Timing of Electronic Health Record Integrated Decision Aid (Idea) for Stroke Prevention in Atrial Fibrillation Matters

Stacey L. Schott, Keren Xu, Julia Berkowitz, Curtis Petersen, Catherine Saunders, Navjot Sobti and Megan Coylewright



Background: Although the use of shared decision making (SDM) in cardiology is increasingly recommended, how to best integrate SDM into clinical encounters is unclear. We tested the efficacy of an electronic health record integrated decision aid (IDeA) to determine its impact on knowledge of personalized stroke risk for patients with atrial fibrillation (AF) and its effect on the clinician-patient relationship.

Methods: A cluster randomized trial of 6 cardiovascular clinicians each paired with 11 new, or previously diagnosed AF patients (N=66) were blindly allocated to use either the IDeA or usual care during an inpatient or outpatient encounter when stroke prevention treatment options were discussed. The primary outcome was knowledge about personalized stroke risk assessed before and after the encounter. Exploratory outcomes included differences in SDM, decisional conflict, confidence, and trust. Time spent on steps in the IDeA was automatically recorded. Semi-structured interviews with clinicians and patients who used the IDeA were conducted.

Results: There is not enough evidence to suggest this IDeA significantly increased patients’ knowledge of their stroke risk, (OR 3.22 95% CI: 0.69-15.01) and there was no significant difference in SDM, decisional conflict, confidence or trust. Despite training, each clinician used the IDeA differently. Associate providers spent significantly more time on the step displaying % stroke-risk in 1-year (50.22 seconds SD=46.27) than attendings (14.55 seconds SD = 7.87) (p = 0.03), but time spent on any step did not impact patient knowledge. Qualitative analysis revealed the IDeA’s usefulness is dependent upon context, existing patient knowledge and clinician skill. Patients suggested the IDeA may be most helpful at diagnosis.

Conclusion: Previous trials of non-integrated DAs have shown increases in patient knowledge, but in this study, which utilized an integrated DA in real-world clinical encounters, with a majority of previously diagnosed AF patients, these results did not hold true. The qualitative findings help explain this outcome and suggest that timing the delivery of IDeAs matters. Further research should examine IDeA use closer to diagnosis.

Github repo

Github repo