Beta lactam antibiotic monotherapy versus beta lactam-aminoglycoside antibiotic combination therapy for sepsis

  • Review
  • Intervention

Authors


Abstract

Background

Optimal antibiotic treatment for sepsis is imperative. Combining a beta-lactam antibiotic with an aminoglycoside antibiotic may have certain advantages over beta-lactam monotherapy.

Objectives

We compared clinical outcomes for beta lactam-aminoglycoside combination therapy versus beta lactam monotherapy for sepsis.

Search methods

We searched the Cochrane Central Register of Controlled Trials (CENTRAL), (The Cochrane Library, Issue 3, 2004); MEDLINE (1966 to July 2004); EMBASE (1980 to March 2003); LILACS (1982 to July 2004); and conference proceedings of the Interscience Conference of Antimicrobial Agents and Chemotherapy (1995 to 2003). We scanned citations of all identified studies and contacted all corresponding authors.

Selection criteria

We included randomized and quasi-randomized trials comparing any beta-lactam monotherapy to any combination of one beta-lactam and one aminoglycoside for sepsis.

Data collection and analysis

The primary outcome was all-cause fatality. Secondary outcomes included treatment failure, superinfections, colonization, and adverse events. Two authors independently collected data. We pooled relative risks (RR) with their 95% confidence intervals (CI) using the fixed effect model. We extracted outcomes by intention-to-treat analysis whenever possible.

Main results

We included 64 trials, randomizing 7586 patients. Twenty trials compared the same beta-lactam in both study arms, while the remaining compared different beta-lactams using a broader spectrum beta-lactam in the monotherapy arm. In studies comparing the same beta-lactam, we observed no difference between study groups with regard to all-cause fatality, RR 1.01 (95% CI 0.75 to 1.35) and clinical failure, RR 1.11 (95% CI 0.95 to 1.29). In studies comparing different beta-lactams, we observed an advantage to monotherapy: all cause fatality RR 0.85 (95% CI 0.71 to 1.01), clinical failure RR 0.77 (95% CI 0.69 to 0.86). No significant disparities emerged from subgroup and sensitivity analyses, including the assessment of patients with Gram-negative and Pseudomonas aeruginosa infections. We detected no differences in the rate of resistance development. Adverse events rates did not differ significantly between the study groups overall, although nephrotoxicity was significantly more frequent with combination therapy, RR 0.30 (95% CI 0.23 to 0.39). We found no heterogeneity for all comparisons. We included a small subset of studies addressing patients with Gram-positive infections, mainly endocarditis. We identified no difference between monotherapy and combination therapy in these studies.

Authors' conclusions

The addition of an aminoglycoside to beta-lactams for sepsis should be discouraged. All-cause fatality rates are unchanged. Combination treatment carries a significant risk of nephrotoxicity.

Résumé

Bêta-lactamine en monothérapie antibiotique par rapport à une combinaison de traitements antibiotiques par bêta-lactamine-aminoglycoside pour le sepsis.

Contexte

Un traitement antibiotique optimal contre le sepsis est impératif. La combinaison d'un antibiotique bêta-lactamine avec un antibiotique aminoglycoside peut présenter certains avantages par rapport à la monothérapie par bêta-lactamine.

Objectifs

Nous avons comparé les résultats cliniques pour l'association bêta lactam-aminoglycoside par rapport à la même bêta-lactamine en monothérapie pour le sepsis.

Stratégie de recherche documentaire

Nous avons effectué des recherches dans le registre Cochrane des essais contrôlés (CENTRAL), ( La Bibliothèque Cochrane , numéro 3, 2004) ; MEDLINE (de 1966 à juillet 2004) ; EMBASE (de 1980 à mars 2003) ; LILACS (de 1982 à juillet 2004) ; et les actes de conférence de l'Interscience Conference of Antimicrobial Agents and Chemotherapy (de 1995 à 2003). Nous avons analysé les références bibliographiques de toutes les études identifiées et contacté les auteurs correspondants.

Critères de sélection

Nous avons inclus les essais randomisés et quasi-randomisés comparant toute bêta-lactamine en monothérapie à toute combinaison d'une bêta-lactamine et d'un aminoglycoside pour le sepsis.

Recueil et analyse des données

Le critère de jugement principal était la létalité toutes causes confondues. Les critères de jugement secondaires incluaient l'échec du traitement, les surinfections, la colonisation, et les événements indésirables. Deux auteurs ont indépendamment collecté les données. Nous avons regroupé les risques relatifs (RR) avec leurs intervalles de confiance (IC) à l'aide du modèle à effets fixes. Nous avons extrait les résultats par une analyse en intention de traiter lorsque cela était possible.

Résultats Principaux

Nous avons inclus 64 essais, randomisant 7 586 patients. Vingt essais comparaient la même bêta-lactamine dans les deux bras d'étude, tandis que les autres ont comparé différentes bêta-lactamines utilisant une bêta-lactamine à spectre plus large dans le bras monothérapie. Dans les études comparant la même bêta-lactamine, nous n'avons observé aucune différence entre les groupes d'étude en ce qui concerne la létalité toutes causes confondues, RR de 1,01 (IC à 95 % 0,75 à 1,35) et l'échec clinique, RR de 1,11 (IC à 95 % 0,95 à 1,29). Dans les études comparant différentes bêta-lactamines, nous avons observé un avantage à la monothérapie : Létalité toutes causes RR 0,85 (IC à 95 % 0,71 à 1,01), l'échec clinique RR 0,77 (IC à 95 % 0,69 à 0,86). Aucune disparité significative n'est ressortie aux analyses de sous-groupes et de sensibilité, y compris dans l'évaluation des patients atteints d'infections à Gram négatif et Pseudomonas aeruginosa. Nous n'avons détecté aucune différence dans le taux de développement de résistances. Les taux d'événements indésirables ne différaient pas significativement entre les groupes d'étude dans l'ensemble, bien que la néphrotoxicité était significativement plus fréquente avec le traitement combiné, RR de 0,30 (IC à 95 % 0,23 à 0,39). Nous n'avons trouvé aucune hétérogénéité pour toutes les comparaisons. Nous avons inclus un petit sous-ensemble d'études portant sur les patients atteints d'infections à Gram positif, principalement d'endocardite. Nous n'avons identifié aucune différence entre la monothérapie et la combinaison de traitements dans ces études.

Conclusions des auteurs

L'ajout d'un aminoglycoside aux bêta-lactamines pour le sepsis devrait être déconseillé. Les taux de létalité toutes causes sont inchangés. La combinaison de traitements comporte un risque significatif de néphrotoxicité.

Plain language summary

Single versus combination antibiotic treatment for severe infections: beta-lactam monotherapy versus beta-lactam-aminoglycoside combination therapy

Infections caused by bacteria are a leading cause of preventable death. The mortality associated with severe infections necessitating hospitalization is about 30%. Antibiotic treatment improves survival.

There are several classes of antibiotics currently in use. The beta-lactam class is one of the most important class in use. Antibiotics belonging to it (penicillins, cephalosporins, and others) kill bacteria by disrupting their cell wall. Aminoglycosides (e.g. gentamicin) act though a different mechanism, inhibiting bacterial protein synthesis. Studies of bacteria in cell cultures have shown that combining a beta-lactam with an aminoglycoside results in bacterial killing superior to the simple additive activity of each of these antibiotics alone, a phenomenon termed 'synergism'.

In humans, combination therapy may have several drawbacks, such as an increased rate of adverse effects. We therefore decided to compile clinical studies that compared treatment with a beta-lactam to treatment with a beta-lactam plus an aminoglycoside. Our objective was to assess whether combination treatment results in better outcomes, mainly survival.

The review included 64 trials randomizing 7586 patients. Patients were hospitalized with urinary tract, intra-abdominal, skin and soft tissue infections, pneumonia, and infection of unknown origin. Antibiotics were administered intravenously.

Combination antibiotic treatment did not improve the clinical efficacy achieved with the beta-lactam antibiotic alone. One set of studies compared a new, broad-spectrum beta-lactam to an older, less potent beta-lactam combined with an aminoglycoside (44 studies). In these studies, mortality and failure were lower with single beta-lactam antibiotic treatment. Mortality was reduced by 15%, but the difference was not statistically significant. The other set of studies compared one beta-lactam to the same beta-lactam combined with an aminoglycoside (20 studies). In these trials, no differences between single and combination antibiotic treatment were seen. The relative risk for mortality was 1.01, denoting equivalence of the two regimens.

Adverse events rates did not differ between the study groups, overall, but renal damage was more frequent with combination therapy. Combination therapy did not prevent the development of secondary infections.

The reviewers conclude that beta-lactam-aminoglycoside combination therapy offers no advantage to beta-lactams alone. Furthermore, combination therapy is associated with an increased risk of renal damage. Paucity of trials comparing the same beta-lactam in both study arms and incompleteness of mortality reporting may limit these conclusions. These results may not apply to locations in which resistance rates to narrow-spectrum beta-lactams are very low, such as Scandinavian counties.

Résumé simplifié

Bêta-lactamine en monothérapie antibiotique par rapport à une combinaison de traitements antibiotiques par bêta-lactamine-aminoglycoside pour le sepsis.

Traitement antibiotique Unique versus combiné pour les infections graves : La monothérapie par bêta-lactamine versus la combinaison par bêta-lactamine-aminoglycoside

Les infections causées par des bactéries sont une cause majeure de décès évitables. La mortalité associée aux infections graves nécessitant une hospitalisation est d'environ 30 %. Le traitement antibiotique améliore la survie.

Il y a plusieurs classes d'antibiotiques actuellement utilisées. La classe des bêta-lactamines est l'une des classes les plus importantes en usage. Les antibiotiques appartenant à celle -ci (pénicillines, céphalosporines, et autres) tuent les bactéries en désorganisant leur paroi cellulaire. Les Aminoglycosides (par ex. la gentamicine) agissent par un mécanisme différent, inhibant la synthèse des protéines bactériennes. Les études sur des bactéries dans les cultures cellulaires ont montré que la combinaison d'une bêta-lactamine avec un aminoglycoside entraîne un effet mortel pour les bactéries supérieur à la simple activité additive de chacun de ces antibiotiques seuls, un phénomène appelé « synergie ».

Chez l'homme, la thérapie combinée peut avoir plusieurs inconvénients, tels que l'augmentation du taux d'effets indésirables. Nous avons donc décidé de compiler les études cliniques qui ont comparé le traitement par une bêta-lactamine à un traitement par une bêta-lactamine plus un aminoglycoside. notre objectif était d'évaluer si le traitement combiné donnaient de meilleurs résultats, principalement en termes de survie.

La revue a inclus 64 essais randomisant 7 586 patients. Les patients étaient hospitalisés pour infections des voies urinaires, intra-abdominales, de la peau et des tissus mous, pneumonie, et infection d'origine inconnue. Les antibiotiques étaient administrés par voie intraveineuse.

Le traitement par une combinaison d'antibiotiques n' a pas amélioré l'efficacité clinique obtenue avec l'antibiotique bêta-lactamine seul. Une série d'études comparaient une nouvelle bêta-lactamine, à large spectre à une bêta-lactamine plus ancienne, moins puissante combinée avec un aminoglycoside (44 études). Dans ces études, la mortalité et l'échec étaient inférieurs avec le traitement antibiotique bêta-lactamine seul. La mortalité était réduite de 15%, mais la différence n'était pas statistiquement significative. L'autre série d'études comparait une bêta-lactamine à la même bêta-lactamine combinée avec un aminoglycoside (20 études). Dans ces essais, aucune différence entre le traitement antibiotique simple et combiné n'a été observée. Le risque relatif pour la mortalité était de 1,01, indiquant une équivalence des deux schémas thérapeutiques.

Les taux d'événements indésirables ne différaient pas entre les groupes d'étude, dans l'ensemble, mais les lésions rénales étaient plus fréquents avec le traitement combiné. La thérapie combinée n'a pas empêché le développement d'infections secondaires.

Les évaluateurs concluent que la combinaison de traitements par bêta-lactamine-aminoglycoside n'offre aucun avantage par rapport aux bêta-lactamines seules. De plus, la thérapie combinée est associée à un risque accru de lésions rénales. Le manque d' essais comparant la même bêta-lactamine dans les deux bras de l'étude et l'insuffisance de notification de la mortalité peuvent limiter ces conclusions. Ces résultats peuvent ne pas s'appliquer aux sites dans lesquels les taux de résistance aux bêta-lactamines à spectre étroit sont très faibles, comme les pays scandinaves.

Notes de traduction

Traduit par: French Cochrane Centre 5th November, 2013
Traduction financée par: Instituts de Recherche en Santé du Canada, Ministère de la Santé et des Services Sociaux du Québec, Fonds de recherche du Québec Santé et Institut National d'Excellence en Santé et en Services Sociaux

Background

Sepsis is defined as the clinical evidence of infection, accompanied by a systemic inflammatory response such as fever. When associated with organ dysfunction, decreased blood flow in an organ (hypoperfusion), or abnormally low blood pressure (hypotension), sepsis is defined as severe (Bone 1992; Mandell 2004). Sepsis may be a response to direct microbial invasion or may be elicited by microbial signal molecules or toxin production. Infections may be lethal, with fatality rates ranging from less than 10% to more than 40% for those with severe sepsis (Moore 2001; Rangel-Frausto 1995; Russell 2000). Appropriate empirical antibiotic treatment, administered to the patient before identification of the pathogen or its antibiotic susceptibilities, has been shown to halve the fatality associated with sepsis (Bryant 1971; Ibrahim 2000; Leibovici 1998; Whitelaw 1992).

Regimens recommended for the empirical treatment of sepsis include: (1) a single broad-spectrum agent, commonly from the beta lactam class of antibiotics; and (2) a combination of a beta lactam antibiotic with an aminoglycoside antibiotic (Mandell 2004). Combination antibiotic therapy has several theoretical advantages. First, it may have a broader antibiotic spectrum. Second, the combination may possess an enhanced potential (synergism), when compared to the additive effect of each of the antibiotics assessed separately (Giamarellou 1986; Klastersky 1982). Third, combination therapy has been claimed to suppress the emergence of subpopulations of microorganisms resistant to the antibiotics (Allan 1985; Milatovic 1987). The disadvantages of combination therapy may include additional costs, enhanced drug toxicity, the possible induction of resistance caused by the broader antibiotic spectrum (Manian 1996; Weinstein 1985), and possible antagonism between specific drug combinations (Moellering 1986).

Aminoglycoside antibiotics are most active against Gram-negative bacteria (Mandell 2004). In addition, synergism between beta lactam antibiotics and aminoglycoside antibiotics has been repeatedly shown in vitro specifically for Gram-negative bacteria (Giamarellou 1986; Klastersky 1976; Klastersky 1982). Consequently, the benefit of combination therapy, if existent, may be more prominent in patients with Gram-negative infections. Other features related to the infection may affect prognosis. These include the site of infection and the specific causative pathogen. For example, infections caused by Pseudomonas aeruginosa have been shown to portend a poor prognosis (Baine 2001; Geerdes 1991; Leibovici 1997). We expect to deal with factors such as these, expected to underlie heterogeneity, using subgroup analysis where appropriate. Specific guidelines have been instituted for the empirical treatment of cancer patient with neutropenia, basing the suspicion of sepsis on fever alone (Hughes 2002). The authors have therefore considered studies addressing these patients in a separate review (Paul 2013).

Numerous studies have been conducted comparing beta lactam monotherapy to beta lactam-aminoglycoside combination therapy in patients with suspected or proven bacterial infections. Some trials have focused specifically on infections commonly caused by Gram-negative bacteria, such as urinary tract infections and hospital acquired infections, where the benefit of combination therapy may be more prominent. Nevertheless, superiority of either monotherapy or combination therapy has not been shown conclusively in these studies.

Objectives

Our objectives were:

  1. to compare beta lactam monotherapy versus beta lactam-aminoglycoside combination therapy in patients with sepsis; and

  2. to estimate the rate of adverse effects with each treatment regimen, including the development of bacterial resistance to antibiotics.

Methods

Criteria for considering studies for this review

Types of studies

We included randomized or quasi-randomized controlled trials.

Types of participants

We included hospitalized patients with sepsis acquired either in the community or in the hospital (nosocomial). We defined sepsis as clinical evidence of infection, plus evidence of a systemic response to infection (Bone 1992). We excluded neonates and preterm babies. We also excluded studies including more than 15% neutropenic patients.

Types of interventions

We considered studies comparing the antibiotic regimens described below.

  1. Any intravenous beta-lactam antibiotic given as monotherapy, including:

    1. penicillins;

    2. beta lactam drugs plus beta lactamase inhibitors (e.g. co-amoxiclav);

    3. cephalosporins (e.g. ceftazidime, cefotaxime);

    4. carbapenems (e.g. imipenem, meropenem).

  2. Combination therapy of a beta lactam antibiotic (as specified) with one of the following aminoglycoside antibiotics:

    1. gentamicin;

    2. tobramycin;

    3. amikacin;

    4. netilmicin;

    5. streptomycin;

    6. isepamicin;

    7. sisomicin.

