Attributable nephrotoxicity of vancomycin in critically ill patients: a marginal structural model study.

Abstract

Although vancomycin nephrotoxicity is recognizable, critically ill patients have other potential reasons for acute kidney injury (AKI) and determining its attributable nephrotoxic risk in this population can be cumbersome. To determine the risk of AKI attributable to vancomycin, controlling for baseline and time-dependent confounders. Time-fixed and daily time-varying variables were extracted from a large public database. The exposures analysed were: (i) IV vancomycin; (ii) serum trough level greater than 15 and 20 mg/L; and (iii) concomitant exposure to vancomycin and piperacillin/tazobactam or other antipseudomonal β-lactams. Censoring and exposure inverse probability of treatment weighting were calculated. Marginal structural models were plotted to evaluate AKI, severe AKI (stage 2/3) and need of renal replacement therapy (RRT). A total of 26 865 patients were included; 19.7% received vancomycin during ICU stay. After adjusting for fixed and time-variable confounders, vancomycin exposure was associated with AKI (HR = 1.24, 95% CI = 1.09-1.38), but not with severe AKI or need of RRT (HR = 1.05, 95% CI = 0.91-1.23 and HR = 0.97, 95% CI = 0.74-1.29, respectively). A serum trough level greater than 20 mg/L was associated with AKI (HR = 1.90, 95% CI = 1.52-2.30) and severe AKI (HR = 1.69, 95% CI = 1.31-2.19), but showed no statistically significant association with need of RRT (HR = 1.48, 95% CI = 0.92-2.56). The vancomycin + piperacillin/tazobactam combination was not associated with a greater risk than vancomycin alone. The attributable nephrotoxicity of vancomycin in critically ill patients is significantly lower than previously suggested and severe AKI is related to vancomycin only when trough serum levels are greater than 20 mg/L.