Empirical antibiotics targeting Gram-positive bacteria for the treatment of febrile neutropenic patients with cancer

  • Review
  • Intervention

Authors


Abstract

Background

The pattern of infections among neutropenic cancer patients has shifted in the last decades to a predominance of Gram-positive infections. Some of these Gram-positive bacteria are increasingly resistant to beta-lactams and necessitate specific antibiotic treatment.

Objectives

To assess the effectiveness of empirical antiGram-positive (antiGP) antibiotic treatment for febrile neutropenic cancer patients in terms of mortality and treatment failure. To assess the rate of resistance development, further infections and adverse events associated with additional antiGP treatment.

Search methods

We searched the Cochrane Central Register of Controlled Trials (CENTRAL) (2013, Issue 7), MEDLINE (1966 to 2013), EMBASE (1982 to 2013), LILACS (1982 to 2013), conference proceedings, and the references of the included studies. First authors of all included and potentially relevant trials were contacted.

Selection criteria

Randomised controlled trials (RCTs) comparing one antibiotic regimen to the same regimen with the addition of an antiGP antibiotic for the treatment of febrile neutropenic cancer patients.

Data collection and analysis

Two review authors independently assessed trial eligibility and risk of bias, and extracted all data. Risk ratios (RR) with 95% confidence intervals (CI) were calculated. A random-effects model was used for all comparisons showing substantial heterogeneity (I2 > 50%). Outcomes were extracted by intention to treat and the analysis was patient-based whenever possible.

Main results

Thirteen trials and 2392 patients or episodes were included. Empirical antiGP antibiotics were tested at the onset of treatment in 11 studies, and for persistent fever in two studies. The antiGP treatment was a glycopeptide in nine trials. Seven studies were assessed in the overall mortality comparison and no significant difference was seen between the comparator arms, RR of 0.82 (95% CI 0.56 to 1.20, 852 patients). Ten trials assessed failure, including modifications as failures, while six assessed overall failure disregarding treatment modifications. Failure with modifications was significantly reduced, RR of 0.76 (95% CI 0.68 to 0.85, 1779 patients) while overall failure was the same, RR of 1.00 (95% CI 0.79 to 1.27, 943 patients). Sensitivity analysis for allocation concealment and incomplete outcome data did not change the results. Both mortality and failure did not differ significantly among patients with Gram-positive infections, but the number of studies in the comparisons was small. Data regarding other patient subgroups likely to benefit from antiGP treatment were not available. Glycopeptides did not increase fungal superinfection rates and were associated with a reduction in documented Gram-positive superinfections. Resistant colonisation was not documented in the studies.

Authors' conclusions

Current evidence shows that the empirical routine addition of antiGP treatment, namely glycopeptides, does not improve the outcomes of febrile neutropenic patients with cancer.

Plain language summary

Spectrum of the initial antibiotic treatment for cancer patients with fever and low leukocyte counts

Background: cancer patients develop neutropenia, a decrease in the subset of leukocytes responsible for protection against bacteria, as a result of chemotherapy or cancer. Neutropenia predisposes the patients to severe bacterial infections. Standard antibiotic regimens for cancer patients with neutropenia and fever are directed at most of the bacteria that can cause infections. However, a subset of resistant bacteria belonging to the the Gram-positive group (Staphylococcus aureus and Streptococci) remain untreated unless specific antibiotics are added to the treatment.

Review question: we assessed whether the addition of specific antiGram-positive antibiotics prior to identification of a causative bacteria improves survival and cure among cancer patients with fever and neutropenia.

Search dates: the evidence is current to August 2013.

Study characteristics: we included randomised controlled trials that compared a standard antibiotic regimen to the same regimen with an antibiotic directed at Gram-positive bacteria. Overall, 13 randomised controlled trials were included with 2392 patients or episodes of infection. The antibiotics were given to cancer patients with neutropenia and fever as first-line treatment (11 trials) or for recurrent fever (two trials).

Study funding sources: in 8/13 of the trials the trial received funding from the industry.

Key results: mortality did not differ significantly between patients groups. Antibiotic treatment was more frequently modified among patients who did not initially receive specific antibiotics against Gram-positive bacteria, but overall treatment failures were not significantly different. We attempted to examine the durations of fever and hospital stay, but these were not consistently reported. The addition of specific antibiotics against Gram-positive bacteria resulted in more adverse events, mainly rash. We conclude that antibiotic treatment directed against resistant Gram-positive bacteria can await identification of bacteria and need not be given routinely prior to bacterial identification.

Quality of the evidence: overall the quality of the evidence was good since it relied on randomised controlled trials, most of which were at low risk of bias. A limitation of the results for mortality was that all-cause mortality was not reported and could not be obtained in 6/13 of the studies. The trials did not examine specific circumstances that might mandate empirical use of antibiotics against Gram-positive bacteria and thus the evidence is relevant to cancer patients with fever, without low blood pressure, or a focus of infection that might be caused by Gram-positive bacteria.

Резюме на простом языке

Спектр стартовой антибактериальной терапии у больных раком с лихорадкой и низким числом лейкоцитов

Актуальность: У больных раком, в результате проведения химиотерапии или самого рака, может развиться нейтропения - снижение числа лейкоцитов [нейтрофилов], отвечающих за защиту против бактерий. Нейтропения способствует развитию тяжелых бактериальных инфекций у пациентов. Стандартные режимы антибиотикотерапии у онкологических больных с нейтропенией и лихорадкой направлены на большинство бактерий, которые могут вызвать инфекции. Однако, некоторая часть устойчивых бактерий, принадлежащих к группе Грам-положительных бактерий ( золотистый стафилококк и стрептококки ) остаются "не вылеченными", если определенные антибиотики не будут добавлены к лечению.

Вопрос обзора : мы оценили, улучшает ли добавление определенных антиГрам-положительных антибиотиков до идентификации бактериальных возбудителей выживаемость и показатели излечения среди больных раком с лихорадкой и нейтропенией.

Дата поиска : доказательства актуальны по август 2013 года.

Характеристика исследований : мы включили рандомизированные контролируемые клинические испытания, в которых сравнивали стандартный режим антибактериальной терапии с таким же режимом в сочетании с антибиотиком, направленным на Грам-положительные бактерии. В целом, было включено 13 рандомизированных контролируемых испытаний с участием 2392 пациентов или эпизодов инфекции. Антибиотики были назначены больным раком с нейтропенией и лихорадкой в качестве терапии ​​первой линии (11 клинических испытаний) или при рецидивирующей [повторной] лихорадке (2 клинических испытания).

Источники финансирования исследований : 8 из 13 клинических испытаний получили финансирование от промышленности [фармацевтического производителя].

Основные результаты : смертность существенно не различалась между группами пациентов. Антибактериальную терапию изменяли чаще у пациентов, которые изначально (в качестве стартовой терапии) не получали специфические антибиотики против Грам-положительных бактерий, но в целом, неудачи лечения существенно не различались. Мы попытались изучить длительность лихорадки и пребывания в больнице, но об этом соответствующим образом не сообщали. Добавление специфических антибиотиков против Грам-положительных бактерий привело к большему числу неблагоприятных событий, в основном, к сыпи. Мы пришли к выводу, что антибактериальную терапию, направленную против устойчивых Грам-положительных бактерий, можно отложить до идентификации бактерий, и их назначение до идентификации бактерий обычно не требуется.

Качество доказательств : в целом, качество доказательств было хорошим, так как они опирались на рандомизированные контролируемые испытания, большинство из которых были с низким риском смещения. Ограничение результатов в отношении смертности было в том, что в 6 из 13 клинических исследований не сообщали о смертности от всех причин, и не было возможности собрать эти данные. В клинических испытаниях не исследовали конкретные обстоятельства, которые могли бы быть показаниями для эмпирического использования антибиотиков, направленных против Грам-положительных бактерий, и, следовательно, доказательства относятся к больных раком с лихорадкой, без низкого артериального давления, или очага инфекции, который мог бы быть вызван Грам-положительными бактериями.

Заметки по переводу

Перевод: Мухаметзянова Алсу Сириновна. Редактирование: Юдина Екатерина Викторовна. Координация проекта по переводу на русский язык: Казанский федеральный университет - аффилированный центр в Татарстане Северного Кокрейновского Центра. По вопросам, связанным с этим переводом, пожалуйста, обращайтесь к нам по адресу: lezign@gmail.com

Background

Advances in therapy for cancer patients are associated with an increased risk of infection. Newer chemotherapeutic regimens, indwelling intravenous catheters, and bone marrow transplantation for both hematological and solid tumor cancer patients constitute major risk factors for infection. These cause bone marrow suppression with resulting neutropenia and damage to the physiological barriers of infection such as skin and mucous membranes. Infections are the most common cause of death among cancer patients and they are a common rate-limiting factor for continuing cancer therapy (Pizzo 2001).

