Implementation of digital patient decision support tools for Atrial Fibrillation treatment: systematic review

Abstract Background/ Introduction Treating atrial fibrillation (AF) involves complex decisions. Digital patient decision-support tools can facilitate shared decision making by significantly reducing decisional conflict and improving patient knowledge [1,2]. Yet, implementation in clinical practice is lagging [3]. Purpose To systematically review trials of digital patient decision-support tools for AF treatment decisions, focusing on factors influencing their implementation in clinical practice. Methods We conducted a systematic review following Cochrane methods. EMBASE, MEDLINE and Scopus were searched from 2005 to 2023. Eligible randomised controlled trials and quasi-experimental trials that evaluated digital patient decision-support tools for AF treatment decisions were included. Two independent reviewers screened title and abstract, extracted data and assessed risk of bias. The implementation outcomes included a description of when and how the tool was used, as well as its characteristics and user satisfaction. Results Thirteen articles reporting on eleven studies (5 RCTs and 6 quasi-experimental trials) met the inclusion criteria (2,714 participants; mean age 71±4.2 years; 26% women; 92% with non-valvular AF, remaining not reported). Risk of bias was low for 2 studies, moderate for one and high for seven studies. All tools (n=11) focused on anticoagulation choice (1: warfarin vs no warfarin; 1: aspirin vs warfarin; 7: Direct Oral Anticoagulants (DOAC) vs antiplatelet or warfarin; 2: unspecified choice). The tools were used either pre-visit (at home or in the waiting room) or during the consultation (n=5). Tools were delivered using a mobile application (n=4), web-based application (n=5), or a desktop (n=2). Of the eleven tools, 2 could automatically calculate individualized stroke risk using data from electronic health records. Most articles reported favourable user feedback regarding use of the digital patient decision-support tools, such as high perceived usefulness, user-friendliness, and overall satisfaction. No significant difference was found in encounter times between intervention and control arms (2 RCTs). Four tools were publicly available and 3 were implemented in healthcare delivery beyond research. Conclusion Despite demonstrated efficacy in reducing decisional conflict and improving patient knowledge, as well as positive clinician and patient experience, only 3 tools were implemented in clinical practice beyond their trial. Future tools could better leverage digital capabilities to integrate with Electronic Health Records, automate stroke and bleeding risk calculations, and personalise content based on patient values and preferences. Additional robust trials and implementation studies are warranted to understand barriers and enablers to the use of these tools.