This is a single-center (conducted in a university hospital), prospective, observational, and longitudinal study with 421 participants (patients) with AF, aged ≥ 18 years, included after granting free and informed consent (FIC), in the period ranging from May 2018 to March 2020.^12,13 All patients underwent clinical and laboratory evaluation (including electrocardiogram, transthoracic echocardiogram, and clinical pathology tests) and clinical follow-up. The AF rate was documented using an electrocardiogram. Three groups were formed according to the EHRA categorization type 1, type 2, and group 3 of participants without primary valvular heart disease or mechanical or biological valve prostheses. Upon enrollment into the study, risk scores for embolism (for groups 2 and 3) and bleeding (for all groups) were calculated according to guidelines,^1,14 as well as renal function. For the severity of symptoms related to AF, the CCS ¹⁵ and EHRA¹⁶ scales were employed. The scores used to estimate the occurrence of embolism were CHA₂DS₂-VASc^1,16 and CHA₂DS₂-VASc-RAF. To estimate bleeding^1,14 the HAS-BLED and ATRIA scores were used.

For patients taking warfarin, the quality of individual OAC was based on the percentage of time during which the International Normalized Ratio (INR) was maintained in the therapeutic range during the 6-month follow-up period. Clinically relevant and non-relevant bleeding resulting from OAC,¹⁷ occurrences of systemic embolism, cardiac death, total mortality, and events (defined as the combination of all, including hospitalization) were considered during the median clinical follow-up of 12 months.

The study was approved by the institution's Research Ethics Committee and all participants signed the informed consent form.

Statistical analysis was performed using the SPSS software for Windows, version 16.0 (SPSS, Inc., Chicago, Illinois). Qualitative variables were presented using frequencies, and the continuous variables using mean and standard deviation (for variables with normal distribution), or median and interquartile ranges (for variables without normal distribution), according to the Kolmogorov-Smirnov test with Lilliefors correction. Comparison between groups was made using the ANOVA test (One-Way) or Student's T-test (unpaired) for continuous variables with normal distribution, or the Kruskal-Wallis test for variables with non-Gaussian distribution. The Tukey test (post hoc) and the Mann-Whitney test (with Bonferroni correction) were used to determine the differences between means and medians, respectively. The chi-square test was used for qualitative variables. The Kaplan-Meier curve was applied to compare the three groups regarding the occurrence of events (cardiovascular, embolism, bleeding) separately and combined and the Log-rank test was used. P-value < 0.05 was the statistical significance criterion.

The study included 421 patients, aged 58.6 ± 14.9 years (between 18 and 92 years) on average, 227 of whom were women. AF presentation at the time of entry into the study was permanent in 244, persistent in 66, and paroxysmal in 111. The median Left Ventricular Ejection Fraction (LVEF) was 0.56 (interquartile range – Q1-Q3: 0.37 – 0.65) and the mean Left Atrial (LA) diameter was 50.8 ± 9.4 mm (ranging from 30 to 84).

According to the group's categorization, 113 patients were of the EHRA type 1, 53 of type 2 and group 3 consisted of 255 patients. Among the EHRA type 1 group, 15 patients had mechanical prostheses; 13 in the mitral position and 2 in the aortic position. The others had moderate to severe MS of rheumatic etiology. The primary etiologies of heart disease in group 3 included hypertension (39.6%), dilated cardiomyopathy of various causes, predominantly ischemic (45.5%), and other etiologies not associated with systolic dysfunction, such as coronary artery disease, congenital heart disease, and brady-tachycardia syndrome. Regarding the use of OAC, 258 patients were taking warfarin, 6 were taking rivaroxaban, 2 were taking apixaban, 2 were taking edoxaban, and 1 was taking dabigatran.

Table 1 shows the comparison between groups regarding clinical variables, laboratory data, and scores for embolism and bleeding at the beginning of the study. In the EHRA type 1 group, patients were younger, with a higher proportion of women and a greater percentage using OAC. Tukey test p-values for age were 0.019 between groups 1 and 2, < 0.0001 between groups 1 and 3, and 0.028 between groups 2 and 3. LA diameter comparisons revealed differences of 0.20 between groups 1 and 2, < 0.001 between groups 1 and 3, and 0.005 between groups 2 and 3, with group 3 exhibiting smaller diameters.