Gram-negative bacteria were the most common cause for bacteriologically documented infections when empirical treatment for neutropenic cancer patients was proposed. Eventually, Gram-positive bacteria have replaced the Gram-negative bacteria as the most commonly documented infection. These include mainly Staphylococci, Streptococcus species, Enterococci and Corynebacterium species. The European Organisation for Research and Treatment of Cancer International Antimicrobial Therapy Cooperative Group (EORTC-IATCG) has conducted several multi-center randomised controlled trials (RCTs) of empirical therapy in cancer patients with fever and neutropenia throughout this period (EORTC 1978; EORTC 1983; EORTC 1986; EORTC 1987; EORTC 1991; EORTC 1993; EORTC 1995; EORTC 1996). In these trials the frequency of Gram-positive isolates has been steadily increasing, from 29% of single-organism bacteremias in the trial conducted between 1973 and 1976 (EORTC 1978) to 41% between 1983 and 1986 (EORTC 1986), 64% between 1986 and 1988 (EORTC 1991), 67% between 1991 and 1992, and 69% between 1993 and 1994 (EORTC 1996). In these same trials the rate of single-agent Gram-negative bacteremias dropped from 71% to 31%. In addition, the overall mortality associated with treated infection has decreased from around 25% to 6% in trials conducted during the recent years (Del Favero 2001; EORTC 1996; Gurwith 1978). Of late, epidemiology might be reverting to a predominance of Gram-negative bacteria, at least in some locations (Montassier 2013).

Several explanations may underlie the changes in the epidemiology of febrile neutropenia. The increase in infections due to Gram-positive bacteria is probably due mainly to the widespread use of centrally placed venous catheters, which have the propensity to be colonised by Gram-positive bacteria (Press 1984). Mucositis induced by intensive chemotherapy is similarly associated with Gram-positive bacteria. Quinolone prophylaxis decreases both the incidence of Gram-negative and Gram-positive infections, but infections occurring despite prophylaxis are more likely to be Gram-positive (Bucaneve 2005; Gafter-Gvili 2005). Multiple reasons account for the decreased mortality. The fact that infections due to Gram-positive bacteria are usually less rapidly fatal than Gram-negative infections may contribute to the decrease (EORTC 1991; Rubin 1988).

Gram-positive bacteria among cancer patients are frequently resistant to the beta-lactams, which are currently recommended for the empirical treatment of febrile neutropenic cancer patients. The observed prevalence of Staphylococcus aureus resistant to methicillin (an antistaphylococcal penicillin) is currently about 30% in the United States and many European countries, reaching 70% in Japan, while that of Staphylococcus epidermidis is greater than 70% worldwide (Diekema 2001).

Current guidelines for the use of antimicrobial agents in febrile neutropenic patients advise against routine empirical treatment with glycopeptides, prior to identification of the causative pathogen or its susceptibilities (Cometta 2007; Freifeld 2011; Penack 2011). Exceptions are recommended for patients with hypotension, severe sepsis and septic shock, and those with severe mucositis. Pre-emptive treatment is advised for patients with suspected catheter-related infections. European guidelines recommend the consideration of empirical glycopeptides in centers where resistant Gram-positive bacteria (that is methicillin-resistant S. aureus or penicillin-resistant streptococci) are predominant (Cometta 2007). Targeted treatments are recommended for patients with documented infections caused by beta-lactam resistant Gram-positive bacteria, and for those with bacteremia caused by Gram-positive bacteria, prior to final identification of the pathogen and susceptibility testing (Freifeld 2011).

Thus, although infections due to Gram-positive bacteria currently dominate, empirical treatment often does not cover these pathogens. However, withholding broad-spectrum antiGP treatment is not necessarily detrimental and may even be advantageous. Early empirical antibiotic treatment for febrile neutropenic patients was suggested when Gram-negative organisms dominated. Such early treatment reduced mortality since Gram-negative infections are notoriously rapidly fatal. Infections due to Gram-positive bacteria, especially those caused by coagulase-negative Staphylococci, may be less rapidly fatal permitting initiation of specific antibiotic treatment when an infection is documented (Viscoli 1991). Administration of glycopeptides may be associated with adverse effects, especially when combined with aminoglycosides or other nephrotoxic agents (Goetz 1993). Moreover, use of glycopeptides has been associated with emergence of glycopeptide-resistant enterococci, and lately with Staphylococcus aureus resistant to glycopeptides (Montecalvo 1994; Sievert 2002; Tenover 2001). Finally, use of glycopeptides may increase the risk of fungal superinfections, a feared complication with persistent neutropenia (Pagano 1999). Widespread empirical treatment with glycopeptides should therefore be very carefully considered and, indeed, in the Center for Disease Control recommendations for preventing the spread of vancomycin resistance, the empirical use of vancomycin for febrile neutropenic patients, is specifically discouraged (CDC 1995).

Considering an overall mortality rate among patients with febrile neutropenia of around 6% (Paul 2003), the sample size needed to assess the effect of antiGP treatment is large. We therefore conducted a meta-analysis of trials comparing the treatment for febrile neutropenia with or without specific antiGP coverage. We looked for specific patient subgroups for whom antiGP treatment may be specifically indicated.

Objectives

  • To assess whether the addition of empirical antiGram-positive (antiGP) antibiotic treatment in febrile neutropenic cancer patients in terms of mortality and treatment failure.

  • To assess the rate of resistance development, further infections and adverse events associated with additional antiGP treatment.

In addition, we planned to compare the effectiveness of empirical treatment with and without additional antiGP antibiotics in the following patient subgroups:

  • patients with Gram-positive infections;

  • patients with central venous catheters;

  • patients having received quinolone prophylaxis.

Methods

Criteria for considering studies for this review

Types of studies

Any randomised or quasi-RCT. We included studies assessing empirical or pre-emptive antibiotic treatment (that is treatment instituted before final identification of causative pathogen(s) and their susceptibilities), both at onset of treatment (empirical' treatment) and for fever persisting beyond 48 to 72 hours after treatment initiation (first modification). Only studies comparing one antibiotic regimen with or without a placebo to the same antibiotic regimen with the addition of an antiGP antibiotic (as defined) were included. Studies comparing different antibiotic regimens, including an antiGP antibiotic in one arm, were excluded.

Types of participants

Febrile neutropenic cancer patients with suspected or documented infections.

Types of interventions

The following antiGP antibiotics were included.

  • Glycopeptides:

    • vancomycin;

    • teicoplanin.

  • Beta-lactams:

    • penicillinase resistant penicillins, oxacillin, cloxacillin, dicloxacillin, flucloxacillin, or nafcillin;

    • first generation cephalosporins, cefazolin.

  • Lincosamines:

    • clindamycin.

  • Streptogramins:

    • quinupristin-dalfopristin.

  • Oxazolidinones:

    • linezolid.

  • Sulphonamides:

    • trimethoprim-sulphamethoxazole.

  • Lipopeptides:

    • daptomycin.

  • Glycylcyclines:

    • tigecycline.

Types of outcome measures

Primary outcomes
  • Overall mortality at end of study follow-up and up to 30 days following end of treatment. We extracted 30-day mortality. If not reported, we used overall mortality data at the latest point of study follow-up when the follow-up did not exceed 30 days.

Secondary outcomes
  • Treatment failure, as defined in study, once including any modification of the empirical antibiotic regimen in the definition of failure (modifications included), and once disregarding treatment modifications (overall failure) (Consensus 1990)

  • Duration of fever and hospital stay among survivors

  • Removal of central catheter

  • Addition of amphotericin (antifungal antibiotic)

  • Superinfection: new, persistent, or worsening symptoms or signs of infection associated with the isolation of a new pathogen (different pathogen, or same pathogen with different susceptibilities), or the development of a new site of infection

  • Colonisation by resistant bacteria: the isolation of bacteria during or following antibiotic therapy, without signs or symptoms of infection

  • Development of resistance: change in susceptibility of pathogens isolated at initiation of antibiotic therapy

  • Adverse events

The adverse events were described as:

(1) any serious adverse events that were fatal, life-threatening, or requiring inpatient hospitalisation or prolongation of existing hospitalisation (death due to adverse event, anaphylaxis, nephrotoxicity requiring renal replacement therapy, pseudomembranous colitis). Serious adverse events were not independent of the primary outcome, overall mortality;
(2) any adverse events that resulted in significant disability or incapacity (e.g. nephrotoxicity, ototoxicity, bleeding severe skin reactions);
(3) any important medical events that might not be immediately life-threatening or result in death or hospitalisation but might jeopardise the patient or require intervention to prevent one of the above outcomes;
(4) any adverse events that required discontinuation of medication;
(5) any adverse event.

