Abstract
Untested psychosocial or economic factors mediate associations between perceived discrimination and suboptimal antihypertensive therapy. This study included 2 waves of data from Health and Retirement Study participants with self-reported hypertension (n = 8,557, 75% non-Hispanic White, 15% non-Hispanic Black, and 10% Hispanic/Latino) over 4 years (baselines of 2008 and 2010, United States). Our primary exposures were frequency of experiencing discrimination, in everyday life or across 7 lifetime circumstances. Candidate mediators were self-reported depressive symptoms, subjective social standing, and household wealth. We evaluated with causal mediation methods the interactive and mediating associations between each discrimination measure and reported antihypertensive use at the subsequent wave. In unmediated analyses, everyday (odds ratio (OR) = 0.86, 95% confidence interval (CI): 0.78, 0.95) and lifetime (OR = 0.91, 95% CI: 0.85, 0.98) discrimination were associated with a lower likelihood of antihypertensive use. Discrimination was associated with lower wealth, greater depressive symptoms, and decreased subjective social standing. Estimates for associations due to neither interaction nor mediation resembled unmediated associations for most discrimination-mediator combinations. Lifetime discrimination was indirectly associated with reduced antihypertensive use via depressive symptomatology (OR = 0.99, 95% CI: 0.98, 1.00). In conclusion, the impact of lifetime discrimination on the underuse of antihypertensive therapy appears partially mediated by depressive symptoms.
Abbreviations
- CES-D
Center for Epidemiological Studies Depression Scale
- CI
confidence interval
- HRS
Health and Retirement Study
- OR
odds ratio
- SES
socioeconomic status
Antihypertensive therapy is a cornerstone of hypertension management, reducing the risk of cardiovascular events (1), kidney disease (2), and death (3). Recent national estimates demonstrate that hypertensive adults who use antihypertensive treatment are 50% more likely to experience adequate blood pressure control relative to those not using such therapy (4). Lack of antihypertensive treatment among people with high blood pressure can contribute to disparities, especially when due to sociodemographic (e.g., income (5)), psychological (e.g., depression (6)), and patient (e.g., multimorbidity, health-care access) factors. Compared with their non-Hispanic White counterparts in a large insurance database, Black and Hispanic adults were, respectively, 47% and 51% more likely to discontinue antihypertensive therapy over 2.5 years (7). Inadequate blood pressure control among marginalized older adults is associated with higher cardiovascular mortality as well as disproportionately more years of potential life lost (8, 9).
These stark disparities should be reframed around modifiable factors downstream from inequities in psychosocial or economic factors to inform targets for intervention. The psychometric instrument developed by Williams et al. (10) to initially assess discrimination experienced by African Americans has great relevance to capturing how older adults belonging to diverse backgrounds face economic, political, and cultural disadvantages in societies framed around the young and able-bodied (11–13). Beyond acting as a stressor, experiences of discrimination may also contribute to suboptimal antihypertensive therapy via established disease risk factors such as adverse mental health or insufficient interpersonal resources. A recent meta-analysis on medication adherence risk identified insufficient understanding of the psychosocial and economic factors on the causal pathway between sociodemographic factors and the likelihood of pharmacological management of chronic conditions (5). As an example, higher versus lower socioeconomic status (SES) was associated with a lower risk (relative risk = 0.89, 95% confidence interval (CI): 0.87, 0.92) of nonadherence to antihypertensive medication among 40 cohorts utilized in 30 studies (5). However, the heterogeneity between included studies (τ = 0.01; I2 = 95%) signifies that income does not adequately capture the complex impact of SES on antihypertensive use disparities among diverse groups (14–16). Compared with traditional metrics of socioeconomic status such as income, perceived position on the cultural hierarchy may more accurately operationalize social stratification experienced by the retired, the widowed, and the marginalized (17–20). Additionally, symptoms of depression due to discriminatory experiences may reduce motivation to pharmacologically control blood pressure (21–28).
A corresponding methodological consideration is that previous studies evaluating pathways between discrimination and health behaviors incorporated mediating psychosocial or economic variables with the same modeling strategies used for confounders; fixing the value of the mediator cannot account for plausible exposure-mediator interactions (29, 30). Furthermore, the mediation method most used in epidemiology generates noninterpretable estimates when the outcome is binary, because the components of an odds ratio do not equal the overall associations even in the absence of confounding (31). Clearer understanding of direct, indirect, and interactive effects of discrimination and psychosocial or economic resources may inform interventions to attenuate racial and ethnic differences in antihypertensive therapy. We have previously identified discrimination as an important risk factor for lack of antihypertensive therapy in a diverse older adult sample (32), building upon similar findings in African-American populations (22, 32). Evaluation of the synergistic and sequential impact of perceived discrimination and psychosocial or economic factors on suboptimal antihypertensive therapy motivates our current analysis.
Accordingly, we decomposed the casual pathway from experiences of discrimination, as it interacts with as well as operates via depressive symptomatology, subjective social standing, or household wealth, to subsequent antihypertensive treatment among a nationally representative sample of older adults with hypertension. We were able to evaluate the components of overall associations between each discrimination measure, each mediating factor, and antihypertensive treatment: 1) the discrimination–antihypertensive treatment associations due to neither interaction with or mediation by the psychosocial or economic factor; 2) the combined effect of discrimination and the psychosocial or economic factor associated with antihypertensive treatment, where discrimination is not a determinant of the psychosocial or economic factor; 3) the discrimination–antihypertensive treatment associations where discrimination is necessary for the psychosocial or economic factor to have a mediating effect; and 4) the discrimination–antihypertensive treatment associations via the mediating psychosocial or economic factor alone. We hypothesized that discrimination–antihypertensive treatment associations were partially mediated by higher depressive symptomatology, lower subjective social standing, or less household wealth; that depressive symptoms and subjective social standing were more informative mediators than household wealth among a diverse group of aging adults; and that discrimination was necessary for these mediated associations.
METHODS
Study sample
The Health and Retirement Study (HRS) obtains health, economic, and social data from a cohort of initially noninstitutionalized US adults at least 51 years of age (33). The baseline study protocol was approved by the University of Michigan Institutional Review Board. The current analyses, conducted using anonymized data, required no additional ethical approval. Between 2008 and 2014, halves of the sample alternated between assignment to telephone interviews and in-person interviews conducted in their homes; during the latter, participants completed physical assessments (e.g., sphygmomanometer readings) and received questions about their psychosocial characteristics to be returned by mail (34). The waves interviewed in 2008 and 2010 (time 1) and in 2012 and 2014 (time 2) were concatenated, with a baseline wave of assessment variable (2008 or 2010) included in adjusted effect estimates to account for period trends. Each participant has therefore provided 4 years of follow-up data, using the updated discrimination measures. As outlined in Figure 1, we used the entire cohort sample with complete discrimination and psychosocial or economic measures during 2008 or 2010, a reported hypertension diagnosis during 2012 or 2014, and had complete covariate data (n = 8,112 for depressive symptomatology, n = 7,470 for subjective social standing, and n = 8,292 for household wealth, Figure 1). We performed a complete-case analysis.
Inclusion criteria for a study of perceived discrimination and subsequent antihypertensive treatment, Health and Retirement Study, United States, 2008–2014. CES-D, Center of Epidemiological Studies Depression Scale.
Exposures
Our 2 main exposures were the frequency of everyday and sum of lifetime discrimination measures (10) assessed in 2008 or 2010, the first cohort waves that contained the updated measures. In accordance with recommendations (34), participants missing at least 4 items on either the everyday or lifetime discrimination scales (n = 563, 6%) were excluded from analyses. With Cronbach α values of 0.82 and 0.71 respectively, the frequency of everyday discrimination and sum of lifetime discrimination are reliable measures among the sample (34).
Frequency of everyday discrimination
Six questions followed the prompt, “In your day-to-day life, how often have any of the following things happened to you?” with possible responses ranging from “never” to “almost every day.” Participants confirmed the regularity of: 1) less respectful treatment, 2) poorer service during service industry interactions, 3) being deemed not intelligent, 4) being feared, 5) harassment or threatening encounters, and 6) poorer treatment in health-care situations. The everyday discrimination score was the averaged response to the items.
Sum of lifetime discrimination
Participants also completed 7 dichotomized questions regarding whether they had experienced lifetime discrimination in: 1) education, 2) health care, 3) law enforcement encounters, 4) bank loans, 5) housing, 6) employment hiring, and 7) occupational promotion. Participant scores could range from 0, meaning they experienced no lifetime discrimination, to 7, meaning they experienced all 7 listed experiences of lifetime discrimination.
