| Popis: |
Vancomycin-associated acute kidney injury (AKI) remains a major challenge for patients and clinicians. This study aimed to construct a risk scoring system for vancomycin-associated AKI. We retrospectively reviewed medical records of patients who underwent therapeutic drug monitoring for vancomycin from June 2018 to July 2019. We selected possible risk factors for AKI by univariate and multivariable logistic regression analyses and developed a scoring system for vancomycin-associated AKI. Machine learning methods were utilized to predict risk factors for the occurrence of AKI. The incidence of vancomycin-associated AKI was 31.7% among 104 patients included in this study. A bodyweight ≤60 kg (two points), a Charlson comorbidity index ≥3 (two points), a vancomycin trough serum level >15 μg/ml (one point), and concomitant use of ≥6 nephrotoxic agents (two points) were included to construct a risk scoring system based on the coefficient from the logistic regression model. The area under the receiver operating characteristic curve (AUROC) (mean, 95% confidence interval (CI)) across 10 random iterations using five-fold cross-validated multivariate logistic regression, elastic net, random forest, support vector machine (SVM)-linear kernel, and SVM-radial kernel models was 0.735 (0.638–0.833), 0.737 (0.638–0.835), 0.721 (0.610–0.833), 0.739 (0.648–0.829), and 0.733 (0.640–0.826), respectively. For total scores of 0–1, 2–3, 4–5, 6–7, the risk of vancomycin-associated AKI was 5, 25, 45, and 65%, respectively. Our scoring system can be applied to clinical settings in which several nephrotoxic agents are used along with vancomycin therapy. |