Search methods for identification of studies

Electronic searches

A comprehensive search strategy was formulated in an attempt to identify all relevant studies regardless of language or publication status, in combination with the search strategy for clinical trials developed by The Cochrane Collaboration and detailed in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). The following databases were searched using the tailored search strategies detailed in Appendix 1, Appendix 2, and Appendix 3.

Cochrane Central Register of Controlled Trials (CENTRAL) (2013, Issue 7).
MEDLINE (1966 to August 2013).
EMBASE (1988 to August 2013).
LILACS (1982 to August 2013).

Searching other resources

The bibliographies of all included studies and pertinent reviews were scanned for additional references. The first or corresponding author of each included study, and the researchers active in the field, were contacted for information regarding unpublished trials or complementary information on their own trials. We searched the following conference proceedings for unpublished trials: Interscience Conference on Antimicrobial Agents and Chemotherapy (ICAAC) (1995 to 2011); European Congress of Clinical Microbiology and Infectious Diseases (ECCMID) (2003 to 2012). We searched for ongoing and unpublished trials in the National Institutes of Health database (http://clinicaltrials.gov/).

Data collection and analysis

Selection of studies

One review author (YD for the 2013 update) inspected the abstract of each reference identified in the search and applied the inclusion criteria. Where potentially relevant articles were identified, the full article was obtained and inspected independently by two review authors (MP, LL).

Data extraction and management

Two review authors extracted the data from the included trials independently into a data extraction sheet (MP, SB). Differences in the data that was extracted were resolved by discussion with a third review author (LL). Justification for excluding studies from the review was documented. Authors of all included trials, and trials in the assessment for inclusion, were contacted for clarifications and further information. Data regarding all-cause mortality and randomisation methods were primarily requested.

For the mortality comparison we extracted results by intention to treat (ITT), including all individuals randomised in the outcome assessment. Where impossible, the data by available case analysis were extracted. We compared the main analysis, including both types of studies, to the ITT analysis. All other outcome data were extracted preferentially by ITT and combined with the available case analysis. For sensitivity analysis, we imputed failure for all dropouts and presented an ITT analysis including all randomised individuals in the denominator. We could not include all studies in this comparison as some trials did not report the number of dropouts per study arm, prohibiting imputation for dropouts.

Some trials allowed the inclusion of several episodes for each patient; these outcomes for different episodes in the same patient are not independent. Ideally, such trials should be analysed allowing for clustering of episodes within patients, but this clustering is often ignored giving rise to spuriously narrow confidence intervals on the estimated effects of treatment. To minimise such problems, we extracted the number of patients and episodes per trial. Where data were available, we used the number of patients with the outcome and number of patients randomised, rather than basing the analysis on episodes.

The following data were extracted, checked, and recorded.

Trial characteristics
  • Year (defined as recruitment initiation year) and country of study

  • Trial sponsor

  • Publication status: published in journal; abstract or proceeding; unpublished

  • ITT analysis: performed; possible to extract; efficacy analysis

  • Randomisation methods: allocation generation and concealment

  • Blinding

  • Failure definition: including time of failure assessment

  • Study follow-up duration

  • Performance of surveillance cultures

Patient characteristics
  • Number of patients with clinically documented infections

  • Number of patients with bacteriologically documented infections

  • Number of patients with documented infections due to Gram-positive bacteria: any Gram-positive, Staphlococcus epidermidis, Staphylococcus aureus; Streptococci

  • Number of patients with bacteremia

  • Number of patients with Gram-positive bacteremia: any Gram-positive, Staphylococcus epidermidis, Staphylococcus aureus; Streptococci

  • Number of patients with Gram-negative bacteremia

  • Number of patients with infections caused by bacteria resistant to the administered antibiotic regimen: methicillin-resistant staphylococci; other

  • Infection characteristics

  • Number of patients with clinically documented infections

  • Number of patients with bacteriologically documented infections

  • Number of patients with documented Gram-positive infections: any Gram-positive, Staphlococcus epidermidis, Staphylococcus aureus; Streptococci

  • Number of patients with bacteremia

  • Number of patients with Gram-positive bacteremia: any Gram-positive, Staphylococcus epidermidis, Staphylococcus aureus; Streptococci

  • Number of patients with Gram-negative bacteremia

  • Number of patients with infections caused by bacteria resistant to the administered antibiotic regimen: methicillin-resistant staphylococci; other

Intervention characteristics
  • Antibiotics type and dose

  • Treatment duration

  • Treatment modifications

  • Previous antibiotic regimen for the first modification trials

Measures of outcome

Measures of outcome as defined under 'Types of outcome measures', extracted as number of patients per group

Assessment of risk of bias in included studies

The risk of bias of the included trials was assessed for allocation sequence, allocation concealment, blinding and incomplete outcome data. Studies with a dropout rate above 30% were excluded, unless an ITT analysis was possible for any outcome. Risk of bias assessment was performed independently by two review authors (MP, SB). Risk of bias assessment was based on the evidence of a strong association between poor allocation concealment and overestimation of effect, and was defined by A (low risk of bias; adequate allocation concealment), B (moderate risk of bias; unclear allocation concealment), and C (high risk of bias; inadequate allocation concealment) (Schulz 1995).

Assessment of heterogeneity

Heterogeneity in the results of the trials was assessed using a Chi2 test of heterogeneity (P less than 0.1) and the I2 statistic. We planned to explore heterogeneity by performing the following subgroup analyses: (1) patients with infections due to Gram-positive bacteria; (2) patients with central venous catheters; (3) patients having received quinolone prophylaxis. Available data permitted subgroup analysis only for patients with infections due to Gram-positive bacteria.

Assessment of reporting biases

A funnel plot of log odds ratio (OR) for efficacy against the sample size was examined in order to assess potential selection bias (publication and language). In addition, the standard normal deviate (SND), defined as the OR divided by its standard error, was regressed against the estimate's precision (regression equation: SND = a + b x precision) in order to summarise any potential selection bias (Egger 1997). In this equation, the SND reflects the degree of funnel plot asymmetry as measured by the intercept from regression of standard normal deviates against precision.

Data synthesis

Dichotomous data were analysed by calculating the risk ratio (RR) for each trial, with the uncertainty of each result being expressed using the 95% confidence interval (CI). We planned to extract time to event data for hospitalisation, fever and treatment durations according to the method described by Parmar (Parmar 1998). Meta-analysis was performed using the fixed effect model for comparisons showing no substantial heterogeneity (I2 less than 50%) and the random-effects model for other comparisons. The effect of risk of bias on results was examined using sensitivity analysis restricting the analysis to trials at low risk of bias for allocation concealment and reporting results by ITT.

Meta-regression using STATA was performed to assess the relationship between the rate of Gram-positive infections in the studies and their estimated treatment effects, in order to assess the hypothesis that antiGP treatment would appear more effective with increasing prevalence of gram-positive infections.

Results

Description of studies

Results of the search

The search strategy resulted in 331 references. After reviewing all abstracts, 42 studies were retrieved for full-text inspection. We updated the search in 2013 and 3515 new references were screened. No new trials were identified for inclusion since the first version of the review, published in 2005.

Included studies

Twenty Included publications represented 13 individual RCTs corresponding to our inclusion criteria.

Glycopeptides were tested in nine trials (vancomycin five, teicoplanin four) (Characteristics of included studies). These trials were performed between 1984 and 2000. Other antiGram-positive drugs were tested in four trials: cephalothin in two (Lawson 1979; Verhagen 1987), flucloxacillin and trimethoprim-sulphamethoxazole in one trial each (De Pauw 1985; Menichetti 1986, respectively). These trials were conducted between 1979 and 1986. The basic antibiotic regimens are specified in the table 'Characteristics of included studies'. Ceftazidime was used alone in five trials and with amikacin in two.

The antiGP antibiotic was tested at the onset of antibiotic treatment as the first line (empirical) regimen in all trials but two, which assessed its addition for persistently febrile patients (first modification) after 72 to 96 hours of imipenem monotherapy (Erjavec 2000) or after 48 to 60 hours of piperacillin-tazobactam monotherapy (Cometta 2003). We did not identify studies assessing pre-emptive antiGP treatment, such as for patients with suspected catheter-related infections.

Eight trials randomised 1603 patients. Five studies allowed patient re-entry for separate neutropenic febrile episodes, thus randomising episodes instead of patients. Three trials included 352 episodes representing 292 patients (Del Favero 1987; Erjavec 2000; Menichetti 1986), and two trials included 437 episodes without specifying the number of patients (Lawson 1979; Marie 1991). Overall, 2392 febrile episodes were included and 2159 were evaluated.

