more than one reason), and the sum of discrimination over the life span. We assessed whether discrimination was associated with a change in measured hypertension status (N = 14 582) and concurrent medication use among reported hypertensives (N = 9 086) over 4 years (2008–2014).
Results
There was no association between the frequency of everyday discrimination and change in measured hypertension. Lifetime discrimination was associated with higher odds of hypertension 4 years later among men (odds ratio [OR]: 1.21, 95% confidence interval [CI]: 1.08–1.36) but not women (OR: 0.98, 95% CI: 0.86–1.13). Only among men, everyday discrimination due to at least 2 reasons was associated with a 1.44-fold (95% CI: 1.03–2.01) odds of hypertension than reporting no everyday discrimination, reporting intersectional discrimination was not associated with developing hypertension among women (OR: 0.91, 95% CI: 0.70–1.20). All 3 discriminatory measures were inversely related to time-averaged antihypertensive medication use, without apparent gender differences (eg, OR for everyday discrimination–antihypertensive use associations: 0.85, 95% CI: 0.77–0.94).
Conclusions
Gender differences in marginalization may more acutely elevate hypertensive risk among older men than similarly aged women. Experiences of discrimination appear to decrease the likelihood of antihypertensive medication use among older adults overall.
Keywords: Cardiovascular, Health disparities, Risk factors, Socioeconomic issues
Background
Non-Hispanic Black and certain Hispanic subgroups receive hypertension diagnoses at younger ages (1,2) and go on to experience disproportionately more adverse events including cardiovascular events (3), disability (4), as well as death (5) than non-Hispanic Whites. The risk of hypertension also appears to be higher among people of lower socioeconomic status (SES) (6). However, factors such as race and SES are insufficient proxies to evaluate the impact of corresponding forms of unfair treatment (eg, racism and classism, respectively) on blood pressure management among older adults.
Experiences of discrimination are theorized to simultaneously produce stress that dysregulates the cardiovascular system (7) and inhibit the ability to control blood pressure (8,9). Cross-sectional evidence of the discrimination–hypertension association has been equivocal (10–12). By focusing on racism among African Americans, the predominance of studies may have insufficiently captured intra- as well as intergroup variability across sociodemographic categories. Epidemiological research should take into account how intersectional discrimination, such as combined classism, ageism, and racism, impairs blood pressure control (13,14). Epidemiologic applications of Crenshaw’s intersectionality theory involve moving beyond evaluations of sociodemographic factors such as race, gender, or class as independent and immutable primary exposures, or assuming consistent societal treatment across such identifiers (15,16). Though initially addressing the distinct inequity experienced by African American women, the intersectionality framework should be extended to diverse populations experiencing discrimination due to reasons experienced potentially since birth (eg, sexism) as well as acquired (eg, ageism). Overlapping explanations as to why intersectional discrimination may acutely induce stress later in the life course include losing the ability to generate income or exert power, in conjunction with an increased comorbidity burden and being classified as frail (17). Compared to discrimination ascribed to a single attribute, discrimination due to multiple aspects of identity may deprive individuals of more socioeconomic and psychological resources to engage in hypertension-preventing behaviors (15). Moreover, the association between discrimination and blood pressure management may vary by gender due to varying exposure to discrimination, relative societal standing over the life course, as well as coping mechanisms (18–20).
This study extends the examination of the link between discrimination and hypertension risk by evaluating the association of intersectional discrimination and blood pressure control in a diverse population of older adults. We assessed experiences of discrimination using 3 measures: the mean frequency of discrimination occurring in one’s daily life, the number of reasons for everyday discrimination, and the amount of systemic discrimination that occurs over an entire life span. As discrimination may influence access to and use of medication, we also evaluated the association of discrimination and antihypertensive medication use among participants with reported hypertension diagnoses. We evaluated whether the magnitude of these associations differs by gender, hypothesizing that the difference in how men and women experience as well as respond to unequal treatment influences blood pressure control.
Method
Study Sample
The Health and Retirement Study (HRS) is an ongoing cohort of initially noninstitutionalized adults, as well as their spouses of any age. Investigators on the HRS collect comprehensive information on the transitions from middle age to older adulthood (21). The 4-stage sampling of the contiguous 48 US states is nationally representative, with oversamples of Black and Hispanic populations as well as Floridians. The University of Michigan Institutional Review Board approved the study protocol. This article, based on analyses of deidentified data, required no further ethical approval. Figure 1 details criteria for inclusion in the current analytic sample.
Figure 1.
Study inclusion criteria.
In 2008, approximately half of the entire HRS sample participated in a face-to-face interview conducted in their home and provided information about their SES (eg, income) and health conditions (eg, hypertension), along with biomarker and physical functioning (eg, sphygmomanometer readings) measures. Meanwhile, the other half of the study sample completed telephone interviews and had their face-to-face interview in 2010. After the face-to-face interview, participants were given a “Left Behind Questionnaire” about psychosocial (eg, discrimination) and lifestyle (eg, physical activity) factors to self-complete and return by mail (22). Accordingly, in the current analyses, an indicator for belonging to half of the study sample completing the baseline wave for 2008 versus 2010 was included to account for potential period trends. As the discrimination measures were updated in 2008, all analyses were conducted on 2008, 2010, 2012, and 2014 waves of the HRS to ensure consistency of exposure assessment and provide 4 years of follow-up for each participant. The current analytic sample was restricted to the pooled 2008 and 2010 waves (Time 1) and pooled 2012 and 2014 waves (Time 2) of the HRS who were at least 51 years of age at study baseline, did not use a proxy during their baseline wave, and had completed at least one of the everyday and lifetime discrimination subscales on the Left Behind Questionnaire (N = 14 582, Figure 1).
