Adriëtte Oostvogels

5 131 Family history of diabetes and childhood outcomes Sensitivity analyses When we limited our characteristics to the Dutch only group for the sensitivity analyses (N=2033), characteristics of the four second-degree FHD groups changed slightly. We found that compared to the no FHD group, maternal FHD had higher maternal BMI, lower gestational age at birth and higher screen time hours; paternal FHD had higher paternal BMI and lower birth weight and both maternal and paternal FHD had lower educated mothers and higher paternal BMI (Supplementary Table 2). Results of the sensitivity analyses on the Dutch children only are presented in Supplementary Table 3. No associations were found between second-degree FHD and body composition at age 5-6 years. In the fully corrected model, children with maternal and paternal FHD had increased C-peptide compared to children with no FHD (0.02 nmol/L, 95%CI: 0.01-0.04), only maternal FHD (0.04 nmol/L, 95%CI: 0.01- 0.08) and only paternal FHD (0.04 nmol/L, 95%CI: 0.01-0.08). When analyses for C-peptide were repeated on non-predicted values only (N=508) differences between no or maternal second-degree FHD and maternal and paternal second-degree FHD were attenuated, although differences between paternal second-degree FHD and both maternal and paternal second-degree FHD remained (Supplementary Table 4): C-peptide (0.08 nmol/L, 95%CI: 0.02-0.15). Discussion This study examined outcomes in body composition and glucose metabolism in children at age 5-6 years with respectively no second-degree FHD, only maternal, only paternal or both maternal and paternal second-degree FHD in a multi-ethnic prospective cohort. We found no differences in body composition or glucose metabolism in children with either maternal or paternal second-degree FHD. However, children with both maternal and paternal second-degree FHD had already higher C-peptide levels than their peers with respectively no, maternal or paternal second-degree FHD, independent of their body composition. Strengths and limitations This study was part of a large population-based multi-ethnic cohort, for which reason we had distinctive outcome measurements for both body composition and glucose metabolism and we were able to correct for different covariates, both in utero and at age 5-6 years. The groups with second-degree FHD were more ethnically diverse than the group without second-degree FHD which increased the statistical power of