Discussion

To our knowledge, this is the first study that evaluated a web-based CDS tool on antibiotic prescribing knowledge and confidence in dental settings in the US, which is now openly available at dds.tips. Our findings demonstrate that releasing a web-based tool was not associated with an increase in confidence or knowledge in antibiotic prescribing among survey respondents, despite regular reminders and education at an academic dental center.

Our findings reinforced the need for dental antibiotic stewardship. The survey responses show that respondents are not particularly confident about their antibiotic prescribing decisions and frequently continue to prescribe antibiotics when they are not necessary based on clinical vignettes. Furthermore, our pooled and correlation results showed that less confident dentists are more likely to prescribe antibiotics—further suggesting that additional education or CDS tools would be beneficial.

However, our findings suggest that additional work, beyond providing a web-based tool, may be necessary to improve antibiotic prescribing confidence and practices. Prior studies have demonstrated that academic detailing and audit and feedback approaches may be beneficial in dentistry. ^{Reference Elouafkaoui, Young and Newlands13,Reference Seager, Howell-Jones, Dunstan, Lewis, Richmond and Thomas19,Reference Chehabeddine, Lahoud, Noujeim, Zeidan, Hleyhel and Saleh20} However, simple interventions such as paper-based tools also appear to be effective at reducing antibiotic prescribing in dental settings. ^{Reference Gross, Hanna, Rowan, Bleasdale and Suda17} It is likely that some of these interventions will be more effective when combined or modified using human-centered design approaches. For example, dentists often feel pressure to prescribe antibiotics when they are recommended by physicians, even when they know it is wrong. ^{Reference Ramanathan, Yan and Suda21} Many antibiotics prescribed in dental settings may fall outside of guidelines, which would fall outside of our tools. ^{Reference Ramanathan, Yan and Suda21} Several additional factors, such as patient expectations, fear, treatment delays, and diagnostic uncertainty also play a role in antibiotic prescribing according to a recent meta analysis. ^{Reference Vázquez-Cancela, Zapata-Cachafeiro, Herdeiro, Figueiras and Rodríguez-Fernández22}

There may have been unique limitations with this study that limited the strength of our findings. We deployed the tool in the first US dental school to implement the Centers for Disease Control and Prevention Core Elements of Outpatient Antibiotic Stewardship. This may have limited the opportunities for additional improvements in antibiotic use. Anecdotally, some of our earlier qualitative work when deploying the tool included quotes from students that they infrequently or have never prescribed antibiotics. ^{Reference Seager, Howell-Jones, Dunstan, Lewis, Richmond and Thomas19} It is also possible that the login requirements reduced utilization from study participants. Low utilization could for a variety of reasons. First, the tool may not have been user friendly. We think that this is unlikely as pilot testing was performed and some aspects of the tool were based on existing online and paper resources. Second, the tool may have been cumbersome from a workflow perspective. This is very possible, as our prior qualitative work suggested that some students preferred a tool that was embedded into the electronic health record. ^{Reference Seager, Howell-Jones, Dunstan, Lewis, Richmond and Thomas19} Third, low uptake could have been a sign of little need for the tool. As discussed before, prior interviews also highlighted that several students reported rarely or never prescribing antibiotics in their training. This lack of prior antibiotic exposure may have decreased the likelihood that dentists would consider opening the tool to help with antibiotic prescribing. Fourth, the tool was deployed alongside a longstanding paper tool available chairside, which may have been easier for participants to access. Fifth, some of the low utilization may be because students have a faculty member present overseeing their work. In this scenario, it may be easier for dental students and residents to simply ask their attending, rather than find their phone or a computer to access the tool. Sixth, it is also possible that the tool needed to be embedded into the electronic health record system, which is what most participants desired during our interviews. Seventh, uptake of the tool was far lower than anticipated and did not improve substantially during the study period. Our study team added new features, including a pain module, without much uptake in users. Eight, here were temporal differences in the study time line, with earlier participants completing the survey 3 years before people completing it in the post period. This, in theory, could contribute to artificially increasing the likelihood of detecting a positive finding based on continued general education about antibiotic stewardship in dentistry. However, our null findings imply that this is unlikely. Ninth, we had originally anticipated tracking users over time using a combination of university and personal email addresses with web registration to measure implementation outcomes. Due to both partial and duplicate registrations, tracking users by website metrics was challenging. Finally, our study was also performed at a single academic dental center, and the results may have been different in diverse settings. Additional work, probably combining academic detailing with CDS tools, integrating CDS into the electronic medical record, or providing just-in-time education based on the clinical scenario should be explored.

In conclusion, releasing a web-based CDS tool at an academic dental center was not associated with an increase in dental provider knowledge and confidence in antibiotic prescribing decisions. Additional research identifying modified, additional, or combined interventions are needed to improve antibiotic prescribing in dental settings.