Abstract 10428: Probability of Atrial Fibrillation (AF) Diagnosis and Error in Estimation of Burden in People With Paroxysmal AF Based on the Duration of Monitoring

Introduction: In individuals with paroxysmal atrial fibrillation (PAF) undergoing heart rhythm monitoring, total duration of monitoring can influence both the probability of recognizing AF and estimation of burden. We estimate the probability of AF detection and quantify the error in burden estimation based on the duration of monitoring up to 10 days. Methods: Data included 5361 ECG patch recordings up to 14 days long with at least 1 episode of AF. 100 samples of pre-specified lengths were drawn at random from each recording. For each sample length, the probability of identifying any AF, and the AF burden over shorter periods of monitoring were determined. The absolute error in burden estimation over short monitoring divided by the true AF burden during up to 14 days of monitoring was used to determine relative deviation. We stratified the expected relative deviation and probability of detecting AF by high (>10%), moderate (2-10%), or low (<2%) true AF burden. Results: Estimated probability of detecting any AF was significantly affected by duration of monitoring (Fig): 1 day of monitoring identified 76% [75-77%], 52% [51-54%] and 35% [33-36%] of individuals with high, moderate, and low burden of AF. Relative deviation in the estimation of AF burden is also significantly impacted by monitoring duration with 1 day recording, with 0.89 [0.87-0.91], 1.45 [1.43-1.47], and 1.65 [1.63-1.67] for high, moderate, and low burden. Even for 7 days recording, the expected deviation is significant, with 0.31 [0.30-0.33], 0.60 [0.58-0.61], 0.79 [0.78-0.80] for high, moderate, and low burden (Fig). Conclusions: In people with PAF, known or unknown, the duration of continuous ECG recording can significantly impact both ability to diagnosis AF and accurate determination of AF burden. Even in individuals with true AF burden >10%, 24 hours of monitoring will not detect AF in >20% of people and provide an almost 90% relative error in the estimate of AF burden.