Relationship Between Direct Oral Anticoagulant Doses and Clinical Outcomes in Elderly Patients With Non-Valvular Atrial Fibrillation　― ANAFIE Registry Sub-Analysis ―

Masaharu Akao; Hiroshi Inoue; Takeshi Yamashita; Hirotsugu Atarashi; Takanori Ikeda; Yukihiro Koretsune; Ken Okumura; Shinya Suzuki; Hiroyuki Tsutsui; Kazunori Toyoda; Atsushi Hirayama; Masahiro Yasaka; Takenori Yamaguchi; Satoshi Teramukai; Tetsuya Kimura; Yoshiyuki Morishima; Atsushi Takita; Wataru Shimizu

doi:10.1253/circj.CJ-23-0143

Abstract

Background: This sub-analysis of the ANAFIE Registry, a prospective, observational study of >30,000 Japanese non-valvular atrial fibrillation (NVAF) patients aged ≥75 years, assessed the prevalence of direct oral anticoagulant (DOAC) under-dose prevalence, identified the factors of under-dose prescriptions, and examined the relationship between DOAC dose and clinical outcomes.

Methods and Results: Patients, divided into 5 groups by DOAC dose (standard, over-, reduced, under-, and off-label), were analyzed for background factors, cumulative incidences, and clinical outcome risk. Endpoints were stroke/systemic embolic events (SEE), major bleeding, and all-cause death during the 2-year follow-up. Of 18,497 patients taking DOACs, 20.7%, 3.8%, 51.6%, 19.6%, and 4.3%, were prescribed standard, over-, reduced, under-, and off-label doses. Factors associated with under-dose use were female sex, age ≥85 years, reduced creatinine clearance, history of major bleeding, polypharmacy, antiplatelet agents, heart failure, dementia, and no history of catheter ablation or cerebrovascular disease. After confounder adjustment, under-dose vs. standard dose was not associated with the incidence of stroke/SEE or major bleeding but was associated with a higher mortality rate. Patients receiving an off-label dose showed similar tendencies to those receiving an under-dose; that is, they showed the highest mortality rates for stroke/SEE, major bleeding, and all-cause death.

Conclusions: Inappropriate low DOAC doses (under- or off-label dose) were not associated with stroke/SEE or major bleeding but were associated with all-cause death.

Atrial fibrillation (AF) is common among elderly patients and is a well-known risk factor for stroke.¹ The use of oral anticoagulants (OACs) in patients with non-valvular AF (NVAF) is paramount for preventing stroke and its sequelae, but is associated with bleeding risk, particularly among elderly patients.²^,³ Thus, it is necessary to optimize and personalize OAC therapy.

Editorial p 1775

Direct OACs (DOACs; dabigatran, rivaroxaban, apixaban, and edoxaban) that selectively inhibit coagulation factors have been used for stroke prevention in patients with AF since 2011 and have become the predominant OAC therapy worldwide during the past 10 years.⁴ However, a recent meta-analysis of real-world use of DOACs in AF reported an overall prevalence of DOAC off-label use (both under- and over-dose) of 24%, with 20% of patients receiving under-doses of DOACs.⁵ By region, Asia had the highest prevalence of off-label use of DOACs (32%; predominantly under-dosing), compared with Europe (22%) and North America (14%).⁵ Although DOAC under-dosing is often prescribed in routine medical care primarily to reduce the risk of bleeding, previous studies have shown that DOAC underdosing is associated with poorer outcomes.⁶^–⁸ However, the current status and the association of inappropriate low doses of DOACs with the prognosis of elderly Japanese patients with NVAF are poorly understood.

The All Nippon AF in the Elderly (ANAFIE) Registry, a prospective large-scale observational study, aimed to elucidate the clinical status and prognosis of >30,000 Japanese NVAF patients aged ≥75 years.⁹^–¹¹ In the Registry, the proportion of patients receiving 4 DOACs at under-doses that did not meet the dose adjustment criteria and off-label doses ranged from 14% to 31% for each type of DOAC.¹² The objectives of this post-hoc sub-analysis of the ANAFIE Registry were 2-fold: (1) to determine the prevalence of DOAC under-dose and identify the causative factors and (2) to examine the relationship between DOAC dose and clinical outcomes during a 2-year follow-up period in Japan.

Methods

Study Design and Patients

The study design and methods of the ANAFIE Registry have been previously described.⁹ Patients were followed up for 2 years after enrollment, and treating physicians prescribed pharmacotherapy based on their judgment, clinical practice, Japanese treatment guidelines, and approved drug indications per the manufacturer labels.

The target population included Japanese men and women aged ≥75 years with NVAF, diagnosed based on ECG, who provided informed consent for study participation, and could attend study follow-up visits. The baseline characteristics of the whole population of the ANAFIE Registry have been previously described.¹⁰ The present study was one of the post-hoc sub-analyses of the ANAFIE Registry, and focused on the population receiving DOACs at enrollment.