Outcome
Our outcome in all analyses was antihypertensive medication use status, ascertained in 2012 or 2014 to ensure temporality between this variable and discrimination as well as psychosocial or economic measures. Participants who reported being diagnosed with hypertension by a doctor were asked, “In order to lower your blood pressure, are you now taking any medication?” (35). Participants who confirmed this in 2012 or 2014 were coded as being treated with antihypertensive medication. Comparison of sphygmomanometer readings and reported hypertension reveals high sensitivity and specificity across sociodemographic categories in the sample (36).
Mediating factors
Each candidate mediator was assessed at the participant baseline (either 2008 or 2010) wave to ensure that the intervening psychosocial or economic factor preceded antihypertensive use.
Depressive symptoms
Study staff administered an abbreviated version of the Center for Epidemiological Studies Depression Scale (CES-D), commonly used to assess symptoms of this mood disorder among older adults outside of clinical settings (37). This inventory, the CES-D 8, consists of 8 dichotomous questions and has high internal consistency (Cronbach α = 0.80) in the sample (38). Participants who affirmed the initial item (“feeling sad, blue, or depressed for two weeks or more in a row”) were then asked about 5 negative symptoms, such as restless sleep and perceiving everything they did to be an effort, as well as 2 reverse-coded positive components, such as feeling of happiness or having a lot of energy. The CES-D 8 score ranged from 0 to 8, with higher scores indicating more depression symptoms.
Subjective social standing
Participants completed the MacArthur Scale of subjective social status by indicating their perceived station on a ladder representative of society (39). Possible scores ranged from 1, having “the least money, least education, and the worst jobs or no jobs,” to 10, having “the most money, most education, and best jobs.” The overall modest correlations with educational attainment (cumulative effect size = 0.25), annual household income (cumulative effect size = 0.34), and occupation (cumulative effect size = 0.33) across 20 studies appear stronger among White than among Black adults surveyed (23). The subjective standing scale captures also sufficiently distinct constructs from other psychometric scales, including the CES-D 8, among adults of middle and older age (40).
Household wealth
The study uses sequential unfolding brackets and cross-wave asset verification of discrepancies to calculate assets in 20 domains, including residence values and retirement funds, minus debts (41). The RAND Corporation developed cross-wave imputation models of household wealth to further minimize systematic nonresponse and error, based on the following variables provided by each participating spouse: age, race, educational attainment, occupational status as well as class, marital status, expected bequests, cognitive function, and health status. Based upon earlier studies of disparities in wealth accumulation (42, 43), household wealth was categorized as quintiles with the following bounds: 1, minus $550,000 to $9,000; 2, $9,001 to $85,000; 3, $85,001 to $221,000; 4, $222,001 to $536,500: 5, above $536,501.
Confounding factors
As the conditional exchangeability assumption becomes stronger in the setting of causal mediation compared with other mediation methods (44), we included variables associated with discrimination (45), psychosocial or economic factors (22, 42), and medication adherence (5). Age, sex as the only available proxy for gender, race/ethnicity, and US nativity were self-reported. Racial and ethnic designation categories were non-Hispanic White, non-Hispanic Black, and Hispanic/Latino ethnicity inclusive; participants not belonging to those identities (n = 208, 2%) were included in the non-Hispanic White category. Highest educational attainment for study participant as well as for the father was categorized as less than high school, high-school diploma/General Educational Development certificate (GED), 2-year college degree, at least a 4-year college degree, or unknown. Marital status was classified as married/partnered, divorced/separated, widowed, never married/partnered, and unknown (46). Housing status categories were owning a home, either renting or living with family, and unknown. Insurance type categories included private (i.e., employer-based or individually purchased), nonprivate only (i.e., Medicare, Medicaid, Civilian Health and Medical Program Uniformed Service, and Department of Veterans Affairs), and unknown (47). Self-rated health was determined by a 5-point Likert scale, with possible answers from “excellent” to “poor.” To capture the burden of multimorbidity and polypharmacy while avoiding the collinearity inherent to adjustment for individual diseases, we derived 2 ordinal scales that corresponded with: 1) the number of self-reported chronic conditions, diagnosed by a doctor, and 2) the count of medications to treat these conditions. The reported comorbidities and polypharmacy in these scales included diabetes, cancer, lung disease, heart disease (i.e., myocardial infarction, coronary heart disease, angina, congestive heart failure, or other heart problem), stroke, psychiatric conditions, arthritis, and memory-related disease.
Trained interviewers took 3 blood pressure readings, 45 and 60 seconds apart, from a seated participant’s supported left arm with the palm facing upwards using Omron HEM-780 Intellisense automated sphygmomanometers with ComFit cuffs (Omron, Bannockburn, Illinois) (48). The averaged systolic and diastolic blood pressure readings were used in the current analyses, excluding readings below 40 mm Hg (n = 102 readings at time 1 and n = 40 readings at time 2) due to probable measurement error.
Statistical analyses
We conducted all analyses using Stata, version 15.1 (StataCorp LLC, College Station, Texas). We presented baseline characteristics of hypertensive participants, testing for differences between experiencing everyday discrimination (ever vs. never) with Pearson χ2 test for categorical variables and the Wilcoxon rank-sum test for continuous variables. We reported the percentages and counts of categorical variables and the medians and interquartile ranges of continuous variables.
We evaluated minimally and multiply adjusted associations of everyday and lifetime discrimination with antihypertensive medication use. Minimal adjustment consisted of age in years, sex, and 2008 baseline wave. Multiply adjusted estimates were derived from models that included the following variables: age, sex, racial/ethnic designation, 2008 baseline wave, US nativity, highest educational attainment of participant, marital status, highest educational attainment of father, self-rated health, the number of self-reported diagnoses, and the count of self-reported medications taken. We next fitted parallel minimally and multiply adjusting logistic regression models to independently evaluate the associations between each discrimination measure as exposure; depression symptoms, subjective social status, or wealth quintiles as mediator; and subsequent antihypertensive medication use as outcome.
We then implemented causal mediation, which improves upon the Baron and Kenny method (49) by accounting for interactive effects between discrimination (exposure) and psychosocial or economic factor (mediator) on antihypertensive medication treatment under pure (allowed to vary) and counterfactual (controlled) conditions (44). Using the Stata med4way command (50), we estimated pure and controlled effects by generating 2 sets of models: a linear regression model for the mediator, conditional on the presence of discrimination and confounders, and a logistic regression model for subsequent antihypertensive treatment conditional on the presence of discriminatory experiences, the mediator, an exposure-mediator interaction, and confounders. We therefore estimated the following components of the total effect: 1) the controlled direct effect: the odds ratios of the discrimination–antihypertensive treatment associations due to neither interaction nor mediation, in a counterfactual population comparing people experiencing no versus any discrimination where the mediating factors are fixed at ideal values; 2) the interactive-reference effect: the combined effect of discrimination and the mediating factor, where discrimination is not a determinant of the mediator; 3) the mediated-interactive effect: the odds ratio for the discrimination-treatment association due to mediation and interaction, where the values are controlled to indicate that discrimination is necessary for the mediating factor to have an indirect effect; and 4) the pure indirect effect: the odds ratio due to mediation only, where the factors mediating discrimination and antihypertensive treatment are allowed to vary (Web Figure 1, available at https://doi.org/10.1093/aje/kwac102).The ideal values specified in the controlled direct effect were 0 for depressive symptoms, 10 for subjective social standing, and 5 for quintile of wealth. The 95% confidence intervals accompanying these estimates were derived from bootstrapped standard errors using 1,000 repetitions.
Since baseline blood pressure could inform hypertension diagnosis and control, we conducted sensitivity analyses adjusting for systolic and diastolic blood pressure at time 1 among participants with complete sphygmomanometer readings (n = 7,129 for depressive symptomatology, n = 6,403 for subjective social standing, and n = 7,085 for wealth quintile).
RESULTS
Among hypertensive participants during the pooled 2008 and 2010 waves, 66% had experienced discrimination in their everyday lives, and 33% had experienced at least one of the 7 specified lifetime-discrimination circumstances. We present the baseline characteristics of participants according to frequency everyday discrimination in Table 1. Compared with participants who had never experienced everyday discrimination, participants who had ever experienced everyday discrimination were younger, less likely to be non-Hispanic White, and less likely to own their own homes. Participants who had ever experienced everyday discrimination also rated their subjective social status as 1 rung lower on the 10-rung ladder and reported 1 more depressive symptom. The average household wealth fell between $85,001 and $221,000. Although median wealth was equivalent across groups, the significance of the rank-sum test indicated inequality at the lowest and highest extremes: For example, 19% of participants experiencing everyday discrimination belonged to the lowest quintile compared with 14% of participants reporting never. Both baseline systolic blood pressure and number of medications other than antihypertensive medications were comparatively lower among participants reporting everyday discrimination. Subsequent antihypertensive medication use was higher among participants never experiencing everyday discrimination.