All trials included patients with hematological malignancies except one trial that was restricted to patients with solid tumours (Molina 1993). Two trials did not specify patients' age. In the remaining, children less than 16 years were included in six trials, and the mean age in these trials ranged between 38 to 48 years. With regard to exclusion of patients at risk for infections due to Gram-positive bacteria, the two first modification trials excluded patients with documented or suspected catheter-related infections (Cometta 2003; Erjavec 2000). Two empirical studies excluded patients with a documented focus of infection (Marie 1991; Novakova 1991), of which one also excluded patients in septic shock (Marie 1991). Otherwise, no such restrictions were imposed on patient inclusion.

The rate of single-agent Gram-positive bacteremia varied between 6% and 28% (Table 1). It did not correlate with study year as expected, possibly due to the differing locations and inclusion criteria of the studies included in the review.

Table 1. Study year - gram-positive bacteremia correlation
Study IDStudy yearGP bacteremia
Empirical  
Menichetti 1986198313%
Del favero 1987198419%
Karp 1986198428%
Nivakova 1991198725%
Ramphal 1992198825%
EORTC 1991198818%
Molina 199319926%
Semi-empirical  
Cometta 2003200011%

Excluded studies

Twenty-one studies (23 publications) were excluded. Two studies were excluded on account of a high percentage of dropouts. An EORTC trial randomised 841 patients and evaluated 419 patients (EORTC 1983). Martino et al reported outcomes for a 10-month period and 158 patients of a trial which was conducted for 15 months and included 232 patients (Martino 1992). The reasons for exclusion of the remaining studies are listed in the Characteristics of excluded studies table.

Risk of bias in included studies

Results are summarised in Figure 1 and Figure 2 and detailed per study in Characteristics of included studies. Generation of the randomisation sequence was described as low risk in 11 trials. Allocation concealment was described as low risk in eight trials, and four additional trials used sealed envelopes that were not described as opaque (classified as unclear). The two first modification trials were double blinded (Cometta 2003; Erjavec 2000), as was a single empirical trial (Karp 1986). All remaining trials were open label.

Figure 1.

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Figure 2.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Full ITT analysis for failure and mortality was reported in three trials (Cometta 2003; De Pauw 1985; Verhagen 1987) and for mortality alone in two (Menichetti 1986; Novakova 1991). Four additional trials provided the number of patients excluded from each study arm, allowing an ITT analysis by imputing failure for dropouts (Del Favero 1987; Erjavec 2000; Karp 1986; Novakova 1991).

Five trials permitted patient re-inclusion, referring to episodes or infections instead of individual patients, as stated above ('Other bias'). Results per patient were unavailable from the publication even when the number of included patients was known. Results from these trials were analysed together with the remaining trials.

The patient's consent was reported in seven trials and approval of the ethics committee in three, all of which required patient consent. Eights trials reported funding by industry, while no external sources of funding were stated in the other trials.

Effects of interventions

Mortality

Seven studies, including 852 participants, reported overall mortality. The adjusted mean mortality rate in these studies was 9.4%. The RR for death was 0.82 (95% CI 0.56 to 1.20, Analysis 1.1), values lower than 1 favouring the antiGP arm. Two trials used a glycopeptide empirically, two used a glycopeptide semi-empirically, and three used another antiGP antibiotic empirically. No difference in mortality was seen in each of these groups. Considering only studies with low-risk allocation concealment, or those reporting mortality by ITT results, the results were similar, below 1 but not significant (Analysis 1.2; Analysis 1.3). Overall, no heterogeneity was seen with this comparison, which was performed using the fixed-effect model (I2 = 0%).

Seven trials compared infection-related fatality, two of which did not report overall mortality. The RR was 1.18 (95% CI 0.72 to 1.92, Analysis 1.4).

Data regarding mortality among patient subgroups were scarce. Only four studies were included in the comparison for patients in whom a Gram-positive infection was documented (Analysis 1.5). Only nine deaths were recorded; hence although overall mortality among patients receiving antiGP treatment was twice that in the control group this difference was not statistically significant. We did observe a relationship between the rate of single Gram-positive bacteremia in the study (Table 1) and the RR for mortality.

Inspecting the forest plot did not reveal a small study effect (Figure 3).

Figure 3.

Failure.

Treatment failure

Overall failure, disregarding treatment modifications, was assessed in six studies including 943 participants and was similar in both study arms (RR 1.00, 95% CI 0.79 to 1.27, Analysis 2.5). When modifications were counted as causes for treatment failure, a significant advantage in favour of empirical antiGP treatment was evident (RR 0.76, 95% CI 0.68 to 0.85, 10 trials, 1779 participants, Analysis 2.1). The advantage originated from studies that assessed the initial empirical administration of antiGP antibiotics. Considering only studies with adequate allocation concealment, or those permitting analysis by ITT, the effect estimate for treatment including modifications was reduced (RR 0.92 and 0.94, respectively) and no longer statistically significant (Analysis 2.2; Analysis 2.3). The funnel plot for failure was centered approximately symmetrically around the effect estimate (Figure 4).

Figure 4.

Mortality.

Duration of fever was not reported comparatively, but three studies compared the number of persistently febrile patients at 72 hours after the initiation of empirical antibiotic treatment (Analysis 2.6). A non-significant advantage to the antiGP arm was seen but the number of patients evaluated was small (312 patients). Amphotericin was added more frequently to the control arm but the difference was not statistically significant (RR 1.23, 95% CI 0.84 to 1.80, 5 studies, 1201 participants, Analysis 2.7).

Substantial heterogeneity was seen in the comparisons of persistent fever and addition of amphotericin, which were analysed using the random-effects model. Some heterogeneity was detected in the comparison of failure with modifications, conducted with the fixed-effect model (I2 = 39.5%). The advantage with the antiGP arm remained significant when the analysis was repeated using a random-effects model (RR 0.79, 95% CI 0.68 to 0.92).

Only four studies were included in the subgroup assessment of patients with infections caused by Gram-positive bacteria (Analysis 2.4). Although the proportion of patients experiencing treatment failure was lower among those receiving antiGP therapy, the effect was not statistically significant. Meta-regression demonstrated no association between the rate of single Gram-positive bacteremia in the studies and their relative risk for failure, ratio of odds ratios = 0.98 (95% CI 0.87 to 1.1). No data were available for the subgroup of patients with central catheters or those having received quinolone prophylaxis. Duration of hospital stay and removal of the central catheter were not assessed as outcomes in these studies.

Superinfections and adverse events

AntiGP treatment did not increase superinfection rates (Analysis 3.1; Analysis 3.2; Analysis 3.3). The rate of fungal superinfections was similar (RR 1.04, 95% CI 0.60 to 1.82). We observed a significant decrease in bacterial superinfections (RR 0.48, 95% CI 0.28 to 0.83), specifically Gram-positive superinfections (RR 0.31, 95% CI 0.16 to 0.59) in the antiGP arm (Analysis 3.4). No study assessed the effect of additional antiGP treatment on the rate of colonisation with resistant microorganisms or development of resistance.

Adverse events were significantly more frequent in the antiGP arm (Analysis 4.1). However, this originated mainly from a difference in skin reactions (Analysis 4.2) rather than in adverse events incurring significant morbidity. Nephrotoxicity did not differ significantly between the study groups (Analysis 4.3).

Other outcomes

Duration of hospital stay was inconsistently reported and summarised heterogeneously, as means or medians without appropriate CIs, in the included trials. Thus results could not be combined. Duration of fever and removal of central catheters were not reported in the studies.

Discussion

We show that the current evidence does not point to a reduction in the risk of death with the empirical addition of antiGP antibiotics. Eleven studies assessed their addition to the initial antibiotic regimen among non-selected febrile neutropenic patients. Two studies assessed their addition at the time of first modification. Both combined and separately, these studies show that there is no significant difference in overall 30-day patient mortality. The relative risk for mortality favoured the empirical addition of antiGP antibiotics, but the difference was not statistically significant (RR 0.82, 95% CI 0.56 to 1.20). The advantage in the antiGP arm originated from older studies that did not use glycopeptides. The last trial included in this review was published in 2003 and updates of the review since then have found no new trials examining the effects of empirical antiGP treatment for febrile neutropenia.

Failure of the empirical antibiotic regimen, denoting mainly the need to add or change antibiotic therapy, was more common in the control arm. An ITT analysis, including studies conducted by ITT and imputing failure to dropouts in other studies, did not show this advantage in the antiGP arm. Most studies included in the review were conducted at the time when practice guidelines suggested the addition of empirical antiGP treatment on day three to five for persistent fever (Hughes 1990). Similarly, amphotericin was added more frequently to the control arm, but the difference was not statistically significant. Current guidelines advise addition of antifungal therapy on day five to seven for persistent fever with neutropenia (Freifeld 2011). Thus, treatment modifications and the addition of amphotericin may represent persistence of fever regardless of the incidence of uncontrolled infection or fungal infections. Overall failure, whether or not antibiotic treatment was modified, was equal in both study arms.