The ethics committees of the participating centers approved the study protocol, and the ANAFIE Registry adhered to the Declaration of Helsinki and local registry and ethical guidelines for conducting clinical studies in Japan. The principal approving ethics committee was the Ethics Committees of The Cardiovascular Institute (Tokyo, Japan; reference no. 299). All patients provided informed consent before enrollment and were free to withdraw from the study at any time. The corresponding author had full access to all the data in the study and takes responsibility for its integrity and the data analysis.

Study Measures and Outcomes

The endpoints were stroke/systemic embolic events (SEE), major bleeding classified according to the International Society on Thrombosis and Haemostasis, intracranial hemorrhage (ICH), cardiovascular death, all-cause death, and net clinical outcomes (composite of stroke/SEE, major bleeding, and all-cause death) for 2 years.

Data Collection and Definitions

The data collected at baseline comprised patient demographics and medical history, including type of AF, date, and method of diagnosis, symptoms, treatment prescribed, type of OAC used, CHADS₂ and CHA₂DS₂-VASc scores for stroke risk, and HAS-BLED score for bleeding risk. During the study visits and follow-up, the data collected consisted of clinical outcomes.

For the present sub-analysis, the population treated with DOACs was divided into 5 groups based on the doses prescribed at the time of enrollment: standard dose, over-dose, reduced dose, under-dose, and off-label dose. The definitions of the DOAC dose groups¹² were as follows. A standard dose complied with the on-label standard-dose regimen: dabigatran 300 mg/day, apixaban 10 mg/day, and edoxaban 60 mg/day. In the case of rivaroxaban, the Japan-specific standard dose is 15 mg rather than 20 mg, based on pharmacokinetic modeling data indicating that the distributions of both the maximum concentration and the 0–24-h area under the curve of a 15-mg once-daily dose of rivaroxaban are comparable to those of a 20-mg once-daily dose in White patients at steady state.¹³ The non-inferiority to warfarin of the 15-mg dose was confirmed in the J-ROCKET-AF study.¹⁴ An over-dose was a standard dose of DOAC that was administered despite the patient meeting dose reduction criteria. A reduced dose was a reduced on-label dose administered to a patient meeting dose reduction criteria according to the prescribing information of each DOAC (dabigatran 220 mg/day, rivaroxaban 10 mg/day, apixaban 5 mg/day, or edoxaban 30 mg/day). An under-dose was a reduced on-label dose that was administered despite the patient meeting the standard-dose regimen criteria (and not meeting dose reduction criteria). An off-label dose was a dose lower than the reduced dose per the drug label. Edoxaban 15 mg was considered an off-label dose it had not yet been approved in Japan during the conduct of the ANAFIE Registry.

With regard to the dose reduction criteria for the DOACs currently being used in Japan, there are no specified dose reduction criteria for dabigatran, but for patients aged ≥70 years, with reduced creatinine clearance (CrCL ≥30 to <50 mL/min), major bleeding history or undergoing treatment with verapamil or other P-glycoprotein inhibitors, a reduced dose (220 mg/day) is suggested. For rivaroxaban, patients with a CrCL of ≥15 to <50 mL/min should receive a reduced dose (10 mg/day), and patients taking apixaban should receive a reduced dose (5 mg/day) if they meet 2 of the following criteria: body weight ≤60 kg, age ≥80 years, or serum creatinine ≥1.5 mg/dL. Patients with a CrCL of ≥15 to <50 mL/min, body weight ≤60 kg, or undergoing concomitant P-glycoprotein inhibitor therapy should receive a reduced dose of edoxaban (30 mg/day).¹²

Statistical Analysis

Summary statistics were used to report on background characteristics, which were compared using analysis of variance or the chi-squared test as appropriate. Univariate and multivariate logistic regression analyses were performed to identify relevant patient background characteristics among the patients prescribed under-doses. The probability of event occurrence of each clinical event for 2 years was estimated by Kaplan-Meier method. Univariate and multivariate logistic regression analyses of factors associated with DOAC under-dose vs. standard dose were performed. Incidence rates per 100 person-years with 95% confidence intervals (CI) were also estimated. Hazard ratios (HRs) and 95% CI were estimated using the Cox proportional hazards model adjusted by potential confounders with the standard dose of DOACs as a reference; the confounders were sex, age, body mass index, history of major bleeding, type of AF, systolic blood pressure, severe hepatic disease, diabetes, hyperuricemia, heart failure and/or reduced left ventricular ejection fraction, myocardial infarction, cerebrovascular disease, thromboembolic disease, active cancer, dementia, fall within 1 year, history of catheter ablation, CrCL, digestive diseases, polypharmacy, and use of antiarrhythmic drugs, antiplatelet agents, proton pump inhibitors, P-glycoprotein inhibitors, and antihyperlipidemia drugs. The chi-squared test was used to compare causes of death (cardiovascular, noncardiovascular, and unknown) by DOAC dose group. Statistical tests were 2-sided, with a significance level of 5%. The software used for statistical analysis was SAS release 9.4 (SAS Institute, Inc., Cary, NC, USA).