Characteristics of Hypertensive Health and Retirement Study Participants, by Frequency of Everyday Discrimination (n = 8,557), United States, 2008–2010
| Never Experienced Everyday Discrimination (n = 3,328) | Ever Experienced Everyday Discrimination (n = 5,229) | P Valuea | |||
|---|---|---|---|---|---|
| Characteristic | No. | % | No. | No. | |
| Age, yearsb | 71 (64–77) | 68 (60–74) | <0.001 | ||
| Male sex | 1,247 | 37 | 2,,262 | 43 | <0.001 |
| Racial/ethnic designation | <0.001 | ||||
| Non-Hispanic White | 2,487 | 75 | 3,761 | 72 | |
| Non-Hispanic Black | 493 | 15 | 981 | 19 | |
| Latino/Hispanic | 348 | 10 | 487 | 9 | |
| US nativity | 2,959 | 89 | 4,787 | 92 | <0.001 |
| Marital/partnership status | <0.001 | ||||
| Married/partnered | 1,998 | 60 | 3,130 | 60 | |
| Divorced/separated | 415 | 13 | 881 | 17 | |
| Widowed | 807 | 24 | 991 | 19 | |
| Never married/partnered | 104 | 3 | 226 | 4 | |
| Housing status | <0.001 | ||||
| Owns home | 1,962 | 59 | 2,871 | 55 | |
| Rents or lives with family | 518 | 16 | 980 | 19 | |
| Other | 848 | 25 | 1,378 | 26 | |
| Highest educational attainment | <0.001 | ||||
| Less than high school | 736 | 22 | 960 | 18 | |
| High-school diploma/GED | 1,828 | 55 | 2,916 | 56 | |
| 2-year college degree | 131 | 4 | 252 | 5 | |
| At least 4-year college degree | 631 | 19 | 1,088 | 21 | |
| Insurance type | <0.001 | ||||
| Private | 1,852 | 56 | 2,970 | 57 | |
| Only federal/public | 1,307 | 39 | 1,882 | 36 | |
| Neither public or private | 152 | 5 | 354 | 7 | |
| Paternal educational attainment | 0.31 | ||||
| Less than high school | 1,819 | 54 | 2,712 | 52 | |
| High-school diploma/GED | 659 | 20 | 1,104 | 21 | |
| 2-year college degree | 160 | 5 | 259 | 5 | |
| At least 4-year college degree | 201 | 6 | 340 | 6 | |
| Unknown | 489 | 15 | 814 | 15 | |
| Self-rated health | <0.001 | ||||
| Excellent | 250 | 7 | 257 | 5 | |
| Very good | 1,077 | 33 | 1,418 | 28 | |
| Good | 1,112 | 34 | 1,921 | 37 | |
| Fair | 671 | 20 | 1,189 | 23 | |
| Poor | 218 | 6 | 443 | 8 | |
| Quintiles of wealthb–d | 3 (2–4) | 3 (2–4) | <0.001 | ||
| Social standingb | 7 (6–8) | 6 (5–8) | <0.001 | ||
| Depressive symptomatologyb | 1 (1–1) | 2 (1–2) | <0.001 | ||
| Number of comorbiditiesb | 2 (1–2) | 2 (1–3) | <0.001 | ||
| Number of medicationsb | 7 (6–7) | 6 (6–6) | <0.001 | ||
| SBP, mg/Lb | 135 (122–148) | 132 (120–146) | <0.001 | ||
| DBP, mg/Lb | 80 (73–88) | 80 (73–88) | 0.36 | ||
| Antihypertensive use at time 2e | 3,065 | 92 | 4,703 | 90 | 0.003 |
| Never Experienced Everyday Discrimination (n = 3,328) | Ever Experienced Everyday Discrimination (n = 5,229) | P Valuea | |||
|---|---|---|---|---|---|
| Characteristic | No. | % | No. | No. | |
| Age, yearsb | 71 (64–77) | 68 (60–74) | <0.001 | ||
| Male sex | 1,247 | 37 | 2,,262 | 43 | <0.001 |
| Racial/ethnic designation | <0.001 | ||||
| Non-Hispanic White | 2,487 | 75 | 3,761 | 72 | |
| Non-Hispanic Black | 493 | 15 | 981 | 19 | |
| Latino/Hispanic | 348 | 10 | 487 | 9 | |
| US nativity | 2,959 | 89 | 4,787 | 92 | <0.001 |
| Marital/partnership status | <0.001 | ||||
| Married/partnered | 1,998 | 60 | 3,130 | 60 | |
| Divorced/separated | 415 | 13 | 881 | 17 | |
| Widowed | 807 | 24 | 991 | 19 | |
| Never married/partnered | 104 | 3 | 226 | 4 | |
| Housing status | <0.001 | ||||
| Owns home | 1,962 | 59 | 2,871 | 55 | |
| Rents or lives with family | 518 | 16 | 980 | 19 | |
| Other | 848 | 25 | 1,378 | 26 | |
| Highest educational attainment | <0.001 | ||||
| Less than high school | 736 | 22 | 960 | 18 | |
| High-school diploma/GED | 1,828 | 55 | 2,916 | 56 | |
| 2-year college degree | 131 | 4 | 252 | 5 | |
| At least 4-year college degree | 631 | 19 | 1,088 | 21 | |
| Insurance type | <0.001 | ||||
| Private | 1,852 | 56 | 2,970 | 57 | |
| Only federal/public | 1,307 | 39 | 1,882 | 36 | |
| Neither public or private | 152 | 5 | 354 | 7 | |
| Paternal educational attainment | 0.31 | ||||
| Less than high school | 1,819 | 54 | 2,712 | 52 | |
| High-school diploma/GED | 659 | 20 | 1,104 | 21 | |
| 2-year college degree | 160 | 5 | 259 | 5 | |
| At least 4-year college degree | 201 | 6 | 340 | 6 | |
| Unknown | 489 | 15 | 814 | 15 | |
| Self-rated health | <0.001 | ||||
| Excellent | 250 | 7 | 257 | 5 | |
| Very good | 1,077 | 33 | 1,418 | 28 | |
| Good | 1,112 | 34 | 1,921 | 37 | |
| Fair | 671 | 20 | 1,189 | 23 | |
| Poor | 218 | 6 | 443 | 8 | |
| Quintiles of wealthb–d | 3 (2–4) | 3 (2–4) | <0.001 | ||
| Social standingb | 7 (6–8) | 6 (5–8) | <0.001 | ||
| Depressive symptomatologyb | 1 (1–1) | 2 (1–2) | <0.001 | ||
| Number of comorbiditiesb | 2 (1–2) | 2 (1–3) | <0.001 | ||
| Number of medicationsb | 7 (6–7) | 6 (6–6) | <0.001 | ||
| SBP, mg/Lb | 135 (122–148) | 132 (120–146) | <0.001 | ||
| DBP, mg/Lb | 80 (73–88) | 80 (73–88) | 0.36 | ||
| Antihypertensive use at time 2e | 3,065 | 92 | 4,703 | 90 | 0.003 |
Abbreviations: DBP, diastolic blood pressure; GED, General Educational Development certificate; SBP, systolic blood pressure.
aP value for difference between any vs. never experiencing everyday discrimination.
b Values are expressed as median (interquartile range).
c Reported private insurance indicates employer-based or individually purchased, while federal/public indicates Medicare, Medicaid, Civilian Health and Medical Program Uniformed Services, and Department of Veterans Affairs.
d Household net wealth quintiles have the following bounds, in US dollars: 1, minus $550,000 to $9,000; 2, $9,001 to $85,000; 3, $85,001 to $221,000; 4, $222,001 to $536,500; 5, above $536,501.
e Antihypertensive use is only among participants self-reporting having been diagnosed with hypertension by a doctor.