Types of outcome measures

Primary outcomes

All-cause fatality by the end of the study follow-up.

Secondary outcomes
  1. Treatment failure defined as death and/or one or more serious morbid events (persistence, recurrence, or worsening of clinical signs or symptoms of presenting infection; any modification of the assigned empirical antibiotic treatment; or any therapeutic invasive intervention required not defined in the protocol).

  2. Length of hospital stay.

  3. Dropouts: number of patients excluded from the outcome assessment after randomization.

  4. Superinfection: recurrent infections defined as new, persistent, or worsening symptoms and/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.

  5. Colonization by resistant bacteria: the isolation of bacteria resistant to the beta lactam antibiotic, during or following antibiotic therapy, with no signs or symptoms of infection.

  6. Adverse effects:

    1. life-threatening or associated with permanent disability (severe nephrotoxicity; ototoxicity; anaphylaxis; severe skin reactions);

    2. serious: requiring discontinuation of therapy (other nephrotoxicity; seizures; pseudomembranous colitis; other allergic reactions);

    3. any other (other gastrointestinal; other allergic reactions).

Search methods for identification of studies

Electronic searches

We formulated a comprehensive search strategy in an attempt to identify all relevant studies regardless of language or publication status (published, unpublished, in press, and in progress). The key words used for the search strategy are shown in Appendix 1.

We searched the Cochrane Infectious Diseases Group specialized trials register for relevant trials up to December 2002 using the search terms: ((aminoglycoside* OR netilmicin* OR gentamicin* OR amikacin* OR tobramycin* OR streptomycin* OR isepamicin* OR sisomicin*) AND (pneumonia* OR infection OR infect* OR sepsis OR bacter* OR bacteremia OR septicemia).

We searched the Cochrane Controlled Trials Register, (CENTRAL), (The Cochrane Library, Issue 3, 2004) using the same search terms.

We searched the following electronic databases in combination with the search strategy developed by The Cochrane Collaboration and detailed in the Cochrane Handbook for Systematic Reviews of Interventions to limit the search for randomized or quasi-randomized trials (Higgins 2005):

  1. MEDLINE (1966 to July 2004) using the search: (aminoglycoside* OR netilmicin* OR gentamicin* OR amikacin* OR tobramycin* OR streptomycin* OR isepamicin* OR sisomicin*) AND (combination OR combi*). In a second search, the terms (combination OR combi*) were replaced by endocarditis, Staphylococcus, Streptococcus or pneumonia to enhance the sensitivity and specificity of our search to these infections.

  2. EMBASE (1980 to March 2003) using the same search terms.

  3. LILACS (1982 to July 2004) using the same search terms.

Searching other resources

We searched the Interscience Conference of Antimicrobial Agents and Chemotherapy conference proceedings (1995 to 2003) for relevant abstracts.

We contacted the first or corresponding author of each included study, and the researchers active in the field, for information regarding unpublished trials or complementary information on their own trials.

We also checked the citations of major reviews and of all trials identified by the above methods for additional studies.

We did not have a language restriction.

Data collection and analysis

Study selection

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

Quality assessment

We assessed the quality of the trials to be included for allocation sequence, allocation concealment, blinding, fatality outcome reporting, intention-to-treat analysis, and number of patients excluded from outcome assessment. Two authors (MP, IS or KSW) independently performed quality assessment. We based methodological quality classification on the evidence of a strong association between poor allocation concealment and over estimation of effect. We defined it as: A (low risk of bias; adequate allocation concealment); B (moderate risk of bias; some doubt about allocation concealment); and C (high risk of bias; inadequate allocation concealment) (Schulz 1995). We performed sensitivity analyses to assess the effect of study quality measures on effect estimates. We intend to assess the effect of number of exclusions on effect estimates (above or below 20%) in future updates of the review.

Data collection

Two authors (MP, IS or SG) independently extracted data from included trials. In case of disagreement between the two authors, a third author (KSW, LL) independently extracted the data. A third author (KSW or LL) also extracted the data in 10% of the studies, selected at random. We discussed data extraction, documented decisions, and contacted authors of all studies for clarification. We resolved differences in the data extracted by discussion. We also documented the justification for excluding studies from the review.

We identified the trials by the name of the first author and the year in which the trial was first published, and listed in chronological order. We extracted, checked and recorded the following data.

Characteristics of trials:

  1. date, location, and setting of trial;

  2. publication status;

  3. country of origin;

  4. design (intention-to-treat, method of randomization);

  5. duration of study follow-up;

  6. performance of surveillance cultures (routine cultures for the detection of colonization);

  7. sponsor of trial.

Characteristics of patients:

  1. number of participants in each group;

  2. age (mean and standard deviation, or median and range);

  3. number of patients with renal failure before treatment;

  4. number of patients with shock.

Characteristics of infection:

  1. number of patients with infections caused by bacteria resistant to the administered beta lactam antibiotic;

  2. number of patients with nosocomial infections;

  3. number of patients with bacteraemia;

  4. number of patients with bacteriologically documented infection;

  5. number of patients with infections caused by Gram-negative bacteria;

  6. number of patients with Gram-negative bacteraemia;

  7. number of patients with documented Pseudomonas infections (Pseudomonas isolated in the blood or specimen(s) obtained from suspected site(s) of infection);

  8. number of patients with:

    1. urinary tract infection;

    2. pneumonia;

    3. intra-abdominal infection;

    4. skin and soft tissue infection; and

    5. infection of unknown origin.

Characteristics of interventions:

  1. antibiotic type and dose;

  2. duration of therapy (mean).

Characteristics of outcome measures:

  1. number of deaths at the end of the follow-up period;

  2. number of patients failing treatment (as defined);

  3. adverse reactions (as defined) in each group;

  4. loss of follow-up (dropouts) before the end of the study in each group;

  5. number of patients developing super-infection;

  6. number of patients developing colonization (as defined) with resistant bacteria;

  7. duration of fever and hospital stay.

We collected outcome measures on an intention-to-treat basis whenever possible. Where such data were not presented, we sought information from the authors, and if unavailable, per-protocol results were used. For failure outcome, we performed sensitivity analyses comparing these results with a 'presumed all intention to treat', which we achieved by counting all dropouts as failures. We could not make such an assumption in studies that did not specify the number of dropouts per study arm, and we analysed these studies separately.

Data synthesis

We calculated relative risks for dichotomous data. Continuous outcomes were unavailable for this review. We will use weighted mean differences for continuous outcomes in future updates of the review. We initially assessed heterogeneity in the results of the trials using a chi-squared test of heterogeneity (P < 0.1). We used a fixed effect model throughout the review, as the I2 measure of inconsistency was low for all comparisons. We compared results obtained by the fixed effect model to those obtained by a random effect model for the major outcomes. We explored the following factors to explain heterogeneity in relation to the major outcomes:

  1. infections caused by Pseudomonas sp. versus all other infections;

  2. Gram-negative versus all other infections; and

  3. urinary tract infections versus other sites of infection.

We performed subgroup analysis by these factors where data were available. For subgroup analyses we extracted all-cause fatality and treatment failures outcomes. We adjusted the descriptive mean mortality rate in included studies to the inverse of the mortality variance between the trials.

We examined a funnel plot of SE(log(relative risk)) versus relative risk of each study in order to estimate potential selection bias (publication and language).

Results

Description of studies

The search strategy resulted in 5568 references. We filtered double references, and screened 2805 different abstracts for inclusion. We did not evaluate studies in which the comparator antibiotic regimens were clearly incompatible with inclusion criteria in depth. We similarly excluded non-randomized and non-human studies.

We retrieved 145 studies for full-text inspection, of which we excluded 67 publications, representing 63 studies (see table of ' Characteristics of excluded studies'), and categorized two as awaiting assessment (see Additional Table 1, and 'Table of studies awaiting assessment'). Several studies compared monotherapy versus combination therapy among patients with cystic fibrosis. Patients in these studies typically do not have fever or other signs of sepsis when entering the trial, and thus did not fulfil inclusion criteria for this review. These studies are included in a separate review (Elphick 2005). Seventy-eight studies fulfilled inclusion criteria. Fourteen were double publications, and thus we have included 64 trials in this review. We requested complementary information from nearly all the authors, and included complementary data in 22 studies (see references to studies).

Table 1. Table of studies awaiting assessment
Study IDExplanationContact details
Luis-Alberto 1999According to abstract (LILACS) patients with community acquired pneumonia were divided ('al azar') randomly to ceftazidime versus penicillin + amikacin. Awaiting full text availability for inclusion and data extraction. 
Figueroa-Damian 1996According to abstract (LILACS) patients were included sequentially (Aleatoriamente), and given piperacillin-tazobactam or ampicillin + gentamicin for postcaesarean endometiris. Awaiting full text availability to inspect whether study indeed randomized. 

We have detailed study characteristics in the table of ' Characteristics of included studies'. The included studies were performed between the years 1968 to 2001. Twenty-two were multi-centred. Twenty-one were performed in the USA or Canada, 34 in Europe, and 10 in other countries.

The studies included 7586 patients. The median number of included patients per trial was 87.5 (range 20 to 580). Two trials (Cardozo 2001; Naime Libien 1992) included children, while all other trials were restricted to or included mostly adults.

The studies differed by the type of population and infection targeted (see table of ' Characteristics of included studies'). Most trials (designated 'sepsis') included patients with severe sepsis, suspected Gram-negative infections (25 trials), or pneumonia (16 trials). The adjusted mean fatality rate in these studies was 8.6%. Eleven trials included patients with intra-abdominal infections, related mainly to the biliary tract (designated 'abdominal'). The mean fatality in these trials was 1.7%. Seven trials were restricted to patients with urinary tract infections (UTIs), all hospitalized, mainly women (UTI). Five of these studies reported fatality, and no deaths occurred in four. Finally, five of the studies included in the review targeted patients with Gram-positive infections, mainly endocarditis. We will present results for these infections separately, in addition to their inclusion in the overall analysis.

Most studies compared the initial, empirical antibiotic treatment administered to the patients. Four studies assessed the empirical treatment of a specific infection by randomizing patients empirically and evaluating only those subsequently fulfilling criteria for the specific infection. Two such studies randomized patients with suspected endocarditis and evaluated only those with Staphylococcus aureus bacteraemia and proven endocarditis (Abrams 1979; Korzeniowski 1982). The other two randomized patients with suspected biliary tract infections and evaluated only patients with a surgically proven diagnosis (Gerecht 1989; Yellin 1993). Non-evaluated patients in these studies were not counted as dropouts, since the study design defined evaluation only for patients fulfilling definitive criteria. Eight studies, focusing on patients with specific infections or pathogens (e.g., cholecystitis, Staphylococcal infections, etc.), tested the effect of monotherapy versus combination therapy semi-empirically. In these studies (designated 'semi-empirical', see table of ' Characteristics of included studies') randomization occurred after the specific infection was documented, and patients could have received prior antibiotic treatment for this infection. Analysis of empirical and semi-empirical studies was not separated.

The specific antibiotic regimens used are detailed in the table of ' Charcteristics of included studies'. Forty-four studies compared a single beta-lactam drug to a different, narrower spectrum, beta-lactam combined with an aminoglycoside (designated 'different BL'). Sixteen 'different BL' studies reported baseline susceptibility rates of the pathogens isolated on admission to the beta-lactam. The beta-lactam used in the combination arm covered less pathogens than the monotherapy beta-lactam in 13 studies, while the opposite occurred in two studies only. Twenty studies compared the same beta-lactam (designated 'same BL'). Results obtained from studies comparing same and different beta-lactams were kept separated throughout all efficacy analyses. The aminoglycoside was administered once daily in six trials (Cardozo 2001; Jaspers 1998; Rubinstein 1995; Sandberg 1997; Sexton 1998; Speich 1998). Other trials administered the aminoglycosides multiple daily (47 trials), or did not specify the administration schedule (11 trials). Mean antibiotic treatment duration ranged between 4 to 17.5 days in the sepsis studies, 6.8 to 11.9 in the abdominal studies, 4.1 to 7 days in the UTI studies, and 2 to 4 weeks in the endocarditis studies.

Risk of bias in included studies

(See Additional Table 2: Study quality assessment table.)

Table 2. Study quality assessment
Study IDAlloc. generationAlloc. concealmentBlindingIntention to treatLost to follow up
Abrams 1979No informationNo informationNoneNo12 of 36 randomized patients, but none out of patients fulfilling pre-specified inclusion criteria (staphylococcus aureus bacteraemia)
Aguilar 1992No informationNo informationNoneUnknownNo reference to drop-outs in study
Alvarez-Lerma 2001aComputer generated in blocks of 6 patientsCentral and sealed opaque envelopes For mortality only24 of 140 randomized patients for failure
Arich 1987Table of random numbersSealed opaque envelopesNoneNo18 of 65 randomized patients
Bergeron 1988No informationNo informationNoneNo11 of 77 randomized patients
Biglino 1991No informationNo informationNoneUnknownNo reference to drop-outs in study
Brown 1984Random tableNo informationSingle, outcome assessorSo14 of 48 randomized patients
Carbon 1987No informationNo informationNoneUnknownNo reference to drop-outs in study
Cardozo 2001Numerical assignationNo informationNoneUnknownNo reference to drop-outs in study
Cometta 1994Table of random numbersSealed, opaque numbered envelopesNoneNo33 of 313 randomized patients
Cone 1985No informationNo informationNoneNo17 of 57 randomized patients
Coppens 1983Consecutively numbered envelopesEnvelopes (sealed or opaque not mentioned)NoneNo14 of 80 randomized patients
D'Antonio 1992Table of random numbers, stratified according to underlying malignancySealed opaque envelopes.NoneFor mortality only17 of 300 randomized patients for failure
Duff 1982Based on the last digit of the hospitalization number - odds/ evensNo informationNoneYesNone
Dupont 2000Computer generated in blocks of 4 patientsCentralSingle, evaluation committeeNo14 of 241 randomized patients
Felisart 1985Table of random numbersNo informationNoneYesNone
Finer 1992Computer generatedSealed, opaque envelopesNoneFor mortality only56 of 471 randomized patients for failure
Gerecht 1989Computer generated table of random numbersNo informationNoneNo4 of 50 randomized patients fulfilling operative and bacteriological criteria for cholangitis
Gomez 1990aComputer generatedSealed, closed envelopesNoneNo119 of 197 randomized patients, but none out of patients fulfilling pre-specified inclusion criteria (gram-negative bacteraemia)
Havig 1973Consecutive according to a randomized list prepared in advanceNo informationNoneNo22 of 90 randomized patients
Hoepelman 1988Randomization lists made by handSealed opaque envelopesnoneYesNone
Holloway 1985No informationNo informationNoneNo10 of 43 randomized patients
Iakovlev 1998Parallel groups, using the envelope methodEnvelopes (sealed or opaque not mentioned)NoneYesNone
Jaspers 1998Table of random numbers in consecutive envelopesSealed opaque envelopesNoneYesNone
Klastersky 1973No informationNo informationNoneNo7 of 75 randomized patients
Kljucar 1990Computer generated in consecutively numbered envelopesClosed envelopesNoneNo1 of 150 randomized patients
Koehler 1990No informationNo informationNoneFor mortality only17 of 144 randomized patients for failure
Korzeniowski 1982Table of random numbersCentralNoneNo82 of 156 randomized patients, but only 4 of 78 patients fulfilling pre-specified inclusion criteria (Staphylococcus aureus endocarditis)
Landau 1990According to last digit of identification numberNo informationNoneUnknownNo reference to drop-outs in study
Limson 1988No informationNo informationNoneNo14 of 54 randomized patients
Mandell 1987Consecutive in blocks of fourSealed envelopes (opaque not mentioned)NoneNo19 of 129 randomized patients
Martin 1991Randomization tableNo informationNoneNo22 of 116 randomized patients
McCormick 1997Table of random numbersSealed, opaque envelopesNoneNo19 of 147 randomized patients
Mergoni 1987No informationSealed opaque envelopesNoneUnknownNo reference to drop-outs in study
Moreno 1997No informationNo informationNoneNo12 of 70 randomized patients
Mouton 1990No informationNo informationNoneYesNone
Mouton 1995No informationNo informationNoneFor mortality only43 of 272 randomized patients for failure
Muller 1987Computer generated listsNo informationNoneNo25 of 131 randomized patients
Naime Libien 1992No informationNo informationNoneUnknownNo reference to drop-outs in study
Piccart 1984No informationNo informationNoneNo20 of 105 randomized patients
Rapp 1984No informationNo informationNoneYesNone
Rasmussen 1986Table of random numbers NoneNo3 of 62 randomized patients
Ribera 1996Table of random numbersSealed, opaque envelopesNoneYesNone
Rubinstein 1995Computer generatedSealed, opaque, numbered envelopesSingle, outcome assessors blindedFor mortality only75 of 580 randomized patients for failure
Sage 1987No informationEnvelopesNoneNo13 of 61 randomized patients from arms used in review
Sandberg 1997Computer generated lists in blocks of fourSealed opaque envelopesNoneFor mortality only22 of 73 randomized patients for failure
Sanfilippo 1989Computer generated codeCentralDouble blind, placebo controlledUnknownNo reference to drop-outs in study
Sculier 1982No informationNo informationNoneYesNone
Sexton 1998No informationNo informationNoneNo16 of 67 randomized patients
Sieger 1997No informationNo informationNoneYesNone
Smith 1984Table of random numbersCentralDouble blind, placebo controlledNo5 of 200 randomized patients for failure; 13 of 200 randomized patients for mortality
Speich 1998Computer generatedSealed opaque envelopes.NoneFor mortality only5 of 89 randomized patients for failure
Stille 1992Computer generated list of blocks of 16 patientsNo informationNoneYesNone
Sukoh 1994No informationEnvelopes (sealed or opaque not mentioned).NoneUnknownNo reference to drop-outs in study
Takamoto 1994Computer generatedEnvelopes (sealed or opaque not mentioned).NoneNo14 of 171 randomized patients
Thompson 1990Computer generatedNo informationNoneNo10 of 106 randomized patients
Thompson 1993Computer generatedNo informationNoneNo27 of 147 randomized patients
Trujillo 1992No informationNo informationNoneUnknownNo reference to drop-outs in study
Vergnon 1985Tirage a sortNo informationNoneUnknownNo reference to drop-outs in study
Verzasconi 1995No informationNo informationSingleNo6 of 93 randomized patients
Warren 1983Random numbers contained within consecutively numbered envelopesSealed envelopes (opaque not mentioned).NoneNo3 of 123 randomized patients
Wiecek 1986No informationNo informationNoneUnknownNo reference to drop-outs in study
Wing 1998Computer generated random number tableSealed opaque envelopes.NoneYesNone
Yellin 19932:1 (mono.: combi.), through table of random numbersCentralSingle, providerno59 of 149 randomized patients, but only 22 of 112 patients fulfilling pre-specified inclusion criteria (infection proven at surgery)

Allocation concealment and generation

Thirty-three percent of the studies (21/64) reported adequate allocation concealment. Two studies were graded as C (Duff 1982; Landau 1990). No information was available for the other studies (34 studies), or envelopes were used but not described as sealed or opaque (7 studies).

