Meta-analysis of bacterial resistance to macrolides

OBJECTIVES: Understanding changing resistance patterns is important in determining appropriate antibiotic treatments. This meta-analysis systematically evaluated resistance of Streptococcus pneumoniae and Streptococcus pyogenes to macrolide antibiotics among patients with community-acquired respiratory tract infections. METHODS: MEDLINE and EMBASE databases were searched and experts were consulted to identify published and unpublished literature reporting macrolide resistance rates. Identified studies were evaluated by two independent reviewers; those meeting a priori specified criteria (resistance by patient condition and strain, resistance thresholds, 1997-2003 isolates) were included. Data from included studies were abstracted by two independent reviewers using a standard review form. Discrepancies in abstracted data were resolved by the study investigator. RESULTS: Random-effects meta-analysis was performed for outcomes present in at least four studies overall and for specified subgroups. We identified 3849 studies and performed detailed review on 407; of these 29, published between 1998-2003, met the inclusion criteria. Mean resistance of S. pneumoniae isolates to azithromycin was 27.2% [95% confidence interval (CI) 24.6-29.7]; mean resistance to erythromycin was statistically equivalent (30.4%; 95% CI 28.1-32.7). Resistance of S. pyogenes to erythromycin (30.0%; CI 18.6-41.5) was similar to that of S. pneumoniae. Too few studies of clarithromycin were included to allow evaluation of resistance. In subgroup analyses, substantial variation in resistance to erythromycin was seen by geographic area. CONCLUSIONS: Reported macrolide resistance of S. pneumoniae varies substantially and may be a significant issue in certain regions. Use of meta-analysis to aggregate individual studies enabled determination of robust values for macrolide resistance. This information is useful for clinical and policy decision makers in developing appropriate antibiotic strategies.