Background

Exposure to discrimination may be a lifelong stressor with serious health consequences for racial/ethnic minority groups [1–3]. Discrimination is considered the behavioral manifestation of racism in which racial slurs, unequal treatment, harassment, and other forms of differential treatment are directed toward ethnic minority groups. Further, discrimination may operate on institutional or individual levels in which ethnic minority groups are disadvantaged as based on race [2]. Among African Americans, reports of discrimination have been associated with a wide range of cardiovascular-related health outcomes such as depressive symptoms, psychological distress, hypertension, obesity, C-reactive protein, coronary artery calcification, and intima medial wall thickness, and poorer self-rated health [3–7]. Links between discrimination and self-rated health appear to be particularly strong and enduring among a wide range of ethnic minority groups which include Latinos, African Americans, and Asian Americans [2, 3]. Therefore, self-rated health may be one of the most salient measures to evaluate how discrimination impacts the health and lives of ethnic minorities. However, Asian Americans, in particular South Asians (SA), have been among the least studied in discrimination and health studies. Discrimination and health studies with SA are important given they are the second fastest growing ethnic group in the US. South Asians are also vulnerable to discrimination given they are often publically perceived as a “Model Minority” [8] group that is not susceptible to social, economic, or health-related problems. However, discrimination is a potent social problem SA face as evidenced by decades of documented hate crimes and immigration-related discriminatory experiences [9, 10]. Most recently, SA have been misperceived as threatening or menacing given inaccurate associations with individuals who act on the behalf of terrorist groups [10]. SA also have a high prevalence of cardiovascular disease and diabetes that is not explained by traditional risk factors, and because they have experienced structural and interpersonal discrimination that has been largely under-recognized [10–12]. For these reasons, discrimination and health studies among SA are needed.

Conceptual Framework

A large body of literature demonstrates that stress is related to the development of ASCVD [13–15]. Chronic exposure to discrimination is conceptualized as a stressor for racial/ethnic minority groups which may trigger physiological stress responses, and ultimately lead to poorer cardiovascular health outcomes [16].

Objective/Purpose

The purpose of this study was to explore the cross-sectional association between discrimination and cardiovascular (CV) health among Asian Indian (AI) participants in the Mediators of Atherosclerosis in South Asians Living in America (MASALA) Study. The primary of the ongoing MASALA Study is to determine social, cultural, behavioral, and physiological risk factors for subclinical atherosclerosis among South Asians in the US. Further, MASALA aims to compare Study findings with the racial/ethnic groups represented in the Multi-Ethnic Study of Atherosclerosis. Discrimination was one psychosocial factor measured in MASALA and it is plausible that discrimination-related stress would be associated with several indicators of poorer cardiovascular health (e.g., hypertension, elevated body-mass index, elevated hemoglobin A1C, and high cholesterol) and poorer self-rated health [17–20] among SA studied.

Self-reported health is an important aspect of CV health and may reflect underlying cardiovascular disease that is not detected clinically [18]. In a systematic review and meta-analysis by Mavaddat et al. [18], self-rated health was highly associated with cardiovascular disease mortality and non-fatal cardiovascular disease events. Therefore, we examined associations between discrimination and (1) a physiological-based cardiovascular health risk score as based on blood pressure, body-mass index, hemoglobin A1C, and total cholesterol measurements and (2) overall self-rated health. A positive relationship between discrimination and cardiovascular risk and an inverse relationship between discrimination and self-rated health were hypothesized.

Methods

Sample

The Mediators of Atherosclerosis in South Asians Living in America (MASALA) Study (N = 906) included Bangladeshi, Pakistani, Sri Lankan, Nepali and Asian Indians. Sampling and recruitment methods have been described in detail previously [21]. A random sampling method was applied to several counties surrounding the San Francisco and Chicago areas. Baseline surveys were completed between 2010 and 2013. The MASALA study was approved by institutional review boards at Northwestern University and University of California at San Francisco.

Eligibility criteria for the MASALA baseline study included those who had at least three grandparents originating from South Asia and those who identified as SA. Participants were between the ages of 40 and 84 years of age and were able to speak or read English, Hindi, or Urdu. Exclusion criteria include those free of cardiovascular disease and are detailed in Kanaya et al. [21].

There were two differences between the inclusion criteria for the broader MASALA Study and the current study. For the current study, Asian Indian (AI) MASALA Study participants (n = 757) were isolated for analysis given their experiences and exposure to discrimination may differ from other, unique SA groups. Similarly, only AI born in India were included because there is likely to be variation in discrimination experiences dependent upon foreign born versus US born status [22].