Adverse events were more common in the antiGP arm as expected, but the difference was in minor adverse events. The most feared adverse outcome of adding an antibiotic, especially a glycopeptide, is the induction of resistance. Studies conducted in other settings have shown that excessive use of glycopeptides is associated with increased rates of vancomycin-resistant enterococci and, lately, vancomycin-resistant staphylococci (CDC 1995). Studies included in this review assessed superinfection rates and these did not increase in the antiGP arm. Rather, Gram-positive bacterial superinfections were reduced in the treatment arm, possibly reflecting reduced detection of these infections in the presence of antiGP antibiotic treatment. However, there are no data on colonisation from these studies. Therefore, we do not know whether patients treated with glycopeptides were more likely to carry resistant Gram-positive bacteria, an important factor when considering future infections and the environment. The assessment of resistance induction may require a longer timescale than possible in randomised trials.

Several limitations of our analysis should be noted. Firstly, all-cause mortality was reported only in seven of 13 included studies. We contacted the authors of the six studies with missing mortality data, of which four replied that the data could no longer be retrieved. Secondly, the definition for treatment failure varied between studies such that we could not combine all studies to assess treatment failure, with or without treatment modifications. We have encountered methodological deficiencies in included studies, which we could not correct for in the meta-analysis. Randomising patients more than once creates episode clusters in which individual outcomes are not independent. Since data could not be extracted only for the first episode of each included patient we could not enter the data correctly for the analysis. ITT analysis was frequently missing and could not be reproduced since the number of patients excluded from each study arm was not consistently reported. While the handling of loss to follow-up with regard to measurable outcomes must entail some assumptions (carry-over, imputations, etc.), all randomised patients can be included in the all-cause mortality comparison. Adequate randomisation should ensure that deaths unrelated to infection are equally distributed. Finally, our results pertain to patients with uncomplicated low or high-risk febrile neutropenia, that is patients presenting without specific risk factors such as catheter-related infection, skin or soft-tissue infection, pneumonia, or hemodynamic instability for whom empirical glycopeptide treatment is recommended (Freifeld 2011).

Authors' conclusions

Implications for practice

Our conclusions are in accordance with current practice guidelines (Freifeld 2011). Non-selective empirical use of glycopeptides, initially or for persistent fever, is discouraged. Data from these studies cannot aid in the selection of patient subgroups for whom an advantage does exist.

Implications for research

The RR for mortality favoured empirical antiGP treatment but did not reach statistical significance, with wide confidence intervals. This raises the question whether further studies are needed to obtain a more precise effect estimate or narrow the confidence interval. The more recent studies assessing empirical glycopeptides did not show a reduction in mortality, but the numbers of included patients and events were small. Thus further trials may be justified, especially if the prevalence of resistant Gram-positive infections increases further. Only two trials (Cometta 2003; Erjavec 2000) tested the addition of a glycopeptide for fever persisting more than three days. These studies do not show an advantage with regard to mortality or failure. However, the confidence intervals are again wide as the studies included only 22 deaths, and further trials assessing the empirical addition of glycopeptides or newer agents against multi-drug resistant Gram-positive bacteria for persistent fever may be warranted.

Further research should focus on risk factors defining specific patient groups who will benefit from the addition of glycopeptides prior to microbiological documentation of these infections.

Our analysis highlights the pitfalls of assessing treatment failure in these and similar studies. Results are dependent on the definition of failure. In most studies failure was defined as a change in the empirical antibiotic regimen, an outcome that is not necessarily associated with patient morbidity. Survival is the ultimate goal of chemotherapy in cancer patients. Usually not chosen as a primary outcome due to the sample size calculation considerations, all-cause mortality should be reported in all these studies. We also suggest the addition of number of hospital days within the study follow-up, for all patients and for those surviving the infectious episode, as an objective measure of patient morbidity.

We showed that the use of glycopeptides was associated with fewer Gram-positive superinfections. However, we do not rule out the possibility of resistance induced by their use by this as the trials did not assess the rates of colonisation with resistant microorganisms. Future studies must incorporate methods for surveillance of colonisation to correctly represent the effects of glycopeptide use on future infections and the environment.

All future studies should adhere to better methodological standards (Consort statement). Specifically, patients should be included in the study only once, data regarding overall mortality should be reported by ITT, and the number of exclusions after randomisation for all other outcomes should be reported per study arm.

Acknowledgements

We thank the authors who have responded to our request for further data (Ben E De Pauw, Judith E Karp, Gerald P Bodey). We would like to express our appreciation to Heather Dickinson for her thoughtful and educative review of our work. We thank the Cochrane Gynaecological Cancer Review Group for their support throughout the review.

We would like to thank Abigail Fraser and Michal Cohen who contributed to the first version of the review.

The National Institute for Health Research (NIHR) is the largest single funder of the Cochrane Gynaecological Cancer Group. The views and opinions expressed herein are those of the authors and do not necessarily reflect those of the NIHR, NHS or the Department of Health.

Data and analyses

Download statistical data

Comparison 1. Mortality
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Overall mortality7852Risk Ratio (M-H, Fixed, 95% CI)0.82 [0.56, 1.20]
1.1 Glycopeptide empirical2247Risk Ratio (M-H, Fixed, 95% CI)0.93 [0.47, 1.84]
1.2 Glycopeptide first modification2279Risk Ratio (M-H, Fixed, 95% CI)0.81 [0.36, 1.80]
1.3 Other antiGP empirical3326Risk Ratio (M-H, Fixed, 95% CI)0.76 [0.44, 1.33]
2 Overall mortality (adequate allocation concealment)6728Risk Ratio (M-H, Fixed, 95% CI)0.88 [0.55, 1.39]
2.1 Glycopeptide empirical2247Risk Ratio (M-H, Fixed, 95% CI)0.93 [0.47, 1.84]
2.2 Glycopeptide first modification2279Risk Ratio (M-H, Fixed, 95% CI)0.81 [0.36, 1.80]
2.3 Other antiGP empirical2202Risk Ratio (M-H, Fixed, 95% CI)0.89 [0.34, 2.36]
3 Overall mortality (intention to treat)5611Risk Ratio (M-H, Fixed, 95% CI)0.71 [0.46, 1.09]
3.1 Glycopeptide empirical1120Risk Ratio (M-H, Fixed, 95% CI)0.78 [0.31, 1.95]
3.2 Glycopeptide first modification1165Risk Ratio (M-H, Fixed, 95% CI)0.46 [0.14, 1.47]
3.3 Other antiGP empirical3326Risk Ratio (M-H, Fixed, 95% CI)0.76 [0.44, 1.33]
4 Infection-related fatality71420Risk Ratio (M-H, Fixed, 95% CI)1.18 [0.72, 1.92]
4.1 Glycopeptide empirical41030Risk Ratio (M-H, Fixed, 95% CI)1.16 [0.62, 2.17]
4.2 Glycopeptide first modification2279Risk Ratio (M-H, Fixed, 95% CI)1.97 [0.51, 7.59]
4.3 Other antiGP1111Risk Ratio (M-H, Fixed, 95% CI)0.88 [0.33, 2.35]
5 Mortality in Gram-positive infections4107Risk Ratio (M-H, Fixed, 95% CI)2.15 [0.56, 8.25]
Analysis 1.1.

Comparison 1 Mortality, Outcome 1 Overall mortality.

Analysis 1.2.

Comparison 1 Mortality, Outcome 2 Overall mortality (adequate allocation concealment).

Analysis 1.3.

Comparison 1 Mortality, Outcome 3 Overall mortality (intention to treat).

Analysis 1.4.

Comparison 1 Mortality, Outcome 4 Infection-related fatality.

Analysis 1.5.

Comparison 1 Mortality, Outcome 5 Mortality in Gram-positive infections.

