Relative Influence of Antibiotic Therapy Attributes on Physician Choice in Treating Acute Uncomplicated Pyelonephritis

Background. Reducing excess duration of antibiotic therapy is a strategy for limiting the spread of antibiotic resistance, but altering physician practice to accomplish this requires knowledge of the factors that influence physician antibiotic choice. The authors aimed to quantify physician willingness to trade between 4 attributes of antibiotic therapies: different therapy durations, failure rates, dosing frequencies, and days of diarrhea as a side effect when treating acute uncomplicated pyelonephritis. Methods. The authors distributed conjoint analysis questionnaires to physicians enrolling patients in a randomized trial comparing 2 antibiotics in pyelonephritis treatment. For each question, respondents were required to select 1 of 2 antibiotics based on the values of the 4 attributes. Proportional hazards regression was used to model predictors of physician choice. Results. Eighty-seven of 88 physicians completed the questionnaire. Duration of therapy, days of diarrhea, and failure rate were significant predictors of choice (P < 0.05), but dosing frequency (once daily v. twice daily) was not. Increasing days of diarrhea greatly reduced the probability of an antibiotic being chosen. If failure and side effects were equivalent, physicians were more likely to prescribe a 5- v. 10-day duration of therapy (odds ratio = 4.18, P < 0.01). Conclusion. Antibiotic choice is most influenced by physicians' desires to limit treatment failure and side effects, although physicians were willing to accept increases in treatment failure to obtain reduced days of diarrhea as a side effect. Because shorter-course therapy is frequently associated with fewer side effects, efforts to encourage physicians to choose shorter treatment durations should include mention of reduced treatment-associated side effects.