Abstract

Objectives: Early, empirical broad-spectrum antibiotic treatment is the established practice for febrile neutropenia. Several β-lactams are accepted for monotherapy. We asked whether patients' outcomes are influenced by the chosen β-lactam.

Methods: Systematic review and meta-analysis of randomized controlled trials comparing anti-pseudomonal β-lactams administered as empirical monotherapy for febrile neutropenia, with or without vancomycin. The search included The Cochrane Library, PubMed, Embase, Lilacs databases, bibliography, conference proceedings, trial registries and FDA new drug approvals. Two reviewers independently applied selection criteria, performed quality assessment and extracted the data. Trials assessing the same β-lactam were pooled using the fixed effect model. Relative risks (RRs) with 95% confidence intervals (CIs) were calculated. The primary outcome assessed was all-cause mortality.

Results: Thirty-three trials fulfilled inclusion criteria. Cefepime was associated with higher all-cause mortality at 30 days than other β-lactams (RR 1.44, 95% CI 1.06–1.94, 3123 participants). Carbapenems were associated with fewer treatment modifications, including addition of glycopeptides, than ceftazidime or other comparators. Adverse events were significantly more frequent with carbapenems, specifically pseudomembranous colitis (RR 1.94, 95% CI 1.24–3.04, 2025 participants). All-cause mortality was unaltered. Piperacillin/tazobactam was compared only with cefepime and carbapenems, in six trials. No significant differences were demonstrated with paucity of data for all-cause mortality.

Conclusions: The use of cefepime for febrile neutropenia is associated with increased mortality and should be carefully considered pending further analysis. Empirical use of carbapenems entails fewer treatment modifications, but an increased rate of pseudomembranous colitis. Ceftazidime, piperacillin/tazobactam, imipenem/cilastatin and meropenem appear to be suitable agents for monotherapy.

Background

Febrile neutropenia is a common complication of cancer treatment associated with significant morbidity and mortality. Early institution of broad-spectrum antibiotic treatment reduces mortality.1

Monotherapy using a broad-spectrum β-lactam is as effective and safer than the classical combination of a β-lactam and an aminoglycoside.2 Current guidelines propose ‘cefepime or ceftazidime or imipenem or meropenem’, empirically, as single β-lactams, with or without vancomycin.3 The potential of piperacillin/tazobactam is pointed at, stipulating further experience.

Common to all monotherapies is an in vitro spectrum of coverage including Pseudomonas aeruginosa. Differences exist with regard to the spectrum of coverage against Gram-positive bacteria and multidrug-resistant Gram-negative bacteria. Ceftazidime, with the longest experience in clinical trials, provides broad-spectrum coverage against Gram-negative bacteria. Comparatively, cefepime, piperacillin/tazobactam and carbapenems offer better coverage in vitro against methicillin-susceptible Staphylococcus aureus, Streptococcus viridans and Streptococcus pneumoniae.4 Cefepime and piperacillin/tazobactam provide enhanced activity against certain extended spectrum β-lactamases (ESBLs),5 whereas carbapenems are the treatment of choice for ESBL-producing Gram-negative bacteria.6

Antibiotic-related adverse events may also differ, affecting morbidity, cancer treatment schedule and ultimately patients' prognosis. Significant adverse events include pseudomembranous colitis, colonization with resistant bacteria or fungi and resultant superinfections.

We speculated whether these differences result in different clinical outcomes. We performed a systematic review with meta-analysis of randomized controlled trials that compared these antibiotic monotherapies for febrile neutropenia.

Methods

Our objective was to assess whether there is an advantage to one of the β-lactam monotherapies recommended for the empirical treatment of febrile neutropenia.3 We included randomized controlled trials comparing ceftazidime, cefepime, carbapenems or piperacillin/tazobactam with any different β-lactam for the empirical treatment of febrile neutropenia. We restricted the antibiotics assessed to anti-pseudomonal β-lactams. We regarded ceftazidime, the first established monotherapy for febrile neutropenia, as reference and compared other monotherapies with it when relevant. Only the addition of a glycopeptide to both study arms was permitted, since monotherapy is recommended in current guidelines with or without a glycopeptide.3 The primary outcome assessed was all-cause mortality, 30 days following end of treatment. When unavailable, mortality at end of study follow-up was used. Secondary outcomes included treatment failure, defined as non-resolved infection, treatment modification or death owing to infection; infection-related mortality; treatment modifications assessing specifically the need to add glycopeptides and antifungals; microbiological failure, defined as failure to eradicate causative pathogen/s; superinfection; and adverse events. We planned to compare the duration of hospital stay, development of resistance among pre-treatment isolates and the rate of colonization with resistant bacteria. However, these outcomes were not assessed in all included studies.

We searched The Cochrane Library, Medline, Embase and Lilacs databases using specific antibiotic names combined with the terms (neutrop?en* OR granulocytop?en*) and a filter for randomized trials.7 Unpublished trials were sought in references of all included studies; relevant conference proceedings; trial registries and ongoing trial databases; new drug application (NDA) documents of the US Food and Drug Administration; and through personal contact with the authors and sponsoring pharmaceutical companies of included studies. No language or date restrictions were imposed. The last search was performed in February 2005.

Two reviewers independently performed the search, applied inclusion criteria and extracted the data. Whenever mortality data or randomization methods were not reported in the primary reference we contacted the authors and the sponsor requesting these data. Quality assessment was performed using the individual component approach assessing allocation sequence generation, allocation concealment, blinding, intention to treat analysis and number of patients excluded from outcome assessment. Allocation concealment and generation were graded as adequate (A), unclear (B) or inadequate (C) using criteria suggested in the Cochrane handbook. To assess the effect of study quality on outcomes we performed sensitivity analyses by individual components. We regarded primarily the quality of allocation concealment that has been shown to correlate with bias.8,9

Relative risks (RRs) and 95% confidence intervals (CI) were calculated for individual studies. Studies comparing the same β-lactam were combined using the fixed effect model to obtain the combined RR of that β-lactam versus comparators. Within each comparison studies were subgrouped by the comparator. Heterogeneity in the results of the trials was assessed using a χ2 test of heterogeneity and the I2 measure of inconsistency quantified according to Higgins et al.10 Analyses were conducted using RevMan 4.2 software. A funnel plot and Egger's test of the intercept were used to assess small study effect (publication bias or other).11

The full search strategy and detailed methodology are available in The Cochrane Library.12

Results

The trial flow is depicted in Figure 1. Thirty-three trials fulfilled inclusion criteria (Table 1).1347 The monotherapies assessed were ceftazidime (19 trials); cefepime (18 trials); imipenem (12 trials); meropenem (8 trials); and piperacillin/tazobactam (7 trials). Single trials compared piperacillin13 or ticarcillin/clavulanate18 with ceftazidime and three trials compared cefoperazone/sulbactam with imipenem.16,19,47 Vancomycin was added in both study arms to all patients in four trials,13,14,18,19 and as needed in two trials.33,40 Most studies were conducted during the last decade. The primary outcome assessed in all studies was treatment failure. Data regarding all-cause mortality were reported in the primary reference in 20 studies and obtained from the authors in eight additional studies. The adjusted mean all-cause mortality rate at end of study follow-up was 3.9% for adults (23 studies, 4938 patients) and 1.2% for children (4 studies, 605 patients). Seven trials are unpublished—three were found in the NDA application for cefepime14,33,41 and four in conference proceedings.2224,28,39,43

Figure 1.

