Risk factor analysis and nomogram for predicting 28-days mortality in ICU patients with sepsis and diabetes

Background: Diabetes is a common comorbidity in septic patients, often leading to immune dysfunction and metabolic disturbances, associated with an increased mortality rate. Therefore, this study aims to identify risk factors in septic patients with concomitant diabetes, establish a useful predictive model to accurately assess the prognosis, and provide treatment recommendations. Methods: Data were sourced from the open-access clinical database, Medical Information Mart for Intensive Care (MIMIC-IV). A total of 4252 cases of septic patients with concomitant diabetes were extracted for the study. Participants were randomly divided into training and validation sets in a 7:3 ratio. A predictive nomogram model was constructed in the training set using COX regression analysis. Specificity and sensitivity analyses were conducted using the area under the receiver operating characteristic curve (AUC). Internal validation and evaluation of the nomogram were performed through integrated discrimination improvement (IDI), calibration curve, and decision curve analysis (DCA). Results: According to the COX regression analysis, factors including Norepinephrine maximum, Temperature, Antibiotic, Heart Rate, SAPS II, Age, BMI, Bilirubin total, Stroke, Troponin, Metastatic solid tumor, and APTT were incorporated into the nomogram. In comparison with the Sequential Organ Failure Assessment (SOFA) and Simplified Acute Physiology Score II (SAPS II), the nomogram demonstrated superior discriminative ability, with AUCs of 0.797 (95% CI: 0.779-0.816) and 0.782 (95% CI: 0.754-0.81) for the training and validation sets, respectively. Decision curve analysis (DCA) of the nomogram showed a net benefit in clinical use compared to the use of SOFA and SAPS II in both groups. Conclusion: The predictive model in this study exhibits excellent discriminative and calibration abilities in predicting mortality among septic patients with concomitant diabetes. Implementing treatment strategies targeting factors identified in the model could improve the prognosis of these patients.