Measures
Mean frequency of everyday discrimination
The 6 questions from the frequency of everyday discrimination scale concerned unfair treatment within the past 12 months, following the prompt: “In your day-to-day life, how often have any of the following things happened to you?” Items determined whether participants (a) were treated with less respect, (b) received poorer customer service, (c) were treated as unintelligent, (d) evoked others acting as if fearful of them, (e) threats or harassment, and (f) received disproportionately poor health care treatment. The possible answer choices, a Likert scale ranging from “never” to “almost every day,” were reverse-coded and averaged for the current analyses. The frequency of everyday discrimination has a Cronbach’s alpha of 0.80 for the 2008 HRS wave and 0.82 for the 2010 HRS wave (22), indicating high reliability of this measure among this study sample.
Intersectional everyday discrimination
In a subsequent question, participants could ascribe the experiences of everyday discrimination to any combination of the following reasons: ancestry or national origin, gender, race, age, religion, weight, physical disability, aspect of physical appearance, sexual orientation, financial status, or other. Intersectional everyday discrimination was operationalized into 3 categories: ascribing everyday discrimination to at least 2 reasons, one reason, or no everyday discrimination.
Sum of lifetime discrimination
The questionnaire also contained 7 dichotomous questions about lifetime discrimination, after specifying, “For each of the following events, please indicate whether the event occurred at any point in your life.” In the current analyses, lifetime discrimination was the summed affirmatives to questions about unfairly (a) being prevented from moving to a neighborhood, (b) not being hired, (c) not being promoted, (d) losing employment, (e) receiving adequate treatment in health care settings, (f) denied a bank loan, and (g) encountering harassment or abuse from the police. Possible scores on this scale ranged from “no lifetime discrimination” (0) to “experienced all 7 domains of lifetime discrimination” (7). The lifetime discrimination subscale has moderate reliability in this study sample (Cronbach’s alpha = 0.71) (23). Per HRS guidance (22), participants missing more than 3 items on either the everyday or lifetime discrimination scale were excluded (N = 563).
Measured hypertension
Trained interviewers used Omron HEM-780 Intellisense automated sphygmomanometers with ComFit cuffs to take systolic and diastolic readings from participants seated with both feet on the floor (24). Three sets of readings, taken between 45 and 60 seconds apart from a participant’s supported left arm with the palm facing upward, were averaged for the current analyses. Blood pressure readings below 40 mmHg (102 readings at Time 1 and 40 readings at Time 2) were recoded as missing due to probable measurement error. Measured hypertension was defined based on mean sphygmomanometer readings and classified according to The Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC 7) criteria (25), the guideline during the period of the study: at least 140 mmHg for systolic blood pressure and/or at least 90 mmHg for diastolic blood pressure. The use of JNC 7 as outcome accounts for the period trend of blood pressure control awareness and practices known to health care providers and study participants between 2008 and 2014.
Reported antihypertensive medication use
Only the participants who responded in the affirmative to the question, “Has a doctor ever told you that you have high blood pressure or hypertension?” (N = 9 086), were asked a subsequent question about taking antihypertensive medication: “In order to lower your blood pressure, are you now taking any medication?” An indicator variable of antihypertensive medication use among reported hypertensive HRS participants was derived from an affirmative to this question. People without reported hypertension were excluded from the medication analyses.
Confounders
Selection and categorization of baseline confounding variables were considered based on prior evidence. Age, sex, race/ethnicity, and US nativity were assessed by self-report. Racial and ethnic designation categories were non-Hispanic White, non-Hispanic Black, and Hispanic/Latino ethnicity inclusive; less than 2% of the sample, who did not belong to those identities were recoded as missing. Highest educational attainment for study participant, mother, and father was categorized as less than high school, high school degree/General Education Diploma (GED), 2-year college degree, at least a 4-year college degree (26); the systematic missingness in parental education was addressed with a distinct category. Marital status was classified as married/partnered, divorced/separated, widowed, never married/partnered, and unknown (27). Household net income was classified as none, unknown, equaling to twice above the 2000 Current Population Survey (CPS) poverty threshold, more than twice to 4 times above the CPS poverty threshold, and more than 4 times above the CPS poverty threshold (28). Housing status categories included home ownership, renting or living rent free with family, or unknown. Insurance type was categorized as private (ie, employer-based or individually purchased) nonprivate only (ie, Medicare, Medicaid, Civilian Health and Medical Program Uniformed Service, and Department of Veterans Affairs), or unknown (29). Self-rated health was determined by a 5-point Likert scale, with possible answers from “excellent” to “poor” (30).
Statistical Analyses
We conducted analyses in 2020 using Stata version 15.1 (StataCorp LP, College Station, TX). All estimates were weighted by the probability of being eligible for completing the Left Behind Questionnaire to account for the complex sampling methodology of the HRS. We first described baseline characteristics of the study population, across categories of a number of reasons for everyday discrimination (none vs 1 vs at least 2) and tested for differences using the Wilcoxon rank-sum test for continuous variables and Pearson chi-squared test for categorical variables. We reported the weighted medians and weighted interquartile ranges of continuous variables, along with the weighted percentages and unweighted counts of categorical variables. The 882 participants who did not ascribe their everyday discrimination to any reason were combined with the no everyday discrimination category. People reporting either never experiencing everyday discrimination or no reasons for everyday discrimination were compared to those reporting one as well as those reporting at least 2 of these reasons.