Study flow. RCT, randomized controlled trial. References refer to the list of references to excluded studies provided as Supplementary data at http://jac.oxfordjournals.org

Figure 1.

Study flow. RCT, randomized controlled trial. References refer to the list of references to excluded studies provided as Supplementary data at http://jac.oxfordjournals.org

Table 1.

Characteristics of included studies

Study IDa
 
Years
 
Locationb
 
Participants
 
Interventions
 
Allocation generationc
 
Allocation concealmentc
 
Blindingc
 
Anaissie (1988)13 1985 USA 339 episodes; age:d median 42, range: 16–85 piperacillin 4 g six times daily + vancomycin 1 g twice daily versus ceftazidime 1 g six times daily + vancomycin 1 g twice daily (an additional arm not included in this review assessed all three antibiotics combined) outcome assessor 
Aoun (1997)14 1993–94 MC Belgium 111 patients (128 episodes); age: mean 48.8 ± 14.9 cefepime 2 g three times daily + vancomycin 30 mg/kg/day versus ceftazidime 2 g three times daily + vancomycin 30 mg/kg/day open 
Aparicio (1996)15 1991–94 Spain 102 patients (118 episodes); age: median 58, range: 15–75 ceftazidime 2 g three times daily versus imipenem 500 mg four times daily open 
Bickers (1990)16  USA 135 patients; age: mean 41 cefoperazone/sulbactam 2 g/1 g twice daily versus imipenem 500 mg four times daily ND ND ND 
Biron (1998)17 1995–96 MC France 380 patients (400 episodes); age: no data cefepime 2 g twice daily versus imipenem 1 g three times daily outcome assessor 
Bodey (1990)18 1985–86 USA 352 episodes; age: mean 46, range: 15–82 ticarcillin/clavulanate 4 g/0.1 g six times daily + vancomycin 1 g twice daily versus ceftazidime 1 g six times daily + vancomycin 1 g twice daily (an additional arm not included in this review assessed all three antibiotics combined) ND outcome assessor 
Bodey (1996)19 1990–93 USA 407 patients (457 episodes); age: median 51, range: 17–84 cefoperazone/sulbactam 2 g/1 g three times daily + vancomycin 1 g twice daily versus imipenem 500 mg/m2 four times daily + vancomycin 1 g twice daily outcome assessor 
Bohme (1998)20 1996 Germany 88 patients (102 episodes); age: median 45, range: 18–68 piperacillin/tazobactam 4.5 g three times daily versus cefepime 2 g three times daily ND ND open 
Boogaerts (1995)21 1991–94 MC Europe 248 patients; age: mean 42, range: 14–78 meropenem 1 g three times daily versus ceftazidime 2 g three times daily ND 
Bow (2003)2224 no data MC USA, Canada, Australia 528 patients; age not given piperacillin/tazobactam 4.5 g four times daily versus cefepime 2 g three times daily open 
Chandrasekar (2000)25 1993–94 MC USA 276 patients (315 episodes); age: median 56, range: 18–82 cefepime 2 g three times daily versus ceftazidime 2 g three times daily triple-blind 
Cherif (2004)26 1996–98 MC Sweden 180 patients (207 episodes); age: median 55, range: 16–83 cefepime 2 g three times daily versus imipenem 500 mg four times daily ND open 
Chuang (2002)27 2000–01 Taiwan 95 patients (120 episodes); age: mean 5.5, range: 0.3–15.5 cefepime 25 mg/kg twice daily versus ceftazidime 25 mg/kg twice daily ND ND open 
Cornely (2001)28 no data MC Germany 602 patients assessed; age: adults meropenem 1 g three times daily versus cefepime 2 g three times daily (an additional arm not included in this review assessed piperacillin/tazobactam + aminoglycoside) ND ND ND 
Feld (2000)29 1991–93 MC (USA, Canada, The Netherlands) 411 patients (471 episodes); age: mean 48.5; range: 17–85 meropenem 1 g three times daily versus ceftazidime 2 g three times daily ND ND double- blind 
Figuera (2001)30 1994–96 Spain 150 patients; age: mean 38.3; range: 16–73 imipenem 500 mg four times daily versus piperacillin/tazobactam 4.5 g four times daily ND ND open 
Fleischhack (2001)31 1998–2000 MC Germany 169 patients (375 episodes); age: median 7.4, range: 0.3–32.6 meropenem 60 mg/kg three times daily versus ceftazidime 100 mg/kg three times daily open 
Freifeld (1995)32 1986–90 MC USA 501 episodes; age: mean 32, range: 4–80 ceftazidime 30 mg/kg three times daily versus imipenem 15 mg/kg (or 12.5 mg/kg) four times daily ND ND open 
Glauser (1997)33 1993–95 MC Europe 281 patients (324 episodes); age: median 52, mean 50.1 ± 16.6 cefepime 2 g three times daily versus ceftazidime 2 g times daily [additional glycopeptides administered to <1% of patients (in both arms)] ND ND outcome assessor 
Kebudi (2001)34 1998 Turkey 33 patients (63 episodes); age: median 7, range: 1 month–14 years ceftazidime 33 mg/kg three times daily versus cefepime 50 mg/kg three times daily open 
Kutluk (2004)35 no data Turkey 30 patients (63 episodes); age: mean 7.4 ± 3.74 meropenem 20 mg/kg three times daily versus cefepime 50 mg/kg three times daily single 
Liang (1990)36 1988–89 Hong Kong 92 patients (103 episodes); age: mean 38, range: 16–76 ceftazidime 1 g three times daily versus imipenem 500 mg four times daily ND open 
Lindblad (1998)37 no data MC Sweden 192 patients; age: mean 60, range: 17–85 meropenem 1 g three times daily versus ceftazidime 2 g three times daily ND ND 
Mustafa (2001)38 no data USA 104 patients (149 episodes); age: median 6, range: 5 months–19 years cefepime 50 mg/kg three times daily versus ceftazidime 50 mg/kg three times daily ND ND outcome assessor 
Oppenheim (2000)39 no data UK. Study terminated early because of problems with supply of one of the study drugs 80 episodes; age: mean 45.6 piperacillin/tazobactam 4.5 g three times daily versus meropenem 1 g three times daily ND ND open 
Raad (2003)40 1996–2001 USA 251 patients; age: median 55, range: 15–84 cefepime 2 g three times daily versus imipenem 500 mg four times daily [in addition, glycopeptides added to 25% and aminoglycosides to 6% of patients (in both study arms)] outcome- assessor 
Ramphal (1996)41 1989–91 USA 90 patients (104 episodes); age: median 41.5, mean 43.5 ± 14.6 cefepime 2 g three times daily versus ceftazidime 2 g three times daily open 
Rolston (1992)42 1986–88 USA 378 episodes assessed; age: median 46, range: 16–84 imipenem 12.5 mg/kg four times daily versus ceftazidime 1 g six times daily (two additional arms not included in this review assessed imipenem + amikacin and ceftazidime + amikacin) ND outcome assessor 
Shichmanter (2004)43 2000–02 Israel 44 patients; age: mean 52 ± 13.6 piperacillin/tazobactam 4.5 g three times daily versus cefepime 2 g twice daily (two additional arms not included in this review compared the same drugs with amikacin added to both arms) open 
Tamura (2002)44 1999–2000 MC Japan 83 patients; age: mean 52.4 cefepime 1–2 g twice daily versus carbapenem (imipenem, panipenem or meropenem) 0.5–1 g twice daily (an additional arm not included in this review assessed cefepime + aminoglycosides) open 
Vandercam (2000)45 1992–94 Belgium 101 patients (121 episodes); age: mean 53, range: 17–86 meropenem 1 g three times daily versus ceftazidime 2 g three times daily open 
Wang (1999)46 1996–97 Taiwan 38 patients (45 episodes); age: mean 44 ± 16 cefepime 2 g thee times daily versus ceftazidime 2 g three times daily open 
Winston (1998)47 no data USA 209 patients (213 episodes); age: median 43, range: 16–81 cefoperazone/sulbactam 2 g/4 g twice daily versus imipenem 500 mg four times daily ND 
Study IDa
 
