Criteria-based diagnosis and antibiotic overuse for upper respiratory infections

Studies of antibiotic overuse often rely on physicians' reports of diagnoses, which can overestimate bacterial illness. To assess potential overdiagnosis, we determine bacterial upper respiratory infection diagnoses by direct observation of visit videotapes.

From an observational study of videotaped visits for upper respiratory symptoms (N = 66), coders assessed diagnostic criteria (symptoms, physician description of physical examination findings, and diagnostic tests), physician diagnosis, and prescribing. Survey data included patient demographics and health care utilization as well as physician/practice characteristics (n = 15).

Criteria-based diagnoses were determined from coded diagnostic criteria. Interrater reliabilities were determined for 33% (n = 22) of visits. Chi-square tests assessed concordance between the physician's diagnosis and the criteria-based diagnosis and compared rates of antibiotic overuse as determined from physician and criteria-based diagnoses.

The criteria-based diagnosis agreed with 100% of physicians' diagnoses of streptococcal pharyngitis and 73% of physicians' acute otitis media diagnoses but with only 17% of physicians' sinusitis diagnoses. Antibiotic overuse occurred in 11% of visits based on physicians' diagnoses but in 32% of visits when criteria-based diagnoses were considered, a difference of 21% (95% confidence interval, 2%-38%; P < 0.05).

Criteria-based diagnoses revealed that antibiotic overuse occurred 3 times more frequently than suggested by physician diagnoses. Concordance between physician and criteria-based diagnoses was lowest for sinusitis. Future studies should consider the contribution of overdiagnosis to antibiotic overuse and target this practice to further reduce overuse.