Statistical analyses
We estimated perinatal mortality rates, expressed per 1000 births, for chronic hypertensive versus normotensive pregnancies. We examined the association between chronic hypertension and perinatal mortality on both the additive (risk difference, RD) and multiplicative scales (risk ratio, RR), derived from fitting log-linear Poisson regression models with robust variance and the identity-link for RD and log-link for RR (29). Associations with their corresponding 95% confidence intervals (CI) were derived before and after adjusting for confounders. We also examined associations between chronic hypertension and the risks of stillbirth, early neonatal deaths, and late neonatal deaths.