Measures

Discrimination was measured using the widely cited and applied Everyday Discrimination Scale (EDS; [2, 23, 24]). The measurement is comprised of 9 items reflecting general, non-race based experiences of unfair treatment which include receiving poorer service in restaurants/stores or being treated with less courtesy/respect. The EDS was initially scored on a 9–45 point Likert scale with 9 indicating no experiences of discrimination and 45 indicating experiencing various types of discrimination almost every day. The EDS score was then divided into tertiles to evaluate how discrimination experienced at varying levels of intensity may impact cardiovascular health [25]. Low discrimination scores (9–14) were considered those ranging from 9 to 1 standard deviation below mean discrimination scores, medium discrimination scores ranged from 15 to 21 with the upper limit reflecting 1 standard deviation above mean discrimination scores, and high discrimination scores ranged from 22 to 54. Hahm et al. [25] also utilized the tertile approach and established similar cut points in using the EDS to measure discrimination among an Asian American cohort. Although the EDS is often used as a continuous low-to-high discrimination measure, the tertile approach allowed for more nuance in gauging significant differences between low, medium, and high levels of exposure to discrimination.

The majority of discrimination and cardiovascular health studies have focused on individual markers of cardiovascular health (i.e., blood pressure, body-mass index) and have not considered comprehensive cardiovascular health measurements as potential outcomes [26]. A cardiovascular health scale may capture nuances relative to cardiovascular risk that are not detectable through single measures. Therefore, for the purposes of our study, four physiological-based cardiovascular health factors were established using the American Heart Association Simple 7 guidelines [27] to develop the study’s cardiovascular risk scale (CRS). Body-mass index, systolic blood pressure, total cholesterol, and fasting blood glucose levels were coded into poor, intermediate, and ideal categories (see Table 1 for coding scheme [27–30]). All physiological values categorized into the poor range were assigned “2s,” intermediate values were assigned “1s,” and ideal values were assigned “0s.” A CRS was obtained by summing across all four physiological health factor values. The final CRS ranged from 0 to 8, with 0 representing ideal cardiovascular health and 8 representing the poorer cardiovascular health. Additionally, if participants reported using medications for hypertension, diabetes, and high cholesterol and had ideal lab values in these categories their scores were coded as 1s (intermediate category) as recommended in the Simple 7 guidelines.

Table 1 Cardiovascular risk scale components
Full size table

Given established links between self-rated health and cardiovascular indicators, self-rated health is conceptualized as a potentially understudied aspect of cardiovascular health for AI [4]. Self-rated health was measured by asking participants to rate their health on a continuous scale of 1–10, with 1 being poor health and 10 being excellent health. A categorical measure of self-rated health (excellent, very good, good, fair, and poor) was avoided given there is a wide range of variability in how those of ethnic minority status and foreign born perceive these categories of health [31].

Covariates

We included several factors that may be related to cardiovascular health as covariates in the analysis: age in years, sex, education (having earned a bachelor’s degree versus not), working outside the home versus not, study site (Chicago versus San Francisco), total combined family income per household (low as <$39,999/year, moderate as between $40,000–$74,999, and high as >$75,000), being married or living with a partner versus not, depressive symptoms, number of years lived in the US, a traditional cultural beliefs scale, social support, and English language proficiency.

The traditional beliefs scale was a continuous measure asking participants how much they wished SA cultural traditions would be practiced in the U.S. Examples of these cultural traditions centered upon food related activities (fasting, eating traditional SA foods like chapattis and daal) and partaking in arranged marriage practices [32]. The scale had a Cronbach’s alpha coefficient was .81 and ranged from 0 to 28 with lower scores reflecting high cultural beliefs (low acculturation) and higher scores reflecting lower cultural beliefs (higher acculturation; [32]).

Social Support was measured using a continuous scale assessing low social support (0) to high social support (24). Questions evaluated whether participants had someone there when they needed help with various problems [33]. A Cronbach’s alpha of .88 was established for this scale. The Center for Epidemiological Studies Depression (CES-D), 20-item scale was utilized to evaluate depressive symptoms among participants and ranged from 0 to 60 points [34]. The CES-D had a Cronbach’s alpha of .65. The English language proficiency scale evaluated participants’ ability to speak English, read hospital materials with ease, and learn about their medical condition. A continuous measure reflected low English language proficiency to high English language proficiency.

Health behaviors and chronic health conditions were also measured and included as covariates in analyses. These behaviors included smoking status, fruit and vegetable intake per day, and physical activity. Smoking status was treated as a 0–2 continuous scale where 0 = never smoked, 1 = former smoker, and 2 = current smoker. Fruit and vegetable intake was measured by summing the number of fruits and vegetables consumed per week with a list of 34 fruits and vegetables which included “other fruits” and “other vegetables” items. Physical activity was measured on a 0–2 point scale by assigning a point for participants who walked for exercise per day and a point for those who participated in either moderate or heavy effort in conditioning activities. Scores were then reverse coded to reflect 0 as ideal physical activity, 1 as intermediate physical activity, and 2 as a poor level of physical activity. Fruit and vegetable intake and the physical activity scales were treated as continuous covariates. Chronic health conditions included the incidence if having or not having asthma, arthritis, depressive symptoms, cancer, diabetes, hypertension, liver disease, and kidney disease.

