Development of a nomogram to predict 30-day mortality of sepsis patients with gastrointestinal bleeding: An analysis of the MIMIC-IV database.

Background: We aimed to establish and validate a prognostic nomogram model for improving the prediction of 30-day mortality of gastrointestinal bleeding (GIB) in critically ill patients with severe sepsis. Methods: In this retrospective study, the current retrospective cohort study extracted data from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database, then partitioned the cohort randomly into training and validation subsets. The cohort was partitioned into training and validation subsets randomly. Our primary endpoint was 30-day all-cause mortality. To reduce data dimensionality and identify predictive variables, the least absolute shrinkage and selection operator (LASSO) regression was employed. A prediction model was constructed by multivariate logistic regression. Model performance was evaluated using the concordance index (C-index), receiver operating characteristic (ROC) curve, and decision curve analysis (DCA). Results: The analysis included 1435 total patients, comprising 1005 in the training cohort and 430 in the validation cohort. We found that age, smoking status, glucose, (BUN), lactate, Sequential Organ Failure Assessment (SOFA) score, mechanical ventilation≥48h (MV), parenteral nutrition (PN), and chronic obstructive pulmonary disease (COPD) independently influenced mortality in sepsis patients with concomitant GIB. The C-indices were 0.746 (0.700-0.792) and 0.716 (0.663-0.769) in the training and validation sets, respectively. Based on the area under the curve (AUC) and DCA, the nomogram exhibited good discrimination for 30-day all-cause mortality in sepsis with GIB. Conclusions: For sepsis patients complicated with GIB, we created a unique nomogram model to predict the 30-day all-cause mortality. This model could be a significant therapeutic tool for clinicians in terms of personalized treatment and prognosis prediction.