Abstract 13826: Atrial Fibrillation Symptom Clusters Increase Hospitalizations and Emergency Department Visits

Abstract

Introduction: Symptoms are a predictor of hospitalizations in adults with atrial fibrillation (AF). Symptom clusters are groups of two or more symptoms that are related due to shared mechanisms or a shared impact on outcomes. The impact of AF specific symptom clusters on hospitalization and emergency department (ED) visits has not previously been examined. Hypothesis: We assessed the hypothesis that symptom clusters were associated with AF-related hospitalizations and ED visits. Methods: We examined pre-existing data for 1,501 adults in the Vanderbilt AF Registry, a clinical biorepository that prospectively enrolls both AF patients and their family members. Our sample consists of subjects with confirmed AF and a completed symptom survey (derived from the University of Toronto AF Severity Scale, Chronbach’s α for symptom burden is 0.94). Seven common AF symptoms were measured on a 5 point Likert scale. Our outcomes were self-reported number of AF-related hospitalizations and ED visits in the past 12 months. To identify symptom clusters we used hierarchical cluster analysis, Ward’s method, and Euclidean distance as the dissimilarity measure. The ideal number of clusters was determined with dendrograms, pseudo F, and pseudo T. We used Poisson regression to examine whether the identified clusters were associated with hospitalization and ED visit rates controlling for clinical characteristics. Results: Our sample was primarily male (67%) and Caucasian (96%) with a mean age of 58.4 (± 12.2) years. Two symptom clusters were identified, which we labeled the At Rest cluster (fatigue at rest, shortness of breath at rest, chest pain, and dizziness) and the With Activity cluster (shortness of breath with activity and exercise intolerance). Palpitations was the only symptom that did not cluster with other symptoms. Subjects who experienced all the symptoms in the At Rest cluster had an adjusted incident rate ratio for hospitalizations of 1.9 (p<0.0001) and for ED visits of 2.8 (p<0.0001). There was no statistically significant association between the With Activity cluster and either clinical outcome. Conclusion: We identified two AF-specific symptom clusters. Only the At Rest cluster was significantly associated with AF-related hospitalizations and ED visits.