|
Ceftriaxone versus beta-lactams with antipseudomonal activity for empirical, combined antibiotic therapy in febrile neutropenia: a meta-analysis |
Furno P, Dionisi M S, Bucaneve G, Menichetti F, Del Favero A |
|
|
Authors' objectives To compare the efficacy of antibiotic combinations including ceftriaxone (CRO) with that of combinations including an antipseudomonal beta-lactam, for the empirical treatment of febrile neutropenia in cancer patients.
Searching MEDLINE was searched from 1966 to 1997 using the following principal keywords and subject headings: 'ceftriaxone', 'ceftriaxon', 'granulocytopenia', 'neutropenia', 'fever', 'controlled', 'randomised', 'clinical trial', 'antipseudomonal', 'beta-lactam', 'antibiotic' and 'bacteraemia'. The drug manufacturer was contacted for all published studies and unpublished files from the central internal product information database on ceftriaxone, which would be relevant to this meta-analysis. A manual search of the bibliographies of pertinent studies and reviews was also performed.
Study selection Study designs of evaluations included in the reviewRandomised controlled studies were included.
Specific interventions included in the reviewAntibiotic combinations containing either CRO as the beta-lactam for the experimental group or an antipseudomonal beta-lactam for the control group.
Participants included in the reviewPatients with fever (greater than or equal to 38 degrees C) and neutropenia (less than 1000 cells/mm3) were included. Some individual trials allowed patients to enrol in the given study more than once during multiple neutropenic febrile episodes; all episodes were then included in the meta-analysis.
In 5 of the 8 studies, patients were eligible with a neutrophil count of 500/mm3 or less, whereas the remaining 3 studies required a neutrophil count of 1000/mm3 or less. The overall age of the patients enrolled ranged from 1 to 84 years, although only 3 trials allowed enrolment of patients younger than 14 years. All trials included patients with both leukaemias and solid tumours, whilst 4 studies also included bone marrow-transplanted patients. In all trials, the assessment of pre-treatment characteristics of the patient population failed to find any significant imbalance to suggest a bias in treatment allocation.
Outcomes assessed in the reviewThe outcome measure was failure of antibiotic treatment. The data set were analysed differently according to two different definitions of treatment failure. The primary outcome measure was failure of empirical treatment without modification, i.e. an inadequate clinical response requiring a change or addition of antibiotics to the initial regimen. The secondary outcome measure was defined as overall death, with or without modification of the initial therapy.
In 6 of the 8 studies, the antibiotic combination regimes to be compared were CRO plus amikacin versus ceftazidime plus amikacin. Another study compared CRO plus tobramycin with azlocillin plus tobramycin, whilst the remaining study compared CRO plus teicoplanin with ceftazidime plus teicoplanin.
How were decisions on the relevance of primary studies made?The authors do not state how the papers were selected for the review, or how many of the reviewers performed the selection.
Assessment of study quality The authors used a modified version of a published quality assessment tool (see Other Publications of Related Interest nos.1-2) to assess validity. A total of 20 items were considered: 9 on study design, and 11 on data analysis and results presentation. The maximum possible score achievable by each study was 75 points; 64, 28 and 8% of this total score was for study design, data analysis and results presentation, respectively.
A score was assigned to each item when the item was applicable. The total score gained in each study was then divided by the maximum possible score applicable for the given study. Scores of non-applicable items were not counted in the denominator. Two reviewers independently assigned a score to each item, and any disagreements were resolved by consensus.
Data extraction Data from each study were extracted independently by two people familiar with the clinic topic and blinded to author, title and journal. Their reports were cross-checked to avoid possible errors. The items of information extracted from each study were: the number and ages of patients enrolled; the drugs, their dose and schedule of administration in the control and experimental groups; the type of underlying disease; the definition of neutropenia in terms of degree and duration; adverse effects; the number of failures and successes in each treatment group; the number of deaths; and the number of exclusions or withdrawals.
Methods of synthesis How were the studies combined?The binary data were combined using Peto's modification of the Mantel-Haenszel fixed-effect method (see Other Publications of Related Interest no.3).
A point estimate of odds ratio (OR) less than 1.0 favoured CRO-containing combinations.
How were differences between studies investigated?A chi-squared test for heterogeneity of treatment effects between different trials was performed to assess the appropriateness of pooling.
A subgroup analysis was performed to examine how the pooled OR of failure of CRO combinations might change when calculated for specific subsets of trials, based on selected characteristics of the study population, study design and treatment regimes. Subgroups included adult patients, i.e. patients older than 14 years; the number of bacteraemic episodes; the severity of neutropenia, i.e. less than 500 neutrophils/mm3; high-quality trials, i.e. only trials with quality scores equal to, or above, the mean score for all trials; and aminoglycoside-containing regimes. Only the primary end point was analysed in these subgroups.
Results of the review Eight published trials involving 1,537 evaluable febrile episodes (ranging from 17 to 694 per trial) in neutropenic patients, met the inclusion criteria.
The study quality scores ranged from 0.18 to 0.68, with a mean score of 0.36. Most of the studies achieved the maximum possible score in only four items: selection criteria, listing of withdrawals, dates of study, and tabulation of events employed as end points. There were three items for which all studies achieved the minimum possible score: list of patients excluded, blinding of patients to treatment, and blinding of physicians to treatment.