Allocation generation was described as adequate in 53% of the studies (34/64). No information was available for 28 studies. Two studies were quasi-randomized, using patient identification numbers (Duff 1982; Landau 1990).
Both allocation generation and concealment were considered adequate in 30% of the studies (19/64).

Blinding

Most studies were open. Two studies, including 226 patients, were double blinded (Sanfilippo 1989; Smith 1984). Outcome assessors were blinded in four studies (Brown 1984; Dupont 2000; Rubinstein 1995; Verzasconi 1995). Clinicians were blinded to the treatment in one study (Yellin 1993).

Intention-to-treat versus per-protocol analysis

We separated included studies into four different study types with relation to outcome reporting:

  1. full Intention-to-treat analysis;

  2. per-protocol analysis, in which the number of dropouts was given per study arm;

  3. per-protocol analysis, in which the number of dropouts was known, but not given per study arm;

  4. studies which did not distinguish between the number of randomized and number of evaluated patients. These studies did not refer to dropouts, yet did not define the study explicitly as intention-to-treat.

The distribution of included studies by study type was as follows:
All cause fatality (reported in 43 studies):
Type 1: 19 studies (44%);
Type 2 and 3: 18 studies (42%). As authors cannot make assumptions can be made regarding dropouts for mortality, we have joined study groups 2 and 3 are joined for mortality;
Type 4: 6 studies (14%).

Treatment failure: (reported In 63 studies);
Type 1: 13 studies (21%);
Type 2: 23 studies (37%);
Type 3: 16 studies (25%);
Type 4: 11 studies (17%).

Follow-up

Forty-three studies (67%) specified follow-up duration, while only 18 studies defined a specific time for outcome collection (28%). Follow-up ranged from 48 hours following treatment cessation to 6 months. Outcomes were extracted preferentially at up to 30 days, with the exception of the Gram-positive infection studies, in which the type of infection mandated a longer follow-up (3 to 6 months).

Effects of interventions

All cause fatality

(see Analysis 1)

Forty-three trials including 5527 patients were included in this comparison (see Analysis 1.1). Twelve studies, including 1381 patients, compared the same beta-lactam. These studies showed near equivalence, RR 1.01 (95%CI 0.75-1.35), while studies comparing different beta-lactams tended non-significantly in favour of monotherapy, RR 0.85 (95%CI 0.71-1.01). Analysis was further broken down according to the main study population, excluding Gram-positive infection studies (see Analysis 1.2). The advantage to the monotherapy among studies comparing different beta-lactams was statistically significant in studies addressing 'sepsis', RR 0.83 (95% CI 0.69 to 0.99). No heterogeneity was present for these comparisons (I2 = 0% for the same beta-lactam comparison, I2 = 19.4 for different beta-lactams).

Subgroup analyses

No significant difference between monotherapy and combination therapy was apparent when analysis was restricted to patients with any Gram-negative infection (eight studies) or Gram-negative bacteraemia (four studies, see Analysis 1.3 to Analysis 1.4). Only three studies permitted mortality outcome extraction among patients with Pseudomonas aeruginosa infections, and these did not show a differences, either alone or when combined (graph not shown). Five UTI studies reported mortality, and mortality was null in three studies. Excluding patients with urinary tract infection from the analysis ('non-UTI' subgroup, see Analysis 1.5) strengthened the advantage to monotherapy in studies comparing different beta-lactams (RR 0.70, 95%CI 0.52-0.95).

Sensitivity analyses

Adequate allocation concealment and generation were associated with relative risk closer to one, both for studies comparing the same and different beta-lactams. (See Analysis 7.1 and Analysis 7.2). Combination therapy was significantly better among studies comparing different beta-lactams classified as B. Blinding was performed in too few studies to assess its effect on mortality. The combined RR for studies comparing the same beta-lactam reporting fatality by intention-to-treat was 0.62 (95% CI 0.27 to 1.43), compared to 1.09 (95% CI 0.80 to 1.51) for studies reporting fatality per-protocol (Analysis 7.3). Comparing intention to treat to per-protocol studies for different beta-lactams did not reveal a difference. Re-analysis of the mortality comparison by the random effect model was very similar (RR 1.02, 95% CI 0.76-1.38 for same beta-lactam, RR 0.85 95% CI 0.69 to 1.05 for different beta-lactam).

Treatment failure

(see Analysis 2)

We included all trials but one (Wiecek 1986) in the clinical failure analysis, comprising 6616 patients (see Analysis 2.1). We found a significant advantage to monotherapy among studies comparing different beta-lactams, RR 0.77 (95% CI 0.69 to 0.86). We detected no difference between monotherapy and combination therapy among studies comparing the same beta-lactam, RR 1.11 (95% CI 0.95 to 1.29). No heterogeneity was present (I2=0% for both comparisons).

Grouping studies according to study population highlighted an advantage to combination therapy among the 'sepsis' studies that compared the same beta-lactam, RR 1.25 (95%CI 1.01 tp 1.55). This group of studies also accentuated the opposing advantage to monotherapy among studies comparing different beta-lactams (see Analysis 2.2).

Bacteriological cure occurred more frequently with monotherapy among studies comparing different beta-lactams, RR 0.81 (95% CI 0.69 to 0.94), but did not differ significantly in studies comparing the same beta-lactam (see Analysis 2.3).

Assessment of efficacy for urinary tract infections included re-infections and relapse as outcomes (see Analysis 2.4). We noted no significant difference between monotherapy and combination therapy , with six trials and 458 patients included in this comparison.

Subgroup analyses

We analysed 28 studies including 1835 patients with Gram-negative infections and 18 studies including 426 patients with Pseudomonas aeruginosa infections were analysed (see Analysis 2.5 and Analysis 2.7). We observed no significant differences between the study groups, either for studies comparing the same or different beta-lactams. For studies comparing the same beta-lactam the RR was 1.23 (95% CI 0.90 to 1.68) for Gram-negative infections and 1.02 (95% CI 0.68 to 1.51) for Pseudomonas aeruginosa infections. We observed no difference between study groups among patients with Gram-negative bacteraemia or any bacteraemia (see Analysis 2.6 and Analysis 2.8). The latter comparison mainly comprised of patients with Gram-negative bacteremias but was available from a larger number of studies, and showed an advantage to combination therapy among studies comparing different beta-lactams. Both the subgroups of patients with urinary tract infections (see Analysis 2.8), and patients without urinary tract infections maintained the trends seen previously (Analysis 2.9).

Sensitivity analyses

The quality of allocation concealment and generation did not affect the relative risks for treatment failure, either among studies comparing the same or different beta-lactams. The two studies graded as C compared different beta-lactams, and were non-significantly closer to one than the truly randomized studies (see Analysis 7.4 to Analysis 7.5).

Several studies comparing different beta-lactams used some type of blinding. The advantage to monotherapy was non-significantly larger among these studies, compared to non-blinded studies (see Analysis 7.6).

Among studies comparing the same beta-lactam, we observed an advantage to combination therapy in the presumed intention to treat group (type 2 studies), in which we imputed failure for dropouts. Among studies comparing different beta-lactams, intention to treat, presumed intention to treat, and per-protocol results were similar, favouring monotherapy (see Analysis 7.7). Analysis by the random effect model did not change results (RR 1.09, 95% CI 0.94-1.27 for same beta lactams, RR 0.76, 95% CI 0.68-0.97, for different beta-lactams).

Length of hospital stay

Only four studies contained usable information for the comparison of hospital stay. Significant heterogeneity precluded their combination. Duration of hospitalization was longer with monotherapy in one study (McCormick 1997, 128 patients), shorter in another (Arich 1987, 47 patients), and similar in two (Wing 1998; Yellin 1993, 269 patients).

Summary of gain

Among studies comparing the same beta-lactam there was no benefit to the combination arm for all mortality comparisons, including subgroup and sensitivity analyses. Treatment failure tended to favour the combination arm reaching statistical significance only among studies addressing 'sepsis' and when an intention to treat analysis was imposed on studies performed per-protocol, imputing failure for dropouts.

Studies using different beta-lactam usually compared a broad-spectrum beta-lactam to a narrower spectrum beta-lactam combined with an aminoglycoside. The mortality comparisons favoured monotherapy reaching statistical significance in several subgroups. Treatment failure was significantly in favour of monotherapy overall, among the 'sepsis' studies, the non-UTI subgroup and in all the methodology sensitivity analyses. No comparison favoured the combination arm.

Resistance development and adverse events

(see 'Analysis' 3 and 4)

We compared studies comparing same and different beta-lactams for the assessment of resistance development and adverse events. These outcomes are intended to assess the antibiotic class effect of aminoglycoside-beta-lactam combinations versus beta-lactams alone, whether same or different.

We detected no significant differences between the rates of bacterial or fungal superinfections (see Analysis 3.1 to Analysis 3.4). Bacterial superinfections occurred more frequently with combination therapy, RR 0.76 (95% CI 0.57 to 1.01). This was the largest comparison, including 27 studies and 3085 patients. In outcome 5 we compared bacterial colonization rates only in patients from whom surveillance cultures were taken (7 studies, 751 patients). Colonization was, again, non-significantly more frequent with combination therapy, RR 0.78 (95% CI 0.60-1.01). Few studies monitored development of resistance among pathogens isolated initially (Analysis 3.6). We observed no difference between monotherapy and combination therapy.

Any adverse event occurred non-significantly more frequently with combination therapy, RR 0.92 (95% CI 0.83 to 1.01; see Analysis 4.1). We found nephrotoxicity to be more common in the combination arm in nearly all studies, with a highly significant combined relative risk in favour of monotherapy, RR 0.30 (95% CI 0.23 to 0.39, Analysis 4.3). A significantly increased rate of nephrotoxicity was seen both in studies administering the aminoglycoside once daily and in those with a multiple-day regimen. Vestibular and ototoxicity, other known serious side effects of aminoglycoside treatment, were not reported routinely and could not be analysed. Different definitions and detailing of specific adverse events precluded a meaningful meta-analysis of other adverse events, individually or grouped.

Dropouts and selection bias

(see 'Analysis' 5)

The number of patients excluded from each study arm was nearly equal, both for mortality (RR 1.00, 95% CI 0.66 to 1.49, Analysis 5.1), and failure (RR 1.04, 95% CI 0.88 to 1.23, Analysis 5.2) outcomes assessment. This comparison included studies in which these outcomes could only be collected per-protocol, and reported the number of dropouts per study arm. It should be noted that counting dropouts as failures did affect the combined failure results (failure sensitivity analysis above). This is because among studies comparing the same beta-lactam, a slightly higher rate of dropouts occurred in the monotherapy arm, while the opposite occurred among studies comparing different beta-lactams.

The funnel plot for treatment failure generated a nearly symmetric 'funnel distribution' (Figure 1). Funnel plot analysis for all-cause fatality showed that small studies favouring combination therapy may be missing (Figure 2). Mortality outcome was unavailable from 33% of the trials.

Figure 1.

Funnel failure.

Figure 2.

Funnel mortality.

All cause mortality

Gram positive-infections

(see 'Analysis' 6)

Five studies assessed Gram-positive infections specifically. Four studies addressed patients with endocarditis caused by Staphylococcus aureus (Abrams 1979; Korzeniowski 1982; Ribera 1996), or streptococci (Sexton 1998). One study included any staphylococcal infection (Coppens 1983). All of these compared the same beta-lactam, with or without an aminoglycoside. Although small, we chose to separate this subset of studies and present its meta-analysis, since the rationale and clinical practice of adding an aminoglycoside to the beta-lactam in these infections differ from those underlying combination use in other infections.

The comparison included four outcomes: all cause fatality (three studies, outcome 1), clinical and bacteriological failure (five studies, outcomes 2 to 3), and the need for surgery (four endocarditis studies, outcome 4). None of these comparisons showed an advantage to combination therapy. The combined relative risk consistently favoured monotherapy, although differences were non-significant. The combined relative risk for clinical failure was 0.69 (95% CI 0.40 to 1.19, 5 studies, 305 patients). Clinical failure in these studies could be and indeed was defined more rigorously than in other studies. The time of outcome determination was pre-defined in all the trials and the follow-up was longer (1 to 6 months). Measures of treatment failure included persistence of bacteraemia or signs of endocarditis, relapse, need for valve replacement, and death.

Discussion

This present review compares beta-lactam-aminoglycoside antibiotic combinations to beta-lactam monotherapy. The primary outcome we assessed was all-cause fatality. Most studies compared one beta-lactam to a different, narrower spectrum beta-lactam, combined with an aminoglycoside. Twenty of the 64 included studies used the same beta-lactam in both study arms.

A special emphasis should be placed on studies comparing the same beta-lactam. These are the studies directly testing the hypothesis that the addition of an aminoglycoside to the beta-lactam is beneficial. Among these studies, all-cause fatality did not differ between study arms (RR 1.02, 95% CI 0.76 to 1.38). Treatment failure occurred more frequently in the monotherapy arm, reaching statistical significance only in subgroup analyses.

In studies comparing different beta-lactams, both failure and mortality were more common in the combination treatment arm. Failure was highly significant, while mortality reached significance only with subgroup analyses. These studies demonstrate an advantage to broad-spectrum beta-lactam monotherapy when compared to a narrower spectrum beta-lactam combined with an aminoglycoside, despite an equal in-vitro coverage of the culprit pathogens in both arms.

Development of resistance was assessed by the occurrence of superinfections and colonization, assuming that bacteria appearing under antibiotic treatment are resistant to the antibiotic administered. No difference between monotherapy and combination therapy was detected. Adverse events occurred more frequently with combination therapy. Specifically, nephrotoxicity occurred significantly more frequently in the combination treatment arm (RR 0.30, 95% CI 0.23 to 0.39).

We defined all-cause fatality as the primary outcome, while most studies assessed and reported treatment failure as a main outcome. Obviously, the most significant outcome for the patient is survival following the infectious episode. Available evidence shows that the addition of an aminoglycoside to a beta-lactam does not reduce mortality. Replacing beta-lactam monotherapy with a narrower spectrum beta-lactam combined with an aminoglycoside may be associated with increased mortality.

Failure was commonly defined as lack of clinical improvement, deterioration, relapse, and/or modifications to the antibiotic treatment. These endpoints are highly subjective and do not necessarily translate to detriments experienced by the patient. Detection bias is a concern in open trials that compared the same beta-lactam, or in trials comparing a 'new' broad spectrum monotherapy to a conventional antibiotic regimen. Thus, the advantage to monotherapy therapy in studies comparing different beta-lactams, and the opposing advantage to combination therapy in studies comparing the same beta-lactams, may be largely biased.

The major adverse event associated with combination therapy was nephrotoxicity. We did not observe a protective effect of the combination with regard to resistance development. During the last decade, once daily administration of aminoglycosides has entered into use, with similar efficacy but lower nephrotoxicity (Barza 1996). Most studies in our review used multiple-day administration schedules for the complete duration of antibiotic therapy or until modification. The RR of 0.30 for any nephrotoxicity we observed may, therefore, be an overestimation. However, the RR among the few studies that did administer the aminglycoside once daily was also highly significant in favour of monotherapy (0.17, 0.06 to 0.53).

