Establishment of a Charlson comorbidity index-based model for predicting the preterm premature rupture of the membrane.

Background: The Charlson comorbidity index (CCI) is a tool used to quantitatively assess the severity of comorbidities in patients. This study aimed to explore the predictive value of the CCI in preterm premature rupture of the membrane (PPROM). Methods: Data were collected from the Medical Information Mart for Intensive Care-IV (MIMIC-IV) database. The continuous variables were analyzed using the T-test/one-way analysis of variance (ANOVA). Comparisons of categorical variables between groups were performed using the chi-square test. Colinearity analysis was used to exclude collinear variables. Receiver operating characteristic (ROC) and decision curve analysis (DCA) were used to assess the prediction efficiency. The prediction model was established by the extreme gradient boosting (XGBoost), logistic, and Random Forest on the independent impact factors, and the areas under the curve (AUC), accuracy, precision, specificity, and F1 were used to assess the prediction efficiency of the model. Results: CCI was significantly higher in the PPROM group than in the term premature rupture of membranes (TPROM) group. The ROC (AUC: 0.535) and DCA indicated that CCI was limited in predicting the PPROM in clinical practice. The results of the logistic analysis showed that the CCI, platelet, age, red blood cell distribution width (RDW), and C-reactive protein (CRP) were the independent influencing factors of PPROM. The ROC results showed that the CCI-based model had a better prediction efficiency in predicting the occurrence of PPROM (AUC: 0.627). Conclusions: CCI was a risk factor for the PPROM. In addition, a risk prediction model based on the CCI and clinical features had a good prediction efficiency for PPROM.