Characteristics of Hypertensive Health and Retirement Study Participants, by Frequency of Everyday Discrimination (n = 8,557), United States, 2008–2010
| Never Experienced Everyday Discrimination (n = 3,328) | Ever Experienced Everyday Discrimination (n = 5,229) | P Valuea | |||
|---|---|---|---|---|---|
| Characteristic | No. | % | No. | No. | |
| Age, yearsb | 71 (64–77) | 68 (60–74) | <0.001 | ||
| Male sex | 1,247 | 37 | 2,,262 | 43 | <0.001 |
| Racial/ethnic designation | <0.001 | ||||
| Non-Hispanic White | 2,487 | 75 | 3,761 | 72 | |
| Non-Hispanic Black | 493 | 15 | 981 | 19 | |
| Latino/Hispanic | 348 | 10 | 487 | 9 | |
| US nativity | 2,959 | 89 | 4,787 | 92 | <0.001 |
| Marital/partnership status | <0.001 | ||||
| Married/partnered | 1,998 | 60 | 3,130 | 60 | |
| Divorced/separated | 415 | 13 | 881 | 17 | |
| Widowed | 807 | 24 | 991 | 19 | |
| Never married/partnered | 104 | 3 | 226 | 4 | |
| Housing status | <0.001 | ||||
| Owns home | 1,962 | 59 | 2,871 | 55 | |
| Rents or lives with family | 518 | 16 | 980 | 19 | |
| Other | 848 | 25 | 1,378 | 26 | |
| Highest educational attainment | <0.001 | ||||
| Less than high school | 736 | 22 | 960 | 18 | |
| High-school diploma/GED | 1,828 | 55 | 2,916 | 56 | |
| 2-year college degree | 131 | 4 | 252 | 5 | |
| At least 4-year college degree | 631 | 19 | 1,088 | 21 | |
| Insurance type | <0.001 | ||||
| Private | 1,852 | 56 | 2,970 | 57 | |
| Only federal/public | 1,307 | 39 | 1,882 | 36 | |
| Neither public or private | 152 | 5 | 354 | 7 | |
| Paternal educational attainment | 0.31 | ||||
| Less than high school | 1,819 | 54 | 2,712 | 52 | |
| High-school diploma/GED | 659 | 20 | 1,104 | 21 | |
| 2-year college degree | 160 | 5 | 259 | 5 | |
| At least 4-year college degree | 201 | 6 | 340 | 6 | |
| Unknown | 489 | 15 | 814 | 15 | |
| Self-rated health | <0.001 | ||||
| Excellent | 250 | 7 | 257 | 5 | |
| Very good | 1,077 | 33 | 1,418 | 28 | |
| Good | 1,112 | 34 | 1,921 | 37 | |
| Fair | 671 | 20 | 1,189 | 23 | |
| Poor | 218 | 6 | 443 | 8 | |
| Quintiles of wealthb–d | 3 (2–4) | 3 (2–4) | <0.001 | ||
| Social standingb | 7 (6–8) | 6 (5–8) | <0.001 | ||
| Depressive symptomatologyb | 1 (1–1) | 2 (1–2) | <0.001 | ||
| Number of comorbiditiesb | 2 (1–2) | 2 (1–3) | <0.001 | ||
| Number of medicationsb | 7 (6–7) | 6 (6–6) | <0.001 | ||
| SBP, mg/Lb | 135 (122–148) | 132 (120–146) | <0.001 | ||
| DBP, mg/Lb | 80 (73–88) | 80 (73–88) | 0.36 | ||
| Antihypertensive use at time 2e | 3,065 | 92 | 4,703 | 90 | 0.003 |
| Never Experienced Everyday Discrimination (n = 3,328) | Ever Experienced Everyday Discrimination (n = 5,229) | P Valuea | |||
|---|---|---|---|---|---|
| Characteristic | No. | % | No. | No. | |
| Age, yearsb | 71 (64–77) | 68 (60–74) | <0.001 | ||
| Male sex | 1,247 | 37 | 2,,262 | 43 | <0.001 |
| Racial/ethnic designation | <0.001 | ||||
| Non-Hispanic White | 2,487 | 75 | 3,761 | 72 | |
| Non-Hispanic Black | 493 | 15 | 981 | 19 | |
| Latino/Hispanic | 348 | 10 | 487 | 9 | |
| US nativity | 2,959 | 89 | 4,787 | 92 | <0.001 |
| Marital/partnership status | <0.001 | ||||
| Married/partnered | 1,998 | 60 | 3,130 | 60 | |
| Divorced/separated | 415 | 13 | 881 | 17 | |
| Widowed | 807 | 24 | 991 | 19 | |
| Never married/partnered | 104 | 3 | 226 | 4 | |
| Housing status | <0.001 | ||||
| Owns home | 1,962 | 59 | 2,871 | 55 | |
| Rents or lives with family | 518 | 16 | 980 | 19 | |
| Other | 848 | 25 | 1,378 | 26 | |
| Highest educational attainment | <0.001 | ||||
| Less than high school | 736 | 22 | 960 | 18 | |
| High-school diploma/GED | 1,828 | 55 | 2,916 | 56 | |
| 2-year college degree | 131 | 4 | 252 | 5 | |
| At least 4-year college degree | 631 | 19 | 1,088 | 21 | |
| Insurance type | <0.001 | ||||
| Private | 1,852 | 56 | 2,970 | 57 | |
| Only federal/public | 1,307 | 39 | 1,882 | 36 | |
| Neither public or private | 152 | 5 | 354 | 7 | |
| Paternal educational attainment | 0.31 | ||||
| Less than high school | 1,819 | 54 | 2,712 | 52 | |
| High-school diploma/GED | 659 | 20 | 1,104 | 21 | |
| 2-year college degree | 160 | 5 | 259 | 5 | |
| At least 4-year college degree | 201 | 6 | 340 | 6 | |
| Unknown | 489 | 15 | 814 | 15 | |
| Self-rated health | <0.001 | ||||
| Excellent | 250 | 7 | 257 | 5 | |
| Very good | 1,077 | 33 | 1,418 | 28 | |
| Good | 1,112 | 34 | 1,921 | 37 | |
| Fair | 671 | 20 | 1,189 | 23 | |
| Poor | 218 | 6 | 443 | 8 | |
| Quintiles of wealthb–d | 3 (2–4) | 3 (2–4) | <0.001 | ||
| Social standingb | 7 (6–8) | 6 (5–8) | <0.001 | ||
| Depressive symptomatologyb | 1 (1–1) | 2 (1–2) | <0.001 | ||
| Number of comorbiditiesb | 2 (1–2) | 2 (1–3) | <0.001 | ||
| Number of medicationsb | 7 (6–7) | 6 (6–6) | <0.001 | ||
| SBP, mg/Lb | 135 (122–148) | 132 (120–146) | <0.001 | ||
| DBP, mg/Lb | 80 (73–88) | 80 (73–88) | 0.36 | ||
| Antihypertensive use at time 2e | 3,065 | 92 | 4,703 | 90 | 0.003 |
Abbreviations: DBP, diastolic blood pressure; GED, General Educational Development certificate; SBP, systolic blood pressure.
aP value for difference between any vs. never experiencing everyday discrimination.
b Values are expressed as median (interquartile range).
c Reported private insurance indicates employer-based or individually purchased, while federal/public indicates Medicare, Medicaid, Civilian Health and Medical Program Uniformed Services, and Department of Veterans Affairs.
d Household net wealth quintiles have the following bounds, in US dollars: 1, minus $550,000 to $9,000; 2, $9,001 to $85,000; 3, $85,001 to $221,000; 4, $222,001 to $536,500; 5, above $536,501.
e Antihypertensive use is only among participants self-reporting having been diagnosed with hypertension by a doctor.
In unmediated estimates derived from multiply adjusted models presented in Table 2, experiencing more frequent everyday discrimination was associated with lower odds (odds ratio (OR) = 0.86, 95% CI: 0.78, 0.95) of antihypertensive use 4 years later. Similarly, participants reporting more experiences of lifetime discrimination had a lower likelihood of subsequent antihypertensive medication use (OR = 0.91, 95% CI: 0.85, 0.98). Comparable estimates were observed among the sample with available baseline systolic and diastolic readings (OR for experiencing more frequent everyday discrimination = 0.86 (95% CI: 0.77, 0.96) and for experiencing a higher sum of lifetime discrimination = 0.90 (95% CI: 0.83, 0.97); Web Table 1). The associations between discrimination and mediating psychosocial or economic factors are presented in Table 3. Both everyday discrimination and lifetime discrimination were independently associated with modest increases in depressive symptomatology, decreases in subjective social status, and belonging to a lower wealth quintile. As examples, reporting more frequent experiencing of everyday discrimination was associated with a higher CES-D 8 score (β = 0.45, 95% CI: 0.40, 0.51), while experiencing a higher sum of lifetime discrimination was associated with a subjective social standing lower than the sample mean (β = –0.16, 95% CI: −0.19, −0.12).
