Efficacy and validation of a clinical model to predict acute kidney injury in severe pneumonia requiring mechanical ventilation in elderly patients: a multicenter retrospective observational analysis.

Background: The objective of our retrospective multicenter analysis was to identify risk factors and construct a statistical model for predicting acute kidney injury (AKI) among elderly patients with severe pneumonia requiring mechanical ventilation (SPRMV) in elderly patients in different intensive care units (ICU). Methods: We aimed to utilize a multi-center retrospective analysis, including 353 cases of SPRMV patients diagnosed and treated in the ICU of the Hefei Second People's Hospital between May 2018 and February 2025 as a training dataset, and 151 participants were admitted to the ICU of the First Affiliated Hospital of Anhui Medical University between June 2020 and March 2025, considered as a validation dataset. Both univariate and multivariate logistic regression analyses were utilized to investigate the risk factors of SPRMV with AKI. After that, our predictive model was evaluated using a nomogram, a receiver operating characteristic (ROC) curve for discrimination, calibration curves, and decision curve analysis (DCA) curves for clinical validity. Results: Our multivariate logistic regression analysis indicated that CREA, SOFA, APACHE II, driving pressure, mechanical kinetic energy, CRP/ALB, and MAP are independent risk factors of SPRMV in the elderly patients with AKI. A nomogram of SPRMV in elderly patients with AKI was constructed. The ROC curve revealed that our predictive model showed good predictive efficacy with an area under curve (AUC) of 0.920 (95% confidence intervals (CI) = 0.892-0.948) with a specificity of 0.993 and sensitivity of 0.763 in the training dataset and an AUC value of 0.938 (95%CI = 0.899-0.977) with a specificity of 0.952 and sensitivity of 0.854 in the validation dataset. Moreover, calibration and DCA curves demonstrated that our predictive model had a good fit, better net benefit, and higher predictive efficiency for SPRMV in the elderly patients with AKI. Conclusion: Our predictive model demonstrated that CREA, SOFA, APACHE II, driving pressure, mechanical kinetic energy, CRP/ALB, and MAP were the independent risk factors of AKI in SPRMV in the elderly patients with high accuracy and good calibration.