The rationale for administering combination therapy arose from in-vitro studies showing synergistic bactericidal activity of specific beta-lactam-aminoglycoside antibiotic combinations. Synergy has been observed for Pseudomonas aeruginosa (Giamarellou 1984), other Gram-negative bacteria (Giamarellou 1986; Klastersky 1976), and Staphylococci (Sande 1975; Sande 1976). Assessment of antibiotic efficacy against specific infections in randomized trials must either be limited to definitive treatment (randomisation performed when infection is microbiologically documented), or be performed as a subgroup analysis to assess empirical treatment (randomizing patients empirically and assessing those with documented infections). Eight studies assessed definitive treatment (semi-empirical studies), while most assessed empirical treatment. We did not find an advantage to combination therapy among patients with any Gram-negative infection, Gram-negative bacteraemia, or Pseudomonas aeruginosa infections. Lack of data precluded the assessment of Pseudomonas aeruginosa bacteraemia.

In a previous non-randomized prospective study of bacteraemic patients, we showed that appropriate beta-lactam monotherapy was as effective as appropriate beta-lactam aminoglycoside combination therapy, both empirically and semi-empirically. Appropriate single aminoglycoside monotherapy was associated with increased mortality (Leibovici 1997). Combination therapy was claimed superior to monotherapy in a prospective observational study of patients with Pseudomonas aeruginosa bacteraemia, but most patients in the monotherapy group received aminoglycosides (Hilf 1989). In a meta-analysis including non-randomized trials (mostly retrospective cohort studies), Safdar and colleagues found a reduction in mortality with combination therapy for patients with Pseudomonas aeruginosa bacteraemia (five studies; OR 0.50, 95%CI 0.32 to 0.79), but not for patients with Gram-negative bacteraemia (17 studies; OR 0.96, 95% CI 0.79 to 1.32). Monotherapy, however, included single aminoglycoside treatment, and analysis was not performed separately for beta-lactam monotherapy (Safdar 2004). Finally, in a previous systematic review and meta-analysis of randomized trials comparing combination therapy to beta-lactam monotherapy for febrile neutropenic patients, no advantage was seen for the combination (Paul 2013). Overall, empirical evidence does not show the synergy effect when adding an aminoglycoside to a beta-lactam in the clinical setting. Why does synergy, observed in-vitro, not translate into clinical benefit? Specific growth conditions in-vitro, unattainable in-vivo, may induce synergism. Pharmacokinetic and pharmacodynamic properties involving specific antibiotics, sites of infection, timing and intervals of administration may prevent synergism in-vivo. Adverse events related directly to the aminoglycoside, or to the combination, may interfere with an in-vivo benefit, amounting altogether to no benefit.

A small subset of studies in our review addressed patients with Gram-positive infections, mainly Staphylococcus aureus endocarditis. No study assessed enterococcal infections specifically. In these, also, no outcome was improved by the addition of an aminoglycoside. Current guidelines for the treatment of Staphylococcus aureus endocarditis advise the addition of an aminoglycoside to the beta-lactam, at least initially (Bayer 1998). These recommendations rely mainly on in-vitro data (Sande 1975; Sande 1976). Animal studies have shown that sterilization of cardiac vegetations may be achieved more rapidly with combination therapy (Sande 1975; Sande 1976). One clinical study included in our review showed that combination therapy shortened the duration of bacteraemia, but this comparison was performed according to the empirical antibiotic regimen, while randomization occurred either empirically or semi-empirically (Korzeniowski 1982). We could not show an advantage to combination therapy combining all trials in humans. On the contrary, all outcomes tended to favour monotherapy, although statistical significance was not reached.

The limitations of our analysis may originate from the quality of data reported in available studies and from our analysis of these data. Of these, we emphasize the lack of data for all-cause fatality from a third of included studies. Survival, with or without the more subjective assessment of infection-related mortality, must be reported comparatively in all trials. Data for subgroups most likely to benefit from combination therapy were also not available from all studies. In our analysis, we did not correct for the appropriateness of antibiotic treatment, which has been shown conclusively to correlate with survival (Ibrahim 2000; Leibovici 1998). Data were not fully available to perform such an analysis. However, among studies comparing the same beta-lactam, combination therapy by definition broadened the spectrum of coverage, without improving outcomes. In studies comparing different beta-lactams, inappropriate beta-lactam was used more frequently in the combination arm, which may partially explain the advantage to monotherapy.

We conclude that the addition of an aminoglycoside to a beta-lactam does not improve the clinical efficacy achieved with the beta-lactam alone. Substituting a narrow-spectrum beta-lactam with an aminoglycoside for a single broad-spectrum beta-lactam, will result in increased failure rates and may be associated with increased mortality. Adverse events occur more frequently with combination treatment. Short-term combination therapy for sepsis does not prevent development of resistant bacteria, as assessed by superinfection or colonization rates following antibiotic treatment. Thus, the use of beta-lactam-aminoglycoside combination therapy for sepsis should be discouraged.

Authors' conclusions

Implications for practice

Clinicians usually face the dilemma of selecting an antibiotic treatment on two occasions during an un-complicated infectious episode. On the initial encounter with a patient the clinician must prescribe empirical antibiotic treatment, since the causative pathogen and its susceptibilities are generally unknown. Most studies addressed this situation, and the results show that there is no difference in overall mortality whether monotherapy or combination therapy is used. Adverse effects, most significantly nephrotoxicity, will occur more frequently with combination therapy. If the choice is between a narrower-spectrum beta-lactam combined with an aminoglycoside versus a broad-spectrum beta-lactam, our results show that treatment will ultimately have to be modified more frequently if the combination is chosen. We have not identified a specific site of infection, or disease severity, where combination treatment has an advantage.

The second decision point occurs when the causative pathogen is identified. Here, the choice of the antibiotic treatment is dictated by known susceptibility results. However, the question remains, whether for specific bacteria beta-lactam-aminoglycoside combination treatment offers an advantage over single beta-lactam treatment. We addressed this question by subgroup analyses of patients with documented infections caused by specific pathogens (Gram-negatives, Pseudomonas aeruginosa, Staphylococcus aureus). In addition, several semi-empirical studies addressed this question specifically. We have not identified a specific pathogen, or pathogen group, where combination therapy is advantageous.

Overall, appropriate beta-lactam monotherapy should be used. Beta-lactam-aminoglycoside combination therapy does not offer an advantage, and is associated with an increased rate of adverse events.

Implications for research

We cannot point to a specific patient subgroup that showed a trend for benefit with combination therapy. The design of existing studies did not permit a comparison between monotherapy and combination therapy for specific pathogens when all the antibiotics administered matched the in-vitro susceptibility of the pathogen. However the large body of studies that were performed did not point towards any benefit. Thus we do not see the justification for such future trials.

Exceptions to this are trials addressing patients with endocarditis. Prolonged combination treatment for endocarditis, including an aminoglycoside, is well accepted in clinical practice, but does not seem grounded in clinical evidence. Future trials must examine the justification for this practice.

Further comparisons between monotherapy and combination, if performed, should be limited to comparisons involving the same beta-lactam. This is the only design that explores the benefit of beta-lactam-aminoglycoside combination therapy. Studies comparing broad-spectrum monotherapy, such as new antibiotics, to an older generation beta-lactam with an aminoglycoside should not be performed. Patients may be harmed by combination therapy in such trials.

Appropriate antibiotic treatment has been shown to significantly reduce mortality, and should therefore be reported with results adjusted to it. Outcomes relevant to patients, such as survival and hospitalisation duration should be assessed. Survival, if not assessed as a primary outcome, must at least be reported.

Acknowledgements

We would like to thank all the authors who responded to our requests for additional data (see 'unpublished data' and 'unpublished data sought but not used', 'References to studies'). Dr Solomkin (Solomkin 1986) and Dr. Sexton (Sexton 1984) supplied supplementary data for their studies, which were not included in the review. Dr. Finer and Dr. Goustas of the GlaxoSmithKline Company supplied detailed data for their study (Finer 1992). Dr Kora Huber sent completed trial results for Kljucar 1990 and supplied requested additional information. Ms Mary Forrest (Managing editor, Journal of Chemotherapy), sent several publications that were not available to us. We would also like to warmly thank Ms Rika Fujiya who translated the Japanese studies (Sukoh 1994; Takamoto 1994).

We thank Dr Vittoria Lutje, Dr Harriet G. MacLehose, and Ms Rieve Robb (Review Group Co-ordinator) of the Cochrane Infectious Diseases Group. We thank Dr Harald Herkner, Prof. Nathan Pace, Kathie Godfrey, Janet Wale and Jane Cracknell (Review Group Co-ordinator) of the Cochrane Anaesthesia Review Group. Both groups supported and provided helpful revisions for this review.

This review was initially developed within the Infectious Diseases Group and supported by a grant from the Department for International Development, UK. The review was transferred to the Anaesthesia Group in May 2005.

Data and analyses

Download statistical data

Comparison 1. Monotherapy versus combination therapy
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 All cause fatality43 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
1.1 Same BL121381Risk Ratio (M-H, Fixed, 95% CI)1.01 [0.75, 1.35]
1.2 Different BL314146Risk Ratio (M-H, Fixed, 95% CI)0.85 [0.71, 1.01]
2 All cause fatality by study groups40 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
2.1 Same sepsis6789Risk Ratio (M-H, Fixed, 95% CI)1.15 [0.79, 1.67]
2.2 Same abdominal2331Risk Ratio (M-H, Fixed, 95% CI)0.91 [0.54, 1.55]
2.3 Same UTI173Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]
2.4 Different sepsis213298Risk Ratio (M-H, Fixed, 95% CI)0.83 [0.69, 0.99]
2.5 Different abdominal6550Risk Ratio (M-H, Fixed, 95% CI)1.09 [0.56, 2.15]
2.6 Different UTI4298Risk Ratio (M-H, Fixed, 95% CI)1.33 [0.34, 5.21]
3 All cause fatality (Gram negative infections)8 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
3.1 Same BL3117Risk Ratio (M-H, Fixed, 95% CI)0.56 [0.08, 4.07]
3.2 Different BL5313Risk Ratio (M-H, Fixed, 95% CI)1.25 [0.80, 1.95]
4 All cause fatality (Gram negative bacteremia)5 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
4.1 Same BL385Risk Ratio (M-H, Fixed, 95% CI)1.62 [0.30, 8.75]
4.2 Different BL2125Risk Ratio (M-H, Fixed, 95% CI)1.31 [0.63, 2.70]
5 All cause fatality (non urinary tract infections)16 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
5.1 Same BL3351Risk Ratio (M-H, Fixed, 95% CI)0.88 [0.53, 1.47]
5.2 Different BL131458Risk Ratio (M-H, Fixed, 95% CI)0.70 [0.52, 0.95]
Analysis 1.1.

Comparison 1 Monotherapy versus combination therapy, Outcome 1 All cause fatality.

Analysis 1.2.

Comparison 1 Monotherapy versus combination therapy, Outcome 2 All cause fatality by study groups.

Analysis 1.3.

Comparison 1 Monotherapy versus combination therapy, Outcome 3 All cause fatality (Gram negative infections).

Analysis 1.4.

Comparison 1 Monotherapy versus combination therapy, Outcome 4 All cause fatality (Gram negative bacteremia).

Analysis 1.5.

Comparison 1 Monotherapy versus combination therapy, Outcome 5 All cause fatality (non urinary tract infections).

Comparison 2. Monotherapy versus combination therapy
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Clinical failure63 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
1.1 Same BL201870Risk Ratio (M-H, Fixed, 95% CI)1.11 [0.95, 1.29]
1.2 Different BL434746Risk Ratio (M-H, Fixed, 95% CI)0.77 [0.69, 0.86]
2 Clinical failure by study groups58 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
2.1 Same sepsis121196Risk Ratio (M-H, Fixed, 95% CI)1.25 [1.01, 1.55]
2.2 Same abdominal2308Risk Ratio (M-H, Fixed, 95% CI)1.03 [0.80, 1.32]
2.3 Same UTI161Risk Ratio (M-H, Fixed, 95% CI)0.98 [0.46, 2.09]
2.4 Different sepsis293612Risk Ratio (M-H, Fixed, 95% CI)0.75 [0.66, 0.84]
2.5 Different abdominal9675Risk Ratio (M-H, Fixed, 95% CI)0.82 [0.59, 1.13]
2.6 Different UTI5459Risk Ratio (M-H, Fixed, 95% CI)1.12 [0.65, 1.91]
3 Bacteriological failure - all43 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
3.1 Same BL14751Risk Ratio (M-H, Fixed, 95% CI)1.15 [0.88, 1.51]
3.2 Different BL292760Risk Ratio (M-H, Fixed, 95% CI)0.81 [0.69, 0.94]
4 UTI relapse or re-infection6458Risk Ratio (M-H, Fixed, 95% CI)1.01 [0.61, 1.67]
5 Clinical failure (Gram negative infections)28 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
5.1 Same BL10432Risk Ratio (M-H, Fixed, 95% CI)1.23 [0.90, 1.68]
5.2 Different BL181403Risk Ratio (M-H, Fixed, 95% CI)0.85 [0.66, 1.09]
6 Clinical failure (Gram negative bacteremia)11 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
6.1 Same BL4101Risk Ratio (M-H, Fixed, 95% CI)1.07 [0.45, 2.56]
6.2 Different BL7198Risk Ratio (M-H, Fixed, 95% CI)0.75 [0.38, 1.48]
7 Clinical failure (Pseudomonas aeruginosa infections)18 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
7.1 Same BL6124Risk Ratio (M-H, Fixed, 95% CI)1.02 [0.68, 1.51]
7.2 Different BL12302Risk Ratio (M-H, Fixed, 95% CI)1.20 [0.80, 1.82]
8 Clinical failure (bacteremia)22 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
8.1 Same BL5141Risk Ratio (M-H, Fixed, 95% CI)1.43 [0.77, 2.66]
8.2 Different BL17624Risk Ratio (M-H, Fixed, 95% CI)0.64 [0.46, 0.89]
9 Clinical failure (urinary tract infections)17 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
9.1 Same BL484Risk Ratio (M-H, Fixed, 95% CI)1.12 [0.59, 2.13]
9.2 Different BL13708Risk Ratio (M-H, Fixed, 95% CI)1.22 [0.80, 1.87]
10 Clinical failure (non urinary tract infections)41 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
10.1 Same BL101248Risk Ratio (M-H, Fixed, 95% CI)1.18 [0.99, 1.42]
10.2 Different BL312945Risk Ratio (M-H, Fixed, 95% CI)0.70 [0.61, 0.81]
Analysis 2.1.

Comparison 2 Monotherapy versus combination therapy, Outcome 1 Clinical failure.

Analysis 2.2.

Comparison 2 Monotherapy versus combination therapy, Outcome 2 Clinical failure by study groups.

Analysis 2.3.

Comparison 2 Monotherapy versus combination therapy, Outcome 3 Bacteriological failure - all.

Analysis 2.4.

Comparison 2 Monotherapy versus combination therapy, Outcome 4 UTI relapse or re-infection.

Analysis 2.5.

Comparison 2 Monotherapy versus combination therapy, Outcome 5 Clinical failure (Gram negative infections).

Analysis 2.6.

Comparison 2 Monotherapy versus combination therapy, Outcome 6 Clinical failure (Gram negative bacteremia).

Analysis 2.7.

Comparison 2 Monotherapy versus combination therapy, Outcome 7 Clinical failure (Pseudomonas aeruginosa infections).

Analysis 2.8.

Comparison 2 Monotherapy versus combination therapy, Outcome 8 Clinical failure (bacteremia).

Analysis 2.9.

Comparison 2 Monotherapy versus combination therapy, Outcome 9 Clinical failure (urinary tract infections).

Analysis 2.10.

Comparison 2 Monotherapy versus combination therapy, Outcome 10 Clinical failure (non urinary tract infections).

Comparison 3. Monotherapy versus combination therapy
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Bacterial superinfections273085Risk Ratio (M-H, Fixed, 95% CI)0.76 [0.57, 1.01]
2 Fungal superinfections111119Risk Ratio (M-H, Fixed, 95% CI)0.79 [0.42, 1.48]
3 Bacterial colonization141635Risk Ratio (M-H, Fixed, 95% CI)0.85 [0.65, 1.10]
4 Fungal colonization71132Risk Ratio (M-H, Fixed, 95% CI)1.39 [0.93, 2.09]
5 Bacterial colonization - surveillance cultures6751Risk Ratio (M-H, Fixed, 95% CI)0.78 [0.60, 1.01]
6 Bacterial resistance development91370Risk Ratio (M-H, Fixed, 95% CI)0.88 [0.54, 1.45]
Analysis 3.1.

Comparison 3 Monotherapy versus combination therapy, Outcome 1 Bacterial superinfections.

Analysis 3.2.

Comparison 3 Monotherapy versus combination therapy, Outcome 2 Fungal superinfections.

Analysis 3.3.

Comparison 3 Monotherapy versus combination therapy, Outcome 3 Bacterial colonization.