Results

Patient Disposition

Figure 1 shows the patient flow. The analysis set for this study included 29,818 patients from the ANAFIE Registry; 2,457 patients from the entire ANAFIE cohort were excluded because they did not receive OAC treatment. Of the 29,818 patients receiving OACs, 7,896 were excluded for receiving warfarin. A further 3,425 were excluded for other reasons. The DOAC group (n=18,497) was then divided into 5 groups according to the dose prescribed: standard dose (n=3,950 [21.4%]), over-dose (n=574 [3.1%]), reduced dose (n=9,548 [51.6%]), under-dose (n=3,630 [19.6%]), and off-label dose (n=795 [4.3%]). Thus, 23.9% of patients were receiving inappropriate low doses of DOACs (under-dose or off-label dose).

Figure 1.

Patient disposition. *Reasons for exclusion included lacking information on daily dose of direct oral anticoagulant (DOAC), the dose was prescribed without meeting any of the specified DOAC dose group definitions, and creatinine clearance <30 mL/min for dabigatran or <15 mL/min for other DOACs.

Of the 3,630 patients receiving an under-dose, 1,346 (37.1%), 1,721 (47.4%) and 563 (15.5%) were given rivaroxaban, apixaban, and edoxaban respectively. Of the 795 patients receiving off-label doses, 12 (1.5%), 357 (44.9%), 243 (30.6%) and 183 (23.0%) were given rivaroxaban, apixaban, edoxaban, and dabigatran respectively.

Patients’ Characteristics

Patients’ characteristics were summarized by dose group (Table 1), and there were significant differences in the distribution of most characteristics. Patients receiving a reduced dose were older, more often female, and had lower body weight and lower CrCL compared with those receiving a standard dose. Those receiving a reduced dose also had higher CHADS₂ and CHA₂DS₂-VASc scores, less often underwent catheter ablation, and more often had dementia and a history of a fall within 1 year than patients receiving a standard dose. Data for age, sex, body weight, and CrCL in patients receiving under-doses fell between that of patients receiving a standard dose and those receiving a reduced dose. Patients receiving off-label doses were the oldest (i.e., highest proportion of those aged ≥85 years), received catheter ablation least frequently, and had the highest frequency of dementia and a history of falls.

Table 1.

Characteristics at Baseline of Patients Receiving DOACs in the ANAFIE Registry (n=18,497) by DOAC Dose Group

Characteristics	Standard dose (n=3,950)	Over-dose (n=574)	Reduced dose (n=9,548)	Under-dose (n=3,630)	Off-label dose (n=795)	P value*
Sex, male	2,860 (72.4)	308 (53.7)	4,417 (46.3)	2,414 (66.5)	393 (49.4)	<0.001
Age, years	78.2±3.0	81.3±4.1	82.8±4.7	79.9±3.8	83.4±5.3	<0.001
≥85	170 (4.3)	110 (19.2)	3,377 (35.4)	468 (12.9)	340 (42.8)	<0.001
Body weight, kg	65.2±10.0	55.0±8.1	53.3±9.4	63.3±9.8	54.7±11.7	<0.001
<60	1,018 (25.8)	429 (74.7)	7,497 (78.5)	1,066 (29.4)	509 (64.0)	<0.001
Body mass index, kg/m²	24.9±3.5	22.5±2.8	22.4±3.2	24.7±3.5	22.7±3.7	<0.001
Creatinine clearance, mL/min	63.4±19.0	44.9±10.2	42.9±14.4	57.3±14.0	44.1±18.3	<0.001
<30	9 (0.2)	36 (6.3)	1,552 (16.3)	65 (1.8)	154 (19.4)	<0.001
≥30 to <50	524 (13.3)	404 (70.4)	5,297 (55.5)	829 (22.8)	309 (38.9)
≥50	3,219 (81.5)	103 (17.9)	2,193 (23.0)	2,633 (72.5)	227 (28.6)
CHADS₂ score	2.8±1.2	3.0±1.2	2.9±1.2	2.8±1.2	2.8±1.2	<0.001
CHA₂DS₂-VASc score	4.2±1.4	4.6±1.4	4.6±1.4	4.3±1.3	4.5±1.4	<0.001
HAS-BLED score	1.8±0.8	1.8±0.9	1.8±0.9	1.8±0.8	1.8±0.8	0.862
History of major bleeding	154 (3.9)	20 (3.5)	362 (3.8)	189 (5.2)	40 (5.0)	0.003
AF type
Paroxysmal	1,732 (43.8)	256 (44.6)	4,210 (44.1)	1,542 (42.5)	316 (39.7)	0.095
Persistent/long-standing persistent/permanent	2,218 (56.2)	318 (55.4)	5,338 (55.9)	2,088 (57.5)	479 (60.2)
Polypharmacy (≥5 medications)	2,565 (64.9)	427 (74.4)	7,106 (74.4)	2,550 (70.2)	577 (72.6)	<0.001
Antplatelet agents	584 (14.8)	79 (13.8)	1,386 (14.5)	610 (16.8)	118 (14.8)	0.025
Catheter ablation	546 (13.8)	56 (9.8)	770 (8.1)	343 (9.4)	50 (6.3)	<0.001
Hypertension	2,987 (75.6)	443 (77.2)	7,230 (75.7)	2,772 (76.4)	583 (73.3)	0.413
Diabetes mellitus	1,196 (30.3)	170 (29.6)	2,405 (25.2)	1,071 (29.5)	184 (23.1)	<0.001
Myocardial infarction	205 (5.2)	35 (6.1)	469 (4.9)	187 (5.2)	44 (5.5)	0.687
Heart failure	1,109 (28.1)	213 (37.1)	3,982 (41.7)	1,230 (33.9)	313 (39.4)	<0.001
History of cerebrovascular disease	934 (23.6)	159 (27.7)	2,230 (23.4)	760 (20.9)	168 (21.1)	<0.001
Gastrointestinal diseases	1,084 (27.4)	177 (30.8)	2,921 (30.6)	1,038 (28.6)	208 (26.2)	<0.001
Active cancer	454 (11.5)	67 (11.7)	1,053 (11.0)	444 (12.2)	99 (12.5)	0.321
Dementia	155 (3.9)	52 (9.1)	949 (9.9)	203 (5.6)	96 (12.1)	<0.001
Fall within 1 year	186 (4.7)	47 (8.2)	781 (8.2)	212 (5.8)	83 (10.4)	<0.001