Statistical Analyses

The SPSS version 22 software was used for all analyses. Descriptive statistics and normality assumptions were established. Bivariate correlations in the Pearson’s correlation matrix did not exceed the .70 level, signifying a lack of multicollinearity among variables [35]. Low, medium, and high levels of discrimination were dummy coded. Low levels of discrimination category was used as the referent group for all multivariate analyses.

Two separate multiple linear regression analyses were examined to evaluate relationships between (1) self-reported discrimination and the physiologically-based CRS, and (2) self-reported discrimination and self-rated health. Moderation effects for gender were tested in analyses with significant main findings. Four additional multiple logistic regressions were employed to evaluate potential relationships between discrimination and the odds of having any one of the four individual components of the CRS (elevated body-mass index, elevated systolic blood pressure, elevated total cholesterol, and elevated fasting blood glucose levels). The following control measures were included in analyses: age, sex, education, working outside the home versus not, study cite, family income, use of antidepressant medications, years in the US, traditional cultural beliefs, social support, and English language proficiency. The discrimination and CRS analysis was also adjusted for health behaviors related to cardiovascular health. The discrimination and self-rated health analysis was adjusted for chronic conditions.

Results

Slightly more men (n = 408) than women (n = 349) participated in the study. The age range of participants was 40–83 years old and the average age was 56 (SD = 9) years. The majority of participants were highly educated (93.9% with bachelor’s degrees) and all participants were born in India (Table 2). Due to 24 individuals missing family income data, 733 participants were included in the final analyses. The average level of discrimination reported was M = 14.90 (SD = 5.96); 56% of participants scored low (range 9–14), 33 % of participants reported medium levels of discrimination (range 15–21), and 11 % of participants scored high (range 22–54) on the EDS. Participants scored an average of 3.33 (SD = 1.46) on the CRS and an average of 7.66 (SD = 1.35) on the self-rated health scale.

Table 2 Demographics of sample
Full size table

Table 3 reflects that the CRS and self-rated health scales were correlated (r = −.076, p < .05). Discrimination was significantly associated with age (−.084, p < .05), social support (−.35, p < .05), depressive symptoms (.39, p < .001), and self-rated health −.17, p < .001). Table 4 presents the F-values and relevant coefficients for the main discrimination and health (i.e., physiological and self-rated health) analyses. The results from the first regression analysis (i.e., Model 1) show that discrimination was not associated with the CRS, F (18, 688) = 2.317, p = .002, [for participants reporting medium levels of discrimination relative to those reporting low levels of discrimination: B = .18 (SE = .13), p = .14; for those reporting high discrimination relative to those reporting low discrimination: B = −.010 (.19), p = .96]. In an additional analysis with medium levels of discrimination as the referent group, there was no significant relationship between those reporting medium levels of discrimination relative to those reporting high levels of discrimination on the discrimination and CRS pathway (p > .05).

Table 3 Pearson’s correlations between continuous variables
Full size table
Table 4 Regression analyses for relationships between discrimination and health (n = 733)
Full size table

Model 2 presents the results for self-rated health. In the second regression analysis, participants reporting high levels of discrimination had poorer self-rated health relative to those reporting low discrimination (B = −.42 (SE = .17), p = .02). However, for participants reporting medium levels of discrimination self-rated health was not poorer relative to those reporting low levels of discrimination: B = −.11 (SE = .11), p = .31. In an additional analysis with medium levels of discrimination as the referent group, there was no significant relationship between those reporting medium levels of discrimination relative to those reporting high levels of discrimination on the discrimination and self-rated health pathway (p > .05). Sex did not modify the relationship between discrimination and self-rated health (p > .05).

Finally, links between discrimination and the individual components of the CRS were evaluated using logistic regression analyses. When compared to low levels of discrimination, medium levels of discrimination were associated with an increased odds of hypertension as defined by >120 mmHg systolic, >80 mmHg diastolic, or those treated to goal [Chi square = 121.47, p < .001 with df = 18; Exp(B) = 1.61, (SE = .30), p = .02]. Higher levels of discrimination were not associated with increased odds of hypertension relative to low levels of discrimination p > .05). Discrimination was not associated with elevated BMI, fasting blood glucose, or total cholesterol at the p = .05 level.

