Adriëtte Oostvogels

5 127 Family history of diabetes and childhood outcomes for sex and age at the time of measurements, model 2 additionally adjusted for the constitutional factors maternal age, education, ethnicity and standardized birth weight, and model 3 additionally adjusted for the lifestyle factors energy intake, screen time, fat% (only for glucose and C-peptide), maternal BMI and paternal BMI at age 5-6 years. By changing the reference group, we first compared the children with no second-degree FHD to children with a FHD, then we compared children with maternal FHD to children with paternal and both maternal and paternal FHD and lastly we compared children with paternal FHD to children with both maternal and paternal FHD. Boy-girl differences were tested by adding an interaction term between sex and second-degree FHD to the models. Sensitivity analyses were performed on the Dutch children only to exclude residual confounding by ethnic differences in for example lifestyle. Lifestyle differences between the four second-degree FHD groups were much smaller in the Dutch only group, compared to the multi-ethnic group. Moreover, we repeated the analyses with C-peptide as outcome without the imputed values in the complete study population. All statistical analyses were performed using SPSS 22.0 for Windows. A p-value of <0.05 was considered significant. Results The final sample consisted of 3024 children. Study population characteristics are presented in table 1, stratified per second-degree FHD group. Compared to children with no FHD, children with maternal, paternal or both maternal and paternal FHD had higher screen time hours and the groups were more ethnically diverse. Children with maternal FHD and children with both maternal and paternal FHD had, compared to the no FHD group, lower educated mothers with a higher BMI. Children with paternal and both maternal and paternal FHD had lower birth weight and fathers with a higher BMI compared to the no FHD group. The group with both maternal and paternal FHD had younger mothers, and the children had a higher energy intake per day, compared to the no FHD group. Mean values per outcome measurement corrected for age at the time of the measurements are presented in table 2. We compared the characteristics of the children included in this study with non-responders (invited to participate at age 5-6 years, but no questionnaire or health check data available; n=2855). We found that mothers of children included in this study had lower BMI, were older, higher educated and more often of Dutch origin. The children had a higher gestational age at birth and had a higher birth weight (Supplementary Table 1).