Associations Between Perceived Discrimination and Subsequent Antihypertensive Treatment, Health and Retirement Study, United States, 2008–2014
| Perceived Discrimination Subscale | Minimally Adjusteda | Multivariable-Adjustedb | ||||
|---|---|---|---|---|---|---|
| OR | 95% CI | P Value | OR | 95% CI | P Value | |
| Frequency of everyday discrimination (1–6) | 0.89 | 0.81, 0.98 | 0.02 | 0.86 | 0.78, 0.95 | 0.003 |
| Sum of lifetime discrimination (0–7) | 0.94 | 0.88, 1.01 | 0.08 | 0.91 | 0.85, 0.98 | 0.01 |
| Perceived Discrimination Subscale | Minimally Adjusteda | Multivariable-Adjustedb | ||||
|---|---|---|---|---|---|---|
| OR | 95% CI | P Value | OR | 95% CI | P Value | |
| Frequency of everyday discrimination (1–6) | 0.89 | 0.81, 0.98 | 0.02 | 0.86 | 0.78, 0.95 | 0.003 |
| Sum of lifetime discrimination (0–7) | 0.94 | 0.88, 1.01 | 0.08 | 0.91 | 0.85, 0.98 | 0.01 |
Abbreviations: CI, confidence interval; GED, General Educational Development certificate; OR, odds ratio.
a Adjusted for 2008 versus 2010 baseline wave, age in years, and sex.
b Adjusted for 2008 versus 2010 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 diploma/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 father (less than high school, high-school diploma/GED, 2-year college degree, 4-year college degree, unknown), household net income relative to 2000 poverty line (no income, unknown, up to twice above, 2–4 times above, more than 4 times above), insurance type (private, only federal/public, neither), self-rated health (excellent, very good, good, fair, poor), and number of self-reported diagnoses and medication use for the following comorbidities: diabetes, cancer, lung disease, heart disease (e.g., heart attack, angina, congestive heart failure), stroke, psychiatric conditions, arthritis, and memory-related disease.
Associations Between Perceived Discrimination and Subsequent Antihypertensive Treatment, Health and Retirement Study, United States, 2008–2014
| Perceived Discrimination Subscale | Minimally Adjusteda | Multivariable-Adjustedb | ||||
|---|---|---|---|---|---|---|
| OR | 95% CI | P Value | OR | 95% CI | P Value | |
| Frequency of everyday discrimination (1–6) | 0.89 | 0.81, 0.98 | 0.02 | 0.86 | 0.78, 0.95 | 0.003 |
| Sum of lifetime discrimination (0–7) | 0.94 | 0.88, 1.01 | 0.08 | 0.91 | 0.85, 0.98 | 0.01 |
| Perceived Discrimination Subscale | Minimally Adjusteda | Multivariable-Adjustedb | ||||
|---|---|---|---|---|---|---|
| OR | 95% CI | P Value | OR | 95% CI | P Value | |
| Frequency of everyday discrimination (1–6) | 0.89 | 0.81, 0.98 | 0.02 | 0.86 | 0.78, 0.95 | 0.003 |
| Sum of lifetime discrimination (0–7) | 0.94 | 0.88, 1.01 | 0.08 | 0.91 | 0.85, 0.98 | 0.01 |
Abbreviations: CI, confidence interval; GED, General Educational Development certificate; OR, odds ratio.
a Adjusted for 2008 versus 2010 baseline wave, age in years, and sex.
b Adjusted for 2008 versus 2010 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 diploma/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 father (less than high school, high-school diploma/GED, 2-year college degree, 4-year college degree, unknown), household net income relative to 2000 poverty line (no income, unknown, up to twice above, 2–4 times above, more than 4 times above), insurance type (private, only federal/public, neither), self-rated health (excellent, very good, good, fair, poor), and number of self-reported diagnoses and medication use for the following comorbidities: diabetes, cancer, lung disease, heart disease (e.g., heart attack, angina, congestive heart failure), stroke, psychiatric conditions, arthritis, and memory-related disease.
Associations Between Perceived Discrimination and Mediating Psychosocial or Economic Factors, Health and Retirement Study, United States, 2008–2010
| Perceived Discrimination Subscale | Minimally Adjusteda | Multivariable-Adjustedb | ||||
|---|---|---|---|---|---|---|
| β | 95% CI | P Value | β | 95% CI | P Value | |
| Frequency of everyday discrimination (1–6) | ||||||
| Depressive symptoms (0–8)c | 0.67 | 0.61, 0.73 | <0.001 | 0.45 | 0.40, 0.51 | <0.001 |
| Social standing (1–10)d | −0.46 | −0.52, −0.41 | <0.001 | −0.30 | −0.35, −0.25 | <0.001 |
| Quintiles of wealth (1–5)e | −0.26 | −0.30, −0.22 | <0.001 | −0.10 | −0.14, −0.07 | <0.001 |
| Sum of lifetime discrimination (0–7) | ||||||
| Depressive symptoms (0–8)c | 0.25 | 0.21, 0.29 | <0.001 | 0.15 | 0.11, 0.19 | <0.001 |
| Social standing(1–10)d | −0.21 | −0.25, −0.16 | <0.001 | −0.16 | −0.19, −0.12 | <0.001 |
| Quintiles of wealth (1–5)e | −0.16 | −0.19, −0.13 | <0.001 | −0.10 | −0.12, −0.07 | <0.001 |
| Perceived Discrimination Subscale | Minimally Adjusteda | Multivariable-Adjustedb | ||||
|---|---|---|---|---|---|---|
| β | 95% CI | P Value | β | 95% CI | P Value | |
| Frequency of everyday discrimination (1–6) | ||||||
| Depressive symptoms (0–8)c | 0.67 | 0.61, 0.73 | <0.001 | 0.45 | 0.40, 0.51 | <0.001 |
| Social standing (1–10)d | −0.46 | −0.52, −0.41 | <0.001 | −0.30 | −0.35, −0.25 | <0.001 |
| Quintiles of wealth (1–5)e | −0.26 | −0.30, −0.22 | <0.001 | −0.10 | −0.14, −0.07 | <0.001 |
| Sum of lifetime discrimination (0–7) | ||||||
| Depressive symptoms (0–8)c | 0.25 | 0.21, 0.29 | <0.001 | 0.15 | 0.11, 0.19 | <0.001 |
| Social standing(1–10)d | −0.21 | −0.25, −0.16 | <0.001 | −0.16 | −0.19, −0.12 | <0.001 |
| Quintiles of wealth (1–5)e | −0.16 | −0.19, −0.13 | <0.001 | −0.10 | −0.12, −0.07 | <0.001 |
Abbreviations: CI, confidence interval; GED, General Educational Development certificate.
a Adjusted for 2008 versus 2010 baseline wave, age in years, and sex.
b Adjusted for 2008 versus 2010 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 diploma/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 father (less than high school, high-school diploma/GED, 2-year college degree, 4-year college degree, unknown), household net income relative to 2000 poverty line (no income, unknown, up to twice above, 2–4 times above, more than 4 times above), insurance type (private, only federal/public, neither), self-rated health (excellent, very good, good, fair, poor), and number of self-reported diagnoses and medication use for the following comorbidities: diabetes, cancer, lung disease, heart disease (e.g., heart attack, angina, congestive heart failure), stroke, psychiatric conditions, arthritis, and memory-related disease.
cn = 8,112.
dn = 7,470.
en = 8,292.
Associations Between Perceived Discrimination and Mediating Psychosocial or Economic Factors, Health and Retirement Study, United States, 2008–2010
| Perceived Discrimination Subscale | Minimally Adjusteda | Multivariable-Adjustedb | ||||
|---|---|---|---|---|---|---|
| β | 95% CI | P Value | β | 95% CI | P Value | |
| Frequency of everyday discrimination (1–6) | ||||||
| Depressive symptoms (0–8)c | 0.67 | 0.61, 0.73 | <0.001 | 0.45 | 0.40, 0.51 | <0.001 |
| Social standing (1–10)d | −0.46 | −0.52, −0.41 | <0.001 | −0.30 | −0.35, −0.25 | <0.001 |
| Quintiles of wealth (1–5)e | −0.26 | −0.30, −0.22 | <0.001 | −0.10 | −0.14, −0.07 | <0.001 |
| Sum of lifetime discrimination (0–7) | ||||||
| Depressive symptoms (0–8)c | 0.25 | 0.21, 0.29 | <0.001 | 0.15 | 0.11, 0.19 | <0.001 |
| Social standing(1–10)d | −0.21 | −0.25, −0.16 | <0.001 | −0.16 | −0.19, −0.12 | <0.001 |
| Quintiles of wealth (1–5)e | −0.16 | −0.19, −0.13 | <0.001 | −0.10 | −0.12, −0.07 | <0.001 |
| Perceived Discrimination Subscale | Minimally Adjusteda | Multivariable-Adjustedb | ||||
|---|---|---|---|---|---|---|
| β | 95% CI | P Value | β | 95% CI | P Value | |
| Frequency of everyday discrimination (1–6) | ||||||
| Depressive symptoms (0–8)c | 0.67 | 0.61, 0.73 | <0.001 | 0.45 | 0.40, 0.51 | <0.001 |
| Social standing (1–10)d | −0.46 | −0.52, −0.41 | <0.001 | −0.30 | −0.35, −0.25 | <0.001 |
| Quintiles of wealth (1–5)e | −0.26 | −0.30, −0.22 | <0.001 | −0.10 | −0.14, −0.07 | <0.001 |
| Sum of lifetime discrimination (0–7) | ||||||
| Depressive symptoms (0–8)c | 0.25 | 0.21, 0.29 | <0.001 | 0.15 | 0.11, 0.19 | <0.001 |
| Social standing(1–10)d | −0.21 | −0.25, −0.16 | <0.001 | −0.16 | −0.19, −0.12 | <0.001 |
| Quintiles of wealth (1–5)e | −0.16 | −0.19, −0.13 | <0.001 | −0.10 | −0.12, −0.07 | <0.001 |
Abbreviations: CI, confidence interval; GED, General Educational Development certificate.
