Alcohol intake and hypertensive disorders of pregnancy: a negative control analysis in the ALSPAC cohort

Alcohol intake increases blood pressure, yet estimates of associations between maternal intake and hypertensive disorders of pregnancy (HDP) are sparse and range from null to a protective effect. Here we estimated the association of maternal drinking during pregnancy with preeclampsia and gestational hypertension (separately and jointly, as HDP). We used partner9s alcohol intake as a negative control exposure, beverage type-specific models, and a range of sensitivity analyses to strengthen causal inference and reduce the influence of bias. We used data on self-reported alcohol intake in the UK Avon Longitudinal Study of Parents And Children (ALSPAC) and HDP ascertained from obstetric notes. Multivariable multinomial regression models were adjusted for confounders and mutually adjusted for partner9s or maternal alcohol intake in the negative control analysis. We also performed a beverage type analysis of the effect of beer and wine separately on HDP risk, due to different social patterning associated with different drinks. Sensitivity analyses assessed the robustness of results to assumptions of no recall bias, no residual confounding, and no selection bias. Of the 8,999 women eligible for inclusion, 1,490 developed HDP (17%). Both maternal and partner9s drinking were associated with decreased HDP odds (mutually adjusted odds ratio 0.86, 95% confidence interval 0.77 to 0.96, P-value=0.008 and 0.82, 0.70 to 0.97, P=0.018, respectively). We demonstrate the validity of the negative control analyses using the same approach for smoking as the exposure. This confirmed an inverse association for maternal but not partner9s smoking, as expected. Estimates were more extreme for increasing levels of wine intake compared to increasing levels of beer. Multiple sensitivity analyses did not alter our conclusions. We observed an inverse relationship between alcohol intake during pregnancy and risk of HDP for both maternal and, more surprisingly, partner9s drinking. We speculate that this is more likely to be due to common environmental exposures shared between pregnant women and their partners, rather than a true causal effect. This warrants further investigation using different study designs, including Mendelian randomisation.