Objective: To investigate the development of nephrotoxicity associated with colistin dose, and whether this relationship differs depending on renal function.
Design: A retrospective cohort study of patients who received intravenous colistin to treat infections caused by extensively drug-resistant Gram-negative microorganisms. Adult patients receiving colistin for 72 hours or longer were included in this study. Patients who received renal replacement therapy at baseline or were administered colistin for less than 3 days were excluded. Colistin-induced nephrotoxicity was defined as a doubling of baseline serum creatinine. Colistin dosing was evaluated based on both actual body weight and ideal body weight.
Setting: Single general hospital between 2010 and 2013.
Patients: A total number of 475 patients received colistin therapy. Of these patients, 329 met the inclusion criteria and were included in the analysis.
Measurements and main results: One hundred forty-three patients (43.5%) experienced nephrotoxicity during colistin treatment. The median onset time of nephrotoxicity was 6 days (interquartile range, 4-8 days). The patients with nephrotoxicity were older. Hematocrit and serum albumin levels were lower in patients with nephrotoxicity. Median daily dosing of colistin based on ideal body weight was significantly higher in patients with nephrotoxicity than in those without nephrotoxicity (4.55 vs 4.43 mg/kg/d, respectively; p=0.021). The cumulative dose was not different between patients with and without nephrotoxicity. In multiple logistic regression analysis, daily colistin dosing based on ideal body weight was only significantly associated with the development of nephrotoxicity in patients with an estimated glomerular filtration rate<60 mL/min/1.73 m2 (odds ratio, 2.34; 95% CI, 1.22-4.5). In these affected patients, based on a receiver operating characteristic plot, the optimal predictive cutoff of colistin dose for the development of nephrotoxicity was 2.87 mg/kg/d of colistin, with a sensitivity of 92.3% and a specificity of 76.7%. In patients with estimated glomerular filtration rate≥60 mL/min/1.73 m, age, serum albumin, hematocrit, and use of glycopeptide were associated with the development of nephrotoxicity.
Conclusions: Development of nephrotoxicity was significantly more strongly associated with the dose of colistin, but only in patients with an estimated glomerular filtration rate<60 mL/min/1.73 m2 and not in those with normal renal function.