To minimize selection bias due to substantial incomplete covariate and outcome data, we implemented multiple imputation by chained equations (MICE) in sensitivity analyses (31). Complete data predictively provided values to variables with continuous, binary, and ordinal distributions, conditional on all other variables in the chained equation. Results across 35 imputed data sets were pooled, with standard errors adjusting for intra- and intervariability using Rubin’s rule (32). We made the following specifications to the imputation chains: (a) variables that had no missingness: age, baseline wave, highest educational attainment of father, as well as the discrimination measure (eg, mean everyday frequency, everyday attribution, and lifetime sum); (b) variables that had missingness (percentage noted parenthetically): racial/ethnic designation (3%), highest educational attainment of participant (<1%), US nativity (<1%), marital status (<1%), household net income relative to 2000 poverty line (1%), insurance type (<1%), and self-rated health (<1%), systolic blood pressure at Time 1 (15%), systolic blood pressure at Time 2 (35%), diastolic blood pressure at Time 1 (15%), and diastolic blood pressure at Time 2 (35%); and (c) interaction terms between the discrimination measures and sex. Hypertensive status was passively generated from the imputed blood pressure readings. Separate models with the same specifications were used for medication use among reported hypertensives at Time 1 (<1%) and Time 2 (14%). We compared the imputed and unimputed estimates as qualitative evaluations of accurate MICE specification as well as nonsystematic data missingness (33).
The everyday, everyday attribution, and lifetime measures of perceived discrimination were evaluated individually as predictors in 3 sets of models. The associations of perceived discrimination and hypertensive outcomes were first estimated without adjustment for confounding variables. To obtain adjusted estimates in a research area without conclusive guidelines for confounder selection, the models of each hypertensive outcome were then fitted using a backward stepwise process, based on a p value of .20 for retention. The following variables were commonly retained across hypertensive outcomes: 2008 baseline wave, age in years, sex (used here as the only available proxy for gender), racial/ethnic designation (non-Hispanic White, non-Hispanic Black, Hispanic/Latino), study participant highest educational attainment (less than high school, high school degree/GED, 2-year college degree, 4-year college degree), US nativity, marital status (married/partnered, divorced/separated, widowed, never married/partnered, unknown), highest educational attainment of the father (less than high school, high school degree/GED, 2-year college degree, 4-year college degree, unknown), household net income relative to 2000 poverty line (no income, unknown, at to twice above, more than twice to 4 times above, more than 4 times above), insurance type (private, only federal/public, neither), and self-rated health (excellent, very good, good, fair, poor).
We used generalized estimating equations to evaluate the time-averaged associations between the 3 discrimination measures and antihypertensive use among reported hypertensives (N = 9 086), specifying a binomial distribution, a logistic link function, and an independent correlation structure. We evaluated the longitudinal changes in measured hypertension, specifying an independent covariance structure and robust standard errors in the logistic mixed models. Depending on the participant baseline wave, the 4-year changes in hypertensive outcomes were either between 2008 and 2012 or 2010 and 2014. We added an interaction term for sex to all models to evaluate effect modification. When the sex differences were significant, we presented stratified estimates with corresponding 95% confidence intervals (CIs), as well as the p value for the 3-way interaction term (Discrimination measure × Sex × Time).
Results
Sixty-one percent of the HRS sample reported experiencing everyday discrimination. Of those participants, 25% ascribed these experiences to 1 reason and another 25% to at least 2 reasons. Thirty-three percent of participants had experienced discrimination in any of the 7 domains specified in the lifetime discrimination scale. We presented the baseline characteristics of the combined 2008 and 2010 study waves in Table 1. Compared to those reporting no everyday discrimination, participants reporting one or more reasons for everyday discrimination within the past 12 months of assessment were more likely to be younger. Participants reporting one reason for everyday discrimination were more likely to be men, non-Hispanic White, born in the United States, or currently married than participants reporting no everyday discrimination as well as at least 2 reasons for everyday discrimination. Participants experiencing at least 2 reasons for everyday discrimination were less likely than participants experiencing no everyday discrimination to have measured or reported hypertension and, if hypertensive, were less likely to report antihypertensive use.
Table 1.
