Skip to main content
. Author manuscript; available in PMC: 2022 Jul 1.
Published in final edited form as: J Psychosom Res. 2021 Mar 24;146:110434. doi: 10.1016/j.jpsychores.2021.110434

Table 5.

Analytic Approach and Main Findings

First Author, Year Discrimination Operationalization AL Summary Score Summary Score Distributions, Mean(SD)/% Analytic Approach Confounders/ Covariates Main Findings
Allen, 2019 Modified Version of Experiences of Discrimination Scale. Items (n=8) were scored on a 5-point Likert scale and summed to compute a score that ranged from 8 to 40. Categorized into none (8), low (9–16), moderate (17–24), high (25–32) and very high (33–40) levels of discrimination for analyses, with moderate being the reference group. 15 biomarkers. Four composite measures of AL were calculated. The first and second were a sum of biomarkers that were above a subclinical cut-point or above the 75th (AL75) or 90th percentile (AL90), respectively. The third was a sum score based on established cut-points for subclinical and clinical risk or the IQR (ALIQR). The fourth was a sum of z-scores for each biomarker (ALZ). Discrimination:

  • None=10.6;

  • Low=34.3; Moderate=30.4; High=14.0; Very high=10.6

Allostatic Load:

  • AL75= 6.10 (2.2);

  • AL90=2.32 (1.6);

  • ALIQR=11.45 (3.8);

  • ALZ=NR

 Multivariable regression. Ordinary least squares and ordered logistic models were used depending on the dependent variable. Age, educational attainment, and medication use were included in models regardless of significance. Other variables, including health insurance, poverty status, adjusted household size, marital status and neuroticism were included in models if significance was reached at p<0.10.

  • Low (b=− 1.09, p=0.02, 95% CI=− 1.99, −0.18) and very high (b=− 1.88, p=0.003, 95% CI −3.11, −0.65) racial discrimination were associated with lower levels of AL75 in individuals with a high school diploma or more.

  • Low racial discrimination was associated with high AL75 (b=2.05, p<.01, 95% CI=0.55, 3.56) in individuals with lower education.

  • Associations also varied by education in AL90 and ALIQR

  • Associations varied by poverty status in ALiqR
Cuevas, 2019 Modified versions of the Major Experiences of Discrimination Scale and Everyday Experiences of Discrimination Scale. Both were scales scored on a 4-point Likert scale. Items were averaged to compute score that rangd from 0 to 3. 11 biomarkers. One point was assigned to biomarkers outside of normal values or to account for medication use. Points were summed to compute an AL summary score, resulting in a possible range of 0 to 11. Discrimination:

  • Major= 0.21 (0.42);

  • Everyday= 0.29 (0.49)

Allostatic Load:

  • 5.11(1.76)

 Multivariable linear regression Age, sex, marital status, language acculturation, years living in mainland US, educational attainment, income-to-poverty ration, employment status, work history, current employment status, alcohol consumption, smoking status, physical activity, diet quality, sleep quality and depressive symptoms

  • Major lifetime discrimination was associated with higher AL (b=0.56, p<0.01, 95% CI=0.19, 0.92), adjusting for covariates.

  • Everyday discrimination was associated with lower AL but did not show statistical significance when adjusted for covariates.

  • No significant interactions between discrimination and sex.
Currie, 2019a Adapted one question from the Experiences of Discrimination Scale. Response options: rare/never, some of the time and most of the time. Operationalized as an ordinal measure in analysis. Two additional questions from the scale were used for descriptive purposes. 7 biomarkers. One point was assigned to each biomarker if the biomarker was in a high-risk quartile. Points were summed to compute an AL summary score, resulting in a possible range of 0 to 6. Discrimination:

  • Rarely/Never= 27.6;

  • Some of the time= 57.1; Most of the time= 15.2

Allostatic Load:

  • 2.5 (1.3)

 Bootstrapped linear regression (k=5000) Age, gender, indigenous group, current income group, marital status, number of children, being a single parent, attendance at a residential school and medication use were tested in individual models before entry into main model. Variables that were associated with AL at p<0.20 were retained. These including age and current income.

  • Childhood racial discrimination was associated with increased AL (β=0.23, p<0.01, B=0.47 95% CI=0.11, 0.83).

