External validation of preexisting first trimester preeclampsia prediction models

Area of Science:

• Obstetrics and Gynecology
• Reproductive Medicine
• Clinical Prediction Modeling

Background:

• Preeclampsia is a significant cause of maternal and fetal morbidity worldwide.
• Numerous prognostic models have been developed for early prediction of preeclampsia.
• External validation of these models is crucial for reliable clinical application.

Purpose of the Study:

• To externally validate existing prognostic models for preeclampsia prediction.
• To assess the performance of selected models in a new prospective cohort.
• To determine the generalizability of current preeclampsia prediction models.

Main Methods:

• Systematic literature review to identify first-trimester preeclampsia prediction models.
• Selection of models based on predefined criteria including biomarkers, uterine artery Doppler, and maternal characteristics.
• Validation using regression coefficients applied to a prospective cohort, assessing discrimination (Area Under the Curve) and calibration.

Main Results:

• Twenty preeclampsia prediction models were identified and validated.
• Observed discrimination (AUC) varied significantly when models were applied to the validation cohort (0.504–0.833) compared to derivation studies (0.70–0.96).
• Statistically significant differences were found between derivation and validation AUCs for several models.

Conclusions:

• No current preeclampsia prediction model demonstrates adequate discrimination and calibration across different populations.
• The large number of existing models may limit the value of developing new ones.
• Future research should prioritize validating existing models and evaluating their impact on patient care and outcomes.