The risk of incident atrial fibrillation in patients with inflammatory arthritis: a hierarchical clustering approach from the Hong Kong Hospital Authority Data Collaboration Laboratory (HADCL)

Abstract Background Patients with inflammatory arthritis have a higher risk of developing atrial fibrillation (AF). Risk factors do not act independently but tend to interact, forming clusters of interrelated factors. These aggregated risks may create distinct clinical profiles leading to varying levels of AF risk in patients with inflammatory arthritis. Purpose To identify different clusters in patients with inflammatory arthritis that may correspond to varying risks of AF. Methods This retrospective cohort study utilised data from the Hospital Authority Data Collaboration Laboratory (HADCL) in Hong Kong. Patients diagnosed with rheumatoid arthritis (RA) or psoriatic arthritis (PsA) were included in the study. Hierarchical clustering, an unsupervised machine learning technique, was used to identify distinct clusters within the population. The incidence rates of AF were compared across different clusters, and survival probabilities were analysed using Kaplan-Meier survival curves and log-rank test. Results A total of 18,034 patients (mean age 57.1, SD 14.1 years; 75.2% female) were included in the study. Four clusters were identified: Cluster 1 "High Cardiometabolic Burden Cluster 2 "Low Cardiometabolic Burden Cluster 3 "High Cardiometabolic Burden Cluster 4 "Low Cardiometabolic Burden & Low Inflammatory Burden". Detailed characteristics of each cluster are presented in Table 1. The prescription rate of glucocorticoid was extremely low in cluster 3 and cluster 4, while it was more predominant in cluster 1 and cluster 2. After a mean 8.4 (SD: 4.5) years of follow-up, 484 (2.64%) AF were observed. Cluster 1 exhibited the highest incidence of AF, followed by cluster 2 and cluster 3, whereas cluster 4 had the lowest incidence (Figure 1). Conclusion Patients with inflammatory arthritis exhibit differential risks of incident AF based on distinct clinical profiles. Individualized management tailored to these profiles may provide an optimal strategy for AF risk mitigation in this population.