Representation of Race and Ethnicity in the Contemporary US Health Cohort All of Us Research Program

Abstract

Importance: To address systemic disparities in biomedical research, the All of Us (AoU) Research Program was created to identify the root causes and consequences of health outcomes in the US. However, the extent of AoU's racial and ethnic diversity is unknown.

Objective: To quantify representation of key racial and ethnic groups in the accruing AoU nationwide health cohort and compare with their actual representation in the US.

Design, setting, and participants: This cohort study compared the AoU program from May 2017 to June 2022 for individuals 18 years and older with the Decennial Survey 2020 (DEC) collected by the US Census Bureau.

Exposures: Representation of non-Hispanic Asian, non-Hispanic Black or African American, Hispanic or Latino, non-Hispanic White, and uncategorized or multiple races in AoU.

Main outcomes and measures: The extent of underrepresentation or overrepresentation of each racial group in the AoU program at both nationwide and state-level relative to DEC.

Results: Of the 358 705 US adults in the AoU to date, individuals identified with the following race and ethnicity categories: 12 710 non-Hispanic Asian (3.5%), 73 348 non-Hispanic Black or African American (20.5%), 58 488 Hispanic or Latino (16.3%), 205 457 non-Hispanic White (57.3%), and 8702 uncategorized or reporting multiple categories (2.4%). Of 355 413 participants with available sex at birth and age data, 218 981 (61.6%) were female and had a mean (SD) age of 53.1 (17.0) years, 136 037 (38.28%) were male and had a mean (SD) age of 56.7 (17.0) years, and 395 reported nonbinary sex (0.1%), with a mean (SD) age of 55.4 (15.8) years. Compared with the referent US, non-Hispanic Black or African American individuals were overrepresented in the AoU by 8.73% (AoU, 20.5% [73 348 of 358 705] vs DEC, 11.7% [30 266 080 of 258 343 281]) and by relative scale, 1.94-fold. Non-Hispanic White individuals accounted for the greatest participation in the AoU with generally consistent dominance across all regions yet numerically underrepresented by absolute difference of -3.54% (95% CI, -3.70 to -3.38). Uncategorized or multiracial group in the AoU (2.4% [8702 of 358 705]) was 0.43-fold likely to be represented relative to the DEC (4.6% [11 922 096 of 258 343 281]) with an absolute difference of -2.19% (95% CI, -2.24 to -2.14). Moreover, non-Hispanic Asian individuals were underrepresented by -2.54% (95% CI, -2.60 to -2.48) prominently in most states. Individuals identifying as Hispanic or Latino were nominally underrepresented by -0.46% (95% CI, -0.58 to -0.34) (AoU, 16.3% [58 488 of 358 705] vs DEC, 16.8% [43 322 792 of 258 343 281]).

Conclusions and relevance: Recruitment trends for the ongoing AoU show relatively improved representation of some major race groups with geographic trends. These findings underscore the need to further tailor and augment recruitment and participation initiatives for diverse populations.

Figure 1.. Comparison of Racial and Ethnic Proportion in the US and All of Us (AoU)

The US data on racial and ethnic distributions are projected by the Decennial Census 2020. Race and ethnicity are self-reported from fixed categories of non-Hispanic Asian, non-Hispanic Black or African American, Hispanic or Latino, non-Hispanic White, and uncategorized or multiple categories in accordance with the US Census Bureau scheme. Absolute difference (95% CI) is calculated using the 2-sample test for equality of proportions. Green denotes overrepresentation in AoU; light gray denotes underrepresentation in AoU. OR indicates odds ratio.

Figure 2.. Mapping of Genetic Ancestry in 1000 Genomes Project and Genetic Similarity Within Self-Reported Racial and Ethnic Category in All of Us

Principal component analysis visualizes variations in geographic and ethnic structure based on a dimensionality reduction method. It projects individuals onto a number of orthogonal axes, each of which is constructed based on linear combinations of allelic scores across single nucleotide variations. The first principal component (dimensional data) shows the maximal possible variance among all possible axes; the second principal component maximizes the remaining variance for all possible axes perpendicular to the first principal component. For the 1000 Genomes Project, in the reference data set of individuals ascertained at country of origin, the principal components of ancestry are color coded by the first 2 genetic principal component coordinates based on provided classifications. For All of Us, the principal components of genetic similarity are illustrated among 96 268 of 358 705 participants with whole-genome sequencing data. Race and ethnicity are self-reported from fixed categories of non-Hispanic Asian, non-Hispanic Black or African American, Hispanic or Latino, non-Hispanic White, and uncategorized or multiple categories in accordance with the US Census Bureau scheme.