Years
 
Locationb
 
Participants
 
Interventions
 
Allocation generationc
 
Allocation concealmentc
 
Blindingc
 
Anaissie (1988)13 1985 USA 339 episodes; age:d median 42, range: 16–85 piperacillin 4 g six times daily + vancomycin 1 g twice daily versus ceftazidime 1 g six times daily + vancomycin 1 g twice daily (an additional arm not included in this review assessed all three antibiotics combined) outcome assessor 
Aoun (1997)14 1993–94 MC Belgium 111 patients (128 episodes); age: mean 48.8 ± 14.9 cefepime 2 g three times daily + vancomycin 30 mg/kg/day versus ceftazidime 2 g three times daily + vancomycin 30 mg/kg/day open 
Aparicio (1996)15 1991–94 Spain 102 patients (118 episodes); age: median 58, range: 15–75 ceftazidime 2 g three times daily versus imipenem 500 mg four times daily open 
Bickers (1990)16  USA 135 patients; age: mean 41 cefoperazone/sulbactam 2 g/1 g twice daily versus imipenem 500 mg four times daily ND ND ND 
Biron (1998)17 1995–96 MC France 380 patients (400 episodes); age: no data cefepime 2 g twice daily versus imipenem 1 g three times daily outcome assessor 
Bodey (1990)18 1985–86 USA 352 episodes; age: mean 46, range: 15–82 ticarcillin/clavulanate 4 g/0.1 g six times daily + vancomycin 1 g twice daily versus ceftazidime 1 g six times daily + vancomycin 1 g twice daily (an additional arm not included in this review assessed all three antibiotics combined) ND outcome assessor 
Bodey (1996)19 1990–93 USA 407 patients (457 episodes); age: median 51, range: 17–84 cefoperazone/sulbactam 2 g/1 g three times daily + vancomycin 1 g twice daily versus imipenem 500 mg/m2 four times daily + vancomycin 1 g twice daily outcome assessor 
Bohme (1998)20 1996 Germany 88 patients (102 episodes); age: median 45, range: 18–68 piperacillin/tazobactam 4.5 g three times daily versus cefepime 2 g three times daily ND ND open 
Boogaerts (1995)21 1991–94 MC Europe 248 patients; age: mean 42, range: 14–78 meropenem 1 g three times daily versus ceftazidime 2 g three times daily ND 
Bow (2003)2224 no data MC USA, Canada, Australia 528 patients; age not given piperacillin/tazobactam 4.5 g four times daily versus cefepime 2 g three times daily open 
Chandrasekar (2000)25 1993–94 MC USA 276 patients (315 episodes); age: median 56, range: 18–82 cefepime 2 g three times daily versus ceftazidime 2 g three times daily triple-blind 
Cherif (2004)26 1996–98 MC Sweden 180 patients (207 episodes); age: median 55, range: 16–83 cefepime 2 g three times daily versus imipenem 500 mg four times daily ND open 
Chuang (2002)27 2000–01 Taiwan 95 patients (120 episodes); age: mean 5.5, range: 0.3–15.5 cefepime 25 mg/kg twice daily versus ceftazidime 25 mg/kg twice daily ND ND open 
Cornely (2001)28 no data MC Germany 602 patients assessed; age: adults meropenem 1 g three times daily versus cefepime 2 g three times daily (an additional arm not included in this review assessed piperacillin/tazobactam + aminoglycoside) ND ND ND 
Feld (2000)29 1991–93 MC (USA, Canada, The Netherlands) 411 patients (471 episodes); age: mean 48.5; range: 17–85 meropenem 1 g three times daily versus ceftazidime 2 g three times daily ND ND double- blind 
Figuera (2001)30 1994–96 Spain 150 patients; age: mean 38.3; range: 16–73 imipenem 500 mg four times daily versus piperacillin/tazobactam 4.5 g four times daily ND ND open 
Fleischhack (2001)31 1998–2000 MC Germany 169 patients (375 episodes); age: median 7.4, range: 0.3–32.6 meropenem 60 mg/kg three times daily versus ceftazidime 100 mg/kg three times daily open 
Freifeld (1995)32 1986–90 MC USA 501 episodes; age: mean 32, range: 4–80 ceftazidime 30 mg/kg three times daily versus imipenem 15 mg/kg (or 12.5 mg/kg) four times daily ND ND open 
Glauser (1997)33 1993–95 MC Europe 281 patients (324 episodes); age: median 52, mean 50.1 ± 16.6 cefepime 2 g three times daily versus ceftazidime 2 g times daily [additional glycopeptides administered to <1% of patients (in both arms)] ND ND outcome assessor 
Kebudi (2001)34 1998 Turkey 33 patients (63 episodes); age: median 7, range: 1 month–14 years ceftazidime 33 mg/kg three times daily versus cefepime 50 mg/kg three times daily open 
Kutluk (2004)35 no data Turkey 30 patients (63 episodes); age: mean 7.4 ± 3.74 meropenem 20 mg/kg three times daily versus cefepime 50 mg/kg three times daily single 
Liang (1990)36 1988–89 Hong Kong 92 patients (103 episodes); age: mean 38, range: 16–76 ceftazidime 1 g three times daily versus imipenem 500 mg four times daily ND open 
Lindblad (1998)37 no data MC Sweden 192 patients; age: mean 60, range: 17–85 meropenem 1 g three times daily versus ceftazidime 2 g three times daily ND ND 
Mustafa (2001)38 no data USA 104 patients (149 episodes); age: median 6, range: 5 months–19 years cefepime 50 mg/kg three times daily versus ceftazidime 50 mg/kg three times daily ND ND outcome assessor 
Oppenheim (2000)39 no data UK. Study terminated early because of problems with supply of one of the study drugs 80 episodes; age: mean 45.6 piperacillin/tazobactam 4.5 g three times daily versus meropenem 1 g three times daily ND ND open 
Raad (2003)40 1996–2001 USA 251 patients; age: median 55, range: 15–84 cefepime 2 g three times daily versus imipenem 500 mg four times daily [in addition, glycopeptides added to 25% and aminoglycosides to 6% of patients (in both study arms)] outcome- assessor 
Ramphal (1996)41 1989–91 USA 90 patients (104 episodes); age: median 41.5, mean 43.5 ± 14.6 cefepime 2 g three times daily versus ceftazidime 2 g three times daily open 
Rolston (1992)42 1986–88 USA 378 episodes assessed; age: median 46, range: 16–84 imipenem 12.5 mg/kg four times daily versus ceftazidime 1 g six times daily (two additional arms not included in this review assessed imipenem + amikacin and ceftazidime + amikacin) ND outcome assessor 
Shichmanter (2004)43 2000–02 Israel 44 patients; age: mean 52 ± 13.6 piperacillin/tazobactam 4.5 g three times daily versus cefepime 2 g twice daily (two additional arms not included in this review compared the same drugs with amikacin added to both arms) open 
Tamura (2002)44 1999–2000 MC Japan 83 patients; age: mean 52.4 cefepime 1–2 g twice daily versus carbapenem (imipenem, panipenem or meropenem) 0.5–1 g twice daily (an additional arm not included in this review assessed cefepime + aminoglycosides) open 
Vandercam (2000)45 1992–94 Belgium 101 patients (121 episodes); age: mean 53, range: 17–86 meropenem 1 g three times daily versus ceftazidime 2 g three times daily open 
Wang (1999)46 1996–97 Taiwan 38 patients (45 episodes); age: mean 44 ± 16 cefepime 2 g thee times daily versus ceftazidime 2 g three times daily open 
Winston (1998)47 no data USA 209 patients (213 episodes); age: median 43, range: 16–81 cefoperazone/sulbactam 2 g/4 g twice daily versus imipenem 500 mg four times daily ND 
a

