Single-lead patch-based ambulatory external ECG (AECG) monitoring devices are being used for diagnosis of atrial fibrillation (AF). These AECG systems enable continuous long-term monitoring outside of the clinic. AECG service providers leverage human technicians and algorithms to analyze raw data and distill clinically relevant metrics like AF burden into a daily or end-of-study report for the prescribing clinician. The AECG monitoring devices can be used for up to a month; however, in many instances, monitoring is terminated after just 1-2 weeks. The loss of diagnostic yield with respect to AF remains unknown.
In a retrospective analysis, AF statistics were noted for 25,457 randomly selected patients who had undergone AECG monitoring in the mobile cardiac telemetry (MCT) mode for at least 28 days with the BodyGuardian® Heart device (Preventice Solutions, Inc.). MCT is the only non-invasive ambulatory service that provides comprehensive holter-like data for up to 30 days. For these patients, each detected AF episode duration was noted, to calculate the AF burden, as well as the time it took from the start of the monitoring period for AF episodes of different minimum durations to manifest. The data was analyzed altogether, as well as in groups by diagnosis code. The groups formed based on diagnosis codes were palpitations patients, syncope patients, stroke patients and AF patients.
Atrial Fibrillation (AF) episode detection was performed by Preventice which combines certified ECG technician oversight with algorithmic AF detection using the Preventice BeatLogicTM deep learning platform.
At least one AF event was present for 4,033 patients (15.8%), with the majority of detections occurring on the first monitoring day. For 424 patients, the first AF event detection occurred after monitoring day 14. Out of the 4,033 patients with at least one AF episode, 2,250 were captured after one day of monitoring. Had the monitoring study ended at 14 days, the first AF event would have been missed for 424 (10.5%) patients.
Filtering AF events by duration highlighted cases where the first AF detection was sustained for several minutes or hours and occurred after 14 days. In this analysis, AF episodes with duration shorter than the specified minimum specified were excluded. For example, for ‘2 min’ minimum duration, all AF episodes that were shorter than 2 min were filtered out, for ‘6 min’, all AF episodes that were shorter than 6 min were filtered out, and so on. Patients had sub-clinical (< 2 min) episodes of AF more consistently throughout their studies compared to longer episodes of AF which leads to the higher additional patients detected at 28 days vs 14 days for the longer AF episodes. Monitoring for 28 days verses 14 days resulted in 352 (8.7%) more patients with at least one AF episode that was longer than 1 hour, and 181 (4.5%) more patients with at least one AF episode that was longer than 6 hours.
Figure 4 shows a scatter plot of the AF burden in days 15-28 vs that in days 1-14. The deviation from the diagonal here indicates different burdens in the two time periods. Additional analysis was carried out by grouping the data by the diagnosis code for the patient study. Four broad categories were considered – palpitations patients (n=10012), syncope patients (n=4072), stroke patients (n=1219) and AF patients (n=4678). Kaplan Meier curves for the first documented AF episode for each of these subpopulations are shown in the figures below. Each plot contains 6 curves for the different minimum duration requirements of an AF episode.
In an abbreviated (14 day) monitoring study, the first occurrence of AF would have been missed in 10%-32% of the patient population, depending on the minimum AF episode duration. First AF detections that occurred after 14 days were most common for patients with the diagnosis code, AF, but also occurred for patients with a diagnosis code of palpitations, syncope and stroke. Additionally, many patients were observed to have different AF burden in the monitoring days 1-14 and 15-28, which may have implications for diagnosis and treatment. These results establish the importance of monitoring beyond 14 days.
In patients undergoing AECG monitoring, if detection of AF is the goal, monitoring should be performed for 28 days. While there may be value to longer monitoring, 28 days is the longest feasible duration based on current coverage guidelines. Monitoring for 14 days would have resulted in missing the first AF event for more than 10% of patients, including many of which completed the full 28-day duration study with sustained AF events lasting greater than 1 hour.