Analysis 3.4.

Comparison 3 Monotherapy versus combination therapy, Outcome 4 Fungal colonization.

Analysis 3.5.

Comparison 3 Monotherapy versus combination therapy, Outcome 5 Bacterial colonization - surveillance cultures.

Analysis 3.6.

Comparison 3 Monotherapy versus combination therapy, Outcome 6 Bacterial resistance development.

Comparison 4. Monotherapy versus combination therapy
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Any adverse event394945Risk Ratio (M-H, Fixed, 95% CI)0.92 [0.83, 1.01]
2 Adverse events requiring treatment discontinuation193042Risk Ratio (M-H, Random, 95% CI)0.89 [0.52, 1.52]
3 Any nephrotoxicity455213Risk Ratio (M-H, Fixed, 95% CI)0.30 [0.23, 0.39]
3.1 Once daily aminoglycoside5865Risk Ratio (M-H, Fixed, 95% CI)0.17 [0.06, 0.53]
3.2 Twice daily aminoglycoside71127Risk Ratio (M-H, Fixed, 95% CI)0.43 [0.24, 0.77]
3.3 Thrice daily aminoglycoside232082Risk Ratio (M-H, Fixed, 95% CI)0.28 [0.20, 0.39]
3.4 Non specified aminoglycoside regimen101139Risk Ratio (M-H, Fixed, 95% CI)0.34 [0.19, 0.58]
Analysis 4.1.

Comparison 4 Monotherapy versus combination therapy, Outcome 1 Any adverse event.

Analysis 4.2.

Comparison 4 Monotherapy versus combination therapy, Outcome 2 Adverse events requiring treatment discontinuation.

Analysis 4.3.

Comparison 4 Monotherapy versus combination therapy, Outcome 3 Any nephrotoxicity.

Comparison 5. Monotherapy versus combination therapy
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Drop-outs for all cause fatality8910Risk Ratio (M-H, Fixed, 95% CI)1.00 [0.66, 1.49]
1.1 Same BL4541Risk Ratio (M-H, Fixed, 95% CI)1.08 [0.55, 2.11]
1.2 Different BL4369Risk Ratio (M-H, Fixed, 95% CI)0.95 [0.57, 1.58]
2 Drop-outs for clinical failure243631Risk Ratio (M-H, Fixed, 95% CI)1.04 [0.88, 1.23]
2.1 Same BL101244Risk Ratio (M-H, Fixed, 95% CI)1.26 [0.92, 1.72]
2.2 Different BL142387Risk Ratio (M-H, Fixed, 95% CI)0.96 [0.78, 1.17]
Analysis 5.1.

Comparison 5 Monotherapy versus combination therapy, Outcome 1 Drop-outs for all cause fatality.

Analysis 5.2.

Comparison 5 Monotherapy versus combination therapy, Outcome 2 Drop-outs for clinical failure.

Comparison 6. Monotherapy versus combination therapy
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 All cause fatality (Gram positive infections)3188Risk Ratio (M-H, Fixed, 95% CI)0.44 [0.12, 1.58]
2 Clinical failure (Gram positive infections)5305Risk Ratio (M-H, Fixed, 95% CI)0.69 [0.40, 1.19]
3 Bacteriological failure (Gram positive infections)5300Risk Ratio (M-H, Fixed, 95% CI)0.89 [0.47, 1.69]
4 Need for operation (endocarditis)4243Risk Ratio (M-H, Fixed, 95% CI)0.76 [0.41, 1.39]
Analysis 6.1.

Comparison 6 Monotherapy versus combination therapy, Outcome 1 All cause fatality (Gram positive infections).

Analysis 6.2.

Comparison 6 Monotherapy versus combination therapy, Outcome 2 Clinical failure (Gram positive infections).

Analysis 6.3.

Comparison 6 Monotherapy versus combination therapy, Outcome 3 Bacteriological failure (Gram positive infections).

Analysis 6.4.

Comparison 6 Monotherapy versus combination therapy, Outcome 4 Need for operation (endocarditis).

Comparison 7. Monotherapy versus combination therapy (sensitivity analyses)
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 All cause fatality by allocation concealment43 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
1.1 A same BL61068Risk Ratio (M-H, Fixed, 95% CI)0.96 [0.71, 1.31]
1.2 B same BL6313Risk Ratio (M-H, Fixed, 95% CI)1.56 [0.58, 4.18]
1.3 A different BL122154Risk Ratio (M-H, Fixed, 95% CI)0.95 [0.75, 1.19]
1.4 B different BL181952Risk Ratio (M-H, Fixed, 95% CI)0.70 [0.53, 0.93]
1.5 C different BL140Risk Ratio (M-H, Fixed, 95% CI)1.33 [0.34, 5.21]
2 All cause fatality by allocation generation43 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
2.1 A same BL61068Risk Ratio (M-H, Fixed, 95% CI)0.96 [0.71, 1.31]
2.2 B same BL6313Risk Ratio (M-H, Fixed, 95% CI)1.56 [0.58, 4.18]
2.3 A different BL192957Risk Ratio (M-H, Fixed, 95% CI)0.89 [0.72, 1.09]
2.4 B different BL111149Risk Ratio (M-H, Fixed, 95% CI)0.72 [0.50, 1.04]
2.5 C different BL140Risk Ratio (M-H, Fixed, 95% CI)1.33 [0.34, 5.21]
3 All cause fatality by ITT vs. per-protocol analysis43 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
3.1 ITT - same BL (type 1 studies)4469Risk Ratio (M-H, Fixed, 95% CI)0.62 [0.27, 1.43]
3.2 per-protocol - same BL (type 2 and 3 studies)6761Risk Ratio (M-H, Fixed, 95% CI)1.09 [0.80, 1.51]
3.3 unknown - same BL (type 4 studies)2151Risk Ratio (M-H, Fixed, 95% CI)0.88 [0.06, 13.25]
3.4 ITT - different BL (type 1 studies)152989Risk Ratio (M-H, Fixed, 95% CI)0.87 [0.71, 1.07]
3.5 per-protocol - different BL (type 2 and 3 studies)121037Risk Ratio (M-H, Fixed, 95% CI)0.76 [0.54, 1.07]
3.6 unknown - different BL (type 4 studies)4120Risk Ratio (M-H, Fixed, 95% CI)1.33 [0.34, 5.21]
4 Clinical failure by allocation concealment63 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
4.1 A same BL81138Risk Ratio (M-H, Fixed, 95% CI)1.11 [0.93, 1.32]
4.2 B same BL12732Risk Ratio (M-H, Fixed, 95% CI)1.09 [0.79, 1.50]
4.3 A different BL132028Risk Ratio (M-H, Fixed, 95% CI)0.72 [0.60, 0.86]
4.4 B different BL282604Risk Ratio (M-H, Fixed, 95% CI)0.79 [0.68, 0.92]
4.5 C different BL2114Risk Ratio (M-H, Fixed, 95% CI)1.09 [0.63, 1.88]
5 Clinical failure by allocation generation63 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
5.1 A same BL91319Risk Ratio (M-H, Fixed, 95% CI)1.09 [0.91, 1.29]
5.2 B same BL11551Risk Ratio (M-H, Fixed, 95% CI)1.18 [0.83, 1.69]
5.3 A different BL253217Risk Ratio (M-H, Fixed, 95% CI)0.76 [0.66, 0.88]
5.4 B different BL161415Risk Ratio (M-H, Fixed, 95% CI)0.75 [0.62, 0.92]
5.5 C different BL2114Risk Ratio (M-H, Fixed, 95% CI)1.09 [0.63, 1.88]
6 Clinical failure by blinding63 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
6.1 Non-blinded - same BL191666Risk Ratio (M-H, Fixed, 95% CI)1.12 [0.93, 1.35]
6.2 Any blinding - same BL1204Risk Ratio (M-H, Fixed, 95% CI)1.06 [0.82, 1.37]
6.3 Non-blinded - different BL373809Risk Ratio (M-H, Fixed, 95% CI)0.82 [0.72, 0.94]
6.4 Any blinding - different BL6937Risk Ratio (M-H, Fixed, 95% CI)0.62 [0.50, 0.77]
7 Clinical failure by ITT versus per-protocol analysis63 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
7.1 ITT - same BL (type 1)2110Risk Ratio (M-H, Fixed, 95% CI)0.78 [0.43, 1.40]
7.2 ITT assuming failure for drop-outs - same BL (type 2)9902Risk Ratio (M-H, Fixed, 95% CI)1.32 [1.09, 1.60]
7.3 Per protocol - same BL (type 3 studies)4580Risk Ratio (M-H, Fixed, 95% CI)1.10 [0.91, 1.33]
7.4 Type 4 studies - same BL5278Risk Ratio (M-H, Fixed, 95% CI)0.95 [0.56, 1.61]
7.5 ITT - different BL (type 1)111458Risk Ratio (M-H, Fixed, 95% CI)0.76 [0.64, 0.92]
7.6 ITT assuming failure for drop-outs - different BL (type 2)142065Risk Ratio (M-H, Fixed, 95% CI)0.83 [0.73, 0.94]
7.7 Per protocol - different BL (type 3 studies)121031Risk Ratio (M-H, Fixed, 95% CI)0.76 [0.62, 0.95]
7.8 Type 4 studies - different BL6192Risk Ratio (M-H, Fixed, 95% CI)0.79 [0.40, 1.56]
Analysis 7.1.

Comparison 7 Monotherapy versus combination therapy (sensitivity analyses), Outcome 1 All cause fatality by allocation concealment.

Analysis 7.2.

Comparison 7 Monotherapy versus combination therapy (sensitivity analyses), Outcome 2 All cause fatality by allocation generation.

Analysis 7.3.

Comparison 7 Monotherapy versus combination therapy (sensitivity analyses), Outcome 3 All cause fatality by ITT vs. per-protocol analysis.

Analysis 7.4.

Comparison 7 Monotherapy versus combination therapy (sensitivity analyses), Outcome 4 Clinical failure by allocation concealment.

Analysis 7.5.

Comparison 7 Monotherapy versus combination therapy (sensitivity analyses), Outcome 5 Clinical failure by allocation generation.

Analysis 7.6.

Comparison 7 Monotherapy versus combination therapy (sensitivity analyses), Outcome 6 Clinical failure by blinding.

Analysis 7.7.

Comparison 7 Monotherapy versus combination therapy (sensitivity analyses), Outcome 7 Clinical failure by ITT versus per-protocol analysis.

Appendices

Appendix 1. Search strategy

Type of patientsInterventionsRCT filter
pneumonia* OR(aminoglycoside* ORrandom* OR
infection* OR infect* ORnetilmicin* ORcontrol* OR
sepsis OR septic?emia* ORgentamicin* ORsingle OR double OR blind* OR
bacter* OR bacter?emia*amikacin* ORplacebo OR
 tobramycin* ORclinical OR
 streptomycin* ORcomparative OR
 isepamicin* ORprospectiv*
 sisomicin*) 

Feedback

Obtaining data on all-cause mortality, 16 June 2013

Summary

Thank you for taking on the large amount of data surrounding topic of aminoglycoside and beta-lactam combination therapy in the treatment of sepsis. With the large volume of studies spanning such a long time period this was no small task. With this is mind we still have some one question regarding the primary outcome analysis of all cause mortality between the two treatment arms.

The primary mortality analysis contained 43 of the total 64 studies included in the review and they were split into two separate subgroups, using wither the same or a different beta-lactam agent as monotherapy as in combination therapy. Our concern is centered on the outstanding 21 studies not included in this analysis. We were wondering what attempts were made to collect mortality data from these remaining trials questioning whether the inclusion of those results would statistically alter the outcomes. We understand that a number of these studies were completed over 40 years ago and the data may be very difficult to obtain.

In the subgroup where a different beta-lactam was used the risk of mortality was non-significantly lowered in the monotherapy arm RR 0.85 (95% CI 0.71, 1.01). With the results being close to statistical significance we were wondering if the addition of data from the outstanding studies would actually make a statistical difference. If this were truly the case then your conclusion of “The addition of an aminoglycoside to beta-lactams for sepsis should be discouraged. All-cause fatality rates are unchanged. Combination treatment carries a significant risk of nephrotoxicity” would change and the call to avoid the use of these antibiotics would be much stronger.

We also pooled all of the data from both subgroups (same and different beta-lactams) and found that it did not change the results of the different beta-lactam group but greatly narrowed the confidence interval of the same beta-lactam group with a RR of 1.13 (0.97, 1.31) - increased risk of mortality in the combination group versus monotherapy. With this analysis we also found very little heterogeneity between same and different beta-lactam studies (I2 = 8%).

We understand the beta-lactam agent selected, specifically in regards the spectrum of activity, greatly impacts the effects of empirical therapy but with this potentially increased risk of mortality that is consistent across this large number of studies leads us to believe that although statistically non-significant it seems plausible that the risk of mortality with combination therapy over beta-lactam monotherapy is real.

We also believe that the possibility of “emotional based medicine” is real in this patient population. As the majority of these studies were open-label despite being randomized, it is not unlikely that “sicker” patients would receive more drugs (i.e. combination therapy). If this were true then it is plausible that patients who were more likely to die received more antibiotics and were in the combination groups but with the effect remaining relatively consistent across this large number of studies, we feel that a true risk may actually exist.

After this long discussion, our question returns to whether or not mortality data is available from the remaining 21 studies and what attempts have been made to retrieve this information. A statistically significant increase in mortality, along with the increase in adverse events see with combination therapy would likely facilitate a rapid change in practice and removal of this therapeutic option. Just as an exercise we inserted the data provided by your review into Review Manager to test how many events it would take to make the difference in mortality. We understand this is not a truly scientific exercise but one based on curiosity.

What we found was that when we added two events (deaths) to the combination group in the most heavily weighted study (Felisart 1985) the outcome of mortality became statistically significantly higher in the combination group. We also combined all of the data between the same different beta-lactam subgroups and found that only 8 more deaths in the combination group made the entire analysis (all 43 studies) statistically significant for an increase in mortality in the combination group. On the flip side, it took 80 events in the monotherapy group to swing the analysis the other way and statistically favour combination therapy in the outcome of mortality.

Thank you for your time

Reply

Dear Dr. Amadio,

Thank you for your kind attention to our work and your input for the data analysis.

In response to your question regarding obtaining data on all-cause mortality, we mailed all authors of trials that did not report on this outcome asking for the data, as is routine in Cochrane reviews. We agree with you on the importance of the missing data on mortality and for this reason we made extra efforts to obtain the data. If we did not establish contact with the corresponding author, we tried to contact a second and third author. The data presented in our review are the result of this process and still we miss mortality data from a third of all randomized controlled trials (RCTs) that were conducted.

Selection bias should not occur in adequately conducted RCTs, those using appropriate allocation concealment. Allocation concealment is the procedure ensuring that no one is aware of the treatment assignment when the patient is recruited into the trial and before the patient is allocated to an intervention. We observed in our review that the advantage to combination therapy was larger in trials with unclear methods for allocation concealment (studies not reporting the methods for this procedure) compared to trials that used methods ensuring adequate allocation concealment. Therefore, it is possible that results were affected by selection of sicker patients to the combination therapy group. However the difference between trials with low and unclear risk of bias was not statistically significant and we have no actual data on whether bias could occur in the trials with unclear risk of bias. Most importantly to our view, the trials comparing different beta-lactams usually compared a new, broad-spectrum beta-lactam to an old, classical regimen; we believe that if selection bias crept in to some trials it would have worked in the opposite direction of recruiting the sicker patients to the novel monotherapy arm. The fact that most of the trials were open, might have led to a different type of bias, and dilution of effects, because physicians could add an aminoglycoside to failing patients in the monotherapy arm, while this could not occur in the combination therapy arm.

Methods exist to formally examine the possible effects of missing data in meta-analysis. We will consider adding such an analysis to an update of our review. More importantly, we will highlight the issue of missing data on all-cause mortality. Should your important correspondence result in any authors sending further data from their trials on mortality, these will be added to our review.

Contributors

Anthony Amadio, BSc. Pharm, ACPR, RPh

Doctor of Pharmacy Student

Faculty of Pharmaceutical Sciences

University of British Columbia

Vancouver BC

Canada

Aaron M Tejani, BSc Pharm, PharmD
Researcher
Therapeutics Initiative, University of British Columbia
2176 Health Sciences Mall
Vancouver, BC, Canada

Reply

Mical Paul, corresponding author

What's new

DateEventDescription
14 August 2013AmendedAdditional contributor added to feedback section.

History

Protocol first published: Issue 4, 2001
Review first published: Issue 1, 2006

DateEventDescription
13 August 2013Feedback has been incorporated

Feedback submitted and responded to.

Two Cochrane references updated and typos corrected.

2 September 2008AmendedConverted to new review format.

Contributions of authors

Mical Paul (MP): Performed the search and scanned abstracts; retrieved full-text articles and applied inclusion and exclusion criteria; performed quality assessment, data extraction,and analysis. MP communicated with authors; wrote protocol and review.

Ishay Silbiger (IS): Applied inclusion and exclusion criteria, and performed quality assessment, data extraction and analysis.

