Abstract Background Physicians prescribing antibiotics for common infections must weigh trade-offs across drug attributes such as efficacy, dosing, side effects, and resistance. Understanding these priorities can help inform drug development, regulatory decisions, and insurance coverage determinations. Methods We conducted a discrete choice experiment between June 5 and 9 July 2024, among a national sample of US physicians who prescribed antibiotics for community-acquired pneumonia (CAP) or complicated urinary tract infections (UTIs) in the past year. Respondents evaluated paired hypothetical antibiotics varying by 5 attributes: time to symptom improvement, dosing frequency, risk of nonserious and serious side effects, and risk of future resistance to the patient. Preference weights and relative importance scores were estimated using conditional logistic models. Subgroup analyses were conducted by disease, care setting, and specialty. Results Of 880 enrolled physicians, 756 (86%) completed the survey. Respondents had a mean age of 51.5 years; 60% were male, and most practiced general internal medicine (64%) or infectious disease (15%). The most influential attributes overall were symptom improvement (relative importance score: 28%) and dosing convenience (relative importance score: 27%). In inpatient settings, physicians prioritized symptom improvement (relative importance score: 33%), while outpatient physicians prioritized dosing frequency (relative importance score: 31%). Risk of future antibiotic resistance to the patient was consistently the least influential attribute (relative importance score: 7%–13%) across disease-types and clinical settings. Conclusions In this national survey study, physicians prioritized rapid symptom relief and dosing convenience over other drug attributes when prescribing antibiotics for CAP and UTIs. Understanding physician priorities can help inform stewardship strategies and clinician-facing decision support, and encourage regulators and sponsors to prioritize clinically meaningful trial endpoints.
Hong et al. (Fri,) studied this question.