Table 1

Variables	EHRA type 1	EHRA type 2	Group 3	P-value
Age (mean ± SD) in years	50.6 ± 13.0	56.8 ± 13.3	62.5 ± 14.6	< 0.0001
Proportion of females	77.8	58.4	42.3	< 0.0001
Permanent AF (%)	68.1	62.2	52.5	0.004
Basal HR (bpm) Q2	80.0	78.0	76.5	0.26
Basal BP (mmHg) Q2	114/70	113/71	120/74	0.09/0.36
Average LA diameter (mm)	55.3	52.5	48.3	< 0.0001
LV systolic diameter (mm) Q2 [Q1-Q3]	32.0 [28-39]	38 [31-48]	43 [33-55]	< 0.0001
LVEF Q2 [Q1-Q3]	0.62 [0.54-0.66]	0.61 [0.41-0.66]	0.48 [0.31-0.63]	< 0.0001
OAC usage (%)	87.6	62.3	52.2	< 0.0001
Patients with PM (%)	3.5	5.6	18.8	< 0.0001
CHA2DS2-VASc	-	2.3	2.9	< 0.017
GFR (mL/min/1.73 m2)	96.4	88.3	76.5	< 0.0001
CHA2DS2-VAScRAF	-	9.9	10.1	0.85
CCS Symptom Scale	2.6	2.7	2.7	0.39
EHRA Symptom Scale	2.7	2.7	2.80	0.73
OAC Quality (%)	47.5	40.9	52.7	0.08
HAS-BLED Score	1.4	1.6	1.7	0.06
ATRIA Score	1.4	1.1	2.0	0.001

SD: standard deviation; AF: atrial fibrillation; HR: heart rate; BP: blood pressure; LA: left atrium; LV: left ventricle; EF: ejection fraction; Q2: median; Q1-Q3: interquartile range; OAC: oral anticoagulant; PM: cardiac pacemaker; GFR: MDRD (Diet Modification in Kidney Disease) glomerular filtration rate; CCS: Canadian Cardiovascular Society; EHRA; Evaluated Heart valves, Rheumatic or Artificial; LVEF: Left Ventricular Ejection Fraction.

Group 3 had a larger left ventricular systolic diameter compared to group 1, and lower LVEF compared to groups 1 and 2. Mann-Whitney test p-values for left ventricular systolic diameter were 0.004 between groups 1 and 2, < 0.001 between groups 1 and 3, and 0.09 between groups 2 and 3. Regarding LVEF, the Mann-Whitney test p-values were 0.34 between groups 1 and 2, < 0.001 between groups 1 and 3, and 0.003 between groups 2 and 3. Group 3 showed a higher CHA₂DS₂-VASc score and ATRIA score (compared to other groups) and a lower glomerular filtration rate in relation to group 1. The p-values referring to the Tukey test regarding the ATRIA score were 0.74 between groups 1 and 2, 0.015 between groups 1 and 3, and 0.011 between groups 2 and 3. Regarding the glomerular filtration rate, it was 0.46 between groups 1 and 2; 0.001 between groups 1 and 3, and 0.14 between groups 2 and 3.

There were no differences between the groups regarding baseline data on heart rate, blood pressure level, symptom scales, and OAC quality.

During the median follow-up time of 12.0 months [5.0-22.0], bleeding occurred in 76 patients. Current or previous history of systemic embolism occurred in 93 patients. The percentage of patients using OAC was 87.6% in EHRA type 1, 62.2% in type 2, and 56.0% in group 3 (< 0.0001). The quality of OAC with INR in the therapeutic range ≥ 60% was 31.8% in EHRA type 2, 22.5% in type 2, and 38.9% in group 3 (p = 0.14).

The outcome was cardiac death in 45 patients, totaling 62 deaths from all causes. A higher cardiac mortality rate was observed in patients with LVEF < 0.40 (24.4%) compared to patients with preserved LVEF (4.7%) and intermediate LVEF (9.5%), with p-value < 0.0001 by the Log-rank test.

No difference was observed between the groups regarding the occurrence of bleeding, embolism, cardiac death, or combined events. There was higher mortality from all causes in group 3 (Table 2). Kaplan-Meier survival curves based on events are shown in Figures 1 to 5. Also, Table 3 shows the accumulated survival percentages of the three groups in relation to the occurrence of general death. The estimated mean survival time for the entire population was 49.9 ± 2.5 months, 95% confidence interval of 45.0 to 54.8 months. There was no difference between sexes regarding cardiac death (p = 0.33) in the entire population.

Table 2

Variables	EHRA type 1	EHRA type 2	Group 3	P-value
Bleeding (%)	21.2	24.5	15.3	0.44
Systemic TE (%)	24.7	15.0	22.3	0.61
Cardiac death (%)	7.1	7.5	12.9	0.13
Total death (%)	8.8	9.4	18.4	0.02
Combined events (%)	30.1	41.5	44.7	0.14

P-value by log-rank test (Mantel-Cox); TE: prior thromboembolism and during follow-up; EHRA: Evaluated Heart valves, Rheumatic or Artificial.

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Table 3

Groups/ time (months)	6	12	24	30
Cumulative survival percentage (%); EHRA type 1	97.3	94.5	89.1	89.1
Cumulative survival percentage (%); EHRA type 2	97.8	89.4	89.4	83.8
Cumulative survival percentage (%) group 3	87.8	84.7	76.3	72.5

EHRA: Evaluated Heart valves, Rheumatic or Artificial

The main limitation was the population size, with a different distribution in terms of group size. There was no assessment of inappropriate use of oral anticoagulants, including data on dosing and adherence, and other approaches such as ablation, carotid filter placement, and percutaneous occlusion of the LA appendage were not evaluated. Dunn's test was not performed for multiple comparisons related to the Kruskal-Wallis test.