  • No significant interactions between discrimination and confounders.
Currie 2019b Experiences of Discrimination Scale. Score was derived by counting the number of situations (1 to 9) in which racial discrimination was experienced in the past 12 months. Participants were also asked two questions for descriptive purposes. 7 biomarkers. One point was assigned to each biomarker if the biomarker was in a high-risk quartile. Points were summed to compute an AL summary score, resulting in a possible range of 0 to 6. Discrimination:

  • 2.3 (2.1)

Allostatic Load:

  • 2.5 (1.3)

 Bootstrapped linear regression (k=5000) Age, sex and socioeconomic status

  • Past year racial discrimination was associated with increased AL (β=0.23, p=0.01, B=0.15 95% CI= 0.04, 0.26), adjusting for covariates. This relationship appeared to have linear, dose-response pattern.

  • No significant interactions between discrimination and confounders.
Daly, 2019 Modified version of the Everyday Discrimination Scale. Defined weight-based discrimination as attributing any form of discrimination to weight. Operationalized as binary variable in analyses. 9 biomarkers. AL summary score was calculated by avering the computed z-scores for each biomarker. Summary scores were standardized with a mean of 0 and a standard deviation of 1. Discrimination:

  • 4.6

Allostatic Load:

  • Time 1= −0.01 (0.98);

  • Time 2= 0.01 (1.02)

 Latent-change-score (LCS) approach Baseline physiological dysregulation, age, sex, ethnicity, degree qualification, household weather, medication use, chronic conditions, and depressive symptoms.

  • Perceived weight discrimination predicted an increase in 0.23 standard deviations in AL over the study period (d=0.23, 95% CI=0.07, 0.40, p<0.01), adjusting for covariates.

  • This relationship was not moderated by gender.
Ong, 2017 Everyday unfair treatment was assessed using the Everyday Discrimination Scale. Items (n=9) were scored on a 4-point Likert scale and summed. Lifetime unfair treatment was assessed using the Major Experiences of Discrimination Scale. Summary score calculated by coding responses into categories including none, 1 to 2, and 3 or more instances for analyses. 22 biomarkers. Risk scores (0–1) were created for seven systems depending on the percent of biomarkers in that system that were high-risk. System level risk scores were summed to compute an AL summary score resulting in a possible range of 0 to 7. Discrimination:

  • Every day=14.97 (6.58); Lifetime: never=23.6, One to two instances=26.6, Three or more instances=49.8

Allostatic Load:

  • 1.91 (1.01)

 Multivariate linear regression Age (in years), gender, and educational attainment, medication use, smoking status, presence of alcohol problems, depression symptomatology and global perceived stress

  • Everyday unfair treatment was associated with increased AL (B=0.019, p<0.05, 95% CI=0.001, 0.038) adjusting for covariates, including lifetime unfair treatment.

  • Lifetime unfair treatment was not significantly associated with AL in this model.

  • Relationship did not vary as a function of age, gender or education.
Rosemberg, 2019 Everyday Discrimination Scale. Items (n=9) were scored on a 6-point Likert scale. Items that were experienced a few times a year or more were scored “1”. Dichotomized item scores were summed. For some analyses, discrimination was dichomotized as high (experiencing any item a few times a year or more) verse low. 9 biomarkers. Two composite measures of AL were calculated. First, the number of high-risk biomarkers were summed to create an AL-clinical score. Second, high-risk quartiles were summed to compute an AL-quartile score. Both summary scores had a possible range of 0 to 9. Medication use was adjusted for. Discrimination:

  • 2.02 (2.59); High discrimination= 52.1, Low discrimination= 47.9

Allostatic Load:

  • AL-clinical =2.3 (1.3),

  • AL-quartile=2.1 (1.5)

 Multiple linear regression, Pearson correlation analyses, t-tests Age, education, income, hourly wage, insurance, race, foreign born status, marital status, smoking status and alcohol use

  • Everyday discrimination was associated with increased AL-quartile (β=0.36, p=0.04). Significance did not remain after adjusting for age, foreign born, marital status, education, hourly wage, and insurance.

  • Everyday discrimination was associated with high AL-quartile (r=0.41, p=0.004).

