PCV70 CLINICAL DECISION SUPPORT INTERVENTIONS FOR PROMOTING GUIDELINE-CONCORDANT ANTICOAGULATION IN PATIENTS WITH ATRIAL FIBRILLATION: TARGETED LITERATURE REVIEW OF KEY FEATURES AND GAPS IN EVIDENCE

To inform the design and implementation of a care improvement strategy using clinical decision support (CDS) in patients with atrial fibrillation (AF), we conducted a targeted literature review to synthesize existing evidence regarding use of CDS in promoting evidence-based anticoagulation of patients with AF. The search was conducted using PubMed in August 2019. The search algorithm aimed to identify articles examining CDS in AF related to anticoagulation published within the last 5 years. These articles were reviewed using a two-step screening method ([1]title/abstract;[2]full text) with inclusion/exclusion criteria refined at each step and outstanding items resolved by consensus. A data extraction table was used to identify trends and potential gaps in existing literature. The initial search resulted in 178 articles with 151 excluded at Step-1 and 21 excluded at Step-2 for a final list of 6 publications. Studies were conducted in Western regions (2 US; 4 Europe), were distributed equally between hospitalized:primary care settings (3:3), referenced treatment guidelines related to region and publication year, and featured varied study designs (3 cluster-randomized; 2 randomized controlled; 1 cohort), and AF definitions. CDS tools described included stroke-risk calculators (4 fully automated; 1 partially), electronic alerts (3 forced interaction; 1 optional interaction; 2 passive reminder), and order sets. More AF patients at high risk for stroke were anticoagulated following active alerts compared to passive alerts, and only alerts that forced action/interaction resulted in statistically significant improvements (p< 0.05). Effective CDS interventions to improve guideline-concordant prescribing in AF should feature active alerts with forced real-time provider action/interaction. The body of literature would be strengthened by additional studies of the accuracy of AF identification through computerized record abstraction, accuracy of stroke-risk autocalculators, important factors for CDS usability such as appropriate alert triggers, suggested actions, and features for provider feedback, and the effect of these tools.