Editor’s Choice – The Effectiveness and Safety of High-Dose Colistin: Prospective Cohort Study
Clinical Infectious Diseases December 15, 2016 V.63 N.12 P.1605-1612
Yael Dishon Benattar, Muna Omar, Oren Zusman, Dafna Yahav, Yael Zak-Doron, Sergey Altunin, Michal Elbaz, Vered Daitch, Michal Granot, Leonard Leibovici, and Mical Paul
1Infectious Diseases Institute, Rambam Health Care Campus
2Cheryl Spencer Department of Nursing, University of Haifa
3Sackler Faculty of Medicine, Tel Aviv University
5Department of Medicine E
6Unit of Infectious Diseases, Rabin Medical Center, Beilinson Hospital, Petah Tikva
7Ruth and Bruce Rappaport Faculty of Medicine, Technion–Israel Institute of Technology, Haifa, Israel
Optimizing colistin dosing should translate to improved patient outcomes.
We used data from 2 prospective cohort studies performed between 2006 and 2009 and between 2012 and 2015. In the latter period, a new policy of high-dose colistin (9 million international units [MIU] loading dose followed by 9 MIU daily for normal renal function) was introduced in 2 participating hospitals. We included adult inpatients with invasive infections caused by carbapenem-resistant gram-negative bacteria treated with colistin. Our primary exposure variable was colistin dose, dichotomized to high-dose vs other regimens. The primary outcome was 28-day mortality. We generated a propensity score for high-dose colistin and conducted propensity-adjusted multivariable and matched-cohort analyses for mortality.
Of 529 consecutive patients fulfilling inclusion criteria, 144 were treated with high-dose colistin and 385 with lower-dose colistin regimens. The median daily dose in the high-dose group was 9 MIU (interquartile range [IQR], 9–9) vs 4 MIU (IQR, 3–6) with other regimens. There were 50 of 144 (34.7%) deaths with high-dose colistin vs 165 of 385 (42.9%) with low-dose colistin (P = .1). The propensity-adjusted odds ratio (OR) for mortality was 1.07 (95% confidence interval [CI], .63–1.83) for high-dose colistin. Similar results were obtained when using the study period as the exposure variable, in the subgroup of bacteremic patients (n = 207) and in the propensity-matched cohort (OR, 1.11 [95% CI, .67–1.82]). Nephrotoxicity (RIFLE injury or higher; OR, 2.12 [95% CI, 1.29–3.48]; n = 396) and seizures were significantly more common with high-dose colistin.
In a large cohort, we found no association between high colistin dosing and all-cause mortality. High dosing was associated with more nephrotoxicity.