a Adjusted for 2008 versus 2010 baseline wave, age in years, and sex.
b Adjusted for 2008 versus 2010 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 diploma/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 father (less than high school, high-school diploma/GED, 2-year college degree, 4-year college degree, unknown), household net income relative to 2000 poverty line (no income, unknown, up to twice above, 2–4 times above, more than 4 times above), insurance type (private, only federal/public, neither), self-rated health (excellent, very good, good, fair, poor), and number of self-reported diagnoses and medication use for the following comorbidities: diabetes, cancer, lung disease, heart disease (e.g., heart attack, angina, congestive heart failure), stroke, psychiatric conditions, arthritis, and memory-related disease.
cn = 8,112.
dn = 7,470.
en = 8,292.
The decomposed components of the total associations between discrimination, mediating factors, and subsequent antihypertensive medication use are presented in Table 4. The significant overall associations of both discrimination measures and antihypertensive use, as mediated by depressive symptoms, were corollaries of the “neither interaction nor mediation” components (controlled direct effect for everyday discrimination, OR = 0.90 (95% CI: 0.79, 1.03); for lifetime discrimination, OR = 0.95 (95% CI: 0.86, 1.04)). The associations between lifetime discrimination, subjective social standing, and antihypertensive use followed similar trends. Experiencing lifetime discrimination was indirectly associated with antihypertensive use due to depressive symptomatology (pure indirect effect, OR = 0.99, 95% CI: 0.98, 1.00), although the estimate was imprecise. There was not an indication that subjective social standing or household wealth had interactive or mediating effects on either discrimination–antihypertensive use association. We did not observe significant associations between either discrimination measure and antihypertensive use due to interaction in conjunction with mediation by any factor. Further adjustment for baseline systolic and diastolic blood pressure among participants with available sphygmomanometer readings did not significantly change unmediated or decomposed estimates (Web Tables 1–3).
Decomposed Effects of Perceived Discrimination and Odds of Subsequent Antihypertensive Treatment, Health and Retirement Study, United States, 2008–2014
| Perceived Discrimination Subscale | CDEa | Reference Interaction | Mediated Interaction | PIE | TE | |||||
|---|---|---|---|---|---|---|---|---|---|---|
| OR | 95% CIb | OR | 95% CIb | OR | 95% CIb | OR | 95% CIb | OR | 95% CIb | |
| Frequency of Everyday Discrimination (1–6) | ||||||||||
| Minimally adjustedc | ||||||||||
| Depressive symptoms (0–8)d | 0.91 | 0.80, 1.03 | 1.00 | 0.80, 1.03 | 1.01 | 0.98, 1.03 | 0.97 | 0.92, 1.02 | 0.89 | 0.78, 1.02 |
| Social standing (1–10)e | 1.05 | 0.85, 1.29 | 0.90 | 0.79, 1.01 | 0.98 | 0.96, 1.00 | 1.03 | 0.99, 1.08 | 0.95 | 0.83, 1.08 |
| Quintiles of wealth (1–5)f | 0.80 | 0.66, 0.98 | 1.07 | 0.95, 1.21 | 1.01 | 0.99, 1.03 | 0.99 | 0.95, 1.02 | 0.86 | 0.77, 0.96 |
| Multivariable-adjustedg | ||||||||||
| Depressive symptoms (0–8)d | 0.90 | 0.79, 1.03 | 1.00 | 0.79, 1.03 | 1.00 | 0.98, 1.03 | 0.97 | 0.93, 1.02 | 0.88 | 0.78, 0.99 |
| Social standing (1–10)e | 0.99 | 0.79, 1.23 | 0.91 | 0.79, 1.05 | 0.99 | 0.97, 1.01 | 1.01 | 0.99, 1.04 | 0.90 | 0.80, 1.02 |
| Quintiles of wealth (1–5)f | 0.75 | 0.60, 0.94 | 1.12 | 0.95, 1.31 | 1.01 | 1.00, 1.01 | 0.99 | 0.98, 1.01 | 0.84 | 0.76, 0.92 |
| Sum of Lifetime Discrimination (0–7) | ||||||||||
| Minimally adjustedc | ||||||||||
| Depressive symptoms (0–8)d | 0.95 | 0.87, 1.03 | 1.00 | 0.97, 1.03 | 1.00 | 0.99, 1.01 | 0.99 | 0.98, 1.00 | 0.86 | 0.75, 0.99 |
| Social standing (1–10)e | 0.94 | 0.79, 1.11 | 1.00 | 0.88, 1.15 | 1.00 | 0.99, 1.01 | 1.00 | 0.99, 1.01 | 0.94 | 0.88, 1.01 |
| Quintiles of wealth (1–5)f | 0.88 | 0.77, 0.99 | 1.05 | 0.97, 1.14 | 1.00 | 1.00, 1.01 | 1.00 | 0.99, 1.01 | 0.93 | 0.86, 0.99 |
| Multivariable-adjustedg | ||||||||||
| Depressive symptoms (0–8)d | 0.95 | 0.86, 1.04 | 1.00 | 0.97, 1.04 | 1.00 | 1.00, 1.00 | 0.99 | 0.98, 1.00 | 0.86 | 0.74, 0.98 |
| Social standing (1–10)e | 0.90 | 0.75, 1.08 | 1.03 | 0.89, 1.20 | 1.01 | 1.00, 1.01 | 1.00 | 0.99, 1.01 | 0.93 | 0.86, 1.00 |
| Quintiles of wealth (1–5)f | 0.86 | 0.76, 0.98 | 1.06 | 0.97, 1.15 | 1.00 | 1.00, 1.01 | 1.00 | 0.99, 1.01 | 0.91 | 0.74, 0.98 |
| Perceived Discrimination Subscale | CDEa | Reference Interaction | Mediated Interaction | PIE | TE | |||||
|---|---|---|---|---|---|---|---|---|---|---|
| OR | 95% CIb | OR | 95% CIb | OR | 95% CIb | OR | 95% CIb | OR | 95% CIb | |
| Frequency of Everyday Discrimination (1–6) | ||||||||||
| Minimally adjustedc | ||||||||||
| Depressive symptoms (0–8)d | 0.91 | 0.80, 1.03 | 1.00 | 0.80, 1.03 | 1.01 | 0.98, 1.03 | 0.97 | 0.92, 1.02 | 0.89 | 0.78, 1.02 |
| Social standing (1–10)e | 1.05 | 0.85, 1.29 | 0.90 | 0.79, 1.01 | 0.98 | 0.96, 1.00 | 1.03 | 0.99, 1.08 | 0.95 | 0.83, 1.08 |
| Quintiles of wealth (1–5)f | 0.80 | 0.66, 0.98 | 1.07 | 0.95, 1.21 | 1.01 | 0.99, 1.03 | 0.99 | 0.95, 1.02 | 0.86 | 0.77, 0.96 |
| Multivariable-adjustedg | ||||||||||
| Depressive symptoms (0–8)d | 0.90 | 0.79, 1.03 | 1.00 | 0.79, 1.03 | 1.00 | 0.98, 1.03 | 0.97 | 0.93, 1.02 | 0.88 | 0.78, 0.99 |
| Social standing (1–10)e | 0.99 | 0.79, 1.23 | 0.91 | 0.79, 1.05 | 0.99 | 0.97, 1.01 | 1.01 | 0.99, 1.04 | 0.90 | 0.80, 1.02 |
| Quintiles of wealth (1–5)f | 0.75 | 0.60, 0.94 | 1.12 | 0.95, 1.31 | 1.01 | 1.00, 1.01 | 0.99 | 0.98, 1.01 | 0.84 | 0.76, 0.92 |
| Sum of Lifetime Discrimination (0–7) | ||||||||||
| Minimally adjustedc | ||||||||||
| Depressive symptoms (0–8)d | 0.95 | 0.87, 1.03 | 1.00 | 0.97, 1.03 | 1.00 | 0.99, 1.01 | 0.99 | 0.98, 1.00 | 0.86 | 0.75, 0.99 |
| Social standing (1–10)e | 0.94 | 0.79, 1.11 | 1.00 | 0.88, 1.15 | 1.00 | 0.99, 1.01 | 1.00 | 0.99, 1.01 | 0.94 | 0.88, 1.01 |
| Quintiles of wealth (1–5)f | 0.88 | 0.77, 0.99 | 1.05 | 0.97, 1.14 | 1.00 | 1.00, 1.01 | 1.00 | 0.99, 1.01 | 0.93 | 0.86, 0.99 |
| Multivariable-adjustedg | ||||||||||
| Depressive symptoms (0–8)d | 0.95 | 0.86, 1.04 | 1.00 | 0.97, 1.04 | 1.00 | 1.00, 1.00 | 0.99 | 0.98, 1.00 | 0.86 | 0.74, 0.98 |
| Social standing (1–10)e | 0.90 | 0.75, 1.08 | 1.03 | 0.89, 1.20 | 1.01 | 1.00, 1.01 | 1.00 | 0.99, 1.01 | 0.93 | 0.86, 1.00 |
| Quintiles of wealth (1–5)f | 0.86 | 0.76, 0.98 | 1.06 | 0.97, 1.15 | 1.00 | 1.00, 1.01 | 1.00 | 0.99, 1.01 | 0.91 | 0.74, 0.98 |
Abbreviations: CDE, controlled direct effect; CI, confidence interval; GED, General Educational Development certificate; OR, odds ratio; PIE, pure indirect effect; TE, total effect.