Comparison 2. Treatment failure
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Failure, modifications included101779Risk Ratio (M-H, Fixed, 95% CI)0.76 [0.68, 0.85]
1.1 Glycopeptide empirical51178Risk Ratio (M-H, Fixed, 95% CI)0.70 [0.61, 0.80]
1.2 Glycopeptide first modification2279Risk Ratio (M-H, Fixed, 95% CI)0.98 [0.79, 1.22]
1.3 Other antiGP empirical3322Risk Ratio (M-H, Fixed, 95% CI)0.74 [0.50, 1.08]
2 Failure, modifications included (adequate allocation concealment)6711Risk Ratio (M-H, Fixed, 95% CI)0.92 [0.78, 1.09]
2.1 Glycopeptide empirical2230Risk Ratio (M-H, Fixed, 95% CI)0.80 [0.59, 1.08]
2.2 Glycopeptide first modification2279Risk Ratio (M-H, Fixed, 95% CI)0.98 [0.79, 1.22]
2.3 Other antiGP empirical2202Risk Ratio (M-H, Fixed, 95% CI)1.02 [0.62, 1.67]
3 Failure, modifications included (intention to treat)6678Risk Ratio (M-H, Fixed, 95% CI)0.94 [0.80, 1.10]
3.1 Glycopeptide empirical2186Risk Ratio (M-H, Fixed, 95% CI)0.81 [0.61, 1.08]
3.2 Glycopeptide first modification2290Risk Ratio (M-H, Fixed, 95% CI)1.00 [0.81, 1.23]
3.3 Other antiGP empirical2202Risk Ratio (M-H, Fixed, 95% CI)1.02 [0.62, 1.67]
4 Failure in Gram-positive infections487Risk Ratio (M-H, Fixed, 95% CI)0.79 [0.42, 1.48]
5 Overall failure7943Risk Ratio (M-H, Fixed, 95% CI)1.00 [0.79, 1.27]
5.1 Glycopeptide empirical3293Risk Ratio (M-H, Fixed, 95% CI)1.09 [0.28, 4.20]
5.2 Glycopeptide first modification1165Risk Ratio (M-H, Fixed, 95% CI)0.61 [0.18, 2.09]
5.3 Other antiGP empirical3485Risk Ratio (M-H, Fixed, 95% CI)1.03 [0.81, 1.32]
6 Febrile at 72 hrs. on empirical Tx3312Risk Ratio (M-H, Random, 95% CI)0.72 [0.44, 1.17]
6.1 Glycopeptide empirical3312Risk Ratio (M-H, Random, 95% CI)0.72 [0.44, 1.17]
7 Addition of amphotericin51201Risk Ratio (M-H, Random, 95% CI)1.23 [0.84, 1.80]
7.1 Non-blinded3976Risk Ratio (M-H, Random, 95% CI)1.51 [0.80, 2.83]
7.2 Double blind2225Risk Ratio (M-H, Random, 95% CI)0.99 [0.75, 1.33]
Analysis 2.1.

Comparison 2 Treatment failure, Outcome 1 Failure, modifications included.

Analysis 2.2.

Comparison 2 Treatment failure, Outcome 2 Failure, modifications included (adequate allocation concealment).

Analysis 2.3.

Comparison 2 Treatment failure, Outcome 3 Failure, modifications included (intention to treat).

Analysis 2.4.

Comparison 2 Treatment failure, Outcome 4 Failure in Gram-positive infections.

Analysis 2.5.

Comparison 2 Treatment failure, Outcome 5 Overall failure.

Analysis 2.6.

Comparison 2 Treatment failure, Outcome 6 Febrile at 72 hrs. on empirical Tx.

Analysis 2.7.

Comparison 2 Treatment failure, Outcome 7 Addition of amphotericin.

Comparison 3. Superinfections
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Any superinfections101896Risk Ratio (M-H, Fixed, 95% CI)0.84 [0.66, 1.08]
1.1 Glycopeptide61281Risk Ratio (M-H, Fixed, 95% CI)0.87 [0.67, 1.13]
1.2 Other antiGP4615Risk Ratio (M-H, Fixed, 95% CI)0.73 [0.38, 1.40]
2 Bacterial superinfections81628Risk Ratio (M-H, Fixed, 95% CI)0.41 [0.27, 0.62]
2.1 Glycopeptide51296Risk Ratio (M-H, Fixed, 95% CI)0.38 [0.24, 0.59]
2.2 Other antiGP3332Risk Ratio (M-H, Fixed, 95% CI)0.67 [0.24, 1.89]
3 Fungal superinfections91637Risk Ratio (M-H, Fixed, 95% CI)1.10 [0.69, 1.77]
3.1 Glycopeptide61305Risk Ratio (M-H, Fixed, 95% CI)1.18 [0.70, 1.99]
3.2 Other antiGP3332Risk Ratio (M-H, Fixed, 95% CI)0.82 [0.27, 2.48]
4 Gram-positive superinfections91688Risk Ratio (M-H, Fixed, 95% CI)0.24 [0.14, 0.40]
4.1 Glycopeptide61356Risk Ratio (M-H, Fixed, 95% CI)0.21 [0.11, 0.37]
4.2 Other antiGP3332Risk Ratio (M-H, Fixed, 95% CI)0.46 [0.15, 1.45]
Analysis 3.1.

Comparison 3 Superinfections, Outcome 1 Any superinfections.

Analysis 3.2.

Comparison 3 Superinfections, Outcome 2 Bacterial superinfections.

Analysis 3.3.

Comparison 3 Superinfections, Outcome 3 Fungal superinfections.

Analysis 3.4.

Comparison 3 Superinfections, Outcome 4 Gram-positive superinfections.

Comparison 4. Adverse events
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Any adverse events81546Risk Ratio (M-H, Fixed, 95% CI)1.79 [1.55, 2.08]
1.1 Glycopeptide51195Risk Ratio (M-H, Fixed, 95% CI)1.78 [1.53, 2.07]
1.2 Other antiGP3351Risk Ratio (M-H, Fixed, 95% CI)1.88 [1.11, 3.20]
2 Rash/ allergy71526Risk Ratio (M-H, Fixed, 95% CI)2.31 [1.47, 3.63]
2.1 Glycopeptide51336Risk Ratio (M-H, Fixed, 95% CI)2.33 [1.43, 3.80]
2.2 Other antiGP2190Risk Ratio (M-H, Fixed, 95% CI)2.17 [0.65, 7.30]
3 Any nephrotoxicity101916Risk Ratio (M-H, Fixed, 95% CI)1.28 [0.96, 1.70]
3.1 Glycopeptide61282Risk Ratio (M-H, Fixed, 95% CI)1.43 [1.06, 1.94]
3.2 Other antiGP4634Risk Ratio (M-H, Fixed, 95% CI)0.84 [0.40, 1.75]
Analysis 4.1.

Comparison 4 Adverse events, Outcome 1 Any adverse events.

Analysis 4.2.

Comparison 4 Adverse events, Outcome 2 Rash/ allergy.

Analysis 4.3.

Comparison 4 Adverse events, Outcome 3 Any nephrotoxicity.

Appendices

Appendix 1. CENTRAL search strategy

#1     MeSH descriptor Neoplasms explode all trees
#2     (neoplasm* or tumor* or tumour* or cancer* or malignan* or carcinoma* or adenocarcinoma* or lymphoma* or leukem* or luekaem*)
#3     (#1 OR #2)
#4     MeSH descriptor Neutropenia, this term only
#5     neutropeni*
#6     (granulop?en* or granulocytop?en*)
#7     (immunosuppress* or immuno-suppress*)
#8     (#4 OR #5 OR #6 OR #7)
#9     MeSH descriptor Gram-Positive Bacterial Infections explode all trees with qualifier: DT
#10   MeSH descriptor Anti-Bacterial Agents explode all trees
#11   (antibiotic* or anti-bacterial or antibacterial)
#12   (vancomycin or teicoplanin or oxacillin or cloxacillin or dicloxacillin or flucloxacillin or nafcillin or cefazolin or clindamycin or quinupristin-dalfopristin or linezolid or trimethoprim-sulphamethoxazole or daptomycin or tigecycline)
#13   (#9 OR #10 OR #11 OR #12)
#14   (#3 AND #8 AND #13)

Appendix 2. MEDLINE search strategy

1   exp Neoplasms/
2   (neoplasm* or tumor* or tumour* or cancer* or malignan* or carcinoma* or adenocarcinoma* or lymphoma* or leukem* or luekaem*).mp.
3   1 or 2
4   Neutropenia/
5   neutropeni*.mp.
6   (granulop?en* or granulocytop?en*).mp.
7   (immunosuppress* or immuno-suppress*).mp.
8   4 or 5 or 6 or 7
9   exp Gram-Positive Bacterial Infections/dt [Drug Therapy]
10  exp Anti-Bacterial Agents/
11  (antibiotic* or anti-bacterial or antibacterial).mp.
12  (vancomycin or teicoplanin or oxacillin or cloxacillin or dicloxacillin or flucloxacillin or nafcillin or cefazolin or clindamycin or quinupristin-dalfopristin or linezolid or trimethoprim-sulphamethoxazole or daptomycin or tigecycline).mp.
13  9 or 10 or 11 or 12
14  3 and 8 and 13
15  randomized controlled trial.pt.
16  controlled clinical trial.pt.
17  randomized.ab.
18  placebo.ab.
19  clinical trials as topic.sh.
20  randomly.ab.
21  trial.ti.
22  15 or 16 or 17 or 18 or 19 or 20 or 21
23  14 and 22 

key:
mp=title, abstract, original title, name of substance word, subject heading word, protocol supplementary concept, rare disease supplementary concept, unique identifier
pt=publication type
ab=abstract
sh=subject heading
ti=title