Data are n (%) or mean±standard deviation. *Comparison among the 5 groups. AF, atrial fibrillation; DOAC, direct oral anticoagulant.

Table 2 shows the patients’ characteristics associated with under-doses of DOAC in patients meeting the standard-dose criteria for each DOAC prescribed. The significant factors associated with DOAC under-dosing in the multivariate analyses were female sex, age ≥85 years, reduced CrCL (<30 and ≥30 to <50 mL/min), a history of major bleeding, polypharmacy (≥5 medications), antiplatelet agent use, heart failure or LV systolic dysfunction, and dementia. In contrast, patients with a history of catheter ablation, and a history of cerebrovascular disease had a higher likelihood of receiving a standard DOAC dose than those receiving an under-dose.

Table 2.

Logistic Regression Analysis of Factors Associated With DOAC Under-Dose vs. Standard Dose

	Under-dose	Standard dose	Univariate		Multivariate
	Under-dose	Standard dose	OR (95% CI)	P value	OR (95% CI)	P value
Overall	3,630 (47.9)	3,950 (52.1)	–	–	–	–
Sex
Male	2,414 (45.8)	2,860 (54.2)	–	–	–	–
Female	1,216 (52.7)	1,090 (47.3)	1.32 (1.20–1.46)	<0.001	1.30 (1.17–1.46)	<0.001
Age, years
<85	3,162 (45.5)	3,780 (54.5)	–	–	–	–
≥85	468 (73.4)	170 (26.6)	3.29 (2.74–3.95)	<0.001	2.94 (2.44–3.54)	<0.001
Body weight, kg
<60	1,066 (51.2)	1,018 (48.8)	–	–	–	–
≥60	2,513 (46.9)	2,840 (53.1)	0.85 (0.76–0.94)	0.001	0.98 (0.87–1.10)	0.693
Creatinine clearance, mL/min
<30 or received dialysis	81 (83.5)	16 (16.5)	6.19 (3.61–10.61)	<0.001	5.50 (3.19–9.49)	<0.001
≥30 to <50	816 (61.1)	519 (38.9)	1.92 (1.70–2.17)	<0.001	1.71 (1.51–1.95)	<0.001
≥50	2,632 (45.0)	3,218 (55.0)	–	–	–	–
History of major bleeding
Yes	189 (55.1)	154 (44.9)	1.35 (1.09–1.68)	0.006	1.48 (1.17–1.85)	<0.001
No	3,441 (47.5)	3,796 (52.5)	–	–	–	–
AF type
Paroxysmal	1,542 (47.1)	1,732 (52.9)	–	–	–	–
Persistent/long-standing (persistent/permanent)	2,088 (48.5)	2,218 (51.5)	1.06 (0.97–1.16)	0.230	1.00 (0.90–1.10)	0.921
Polypharmacy (≥5 medications)
Yes	2,550 (49.9)	2,565 (50.1)	1.28 (1.16–1.41)	<0.001	1.14 (1.02–1.27)	0.020
No	1,008 (43.8)	1,293 (56.2)	–	–	–	–
Antiplatelet agents
Yes	610 (51.1)	584 (48.9)	1.16 (1.03–1.32)	0.016	1.15 (1.01–1.32)	0.038
No	3,020 (47.3)	3,366 (52.7)	–	–	–	–
Catheter ablation
Yes	343 (38.6)	546 (61.4)	0.65 (0.56–0.75)	<0.001	0.67 (0.58–0.78)	<0.001
No	3,287 (49.1)	3,404 (50.9)	–	–	–	–
Hypertension
Yes	2,772 (48.1)	2,987 (51.9)	1.04 (0.94–1.16)	0.449	0.99 (0.89–1.11)	0.914
No	858 (47.1)	963 (52.9)	–	–	–	–
Diabetes mellitus
Yes	1,071 (47.2)	1,196 (52.8)	0.96 (0.87–1.06)	0.462	0.91 (0.82–1.02)	0.097
No	2,559 (48.2)	2,754 (51.8)	–	–	–	–
Heart failure or LV systolic dysfunction
Yes	1,251 (52.6)	1,129 (47.4)	1.31 (1.19–1.45)	<0.001	1.17 (1.06–1.30)	0.003
No	2,379 (45.8)	2,821 (54.3)	–	–	–	–
History of cerebrovascular disease
Yes	760 (44.9)	934 (55.1)	0.86 (0.77–0.95)	0.005	0.78 (0.70–0.88)	<0.001
No	2,870 (48.8)	3,016 (51.2)	–	–	–	–
Active cancer
Yes	444 (49.4)	454 (50.6)	1.07 (0.93–1.23)	0.320	1.08 (0.94–1.25)	0.273
No	3,186 (47.7)	3,496 (52.3)
Dementia
Yes	203 (56.7)	155 (43.3)	1.45 (1.17–1.80)	<0.001	1.27 (1.01–1.59)	0.040
No	3,427 (47.5)	3,795 (52.5)	–	–	–	–
Fall within 1 year
Yes	212 (53.3)	186 (46.7)	1.25 (1.02–1.54)	0.029	1.12 (0.90–1.38)	0.304
No	3,040 (47.6)	3,343 (52.4)	–	–	–	–