Studies are identified by the name of the first author and year of publication.

b

MC, multicentre.

c

A, adequate; ND, no description; C, inadequate. See supplementary data available at JAC Online for detailed study quality assessment.

d

Age shown in years.

Cefepime versus comparators

Cefepime was assessed in 17 trials, comparing it with ceftazidime in eight trials, imipenem in four, piperacillin/tazobactam in three and meropenem in two. Data regarding all-cause mortality were obtained for all studies. All-cause mortality was significantly higher with cefepime compared with other β-lactams (RR 1.44, 95% CI 1.06–1.94, P = 0.02), with no heterogeneity (I2 = 0%, Figure 2). The corresponding funnel plot was symmetric (intercept −0.262, 95% CI −1.350 to 0.825, P = 0.61, Figure 3).

Figure 2.

Cefepime versus comparators, all-cause mortality.

Figure 2.

Cefepime versus comparators, all-cause mortality.

Figure 3.

Funnel plot: cefepime versus comparators, all-cause mortality. Plot of the standard error of the log (effect estimate) versus effect estimate. The overall effect estimate is indicated by the vertical dotted line. Intercept (B0) is −0.262, 95% CI (−1.350 to 0.825), with t = 0.52081, df = 13, P = 0.61126.

Figure 3.

Funnel plot: cefepime versus comparators, all-cause mortality. Plot of the standard error of the log (effect estimate) versus effect estimate. The overall effect estimate is indicated by the vertical dotted line. Intercept (B0) is −0.262, 95% CI (−1.350 to 0.825), with t = 0.52081, df = 13, P = 0.61126.

Sensitivity analyses (Figure 4) show a larger mortality difference with adequate versus inadequate allocation concealment (adequate concealment RR 1.81, 95% CI 1.22–2.70), adequate versus inadequate allocation generation, among studies reporting mortality by intention to treat and in published versus unpublished studies. Four studies recruited only children and mortality was higher with cefepime when compared with ceftazidime (RR 2.28, 95% CI 0.53–9.79, 3 studies,27,34,38 263 children) and equal when compared with meropenem (1 study35). The effect estimate was higher among studies that used less than the registered recommended dose for cefepime in febrile neutropenia (2 g three times daily or 50 mg/kg three times daily for adults and children, respectively), but did not reach statistical significance (RR 2.01, 95% CI 0.87–5.08). Mortality difference remained significant in studies using the full-recommended dose with adequate allocation concealment (RR 1.73, 95% CI 1.12–2.66).

Figure 4.

Sensitivity analyses: cefepime versus comparators, all-cause mortality. Subgroup assessed (RR, 95% CI; number of studies, number of patients). Asterisk represents recommended dose, allocation concealment A or allocation concealment B.

Figure 4.

Sensitivity analyses: cefepime versus comparators, all-cause mortality. Subgroup assessed (RR, 95% CI; number of studies, number of patients). Asterisk represents recommended dose, allocation concealment A or allocation concealment B.

No significant differences were detected between cefepime and comparators with regard to all secondary efficacy outcomes (Table 2). Treatment failure, as defined, was assessed in 16 trials (RR 1.03, 95% CI 0.96–1.10, I2 = 0.1%). With adequate allocation concealment the RR for failure was 1.06 (95% CI 0.97–1.15), and with unclear concealment 0.96 (95% CI 0.85–1.07). Infection-related mortality was assessed in 13 trials (RR 1.24, 95% CI 0.78–1.97). Microbiological failure and need for drug modifications were equal. Bacterial superinfections occurred more frequently in the cefepime arm (RR 1.70, 95% CI 0.94–3.09).

Table 2.

Comparisons and outcomes (RR with 95% CIs are shown for the combined analysis of individual monotherapies versus comparators)

 Cefepime versus comparators
 
  Carbapenems versus comparators
 
  Piperacillin/tazobactam versus comparators
 
  

 
no. of studies
 
no. of participants
 
RR [95% CI]
 
no. of studies
 
no. of participants
 
RR [95% CI]
 
no. of studies
 
no. of participants
 
RR [95% CI]
 
All-cause mortality 17 3123 1.44 [1.06, 1.94] 17 3767 0.93 [0.68, 1.29] 843 0.62 [0.34, 1.13] 
Treatment failure (clinical) 16 2630 1.03 [0.96, 1.10] 19 4242 0.94 [0.89, 1.01] 887 0.93 [0.84, 1.03] 
Microbiological failure 12 584 0.98 [0.82, 1.17] 15 1127 0.98 [0.88, 1.09] 156 0.95 [0.74, 1.20] 
Infection-related mortality 13 2274 1.24 [0.78, 1.97] 13 2889 0.85 [0.51, 1.41] 750 0.52 [0.22, 1.23] 
Any antibiotic modification 816 1.01 [0.89, 1.16] 1869 0.88 [0.81, 0.94] 224 1.06 [0.90, 1.26] 
Glycopeptides addition 1080 0.96 [0.78, 1.19] 1768 0.83 [0.70, 0.99] 382 1.02 [0.77, 1.35] 
Antifungal drugs addition 780 0.94 [0.67, 1.31] 11 2110 1.14 [0.94, 1.38] 137 1.33 [0.94, 1.87] 
Any superinfection 1025 0.87 [0.57, 1.33] 2605 0.89 [0.71, 1.12]  no data  
Bacterial superinfection 780 1.70 [0.94, 3.09] 1405 0.86 [0.50, 1.47]  no data  
Any adverse event 11 1713 0.87 [0.79, 0.97] 10 2837 1.65 [1.40, 1.94] 137 0.25 [0.12, 0.53] 
 Cefepime versus comparators
 