Appendix 3. EMBASE search strategy

1   exp neoplasm/
2   (neoplasm* or tumor* or tumour* or cancer* or malignan* or carcinoma* or adenocarcinoma* or lymphoma* or leukem* or luekaem*).mp.
3   1 or 2
4   exp neutropenia/
5   neutropeni*.mp.
6   (granulop?en* or granulocytop?en*).mp.
7   (immunosuppress* or immuno-suppress*).mp.
8   4 or 5 or 6 or 7
9   Gram positive infection/dt [Drug Therapy]
10  exp antibiotic agent/
11  (antibiotic* or anti-bacterial or antibacterial).mp.
12  (vancomycin or teicoplanin or oxacillin or cloxacillin or dicloxacillin or flucloxacillin or nafcillin or cefazolin or clindamycin or quinupristin-dalfopristin or linezolid or trimethoprim-sulphamethoxazole or daptomycin or tigecycline).mp.
13  9 or 10 or 11 or 12
14  3 and 8 and 13
15  crossover procedure/
16  double-blind procedure/
17  randomized controlled trial/
18  single-blind procedure/
19  random*.mp.
20  factorial*.mp.
21  (crossover* or cross over* or cross-over*).mp.
22  placebo*.mp.
23  (double* adj blind*).mp.
24  (singl* adj blind*).mp.
25  assign*.mp.
26  allocat*.mp.
27  volunteer*.mp.
28  15 or 16 or 17 or 18 or 19 or 20 or 21 or 22 or 23 or 24 or 25 or 26 or 27
29  14 and 28

key
mp=title, abstract, subject headings, heading word, drug trade name, original title, device manufacturer, drug manufacturer, device trade name, keyword

What's new

DateEventDescription
11 February 2015AmendedContact details updated.

History

Protocol first published: Issue 4, 2002
Review first published: Issue 3, 2005

DateEventDescription
26 February 2014AmendedContact details updated.
2 December 2013AmendedTitle amended as a result of recent feedback.
16 August 2013New search has been performedText updated and new search dates added. Two review authors removed (Abigail Fraser and Michal Cohen) and one new author (Yaakov Dickstein) added.
10 August 2013New citation required but conclusions have not changedSearch updated, no new studies identified for inclusion.
8 November 2007New search has been performedMinor update
15 August 2007New search has been performedNew studies sought but none found. We updated the search in August 2007 and no new studies were found. We added new anti-Gram positive antibiotics to included interventions.
17 April 2005New citation required and conclusions have changedSubstantive amendment

Contributions of authors

Mical Paul (contact reviewer) - reference search, article retrieval, study inclusion and exclusion, data extraction, analysis, and writing.
Yaakov Dickstein - 2013 review update, writing.
Sara Borok - reference search, article retrieval, study inclusion and exclusion, data extraction, analysis, and review.
Liat Vidal - analysis and review.
Leonard Leibovici - reference search, study inclusion and exclusion, data extraction, analysis, writing and review.

Declarations of interest

None

Sources of support

Internal sources

  • Rabin Medical Center, Beilinson Campus, Israel.

  • Tel-Aviv University, Sackler Faculty of Medicine, Israel.

External sources

  • EU 5th Framework - TREAT project (grant number: 1999-11459), Not specified.

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Cometta 2003

MethodsRandomised controlled trial, double blinded
ParticipantsPatients with neutropenia <1000/mm3 anticipated to fall to 500/mm3 and fever ≥38.5 or >38 in two measurements
Mean age 42(4-78) with all types of cancer
InterventionsVancomycin
15mg/kg x2 versus placebo for persistent fever at 48-60 hrs
Non-intervention antibiotics: piperacillin-tazobactam
OutcomesFailure
Mortality (all-cause and infection-related)
Superinfections
Adverse events
NotesFirst modification design
Multicenter - Europe, Middle East, North America
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskComputer generated randomisation. Randomisation was dynamically performed after the application of a randomisation algorithm, which used the minimisation technique of a global imbalances function between the 2 treatment arms, with the following 3 stratification variables: name and location of study center, infection documentation at randomisation, and underlying disease
Allocation concealment (selection bias)Low riskCentral randomisation
Blinding of participants and personnel (performance bias)
All outcomes
Low riskDouble blind
Blinding of outcome assessment (detection bias)
All outcomes
Low risk 
Incomplete outcome data (attrition bias)
All outcomes
Low riskAll patients included in mortality and failure analyses
Other biasUnclear riskPatients randomised only once

De Pauw 1985

MethodsRandomised controlled trial, open label
ParticipantsPatients with neutropenia <1000/mm3 and fever >38.5 in two measurements, associated with chills. Mean age 34 (16-75), with any type of cancer
InterventionsFlucloxacillin 2grx4 added to one arm
Non-intervention antibiotics: ceftazidime
OutcomesFailure
Superinfections
Mortality (all-cause)
Adverse events
NotesEmpirical design
Single center, open, Netherlands.
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskComputer generated randomisation
Allocation concealment (selection bias)Low riskConsecutive opaque and sealed envelopes were opened
Blinding of participants and personnel (performance bias)
All outcomes
High riskOpen
Blinding of outcome assessment (detection bias)
All outcomes
High riskOpen
Incomplete outcome data (attrition bias)
All outcomes
Low riskAll patients included in mortality and failure analyses
Other biasLow riskPatients randomised only once

Del Favero 1987

MethodsOpen.
ParticipantsPatients with neutropenia <1000/mm3 and fever >38. Mean age 39 (8-71), with acute leukemia
InterventionsTeicoplanin 5mg/kgx1 added to one arm
Non-intervention antibiotics: ceftazidime+amikacin
OutcomesFailure
Adverse events
NotesEmpirical design
Single center, Italy
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskRandom permuted blocks
Allocation concealment (selection bias)Unclear riskConsecutive, sealed envelopes (opacity not mentioned)
Blinding of participants and personnel (performance bias)
All outcomes
High riskOpen
Blinding of outcome assessment (detection bias)
All outcomes
High riskOpen
Incomplete outcome data (attrition bias)
All outcomes
High risk 
Other biasHigh riskPatients included for different episodes and analysis by episode

EORTC 1991

MethodsRandomised controlled trial, open label
ParticipantsPatients with neutropenia <1000/mm3 and fever>38. Mean age 38 (1-88), with any type of cancer
InterventionsVancomycin 500mgX4 added to one arm
Non-intervention antibiotics: ceftazidime+amikacin
OutcomesFailure
Mortality (infection-related only)
Superinfections
Adverse events
NotesEmpirical design
Multicenter, Canada
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskRandomised sequence stratified by groups of six patients
Allocation concealment (selection bias)Unclear riskConsecutive sealed envelopes (opacity not mentioned)
Blinding of participants and personnel (performance bias)
All outcomes
High riskOpen
Blinding of outcome assessment (detection bias)
All outcomes
Low riskOutcome assessors blinded
Incomplete outcome data (attrition bias)
All outcomes
High risk 
Other biasLow riskPatients included only once

Erjavec 2000

MethodsRandomised controlled trial, double blind
ParticipantsPatients with neutropenia <500/mm3 or <1000 and decreasing. Fever ≥38
Mean age 48.2 yrs with any type of cancer
InterventionsTeicoplanin 400mgx1 versus placebo for persistent fever at 72-96hrs
Non-intervention antibiotics: imipenem
OutcomesFailure
Mortality (all-cause and infection-related)
Superinfections
NotesFirst modification design
Single center, Netherlands
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskComputer assisted randomisation
Allocation concealment (selection bias)Low riskRandomisation performed by the hospital pharmacy
Blinding of participants and personnel (performance bias)
All outcomes
Low risk 
Blinding of outcome assessment (detection bias)
All outcomes
Low risk 
Incomplete outcome data (attrition bias)
All outcomes
High risk 
Other biasHigh riskPatients included for different episodes and analysis by episode

Karp 1986

MethodsRandomised controlled trial, double blind
ParticipantsPatients with neutropenia
<500/mm3 and fever >38.3. Mean age 40 (19-63) with acute leukemia or post autologous bone marrow rescue transplantation
InterventionsVancomycin 500mgx4
added to one arm
Non-intervention antibiotics: ticarcillin+gentamicin
OutcomesFailure
Superinfections
Adverse events
NotesEmpirical design
Single center, USA
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskRandomisation by table of random numbers
Allocation concealment (selection bias)Low riskRandomisation concealed centrally within the oncology pharmacy
Blinding of participants and personnel (performance bias)
All outcomes
Low risk 
Blinding of outcome assessment (detection bias)
All outcomes
Low risk 
Incomplete outcome data (attrition bias)
All outcomes
High risk 
Other biasLow riskPatients included only once

