Comparison of four prognostic scales for predicting mortality in patients with severe maternal morbidity

Abstract

ObjectiveTo compare the prognostic validity of the APACHE II-M and O-SOFA scales versus the APACHE II and SOFA to predict mortality in patients with severe maternal morbidity. DesignA retrospective, longitudinal and analytical cohort study was carried out. SettingMedical-surgical intensive care unit (ICU) of a tertiary hospital. PatientsPregnant or puerperal patients of any age admitted to the ICU. InterventionsCalculation of prognostic scores upon admission. Variables of interestAPACHE II, SOFA, APACHE II-M and O-SOFA scores and maternal mortality. ResultsA total of 141 patients were included. The majority (70.2%) were puerperal. The most frequent diagnosis was gestational hypertensive disease (50 cases). The discrimination of each prognostic model was estimated with the area under the ROC curve (AUC-ROC). The calibration was estimated using the mortality ratio and the Hosmer-Lemeshow statistic. The four scales discriminated between survivors and non-survivors with areas under the curve > 0.85. The APACHE II-M model was the predictive model with the highest discrimination and calibration. In the Hosmer-Lemeshow regression analysis, mortality as predicted by the APACHE II and O-SOFA was significantly different from the observed mortality. ConclusionsThe APACHE II-M exhibited the greatest prognostic validity in predicting maternal mortality. This difference was given by its improvement in calibration.