Measuring the comparative efficacy of antibacterial agents for acute otitis media: the "Pollyanna phenomenon"

In randomized, double-blind trials of antibiotic therapy for acute otitis media that determined both clinical and bacteriologic outcomes, clinical success rates were (93%) 236 of 253 for patients with bacteriologic success, (62%) 25 of 40 for those with bacteriologic failure, and (80%) 124 of 155 for those with nonbacterial acute otitis media. These rates were used to calculate the effectiveness of three strategies for assessing drug efficacy: (1) tympanocentesis and culture before and during therapy (bacteriologic efficacy), (2) tympanocentesis before therapy and assessment of clinical efficacy in bacterial acute otitis media, and (3) no tympanocentesis and assessment of clinical efficacy in clinical (total) acute otitis media. For a drug with a bacteriologic efficacy of 100%, calculated clinical efficacy was 93% for bacterial acute otitis media and 89% for clinical acute otitis media. For a drug with bacteriologic efficacy of 27%, a rate consistent with no antibacterial therapy, efficacy was 71% for bacterial acute otitis media and 74% for clinical acute otitis media. We conclude that if efficacy is measured by symptomatic response, drugs with excellent antibacterial activity will appear less efficacious than they really are and drugs with poor antibacterial activity will appear more efficacious than they really are. The predominant phenomenon is that drugs with poor antibacterial activity will appear to be clinically effective in the treatment of acute otitis media.