Adriëtte Oostvogels

52 Chapter 2 Moreover, we were able to filter out the BP measurements taken after medication was started for high BP. This resulted in a more reliable estimate of the BP course based on maternal pBMI and lipids, and not influenced by medical interventions. On the other hand, as women with gestational diabetes have higher BPs during pregnancy, 24 and women who used hypertensive medication in our study were more often overweight, excluding them most likely resulted in an underestimation of the effects of maternal weight status on BP course during pregnancy. Most other studies also excluded women with pre-existent/gestational hypertension and/or diabetes. 2,5,6 Furthermore, results of a sensitivity analysis in a subsample without complications yielded similar results as in the whole study sample. However, in a sensitivity analysis on a group with complications no associations with maternal lipids were found. This could be the result of other unknown underlying mechanisms causing hypertensive disorders and/ or preterm birth. Because the present study had a large study population, we were able to correct for a wide range of covariates. Nevertheless, we did not measure some important covariates such as, for instance, nutritional status/dietary patterns and gestational weight gain; 5 thus, potentially, residual confounding might overestimate our results. Our non-response analysis also showed that the women included in this study were slightly healthier compared to the sample that lacked BP and/or lipid measurements; this could also have led to an underestimation of the true effect. Moreover, also the use of self-reported data could have led to an underestimation of the true effect as women tend to overestimate their height and underestimate their weight. 25 This could have resulted in an underestimation of BMI and misclassification of women with overweight in the normal weight group, diminishing the differences found between the normal weight and overweight groups. Finally, maternal lipids were only measured once, non-fasting in early pregnancy, and randomly during the day. Since maternal lipids fluctuate during pregnancy, more extreme increases in the second half of pregnancy could, for instance, be more directly related to BP increases during this second half of pregnancy. However, other studies found that increased maternal lipids in early pregnancy were already associated with an increased risk of developing hypertensive disorders later in pregnancy. 14, 16 Therefore, we expect that it is possible to relate early pregnancy lipids to BP course during pregnancy; however, our evidence would have been strengthened by multiple lipid measurements in pregnancy. Also, the non-fasting state could have diluted our results for TG and TC as normal food intake is known to alter TG and TC, but not ApoA1 and ApoB. 26 Furthermore, it has been shown that non-fasting levels of TG in women predicted cardiovascular events more strongly than fasting levels. 27 Therefore, using non-fasting lipids could be more predictive of BP course than fasting lipids.