Study participants were from the China-AF study recruited between August 2011 and December 2017. The China-AF is a prospective, multicenter, and hospital-based ongoing registry study. Thirty-one tertiary and non-tertiary hospitals in Beijing providing a clinical service of AF management participated in this registry study [12]. Adult outpatients or inpatients (≥ 18 years old) with a diagnosis of AF were included in the China-AF registry. For the present study, patients were excluded if they met any of the following criteria: (1) transient and reversible AF (e.g. caused by cardiothoracic surgery, hyperthyroidism, and binge drinking), (2) suffering from other serious diseases with a life expectancy < 1 year, (3) diagnosis of rheumatic mitral stenosis or having mitral valve prostheses, (4) HF-related New York Heart Association (NYHA) classification of IV or AF-related European Heart Rhythm Association (EHRA) symptoms classification of IV, (5) follow-up periods less than 6 months.

The study was approved by the Ethics Committee on Human Research in Beijing Anzhen Hospital, the Capital Medical University. All participants provided written informed consent.

Each patient’s demographic data such as age, sex, education level, and health insurance status were collected. AF-specific variables included AF type, duration, the severity of symptoms, and prior treatment. The cardiovascular risk factors were smoking, drinking, and obesity. Current or preexisting comorbid conditions consisted of hypertension (HTN), coronary artery disease (CAD, any history of MI, percutaneous coronary intervention or coronary artery bypass grafting), HF, cardiomyopathy, ischemic stroke/transient ischemic attack (TIA), bleeding, chronic obstructive pulmonary disease (COPD), gastrointestinal disorder, renal dysfunction, hyperthyroidism/hypothyroidism, diabetes mellitus (DM), and hyperlipidemia. The vital signs, laboratory tests, imaging examinations of patients were recorded. Medications of rhythm control agents, rate control agents, antithrombotic drugs, angiotensin-converting enzyme inhibitors (ACEIs)/angiotensin II receptor blockers (ARBs), and cholesterol-lowering agents were collected.

The primary outcome was all-cause hospitalization during follow-up. The secondary outcomes were cause-specific hospitalization and all-cause death. Patients enrolled were followed up every 6 months either by outpatient clinic visit or by telephone interview. The main cause for hospitalization was determined by the site investigators according to the discharge diagnosis, which was broadly classified into AF, other CVD (CVD excluding AF), and non-CVD. Incidence rates for all-cause and cause-specific hospitalization were calculated as the ratio of the number of hospitalizations to the total follow-up person-time using a unit of per 100 patient-years. The index hospitalization of inpatients recruited in China-AF study were not counted. Patients with recurring multiple hospitalizations during follow-up were counted multiple times.

Patients enrolled in this study were stratified by age (< 65 years; 65–74 years; ≥ 75 years). Continuous variables were presented as mean (standard deviation, SD) and categorical variables were shown as counts (proportions). The comparison of variables of patients in different age groups was analyzed using one-way analysis of variance (ANOVA) (for continuous variables), and chi-square tests (for categorical variables). Hospitalization rate (per 100 patient-years) was expressed by age and sex. Kaplan–Meier curve was used to describe the number of patients at risk, those free of hospitalization, and time to the first all-cause or first cause-specific hospitalization. The Fine-Gray modeling was conducted with (a) the first all-cause hospitalization, (b) the first AF hospitalization, and (c) the first other CVD hospitalization as the dependent variable, and death as competing risks, to evaluate factors that are significantly associated with all-cause and cause-specific hospitalizations. All baseline variables were included in the Fine-Gray modeling. Hazard ratios (HRs) [95% confidence interval (CI)] and P-values were presented. P-value < 0.05 was considered statistically significant. All analyses were performed using SAS statistical software version 9.4 (SAS Institute Inc., Cary, NC).

A total number of 23,108 patients from 31 sites from August 2011 to December 2017 were enrolled in the China-AF study. After applying the exclusion criteria, 20,172 AF patients were enrolled in this study (a mean age of 64.1 ± 12.0 years; 38.2% female). Their baseline characteristics stratified by age are presented in Table 1. The majority of patients (95.3%) were fully or partially covered by health insurance. More than 1/3 of patients in each age group have at least earned high school diplomas. Patients aged ≥ 75 years had a higher proportion of being female, having persistent AF, having EHRA class of III (AF-related symptoms), having current or preexisting comorbid conditions, including CVD, DM, COPD, and renal dysfunction, using ventricular rate control drugs, antiplatelet agents, ACEIs/ARBs and statins compared to patients in other age groups (all P < 0.01). Patients aged < 65 years tended to have more paroxysmal AF, more cardiovascular risk factors (47.7% overweight, 24.6% obese, 23.2% smoking and 28.3% drinking), a higher probability of receiving the radiofrequency catheter ablation (RFCA), and a more frequent use of antiarrhythmic drugs and direct oral anticoagulants (DOACs) (P < 0.01). There was no significant difference in the proportion of patients with AF duration ≥ 1 year in each age group.

