Patterns of comorbidities and influence on the natural history of patients with atrial fibrillation: a latent class analysis from a prospective observational european AF registry

Abstract

Abstract Background Patients with Atrial Fibrillation (AF) usually present with multiple comorbidities at diagnosis, and their combinations heterogeneously influence management and outcomes. While the effect of multimorbidity on the natural history of AF is well-known, the influence of patterns of multimorbidity is far less understood. Purpose To identify and analyse subgroups of AF patients according to their comorbidities, through a latent class analysis of a contemporary cohort of AF patients. Methods We used data from a European-wide prospective observational AF registry, and performed a latent class analysis, according to the presence of chronic diseases (including cardiovascular, respiratory, metabolic and other conditions); class membership was assigned according to the modal posterior probability. Multivariable logistic regression models were used to analyse odds of receiving oral anticoagulant (OAC) prescription and type of OAC received; the risk of the primary outcome of all-cause death and MACE was assessed using multivariable Cox-regression model. Results 9,613 AF patients were included in this analysis (mean age 68.9 ± 11.4 years, 40.2% females), and were grouped in 5 classes, which were different according to the burden of comorbidities: low morbidity pattern (46.1%), cardiovascular pattern (25.0%), metabolic pattern (11.3%), heart failure pattern (9.7%) and multisystemic pattern (8.0%) (Figure 1). Patients in the multisystemic pattern were older and more frequently frail and with polypharmacy compared to other classes. Compared to the low morbidity pattern, odds of receiving OAC was lower in those with cardiovascular pattern (OR [95%CI]: 0.73 [0.62-0.86]), heart failure pattern (OR [95%CI]: 0.60 [0.48-0.75] and multisystemic pattern (OR [95%CI]: 0.28 [0.22-0.36]), while those with metabolic pattern were more likely prescribed OAC (OR [95%CI]: 1.55 [1.20-2.00]). All patterns were less likely treated with NOAC compared to VKA (Figure 2). When analysing the risk of the primary composite outcome, all patterns except for the metabolic pattern were associated with higher hazard of events, with highest magnitude observed for the heart failure pattern (HR 2.68 [95%CI: 2.22-3.25]) and the multisystemic pattern (HR 2.67 [2.13-3.35]) (Figure 2). Conclusions Comorbidities are heterogeneously distributed in the AF population, and different phenotypes of AF patients may be identified according to the combination of other chronic diseases, with different OAC use for stroke prevention and an unequal impact on long-term prognosis. These findings reinforce the need for tailored preventive strategies to improve outcomes in AF patients, especially those with multimorbidity.Figure 1Figure 2