Validation of the fullPIERS model in a tertiary hospital in southern Brazil

Abstract

Objective

To validate the fullPIERS model’s efficacy in predicting adverse maternal and neonatal outcomes in women with preeclampsia at a tertiary teaching hospital in southern Brazil.

Methods

This retrospective study analyzed data from women diagnosed with preeclampsia at the Hospital de Clínicas de Porto Alegre (HCPA) between January 2017 and August 2024. Patient data extracted from hospital records was used to calculate the fullPIERS score. Adverse maternal outcomes, including HELLP syndrome, placental abruption and pulmonary edema, were assessed within 48 h and 7 days post-admission, as well as adverse neonatal outcomes. Model performance was assessed by calculating the area under the curve (AUC) from ROC analysis, and the optimal threshold for predicting adverse maternal outcomes was determined.

Results

A total of 665 women with preeclampsia were included in the study, of whom 124 (18.6 %) experienced adverse maternal outcomes. The median fullPIERS was higher among women with maternal complications (1.55 % [IQR 0.55–5.9] vs. (0.6 % [IQR 0.3–1.2]; p < 0.001). The fullPIERS model demonstrated acceptable accuracy for predicting adverse maternal outcomes (AUC = 0.721, CI 0.652 – 0.790, p-value < 0.01). A threshold of 2.26 % yielded the best predictive performance, with an accuracy of 70.26 %, sensitivity of 51.1 % and specificity of 89.4 %. FullPIERS also showed a moderate performance in predicting adverse neonatal outcome (AUC = 0.736, 95 % CI 0.686–0.786, p < 0.001) and for combined maternal and neonatal outcomes (AUC = 0.742, 95 % CI 0.689–0.785, p < 0.001).

Conclusion

The fullPIERS model is a reliable tool for predicting adverse maternal and neonatal outcomes; however, its performance appears suboptimal compared to data from high-income countries.