  Carbapenems versus comparators
 
  Piperacillin/tazobactam versus comparators
 
  

 
no. of studies
 
no. of participants
 
RR [95% CI]
 
no. of studies
 
no. of participants
 
RR [95% CI]
 
no. of studies
 
no. of participants
 
RR [95% CI]
 
All-cause mortality 17 3123 1.44 [1.06, 1.94] 17 3767 0.93 [0.68, 1.29] 843 0.62 [0.34, 1.13] 
Treatment failure (clinical) 16 2630 1.03 [0.96, 1.10] 19 4242 0.94 [0.89, 1.01] 887 0.93 [0.84, 1.03] 
Microbiological failure 12 584 0.98 [0.82, 1.17] 15 1127 0.98 [0.88, 1.09] 156 0.95 [0.74, 1.20] 
Infection-related mortality 13 2274 1.24 [0.78, 1.97] 13 2889 0.85 [0.51, 1.41] 750 0.52 [0.22, 1.23] 
Any antibiotic modification 816 1.01 [0.89, 1.16] 1869 0.88 [0.81, 0.94] 224 1.06 [0.90, 1.26] 
Glycopeptides addition 1080 0.96 [0.78, 1.19] 1768 0.83 [0.70, 0.99] 382 1.02 [0.77, 1.35] 
Antifungal drugs addition 780 0.94 [0.67, 1.31] 11 2110 1.14 [0.94, 1.38] 137 1.33 [0.94, 1.87] 
Any superinfection 1025 0.87 [0.57, 1.33] 2605 0.89 [0.71, 1.12]  no data  
Bacterial superinfection 780 1.70 [0.94, 3.09] 1405 0.86 [0.50, 1.47]  no data  
Any adverse event 11 1713 0.87 [0.79, 0.97] 10 2837 1.65 [1.40, 1.94] 137 0.25 [0.12, 0.53] 

Adverse events, overall, were less frequent with cefepime owing to an advantage when compared with carbapenems (RR 0.68, 95% CI 0.52–0.89). When compared with ceftazidime no difference was seen. Adverse events requiring discontinuation occurred more frequently with cefepime (RR 1.48, 95% CI 0.89–2.46). More patients in the cefepime arm discontinued treatment owing to infections. Discontinuation owing to treatment failure was not assessed separately.

Carbapenems versus comparators (Table 2)

Imipenem was compared with ceftazidime and cefepime in four trials each, with cefoperazone/sulbactam in three trials and with piperacillin/tazobactam in one trial. Meropenem was compared with ceftazidime in five trials, cefepime in two and piperacillin/tazobactam in one trial.

All-cause mortality did not differ significantly between carbapenems and comparators (RR 0.93, 95% CI 0.68–1.29, I2 = 0%, Figure 5). The major comparison was with ceftazidime (RR 0.91, 95% CI 0.54–1.54, 8 studies, 2028 participants).

Figure 5.

Carbapenems versus comparators, all-cause mortality.

Figure 5.

Carbapenems versus comparators, all-cause mortality.

An advantage for carbapenems was observed with regard to treatment failure (Figure 6), any antibiotic modification and glycopeptide addition. Moderate heterogeneity was present in these comparisons (I2 = 25–50%). Again the major comparison was with ceftazidime and the corresponding effect estimates were RR 0.86 (95% CI 0.80–0.94), I2 = 0% for failure, RR 0.86 (95% CI 0.79–0.94), I2 = 32% for any modifications, and RR 0.70 (95% CI 0.57–0.86), I2 = 0% for glycopeptide addition. Addition of antifungals was more common with carbapenems, but the difference was not significant (RR 1.14, 95% CI 0.94–1.38). No significant advantage was observed with regard to infection-related mortality (RR 0.85, 95% CI 0.51–1.41), microbiological failure (RR 0.98, 95% CI 0.88–1.09) and superinfections (RR 0.89, 95% CI 0.71–1.12). Meta-analysis of the three trials comparing imipenem with cefoperazone/sulbactam was consistent with the overall comparison, except for failure with modification, which was non-significantly more frequent with imipenem (RR 1.19, 95% CI 0.99–1.43).

Figure 6.

Carbapenems versus comparators, treatment failure.

Figure 6.

Carbapenems versus comparators, treatment failure.

Imipenem was associated with more frequent adverse events when compared with other β-lactams, RR 1.72, 95% CI 1.45–2.04 for any adverse event (8 studies, 1419 participants) and RR 2.78, 95% CI 1.00–7.76 for seizures (7 studies, 1885 participants). Carbapenems were associated with significantly more frequent pseudomembranous colitis than cephalosporins (RR 1.94, 95% CI 1.24–3.04, Figure 7). All but one trial31 in this comparison assessed imipenem (RR 2.07, 95% CI 1.28–3.34).

Figure 7.

Carbapenems versus cephalosporins, pseudomembranous colitis.

Figure 7.

Carbapenems versus cephalosporins, pseudomembranous colitis.

Sensitivity analysis showed that the advantage to carbapenems with regard to treatment failure was present in studies with unclear allocation concealment (RR 0.90, 95% CI 0.83–0.98, 9 trials, 2094 participants), but not in studies with adequate allocation concealment (RR 1.02, 95% CI 0.92–1.13, 9 trials, 2099 participants). All-cause mortality results were unaffected. No advantage to carbapenems with regard to failure was observed when dropouts were counted as failures in studies conducted per protocol (RR 0.95, 95% CI 0.90–1.01). The RRs for clinical and microbiological failure did not decrease in recent years, as might be expected with increasing prevalence of ESBL-positive Gram-negative bacteria.

Piperacillin/tazobactam versus comparators (Table 2)

Six trials assessed piperacillin/tazobactam comparing it with imipenem and meropenem, (one trial each), and more recently with cefepime (four trials).

All-cause mortality was available in four trials (RR 0.62, 95% CI 0.34–1.13) and infection-related mortality in four (RR 0.52, 95% CI 0.22–1.23). Both favour piperacillin/tazobactam but CIs are wide owing to the small number of trials and participants.

Treatment failure was similar (RR 0.93, 95% CI 0.84–1.03, I2 = 64%, both by intention to treat and per protocol), with heterogeneity originating in a disparity between studies favouring piperacillin/tazobactam when compared with cefepime (RR 0.87, 95% CI 0.77–1.00, I2 = 0%), but not when compared with carbapenems (RR 1.08, 95% CI 0.93–1.25, I2 = 0%). Other secondary outcomes were not significantly different. None of these studies provided data regarding superinfection rates or aetiology. A single study reported adverse event rates.30

Lack of data regarding trials' methodology precluded sensitivity analyses for study quality. A single study with inadequate allocation concealment showed an advantage to cefepime with regard to treatment failure.43

No study comparing piperacillin/tazobactam with ceftazidime monotherapy was identified. Two older studies compared other penicillins with ceftazidime. Piperacillin was found significantly inferior to ceftazidime with regard to failure,13 RR 0.54, 95% CI 0.38–0.78, and ticarcillin/clavulanate was not significantly different.18

Quality assessment (supplementary data are available at JAC Online)

The methods for allocation generation and concealment were unclear in 12 and 17 of the 33 trials, respectively. Two trials were quasi-randomized, using alternation34 or patients' room assignment43 for allocation. Outcome assessors were blinded in eight trials, patients and carers in two, whereas the remaining were open. All-cause mortality was reported by intention to treat in 13 of 28 trials. Follow-up time for mortality assessment was variable, ranging between end of treatment and 30 days following end of treatment. Different lengths of follow-up may introduce bias. Patients were included in the analysis more than once in 26 of 33 trials. This is methodologically incorrect since the statistical tests used assume independence between the observations. Unclear or inadequate allocation concealment was associated with exaggerated treatment effects for failure.