Over a mean follow-up of 37.3 ± 20.4 months, a total number of 14,828 hospitalizations, 1,277 deaths were identified, 7,512 patients (37.2%) had at least one hospitalization. The overall incidence of all-cause hospitalization and all-cause death were 24.0 per 100 patient-years and 2.1 per 100 patient-years, respectively. Cause-specific hospitalization rates of AF patients by age groups and sex were shown in Fig. 1. Hospitalization rates of patients aged < 65, 65–74, and ≥ 75 years were 18.3 per 100 patient-years, 26.0 per 100 patient-years, and 33.5 per 100 patient-years, respectively. Hospitalizations occurred most in female patients aged ≥ 75 years and were more common among females of all age groups compared to their male counterparts. Among patients aged < 65 years, hospitalizations attributed to AF accounted for the largest proportion, reaching 42.1% of the all-cause hospitalizations, while patients aged 65–74 and ≥ 75 years were predominantly hospitalized for non-CVD (43.6% and 49.3% respectively). Kaplan–Meier curves of the first hospitalization for all-cause, AF, other CVD, or non-CVD causes were displayed in Fig. 2.

The multivariate analysis identified several risk factors of first all-cause hospitalizations in AF patients (Table 2). Elderly patients were more likely to have all-cause hospitalization. As indicated by our results, compared with patients in the < 65 age group, patients in the 65–74 age group (HR 1.18, 95% CI 1.12–1.25, P < 0.01) and in the ≥ 75 age group (HR 1.34, 95% CI 1.25–1.43, P < 0.01) were more likely to experience hospitalization. EHRA II (HR 1.13, 95% CI 1.03–1.24, P = 0.01) and EHRA III (HR 1.18, 95% CI 1.06–1.31, P < 0.01) showed a significantly predictive role for all-cause hospitalization. Paroxysmal AF, AF duration ≥ 1 year, and left atrial enlargement were highly associated with risks of all-cause hospitalization. In addition, HTN (HR 1.08, 95% CI 1.02–1.15), HF (HR 1.10, 95% CI 1.02–1.18), CAD (HR 1.24, 95% CI 1.17–1.33), and ischemic stroke/TIA (HR 1.22, 95% CI 1.15–1.30) also contributed to the increased risk of all-cause hospitalization. For AF patients comorbid with HF, there was no difference in all-cause hospitalizations by left ventricular ejection fraction classifications (see Additional file 1: Table 1). Furthermore, non-cardiovascular comorbidities including DM (HR 1.14, 95% CI 1.08–1.20), COPD (HR 1.28, 95% CI 1.02–1.62), gastrointestinal disorder (HR 1.37, 95% CI 1.21–1.55), hyperthyroidism/hypothyroidism (HR 1.25, 95% CI 1.13–1.39), and renal dysfunction (HR 1.24, 95% CI 1.09–1.42) were associated with a higher all-cause hospitalization rate during follow-up. Of note, the use of ACEIs/ARB and history of RFCA were identified as protective factors of all-cause hospitalization. The risk factors of all-cause death in AF patients are shown in Additional file 2: Table 2.

We found that the major predictors of AF hospitalization were severe symptoms (EHRA II, HR 1.32, 95% CI 1.11–1.58, P < 0.01; EHRA III, HR 1.32, 95% CI 1.08–1.61, P < 0.01) and antiarrhythmic drugs (HR 1.38, 95% CI 1.26–1.50, P < 0.01) (Table 3). For other CVD hospitalization, the predictor was complicating CVD (Table 4). Additionally, we observed that RFAC was associated with a decrease of AF-related hospitalization (11%) (HR 0.89, 95% CI 0.81–0.99, P = 0.04) and subsequent CVD hospitalization (14%) (HR 0.86, 95% CI 0.77–0.96, P < 0.01).