Baseline Sociodemographic Characteristics of Health and Retirement Study Participants, Aged 51–101 Years, by Reasons for Everyday Discrimination, 2008–2010 (N = 14 429)
| None (N = 7 179) | One (N = 3 576) | At Least 2 (N = 3 674) | p * | |
|---|---|---|---|---|
| Characteristic | ||||
| Age, years (IQR) | 66 (60, 76) | 62 (57, 71) | 61 (57, 69) | <.001 |
| Sex, male N (%) | 2 918 (44) | 1 601 (48) | 1 463 (45) | .002 |
| Racial/ethnic designation | ||||
| Non-Hispanic White, N (%) | 5 549 (84) | 2 814 (85) | 2 353 (74) | |
| Non-Hispanic Black, N (%) | 816 (8) | 411 (8) | 816 (16) | <.001 |
| Hispanic/Latino, N (%) | 662 (8) | 281 (7) | 371 (10) | |
| US nativity, N (%) | 6 397 (90) | 3 286 (93) | 3 340 (91) | <.001 |
| Marital/partnership status | ||||
| Married/partnered, N (%) | 4 486 (64) | 2 312 (65) | 2 009 (55) | |
| Divorced/separated, N (%) | 855 (13) | 512 (16) | 738 (22) | <.001 |
| Widowed, N (%) | 1 619 (19) | 619 (14) | 717 (16) | |
| Never married/partnered, N (%) | 213 (4) | 133 (5) | 207 (7) | |
| Housing status | ||||
| Owns home, N (%) | 4 142 (58) | 2 054 (57) | 1 910 (52) | |
| Rents or lives with family, N (%) | 1 049 (14) | 505 (14) | 793 (21) | <.001 |
| Other, N (%) | 1 988 (28) | 1 017 (29) | 971 (27) | |
| Highest educational attainment | ||||
| Less than high school, N (%) | 1 444 (17) | 520 (12) | 693 (16) | |
| High school degree/GED N (%) | 3 871 (53) | 1 950 (54) | 1 995 (54) | <.001 |
| Two-year college degree, N (%) | 313 (5) | 182 (6) | 208 (6) | |
| At least 4-year college degree, N (%) | 1 536 (25) | 910 (28) | 771 (24) | |
| Income: Poverty categorization | ||||
| No income, N (%) | 30 (<1) | 20 (<1) | 35(1) | |
| >0 to 2, N (%) | 1 754 (22) | 808 (21) | 1 115 (28) | <.001 |
| 2–4, N (%) | 2 335 (30) | 1 075 (28) | 1 141 (31) | |
| >4, N (%) | 2 998 (47) | 1 634 (50) | 1 353 (40) | |
| Insurance type | ||||
| Private, N (%) | 4 205 (63) | 2 238 (68) | 2 076 (61) | |
| Only federal/public, N (%) | 2 612 (31) | 1 117 (25) | 1 280 (30) | <.001 |
| Neither public or private, N (%) | 327 (6) | 207 (7) | 301 (9) | |
| Paternal educational attainment | ||||
| Less than high school, N (%) | 3 848 (49) | 1 709 (43) | 1 822 (45) | |
| High school degree/GED, N (%) | 1 463 (22) | 841 (27) | 779 (24) | <.001 |
| Two-year college degree, N (%) | 370 (6) | 210 (6) | 173 (5) | |
| At least 4-year college degree, N (%) | 502 (9) | 329 (11) | 260 (9) | |
| Unknown N (%) | 996 (14) | 487 (13) | 640 (17) | |
| Maternal educational attainment | ||||
| Less than high school, N (%) | 3 731 (46) | 1 640 (40) | 1 815 (44) | |
| High school degree/GED, N (%) | 1 979 (32) | 1 157 (36) | 1 030 (32) | <.001 |
| Two-year college degree, N (%) | 417 (6) | 282 (8) | 256 (8) | |
| At least 4-year college degree, N (%) | 375 (7) | 226 (8) | 218 (7) | |
| Unknown, N (%) | 677 (9) | 271 (8) | 355 (9) | |
| Self-rated health | ||||
| Excellent, N (%) | 780 (12) | 339 (11) | 213 (7) | |
| Very good, N (%) | 2 386 (35) | 1 176 (34) | 911 (25) | |
| Good, N (%) | 2 224 (29) | 1 195 (32) | 1 209 (32) | <.001 |
| Fair, N (%) | 1 293 (17) | 649 (16) | 921 (25) | |
| Poor, N (%) | 495 (7) | 217 (7) | 417 (11) | |
| Antihypertensive medication, N (%) | 4 230 (90) | 1 914 (87) | 2 131 (87) | <.001 |
| SBP, mg/L (IQR) | 130 (118, 144) | 127 (116, 141) | 128 (116, 141) | <.001 |
| DBP, mg/L (IQR) | 79 (72, 87) | 80 (72, 87) | 80 (73, 87) | .60 |
| Baseline measured hypertension, N (%) | 2 218 (35) | 973 (31) | 1 033 (32) | .002 |
| Baseline reported hypertension, N (%) | 4 540 (59) | 2 097 (54) | 2 372 (62) | <.001 |
Notes: IQR = interquartile range; GED = General Education Diploma; SBP = systolic blood pressure; DBP = diastolic blood pressure. Household net income classifications are relative to the poverty line as determined by the 2000 Current Population Survey. Reported private insurance indicates employer-based or individually purchased, while Federal/Public indicates Medicare, Medicaid, Civilian Health and Medical Program Uniformed Service, and Department of Veterans Affairs. Antihypertensive use is only among participants self-reporting having been diagnosed with hypertension by a physician.
*p value (Pearson chi-squared test for categorical variables, Wilcoxon rank-sum test for continuous variables) for the difference between reasons for everyday discrimination categories.
There was no association between the frequency of everyday discrimination and change in measured hypertension over 4 years among men or women (Table 2). However, men whose everyday discrimination was ascribed to at least 2 reasons had 1.44-fold odds of measured hypertension over 4 years (95% CI: 1.03–2.01), compared with men reporting no everyday discrimination. The results were significantly different in women, among whom there was no association between the number of reasons for everyday discrimination and hypertension (odds ratio [OR]: 0.91, 95% CI: 0.70–1.20), p value for attribution × Time × Sex interaction = 0.04. We observed a significantly larger association between lifetime discrimination and measured hypertension over 4 years of follow-up in men (OR: 1.21 [95% CI: 1.08–1.36]). This association was not present among women ([OR: 0.98, 95% CI: 0.86–1.13], p value for lifetime discrimination × Time × Sex interaction = 0.03).
Table 2.