Discussion

β-Lactams alone, with or without vancomycin, are commonly used for the empirical treatment of febrile neutropenia.3 These β-lactams have been assessed comparatively prior to and following their introduction into clinical use. We hypothesized that different β-lactams may be associated with different clinical outcomes. To assess this hypothesis, we assembled all randomized studies comparing different anti-pseudomonal β-lactams.

The review included 33 trials. β-Lactams assessed in more than one trial were ceftazidime, cefepime, imipenem, meropenem, piperacillin/tazobactam and cefoperazone/sulbactam. Ceftazidime was assessed most frequently and we compared other β-lactams with it.

Cefepime was associated with a higher all-cause mortality rate than its comparators (RR 1.44, 95% CI 1.06–1.94). The difference was statistically significant. Infection-related mortality, bacterial superinfections and discontinuation of the allocated treatment were more common with cefepime, whereas no differences were observed regarding other secondary outcomes.

Carbapenems were associated with an advantage with regard to treatment failure when compared with ceftazidime. The advantage disappeared in studies with proper allocation concealment and originated from a lower rate of treatment modifications, including significantly less use of glycopeptides in the carbapenem arm. Use of antifungals, however, was not reduced. All-cause mortality and microbiological eradication rates were similar to ceftazidime and other comparators. Carbapenems were associated with a significantly higher rate of pseudomembranous colitis compared with cephalosporins, with an RR of 1.94 denoting a nearly 2-fold risk. Imipenem (used 2 g/day most commonly) was associated with a 3-fold risk of seizures.

Piperacillin/tazobactam has not been compared with ceftazidime monotherapy. Its comparison with cefepime and carbapenems shows no significant differences with regard to clinical and microbiological success. A further study showed similar results for piperacillin/tazobactam monotherapy versus ceftazidime + amikacin.48 The mortality comparison is favourable to piperacillin/tazobactam, but currently available data for this and other outcomes are limited.

The adverse ecological impact of the different antibiotics was infrequently assessed. None of the trials reported on surveillance sampling, thus colonization rates with resistant microorganisms are unknown.

All-cause mortality was defined in our protocol as the primary outcome for this review.12 Undoubtedly the most significant outcome for the patient, all-cause mortality encompasses the many outcomes related to treatment of infections: efficacy, adverse events and superinfections. We extracted mortality at end of treatment and up to 30 days after end of treatment. Most fatal cases within this period are related to the infectious episode, but adequate randomization should ensure an equal distribution of non-infection-related deaths between the study groups. While primary studies are not powered to assess differences in mortality, meta-analyses may detect previously unsuspected trends.49,50

The observation that all-cause mortality is higher with cefepime is unexpected. Ceftazidime and cefepime are both advanced oxyimino-cephalosporins targeting a similar spectrum of Gram-negative bacteria.51 Cefepime was introduced as a ‘fourth generation’ cephalosporin, offering a broader in vitro spectrum of coverage than ceftazidime against Gram-positive bacteria, resistance to some of the ESBLs that inactivate ceftazidime and a lower propensity for selection of resistant (derepressed) mutants.6,51,52 We believe our search identified all randomized trials comparing cefepime with other single β-lactams for febrile neutropenia (6 of 17 studies unpublished). Despite under-reporting of all-cause mortality in the original publications, we obtained these data from the primary investigators in all studies. Adequate randomization methods and adherence to intention to treat analysis accentuated cefepime's disadvantage. Adequate allocation concealment assures against selection of sicker patients to an ‘attractive’ treatment arm.53 Analysis by intention to treat protects against unequal exclusion of cases by evaluators. No heterogeneity was detected within the comparison, attesting to similarity of effect estimates across studies. Finally, the difference persisted when analysis was limited to published studies. Overall, selection bias was minimized and sensitivity analyses support the validity of the comparison of mortality for cefepime.

We could not define the reason for increased all-cause mortality with cefepime within the secondary outcomes assessed. The trends observed with other efficacy outcomes point towards reduced efficacy. The mortality rates we observed were low (mean 3.1%). Comparatively, in-hospital mortality rates in contemporary observational studies range between 5 and 9% in a non-selected population5456 and 13% in high-risk patients.54 Thus, randomized trials tend to recruit rather low-risk patients and our analysis may underestimate differences between treatment regimens.

Implications for practice and further research

Pending further analysis, the increased mortality rate observed in febrile neutropenic patients treated with cefepime should serve as a strong caveat against its use for febrile neutropenia. Further trials are not warranted until individual patient assessment (patient data meta-analysis) of the time and cause of death in existing trials is conducted. Its efficacy for other indications should be re-assessed.

Ceftazidime, piperacillin/tazobactam, imipenem/cilastatin and meropenem appear as suitable agents for monotherapy. Local epidemiology and susceptibility patterns should serve to select the optimal of these agents. Further trials assessing piperacillin/tazobactam are needed.

Empirical use of carbapenems entails fewer treatment modifications than ceftazidime, but adverse events occur more frequently and mortality is similar. In our opinion, the increased adverse events rate, especially pseudomembranous colitis, offsets the advantage in treatment modifications. Bearing in mind that carbapenems are drugs of last resort for multidrug-resistant Gram-negative bacteria, their empirical use should be restricted. To limit glycopeptide overuse with comparators, glycopeptides should not be added empirically for persistent fever alone.57

To permit evidence-based antimicrobial treatment of febrile neutropenia randomized controlled trials should adhere to standard methodology. Allocation must ensure a truly random sequence, patients should not be included more than once and outcomes should be reported by intention to treat. Reporting of all-cause mortality must be mandatory to permit patients' safety assessment.

Transparency declarations

None declared. Complementary data were requested from the pharmaceutical companies that sponsored the studies included. None of the companies provided additional data.

Supplementary data

Supplementary data are available at http://jac.oxfordjournals.org.

We thank all the authors who provided complementary data for their studies (Table 1). In addition, we thank Alice Baruch (Pfizer) and John F. Tomayko (GlaxoSmithKline) for conducting independent searches for additional studies. This work was supported in part by an EC 5th framework IST grant (TREAT project, grant no. 1999-11459). Preliminary results were presented at the Fifteenth European Congress of Clinical Microbiology and Infectious Diseases, 2004. The protocol for this review is published in The Cochrane Library,12 where the complete review will be published and updated.

