Double-blind study comparing erythromycin and mupirocin for treatment of impetigo in children: Implications of a high prevalence of erythromycin- resistant Staphylococcus aureus strains

Staphylococcus aureus has been consistently isolated from a high proportion of impetiginous lesions, and in several recent studies, it was present in the majority of the cases. Since recently a large proportion of S. aureus strains in our community showed erythromycin resistance, we undertook a prospective double-blind controlled study comparing topical mupirocin with oral erythromycin to determine (i) the prevalence of erythromycin-resistant S. aureus strains in impetigo and (ii) whether an increased rate of failure of erythromycin treatment was associated with such resistance. A total of 102 patients 3 to 185 months old (median = 49 months) were enrolled. Culture was positive for 97 of 102 (95%) patients, and S. aureus was present in 93% of the patients for whom cultures were positive. S. aureus was the single pathogen in 64% of these patients. Erythromycin-resistant S. aureus strains were present in 27 of 91 (28%) patients for whom cultures were positive. In all cases but one, S. aureus was resistant to penicillin, and in all cases it was sensitive to mupirocin. A marked difference was observed in favor of mupirocin in the clinical courses of the disease. However, only patients with erythromycin-resistant S. aureus strains had unfavorable courses compared with those treated with mupirocin (failure rate, 47 versus 2%, respectively). Patients with erythromycin-susceptible S. aureus strains who received erythromycin had a failure rate of 8%. In four patients, S. aureus strains initially susceptible to erythromycin became resistant during treatment. We conclude that erythromycin-resistant S. aureus strains are commonly isolated from impetigo in our region. In view of the increasing resistance of S. aureus to various antibiotic drugs, new therapeutic regimens should be sought for this common skin infection. In our setting, mupirocin seems an appropriate alternative.