Development and validation of a nomogram to predict 28-day all-cause mortality in sepsis patients complicated by acute kidney injury

BACKGROUND: Sepsis-induced acute kidney injury (SAKI) carries substantial morbidity and mortality, yet prognostic factors for 28-day all-cause mortality remain inadequately characterized. This study aimed to develop and validate a nomogram for predicting 28-day all-cause mortality in SAKI patients. METHODS: This single-center retrospective cohort study enrolled 857 consecutive SAKI patients from the Affiliated Hospital of Southwest Medical University (December 2018-June 2025), randomly divided into training (n=601, 70%) and internal validation (n=256, 30%) sets. Candidate predictors were initially identified using LASSO regression coupled with ten-fold cross-validation to tune the optimal lambda parameter. Variables possessing non-zero coefficients in the LASSO model were subsequently incorporated into multivariate logistic regression employing backward stepwise elimination to ascertain independent risk factors for 28-day all-cause mortality. A predictive nomogram was then constructed, and model performance was assessed using receiver operating characteristic (ROC), calibration, and decision curve analysis (DCA). RESULTS: LASSO regression identified 11 predictors of 28-day mortality in the training set. Subsequent multivariate logistic regression demonstrated that age, hospital stay (the actual length at the time of SAKI diagnosis), respiratory failure, heart failure, hepatic failure, APTT and mechanical ventilation were independent predictors of 28-day all-cause mortality. The 7-variable nomogram yielded area under the curves (AUCs) of 0.854 and 0.795 in the training and validation cohorts, respectively. Calibration metrics revealed a slope of 1.000 and Brier score of 0.142 in the training set, compared with 0.762 and 0.173 in the validation set. Decision curve analysis (DCA) substantiated the advantageous clinical applicability of the prognostic model.