Fig. 3
Interaction between obesity genetic risk and obesity lifestyle risk index on clinically measured BMI in an electronic health record biobank (n =17,040). A Interactions and associations of the obesity PRS with clinically measured BMI stratified by low and high obesity lifestyle risk index (low vs. high obesogenic behaviors). Effect estimates (Beta) are derived from a multivariable linear regression model for BMI adjusted for age, sex, genotyping array, and 5 PCs of ancestry per SD of the polygenic risk score. Pint value is for the interaction term between the PRS and the obesity lifestyle risk index (both continuous) on BMI in the multivariable linear regression model with the PRS and the index added as covariates. B Interaction and associations stratified by lowest and highest morbidity based on the Charlson Comorbidity Index (10-yr survival probability). Effect estimates are derived from a multivariable linear regression model for BMI adjusted for age, sex, genotyping array, and 5 PCs of ancestry per SD of the PRS. Pint value is for the interaction term between the PRS and the obesity lifestyle risk index (both continuous) on BMI in a multivariable linear regression model with the PRS and the index added as covariates. Abbreviations: confidence interval (CI), polygenic risk score (PRS), principal components (PCs), standard deviation (SD)