Discriminative performance of plasma urea, creatinine, and uric acid for severe preeclampsia: a cross-sectional study

To assess and compare the ability of admission plasma urea, creatinine, and uric acid concentrations to discriminate severe from non-severe preeclampsia.

This cross-sectional study included 190 consecutive women with preeclampsia admitted to Can Tho Obstetrics and Gynecology Hospital, Vietnam, from June 2024 to May 2025. Admission blood samples were obtained before treatment, and severity was classified using the 2024 Vietnamese Ministry of Health guideline. Receiver operating characteristic (ROC) analysis estimated areas under the curve (AUCs) with 95% confidence intervals (CIs), Youden index-derived cut-offs, and test characteristics; correlated AUCs were compared using DeLong tests. Logistic regression adjusted for gestational age, maternal age, and first pregnancy.

Eighty-four women (44.2%) had severe preeclampsia. All three biomarkers were higher in the severe group (p < 0.001 each). AUCs were 0.71 (95% CI 0.63-0.78) for urea, 0.69 (0.62-0.77) for creatinine, and 0.73 (0.65-0.80) for uric acid; pairwise AUC differences were not significant (all p ≥ 0.46). Urea ≥4.34 mmol/L had 63.1% sensitivity and 76.4% specificity; creatinine ≥61.6 µmol/L had 53.6% sensitivity and 81.1% specificity; and uric acid ≥473.5 µmol/L had 39.3% sensitivity and 97.1% specificity. All three remained associated with severe preeclampsia after adjustment, and sensitivity analyses for incorporation bias gave similar AUCs.

These routinely available biomarkers showed modest discrimination of severe preeclampsia. Uric acid was highly specific at the selected cut-off but had low sensitivity, and no biomarker had a superior AUC. The data-driven cut-offs are exploratory and require external validation. They should supplement, not replace, clinical severity assessment.