Clustering Patients with Paroxysmal Atrial Fibrillation

Abstract

Background: Atrial fibrillation is the most common cardiac arrhythmia in emergency departments and intensive care units. Objectives: We aimed to characterize critical care patients with paroxysmal atrial fibrillation who are at a higher risk for mortality or readmission. Methods: A secondary analysis was conducted on data from a large sample of patients admitted between January 2010 and December 2015 to the emergency department of a hospital who had blood electrolytes measured. Data from 520 individuals with paroxysmal atrial fibrillation were analyzed. A clustering algorithm was applied to identify patients with similar electrolyte profiles. Results: Two clusters of patients were identified (N1 = 163, N2 = 357) with significantly different proportions of poor outcomes: 57 (35.0%) vs. 66 (18.5%), respectively (P < 0.001). The high-risk cluster included older patients [74.4 (12.8) vs. 66.6 (14.7) years, P < 0.001] and had higher mean serum glucose levels compared to the low-risk cluster. Additionally, the high-risk group had lower concentrations of serum sodium, potassium, calcium, phosphorus, and magnesium (all P < 0.001). Endocrine and circulatory comorbidities were also more common in the high-risk cluster (both P ≤ 0.001). No significant difference was observed in the female sex ratio between the two clusters [81 (49.7%) vs. 152 (42.6%), P = 0.156]. Conclusions: Based on serum electrolyte concentrations, there are two distinct clusters with different prognoses in patients with paroxysmal atrial fibrillation. The high-risk cluster is characterized by older age, more comorbidities, higher serum glucose levels, and lower serum concentrations of sodium, potassium, calcium, phosphorus, and magnesium.