Clustering of cardiovascular and renal comorbidities and risk of incident atrial fibrillation in the community-based PREVEND cohort

Abstract

Abstract Funding Acknowledgements Type of funding sources: Foundation. Main funding source(s): The PREVEND study is supported by the Dutch Kidney Foundation and the Netherlands Heart Foundation. Background Atrial fibrillation (AF) is a condition that occurs in the presence of comorbidities. With the accumulation of comorbidities over time, certain comorbidities may more often occur together than others, i.e. clustering of comorbidities. Acknowledging clustering of certain comorbidities may have implications for diagnosing previously unrecognized comorbidities, and clusters of comorbidities may carry differential risks of diseases, like AF. Albeit that the individual cardiovascular risk factors and comorbidities are well known, information on the impact of clustering of these on incident AF is sparse. Purpose We aimed to investigate the presence of clusters of cardiovascular and renal comorbidities and study the association between comorbidity clusters and incident AF. Methods We used the community-based Prevention of Renal and Vascular ENd-stage Disease (PREVEND) cohort in which 8,592 individuals participated. We excluded individuals with prior AF or missing ECG data, leaving 8,265 individuals for analysis. Atrial fibrillation was diagnosed if either AF or atrial flutter was present on a 12-lead ECG during one of the study visits, or at an outpatient visit or hospitalization. Latent class analysis was performed to assess clustering of 10 cardiovascular and renal comorbidities. The optimum number of classes was determined by the number of classes for which the Bayesian information criterion (BIC) reached a minimum value. As secondary analysis, latent class analysis was repeated with additionally including age, sex, and ethnicity. Kaplan-Meier analysis was performed to calculate the cumulative probability of incident AF, stratified by the latent classes. Results In the total population, the mean age was 48.9±12.6 years and 50.2% were women. During 9.2±2.1 years of follow-up, 251 individuals (3.0%) developed AF. Latent class analysis resulted in a model with three clusters as the optimal model, with one cluster being young (44.5±10.8 years) and healthy, carrying a low (1.0%) risk of incident AF; one cluster being older (63.0±8.4 years) and multimorbid, carrying a high (16.2%) risk of incident AF and a third middle-aged (57.0±11.3 years), obese and hypertensive cluster carrying an intermediate risk (5.9%) of incident AF. When adding age, sex and ethnicity to the latent class analysis, similar results were observed. Conclusion We identified three clusters of individuals in the community-based PREVEND cohort. The three clusters contained different amounts of comorbidities carrying different risks of incident AF. However, there were no differences between clusters regarding specific combination(s) of comorbidities.