Adriëtte Oostvogels

Study population The ABCD-study is a longitudinal cohort with a multi-ethnic study population, which offers both advantages and disadvantages. First, because of the ethnic diversity, the findings described in this thesis are generalisable to the general population in the Netherlands, which is also multi-ethnic. Second, the ethnic diversity allowed us to study differences in growth patterns of overweight, because children of non-Dutch background are more often overweight. 15 Third, because a family history of diabetes is more common in the non-Dutch population, we had more power to analyse differences in cardiometabolic profile between children with either a maternal or paternal family history of diabetes. On the other hand, due to the multi-ethnic population, it is more challenging to disentangle the effects of maternal pBMI and the effects of lifestyle/genetic factors on the different outcomes. To overcome this issue, sensitivity analyses were performed in the Dutch group only which, however, yielded results similar to those in the complete study population. As with any longitudinal follow-up, the ABCD-study was subject to selection bias from the initiation of the study and also due to attrition over the years. For example, women participating in the ABCD- study were older, more often nulliparous, and more often of Western ethnicity than eligible women who did not participate in the study. 16 Also, compared with the Dutch general population, the prevalence of overweight and obesity was lower in the ABCD-study (overweight 20.4% vs. 17.4%, and obese 9.1% vs. 6.7%, respectively). 17 Attrition over the years resulted in greater numbers of higher-educated Dutch women who continued to participate in the ABCD-study: at study inclusion 52% of the women were Dutch and 39% was higher educated, while of the children from whom a blood sample was collected (at age 5-6 years) 70% were Dutch and 52% had a mother with higher education. This could have consequences for the findings of our study, and for the generalisability of the results to the general population. In the ABCD-study, children of Dutch origin and children of higher-educated women have a healthier metabolic profile than children of non-Dutch origin and children of lower-educated women. 18-20 Moreover, women with a lower socioeconomic status and of non-Dutch ethnicity have a higher prevalence of overweight. 21,22 As both our determinants and outcomes have less extreme values than in the general population, associations between maternal pBMI and early pregnancy lipids, postnatal growth and cardiometabolic profile, might have been underestimated. In other words, the associations found in our study can be generalised, but these associations might be stronger in the Dutch general population. 226 Chapter 9