Skip navigation

Antibiotics for Acute Otitis Media

Ear infections in children are common. In the USA there are 12.8 million episodes a year (about 45,000 per million population), usually in children aged 5 years or younger and most involve seeing a doctor. Despite there being over 170 published clinical trials there is no consensus on the best drug for the initial therapy for acute otitis media (AOM).


A meta-analysis of 35 randomised studies involving antimicrobial drugs in 5400 children may help. This very detailed study [1] is an exemplar of how systematic reviews can be done; the authors started with a written protocol defining the methods and objectives of the meta-analysis.

Study inclusion

Studies to be included had to be randomised controlled trials of antimicrobial drugs for the initial empirical treatment of simple AOM. The authors give exact definitions of what they mean for each of these important words to avoid any doubt about the definitions they used:
  • Simple AOM - new or recurrent episodes of AOM in patients without underlying disorders.
  • AOM - bulging or opacification of the tympanic membrane with or without reddening, with symptoms of acute infection.
  • Initial empirical therapy - treatment of new-onset AOM without knowledge of the specific causative agent.
  • Therapeutic antimicrobial agents - drugs administered to treat established AOM - with individual drugs defined.
  • Randomised controlled studies - allocation of subjects by chance to one or more concurrent treatment groups, at least one of which involving a study drug as defined.
Their initial searching strategy identified 286 studies. The progress from that to the final tally of 30 included papers is set out with the reasons for each exclusion given.

Outcome measures

The primary end point was the clinical response to antimicrobial therapy. This was defined as the absence of all presenting signs and symptoms of AOM at the evaluation point closest to 7 - 14 days after therapy started. The appearance of the tympanic membrane, if reported, should be improved.

Neither middle ear effusion nor the lack of a bacteriologic cure was grounds for considering a specific treatment a failure. All outcomes less than success as defined, including unilateral resolution of bilateral AOM, were considered primary end point failures.

Patients, rather than ears, were the unit of analysis.


Sixty-nine study arms were identified in the 30 trials and gave the following results:-

Pre treatment tympanocentesis increased the primary control rates by 6.5% (95% CI 3 - 10%).

This paper has many different comparisons between different drugs, and a number of sensitivity analyses. The main findings of effect were the comparisons of penicillin, aminopenicillin or any antimicrobial against placebo or no drug controls. The odds ratios and rate difference (RD: same as absolute rate reduction x 100) for these were:-


The number needed to treat is calculated in the paper as 7. Six of every seven children with AOM either do not need antibiotics for primary control or will not respond to antibiotic therapy. All seven children have to be treated because we cannot predict which one of the seven is both at risk for failure and responsive to antibiotics.

Provided that the antibiotic given is safe, well-tolerated and affordable this need to treat many to control a few would be offset by a 14% higher primary control rate and a potentially lower incidence of suppurative complications.


The authors give a qualified yes to the question whether antibiotics should be part of the initial empirical therapy for AOM in children. As always there are qualifications, and these are elegantly dissected. As is so often the case, no single paper, or meta-analysis in this case, gives the entire result. But meta-analyses like this provide policy makers with the raw material from which to forge guidelines based on solid evidence.


  1. RM Rosenfeld, JE Vertrees, J Carr et al. Clinical efficacy of antimicrobial drugs for acute otitis media: metaanalysis of 5400 children from thirty-three randomized trials. Journal of Pediatrics 1994 124: 355-67.

previous or next story in this issue