Longitudinal Associations of Perceived Discrimination and 4-Year Changes in Measured Hypertension, Health and Retirement Study, 2008–2014
| Discrimination Subscale | Unadjusted | Adjusted* | |||||
|---|---|---|---|---|---|---|---|
| Men | Women | Men | Women | ||||
| OR (95% CI) | OR (95% CI) | p † | OR (95% CI) | OR (95% CI) | p † | ||
| Increased frequency of everyday discrimination (1–6) (N = 12 050) | 1.06 (0.88, 1.28) | 0.93 (0.78, 1.11) | .29 | 1.06 (0.88, 1.28) | 0.93 (0.78, 1.11) | .32 | |
| Reasons for everyday discrimination (N = 11 924) | 0 | Ref | |||||
| 1 | 0.89 (0.65, 1.22) | 0.88 (0.66, 1.17) | .97 | 0.88 (0.64, 1.21) | 0.87 (0.66, 1.17) | .99 | |
| >2 | 1.42 (1.02, 1.97) | 0.92 (0.71, 1.21) | .05 | 1.44 (1.03, 2.01) | 0.91 (0.70, 1.20) | .04 | |
| Higher sum of lifetime discrimination (0–7) (N = 11 962) | 1.21 (1.08, 1.36) | 1.00 (0.87, 1.15) | .04 | 1.21 (1.08, 1.36) | 0.98 (0.86, 1.13) | .03 |
*Adjusted for 2008 baseline wave, age in years, sex, racial/ethnic designation (non-Hispanic White, non-Hispanic Black, Hispanic/Latino), and highest educational attainment (less than high school, high school degree/GED, 2-year college degree, 4-year college degree), US nativity, marital status (married/partnered, divorced/separated, widowed, never married/partnered, unknown), highest educational attainment of the father (less than high school, high school degree/GED, 2-year college degree, 4-year college degree, unknown), household net income relative to 2000 poverty line (no income, unknown, at to twice above, more than twice to 4 times above, more than 4 times above), insurance type (private, only federal/public, neither), self-rated health (excellent, very good, good, fair, poor); weighted by the probability of being eligible for Left Behind Questionnaire.
† p Value for discrimination scale × Sex × Time interaction term.
There were consistent inverse associations between discrimination measures and time-averaged antihypertensive use among those participants diagnosed with hypertension (Table 3). For example, among persons with hypertension, each point of increased frequency of everyday discrimination was associated with a 0.85 (95% CI: 0.77–0.94) OR of antihypertensive use. Participants reporting at least 2 reasons for everyday discrimination had a lower likelihood of antihypertensive use than those reporting never experiencing everyday discrimination (OR: 0.82, 95% CI: 0.69–0.98). There were no differences in the association of discrimination and antihypertensive use by sex.
Table 3.
Associations of Perceived Discrimination and Time-Averaged Odds of Antihypertensive Use Among Hypertensives, Health and Retirement Study, 2008–2014
| Discrimination Subscale | Unadjusted | Adjusted* | |||
|---|---|---|---|---|---|
| OR (95% CI) | p | OR (95% CI) | p | ||
| Increased frequency of everyday discrimination (1–6) (N = 8 587†) | 0.79 (0.72, 0.86) | <.001 | 0.85 (0.77, 0.94) | .002 | |
| Reasons for everyday discrimination (N = 8 510†) | 0 | Ref | |||
| 1 | 0.79 (0.65, 0.94) | .14 | 0.87 (0.73, 1.05) | .16 | |
| >2 | 0.72 (0.61, 0.86) | <.001 | 0.82 (0.69, 0.98) | .03 | |
| Higher sum of lifetime discrimination (0–7) (N = 8 499†) | 0.86 (0.80, 0.91) | <.001 | 0.91 (0.84, 0.98) | .01 |
*Adjusted for 2008 baseline wave, age in years, sex, racial/ethnic designation (non-Hispanic White, non-Hispanic Black, Hispanic/Latino), and highest educational attainment (less than high school, high school degree/GED, 2-year college degree, 4-year college degree), US nativity, marital status (married/partnered, divorced/separated, widowed, never married/partnered, unknown), highest educational attainment of the father (less than high school, high school degree/GED, 2-year college degree, 4-year college degree, unknown), household net income relative to 2000 poverty line (no income, unknown, at to twice above, more than twice to 4 times above, more than 4 times above), insurance type (private, nonprivate, unknown), self-rated health (excellent, very good, good, fair, poor); weighted by the probability of being eligible for Left Behind Questionnaire.
†Analyses restricted to those with reported hypertension.
The imputed estimates in Supplementary Tables 1 and 2 generally followed the same trends as the unimputed estimates, though the imputed effect sizes were modestly attenuated compared to the original results (eg, OR: 1.35, 95% CI: 0.97–1.86, for measured hypertension over 4 years among men ascribing everyday discrimination to at least 2 reasons vs no everyday discrimination).
Discussion
Main Results
Among men in our study, everyday discrimination due to at least 2 reasons and a higher amount of discrimination experienced over the lifetime were associated with more detrimental hypertensive trajectories over 4 years. We did not, however, observe similar associations among women. Additionally, hypertensive participants of both sexes who experienced frequent everyday discrimination, everyday discrimination ascribed to at least 2 reasons, and more experiences of discrimination over the lifetime were less likely to report taking medication to manage their condition. Our study is among the first to report a detrimental association between discrimination ascribed to multiple reasons and hypertension outcomes. Moreover, we included longitudinal data and mixed models that allow us to evaluate both the development and resolution of hypertension associated with perceived discrimination.