AF, atrial fibrillation; CI, confidence interval; DOAC, direct oral anticoagulant; LV, left ventricular; OR, odds ratio.

Study Outcomes

Figure 2 shows the probability of clinical outcomes by DOAC dose at 2 years of follow-up. Compared with the standard-dose group, patients receiving a reduced dose showed higher incidences of stroke/SEE, major bleeding, cardiovascular death, all-cause death, and the net clinical outcome. Patients receiving an under-dose showed comparable incidences of stroke/SEE and major bleeding, but all-cause death and the net clinical outcome were higher in this group. Patients receiving an off-label dose showed a higher incidence of stroke/SEE, lower incidence of major bleeding, and the highest all-cause death and net clinical outcome. Significant differences were observed among the 5 dose groups in the cumulative incidences of all-cause death, cardiovascular death, and net clinical outcome, but not in those of stroke/SEE, major bleeding, and ICH. Table 3 summarizes the incidence rates per 100 person-years of clinical outcomes by DOAC dose.

Figure 2.

Incidences of clinical outcomes by DOAC dose. CV, cardiovascular; DOAC, direct oral anticoagulant; ICH, intracranial hemorrhage; SEE, systemic embolic event.

Table 3.

Incidence Rate of Clinical Outcomes by DOAC Dose

Event	By DOAC dose at the time of enrollment	No. of cases	Cases of occurrence	Incidence per 100 person-years	95% CI
Event	By DOAC dose at the time of enrollment	No. of cases	Cases of occurrence	Incidence per 100 person-years	Lower limit	Upper limit
Stroke/systemic embolism	Total	18,497	488	1.41	1.29	1.54
	Standard dose	3,950	94	1.25	1.00	1.50
	Over-dose	574	12	1.12	0.49	1.76
	Reduced dose	9,548	278	1.58	1.39	1.76
	Under-dose	3,630	80	1.17	0.91	1.42
	Off-label dose	795	24	1.66	1.00	2.33
Major bleeding	Total	18,497	335	0.97	0.86	1.07
	Standard dose	3,950	71	0.94	0.72	1.16
	Over-dose	574	8	0.74	0.23	1.26
	Reduced dose	9,548	191	1.08	0.93	1.23
	Under-dose	3,630	54	0.79	0.58	1.00
	Off-label dose	795	11	0.76	0.31	1.20
Intracranial hemorrhage	Total	18,497	230	0.66	0.58	0.75
	Standard dose	3,950	51	0.68	0.49	0.86
	Over-dose	574	6	0.56	0.11	1.00
	Reduced dose	9,548	129	0.73	0.60	0.85
	Under-dose	3,630	38	0.55	0.38	0.73
	Off-label dose	795	6	0.41	0.08	0.74
Cardiovascular death	Total	18,497	309	0.89	0.79	0.99
	Standard dose	3,950	25	0.33	0.20	0.46
	Over-dose	574	6	0.56	0.11	1.00
	Reduced dose	9,548	216	1.21	1.05	1.37
	Under-dose	3,630	40	0.58	0.40	0.76
	Off-label dose	795	22	1.51	0.88	2.13
All-cause death	Total	18,497	1,093	3.14	2.95	3.32
	Standard dose	3,950	111	1.46	1.19	1.73
	Over-dose	574	28	2.59	1.63	3.55
	Reduced dose	9,548	701	3.94	3.64	4.23
	Under-dose	3,630	165	2.38	2.02	2.75
	Off-label dose	795	88	6.02	4.76	7.28
Net clinical outcome	Total	18,497	1,638	4.77	4.54	5.00
	Standard dose	3,950	225	3.01	2.61	3.40
	Over-dose	574	42	3.94	2.75	5.13
	Reduced dose	9,548	999	5.70	5.35	6.05
	Under-dose	3,630	260	3.81	3.35	4.28
	Off-label dose	795	112	7.78	6.34	9.23

CI, confidence interval; DOAC, direct oral anticoagulant.

