3 Results
There were 8,147 participants that met inclusion criteria(Fig. 1 ). Participants’ demographic and obstetric characteristics are shown in Table 1 . Maternal GRSBW values ranged from -0.214 to 0.713 and were positively associated with infant BW (p 0.06, meaning that a change in GRSBW of 1.0 (essentially the entire range of possible GRSBW values) is associated with a 6% increase in BW. The measures of association for terms in the initial model are shown in Table 2 .
Among the included nuMoM2b participants, the largest self-identified racial group was White, (n=5394, 64.1%), followed by Black/African American (n= 1139, 14.0%), unknown (n=699, 8.6%), multiracial (n=508, 6.2%), and Asian (n=358, 4.4%). Genetically predicted continental ancestry groups, in order of decreasing size, were EUR (n=5,099, 62.6%), AFR (n=1,383, 17.0%), AMR (n=1028, 12.6%), EAS (n=274, 3.4%), UNK (n=264, 3.2%), and SAS (n=99, 1.2%). Within each self-identified racial group, the most common genetic ancestry was as follows: White: EUR (91.6%); Black/African American: AFR (98.8%); Unknown/not reported: AMR (69.4%); multiracial: AFR (30.1%); Asian: EAS (66.5%); Native Hawaiian/Pacific Islander: EAS (68.8%); American Indian/Alaska Native: AMR (76.5%). Overlap between race and predicted ancestry is reported in Table 3 .
Figure 2 shows the results of our primary analysis our assessing the generalizability of the association between GRSBW and infant BW by self-identified race. The association between GRSBW and infant BW was only significant among participants who self-identified as White (ꞵ=0.036, 95% CI 0.01-0.062) or more than one race (ꞵ=0.096, 95% CI 0.008-0.184).
Across racial groups, the magnitude of the association between GRSBW and infant weight varied widely; the magnitudes of the association in Asian and multi-racial groups (0.09 and 0.1, respectively) were more than double that of White and Black groups (0.04 for both). The variation in the magnitude of the association and statistical association across racial groups is shown in Figure 2.
Figure 3 demonstrates the results for analyses assessing the association between GRSBW and infant weight within genetically predicted ancestry groups rather than self-identified race. GRSBW was associated with BW in the EUR (ꞵ=0.044, 95% CI 0.017-0.07, p=0.007) and AMR (ꞵ=0.073, 95% CI 0.012-0.135, p=0.007) ancestry groups but not AFR, EAS, SAS, or unknown groups (Figure 3 ).
Two final log-linear models each assessed the association between self-identified race or genetically predicted ancestry and BW after controlling for GRSBW in the entire included cohort, using the largest groups in each category (White race, EUR predicted ancestry) as the referent groups. For all groups except American Indian/Alaska Native, self-identified race was independently associated with lower BW after controlling for GRSBW, gestational age, and infant sex (Table 4 ). All genetically predicted ancestry groups except for UNK remained independently associated with lower BW (Table 4 ). Coefficients and confidence intervals for all included terms in each of the models in Table 4 are shown inSupplementary Tables 1 and 2 .