Evaluating a dynamic sequential pathway for atrial fibrillation detection in a tertiary stroke service

Abstract Background Longer rhythm monitoring in ischaemic stroke patients increases rate of Atrial Fibrillation (AF) detection (1,2). However, prolonged rhythm monitoring is expensive and inappropriate for patients that are already diagnosed with AF. Conversely, pathways reliant on human factors can delay or miss indicated investigations. The extent of inappropriate and missed investigations is unknown as is whether pathways for AF detection can reduce inappropriate requests and improve rate of AF diagnosis. Purpose To evaluate the impact of a dynamic sequential AF detection pathway on outcomes and workload in a large specialist stroke service. Methods Investigations and outcomes were reviewed in patients with ischemic stroke/TIA admitted to a large tertiary stroke service between January-March 2023. AF prevalence and detection on admission ECG and Holter/Zio monitoring were recorded as outcomes to determine appropriateness of onward investigation. Figure 1 outlines the interval pathway introduced to the service on 5th September 2023 based on National Guidance (3,4). Practice was re-evaluated in April-May 2024. The study was prospectively registered and endorsed by the Local Clinical Effectiveness Unit (study number 12024). Results Between January-March 2023, 109 people were admitted with an ischaemic stroke (n=98) or TIA (n=11). 13 patients had new AF detected on admission 12-lead ECG and 14 patients had known AF. 6/27 (22%) with known or newly diagnosed AF on admission ECG were inappropriately referred for further Holter/Zio monitoring for AF detection. 64/82 (78%) patients without an AF diagnosis had Holter/Zio monitoring requested. Of the 35 completed requests, 33 (94%) showed no AF. 0/33 (0%) had documented consideration or referral for implantable continuous monitoring (ICM). Between April-May 2024, after implementation of the AF detection pathway, 85 patients were admitted with an ischaemic stroke (n=81) or TIA (n=4). 2/27 (7%) patients with new or known AF were referred for Holter monitoring (X2 (1, N=130)=1.53, p=0.22). 58/58 (100%) patients without an AF diagnosis underwent Holter/Zio monitoring (X2 (1, N=140)=12.08, p<0.05) but no patients were considered for ICM. AF was detected in 2/35 (6%) patients with Holter/Zio monitoring between January-March 2023 compared to 1/21 (5%) patients in April-May 2024 (X2 (1, N=56)=0.02, p=0.88). 66/122 (54%) Holter/Zio requests from January-March 2023 (n=29) and April-May 2024 (n=37) are still being processed. Conclusion Inappropriate AF investigations are frequent, while indicated prolonged monitoring may be delayed or missed. A dynamic, sequential pathway can reduce unnecessary tests and ensure access to appropriate monitoring. Longer-term follow-up of clinical outcomes and AF detection is warranted.