References

1.
Bodey GP, Elting L, Kassamali H et al. Escherichia coli bacteremia in cancer patients.
Am J Med
 
1986
;
81
:
85
–95.
2.
Paul M, Soares-Weiser K, Leibovici L. β-Lactam monotherapy versus β-lactam-aminoglycoside combination therapy for fever with neutropenia: systematic review and meta-analysis.
Br Med J
 
2003
;
326
:
1111
–15.
3.
Hughes WT, Armstrong D, Bodey GP et al. 2002 guidelines for the use of antimicrobial agents in neutropenic patients with cancer.
Clin Infect Dis
 
2002
;
34
:
730
–51.
4.
Fritsche TR, Sader HS, Jones RN. Comparative activity and spectrum of broad-spectrum β-lactams (cefepime, ceftazidime, ceftriaxone, piperacillin/tazobactam) tested against 12,295 staphylococci and streptococci: report from the SENTRY antimicrobial surveillance program (North America: 2001–2002).
Diagn Microbiol Infect Dis
 
2003
;
47
:
435
–40.
5.
Pfaller MA, Jones RN, Marshall SA et al. Inducible amp C β-lactamase producing gram-negative bacilli from blood stream infections: frequency, antimicrobial susceptibility, and molecular epidemiology in a national surveillance program (SCOPE).
Diagn Microbiol Infect Dis
 
1997
;
28
:
211
–19.
6.
Jacoby GA, Munoz-Price LS. The new β-lactamases.
N Engl J Med
 
2005
;
352
:
380
–91.
7.
Alderson P, Green S, Higgins JPT et al. Highly sensitive search strategies for identifying reports of randomized controlled trials in MEDLINE. Cochrane Reviewers' Handbook 4.2.2 [updated December 2003]. Appendix & Sons, Ltd; 5b. In: The Cochrane Library, Issue 1. Chichester, UK: John Wiley & Sons, Ltd.;
2004
.
8.
Moher D, Pham B, Jones A et al. Does quality of reports of randomised trials affect estimates of intervention efficacy reported in meta-analyses?
Lancet
 
1998
;
352
:
609
–13.
9.
Schulz KF, Chalmers I, Hayes RJ et al. Empirical evidence of bias. Dimensions of methodological quality associated with estimates of treatment effects in controlled trials.
JAMA
 
1995
;
273
:
408
–12.
10.
Higgins JP, Thompson SG, Deeks JJ et al. Measuring inconsistency in meta-analyses.
Br Med J
 
2003
;
327
:
557
–60.
11.
Sterne JA, Gavaghan D, Egger M. Publication and related bias in meta-analysis: power of statistical tests and prevalence in the literature.
J Clin Epidemiol
 
2000
;
53
:
1119
–29.
12.
Paul M, Yahav D, Fraser A et al. Single-agent antibiotic therapy for the initial, empirical, treatment of febrile neutropenia: comparison of β-lactams. (Protocol). The Cochrane Database of Systematic Reviews
2005
, Issue 2. Art. No.: CD005197. doi: 10.1002/14651858.CD005197.
13.
Anaissie EJ, Fainstein V, Bodey GP et al. Randomized trial of β-lactam regimens in febrile neutropenic cancer patients.
Am J Med
 
1988
;
84
:
581
–9.
14.
Aoun M, Boogaerts MA, Bosly AJ et al. A multicenter, randomized, prospective, comparative study of cefepime plus vancomycin versus ceftazidime plus vancomycin as empiric therapy in the treatment of febrile episodes in granulocytopenic patients with hematological malignancies. Joint clinical/statistical review of NDA 50,679/SE1-002. http://www.fda.gov/cder/foi/nda/97/050679s002.htm (18 September 2005, date last accessed).
15.
Aparicio J, Oltra A, Llorca C et al. Randomised comparison of ceftazidime and imipenem as initial monotherapy for febrile episodes in neutropenic cancer patients.
Eur J Cancer
 
1996
;
32A
:
1739
–43.
16.
Bickers J, Gumbart C, Cavalier J et al. A prospective randomized trial of cefoperazone plus sulbactam versus imipenem-cilastatin in febrile neutropenic patients.
Clin Res
 
1990
;
38
:
991A
.
17.
Biron P, Fuhrmann C, Cure H et al. Cefepime versus imipenem-cilastatin as empirical monotherapy in 400 febrile patients with short duration neutropenia. CEMIC (Study Group of Infectious Diseases in Cancer).
J Antimicrob Chemother
 
1998
;
42
:
511
–18.
18.
Bodey GP, Fainstein V, Elting LS et al. β-Lactam regimens for the febrile neutropenic patient.
Cancer
 
1990
;
65
:
9
–16.
19.
Bodey G, Abi-Said D, Rolston K et al. Imipenem or cefoperazone-sulbactam combined with vancomycin for therapy of presumed or proven infection in neutropenic cancer patients.
Eur J Clin Microbiol Infect Dis
 
1996
;
15
:
625
–34.
20.
Bohme A, Shah PM, Stille W et al. Piperacillin/tazobactam versus cefepime as initial empirical antimicrobial therapy in febrile neutropenic patients: a prospective randomized pilot study.
Eur J Med Res
 
1998
;
3
:
324
–30.
21.
Boogaerts MA, Demuynck H, Mestdagh N et al. Equivalent efficacies of meropenem and ceftazidime as empirical monotherapy of febrile neutropenic patients.
J Antimicrob Chemother
 
1995
;
36
:
185
–200.
22.
Bow EJ, Noskin GA, Schwarer AP et al. Efficacy of piperacillin/tazobactam as initial empiric therapy of febrile neutropenia in patients with hematologic malignancy. In: Abstracts of the Forty-fifth Annual Meeting of the American Society of Hematology, San Diego, CA,
2003
. Abstract 1000. American Society of Hematology, Washington, DC, USA.
23.
Bow EJ, Schwarer AP, Laverdiere M et al. Efficacy of piperacillin/tazobactam as initial empiric therapy of febrile neutropenia in patients with hematologic malignancy. In: Abstracts of the Forty-third Interscience Conference on Antimicrobial Agents and Chemotherapy, Chicago, IL,
2003
. Abstract L-114. American Society for Microbiology, Washington, DC, USA.
24.
Noskin GA, Vesole D, Szer J et al. Multicenter efficacy and safety study of piperacillin/tazobactam as empiric treatment for neutropenic fever in patients with hematologic malignancies. In: Abstracts of the Annual Meeting of the Infectious Diseases Society of America, San Diego, CA,
2003
. Abstract no. 373. Infectious Diseases Society of America, Alexandria, VA, USA.
25.
Chandrasekar PH, Arnow PM. Cefepime versus ceftazidime as empiric therapy for fever in neutropenic patients with cancer.
Ann Pharmacother
 
2000
;
34
:
989
–95.
26.
Cherif H, Bjorkholm M, Engervall P et al. A prospective, randomized study comparing cefepime and imipenem-cilastatin in the empirical treatment of febrile neutropenia in patients treated for haematological malignancies.
Scand J Infect Dis
 
2004
;
36
:
593
–600.
27.
Chuang YY, Hung IJ, Yang CP et al. Cefepime versus ceftazidime as empiric monotherapy for fever and neutropenia in children with cancer.
Pediatr Infect Dis J
 
2002
;
21
:
203
–9.
28.
Cornely OA, Reichert D, Buchheidt D et al. Three-armed multicenter randomized study on the empiric treatment of neutropenic fever in a high risk patient population (PEG Study III). In: Abstracts of the Forty-first Interscience Conference on Antimicrobials Agents and Chemotherapy, Chicago, IL,
2001
. Abstract L-775, p. 446. American Society for Microbiology, Washington, DC, USA.
29.
Feld R, DePauw B, Berman S et al. Meropenem versus ceftazidime in the treatment of cancer patients with febrile neutropenia: a randomized, double-blind trial.
J Clin Oncol
 