Discussion

The overall objectives of this study were to examine whether perceived discrimination was associated with cardiovascular and self-rated health among a sample of foreign born AI. A link between discrimination and the CRS was expected given the well-documented history of discrimination AI have endured [10] and the prevalence of pre-established associations between discrimination and CV risk factors among other racial/ethnic minority populations [5, 7, 36, 37]. It was posited that exposure to discrimination would be a chronic stressor for AI and ultimately manifest through poorer cardiovascular health. In contrast to single cardiovascular health measure studies [5, 7], the CRS included several cardiovascular risk factors and was therefore expected to detect more nuance in relating discrimination scores with overall cardiovascular health. One possible explanation for these null findings may be that AI tend to report among the lowest levels of discrimination in comparison to other ethnic minority groups [2, 3, 23]. Low discrimination scores among AI in this study may have lacked the variability necessary to detect relationships between discrimination and the CRS. According to Williams et al. [38], certain levels of discrimination may be related to various alterations in health. Specifically, perhaps higher levels of discrimination exposure trigger physiological stress processes and manifest as cardiovascular risk factors. Conversely, lower levels of discrimination exposure may primarily influence mental health, as measured through self-report. The current study analysis demonstrated that high levels of discrimination (relative to low) were related to poorer self-reported health but medium levels of discrimination (relative to low) were not related to poorer self-reported health. This may suggest that higher exposure to discrimination has worse health effects than lower exposure to discrimination. However, perhaps because discrimination reports were overall low in the current study, discrimination was primary linked with self-rated health.

Importantly, self-rated health was associated with discrimination among AI in this study. This is an important finding given the associations of self-rated health with increased morbidity and mortality [39]. Self-rated health may be thought of as one aspect of health and often provides rich insight into the physiological health of individuals. However, we found no association between discrimination and physiological health as measured by the CRS. The self-rated health measure may have accessed an important aspect of CV health that was not detected through objective, clinical measures in this study [18]. Perhaps additional physiological measures should be tested as outcome variables in further discrimination and AI analyses. In the current study, one significant finding supports continued analyses e.g. for those experiencing medium (versus low) levels of discrimination there appeared to be a link between discrimination and increased odds of hypertension. Although the MASALA Study measures a wide range of physiological markers, perhaps additional biomarkers may be tested in relationship to discrimination. Further studies may measure discrimination in relationship to C-reactive protein or other inflammatory markers associated with stress among AI.

Another explanation for study findings may be that self-rated health captures the impact of discrimination in a way that is not yet manifested through physical health among AI immigrants. Given AIs studied were immigrants, they may not have the same level of lifelong discrimination exposure as AI born in the US. Future studies may evaluate differences between discrimination and physiological health but levels of time lived in the US, or length of time exposure to discrimination.

Strengths

This study has several key strengths. This study identified one South Asian subgroup, AI who shared similar characteristics such as age and immigrant status, for analyses. Although there have been empiric links between discrimination and mental health among SA (e.g., [23, 40]), to our knowledge, this is the only study that has tested an association between discrimination and a continuous, physiological-based measure among AI.

Weaknesses

There are some study limitations. Generalizability of this study is limited to middle-age to older Asian Indians who emigrated from India. The high socioeconomic status of AIs in the MASALA study also does not represent the full spectrum of AI immigrants to the US. Temporal ordering is an issue in that perhaps having a poorer perception of health or hypertension led to increased perceptions of discrimination. The reliability coefficient of .65 for the CES-D scale falls slightly below the .7 α level as recommended by Tabachnick and Fidell [35] and may be less than ideal. However, there are few psychosocial instruments, including the CES-D, specially designed and tailored to the experiences of SA. The development of psychosocial instruments reliable and valid for SA are needed. For example, the EDS may not capture discrimination experiences SA uniquely face (discrimination as based on wearing traditional or ethnic clothing). Longitudinal studies would provide additional support for study findings. Multi-item self-rated health scales would bolster positive findings between discrimination and health among AI.

Contributions to the Literature

Although further studies are needed, significant findings between discrimination and self-rated health and null findings between discrimination and the CRS may suggest that a certain level of discrimination exposure is necessary to be detected through physiological-based outcomes. Future studies may expand upon our findings by using a longitudinal study designs and/or ethically-driven, in-vivo experiments such as those conducted with other ethnic minority groups on this topic [41–43]. Encouraging AI to report discriminatory experiences to appropriate authorities or advocacy organizations may provide necessary data to ameliorate the effect of discrimination among AI. Workplace efforts should focus on implementing anti-discrimination policies. Health promotion efforts are needed given the expanding body of literature confirming how discrimination negatively influences the health of many ethnic minority groups, which now more clearly includes Asian Indians.