A Propensity Score Analysis Shows that Empirical Treatment with Linezolid Does Not Increase the Thirty-Day Mortality Rate in Patients with Gram-Negative Bacteremia

Abstract

ABSTRACTThe role of linezolid in empirical therapy of suspected bacteremia remains unclear. The aim of this study was to evaluate the influence of empirical use of linezolid or glycopeptides in addition to other antibiotics on the 30-day mortality rates in patients with Gram-negative bacteremia. For this purpose, 1,126 patients with Gram-negative bacteremia in the Hospital Clinic of Barcelona from 2000 to 2012 were included in this study. In order to compare the mortality rates between patients who received linezolid or glycopeptides, the propensity scores on baseline variables were used to balance the treatment groups, and both propensity score matching and propensity-adjusted logistic regression were used to compare the 30-day mortality rates between the groups. The overall 30-day mortality rate was 16.0% during the study period. Sixty-eight patients received empirical treatment with linezolid, and 1,058 received glycopeptides. The propensity score matching included 64 patients in each treatment group. After matching, the mortality rates were 14.1% (9/64) in patients who received glycopeptides and 21.9% (14/64) in those who received linezolid, and a nonsignificant association between empirical linezolid treatment and mortality rate (odds ratio [OR], 1.63; 95% confidence interval [CI], 0.69 to 3.82;P= 0.275, McNemar's test) was found. This association remained nonsignificant when variables that remained unbalanced after matching were included in a conditional logistic regression model. Further, the stratified propensity score analysis did not show any significant relationship between empirical linezolid treatment and the mortality rate after adjustment by propensity score quintiles or other variables potentially associated with mortality. In conclusion, the propensity score analysis showed that empirical treatment with linezolid compared with that with glycopeptides was not associated with 30-day mortality rates in patients with Gram-negative bacteremia.