Associations of Self-Reported Race, Social Determinants of Health, and Polygenic Risk With Coronary Heart Disease

Abstract

Background: Social determinants of health (SDOH) influence the risk of common diseases such as coronary heart disease (CHD).

Objectives: This study sought to test the associations of self-reported race/ethnicity, SDOH, and a polygenic risk score (PRS), with CHD in a large and diverse U.S.

Methods: In 67,256 All of Us (AoU) participants with available SDOH and whole-genome sequencing data, we ascertained self-reported race/ethnicity and 22 SDOH measures across 5 SDOH domains, and we calculated a PRS for CHD (PRS_CHD, PGS004696). We developed an SDOH score for CHD (SDOH_CHD). We tested the associations of SDOH and PRS_CHD with CHD in regression models that included clinical risk factors.

Results: SDOH across 5 domains, including food insecurity, income, educational attainment, health literacy, neighborhood disorder, and loneliness, were associated with CHD. SDOH_CHD was highest in self-reported Black and Hispanic people. Self-reporting as Blacks had higher odds of having CHD than Whites but not after adjustment for SDOH_CHD. SDOH_CHD and PRS_CHD were weakly correlated. In the test set (n = 33,628), 1-SD increases in SDOH_CHD and PRS_CHD were associated with CHD in models that adjusted for clinical risk factors (OR: 1.32; 95% CI: 1.23-1.41 and OR: 1.36; 95% CI: 1.28-1.44, respectively). SDOH_CHD and PRS_CHD were associated with incident CHD events (n = 52) over a median follow-up of 214 days (Q1-Q3: 88 days).

Conclusions: Increased odds of CHD in people who self-report as Black are likely due to a higher SDOH burden. SDOH and PRS were independently associated with CHD. Our findings suggest that including both PRS and SDOH in CHD risk models could improve their accuracy.

Keywords: coronary heart disease; polygenic risk scores; social determinants of health.

FIGURE 1. Study Flowchart

The figure describes the study design. CHD = coronary heart disease. PRS = polygenic risk score; SDOH = social determinants of health; WGS = whole-genome sequencing.

FIGURE 2. Associations Between 22 SDOH Measures and CHD Across 5 SDOH Domains

We tested the association between each SDOH measure and CHD separately in various logistic regression models that included the following: 1) age, sex, self-reported race and ethnicity, 10 genetic PCs (basic); 2) basic + PRS_CHD; and 3) basic + PRS_CHD + CHD risk factors. For SDOH measures with multiple levels (eg, annual income), we labeled the reference level “(Ref).” The ORs for CHD are relative to the reference level for these variables. We denote binary variables with a circle and continuous variables with a triangle. For continuous variables, the ORs for CHD correspond to a 1-unit increase in the variables. Abbreviations as in Figure 1.

FIGURE 3. Distribution of SDOH_CHD and Self-Reported Race/Ethnicity in the Test Set

(A) Distribution of (standardized) SDOH_CHD in the test set (n = 33,628), stratified by CHD status. (B) Number of participants in the test set, stratified by self-reported race and ethnicity. (C) Boxplots of (standardized) SDOH_CHD in the test set, stratified by self-reported race and ethnicity. MENA = Middle Eastern and North African; NHPI = Native Hawaiian and Pacific Islander; other abbreviations as in Figure 1.

FIGURE 4. Associations of Self-Reported Race and Ethnicity With CHD

In the test set, we fitted 4 logistic regression models for CHD that included indicator variables for self-reported race (reference: “White”) and various sets of covariates. We show the OR corresponding to the race indicator variable on the x-axis. Different colors indicate different sets of covariates included in the models. The lines denote 95% CIs. We only included self-reported races/ethnicity groups with at least 10 CHD cases. RF = Risk factors; other abbreviations as in Figure 1.