[Construction and validation of a risk nomogram for sepsis-associated acute kidney injury in intensive care unit].

OBJECTIVE

To construct and validate a nomogram model for predicting sepsis-associated acute kidney injury (SA-AKI) risk in intensive care unit (ICU) patients.

METHODS

A retrospective cohort study was conducted. Adult sepsis patients admitted to the department of ICU of the 940th Hospital of Joint Logistic Support Force of PLA from January 2017 to December 2022 were enrolled. Demographic characteristics, clinical data within 24 hours after admission to ICU diagnosis, and clinical outcomes were collected. Patients were divided into training set and validation set according to a 7 : 3 ratio. According to the consensus report of the 28th Acute Disease Quality Initiative Working Group (ADQI 28), the data were analyzed with serum creatinine as the parameter and AKI occurrence 7 days after sepsis diagnosis as the outcome. Lasso regression analysis and univariate and multivariate Logistic regression analysis were performed to construct the nomogram prediction model for SA-AKI. The discrimination and accuracy of the model were evaluated by the Hosmer-Lemeshow test, receiver operator characteristic curve (ROC curve), decision curve analysis (DCA), and clinical impact curve (CIC).

RESULTS

A total of 247 sepsis patients were enrolled, 184 patients developed SA-AKI (74.49%). The number of AKI patients in the training and validation sets were 130 (75.58%) and 54 (72.00%), respectively. After Lasso regression analysis and univariate and multivariate Logistic regression analysis, four independent predictive factors related to the occurrence of SA-AKI were selected, namely procalcitonin (PCT), prothrombin activity (PTA), platelet distribution width (PDW), and uric acid (UA) were significantly associated with the onset of SA-AKI, the odds ratio (OR) and 95% confidence interval (95%CI) was 1.03 (1.01-1.05), 0.97 (0.55-0.99), 2.68 (1.21-5.96), 1.01 (1.00-1.01), all P < 0.05, respectively. A nomogram model was constructed using the above four variables. ROC curve analysis showed that the area under the curve (AUC) was 0.869 (95%CI was 0.870-0.930) in the training set and 0.710 (95%CI was 0.588-0.832) in the validation set. The P-values of the Hosmer-Lemeshow test were 0.384 and 0.294, respectively. In the training set, with an optimal cut-off value of 0.760, a sensitivity of 77.5% and specificity of 88.1% were achieved. Both DCA and CIC plots demonstrated the model's good clinical utility.

CONCLUSIONS

A nomogram model based on clinical indicators of sepsis patients admitted to the ICU within 24 hours could be used to predict the risk of SA-AKI, which would be beneficial for early identification and treatment on SA-AKI.