The chi-squared test for heterogeneity was not significant for the primary (chi-squared 6.1, d.f.=7, p>0.3) or secondary (chi-squared 6, d.f.=6, p>0.3) outcome, indicating the suitability of combining data across individual trials.
Overall, there were 256 treatment failures out of the 782 febrile episodes treated with ceftriaxone-containing combinations (32.7%), and 243 failures out of the 755 treated with antipseudomonal beta-lactam regimes (32.1%). The pooled OR of failure for ceftriaxone-containing combinations was 1.04 (95% confidence interval, CI: 0.84, 1.29, p>0.05) for febrile episodes, and 0.93 (95% CI: 0.58, 1.49) for bacteraemic episodes.
With regard to overall mortality, there were 54 deaths among the 782 febrile episodes treated with ceftriaxone-containing combinations (6.9%), and 62 deaths among the 755 febrile episodes treated with antipseudomonal beta-lactam-containing regimes (8.2%). The pooled OR of death for ceftriaxone regimes was 0.84 (95% CI: 0.57, 1.24, P>0.3).
Subgroup analysis: the pooled ORs for failure of CRO combinations ranged from 0.89 to 1.12 when the analysis was conducted on either studies enrolling exclusively adult patients, those with baseline neutropenia less than 500 cells/mm3, those comparing aminoglycoside-containing combinations, those of higher quality, or those of bacteraemic episodes exclusively; all were non significant. These findings suggested that these variables had no influence on the overall result of the meta-analysis.
The impact of the duration of neutropenia, profound bacteraemic neutropenia less than 100/mm3, and the use of antimicrobial prophylaxis could not be assessed. This was because the disaggregated data were not uniformly reported in individual clinical trials, and therefore, could not be meaningfully combined. When the largest individual study was excluded from the analysis, there was no substantial effect on the results (pooled OR 0.94, 95% CI: 0.71, 1.25).
Authors' conclusions The results of this meta-analysis provide convincing evidence that CRO-containing antibiotic regimes are as effective as combinations in which the beta-lactam exerts specific activity against Pseudomonas aeruginosa, such as ureidopenicillins or ceftazidime, in initial empirical therapy of febrile neutropenic patients.
CRD commentary The review question and inclusion criteria were stated clearly. The validity assessment was thorough and involved an independent assessment by two reviewers. Details of the primary data were clearly tabulated, although no details of adverse effects were provided. The data were pooled appropriately after conducting tests of heterogeneity.
The search strategy only involved one database and unpublished literature was not included. A publication bias cannot, therefore, be ruled out. It was not stated whether foreign language papers were included. It was also unclear how many reviewers assessed the studies for inclusion.
The authors' conclusions follow from the results.
Implications of the review for practice and research Practice: The authors state that the data presented here are not intended to support the indiscriminate use of CRO-containing combinations as first-line treatment in febrile neutropenia. In making the choice of empirical treatment of febrile neutropenic patients, it is wise to consider the individual patient's risk factors, the local epidemiology, and possible modification of the initially selected regimen on the basis of frequent and careful reassessment of clinical response in these critically ill patients.
Research: The authors note that most randomised trials that have compared CRO with an aminoglycoside have involved a lack of statistical power and contradictory reporting of treatment effect directions, or the magnitude between them has made their results difficult to interpret. Trials with a larger number of patients than has generally been employed would be required to detect a moderate treatment effect in favour of either regimen with reasonable statistical certainty. The best way to address the limitations discussed is to perform a definitive, well-designed randomised controlled clinical study of adequate size.
Funding C.N.R. progetto ACRO, grant number 9600551, P.F. 39; Educational Grant of Roche.
Bibliographic details Furno P, Dionisi M S, Bucaneve G, Menichetti F, Del Favero A. Ceftriaxone versus beta-lactams with antipseudomonal activity for empirical, combined antibiotic therapy in febrile neutropenia: a meta-analysis. Supportive Care in Cancer 2000; 8(4): 293-301 Other publications of related interest 1. Chalmers TC, Smith, H, Blackburn B, Silverman B, Schroeder B, Reitman D, et al. A method for assessing the quality of a randomized control trial. Control Clin Trials 1981;2:31-49. 2. Powe NR, Kinnison ML, Steinberg EP. Quality assessment of randomised controlled trials of contrast media. Radiology 1989;170:377-80. 3. Yusuf S, Peto R, Lewis J, Collins R, Sleight P. Beta blockade during and after myocardial infarction: an overview of the randomized trials. Prog Cardiovasc Dis 1985;27:335-71.
Indexing Status Subject indexing assigned by NLM MeSH Adolescent; Adult; Aged; Anti-Bacterial Agents /therapeutic use; Ceftriaxone /therapeutic use; Cephalosporins /therapeutic use; Drug Therapy, Combination /therapeutic use; Fever /chemically induced /drug therapy /microbiology; Humans; Middle Aged; Mortality; Neutropenia /chemically induced /drug therapy; Odds Ratio; Pseudomonas Infections /complications /drug therapy; Pseudomonas aeruginosa /drug effects; Treatment Outcome AccessionNumber 12000001414 Date bibliographic record published 28/02/2002 Date abstract record published 28/02/2002 Record Status This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn. |
|
|
|