Comparison With Other Studies
Our study included a diverse sample of older adults; variance in intersectional discriminatory experience may be limited in work that includes all-minority study samples. For example, a recent analysis among the African American Jackson Heart Study cohort found no significant interactions between everyday or lifetime discrimination and specific reasons (eg, race, age) on the 13-year incidence of hypertension (11). Other studies have used 2-way and 3-way interactions between specific sociodemographic variables and discrimination measures to assess intersectionality among multiethnic cohorts. In an analysis conducted with CARDIA data, though a higher proportion of Black women and Black men reported more discrimination than their White counterparts, the magnitude of discriminatory experiences on worse cardiovascular health was higher among White than Black participants (19).
American male workers at least 50 years in age have reported a decline in age discrimination between 1990 and 2017 (17). However, after spending more of their working years in dominant professional or social positions, men may experience a relatively sharper decline in societal positions. While experiences of discrimination may act as an acute stressor, women may on average spend more time pursuing meaningful coping strategies such as developing interpersonal relationships (34). Compared to women, men may perceive forms of unequal treatment such as classism, ageism, and ableism both later in life as well and more acutely due to declining health as well as a loss of financial resources after retirement from positions at the top of the social hierarchy. Marginalized groups with less legal protection from bias such as low-SES older adult men may be more exposed to and less protected from discriminatory treatment. This social disadvantage may increase negative health behaviors to adhere to traditional gender roles or high effort striving to overcome socioeconomic marginalization (35).
Our findings demonstrate that experiences of discrimination are inversely associated with antihypertensive medication adherence. The authors of a 2017 review identified lower SES and provider–patient miscommunications as reasons that Black patients follow their antihypertensive medication regimens less consistently than Whites (36). Among 806 patients of a northeastern health care system, Black race was no longer statistically significant with the odds of uncontrolled blood pressure in models adjusted for health care discrimination, medication adherence, and SES (37). However, other studies have demonstrated that blood pressure disparities persist when Black and White participants report equal awareness, treatment initiation, and antihypertensive adherence (38,39). As experiences of discrimination include inadequate care from health providers, our work refines the work on disparities in antihypertensive adherence beyond racial differences. Mistrust in the medical community may inhibit blood pressure control among marginalized people.
Strengths and Limitations
By conducting a longitudinal analysis, we were able to establish temporality between discriminatory experiences and hypertensive trajectories. This research adds to the literature on discrimination and hypertension by examining the burden of discrimination as well as the perceived attribution. Additionally, our sensitivity analyses illustrate the application of multiple imputation by chained equations, in order to address the selection bias inherent to limiting analyses to study participants with complete data.
However, the limitations in our study should be considered when interpreting the results. The RAND Corporation contributed model-based imputations for missing income and asset data. Our supplemental MICE estimates should be considered in light of inadequate existing guidance regarding imputing upon imputed data. The HRS discrimination measures used for the exposure were updated in 2008, motivating our use of the combined 2008 and 2010 halves of the sample as our baseline wave. HRS integrated the biomarker and psychosocial measures beginning in 2006, randomly splitting the sample and completing the face-to-face interview for one half of the sample in 2006 and the other half in 2008; the 2006 and 2008 waves are usually pooled. We included the baseline wave in our model to account for potential differences between the 2 waves. Antihypertensive use came from participant self-report, which may systematically fail to capture participants with low health literacy. The wording of the reason for the everyday discrimination items, from which we derived the ordinal number of reasons for everyday discrimination measure, does not specify whether the specific forms of discrimination, such as racism or classism, have the same exposure time frame. This sample attributed their everyday discrimination to more than 300 combinations of reasons, meaning we lacked power in conducting stratified analyses that could assess the differential impact of individual or combined forms of unequal treatment that are experienced earlier (eg, sexism) or later (eg, ageism) in the life course. Our current analytic sample, though comparatively diverse, was more than three quarters non-Hispanic White. As minority participants, particularly men, are harder than non-Hispanic Whites to recruit and retain in population health research due to historical and contemporaneous mistreatment by the medical community (40,41), our results may not be generalizable to people experiencing the most discrimination.
Conclusions
Epidemiological studies have commonly used racial/ethnic designation to determine hypertensive risk. This study builds upon previous work on disparities in blood pressure control by evaluating the independent associations of discrimination and hypertensive outcomes among a multiethnic cohort, as well as assessing sex differences. Future research should factor in the co-occurring reasons for unfair treatment, such as age, national origin, and physical appearance, that compound adverse blood pressure outcomes among older adults. By parsing whether various forms of discrimination operate simultaneously and inseparably, as combinations or permutations of unequal treatment, researchers can more accurately assess the influence of intersectional discrimination on health outcomes. A better understanding of the role of discrimination on hypertension can inform interventions aimed at ameliorating hypertension disparities.
Supplementary Material
Funding
The Health and Retirement Study is conducted by the University of Michigan and sponsored by the National Institute on Aging (NIA U01AG009740).
Conflict of Interest
M.C.O. serves as a consultant for Cricket Health, Inc. K.D.S., E.S., G.D.B., and P.W.H. declare no conflict.
Author Contributions
K.D.S. developed the research question and ran all analyses. All authors contributed to data interpretation and manuscript revision. The final version of the manuscript was reviewed and approved by all authors.
