This observational, multi-center, prospective study was conducted at 30 Brazilian research sites without a specific visit schedule. Data were collected from medical records at four time points over a one-year follow-up period, including patients who discontinued edoxaban. Detailed study design and data collection methods were previously described.¹¹

The study recruited 713 patients (Figure 1) from September 2019 to February 2022 from various care settings. Patients were included if they were assigned to edoxaban for nonvalvular atrial fibrillation (NVAF), were over 18 years old, had been on edoxaban for 14 to 90 days before enrollment, and provided written informed consent.

Figure 1

Primary safety outcomes focused on bleeding events, defined by the International Society of Thrombosis and Haemostasis (ISHT). Major bleeding included fatal bleeding or symptomatic bleeding in critical areas (e.g., retroperitoneal, intracranial). Clinically relevant non-major bleeding (CRNMB) was classified as the ones that required medical intervention, hospitalization, or a face-to-face evaluation.

Secondary outcomes for effectiveness included cardiovascular events like strokes (ischemic and hemorrhagic), systemic embolic events (SEE), transient ischemic attack (TIA), venous thromboembolism (VTE), acute coronary syndrome (ACS), and MACE (major adverse cardiovascular events) – e a composite of non-fatal myocardial infarction (MI), non-fatal stroke, non-fatal SEE, and death due to cardiovascular causes or bleeding. All safety measures and adverse events from enrollment to final follow-up were recorded.

Descriptive analysis of all data collected was conducted, and results were shown as absolute frequencies and percentages, excluding missing data. All descriptive statistics were based on the available information, missing and unknown data were not considered. Continuous variables were described as means and standard deviations (SD) or medians and interquartile ranges, as appropriate. Incidence proportions (IPs) were calculated by dividing the number of patients with an event by the total number of patients (n = 590), and expressed as percentages, with a 95% confidence interval (CI). Kaplan-Meier curves were used for time-to-event analysis. The analysis followed STROBE and CONSORT guidelines.

The study included 705 patients with AF assigned to edoxaban, with 590 considered for full analysis. The mean follow-up was 11.3 months. Patients were mainly from northeast Brazil (42.0%), followed by south (29.4%), southeast (28.2%), and mid-west (0.3%).

At baseline (edoxaban initiation), the median age was 70 years, with a male predominance (60%). Octogenarians represented 18% of the study population with a mean age of 84.96 (SD: 3.95 years).

The most common races/ethnicities were Caucasian (38.06%), Hispanic/Latino (30.97%), and Black or Afro-descendent (28.73%). Most patients (64%) were treated through the public healthcare system (Table 1).

Table 1

Variables	n=590
Demographic features
Gender, N (%)
Valid numbers	590
Male	354 (60%)
Female	236 (40%)
Age, years
Valid number	590
mean ± SD	68.9 ± 12.6
median	70
Ethnicity, N (%)
Valid number	536
Caucasian	204 (38.06%)
Hispanic/Latino	166 (30.97%)
Black or Afro-descendent	154 (28.73%)
Asian	8 (1.49%)
Native	4 (0.75%)
Healthcare Insurance, N (%)
Valid number	508
Public	324 (64%)
Private	184 (36%)
Clinical Features
Body weight, Kg
Valid number	507
mean ± SD	77.48 ± 17.91
BMI (Kg/m2), N (%)
Valid n	496
< 18.5	10 (2%)
18.5 - 24.9	140 (28%)
25 - 29.9	190 (38%)
> 30	156 (32%)
AF type, N (%)
Valid n	543
Paroxysmal	215 (40%)
Persistent	93 (17%)
Long standing persistent	14 (2%)
Permanent	221 (41%)
Signs and symptoms, N of patients (%)
Valid n	525
Asymptomatic	382 (73%)
Symptomatic	143 (27%)
Blood pressure, mmHg
Valid n	519
Systolic, mean ± SD	126.8 ± 19.31
Diastolic, mean ± SD	76.39 ± 13.47
Serum creatinine, mg/dL
Valid n	492
mean ± SD	1.46 ± 5.43
Creatinine clearance, mL/min
Valid n	258
mean ± SD	70.5 ± 35.15
CHA 2 DS 2 -VASc, Risk score
Valid n	360
mean ± SD	3.3 ± 1.6
HAS-BLED, Risk score
Valid n	269
mean ± SD	1.8 ± 1.2
Frailty, N (%)
Valid n	435
Yes	74 (17%)
No	361 (83%)
LV ejection fraction, N (%)
Valid n	229
≥ 40%	156 (68%)
< 40%	73 (32%)
Previous history of CV disease, N (%)
Valid n	590
Hypertension	457 (78%)
Heart Failure	164 (28%)
Myocardial infarction	66 (11%)
Coronary Heart Disease	78 (13%)
Stroke	52 (9%)
Peripheral Artery Disease	18 (3%)
Transient ischemic attack	16 (2.7%)
Diabetes mellitus, N (%)
Valid n	573
Yes	172 (30%)
No	401 (70%)
Obesity, N (%)
Valid n	588
Yes	147 (25%)
No	441 (75%)
Dyslipidemia, N (%)
Valid n	581
Yes	244 (42%)
No	337 (58%)
Chronic obstructive pulmonary disease, N (%)
Valid n	583
Yes	28 (4.8%)
No	555 (95.2%)
Previous history of major bleeding, N (%)
Valid n	579
Yes	11 (1.9%)
No	568 (98.1%)

¹ Unknown and missing data were not considered for the percentage calculation. ² For Previous history of cardiovascular disease, one patient could have more than one previous cardiovascular disease; AF: atrial fibrillation; BMI: body mass index.