a Mediator level in controlled direct effect has been set to the ideal value (i.e., 0 for depressive symptoms, 10 for social standing, and 5 for quintile of wealth).
b Confidence intervals derived from bootstrapped standard errors using 1,000 repetitions.
c Adjusted for 2008 versus 2010 baseline wave, age in years, and sex.
dn = 8,112.
en = 7,470.
fn = 8,292.
g Adjusted for 2008 versus 2010 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 diploma/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 father (less than high school, high-school diploma/GED, 2-year college degree, 4-year college degree, unknown), household net income relative to 2000 poverty line (no income, unknown, up to twice above, 2–4 times above, more than 4 times above), insurance type (private, only federal/public, neither), self-rated health (excellent, very good, good, fair, poor), and number of self-reported diagnoses and medication use for the following comorbidities: diabetes, cancer, lung disease, heart disease (e.g., heart attack, angina, congestive heart failure), stroke, psychiatric conditions, arthritis, and memory-related disease.
Decomposed Effects of Perceived Discrimination and Odds of Subsequent Antihypertensive Treatment, Health and Retirement Study, United States, 2008–2014
| Perceived Discrimination Subscale | CDEa | Reference Interaction | Mediated Interaction | PIE | TE | |||||
|---|---|---|---|---|---|---|---|---|---|---|
| OR | 95% CIb | OR | 95% CIb | OR | 95% CIb | OR | 95% CIb | OR | 95% CIb | |
| Frequency of Everyday Discrimination (1–6) | ||||||||||
| Minimally adjustedc | ||||||||||
| Depressive symptoms (0–8)d | 0.91 | 0.80, 1.03 | 1.00 | 0.80, 1.03 | 1.01 | 0.98, 1.03 | 0.97 | 0.92, 1.02 | 0.89 | 0.78, 1.02 |
| Social standing (1–10)e | 1.05 | 0.85, 1.29 | 0.90 | 0.79, 1.01 | 0.98 | 0.96, 1.00 | 1.03 | 0.99, 1.08 | 0.95 | 0.83, 1.08 |
| Quintiles of wealth (1–5)f | 0.80 | 0.66, 0.98 | 1.07 | 0.95, 1.21 | 1.01 | 0.99, 1.03 | 0.99 | 0.95, 1.02 | 0.86 | 0.77, 0.96 |
| Multivariable-adjustedg | ||||||||||
| Depressive symptoms (0–8)d | 0.90 | 0.79, 1.03 | 1.00 | 0.79, 1.03 | 1.00 | 0.98, 1.03 | 0.97 | 0.93, 1.02 | 0.88 | 0.78, 0.99 |
| Social standing (1–10)e | 0.99 | 0.79, 1.23 | 0.91 | 0.79, 1.05 | 0.99 | 0.97, 1.01 | 1.01 | 0.99, 1.04 | 0.90 | 0.80, 1.02 |
| Quintiles of wealth (1–5)f | 0.75 | 0.60, 0.94 | 1.12 | 0.95, 1.31 | 1.01 | 1.00, 1.01 | 0.99 | 0.98, 1.01 | 0.84 | 0.76, 0.92 |
| Sum of Lifetime Discrimination (0–7) | ||||||||||
| Minimally adjustedc | ||||||||||
| Depressive symptoms (0–8)d | 0.95 | 0.87, 1.03 | 1.00 | 0.97, 1.03 | 1.00 | 0.99, 1.01 | 0.99 | 0.98, 1.00 | 0.86 | 0.75, 0.99 |
| Social standing (1–10)e | 0.94 | 0.79, 1.11 | 1.00 | 0.88, 1.15 | 1.00 | 0.99, 1.01 | 1.00 | 0.99, 1.01 | 0.94 | 0.88, 1.01 |
| Quintiles of wealth (1–5)f | 0.88 | 0.77, 0.99 | 1.05 | 0.97, 1.14 | 1.00 | 1.00, 1.01 | 1.00 | 0.99, 1.01 | 0.93 | 0.86, 0.99 |
| Multivariable-adjustedg | ||||||||||
| Depressive symptoms (0–8)d | 0.95 | 0.86, 1.04 | 1.00 | 0.97, 1.04 | 1.00 | 1.00, 1.00 | 0.99 | 0.98, 1.00 | 0.86 | 0.74, 0.98 |
| Social standing (1–10)e | 0.90 | 0.75, 1.08 | 1.03 | 0.89, 1.20 | 1.01 | 1.00, 1.01 | 1.00 | 0.99, 1.01 | 0.93 | 0.86, 1.00 |
| Quintiles of wealth (1–5)f | 0.86 | 0.76, 0.98 | 1.06 | 0.97, 1.15 | 1.00 | 1.00, 1.01 | 1.00 | 0.99, 1.01 | 0.91 | 0.74, 0.98 |
| Perceived Discrimination Subscale | CDEa | Reference Interaction | Mediated Interaction | PIE | TE | |||||
|---|---|---|---|---|---|---|---|---|---|---|
| OR | 95% CIb | OR | 95% CIb | OR | 95% CIb | OR | 95% CIb | OR | 95% CIb | |
| Frequency of Everyday Discrimination (1–6) | ||||||||||
| Minimally adjustedc | ||||||||||
| Depressive symptoms (0–8)d | 0.91 | 0.80, 1.03 | 1.00 | 0.80, 1.03 | 1.01 | 0.98, 1.03 | 0.97 | 0.92, 1.02 | 0.89 | 0.78, 1.02 |
| Social standing (1–10)e | 1.05 | 0.85, 1.29 | 0.90 | 0.79, 1.01 | 0.98 | 0.96, 1.00 | 1.03 | 0.99, 1.08 | 0.95 | 0.83, 1.08 |
| Quintiles of wealth (1–5)f | 0.80 | 0.66, 0.98 | 1.07 | 0.95, 1.21 | 1.01 | 0.99, 1.03 | 0.99 | 0.95, 1.02 | 0.86 | 0.77, 0.96 |
| Multivariable-adjustedg | ||||||||||
| Depressive symptoms (0–8)d | 0.90 | 0.79, 1.03 | 1.00 | 0.79, 1.03 | 1.00 | 0.98, 1.03 | 0.97 | 0.93, 1.02 | 0.88 | 0.78, 0.99 |
| Social standing (1–10)e | 0.99 | 0.79, 1.23 | 0.91 | 0.79, 1.05 | 0.99 | 0.97, 1.01 | 1.01 | 0.99, 1.04 | 0.90 | 0.80, 1.02 |
| Quintiles of wealth (1–5)f | 0.75 | 0.60, 0.94 | 1.12 | 0.95, 1.31 | 1.01 | 1.00, 1.01 | 0.99 | 0.98, 1.01 | 0.84 | 0.76, 0.92 |
| Sum of Lifetime Discrimination (0–7) | ||||||||||
| Minimally adjustedc | ||||||||||
| Depressive symptoms (0–8)d | 0.95 | 0.87, 1.03 | 1.00 | 0.97, 1.03 | 1.00 | 0.99, 1.01 | 0.99 | 0.98, 1.00 | 0.86 | 0.75, 0.99 |
| Social standing (1–10)e | 0.94 | 0.79, 1.11 | 1.00 | 0.88, 1.15 | 1.00 | 0.99, 1.01 | 1.00 | 0.99, 1.01 | 0.94 | 0.88, 1.01 |
| Quintiles of wealth (1–5)f | 0.88 | 0.77, 0.99 | 1.05 | 0.97, 1.14 | 1.00 | 1.00, 1.01 | 1.00 | 0.99, 1.01 | 0.93 | 0.86, 0.99 |
| Multivariable-adjustedg | ||||||||||
| Depressive symptoms (0–8)d | 0.95 | 0.86, 1.04 | 1.00 | 0.97, 1.04 | 1.00 | 1.00, 1.00 | 0.99 | 0.98, 1.00 | 0.86 | 0.74, 0.98 |
| Social standing (1–10)e | 0.90 | 0.75, 1.08 | 1.03 | 0.89, 1.20 | 1.01 | 1.00, 1.01 | 1.00 | 0.99, 1.01 | 0.93 | 0.86, 1.00 |
| Quintiles of wealth (1–5)f | 0.86 | 0.76, 0.98 | 1.06 | 0.97, 1.15 | 1.00 | 1.00, 1.01 | 1.00 | 0.99, 1.01 | 0.91 | 0.74, 0.98 |
Abbreviations: CDE, controlled direct effect; CI, confidence interval; GED, General Educational Development certificate; OR, odds ratio; PIE, pure indirect effect; TE, total effect.