References
- 1. Fryar CD, Ostchega Y, Hales CM, Zhang G, Kruszon-Moran D. Hypertension prevalence and control among adults: United States, 2015–2016. NCHS Data Brief. 2017;( 289):1–8. PMID: 29155682. [PubMed] [Google Scholar]
- 2. Selby K, Michel M, Gildengorin G, et al. Disparities in hypertension control across and within three health systems participating in a data-sharing collaborative. J Am Board Fam Med. 2018;31(6):897–904. doi: 10.3122/jabfm.2018.06.180166 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Cheng S, Claggett B, Correia AW, et al. Temporal trends in the population attributable risk for cardiovascular disease. Circulation. 2014;130(10):820–828. doi: 10.1161/CIRCULATIONAHA.113.008506 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Hajjar I, Wharton W, Mack WJ, Levey AI, Goldstein FC. Racial disparity in cognitive and functional disability in hypertension and all-cause mortality. Am J Hypertens. 2016;29(2):185–193. doi: 10.1093/ajh/hpv084 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Lackland DT. Racial differences in hypertension: implications for high blood pressure management. Am J Med Sci. 2014;348(2):135–138. doi: 10.1097/MAJ.0000000000000308 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Leng B, Jin Y, Li G, Chen L, Jin N. Socioeconomic status and hypertension: a meta-analysis. J Hypertens. 2015;33(2):221–229. doi: 10.1097/HJH.0000000000000428 [DOI] [PubMed] [Google Scholar]
- 7. Ginty AT, Kraynak TE, Fisher JP, Gianaros PJ. Cardiovascular and autonomic reactivity to psychological stress: neurophysiological substrates and links to cardiovascular disease. Auton Neurosci. 2017;207:2–9. doi: 10.1016/j.autneu.2017.03.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Berger M, Sarnyai Z. “More than skin deep”: stress neurobiology and mental health consequences of racial discrimination. Stress. 2015;18(1):1–10. doi: 10.3109/10253890.2014.989204 [DOI] [PubMed] [Google Scholar]
- 9. Berjot S, Gillet N. Stress and coping with discrimination and stigmatization. Front Psychol. 2011;2:33. doi: 10.3389/fpsyg.2011.00033 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Brondolo E, Love EE, Pencille M, Schoenthaler A, Ogedegbe G. Racism and hypertension: a review of the empirical evidence and implications for clinical practice. Am J Hypertens. 2011;24(5):518–529. doi: 10.1038/ajh.2011.9 [DOI] [PubMed] [Google Scholar]
- 11. Forde AT, Sims M, Muntner P, Lewis T, Onwuka A, Moore K, Diez Roux AV. Discrimination and hypertension risk among African Americans in the Jackson Heart Study. Hypertension. 2020;76(3):715–723. doi: 10.1161/HYPERTENSIONAHA.119.14492. PMID: 32605388; PMCID: PMC8359680. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Cozier Y, Palmer JR, Horton NJ, Fredman L, Wise LA, Rosenberg L. Racial discrimination and the incidence of hypertension in US black women. Ann Epidemiol. 2006;16(9):681–687. doi: 10.1016/j.annepidem.2005.11.008 [DOI] [PubMed] [Google Scholar]
- 13. Turan JM, Elafros MA, Logie CH, Banik S, Turan B, Crockett KB, Pescosolido B, Murray SM. Challenges and opportunities in examining and addressing intersectional stigma and health. BMC Med. 2019;17(1):7. doi: 10.1186/s12916-018-1246-9. PMID: 30764816; PMCID: PMC6376691. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Jackson JW, Williams DR, VanderWeele TJ. Disparities at the intersection of marginalized groups. Soc Psychiatry Psychiatr Epidemiol. 2016;51(10):1349–1359. doi: 10.1007/s00127-016-1276-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Bauer GR, Scheim AI. Methods for analytic intercategorical intersectionality in quantitative research: discrimination as a mediator of health inequalities. Soc Sci Med. 2019;226:236–245. doi: 10.1016/j.socscimed.2018.12.015 [DOI] [PubMed] [Google Scholar]
- 16. Crenshaw, Kimberle. Demarginalizing the Intersection of Race and Sex: A Black Feminist Critique of Antidiscrimination Doctrine, Feminist Theory and Antiracist Policies. University of Chicago Legal Forum; 1989, no. 1(1989):139–167. [Google Scholar]
- 17. Chart data for the state of age discrimination and older workers in the U.S. 50 years after the Age Discrimination in Employment Act (ADEA). U.S. Equal Employment Opportunity Commission. https://www.eeoc.gov/chart-data-state-age-discrimination-and-older-workers-us-50-years-after-age-discrimination. Accessed April 20, 2021. [Google Scholar]
- 18. Richardson LJ, Brown TH. (En)gendering racial disparities in health trajectories: a life course and intersectional analysis. SSM Popul Health. 2016;2:425–435. doi: 10.1016/j.ssmph.2016.04.011 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Bey GS, Jesdale B, Forrester S, Person SD, Kiefe C. Intersectional effects of racial and gender discrimination on cardiovascular health vary among black and white women and men in the CARDIA study. SSM Popul Health. 2019;8:100446. doi: 10.1016/j.ssmph.2019.100446 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Martin AE, North MS, Phillips KW. Intersectional escape: older women elude agentic prescriptions more than older men. Pers Soc Psychol Bull. 2019;45(3):342–359. doi: 10.1177/0146167218784895 [DOI] [PubMed] [Google Scholar]
