Maternal prepregnancy BMI and size at birth: race/ethnicity-stratified, within-family associations in over 500,000 siblings

Purpose: To estimate racial/ethnic-stratified effects of maternal prepregnancy BMI on size for gestational age at birth, by comparing siblings within families. Methods: This study examined linked vital statistics and patient discharge data from 580,960 infants born to 278,770 women in the State of California (2007–2012). To control for family-level confounding, we used fixed effects multinomial regression, modeling size for gestational age (small [SGA], appropriate, large [LGA]) as a function of maternal BMI (underweight, normal weight, overweight, obesity class I, II, III) and time-varying covariates. We conducted overall and race/ethnicity-stratified (non-Hispanic white, black, Asian; Hispanic) analyses. For comparison, we fit analogous random effects models, which do not control for family-level confounding. Results: In fixed effects models, maternal BMI was most strongly associated with LGA in non-Hispanic white women, reaching 6.7 times greater for class III obesity (OR [95% CI]: 6.7 [5.1, 8.7]); and weakest in black women (OR [95% CI]: 3.0 [1.5, 5.7]). Associations with SGA were similar across race/ethnicity. Compared with random effects estimates, fixed effects were most attenuated for LGA associations among racial/ethnic minority women. Conclusions: Maternal prepregnancy BMI was differentially associated with size for gestational age across racial/ethnic groups, with the strongest family-level confounding in racial/ethnic minority women.