Simona Grozinsky (SG): Extracted the data

Karla Soares-Weiser (KSW): Assisted with inclusion and exclusion of studies; performed quality assessment, data extraction and analysis; assisted with the writing and reviewed all versions of protocol and review.

Leonard Leibovici (LL): Assisted with inclusion and exclusion of studies; performed quality assessment, data extraction and analysis; assisted with communication with authors; assisted with the writing and reviewed all versions of protocol and review.

Declarations of interest

We certify that we have no affiliations with or involvement in any organization or entity with a direct financial interest in the subject matter of this review (e.g. employment, consultancy, stock ownership, honoraria, expert testimony).

Sources of support

Internal sources

  • Rabin Medical Center - Beilison Campus, Israel.

External sources

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

  • Department for International Development, UK.

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Abrams 1979

MethodsRCT
Empirical and semi-empirical
Gram positive infections
Participants36 IV drug users with suspected Staphylococcal endocarditis were included. Only those with Staphylococcus aureus bacteraemia and endocarditis according to inclusion criteria were evaluated Patients excluded because they did not fulfil inclusion criteria for bacteraemia were not considered as dropouts for the review
InterventionsOxacillin 12gr/d vs. oxacillin 12gr/d + gentamicin 80mgX3 (gentamicin administered for the first 2 weeks of a 4-week treatment protocol)
OutcomesOverall mortality
Treatment failure (clinical and bacteriological)
Adverse events
Duration of fever
NotesUSA
Outcomes in subgroups: Bacteraemia.
Cephalothin was permitted instead of oxacillin for patients with penicillin allergy, and oxacillin was replaced by penicillin for penicillin-susceptible Staphylococcus aureus.
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Aguilar 1992

MethodsRCT
Sepsis
Participants36 patients > 16 yrs. with severe infections
InterventionsCeftizoxime 60-150 mg/kg/d vs. penicillin 20-30mU/d + gentamicin 3-5mg/kg/d
OutcomesTreatment failure (clinical and bacteriological)
NotesMexico
No outcomes in subgroups
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Alvarez-Lerma 2001a

MethodsRCT
Sepsis
Participants140 adult patients hospitalized in the ICU, mechanically ventilated and diagnosed with pneumonia. All infections were hospital acquired. 66% of patients were on inotropic drugs upon entry to study
InterventionsMeropenem 1grX3 for 9.3 days vs.
ceftazidime 2grX3 + amikacin 7.5mg/kgX2 for 8.3 days
OutcomesOverall mortality
Treatment failure (clinical and bacteriological)
Bacterial superinfections
Adverse events
Duration of treatment
NotesMulticentre
Spain
Outcomes in subgroups: Gram negative and Pseudomonas sp. infections
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Low riskA - Adequate

Arich 1987

MethodsRCT
Partially semi-empirical
Sepsis
ParticipantsAdult patients with enterobacteriacae bacteraemia (at least 2 positive blood cultures with same pathogen). Patients could enter the trial before or at diagnosis of bacteraemia
InterventionsCefotaxime 1grX3-4 for 17.5 days vs. cefazolin 1grX3 + tobramycin 1.5mg/kgX3 for 10 days
OutcomesOverall mortality
Treatment failure (clinical and bacteriological)
Superinfection
Adverse events
Duration of hospitalizations, treatment and fever
NotesFrance (French)
Outcomes in subgroups:
Bacteraemia
Gram-negative infections
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Low riskA - Adequate

Bergeron 1988

MethodsRCT
Abdominal
Participants77 adult patients with severe biliary tract infections (cholecystitis, cholangitis and necrotizing cholecystitis)
InterventionsCefoperazone 2grX2 for 7.2 days vs. ampicillin 1grX4 + tobramycin 1.5mg/kgX3 following loading dose 2mg/kg for 6.8 days (Surgery in addition to medical treatment was performed in 28/36 monotherapy patients and in 19/29 combination patients, not counted as failure)
OutcomesOverall mortality
Treatment failure (clinical and bacteriological)
Superinfections
Colonization
Treatment duration
Dropouts
Adverse events
NotesMulticentre
Canada
Outcomes in subgroups:
Bacteraemia
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Biglino 1991

MethodsRCT
Sepsis
Participants22 patients with severe infections. Patients were compromised by background diseases, including some immune-
compromise in 73%. Randomized to 4 arms monotherapy vs. combination, and high vs. low dose of imipenem
InterventionsImipenem 0.5-1grX4 vs. imipenem 0.5-1grX4 + netilmicin 5mg/kg
OutcomesTreatment failure (clinical)
Adverse events
Duration of fever and hospital stay
NotesItaly
No outcomes in subgroups
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Brown 1984

MethodsRCT
Sepsis
Participants48 adult patients (34 evaluated) with hospital acquired pneumonia of a documented Gram-negative origin (By sputum's Gram stain or cultures). 85% (29/34) acquired infection in the ICU
InterventionsMoxalactam 2grX3 for 10.1 days vs.
carbenicillin 66mg/kgX6 + tobramycin 1.7mg/kgX3 (following a 2-2.5mg/kg loading dose) for 10.6 days
OutcomesOverall mortality
Treatment failure (x-ray non-clearing)
Superinfections
Adverse events
Duration of treatment
NotesUSA
Outcomes in subgroups:
Gram-negative and Pseudomonas sp. infections
4 deaths among 11 excluded patients not included in outcome assessment
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Carbon 1987

MethodsRCT
Probably semi-empirical
Sepsis
Participants74 patients with bacteraemia due to enterobacteriaceae, with at least 3 positive blood cultures entered the study
InterventionsCefotaxime 1grX4 for 12.9 days vs. cefotaxime 1grX4 + amikacin 7.5mg/kg loading dose followed by a renal-function adjusted maintenance dose for 13.2 days
OutcomesOverall mortality
Treatment failure (clinical)
Superinfections
Adverse events
Duration of treatment and fever
NotesMulticentre
France
Outcomes in subgroups:
Gram negative infections
Bacteraemia
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Cardozo 2001

MethodsRCT
Abdominal
Participants110 children <15 years, with acute appendicitis
InterventionsAmoxycillin-sulbactam 33mg/kgX3 vs. amoxycillin
-sulbactam 33mg/kgX3 + gentamicin 5mg/kgX1
OutcomesOverall mortality
Treatment failure
NotesParaguay (Spanish)
No outcomes in subgroups
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Cometta 1994

MethodsRCT
Sepsis
Participants313 adult patients with nosocomial pneumonia, nosocomial sepsis or severe diffuse peritonitis. 73% were in ICU and 48% on mechanical ventilation
InterventionsImipenem 500mgX4 for 10.2 days vs. imipenem 500mgX4 + netilmicin 150mgX2 for 10.5 days
OutcomesOverall mortality
Treatment failure (clinical)
Superinfections
Colonization
Adverse events
Duration of treatment
NotesMulticentre
Switzerland
Outcomes in subgroups: Gram-negative and Pseudomonas sp. infections
A secondary reference, Iten 1992, described 71 patients from this study, for whom surveillance cultures were performed, and detailed data concerning resistance development are given
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Low riskA - Adequate

Cone 1985

MethodsRCT
Sepsis
Participants57 hospitalized patients with pneumonia or bacteraemia. Pneumonia was community acquired or nosocomial. Only patients with positive bacteriological cultures were evaluated
InterventionsCeftazidime 2grX3 vs. ticarcillin 3grX4 + tobramycin 1mg/kgX3
OutcomesOverall mortality
Treatment failure (clinical)
Superinfections
Adverse events
NotesUSA
Outcomes in subgroups: Bacteraemia
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Coppens 1983

MethodsRCT
Semi-empirical
Gram positive infections
Participants80 patients in whom staphylococcal infections were clinically and microbiologically suspected. Inclusion criteria mandated a positive Gram stain showing Staphylococci Patients were randomized to the designated interventions. 24-48 hours following randomisation, patients with documented methicillin-
resistant Staphylococci were switched to vancomycin, only in the monotherapy group (N=14). These were excluded from analysis in the review
InterventionsCefamandole 2grX3 vs.
cefamandole 2grX3 + tobramycin 80mgX3
OutcomesTreatment failure
(clinical and bacteriological)
Bacterial superinfection and colonization
NotesBelgium
Outcomes in subgroups: Bacteraemia
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

D'Antonio 1992

MethodsRCT
Sepsis
ParticipantsNon-neutropenic adult patients with altered immune defence, with fever > 38 lasting > 8 hours. 88% of patients with underlying haematological malignancy
InterventionsCeftriaxone 2grX1 for a median of 12 days vs. ceftriaxone 2grX1 + amikacin 5mg/kgX3 for a median of 11 days
OutcomesOverall mortality
Treatment failure
(clinical and bacteriological)
Superinfection and colonization (bacterial and fungal)
Adverse events
Treatment duration
NotesItaly
Outcomes in subgroups:
Gram-negative and Pseudomonas sp. infections Bacteraemia
Urinary tract infection
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Low riskA - Adequate

Duff 1982

MethodsQuasi-randomized
Abdominal
Participants74 patients included who developed endomyo-
parametritis after caesarian section or vaginal delivery, or who developed pelvic cellulitis after hysterectomy
InterventionsCefoxitin 2grX3
vs. penicillin 5millUX4 + gentamicin 60-80mgX3
OutcomesOverall mortality
Treatment failure
Adverse events
Dropouts
NotesUSA
Outcomes in subgroups: Gram-negative infections
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)High riskC - Inadequate

Dupont 2000

MethodsRCT
Abdominal
Participants227 patients evaluated with severe generalized peritonitis Modified ITT analysis was performed on 204 patients with surgically proven severe intra-abdominal infections
InterventionsPiperacillin- tazobactam 4grX4 for 8.2 days vs.
piperacillin- tazobactam 4grX4 + amikacin 7.5mg/kgX2 for 8.6 days. In addition all patients were operated on
OutcomesOverall mortality
Treatment failure (clinical)
Adverse events
Dropouts
Treatment duration
NotesMulticentre
France
No outcomes in subgroups
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Low riskA - Adequate

Felisart 1985

MethodsRCT
Sepsis
Participants73 adult patients with underlying advanced cirrhosis, presenting with severe bacterial infections. Most patients had spontaneous bacterial peritonitis
InterventionsCefotaxime 2grX6 vs.
ampicillin 2grX6 + tobramycin renal adjusted maintenance dose X3/d following 1.75mg/kg loading dose
OutcomesOverall mortality
Treatment failure (clinical)
Superinfections
Adverse events
NotesSpain
Outcomes in subgroups: Bacteraemia
Urinary tract infections
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Finer 1992

MethodsRCT
Sepsis
Participants471 adult patients hospitalized with signs and symptoms of serious bacterial infections, thought by the physician to require parenteral antibiotic treatment
InterventionsCeftazidime 2grX2 vs. ureidopenillin + aminoglycoside used routinely in specific Center: piperacillin-
gentamicin (73p); ampicillin-
gentamicin (69p); mezlocillin-
netilmicin (44p); piperacillin-
netilmicin (20p)
OutcomesOverall mortality
Treatment failure (clinical and bacteriological)
Superinfections
Colonization
Drop-outs after randomisation
Adverse events
NotesMulticentre
UK
Outcomes in subgroups: Bacteraemia
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Low riskA - Adequate

Gerecht 1989

MethodsRCT
Abdominal
Participants82 patients with suspected cholangitis were randomized empirically. Only those with bacteraemia or positive bile cultures, and fulfilling clinical criteria for cholangitis were evaluated. Patients who were not evaluated because they did not meet inclusion criteria are not considered as dropouts for the review
InterventionsMezlocillin 4grX4 for 11.9 days vs. ampicillin 1grX4 + gentamicin 1.5mg/kgX3 for 10.3 days. In addition to antibiotic therapy all patients underwent surgical intervention
OutcomesTreatment failure (clinical and bacteriological)
Superinfections
Adverse events
Duration of treatment
NotesUSA
No outcomes in subgroups
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Gomez 1990a

MethodsRCT
Sepsis
Participants197 patients with suspected Gram-negative bacteraemia randomized. Patients with proven Gram-negative bacteraemia (78) were analysed. Patients who were not evaluated because they did not meet inclusion criteria for bacteraemia were not considered as dropouts
InterventionsCeftazidime 1grX4 for 10 days vs. cefradine 1grX6 + amikacin 7.5mg/kgX2 for 10 days
OutcomesOverall mortality
Treatment failure (clinical and bacteriological)
Superinfection (bacterial and fungal)
Adverse events
Duration of treatment
NotesSpain (Spanish)
Outcomes in subgroups:
Bacteremia
Gram-negative infections
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Low riskA - Adequate

Havig 1973

MethodsRCT
Abdominal
Participants68 adult patients evaluated with acute cholecystitis verified histologically or by roengten. Trial included 3 arms, of which 2 are included in the review
InterventionsIM ampicillin 0.5grX4
vs. IM chloramphenicol 1grX2 (arm not included in review) vs. IM benzyl-penicillin 400,000IEX2 + IM streptomycin 0.5grX2. In addition 10/24 patients in the ampicillin arm and 15/26 patients in the combination arm were operated on
OutcomesOverall mortality
Treatment failure (clinical)
Duration of fever
NotesNorway
No outcomes in subgroups
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Hoepelman 1988

MethodsRCT
Sepsis
Participants105 patients with serious bacterial infections were included. Of these 18% were neutropenic and are not included for the analysis in this review
InterventionsCeftriaxone 2grX1 vs. cefuroxime 1.5grX3 + gentamicin 80mgX3 (following by an initial 1.5mg/kg dose)
OutcomesOverall mortality Treatment failure (clinical)
Superinfections
Fungal colonization
Adverse events
NotesNetherlands Outcomes for subgroups were not extracted, as they are given in the publication for the whole group including neutropenic patients
Outcomes for non-neutropenic patients were obtained from the author
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Low riskA - Adequate

Holloway 1985

MethodsRCT
Sepsis
Semi-empirical
Participants43 adult patients with suspected Gram-negative septicaemia, or pneumonia, randomized when blood cultures were positive for a Gram-negative pathogen
InterventionsTicarcillin-clavulanic acid 3.1grX4-6 vs. piperacillin 50mg/kgX4-6 + tobramycin 1-1.5mg/kgX3-4
OutcomesTreatment failure (clinical and bacteriological)
Adverse events
NotesUSA
Outcomes in subgroups:
Bacteremia
Gram-negative infections
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Iakovlev 1998

MethodsRCT
Sepsis
Participants95 adult patients with severe nosocomial infections
InterventionsMeropenem 1grX3 for 9 days vs. ceftazidime 1grX3 + amikacin 500mgX2 for 9 days
OutcomesTreatment failure (clinical and bacteriological)
Duration of treatment
Adverse events
NotesMulticentre
Russia (Russian)
Outcomes in subgroups: Urinary tract and Pseudomonas sp. infections
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Jaspers 1998

MethodsRCT
Sepsis
Participants79 elderly patients ( > 65yrs.) with sepsis syndrome and suspected bacteraemia, pneumonia, intra-abdominal sepsis, or complicated urinary tract infection
InterventionsMeropenem 1grX3 for 7.5 days vs. cefuroxime 1.5grX3 + gentamicin 4mg/kgX1 for 7.4 days (metronidazole 500mgX4 added to patients receiving combination in case of abdominal sepsis (15 patients overall)
OutcomesOverall mortality
Treatment failure (clinical and microbiological)
Bacterial superinfections
Adverse events
Duration of treatment
NotesMulticentre
Netherlands
Outcomes in subgroups:
Urinary tract infections
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Low riskA - Adequate

Klastersky 1973

MethodsRCT
Sepsis
Participants75 adult patients with disseminated cancer and life threatening infections, presumed Gram-negative. Randomized to 3 arms, of which 2 are relevant for the review. 18% of patients leukopenic (leukopenia not defined) - no information for neutropenia
InterventionsCarbenicillin 10grX3 for 8.3 days vs. carbenicillin 10grX3 + gentamicin 160mgX3 (IM or IV) for 9 days vs. gentamicin 160mgX3 (3rd arm, not included in review)
OutcomesOverall mortality
Treatment failure (clinical and bacteriological)
Colonization and Superinfection
Duration of treatment
Dropouts
NotesBelgium
Outcomes in subgroups:
Gram-negative and Pseudomonas sp. infections
Bacteremia
Urinary tract infections
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Kljucar 1990

MethodsRCT
Sepsis
Participants150 patients > 14yrs. hospitalized in the intensive care unit and ventilated, with nosocomially acquired pneumonia. Randomized to 3 arms (2 combination and 1 monotherapy)
InterventionsCeftazidime 2grX3 vs.
ceftazidime 2grX3 + tobramycin 80mgX3 vs. azlocillin 5mgX3 + tobramycin 80mgX3, overall for 6.6 days
OutcomesOverall mortality
Treatment failure (clinical and bacteriological)
NotesGermany
No outcomes in subgroups
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Low riskA - Adequate

