Figure 5.

Estimation of the coefficient for BMI with glucose as the outcome by missing data method and weighting approach among non-Hispanic Whites (n=42,999; panels A and B) and non-Hispanic Blacks (n=2,297; panels C and D) in all MGI adults age 40 or older without diabetes. The PRS-imputed subset sample (n=30,942 for non-Hispanic Whites; n=1,437 for non-Hispanic Blacks) was restricted to individuals with non-missing genotype data before multiple imputation. Analyses were adjusted for age, sex, and smoking status (ever/never). Gray shaded regions represent corresponding 95% confidence interval from National Health Interview Survey-weighted All of Us data where weights are calculated separately for non-Hispanic Whites and non-Hispanic Blacks to make All of Us data for each of these groups more representative of their corresponding US population (target population). Results for the full, unstratified cohort are shown in Supplementary Figure 5. Abbreviations: PRS, polygenic risk score.