550. Meta-analysis (MA) of Mortality in Febrile Neutropenia (FN) Trials Associated with Cefepime Use
Session: Poster Session: Hospital-acquired and Transplant Infections
Friday, October 30, 2009: 12:00 AM
Room: Poster Hall A
Background: Two recent MAs showed a statistically significant higher mortality in patients (pts) treated with cefepime (CFP) vs comparator (COM). (Paul JAC 2006: FN alone, Yahav Lancet Infect Dis 2007: FN and other clinical conditions). In the Yahav MA higher mortality was also seen in the FN subset (19 publications). To understand this issue further, FDA conducted additional analyses to: assess if FN pts treated with CFP had higher mortality, further characterize FN pts regarding the cause of death, and identify risk factors for mortality using data from comparative FN trials where patient-level (PL) data were available.
Methods: MAs were performed using the Mantel-Haenzsel risk difference method. Trial-level (TL) analyses were based on summary data from all pts in the trials and PL analyses were based on detailed data from each pt. Additional analyses were performed using Multivariate Bayesian Logistic Regression (MBLR). Case report forms (CRFs) for all deaths in the FN trials were reviewed.
Results: TL MAs based on 88 trials (24 FN) did not show a statistically significant increase in mortality overall or for FN. In the 24 FN trials, the mortality rates were 6.5% (181/2791) for CFP and 5.6% (148/2658) for COM, Adjusted Risk Difference (ARD)/1000 = 9.67 (95% CI: -2.87, 22.21). PL MAs based on 35 trials (7 FN) also showed no statistically significant increase in mortality overall and in FN. In the 7 FN trials with PL data, mortality rates were 7.9% (61/776) for CFP and 6.5% (41/626) for COM, ARD/1000 = 18.1 (95% CI: -9.22, 45.42). COM drugs included β-lactams +/- aminoglycosides. Most pts appeared to die from underlying malignancies and/or co-morbid conditions. Analyses using the MBLR method also did not show increased mortality with CFP; Empirical Bayesian Odds Ratio for mortality CFP vs. COM=1.36 (95% CI: 0.28, 3.6).
Conclusion: Overall, there was no increase in mortality in FN pts treated with CFP vs COM using either TL or PL data. No covariate was identified that was specifically associated with increased mortality among CFP pts and no biologically plausible reasons for mortality related to safety or lack of drug efficacy were identified.
P Kim, MD MS1, Sumathi Nambiar, MD MPH, A Szarfman, MD PhD1, Y Wu, PhD MPH1 and  P. Kim, None..
A. Szarfman, None..
Y. Wu, None..
S. Nambiar, None., (1)FDA, Silver Spring, MD