Koehler 1990

MethodsRCT
Sepsis
Participants144 patients > 18 yrs. with nosocomially acquired pneumonia
InterventionsCeftazidime 1grX3 vs. piperacillin 4grX3 + tobramycin 80mgX3
OutcomesOverall mortality
Treatment failure (clinical and bacteriological)
Bacterial and fungal colonization
Dropouts
NotesMulticentre
Germany
Outcomes in subgroups:
Gram negative and Pseudomonas sp. infections
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Korzeniowski 1982

MethodsRCT
Partially semi-empirical
Gram positive infections
Participants156 patients with clinically suspected infective endocarditis were randomized (prior antibiotic treatment of < 48 hours permitted) 78 patients with Staphylococcus aureus bacteremia and endocarditis were analysed: 48 drug addicts and 30 non-addicts (14 patients randomized semi-empirically)
InterventionsNafcillin 1.5-6grX6 vs. nafcillin 1.5-6grX6 + gentamicin 1mg/kgX3 administered for the first 2 weeks of a 4-week treatment protocol
OutcomesOverall mortality
Treatment failure (clinical and bacteriological)
Dropouts
Need for surgery
Adverse events
Duration of bacteremia and fever are other outcomes shown in the study, but these are shown by groups of empirical treatment regimen which was not always randomly allocated
NotesMulticentre
USA
Outcomes in subgroups:
Bacteremia
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Low riskA - Adequate

Landau 1990

MethodsQuasi-randomized
Urinary tract infections
Participants40 adult patients hospitalized with complicated urinary tract infection
InterventionsCeftriaxone 2grX1
vs. cefazolin 1grX3 + gentamicin 80mgX3
OutcomesOverall mortality
Treatment failure (bacteriological only)
Adverse events
Drop-outs after randomization
Duration of fever
NotesIsrael (Hebrew) Outcomes in subgroups:
Urinary tract and
Gram-negative infections
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)High riskC - Inadequate

Limson 1988

MethodsRCT
Sepsis
Participants54 adult patients randomized, of which 40 patients with severe Gram-negative infections were evaluated
InterventionsCeftazidime 2grX2 vs.
ticarcillin 3grX3-4 + amikacin 500mgX2 (or 15mg/kgX1)
OutcomesTreatment failure (clinical and microbiological)
Fungal superinfections
Adverse events
NotesThe Philippines Outcomes in subgroups:
Bacteremia
Gram negative, and Pseudomonas sp. infections
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Mandell 1987

MethodsRCT
Sepsis
Participants110 patients > 16yrs. evaluated with community acquired or nosocomial pneumonia (2/3 nosocomial)
InterventionsCeftazidime 2grX3 vs.
cefazolin 1.5grX3 or ticarcillin 3grX4 + tobramycin 1.7mg/kgX3
OutcomesTreatment failure (clinical and bacteriological)
Superinfections
Colonization (including resistant development)
Adverse events
NotesMulticentre
Canada
Outcomes in subgroups:
Bacteraemia
Gram-negative infections.
Cefazolin replaced by ticarcillin for combination group patients with documented Pseudomonas infections
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Martin 1991

MethodsRCT
Urinary tract infections
Participants116 patients hospitalized with suspected pyelonephritis
InterventionsCeftriaxone 2grX1 vs. ampicillin 1grX4 + gentamicin 1mg/kgX3
OutcomesTreatment failure (clinical)
Superinfection (relapse and re-infections)
Dropouts
Adverse events
NotesBrussels (French)
Outcomes in subgroups:
Urinary tract infections
Bacteremia
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

McCormick 1997

MethodsRCT
Sepsis
Participants128 adult patients with chronic liver disease (cirrhosis) and suspected or proven sepsis
InterventionsCeftazidime 2grX2 for 5 days vs. mezlocillin 5grX3 + netilmicin 3mg/kgX2 for 4 days
OutcomesOverall mortality
Treatment failure
(clinical)
Adverse events
Duration of treatment and hospital stay
NotesIreland
Outcomes in subgroups:
Bacteremia
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Low riskA - Adequate

Mergoni 1987

MethodsRCT
Sepsis
ParticipantsAdult patients in ICU with severe infections
InterventionsAzlocillin 13+-2.2gr for 6.5 days vs. azloclillin 14.1+-1gr + amikacin 1.16+-0.027gr for 7.2 days (all in for daily doses)
OutcomesTreatment failure (clinical and bacteriological)
Adverse events
Duration of treatment
NotesItaly
Outcomes in subgroups:
Gram negative and Pseudomonas sp. infections
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Low riskA - Adequate

Moreno 1997

MethodsRCT
Sepsis
ParticipantsRenal or (kidney-
pancreas) transplant patients with fever and suspected bacterial infection
InterventionsImipenem-cilastatin 500mgX4 vs. piperacillin 4grX3 + tobramycin 80mgX2
OutcomesTreatment failure (clinical and bacteriological)
NotesSpain
Outcomes in subgroups:
Gram negative and Pseudomonas sp. infections
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Mouton 1990

MethodsRCT
Sepsis
Participants211 adult patients hospitalized in intensive care unit with respiratory tract infections
InterventionsImipenem 500mgX4 for 11.1 days vs. cefotaxime 1grX4 + amikacin 5mg/kgX3 for 10.4 days
OutcomesOverall mortality
Treatment failure (clinical and bacteriological)
Superinfections
Colonization
Hospitalization duration
Duration of treatment
NotesMulticentre
France (French) Outcomes in subgroups:
Bacteremia
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Mouton 1995

MethodsRCT
Sepsis
Participants237 adult patients with community or hospital acquired serious infections, excluding intra-abdominal sepsis (urinary tract infection included)
InterventionsMeropenem 1grX3 for 8.8 days vs. ceftazidime 2grX3 + amikacin 5-7.5mg/kgX2-3 for 8.3 days
OutcomesOverall mortality
Treatment failure (clinical and bacteriological)
Superinfections
Adverse events
Dropouts
Duration of treatment
NotesMulticentre
Europe
Outcomes in subgroups:
Bacteremia
Gram negative and Pseudomonas sp. and urinary tract infections
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Muller 1987

MethodsRCT
Abdominal
ParticipantsTrial includes 3 arms (2 monotherapies, 1 combination treatment)
106 patients evaluated with acute cholecystitis or cholangitis
InterventionsPiperacillin 3grX6 for 7.4 days vs. cefoperazone 2grX3 for 8.1 days vs.
ampicillin 2grX4 + tobramycin 1-1.5mg/kgX3 following 1.5mg/kg loading dose for 11.1 days
OutcomesTreatment failure (clinical)
Adverse events
Duration of treatment
NotesBi-centre
USA
No outcomes in subgroups
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Naime Libien 1992

MethodsRCT
Sepsis
Participants30 children aged 1m - 11yr with severe lower respiratory tract infections
InterventionsCeftizoxime 20-50mg/kgX2-3 vs. penicillin 0.7-1.7 megaunit/kgX3 + gentamicin 1-1.5mg/kgX2
OutcomesOverall mortality
Treatment failure (clinical and bacteriological)
Adverse events
Duration of fever
NotesMexico (Spanish)
No outcomes in subgroups
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Piccart 1984

MethodsRCT
Sepsis
Participants105 adult, non-neutropenic, cancer patients with suspected Gram-negative infections. Study included both neutropenic and non-neutropenic patients, but analysis was completely separated Patients with Gram-positive bacteremia were excluded
InterventionsCefoperazone 6grX2 vs.
cefoperazone 2grX2 + amikacin 500mgX2
OutcomesTreatment failure (clinical and bacteriological)
Superinfections (bacterial and fungal)
Drop-outs after randomization
NotesBelgium
Outcomes in subgroups:
Gram-negative and Pseudomonas sp. infections
Bacteremia
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Rapp 1984

MethodsRCT
Sepsis
Participants35 adult patients hospitalized in a neurosurgical intensive care unit. All with nosocomial pneumonia
InterventionsCeftazidime 2grX3 vs.
ticarcillin 3grX4 + tobramycin pharmacokinetically adjusted doses after 1.75mg/kd loading dose
OutcomesTreatment failure (clinical and bacteriological)
Adverse events
NotesUSA
Outcomes in subgroups:
Gram negative bacteremia
Pseudomonas sp. infections
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Rasmussen 1986

MethodsRCT
Urinary tract infections
Participants62 adult patients hospitalized in a urosurgical department with urinary tract infections, mostly post-operative
InterventionsCefotaxime 3grX3 for 5.4 days vs. ampicillin 1grX4 + netilmicin 150mgX3 for 7 days
OutcomesTreatment failure (clinical)
Relapse
Duration of fever and treatment
Adverse events
NotesDenmark
Outcomes in subgroups:
Urinary tract infections
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Ribera 1996

MethodsRCT
Semi-empirical
Gram-positive infections
ParticipantsSpain
90 intravenous drug users randomized, of which 74 had Staphylococcus aureus right-sided endocarditis. 90.5% of patients were HIV positive. Diagnostic criteria for possible (13% of study patients), probable (34%) and definitive endocarditis (53%) are defined in study
InterventionsCloxacillin 2grX6 vs. cloxacillin 2grX6 + gentamicin 1mg/kgX3
OutcomesOverall mortality
Treatment failure (clinical and bacteriological)
Relapse, re-infection and need for surgery
Duration of treatment
Adverse events
NotesSpain
Journal publication.
Outcomes in subgroups:
Bacteremia
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Low riskA - Adequate

Rubinstein 1995

MethodsRCT
Sepsis
Participants580 adult patients with serious hospital acquired infections and a diagnosis of sepsis, pneumonia or upper urinary tract infection
InterventionsCeftazidime 2grX2 for 9 days vs. ceftriaxone 2grX1 + tobramycin 3-5mg/kgX1 following 2mg/kg loading dose for 9 days
OutcomesOverall mortality
Treatment failure (clinical and bacteriological)
Superinfections
Duration of treatment
Adverse events
NotesMulticentre
Europe, Middle East, Asia, South America
Outcomes in subgroups:
Gram-negative and Pseudomonas sp. infections
Bacteremia
Urinary tract infections
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Low riskA - Adequate

Sage 1987

MethodsRCT
Sepsis
Participants93 patients > 14yrs. randomized to 3 arms, of which 2 are usable in the review. The 3rd arm is aminoglycoside monotherapy. Patients were suspected of a life threatening sepsis, thought to be caused by Enterobacteriaceae or Staphylococci
InterventionsCefotaxime 1-2grX4 for 7.4 days vs. cefotaxime 1-2grX4 + netilmicin 2-3mg/kgX3 (3rd arm, not used - netilmicin 2-3mg/kgX3) for 8.7 days
OutcomesTreatment failure (clinical and bacteriological)
Bacterial and fungal superinfections
Dropouts
Adverse events
Duration of treatment
NotesUK
Outcomes in subgroups:
Bacteremia
Gram negative and urinary tract infections
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Sandberg 1997

MethodsRCT
Urinary tract infections
Participants73 adult female patients with suspected pyelonephritis
InterventionsCefotaxime 1grX2 for 2 days followed by oral cefadroxil 1grX2 vs.
cefotaxime 1grX2 + tobramycin 160mgX1 for 2 days, followed by oral cefadroxil 1 grX2
OutcomesTreatment failure (clinical and bacteriological)
Superinfection and colonization (relapse, re-infections and asymptomatic bacteriuria recurrence)
Adverse events
Drop-outs after randomization
Duration of fever
NotesMulticentre
Sweden
Outcomes in subgroups: Urinary tract infections
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Low riskA - Adequate

Sanfilippo 1989

MethodsRCT
Abdominal
Participants26 female patients aged 16-19 years with acute pelvic inflammatory disease
InterventionsMezlocillin 62.5mg/kgX4 vs. penicillin 480,000U/kgX4 + tobramycin 1mg/kgX3
OutcomesTreatment failure (clinical)
NotesUSA
No outcomes in subgroups
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Low riskA - Adequate

Sculier 1982

MethodsRCT
Sepsis
Participants20 adult, intubated, patients with Gram-negative pneumonia in the neurosurgical intensive-care unit Patients were randomized when presenting with radiographic broncho-
pneumonia, purulent sputum and Gram-negative rods on sputum direct smear
InterventionsMezlocillin 10grX3 vs. mezlocillin 10grX3 + sisomicin 75mgX3.
In addition to allocated systemic treatment, all patients received intra-tracheal sisomycin 25mgX3/d
OutcomesOverall mortality
Treatment failure (clinical and bacteriological)
Bacterial colonization
Resistance development
Adverse events
NotesBelgium
Outcomes in subgroups:
Gram negative and Pseudomonas sp. infections
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Sexton 1998

MethodsRCT
Semi-empirical
Gram-positive infections
Participants67 adult patients randomized, of which 51 with native valve endocarditis (defined by Duke criteria) caused by penicillin-
susceptible Streptococci.
InterventionsCeftriaxone 2grX1 for 4 weeks vs. ceftriaxone 2grX1 + gentamicin 3mg/kgX1 for 2 weeks
OutcomesTreatment failure (clinical and bacteriological)
Relapse and re-infection Adverse events
Dropouts
Duration of hospital stay
Need for surgery
NotesMulticentre
USA
Outcomes in subgroups:
Bacteremia
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Sieger 1997

MethodsRCT
Sepsis
Participants211 adults >18yrs. with hospital-
acquired lower respiratory tract infections. 70% intubated and 27% with severe pneumonia
InterventionsMeropenem 1grX3 for 7.8 days vs. ceftazidime 2grX3 + tobramycin 1mg/kgX3 (following 1.5-2mg/kg loading dose) for 7.4 days
OutcomesOverall mortality
Treatment failure (clinical and bacteriological) Superinfections
Adverse events
Duration of treatment
NotesMulticentre
USA
Outcomes in subgroups:
Gram-negative and Pseudomonas sp. infections.
Study performs both efficacy and ITT analysis, with a drop-out rate of 43% for the efficacy analysis. Outcomes were extracted by ITT. Superinfections and subgroup analyses are given only by efficacy analysis in study
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Smith 1984

MethodsRCT
Sepsis
Participants200 adult patients randomized with suspected or proven serious infections. 195 who actually received study drugs were evaluated for efficacy
InterventionsCefotaxime 2grX6
+ placebo X3 for 5 days vs.
nafcillin 1.5grX6 + tobramycin 2mg/kgX3 for 5.3 days
(Addition of clindamycin 600mgX3 to both groups permitted for suspected anaerobic infections)
OutcomesOverall mortality
Treatment failure
(clinical and microbiological)
Bacterial superinfections
Colonization
Adverse events
Duration of treatment
NotesUSA
Outcomes in subgroups:
Urinary tract and Gram negative infections.
Two additional references refer to the same trial: Moore 1986a (cost-effectiveness analysis), and Moore 1986b (nephrotoxicity analysis). Overall mortality, and treatment duration are taken from Moore 1986a that analysed all patients given study drugs. Cost outcome not included in the review
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Low riskA - Adequate

Speich 1998

MethodsRCT
Sepsis
Participants89 adults >16yrs. with severe pneumonia. Community acquired in 89%
InterventionsPiperacillin-tazobactam 4.5grX3 for 10.2 days vs.
amoxicllin-clavulonic acid 2.2grX3 + gentamicin or netilmicin 3-6mg/kgX1 for 10.1 days
OutcomesOverall mortality
Treatment failure (clinical and bacteriological)
Dropouts
Adverse events
Duration of treatment
NotesMulticentre
Switzerland
No outcomes in subgroups
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Low riskA - Adequate

Stille 1992

MethodsRCT
Sepsis
Participants337 adult patients randomized with non-life-
threatening infections, of abdominal, gynaecological or respiratory tract origin (UTI, skin, bone, and CNS infections excluded)
InterventionsImipenem 500mgX3 for 8.4 days vs. cefotaxime 2grX3 + gentamicin 0.66-1mg/kgX3 for 8.2 days (metronidazile allowed in combination treatment group for suspected anaerobic infection)
OutcomesOverall mortality
Treatment failure (clinical and bacteriological)
Colonization and resistance development
Adverse events
Duration of treatment
NotesMulticentre
Germany and Austria
Outcomes in subgroups:
Gram negative and Pseudomonas sp. infections
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Sukoh 1994

MethodsRCT
Sepsis
Participants63 patients with respiratory tract infections and underlying respiratory disease
InterventionsCefoperazone/ sulbactam 1-4gr/d for 11.7 days vs. Cefoperazone/ sulbactam 2-6gr/d + one of several aminoglycosides in low doses (amikacin 100-400 mg/d 16 patients, tobramycin 40-180 mg/d 15 patients, isepamicin 400 mg/d 1 patient, netilmicin 200 mg/d 1 patient) for 11.1 days
OutcomesTreatment failure (clinical and bacteriological)
NotesJapan (Japanese)
Outcomes in subgroups:
Gram-negative and Pseudomonas sp. infections
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Takamoto 1994

MethodsRCT
Sepsis
Participants171 adult patients with respiratory tract infections
InterventionsImipenem/cilastatin sodium vs.
imipenem/cilastatin sodium + amikacin sulfate
OutcomesTreatment failure (clinical and bacteriological)
Drop-outs after randomization
Adverse events
NotesMulticentre Japan (Japanese) Outcomes in subgroups:
Gram-negative and Pseudomonas sp. infections
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Thompson 1990