Figure 3 shows the adjusted HR for clinical outcomes by DOAC dose. As compared with patients receiving the standard dose, patients receiving a reduced dose showed similar incidences of stroke/SEE (HR 1.14; 95% CI 0.86–1.51) and major bleeding (HR 1.11; 95% CI 0.80–1.54) while showing significantly higher all-cause death (HR 1.56; 95% CI 1.23–1.96) and net clinical outcome (HR 1.35; 95% CI 1.14–1.60). In patients receiving an under-dose, stroke/SEE was similar (HR 0.93; 95% CI 0.69–1.25), major bleeding was nonsignificantly lower (HR 0.80; 95% CI 0.56–1.15), and all-cause death was significantly higher (HR 1.41; 95% CI 1.11–1.80) than those receiving a standard dose. Patients receiving an off-label dose showed similar tendencies to those receiving an under-dose (stroke/SEE HR 1.15; 95% CI 0.71–1.84, major bleeding HR 0.71; 95% CI 0.37–1.38, all-cause death HR 2.06; 95% CI 1.52–2.79). Therefore, inappropriate low doses (i.e., under-doses and off-label doses) were not associated with stroke/SEE or major bleeding but were associated with all-cause death.

Figure 3.

Adjusted HRs for clinical outcomes by DOAC dose. Reference: standard dose; Bars: 95% confidence interval. CV, cardiovascular; DOAC, direct oral anticoagulant; HR, hazard ratio; ICH, intracranial hemorrhage; SEE, systemic embolic event.

Table 4 shows causes of death by DOAC dose. The proportion of patients with cardiovascular and noncardiovascular deaths was 31.5% and 50.3%, respectively; the cause of death was unknown in the remaining 18.2%. The proportions of these 3 categories (cardiovascular, noncardiovascular, and unknown) did not significantly differ among the 5 DOAC dose groups (P=0.138).

Table 4.

Summary of Causes of Death by DOAC Dose Group

	Total (n=1,093)	Standard dose (n=111)	Over-dose (n=28)	Reduced dose (n=701)	Under-dose (n=165)	Off-label dose (n=88)	P value*
Cardiovascular causes	344 (31.5)	30 (27.0)	7 (25.0)	238 (34.0)	46 (27.9)	23 (26.1)	0.138
Heart disease	217 (19.9)	10 (9.0)	4 (14.3)	158 (22.5)	29 (17.6)	16 (18.2)
Heart failure	119 (10.9)	7 (6.3)	2 (7.1)	82 (11.7)	17 (10.3)	11 (12.5)
Coronary artery disease	20 (1.8)	0 (0.0)	1 (3.6)	15 (2.1)	1 (0.6)	3 (3.4)
Arrhythmia	3 (0.3)	0 (0.0)	0 (0.0)	3 (0.4)	0 (0.0)	0 (0.0)
Sudden cardiac death	75 (6.9)	3 (2.7)	1 (3.6)	58 (8.3)	11 (6.7)	2 (2.3)
Vascular disease	127 (11.6)	20 (18.0)	3 (10.7)	80 (11.4)	17 (10.3)	7 (8.0)
Ischemic stroke	39 (3.6)	9 (8.1)	0 (0.0)	22 (3.1)	7 (4.2)	1 (1.1)
Systemic embolism	2 (0.2)	0 (0.0)	1 (3.6)	1 (0.1)	0 (0.0)	0 (0.0)
Intracranial hemorrhage	38 (3.5)	5 (4.5)	1 (3.6)	26 (3.7)	4 (2.4)	2 (2.3)
Extracranial hemorrhage	31 (2.8)	5 (4.5)	1 (3.6)	20 (2.9)	3 (1.8)	2 (2.3)
Others	17 (1.6)	1 (0.9)	0 (0.0)	11 (1.6)	3 (1.8)	2 (2.3)
Noncardiovascular causes	550 (50.3)	61 (55.0)	19 (67.9)	336 (47.9)	82 (49.7)	52 (59.1)
Malignant tumor	214 (19.6)	42 (37.8)	8 (28.6)	112 (16.0)	36 (21.8)	16 (18.2)
Infection	196 (17.9)	8 (7.2)	6 (21.4)	133 (19.0)	31 (18.8)	18 (20.5)
Kidney failure	13 (1.2)	1 (0.9)	0 (0.0)	10 (1.4)	1 (0.6)	1 (1.1)
Other	127 (11.6)	10 (9.0)	5 (17.9)	81 (11.6)	14 (8.5)	17 (19.3)
Unknown	199 (18.2)	20 (18.0)	2 (7.1)	127 (18.1)	37 (22.4)	13 (14.8)

Data are n (%). *The P value comparing cardiovascular/noncardiovascular/unknown categories among DOAC doses was calculated using the chi-square test. DOAC, direct oral anticoagulant.

Discussion

This post-hoc sub-analysis of the ANAFIE Registry, the largest prospective registry of elderly and very elderly Japanese patients with NVAF with over 30,000 patients analyzed,¹⁰^,¹¹ focused on patients receiving DOACs and examined the relationships between DOAC dose (standard dose, over-dose, reduced dose, under-dose or off-label dose) and clinical outcomes during a 2-year follow-up period, and identified factors leading to the prescription of under-doses in Japan.