Lawson 1979

MethodsRandomised controlled trial, open label
ParticipantsPatients with neutropenia <1000/mm3 and fever >38.3
Patients with all types of cancer
InterventionsCephalothin 3grx4 added to one arm
Non-intervention antibiotics: ticarcillin+tobramycin
OutcomesFailure
Superinfections
Adverse events
NotesEmpirical design
Single center, USA
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskComputer generated randomisation
Allocation concealment (selection bias)Low riskRandomisation kept in the pharmacy
Blinding of participants and personnel (performance bias)
All outcomes
High risk 
Blinding of outcome assessment (detection bias)
All outcomes
High risk 
Incomplete outcome data (attrition bias)
All outcomes
High risk 
Other biasHigh riskPatients included for different episodes and analysis by episode

Marie 1991

MethodsRandomised controlled trial, open label
ParticipantsPatients with neutropenia <500/mm3 and fever >38.5 for 3hrs or >38 for 6hrs. Mean age 46yrs, with any type of cancer
InterventionsVancomycin added to one arm
Non-intervention antibiotics: ceftazidime
OutcomesFailure
Superinfections
Adverse events
NotesEmpirical design
Multicenter, France
Only data from protocol one of three consecutive study protocols was extracted
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot stated
Allocation concealment (selection bias)Unclear riskSealed envelopes (opacity not mentioned)
Blinding of participants and personnel (performance bias)
All outcomes
High risk 
Blinding of outcome assessment (detection bias)
All outcomes
High risk 
Incomplete outcome data (attrition bias)
All outcomes
High riskNumber of randomised patients not stated
Other biasHigh riskPatients included for different episodes and analysis by episode

Menichetti 1986

MethodsRandomised controlled trial, single blind
ParticipantsPatients with neutropenia <1000/mm3, fever >38
Mean age 45 (9-82) yrs, with all types of cancer
InterventionsTrimethoprim/sulphamethoxazole 2.5mg/kgx4 (max 640mg per day) added to one arm
Non-intervention antibiotics: piperacillin+amikacin
OutcomesFailure
Mortality (all-cause and infection-related)
Superinfections
Adverse events
NotesEmpirical design
Single center, Italy
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskTable of random numbers
Allocation concealment (selection bias)Unclear riskSealed envelopes (opacity not mentioned)
Blinding of participants and personnel (performance bias)
All outcomes
Unclear riskSingle blind
Blinding of outcome assessment (detection bias)
All outcomes
High risk 
Incomplete outcome data (attrition bias)
All outcomes
High risk 
Other biasHigh riskPatients included for different episodes and analysis by episode

Molina 1993

MethodsRandomised controlled trial, open label
ParticipantsPatients with neutropenia <1000/mm3, fever >38
Solid cancer
InterventionsTeicoplanin 6mg/kg/d added to one arm
Non-intervention antibiotics: piperacillin+amikacin
OutcomesFailure
Mortality (infection-related only)
Superinfections
NotesEmpirical design
Single center, Spain
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot stated
Allocation concealment (selection bias)Unclear riskNot stated
Blinding of participants and personnel (performance bias)
All outcomes
High risk 
Blinding of outcome assessment (detection bias)
All outcomes
High risk 
Incomplete outcome data (attrition bias)
All outcomes
High riskNumber of randomised patients not stated
Other biasLow riskPatients included only once

Novakova 1991

MethodsRandomised controlled trial, open label
ParticipantsPatients with neutropenia <500/mm3, fever>38.3 or >38 in two measurements, without a focus of infection on admission. Mean age 40 (16-69) yrs with all types of cancer
InterventionsTeicoplanin 800mgX2 then 400mgX1 added to GP arm
Non-intervention antibiotics: ceftazidime
OutcomesFailure
Mortality (all-cause and infection-related)
Superinfections
Adverse events
NotesEmpirical design
Single center, Netherlands
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskComputer generated randomisation
Allocation concealment (selection bias)Low riskConsecutive opaque and sealed envelopes
Blinding of participants and personnel (performance bias)
All outcomes
High risk 
Blinding of outcome assessment (detection bias)
All outcomes
High risk 
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskLow risk for mortality, high risk for failure
Other biasHigh riskPatients included for different episodes and analysis by episode

Ramphal 1992

MethodsRandomised controlled trial, open label
ParticipantsPatients with neutropenia <500/mm3 of <1000/mm3 and falling, fever >38.5 or >38 in two measurements. Mean age 41(18-83) yrs, with all types of cancer
InterventionsVancomycin 1grX2 added to one arm
Non-intervention antibiotics: ceftazidime
OutcomesFailure
Mortality (all cause and infection-related)
Superinfections
Adverse events
NotesEmpirical design
Double center, USA
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskRandomisation by a computer random number generator
Allocation concealment (selection bias)Low risk 
Blinding of participants and personnel (performance bias)
All outcomes
High risk 
Blinding of outcome assessment (detection bias)
All outcomes
High risk 
Incomplete outcome data (attrition bias)
All outcomes
Low risk 
Other biasLow riskPatients included only once

Verhagen 1987

MethodsRandomised controlled trial, open label
ParticipantsPatients with neutropenia <1000/mm3 and fever >38.5
Mean age 41(14-78) yrs, with all types of cancer
InterventionsCephalothin 2grx4 added to one arm
Non-intervention antibiotics: ceftazidime
OutcomesFailure
Mortality (all-cause and infection-related)
Superinfections
Adverse events
NotesEmpirical design
Single center, Netherlands
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskComputer generated randomisation
Allocation concealment (selection bias)Low riskConsecutive opaque and sealed envelopes
Blinding of participants and personnel (performance bias)
All outcomes
High risk 
Blinding of outcome assessment (detection bias)
All outcomes
High risk 
Incomplete outcome data (attrition bias)
All outcomes
Low risk 
Other biasLow riskPatients included only once

Characteristics of excluded studies [ordered by study ID]

StudyReason for exclusion
Berger 2002Prospective surveillance study
De Pauw 1997Provides a summary based on the analysis of several trials, results of which appears in other included studies
Dompeling 1996Study includes patients presenting initially with a skin or soft tissue infection, who were assigned non-randomly to an empirical antibiotic regimen which included vancomycin
Elting 1996Study provides a retrospective cohort of 415 neutropenic patients with Gram-positive bacteremia formed from 10 consecutive randomised clinical trials
Elting 1997The study analyses data from 10 consecutive, randomised clinical trials of antibiotic therapy for febrile episodes in neutropenic patients, some of which are included in the review
EORTC 1983Dropout rate of 50%. Study randomised 841 patients, of which 149 were excluded due to protocol violations and 273 were not evaluated because of a doubtful or non-bacterial infection
Fauser 1991Non-comparative study: patients were treated with a cephalosporin, an aminoglycoside and teicoplanin
Granowetter 1988Incompatible comparator antibiotics: ceftazidime versus cephalothin + carbenicillin + gentamicin. Vancomycin was added to ceftazidime treatment arm in the second year of the study as a result of an increase in ceftazidime-resistant gram-positive infections
Jones 1986Incompatible comparator antibiotics: aztreonam + vancomycin versus aztreonam + vancomycin + amikacin versus moxalactam + ticarcillin
Kramer 1986Vancomycin added to ceftazidime regimen at study entry 49 after revealing a preponderance of gram-positive superinfections. Study continued with a 2:1 randomised comparison of ceftazidime + vancomycin versus cephalothin + gentamicin + carbenicillin
Libanore 1991Open prospective study of immunocompromised patients treated with teicoplanin for gram-positive infections
Lim 1990Patients randomised to ceftazidime versus ciprofloxacin, with the addition of teicoplanin to both study arms in cases with clinical suspicion of catheter-associated infection
Link 1994Antibiotic regimens apart from the antiGP antibiotic differed: acylamino penicillin + third generation cephalosporin + vancomycin versus acylamino penicillin + third generation cephalosporin + aminoglycoside
Liu 2000Reference identified in CENTRAL (3rd quarter 2007 search) - full text and abstract not available
Martino 1992Full outcomes are reported for a 10-month period and 158 episodes, of a trial which was conducted for 15 months and included 232 patients and 265 episodes. The original report including all 232 patients describes outcomes for a subgroup of patients with Gram-positive bacteremia
Moroni 1987Antibiotic regimens apart from the antiGP antibiotic differed: ceftazidime + amikacin versus ceftazidime + vancomycin
Novakova 1990Part of study ID Novakova 1991 which was included. The patients treated empirically with teicoplanin were automatically excluded from the present report
Pizzo 1982Incompatible comparator antibiotics: cephalothin, gentamicin and carbenicillin. In the second phase patients were randomised to either continue receiving the same treatment with or without additional amphotericin B or discontinue all treatment
Pizzo 1986Incompatible comparator antibiotics: ceftazidime versus cephalothin + gentamicin + carbenicillin
Rubin 1988The study is a retrospective review of a randomised, prospective study excluded from this review

Ancillary