Construction of a clinical prediction model for pre-eclampsia based on early pregnancy prenatal examination indicators

Background  Preeclampsia is a pregnancy-specific disease. Early prediction and intervention can avoid the occurrence and development of this disease and improve pregnancy outcomes.

Objective To explore factors associated with preeclampsia in early pregnancy and construct a clinical prediction model for the risk of preeclampsia in early pregnancy.

Methods  Clinical data about 1643 pregnant women who received perinatal care and hospitalized in the First Medical Center of Chinese PLA General Hospital from January 2021 to December 2023 were collected. They were divided into modeling group and validation group according to a ratio of 7:3. The factors independently associated with preeclampsia were screened out by the least absolute shrinkage selection operator (LASSO) regression analysis method, and a prediction model was constructed. The discrimination, calibration and clinical practicality of the constructed clinical prediction model were evaluated.

Results  A total of 1 643 pregnant women who had 1 658 deliveries, with an average age of (31.8 ± 3.9) years, were selected. Among them, the early-onset preeclampsia group had 26 pregnancies, and the late-onset preeclampsia group had 36 pregnancies, resulting in a preeclampsia prevalence rate of 3.95%. The modeling group included 1 160 deliveries, and the validation group included 498 deliveries. The results of LASSO regression analysis showed that BMI＞25 kg/m², previous history of gestational hypertension/PE/eclampsia, fibrinogen level and serum uric acid level were independently associated with preeclampsia in early pregnancy. A prediction model for preeclampsia was constructed based on this method. The AUCs of the modeling group and the validation group were 0.920 (95% CI: 0.880-0.971) and 0.884 (95% CI: 0.819-0.949), respectively. The model has certain clinical application value.

Conclusion BMI＞25 kg/m², previous history of gestational hypertension/PE/eclampsia, fibrinogen and serum uric acid are independently associated with preeclampsia. The prediction model for preeclampsia constructed based on the LASSO regression method has good predictive performance.