Prevalence and Clinical Characteristics of Patients Receiving an Under-Dose

Among 18,497 Japanese patients aged ≥75 years taking DOACs, approximately half received reduced doses (51.6%). Of note, among patients who did not meet the dose reduction criteria, only half were receiving the on-label standard dose (21.4%), whereas the remaining 19.6% were receiving an under-dose. Inappropriate low doses were prescribed in approximately 24% of patients, with 19.6% of patients prescribed an under-dose and 4.3% an off-label dose. The international GARFIELD-AF registry, which also followed AF patients for 2 years, reported that 23.2% of patients were under-dosed,⁸ a comparable proportion with that in the present sub-analysis (19.6%). Another study from Israel reported that 39% of patients received off-label reduced doses of DOACs.¹⁵ In the SAKURA AF Registry from Japan, the DOAC dose was inappropriately reduced in 20–30% of users.¹⁶

The present study revealed that factors associated with the prescription of an under-dose were female sex, age ≥85 years, reduced CrCL, history of major bleeding, polypharmacy, antiplatelet agent use, no history of catheter ablation or cerebrovascular disease, heart failure, and dementia. These findings are consistent with previous reports in which DOACs were withheld or under-dosed in elderly patients with frailty, high bleeding risk, and many comorbidities.⁵^,⁸ A previous study characterized patients receiving low doses of DOACs and reported that older age, female sex, Black race, bleeding history, and heart failure were associated with receiving a lower-than-recommended dose in patients receiving dabigatran or rivaroxaban.¹⁷ Another study reported that female sex, older age, non-Caucasian ethnicity, acute coronary syndrome, diabetes mellitus, vascular disease, prior stroke, and antiplatelet agents were the most relevant predictors of under-dosing.⁸

We did not specifically analyze risk factors for stroke by dose group; however, in comparing the present risk factors leading to under-dose prescription and the independent risk factors for stroke/SEE identified in the 2-year outcomes analysis of the overall population of the ANAFIE Registry (i.e., older age ≥85 years, history of cerebrovascular disease, persistent and long-standing persistent/permanent AF, higher systolic blood pressure, higher HbA1c, other thromboembolic diseases, reduced CrCL, and history of falls within 1 year),¹¹ we found 2 overlapping factors: older age and reduced CrCL. Therefore, it is conceivable that these factors may have a strong effect on the onset of stroke, but this needs to be explored in future studies.

Clinical Outcomes of Patients Receiving an Under-Dose

In the present sub-analysis of the ANAFIE Registry, underdosing was, after adjusting for confounders, not associated with the incidence of stroke/SEE or major bleeding but was associated with higher mortality rate. Off-label dosing showed a similar tendency to that of under-dosing, and in particular, showed the highest mortality rate. Over-doses were not associated with adverse events in this sub-analysis, albeit the number of patients receiving an over-dose was low.

Several studies have reported an association between DOAC off-label dosing and patient outcomes. In the GARFIELD-AF registry, non-recommended dosing was associated with worse outcomes, namely increased risk of death (mostly cardiovascular death), compared with recommended dosing, and under-dosing was associated with a significantly reduced risk of bleeding. There was no association between non-recommended dosing and increased stroke/SEE risk.⁸ In another study from Israel evaluating the effectiveness and safety of DOAC off-label reduced dosing vs. per-label standard dosing, off-label reduced dosing was associated with reduced effectiveness (i.e., increased rates of composite outcomes).¹⁵ Similarly, 2 large retrospective cohort studies from the USA that enrolled AF patients aged ≥65 years did not find evidence of worsened thromboembolic or hemorrhagic outcomes among patients eligible for a standard DOAC dose but who received an inappropriate low dose.¹⁷^,¹⁸ An analysis of the SAKURA AF Registry in Japan found that clinical outcomes were not worse for patients receiving an under-dose than for those receiving a standard dose; however, over-dosing of DOACs was independently associated with an increased incidence of net clinical events (stroke/SEE, major bleeding, or all-cause death).¹⁹ In a recent USA study of a similar AF elderly population, under-dosing did not reduce the risk of bleeding, SEE, or all-cause death, but instead increased the risk of all-cause death among patients with AF treated with apixaban.²⁰

There are consistencies and inconsistencies between the present study’s results and those of previous reports, but the consensus may be that there is no clear benefit for using inappropriate low DOAC doses. Such doses are generally expected to reduce bleeding but did not decrease major bleeding compared with the on-label dose in this population. Nonetheless, inappropriate low DOAC doses were associated with a higher risk of all-cause death. Patients receiving inappropriate low DOAC doses are more likely to have a high-risk background, and the risk of events was high even after adjusting for various background factors. The fact that the causes of death were not different among DOAC dose groups supports this hypothesis. To improve the prescription of DOACs based on the manufacturer’s label and approved indications, it may be necessary to increase physicians’ awareness regarding appropriate DOAC dosing, especially for elderly patients with NVAF, and emphasize the risk–benefit of appropriate stroke/SEE prevention.