MethodsRCT
Abdominal
Participants96 patients evaluated with acute cholangitis (cholecystitis not included)
InterventionsPiperacillin 3grX6 for 8.4 days vs. ampicillin 2grX4 + tobramycin 1-1.5mg/kgX3 for 9.1 days (following 1.5mg/kg loading dose). In addition 35/96 patients were operated on
OutcomesOverall mortality
Treatment failure (clinical)
Adverse events
Treatment duration
NotesMulticentre
USA
No outcomes in subgroups
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Thompson 1993

MethodsRCT
Abdominal
Participants120 patients evaluated with acute biliary tract infections (cholecystitis and cholangitis)
InterventionsCefepime 2grX2 for 7.5 days vs. mezlocillin 3grX6 + gentamicin 1.5mg/kgX3 for 7 days. In addition, 118/120 patients were operated on
OutcomesOverall mortality
Treatment failure (clinical)
Adverse events
Treatment and hospitalization duration
NotesMulticenter
USA
No outcomes in subgroups
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Trujillo 1992

MethodsRCT
Sepsis
Participants30 adult patients with severe skin and soft tissue or respiratory tract infections
InterventionsCeftizoxime 1-2grX3 vs. ampicillin 1-3grX4 + gentamicin 3-5mg/kg/d, overall for 10 days
OutcomesOverall mortality
Treatment failure (clinical and bacteriological)
Adverse events
Fever duration
NotesMexico (Spanish)
No outcomes in subgroups
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Vergnon 1985

MethodsRCT
Sepsis
Participants30 adult patients wth severe broncho-
pulmonary infections
InterventionsCefoperazone 2grX2 for 16.8 days vs. ampicillin 1.5grX4 + tobramycin 1mg/kgX3 for 11.8 days
OutcomesTreatment failure (clinical)
Resistant colonization
Adverse events
Duration of treatment
NotesFrance (French)
No outcomes in subgroups
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Verzasconi 1995

MethodsRCT
Urinary tract infections
Participants93 adult patients with acute pyelonphritis or complicated urinary tract infections
InterventionsAmoxicillin-clavulonate 2.2grX3 for 4.1 days vs. amoxicillin 2grX3 + gentamicin 1.5mg/kg loading followed by maintenance for 4.2 days
OutcomesTreatment failure (bacteriological)
Superinfection
Dropouts
Treatment and fever duration
Adverse events
NotesBi-centre
Switzerland (German)
Outcomes in subgroups:
Urinary tract infection
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Warren 1983

MethodsRCT
Sepsis
Participants120 adult patients with suspected or known life-threatening infections caused by Gram-negative bacilli
InterventionsCefoperazone 1.5grX4 for a median of 9 days vs.
cefamandole 2grX6 + tobramycin 1.7mg/kg loading dose, followed by drug- level-adjusted maintenance dose for a median of 8 days
OutcomesOverall mortality
Treatment failure (clinical and bacteriological)
Superinfection
Duration of treatment
Adverse events
Drop-outs after randomization
NotesUSA
Outcomes in subgroups:
Bacteremia
Gram-negative infections
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Wiecek 1986

MethodsRCT
Urinary tract infections
Participants20 adult patient with acute pyelonphritis
InterventionsCeftazidime 1grX3 vs.
cefotaxime 1grX2 + tobramycin 1mg/kgX3
OutcomesTreatment failure (bacteriological)
Re-infection
Adverse events
NotesPoland
Outcomes in subgroups:
Gram negative and Pseudomonas sp. infections
Urinary tract infections
Bacteremia
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Unclear riskB - Unclear

Wing 1998

MethodsRCT
Urinary tract infections
Participants179 pregnant women <24 weeks gestation with pyelonephritis randomized to 2 monotherapy arms and 1 combination therapy arm
InterventionsCefazolin 1grX3 vs. ceftriaxone 1grX1 vs. ampicillin 2grX4 + gentamicin 1.75mg/kgX3 (after 2mg/kg loading)
OutcomesOverall mortality
Treatment failure (clinical and bacteriological)
Re-infection
Fever and hospitalization duration
NotesBi-centre
USA
Outcomes in subgroups:
Urinary tract infections
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Low riskA - Adequate

Yellin 1993

  1. a

    RCT - randomized controlled trial
    vs - versus
    Semi-empirical - comparison of second-line antibiotic treatment, given following establishment of microbiological or clinical diagnosis.
    Treatment duration represents means unless otherwise specified.

MethodsRCT
Abdominal
Participants179 patients with clinically suspected cholecystitisOnly those operated on while on allocated treatment were evaluated (infection proven at surgery) Patients who were not evaluated because surgery was not performed or incorrect diagnosis are not considered as drop-outs for the review
InterventionsCefepime 2grX2 for 7.3 days vs. mezlocillin 4grX4 + gentamicin 1.5mg/kgX3 for 7.2 days. In addition to antibiotic treatment all patients operated
OutcomesOverall mortality
Treatment failure (clinical)
Fever, treatment and hospitalization duration
NotesUSA
No outcomes in subgroups
Risk of bias
BiasAuthors' judgementSupport for judgement
Allocation concealment (selection bias)Low riskA - Adequate

Characteristics of excluded studies [ordered by study ID]

StudyReason for exclusion
Alvarez-Lerma 2001bBeta-lactam-aminoglycoside combination treatment versus beta-lactam-aminoglycoside combination treatment.
Badaro 2002Allocation to additional aminoglycoside treatment not randomized. Patients were randomized to treatment with beta-lactam monotherapy versus 'standard' antibiotic treatment, which was a beta-lactam with or without an aminoglycoside.
Benlloch 1995Antibiotic regimens incompatible with protocol. Randomization to 3 arms: 1) beta-lactam-aminoglycoside-nitroimmidazole combination 2) beta-lactam-aminoglycoside combination 3) double beta-lactam combination.
Blumer 2003No sepsis in inclusion criteria. Study included patients with acute exacerbation of cystic fibrosis, but the definition of exacerbation does answer the criteria for sepsis as defined in review.
Cetto 1983Study includes 71% patients with neutropenia.
Ciftci 1997Antibiotic regimens incompatible with protocol. Randomization to 4 arms: 1) beta-lactam-aminlglycoside-lincosamide combination 2) beta-lactam-aminoglycoside-imidazole combination 3) beta-lactam monotherapy 4) beta-lactam-imidazole combination.
Crenshaw 1983Prophylaxis study. Randomization to beta-lactam monotherapy versus beta-lactam aminoglycoside combination therapy as preventive therapy for patients with penetrating abdominal wounds requiring surgical intervention.
Croce 1993Not a randomized trial. Monotherapy and combination therapy groups were studied consecutively.
De Louvois 1992Included patients were newborns with suspected sepsis.
Extermann 1995Randomization to beta-lactam monotherapy versus best-guess antibiotic treatment as chosen by physician. The best guess treatment group includes monotherapy and various combinations.
Fainstein 1983Study includes 62.5% neutropenic patients. The study randomized 321 episodes, of which 275 were evaluable - 172 in neutropenic patients and 103 episodes in non-neutropenic patients. Although analysis was intended to be separated, the number of evaluated patients in each group, is not separated to neutropenic and non-neutropenic patients. Although outcomes (death and failures) are given for non-neutropenic patients, the number of patients in the group is unknown. Information was unavailable from authors.
Fernandez 1991Randomization to beta-lactam monotherapy versus
combination therapy commonly used in specific centre (multicentre trial). Combinations consisted of different beta-lactams with aminoglycoside antibiotics in 211/273 patients evaluated in the combination group, and other antibiotic combinations in 62/273 patients. Outcomes are given per specific combination (failure), but the study is excluded since the decision as to which combination the patient received was left to the care-taker.
Foord 1985Not a randomized trial. Article describes all patients on Glaxo data files who have been administered Ceftazidime monotherapy in comparative and non-comparative trials. No references in the article.
Gentry 1980Not a randomized trial. Study describes centre's experience with monotherapy versus combination therapy. One study group was previously reported. All prospective, comparative, but no mention of randomization.
Gentry 1984Prophylaxis study. Randomization to 3 arms (2 beta-lactam monotherapy arms and 1 beta-lactam-aminoglycoside combination therapy arm), as perioperative prophylaxis for patients with penetrating injuries of the abdomen.
Gentry 1985Pooled analysis of patients with skin, soft-tissue and bone infections, comparing ceftazidime monotherapy to control regimens, including ticalcillin and tobramycin combination therapy. However, randomized patients cannot be separate from those who entered open comparative trials.
Gerber 1989Prophylaxis study. Antibiotic treatment was administered as prophylaxis and patients did not fulfil the criteria for sepsis when randomized.
Gilbert 1998Study includes 18% neutropenic patients (32/175 evaluable patients). In addition neutropenic patients were not randomized - all were allocated only to the combination regimen. Outcome data was unavailable separating randomized from non randomized (neutropenic) patients.
Giraud 1989Antibiotic regimens incompatible with protocol. Randomization to 2 arms: 1) beta-lactam monotherapy versus 2) beta-lactam-aminoglycoside-nitroimidazole triple combination therapy.
Gold 1985No sepsis in inclusion criteria. Study included patients with acute exacerbation of cystic fibrosis, but the definition of exacerbation does answer the criteria for sepsis as defined in review.
Gomez 1990bObservational study according to author correspondence.
Greco 1989Non-randomized prospective comparative trial.
Gribble 1983Study includes 60% neutropenic patients (30/50 evaluable episodes).
Haffejee 1984Included patients were neonates and children.
Hall 1988Included patients were neonates.
Hammerberg 1989Included patients were premature neonates with risk factors for sepsis (31/72 patients between ages 0-1 months).
Hanson 1982Antibiotic regimens incompatible with protocol. Combination therapy versus combination therapy.
Hoogkamp 1983Not a randomizd trial. Study groups were studied sequentially. In addition study population consists of cystic fibrosis patients with an exacerbation - sepsis not part of inclusion criteria.
Iakovlev 1997Aminoglycoside was added only to patients that did not respond to the initial beta-lactam monotherapy that was administered empirically.
Iakovlev 2000Not a randomized trial.
Ker 1989Prophylaxis study. Randomization to prophylactic antibiotic treatment, patients did not fulfil criteria for sepsis when randomized.
Krumpe 1999Patients first stratified by disease severity to monotherapy (severe disease) or combination therapy. Following stratification, the patients were randomized to 4 arms: 1) quinolone monotherapy 2) 'standard monotherapy'' from a defined choice of various beta-lactams, at investigator's discretion 3) quinolone-beta-lactam combination therapy 4) 'standard combination therapy' which consisted of various possible combinations of beta-lactams and aminoglycosides at investigators discretion.
Ludwig 1980Description of two separate randomized trials: 1) beta-lactam versus aminoglycoside 2) beta-lactam versus beta-lactam. All administered as monotherapies.
Maller 1991Randomization to once daily aminoglycoside treatment versus twice daily aminoglycoside treatment. In addition to the aminoglycoside, a beta-lactam was administered if considered necessary. Administration of the beta-lactam not randomized (interim analysis of a multicentre study).
Mangi 1988Randomization to beta-lactam monotherapy versus combination. The combination group consisted of clindamycin-aminoglycoside or beta-lactam-aminoglycoside combinations. The decision as to which combination treatment was administered was made on a case-by-case basis, according to the sputum's Gram stain. Patients with Gram-negative bacilli in the sputum were administered the beta-lactam based combination while all others received the clindamycin-based regimen. Outcomes for the two different combination treatments are given together.
McArdle 1987Prophylaxis study. Randomization to beta-lactam monotherapy versus beta-lactam aminoglycoside combination therapy for prophylaxis prior to high-risk biliary tract surgery.
McCarty 1988No sepsis in inclusion criteria. Study included patients with acute exacerbation of cystic fibrosis, but the definition of exacerbation does answer the criteria for sepsis as defined in review.
McLaughlin 1983No sepsis in inclusion criteria. Study included patients with acute exacerbation of cystic fibrosis, but the definition of exacerbation does answer the criteria for sepsis as defined in review.
Mondorf 1987Infection or sepsis not mentioned as part of inclusion criteria. Patients were randomized to receive beta-lactam monotherapy versus beta-lactam-aminoglycoside combination therapy, and the only outcome given is urinary enzyme excretion.
Mondorf 1989No outcomes relevant for this review. The study randomized patients with severe infections to beta-lactam monotherapy versus beta-lactam aminoglycoside combination therapy. The only outcomes given are renal functions, mainly urinary enzyme levels, and mean serum creatinine per group. Author contacted to ask number of patients per group developing nephrotoxicity and other outcomes - but did not respond.
Moreno-Martinez 1998Comparison of oral cefixime versus oral amoxicillin and intramuscular netilmicin. By protocol only intravenously administered beta-lactams are included.
Mouton 1985Study published as conference proceeding, comparative without mention on randomization. No further details regarding the study were available.
Oblinger 1982Randomization to beta-lactam monotherapy versus combination of antibiotics as deemed appropriate by the attending physicians.
Odio 1987Included patients were neonates with proven invasive bacterial infections.
Padoan 1987Inclusion criteria did not mandate sepsis for all included patients. Study included patients with acute exacerbation of cystic fibrosis, but the definition of exacerbation does answer the criteria for sepsis as defined in review.
Paoletti 1989Comparison between aminglycoside monotherapy (netimicin) to beta-lactam-aminoglycoside combination therapy (ampicillin + netilmicin).
Rodloff 1998Study randomized patients to imipenem monotherapy versus various combination regimens: beta-lactam-aminoglycoside, two beta-lactams, beta-lactam-beta-lactamsa inhibitor, beta-lactam-anaerobic agent and quinolone-anaerobic agent. Patients allocated to the combination group were analysed as one group.
Romanelli 2002Study randomized patients to beta-lactam monotherapy versus macrolide-aminogycoside or macrolide-beta-lactam combination therapy.
Schoengut 1983Non-randomized, prospective comparative trial.
Schuler 1995Randomization to meropenem versus cefotaxime monotherapy. Aminoglycoside added to the cefotaxime arm for urinary tract infections and metronidazole added to the cefotaxime arm when anaerobic infection was suspected. These additions were performed non-randomly, by protocol.
Scott 1987Randomization to 3 arms comparing beta-lactam monotherapy versus beta-lactam-metronidazole-aminoglycoside triple combination therapy versus beta-lactam-metronidazole combination therapy.
Sexton 1984ICAAC abstract. Twenty-two patients were enrolled in a prospective randomized trial, and 8 patients received monotherapy in an open study. Results are shown for all 30 patients combined. Author contacted and replied that original data are no longer available, and therefore randomized patients cannot be separated from the non-randomized. However, results of these trials were pooled with other trials and are described in Gentry 1985.
Sheftel 1986No relevant outcomes for this review. The study randomized patients with osteomyelitis and provides outcomes only for evaluated patients at a follow up range of 2-38 months (appropriate for the type of infection). The number of randomised patients is unknown and outcomes at 30-days were unavailable.
Smith 1999No sepsis in inclusion criteria. Study included patients with acute exacerbation of cystic fibrosis, but the definition of exacerbation does answer the criteria for sepsis as defined in review.
Solberg 1995Article describes results from 4 separate trials. Randomization to meropenem versus ceftazidime monotherapy. An aminoglycoside was added to patients allocated to ceftazidime when resistance to ceftazidime and severe infections were suspected.
Solomkin 1986Inadequate methodology. A publication stating that 'case report forms from an open multicentre study were reviewed': 69 patients assigned to ceftazidime and 66 patients assigned to ticarcillin and tobramycin with soft tissue infections are reported. Information obtained through author contact: these were the only arms of the trial, all patients included in the trial are reported in the publication, and this is the only report of the trial. However, according to the author, the study was not well designed and considered more as a collection of case reports, as stated in the publication.
Stack 1985No sepsis in inclusion criteria. Study included patients with acute exacerbation of cystic fibrosis, but the definition of exacerbation does answer the criteria for sepsis as defined in review.
Tally 1986Randomization to beta-lactam monotherapy (moxalactam) versus another beta-lactam (cefoxitime). An aminoglycoside could be added to the cefoxitime arm by the attending physician's decision, in consultation with an infectious diseases consultant.
Thompson 1980Oral versus intravenous antibiotic administration. Study randomized women with PID to monotherapy of oral amoxicillin versus combination therapy consisting of IV penicillin + IV gentamicin. Inclusion criteria for the review specify IV administration of the beta-lactam in both arms.
Vazquez 1994Prophylaxis study. Antibiotic treatment administered for prophylaxis, without sepsis. In addition trial is probably not randomized.
Vetter 1987No sepsis in inclusion criteria. Study randomized patients with acute exacerbations of chronic bronchitis. Only 19/102 included patients were febrile.
Vetter 1992Comparison of monotherapy (meropenem) versus monotherapy (ceftazidime)
Watanakunakorn 1997Non-randomized comparison of penicillin versus penicillin + gentamicin for Staphylococcus aureus endocarditis.

Ancillary