Provider Preferences for Interventions for Reducing Inappropriate Antibiotic Prescribing
Antibiotic resistance leads to more than 94,000 life-threatening infections and 18,500 deaths in the U.S. annually. Physicians and other prescribers have resisted efforts for reducing inappropriate antibiotic prescribing. Each year there are more than 41 million antibiotic prescriptions for acute respiratory infections (ARIs) of which 55% (22.6 million) are estimated to be inappropriately ordered for non-bacterial infections.
Methods
As part of a multi-center behavioral economics randomized controlled trial (RCT) testing alternative computerized prescription order entry (CPOE) interventions to reduce ARI antibiotic over-prescribing, a discrete choice experiment (DCE) was developed and implemented to evaluate prescriber stated preferences for alternative health plan interventions to reduce ARI antibiotic over-prescribing. The DCE included the three RCT interventions; i) default CPOE screen with non-antibiotic suggested alternative (SA) prescriptions, ii) accountable justification (AJ) CPOE screen requiring the prescriber to write brief medical justifications for ARI antibiotic orders, and iii) peer comparison (PC) intervention where the prescribers received regular periodic communications matching their own rate of ARI antibiotic prescribing with that of the top 10% of their peers. The DCE also included two non-tested interventions; a) pay-for-performance (P4P) alternative that rewarded prescribers up to $200 per month for substantial reductions in antibiotic over-prescribing and b) five minutes per ARI patient (5PP) of additional office visit time to explain to the ARI patient why antibiotics would be unnecessary or harmful.
Using these five alternatives the DCE fractional factorial design compared subject prescriber binary choice preferences for hypothetical alternatives implemented by health plans. The binary choices compared unique combinations of the alternatives to assess the relative preference strengths in influencing choice. Data were collected in in a web-based Limesurvey format using ten repetitions per subject and analyzed in SPSS Statistics Version 20. Stated choice preferences will ultimately be compared to revealed choice preferences in the RCT study.
Results
The preliminary DCE was implemented with 52 primary care Los Angeles area prescribers (43 physicians, 3 nurse practitioner and 6 physician assistants). The AJ alternative (Odds Ratio 1.03: 95% CI 0.942 to 1.830) and PC alternative (Odds Ratio 1.01 : 95% CI 0.749 to 1.428) did not elicit statistically significant favorable or unfavorable prescriber preference. The SA alternative had a significant positive preference response (Odds Ratio 1.464 : 95% CI 1.046 to 2.045); a much larger effect than those elicited by the P4P (Odds Ratio 1.005 : 95% CI 1.003 to 1.007) or 5PP (Odds Ratio 1.122 : 95% CI 1.050 to 1.200) alternatives which were also statistically significant.
Conclusions
While SA, P4P and 5PP were preferred; providers strongly preferred a suggested alternatives prescribing intervention to pay-for-performance or extending office visit time for ARI patient education on proper antibiotic prescribing. Moreover SA would cost almost nothing at the margin after CPOE implementation while P4P and 5PP would cost thousands of dollars annually per prescriber to achieve the same level of provider acceptance. Providers were indifferent to choosing peer comparison and CPOE accountable justifications as interventions for influencing antibiotic prescribing. These stated preference results are at odds with the findings from the RCT interventions.