A simple nomogram for predicting in-hospital mortality in patients with sepsis-induced coagulopathy in the ICU: a multicenter retrospective study

Background One typical consequence of sepsis is sepsis-induced coagulopathy (SIC), which is linked to a significantly higher risk of in-hospital death for patients in the intensive care unit (ICU). The specificity, sensitivity, and usability of current clinical prognostic evaluation techniques are restricted, and nomogram-based models are still not widely used to predict outcomes in patients with SIC. In order to increase the accuracy of forecasting in-hospital mortality in ICU patients with SIC, this study developed and validated a straightforward and user-friendly nomogram based on multicenter retrospective data. Methods 480 SIC patients who were admitted to the ICUs of Hebei General Hospital and Handan Central Hospital (East District) between March 1, 2021, and March 1, 2024, were retrospectively recruited and randomly assigned to a training set and a testing set at a 7:3 ratio. An external validation set consisted of an additional 229 SIC patients who were hospitalized to Chongqing General Hospital's intensive care unit between March 1, 2021, and January 1, 2026. Eight kinds of variables were gathered, such as laboratory values, comorbidities, and demographics. Eight kinds of information were gathered, including laboratory measurements, comorbidities, and demographics. Using least absolute shrinkage and selection operator (LASSO) regression, core predictors were first chosen. Multivariable logistic regression analysis was then used to find independent risk factors linked to SIC patients’ in-hospital mortality. Using R's rms package, a nomogram was created based on the final set of chosen predictors. Receiver operating characteristic (ROC) curves, calibration plots, and decision curve analysis (DCA) were used to assess the model's performance in terms of discrimination, calibration, and clinical utility. Results Among the 480 SIC patients, the total in-hospital death rate was 42.7%. Five independent predictors—including coronary artery disease, serum potassium levels, prothrombin time–international normalized ratio (PT-INR), platelet count, and platelet transfusion—were identified as being associated with in-hospital mortality among patients with SIC. With area under the ROC curve (AUC) values of 0.873 [95% Confidence Interval (CI): 0.832–0.914] in the training set, 0.776 (95% CI: 0.696–0.856) in the testing set, and 0.756 (95% CI: 0.692–0.819) in the validation set, the nomogram showed high discriminative capacity. DCA demonstrated that the model offered a high net therapeutic benefit over a broad range of threshold probabilities (0.1–1.0), and calibration plots revealed good agreement between projected and observed probability. Conclusions The nomogram created in this work has strong discrimination, good calibration, and acceptable clinical applicability, making it possible to estimate in-hospital mortality risk in ICU patients with SIC in an understandable and accurate manner. By helping doctors identify high-risk patients early on and provide tailored care plans, this tool may maximize the use of medical resources and possibly lower mortality rates among SIC patients.