  • Housekeepers with high everyday discrimination had significantly higher AL-quartile compared to those with low everyday discrimination (t=2.12, p=0.040, Cohen’s d=0.61).
Thomas, 2019 Everyday Discrimination Scale. Items (n=10) were scored on a 6-point Likert and summed to compute a score that ranged from 10 to 60. Categorized into none (≤ 20), low (21–30) moderate (31–40), high (41–50) and very high (51–60). Instiutional discrimination was measured using the. Experiences of Discrimination Scale. Items (n=8) were scored on a 5-point Likert scale and summed to compute a score that ranged from 8 to 40. Categorized into none (8), low (9–16), moderate (17–24), high (25–32) and very high (33–40). 15 biomarkers. One point was assigned to high-risk biomarkers. Biomarkers were summed to compute an AL summary score, resulting in a possible range of 0 to 15. Discrimination:

  • Everyday:

  • None=28.50,

  • Low=31.40,

  • Moderate= 18.36,

  • High=12.56, Very

  • high=9.18;

  • Institutional:

  • None=10.63,

  • Low=34.30,

  • Moderate=30.43,

  • High=14.01, Very

  • high=10.63

Allostatic Load:

  • 5.96 (2.24)

 Multiple linear regression Age, education, employment, poverty, marital/partnership status, medication use

  • Individuals that reported very high everyday discrimination had a lower AL than those reporting moderate everyday discrimination (reference) (B =−1.31, p<0.05, 95% CI= −2.41, −0.20) adjusting for covariates.

  • There was no significant association between institutional discrimination and AL.
Upchurch, 2015 Modified version of the Everyday Discrimination Scale. Items (n=10) were scored on a 4-point Likert scale and averaged within each year and then used as a simple latent variables of discrimination. 11 biomarkers. AL summary score was computed by summing the number of biomarkers in a high-risk quartile, resulting in a possible range of 0 to 11. Discrimination:

  • Baseline= 1.76 (0.47)

  • Year 1=1.75 (0.47),

  • Year 2= 1.70 (0.48),

  • Year 3= 1.67 (0.49)

Allostatic Load:

  • 2.57 (2.27)

 Latent growth curve analysis Race, educational attainment, household income, marital status, men (demographic variables included in latent growth curve)

  • Discrimination was predictive of a higher AL intercept (p<0.05).

  • Being African American (p<0.001, standardized total effect = .25, indirect effect = .02) and having lower income (p<0.001), total effect = −.165, indirect effect = −.013) had an indirect effect on AL through discrimination.
Vadiveloo, 2017 Everyday Discrimination Scale. Items (n=9) were scored on a 4-point Likert scale. A continuous measure of discrimination was computed for individuals who reported weight as a primary reason for discrimination. A categorical indicator variable was created for individuals experienced any verse no weight discrimination. 24 biomarkers. AL summary score was computed by summing risk scores across seven systems, resulting in a possible range of 0 to 7. Individuals that reported use of medication were categorized as high risk for that parameter. A binary indicator variable was created for high versus low AL, in which scores ≥3 were considered “high” and scores <3 were considered “low. Discrimination:

  • Baseline= 0.13 (0.76), 10-year follow-up= 0.22 (1.09)

Allostatic Load:

  • Without medication= 1.72 (1.03); With medication=1.94 (1.10)

 Mixed linear regression and Poisson regression Sex, age, race, household income, educational attainment, smoking status, physical activity, baseline body mass index, and discrimination related to age, race and/or sex

  • Individuals who experienced baseline and long-term weight discrimination had double the risk of high AL scores compared to those individuals who did not experience weight discrimination (RR= 2.07, p<0.001, 95% CI= 1.21, 3.55; RR= 2.16, p<0.001, 95% CI= 1.39,3.36).

  • No significant interactions were detected between perceived weight discrimination and sex, BMI, physical activity, smoking status, or age, sex and race discrimination.
Zilioli, 2017 Everyday Discrimination Scale. Items (n=9) were scored on a 4-point Likert scale and summed to compute a score that ranged from 9 to 30. 24 biomarkers. Risk scores (0–1) were created for seven systems depending on the percent of biomarkers in that system that were high-risk. System level risk scores were summed to compute an AL summary score, resulting in a possible range of 0 to 7. Discrimination:

  • 12.51 (4.14)

Allostatic Load:

  • 1.70 (1.02)

 Multiple linear regression Age, gender, presence of chronic conditions in the past 12 months, and psychological covariates (anger in, anger out, trait anger, positive affect and negative affect)

  • Greater perceived discrimination was associated with higher AL (β =0.071, p=0.031, 95% CI= 0.0067, 0.1357) adjusting for covariates.

Note:β indicates a standardized coefficient; AL: Allostatic Load; NR: Not Reported; CI: Confidence Interval