- 21. Juster FT, Suzman R. An overview of the health and retirement study. J Hum Resour. 1995;30:S7–S56. doi: 10.2307/146277. [DOI] [Google Scholar]
- 22. Smith J, Ryan L, Fisher G, Sonnega A, Weir D.. Psychosocial and Lifestyle Questionnaire 2006–2016: documentation report. Health and Retirement Study, institution: Survey Research Center, Institute for Social Research, University of Michigan. Ann Arbor, MI: HRS Health and Retirement Study; 2017. https://hrs.isr.umich.edu/sites/default/files/biblio/HRS%202006–2016%20SAQ%20Documentation_07.06.17_0.pdf. Accessed November 2020. [Google Scholar]
- 23. Vaghela P, Sutin AR. Discrimination and sleep quality among older US adults: the mediating role of psychological distress. Sleep Health. 2016;2(2):100–108. doi: 10.1016/j.sleh.2016.02.003 [DOI] [PubMed] [Google Scholar]
- 24. National Institute on Aging. Growing Older in America: The Health and Retirement Study. Income and wealth. 2007;56–72. [Google Scholar]
- 25. Chobanian AV, Bakris GL, Black HR, et al. The Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure: the JNC 7 report. JAMA. 2003;289(19):2560–2571. doi: 10.1001/jama.289.19.2560 [DOI] [PubMed] [Google Scholar]
- 26. Alves RF, Faerstein E. Educational inequalities in hypertension: complex patterns in intersections with gender and race in Brazil. Int J Equity Health. 2016;15(1):146. doi: 10.1186/s12939-016-0441-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Iwashyna TJ, Christakis NA. Marriage, widowhood, and health-care use. Soc Sci Med. 2003;57(11):2137–2147. doi: 10.1016/s0277-9536(02)00546-4 [DOI] [PubMed] [Google Scholar]
- 28. Hurd MD, Meijer E, Moldoff M, Rohwedder S. Improved wealth measures in the Health and Retirement Study, 2016.https://www.rand.org/pubs/working_papers/WR1150.html. Accessed April 1, 2019.
- 29. Duru OK, Vargas RB, Kermah D, Pan D, Norris KC. Health insurance status and hypertension monitoring and control in the United States. Am J Hypertens. 2007;20(4):348–353. doi: 10.1016/j.amjhyper.2006.11.007 [DOI] [PubMed] [Google Scholar]
- 30. Barger SD, Muldoon MF. Hypertension labelling was associated with poorer self-rated health in the third US National Health and Nutrition Examination Survey. J Hum Hypertens. 2006;20(2):117–123. doi: 10.1038/sj.jhh.1001950 [DOI] [PubMed] [Google Scholar]
- 31. Harel O, Mitchell EM, Perkins NJ, et al. Multiple imputation for incomplete data in epidemiologic studies. Am J Epidemiol. 2018;187(3):576–584. doi: 10.1093/aje/kwx349 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Mitani AA, Kurian AW, Das AK, Desai M. Navigating choices when applying multiple imputation in the presence of multi-level categorical interaction effects. Stat Methodol. 2015;27:82–99. doi: 10.1016/j.stamet.2015.06.001 [DOI] [Google Scholar]
- 33. White IR, Royston P, Wood AM. Multiple imputation using chained equations: Issues and guidance for practice. Stat Med. 2011;30(4):377–399. doi: 10.1002/sim.4067 [DOI] [PubMed] [Google Scholar]
- 34. Ayalon L, Tesch-Römer C. Taking a closer look at ageism: self- and other-directed ageist attitudes and discrimination. Eur J Ageing. 2017;14(1):1–4. doi: 10.1007/s10433-016-0409-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Gilbert KL, Ray R, Siddiqi A, et al. Visible and invisible trends in black men’s health: pitfalls and promises for addressing racial, ethnic, and gender inequities in health. Annu Rev Public Health. 2016;37:295–311. doi: 10.1146/annurev-publhealth-032315-021556 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Ferdinand KC, Yadav K, Nasser SA, Clayton-Jeter HD, Lewin J, Cryer DR, Senatore FF. Disparities in hypertension and cardiovascular disease in blacks: the critical role of medication adherence. J Clin Hypertens (Greenwich). 2017;19(10):1015–1024. doi: 10.1111/jch.13089. PMID: 28856834; PMCID: PMC5638710. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Kressin NR, Orner MB, Manze M, Glickman ME, Berlowitz D. Understanding contributors to racial disparities in blood pressure control. Circ Cardiovasc Qual Outcomes. 2010;3(2):173–180. doi: 10.1161/CIRCOUTCOMES.109.860841 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Lafata JE, Karter AJ, O’Connor PJ, et al. Medication adherence does not explain black–white differences in cardiometabolic risk factor control among insured patients with diabetes. J Gen Intern Med. 2016;31(2):188–195. doi: 10.1007/s11606-015-3486-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Thorpe RJ, Bowie JV, Smolen JR, et al. Racial disparities in hypertension awareness and management: are there differences among African Americans and Whites living in similar social and healthcare resource environments? Ethn Dis. 2014;24(3):269–275. [PMC free article] [PubMed] [Google Scholar]
- 40. Shedlin MG, Decena CU, Mangadu T, Martinez A. Research participant recruitment in Hispanic communities: lessons learned. J Immigr Minor Health. 2011;13(2):352–360. doi: 10.1007/s10903-009-9292-1 [DOI] [PubMed] [Google Scholar]
- 41. Byrd GS, Edwards CL, Kelkar VA, et al. Recruiting intergenerational African American males for biomedical research studies: a major research challenge. J Natl Med Assoc. 2011;103(6):480–487. doi: 10.1016/s0027-9684(15)30361-8 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.