Atrial fibrillation diagnosis in patients with single- and dual chamber ICD devices in England: a real-world cohort study

FUNDING ACKNOWLEDGEMENTS: Type of funding sources: Private company. Main funding source(s): BIOTRONIK SE & Co. KG BACKGROUND AND PURPOSE: Atrial fibrillation (AF) detection and subsequent treatment is known to be sub-optimal in the general population. It is unclear how implanted devices with atrial sensing contribute to AF diagnosis. This study compares real-world AF diagnosis rates in patients with ICDs without (single chamber ICD, VR ICD) and with atrial sensing (dual chamber ICD, DR ICD) and contrasts them with detection rates in trials which are estimated to be 200 per 1000 person years (Reinhold, 2018). METHODS: Linked primary care data from the Clinical Practice Research Datalink (CPRD), secondary care data from Hospital Episode Statistics (HES) and mortality statistics from the Office of National Statistics (ONS) were used. All patients who received a VR- or DR ICD between 1 April 2016 and 31 October 2020, with at least 12 months primary care registration and no AF diagnosis prior to ICD implant were included in the analysis. Time to AF detection was calculated; in Cox proportional hazards regression analysis comorbidities, age, gender, and disease aetiology were adjusted for. Stroke rates were estimated for the total cohort. RESULTS: 835 patients with a VR ICD and 1034 patients with a DR ICD were identified. Both patient groups had similar comorbidity profiles, with VR ICD patients being on average younger (mean 57.5±13.9 vs 61.5±15.4 years). During 2700 and 3600 person-years follow-up the incidence of new AF diagnoses was 45.6 (CI: 37.6-53.7) per 1000 person years in VR ICD and 54.7 (CI: 47.1-62.4) per 1000 person-years in DR ICD patients. Median time to AF diagnosis was also similar at 36.3 (IQR: 26-49.6) and 36.9 (IQR: 25.2-51.5) months, respectively. In the adjusted analysis, ICD-10 or SNOMED coded diagnosis of AF was not significantly different between groups (HR: 0.88 [95% CI 0.70-1.10]; p-value=0.3). Stroke rates in the total cohort (with or without AF) were 5.7% in the VR ICD and 3.4% in the DR ICD cohort at 36 months. CONCLUSION: The analysis highlights a difference between the rate of clinically relevant AF observed in patients with cardiac implanted electronic devices (CIED) in clinical trials and in real-world practice. The occurrence of stroke appears to also be higher in real world practice than in clinical trial populations and this is particularly marked in patients with VR ICD. These findings could indicate low use of remote monitoring or clinician uncertainty about the clinical relevance of CIED-detected AF in the real world. The latter may be because of a still accumulating body of predictive evidence towards the surveillance of AF. We propose that consistent use of remote monitoring with daily transmission in combination with CIEDs with atrial sensing will enable improvement of care.