2000
;
18
:
3690
–8.
30.
Figuera A, Rivero N, Pajuelo F et al. Initial empiric therapy of febrile neutropenia with antibiotic monotherapy. Piperacillin/tazobactam versus imipenem/cilastatin (1994-1996).
Med Clin
 
2001
;
116
:
610
–11.
31.
Fleischhack G, Hartmann C, Simon A et al. Meropenem versus ceftazidime as empirical monotherapy in febrile neutropenia of paediatric patients with cancer.
J Antimicrob Chemother
 
2001
;
47
:
841
–53.
32.
Freifeld AG, Walsh T, Marshall D et al. Monotherapy for fever and neutropenia in cancer patients: a randomized comparison of ceftazidime versus imipenem.
J Clin Oncol
 
1995
;
13
:
165
–76.
33.
Glauser M, Dekker AW, Van Marwijkkooy M et al. A multicenter comparative study of cefepime vs. ceftazidime as empiric therapy in the treatment of febrile episodes in neutropenic patients. Joint clinical/statistical review of NDA 50,679/SE1-002. http://www.fda.gov/cder/foi/nda/97/050679s002.htm (18 September, 2005, date last accessed).
34.
Kebudi R, Gorgun O, Ayan I et al. Randomized comparison of cefepime versus ceftazidime monotherapy for fever and neutropenia in children with solid tumors.
Med Pediatr Oncol
 
2001
;
36
:
434
–41.
35.
Kutluk T, Kurne O, Akyuz C et al. Cefepime vs. meropenem as empirical therapy for neutropenic fever in children with lymphoma and solid tumours.
Pediatr Blood Cancer
 
2004
;
42
:
284
–6.
36.
Liang R, Yung R, Chiu E et al. Ceftazidime versus imipenem-cilastatin as initial monotherapy for febrile neutropenic patients.
Antimicrob Agents Chemother
 
1990
;
34
:
1336
–41.
37.
Lindblad R, Rodjer S, Adriansson M et al. Empiric monotherapy for febrile neutropenia—a randomized study comparing meropenem with ceftazidime.
Scand J Infect Dis
 
1998
;
30
:
237
–43.
38.
Mustafa MM, Carlson L, Tkaczewski I et al. Comparative study of cefepime versus ceftazidime in the empiric treatment of pediatric cancer patients with fever and neutropenia.
Pediatr Infect Dis J
 
2001
;
20
:
362
–9.
39.
Oppenheim BA, Morgenstern GR, Chang J et al. Safety and efficacy of piperacillin/tazobactam versus meropenem in the treatment of febrile neutropenia.
J Infect
 
2001
;
43
:
A67
.
40.
Raad, II, Escalante C, Hachem RY et al. Treatment of febrile neutropenic patients with cancer who require hospitalization: a prospective randomized study comparing imipenem and cefepime.
Cancer
 
2003
;
98
:
1039
–47.
41.
Ramphal R, Haddow A, McCracken GH. A comparative study of cefepime vs. ceftazidime in the treatment of adult cancer patients with fever and neutropenia. Joint clinical/statistical review of NDA 50,679/SE1-002. http://www.fda.gov/cder/foi/nda/97/050679s002.htm (18 September, 2005, date last accessed).
42.
Rolston KV, Berkey P, Bodey GP et al. A comparison of imipenem to ceftazidime with or without amikacin as empiric therapy in febrile neutropenic patients.
Arch Intern Med
 
1992
;
152
:
283
–91.
43.
Shichmanter R, Miller E, Shtalrid M et al. Empiric antibiotic treatment of adults with neutropenic fever in a general internal medicine ward. In: Abstracts of the Fourteenth European Congress of Clinical Microbiology and Infectious Diseases, Prague,
2004
. Abstract P1452. European Society of Clinical Microbiology and Infectious Diseases, Basel, Switzerland.
44.
Tamura K, Matsuoka H, Tsukada J et al. Cefepime or carbapenem treatment for febrile neutropenia as a single agent is as effective as a combination of 4th-generation cephalosporin + aminoglycosides: comparative study.
Am J Hematol
 
2002
;
71
:
248
–55.
45.
Vandercam B, Gerain J, Humblet Y et al. Meropenem versus ceftazidime as empirical monotherapy for febrile neutropenic cancer patients.
Ann Hematol
 
2000
;
79
:
152
–7.
46.
Wang FD, Liu CY, Hsu HC et al. A comparative study of cefepime versus ceftazidime as empiric therapy of febrile episodes in neutropenic patients.
Chemotherapy
 
1999
;
45
:
370
–9.
47.
Winston DJ, Bartoni K, Bruckner DA et al. Randomized comparison of sulbactam/cefoperazone with imipenem as empirical monotherapy for febrile granulocytopenic patients.
Clin Infect Dis
 
1998
;
26
:
576
–83.
48.
Hess U, Bohme C, Rey K et al. Monotherapy with piperacillin/tazobactam versus combination therapy with ceftazidime plus amikacin as an empiric therapy for fever in neutropenic cancer patients.
Support Care Cancer
 
1998
;
6
:
402
–9.
49.
Juni P, Nartey L, Reichenbach S et al. Risk of cardiovascular events and rofecoxib: cumulative meta-analysis.
Lancet
 
2004
;
364
:
2021
–9.
50.
Psaty BM, Furberg CD. COX-2 inhibitors—lessons in drug safety.
N Engl J Med
 
2005
;
352
:
1133
–5.
51.
Sanders CC. Cefepime: the next generation?
Clin Infect Dis
 
1993
;
17
:
369
–79.
52.
Goldstein FW. Cephalosporinase induction and cephalosporin resistance: a longstanding misinterpretation.
Clin Microbiol Infect
 
2002
;
8
:
823
–5.
53.
Egger M, Altman DG, Smith GD. Systematic Reviews in Health Care: Meta-Analysis in Context. London: BMJ Books;
2001
.
54.
Klastersky J, Paesmans M, Rubenstein EB et al. The Multinational Association for Supportive Care in Cancer risk index: a multinational scoring system for identifying low-risk febrile neutropenic cancer patients.
J Clin Oncol
 
2000
;
18
:
3038
–51.
55.
Ortega M, Rovira M, Filella X et al. Prospective evaluation of procalcitonin in adults with febrile neutropenia after haematopoietic stem cell transplantation.
Br J Haematol
 
2004
;
126
:
372
–6.
56.
Reuter S, Kern WV, Sigge A et al. Impact of fluoroquinolone prophylaxis on reduced infection-related mortality among patients with neutropenia and hematologic malignancies.
Clin Infect Dis
 
2005
;
40
:
1087
–93.
57.
Paul M, Borok S, Fraser A et al. Empirical antibiotics against Gram-positive infections for febrile neutropenia: systematic review and meta-analysis of randomized controlled trials.
J Antimicrob Chemother
 
2005
;
55
:
436
–44.

Author notes

1Department of Medicine E, Rabin Medical Center, Beilinson Campus, 49100 Petah-Tiqva, Israel; 2Sackler Faculty of Medicine, Tel-Aviv University, Tel Aviv, Israel