Objectives: With increasing rates of infections caused by MDR Gram-negative organisms, clinicians resort to older agents such as colistimethate sodium (CMS) despite a significant risk of nephrotoxicity. Several risk factors for CMS-associated nephrotoxicity have been reported, but they have yet to be validated. We compared the performance of published mathematical models in predicting the risk of CMS-associated nephrotoxicity.
Methods: In a multicentre, retrospective, cohort study, adult patients (≥18 years of age) were evaluated from five large academic medical centres in the USA. Patients with normal renal function (baseline serum creatinine ≤1.5 mg/dL) who received intravenous CMS for ≥72 h were followed for up to 30 days. The development of nephrotoxicity was as defined by the RIFLE criteria. Each published model was conditioned using patient-specific variables to predict the risk of nephrotoxicity. The predictive performance of the models was evaluated using the observed-to-expected (O/E) ratio. The most significant cut-off threshold for stratifying patients into high and low risk of nephrotoxicity was identified using classification and regression tree analysis.
Results: A total of 106 patients were examined (mean age 53.3 ± 14.9 years, 66% male); the overall observed nephrotoxicity rate was 52.8%. We identified a simple model demonstrating reasonable overall nephrotoxicity risk assessment [O/E ratio of 1.07 (95% CI = 0.81-1.39)] and high sensitivity (92.9%) in predicting nephrotoxicity development in patients on CMS therapy.
Conclusions: We identified a model that could be incorporated into patient management strategies to reduce the risk of nephrotoxicity in patients requiring CMS therapy.
© The Author 2016. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved. For Permissions, please e-mail: [email protected].