Establishing a clinical tool to predict acute kidney injury in pulmonary tuberculosis: insights from a large-scale retrospective study.

BACKGROUND

Acute kidney injury (AKI) is a potentially serious complication that may arise during the course of anti-tuberculosis therapy in patients with pulmonary tuberculosis (PTB). Reliable tools for predicting AKI risk in this population remain limited. This study aimed to construct a clinical model for early detection of individuals at increased risk of AKI during treatment.

METHODS

This study retrospectively analyzed 1,045 hospitalized PTB patients treated with standard anti-TB regimens. Multivariable logistic regression was used to identify independent predictors of AKI. A clinical nomogram was developed and validated using discrimination, calibration, and decision curve analysis.

RESULTS

AKI occurred in 16.6% of patients. Eight predictors were identified: age, body mass index (BMI), CA-125, microalbuminuria, hematuria, albumin (ALB), cystatin-C (CYS-C), and estimated glomerular filtration rate (eGFR). The model demonstrated excellent predictive performance (AUC: 0.97 in the training cohort and 0.96 in the validation cohort) and good clinical applicability.

CONCLUSIONS

This study developed and internally validated a practical nomogram for risk prediction of AKI in patients with PTB. The tool, based on routinely available variables, may support early risk stratification and targeted renal monitoring.