Back

Prediction of future chronic hypertension from maternal characteristics in early pregnancy

Charakida, M.; Wright, A.; Magee, L. A. A.; Syngelaki, A.; von Dadelszen, P.; Akolekar, R.; Wright, D.; Nicolaides, K. H.

2023-05-01 cardiovascular medicine
10.1101/2023.04.26.23289181 medRxiv
Show abstract

BackgroundPre-eclampsia (PE) and gestational hypertension (GH) identify women at increased risk of chronic hypertension (CH) and cardiovascular disease, but as efforts to prevent PE and GH advance, fewer women at increased cardiovascular risk will be identified. MethodsCohort of 26,511 women seen in two consecutive pregnancies. Included were women without CH, with information on maternal characteristics and blood pressure (BP) at 11-13 weeks gestation, and development of PE or GH in the index pregnancy. Logistic regression models were fitted for prediction of development of future CH by the 20th week of the subsequent pregnancy. Performance of screening and risk calibration of the model were assessed. Results1560 (5.9%) women developed PE or GH (index pregnancy), and 215 (0.8%) developed future CH, a median of 3.0 years later. Predictors from the index pregnancy of development of future CH were: early pregnancy maternal age, weight and BP; Black or South Asian ethnicity; family history of PE; parity; and development of PE or GH. PE or GH accounted for 52.1% (95% confidence interval 45.2-58.9%) of future CH. At a screen-positive-rate of 10%, a model including terms for maternal characteristics and early pregnancy BP accounted for 67.9% (61.2-74.5) of future CH; addition of the development of PE or GH detected 73.5% (67.1-79.3) of future CH. Risks produced from the predictive model were well-calibrated and confirmed by five-fold cross-validation. ConclusionEarly maternal characteristics and BP are effective in predicting development of future CH. As new interventions are expected to reduce the occurrence of PE and GH, our study results offer an alternative strategy for identifying women at increased risk of future CH and are applicable worldwide.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.