The mean body weight was 77.5 Kg and common comorbidities included hypertension (78%), dyslipidemia (42%), and diabetes (30%).

For almost three-quarters of patients, AF was asymptomatic. The most common symptoms reported among the symptomatic patients were palpitations (52%), dyspnea (43%), and fatigue (32%). Permanent AF was the most frequent type (41%), followed by paroxysmal AF (40%). The median CHA2DS2-VASc score was 3, and the median HAS-BLED score was 2. Overall, 17% of patients perceived to be frail.

More than half (51%) of patients received edoxaban as their first-time treatment for NVAF, and 23% received the medication considering the possibility of improved safety or efficacy. A reduced dose (30mg daily) was given to 30% of patients (Table 2).

Table 2

Variables	N = 590
Edoxaban dosage, N(%)
Valid n	590
30mg/once daily	179 (30%)
60mg/once daily	411 (70%)
Reason for prescription, N(%)
Valid n	444
First-time OAC treatment for NVAF	227 (51%)
Expected better efficacy	56 (13%)
Expected better safety	45 (10%)
High variability of former OAC treatment response	35 (8%)
Lack of efficacy of former OAC treatment	13 (3%)
Patient’s request	13 (3%)
Expected better patient compliance	8 (2%)
Drug-drug interaction in the former treatment	2 (0.5%)
Other	45 (10%)
Interruption of edoxaban, N(%)
Number of patients	117 (21%)
Number of interruptions	188 (13%)
Reason for interruption*, N of interruption (%)
Valid n	188
Adverse event	42 (22%)
Patient’s request	33 (18%)
AF intervention	12 (6%)
Healthcare system related	6 (3%)
Drug-drug interaction	1 (0.5%)
Other	94 (50%)
Restarting edoxaban after interruption, N (%)
Valid n	117
Patients that restarted and ongoing	15 (13%)
Patients that interrupted and did not restart (discontinued)	102 (87%)
Treatment duration, months
Valid n	590
Mean ± SD	10.4 ± 3.8
Median	12.25
IQR (25-75)	5.58 (7.46 - 13.04)

*Reason for discontinuation reported by a unique patient; patient could have more than one interruption; NVAF: non-valvular atrial fibrillation; AF: atrial fibrillation; OAC: oral anticoagulants; SD: standard deviation; IQR: interquartile range.

Treatment was interrupted in 21% of patients, with 188 interruptions recorded throughout the study period. Reasons included adverse events (22%), patient request (18%), AF intervention (6%), healthcare service issues (3%), drug interactions (0.5%), and other reasons (50%). Among these, the most reported were financial issues (10%, N = 18).

After discontinuation, 13% restarted edoxaban, while 87% permanently stopped (Table 2). At the one-year follow-up, 83% of patients were still on edoxaban.

Figure 2 presents the occurrence of major bleeding and CRNMB, according to the International Society on Thrombosis and Haemostasis (ISTH), observed during the one-year of follow-up. A total of 13 primary safety events were registered during the study period; nine major bleeding events were reported, of those five were gastrointestinal bleeds (IP 0.85 [95% CI =0.82; 0.88]), 1 was intramuscular with compartment syndrome (IMCS) (IP 0.17 [95% CI =0.16; 0.18]) and 3 events were classified as others (IP 0.51 [95% CI= 0.48; 0.53]). Regarding the causes of the major bleeding events, six of them were considered spontaneous (IP 1.02 [95% CI= 0.98; 1.05]) and three events had the cause reported as unknown. Among all major bleeding events, 2 (IP 0.34 [95% CI =0.32; 0.36]) of them lead to a fatal outcome.

Figure 2

Number of bleeding events (N) [IP (95%CI)]

Regarding effectiveness outcomes, Figure 3 presents the incidence of cardiovascular events among AF patients. Effectiveness assessment showed that overall, 112 (23.9%) patients had 141 cardiovascular events (IP 23.9 [95% CI 23.76;24.04]). Four events of stroke (IP 0.68 [CI 95% 0.65;0.71]) were recorded during the study period –three (IP 0.51 [CI 95% 0.48;0.53]) ischemic stroke and one was considered a stroke of unknown cause (IP 0.17 [CI 95% 0.16;0.18]) (data not shown). ACS (N=8; IP 1.36 [CI 95% 1.32;1.39]), TIA (N=1; IP 0.17 [CI 95% (0.16;0.18]) and VTE (N=1; IP 0.17 [CI 95% (0.16;0.18]) had a lower IP compared to other cardiovascular events, representing less than 2% of the patients. No patient presented SEE.

Figure 3

Congestive heart failure (N=29; IP 4.92 [CI 95% 4.84;4.99]) and MACEs (N=25; IP 4.24 [95% CI 4.17;4.3]) had the highest IP, with approximately 5% of the patients presenting at least one of the events.

Regarding the cardiovascular events related to edoxaban, one TIA (IP 0.17 95% CI [0.16;0.18]) was registered throughout the study.

Overall, 107 patients had 130 events classified as serious adverse events (SAE) (IP 22.03 [95% CI 21.9;22.17]). Among those SAEs, 58 events resulted in death (IP 9.83 [95% CI 9.73;9.93]), however, most of the events (N= 75) resulted in recovery (IP 12.71 [95% CI 12.6;12.82]).

Considering the death occurred during the study, two of them were related to major bleeding events and one was related to TIA (data not shown). Despite the limited number of events observed among study participants, which makes precise time-to-event estimation challenging, Figure 4 presents the Kaplan-Meier analyses of the time (in months) from edoxaban initiation to the occurrence of any cardiovascular events (A). Additionally, it illustrates the Kaplan-Meier estimation from baseline to death related to cardio-vascular events (B).

Figure 4