a Mediator level in controlled direct effect has been set to the ideal value (i.e., 0 for depressive symptoms, 10 for social standing, and 5 for quintile of wealth).
b Confidence intervals derived from bootstrapped standard errors using 1,000 repetitions.
c Adjusted for 2008 versus 2010 baseline wave, age in years, and sex.
dn = 8,112.
en = 7,470.
fn = 8,292.
g Adjusted for 2008 versus 2010 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 diploma/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 father (less than high school, high-school diploma/GED, 2-year college degree, 4-year college degree, unknown), household net income relative to 2000 poverty line (no income, unknown, up to twice above, 2–4 times above, more than 4 times above), insurance type (private, only federal/public, neither), self-rated health (excellent, very good, good, fair, poor), and number of self-reported diagnoses and medication use for the following comorbidities: diabetes, cancer, lung disease, heart disease (e.g., heart attack, angina, congestive heart failure), stroke, psychiatric conditions, arthritis, and memory-related disease.
DISCUSSION
We observed that ever experiencing everyday discrimination such as harassment, as well as discrimination in major life domains, were each associated with reductions in the likelihood of subsequent antihypertensive use. Among our sample, lifetime discrimination indirectly contributed to suboptimal antihypertensive therapy via depressive symptomatology. Subjective social standing or level of wealth did not appear to have mediating or interacting effects on associations between discrimination and antihypertensive use. Our findings ultimately indicate that multifactorial social determinants of health impede effective blood pressure control.
Comparison with other studies
By analyzing data from an older, multiethnic adult cohort, we expand both the scope and generalizability of prior work on how adverse psychosocial or economic factors impair antihypertensive therapy. Our results correspond to research conducted in the Framingham Heart Study, where each 5-unit increase in a 20-item CES-D score was associated with a 45% higher likelihood of medication nonadherence (24). Building upon observational findings (23–25), depressed cardiac patients randomized to a collaborative care intervention that managed their somatic and anhedonia symptoms reported increased medication adherence over 6 months of follow-up (21).
Since educational attainment and occupational standing preceded the time frame for many discriminatory circumstances specified on the survey, contemporaneous wealth was the only available SES measure that could plausibly mediate associations between midlife discriminatory experiences and antihypertensive therapy. Other studies have similarly investigated the limitations of traditional socioeconomic indicators among diverse aging populations with chronic conditions. A pooled analysis found that high- compared with low-SES adults had an OR of 0.89 (95% CI: 0.87, 0.92) for antihypertensive nonadherence; the authors attributed the substantial heterogeneity across the 30 included English and French studies to inconsistent SES assessment (5). Furthermore, while household income or education were not associated with antihypertensive adherence among a rural North Carolina sample (n = 495), Black but not White adults with lower subjective social standing were less likely to adhere to antihypertensive therapy (20).
Intergenerational wealth, historical discrimination, fluctuating earning power, and stability of health-care coverage plausibly first confound and then mediate associations of discrimination with antihypertensive use (14, 51). These key covariates are not explicitly measured in our sample. Beyond inadequate information on life course, class is both a cause and consequence of discrimination; nearly one-fifth of HRS participants attribute everyday discrimination to financial status (52). While adults belonging to groups historically considered dominant in society may receive health benefits from upward social mobility, adults of marginalized backgrounds may both continue to experience unequal treatment and receive reduced cardiovascular health–preserving benefit from higher educational attainment and better occupational opportunities (53, 54). Most data available to assess associations between wealth and antihypertensive therapy remain confounded by multilevel, life-course factors.
Strengths and limitations
In our study, we implemented causal mediation analysis to parse an underexplored aspect of hypertensive disparities: how discrimination contributes, via adverse psychosocial or economic factors, to suboptimal antihypertensive therapy. We established temporality between exposure and outcome, better allowing us to distinguish whether perceived discrimination contributed to antihypertensive use, compared with whether a frail or ill person receiving hypertensive treatment subsequently experienced ageism or ableism.
We recognize further limitations inherent to nonclinical data. The available antihypertensive use information was based on self-report instead of measurement or prescription refill databases. Suspected nondifferential reporting means that the true association between discrimination and medication use is likely stronger. We also do not have complete information about either comorbidity duration or class of medications, which can lower the likelihood of adherence (55, 56). Reported age of hypertension onset or history of antihypertensive use are not available prior to entry into HRS; therefore, neither variable could be factored into our models. Antihypertensive status was assessed only among participants reporting a hypertension diagnosis. The ability to evaluate discordance between reported and measured hypertension in the HRS sample is restricted to the 62% of the sample with both sources of data (36).
The external validity of our findings was limited by issues common to cohorts, namely data collection and participant retention. Health-care access and health literacy may have systematically biased our results, based on condition self-report, towards the null. Furthermore, exclusion of participants with missing covariate data limited the generalizability of our findings to marginalized groups. Due to the distribution of perceived discrimination and mediating psychosocial or economic measures, we had limited power to evaluate small interactive and mediating effect sizes. Although the discrimination scales may be collinear with other psychometric scales beyond the correlated confounding (45), their simultaneous evaluation can inform effective health behavior interventions.
Conclusion
Our work has evaluated the theorized mechanisms linking marginalization with barriers to pharmacological blood pressure control among older adults. Sociodemographic characteristics traditionally considered risk factors for suboptimal antihypertensive therapy strongly correlate with both intrapersonal resources such as depression and interpersonal factors such as inequality. Causal mediation methods present an analytical approach for disaggregating the influence of perceived discrimination and resultant psychosocial or economic factors on health behaviors. More comprehensive data is needed to assess psychological factors that modulate antihypertensive adherence.
ACKNOWLEDGMENTS
Author affiliations: Department of Epidemiology and Biostatistics, University of California San Francisco, San Francisco, California, United States (Kendra D. Sims); Department of Epidemiology and Public Health, University College London, London, United Kingdom (G. David Batty); School of Biological and Population Health Sciences, College of Public Health and Human Sciences, Oregon State University, Corvallis, Oregon, United States (Kendra D. Sims, G. David Batty, Ellen Smit, Perry W. Hystad, Michelle C. Odden); Department of Pharmacy Practice, College of Pharmacy, Oregon State University/Oregon Health and Science University, Portland, Oregon, United States (Jessina C. McGregor); and Department of Epidemiology and Population Health, Stanford University School of Medicine, Stanford, California, United States (Michelle C. Odden).
The data set is available from the Health and Retirement Study (https://hrs.isr.umich.edu/).
Presented at the 2021 Society for Epidemiologic Research annual meeting (online), June 22–25, 2021.
M.C.O. serves as a consultant for Cricket Health, Inc. The other authors report no conflicts.
