Objectives: This systematic review and meta-analysis compared effects of different antibiotics on mortality in patients with bloodstream infections caused by Enterobacteriaceae with chromosomal AmpC β-lactamase.
Methods: Databases were systematically searched for studies reporting mortality in patients with bloodstream infections caused by AmpC producers treated with carbapenems, broad-spectrum β-lactam/β-lactamase inhibitors (BLBLIs), quinolones or cefepime. Pooled ORs for mortality were calculated for cases that received monotherapy with these agents versus carbapenems.
Registration: PROSPERO international prospective register of systematic reviews (CRD42014014992; 18 November 2014).
Results: Eleven observational studies were included. Random-effects meta-analysis was performed on studies reporting empirical and definitive monotherapy. In unadjusted analyses, no significant difference in mortality was found between BLBLIs versus carbapenems used for definitive therapy (OR 0.87, 95% CI 0.32-2.36) or empirical therapy (OR 0.48; 95% CI 0.14-1.60) or cefepime versus carbapenems as definitive therapy (OR 0.61; 95% CI 0.27-1.38) or empirical therapy (0.60; 95% CI 0.17-2.20). Use of a fluoroquinolone as definitive therapy was associated with a lower risk of mortality compared with carbapenems (OR 0.39; 95% CI 0.19-0.78). Three studies with patient-level data were used to adjust for potential confounders. The non-significant trends favouring non-carbapenem options in these studies were diminished after adjustment for age, sex and illness severity scores, suggestive of residual confounding.
Conclusions: Despite limitations of available data, there was no strong evidence to suggest that BLBLIs, quinolones or cefepime were inferior to carbapenems. The reduced risk of mortality observed with quinolone use may reflect less serious illness in patients, rather than superiority over carbapenems.
© The Author 2015. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com.