Multimorbidity and risk of atrial fibrillation in the Lifelines Cohort

Abstract Background Atrial fibrillation (AF) occurs in the presence of comorbidities. With the increasing incidence of both AF and comorbidities over the last decades, multimorbidity is currently present in the majority of patients at the time of AF diagnosis. Associations of individual, mainly cardiovascular comorbidities with incident AF are well-studied. However, the impact of multimorbidity, the role of non-cardiovascular comorbidities, and potentially clustering of comorbidities on incident AF remains unclear. Purpose To investigate the association between the number and clustering of (non-)cardiovascular comorbidities with incidence of AF in a population-based observational cohort. Methods Lifelines is a prospective community-based cohort study designed to investigate health and health-related behaviours. Baseline assessment was performed between 2007-2013. For the present study, we studied 25 (non-)cardiovascular comorbidities in Lifelines participants. Logistic regression was used to study the association between the number of comorbidities and incident AF. Latent class analysis was used to identify comorbidity clusters. Results In the study population of 76,648 participants, mean age was 46.4±2.6 years and 59.3% were women. In this population, 56,034 (73.1%) participants had ≥2 comorbidities, 42,575 (55.5%) ≥2 cardiovascular comorbidities and 14,612 (19.1%) ≥2 non-cardiovascular comorbidities. After a mean follow-up of 3.70±0.95 years, 188 (0.2%) participants developed incident AF. After adjusting for age and sex, a greater total number of comorbidities (OR 1.10 for each additional comorbidity, 95% CI [1.01-1.19]) and cardiovascular comorbidities (OR 1.18 for each additional comorbidity, 95% CI [1.06-1.31]) were associated with incident AF; however no association was detected for the number of non-cardiovascular comorbidities (Figure 1). We identified 12 comorbidity clusters associated with varying risks of incident AF (AF incidence rate range 0.00 to 0.58 per 100 person-years, P<0.001) (Figure 2). Rather than specific combinations of comorbidities, the clusters seemed to be based on the number of comorbidities present. Some clusters were relatively healthy with low relative risk of comorbidities (green squares), while other clusters were relatively unhealthy with higher relative risk of comorbidities (orange/red squares). The median number of comorbidities ranged from one to seven. Clusters with higher AF incidence rates were predominantly determined by the presence of cardiovascular comorbidities. Conclusion There was a dose-dependent relationship between the number of total comorbidities and cardiovascular comorbidities and risk of incident AF, but not for non-cardiovascular comorbidities. We identified 12 comorbidity clusters with different risks of incident AF, however, the number of comorbidities rather than a specific pattern of comorbidities was associated with an elevated risk AF.