Study Limitations

The main limitations of the ANAFIE Registry have been described previously.¹⁰ Data on DOAC doses were collected only at the time of enrollment; thus, the present analysis does not account for changes in either DOAC or dose during follow-up. Edoxaban 15 mg was included as an off-label dose in this study, but has been recently approved to prevent stroke/SEE in NVAF patients with high bleeding risk.²¹ Comparison among different DOACs can be misleading; therefore, we decided not to conduct such comparisons in the present analysis. Finally, this analysis did not account for the competing risk of death in the time-to-event analyses of nonfatal outcomes, and an analysis of risk factors for stroke by dose group was not conducted.

Conclusions

Inappropriate low DOAC doses (under- or off-label dosing) were prevalent in the routine management of elderly Japanese patients with AF and associated with a higher risk of all-cause death. This study is based on one of the largest AF registries in Japan during the DOAC era, and because it covers the entirety of Japan rather than specific regions, it is considered to more accurately reflect the real-world management of AF patients in Japan.

Acknowledgments

The authors thank all the centers that participated in the ANAFIE Registry and all the patients who consented to participate, as well as Keyra Martinez Dunn, MD, of Edanz (www.edanz.com) for providing medical writing support, which was funded by Daiichi Sankyo in accordance with Good Publication Practice (GPP 2022) guidelines (https://www.ismpp.org/gpp-2022). In addition, the authors thank Daisuke Chiba, of Daiichi Sankyo Co., Ltd., for supporting preparation of the manuscript.

Sources of Funding

This research was supported by Daiichi Sankyo.

Disclosures

M.A. received research funding from Bayer and Daiichi Sankyo, and remuneration from Bristol-Myers Squibb, Nippon Boehringer Ingelheim, Bayer, and Daiichi Sankyo. H.I. received remuneration from Daiichi Sankyo and Nippon Boehringer Ingelheim, and consultancy fee from Daiichi Sankyo. T. Yamashita received research funding from Bristol-Myers Squibb, Bayer, and Daiichi Sankyo, manuscript fees from Daiichi Sankyo and Bristol-Myers Squibb, and remuneration from Daiichi Sankyo, Bayer, Pfizer Japan, and Bristol-Myers Squibb. H.A. received remuneration from Daiichi Sankyo. T.I. received research funding from Daiichi Sankyo and Bayer, and remuneration from Daiichi Sankyo, Bayer, and Pfizer Japan. Y.K. received remuneration from Daiichi Sankyo, Bristol-Myers Squibb, and Nippon Boehringer Ingelheim. K.O. received remuneration from Nippon Boehringer Ingelheim, Daiichi Sankyo, Johnson & Johnson, and Medtronic. S.S. received research funding from Daiichi Sankyo, and remuneration from Bristol-Myers Squibb and Daiichi Sankyo. H.T. received research funding from Daiichi Sankyo and Nippon Boehringer Ingelheim, remuneration from Daiichi Sankyo, Bayer, Nippon Boehringer Ingelheim, and Pfizer Japan, scholarship funding from Daiichi Sankyo, and consultancy fees from Pfizer Japan, Bayer, and Nippon Boehringer Ingelheim. K.T. received remuneration from Daiichi Sankyo, Bayer, Bristol-Myers Squibb, Otsuka, Novartis, and Abbott Medical. A.H. participated in a course endowed by Boston Scientific Japan, received research funding from Daiichi Sankyo and Bayer, and remuneration from Bayer, Daiichi Sankyo, Bristol-Myers Squibb, and Nippon Boehringer Ingelheim. M.Y. received research funding from Nippon Boehringer Ingelheim, and remuneration from Nippon Boehringer Ingelheim, Daiichi Sankyo, Bayer, Bristol-Myers Squibb, and Pfizer Japan. T. Yamaguchi acted as an advisory board member for Daiichi Sankyo and received remuneration from Daiichi Sankyo and Bristol-Myers Squibb. S.T. received research funding from Nippon Boehringer Ingelheim and remuneration from Daiichi Sankyo, Sanofi, Takeda, Chugai Pharmaceutical, Solasia Pharma, Bayer, Sysmex, Nipro, NapaJen Pharma, Gunze, Kaneka, Kringle Pharma and Atworking. T.K., Y.M., and A.T. are employees of Daiichi Sankyo. W.S. received research funding from Daiichi Sankyo, and Nippon Boehringer Ingelheim, and remuneration from Daiichi Sankyo, Pfizer Japan, Bristol-Myers Squibb, Bayer, and Nippon Boehringer Ingelheim.

T. Yamashita, H.T., W.S., T.I., and A.H. are members of Circulation Journal’s Editorial Team.

IRB Information

Ethical approval was obtained from all relevant institutional review boards, and all patients provided written informed consent and were free to withdraw from the registry at any time. The name of the principal ethics committee was the Ethics Committees of The Cardiovascular Institute (Tokyo, Japan), and the approval number was 299.

Data Availability

The study protocol will be made available. The deidentified participant data used in this study will be shared with researchers who participated in the study and provide a methodologically sound proposal for 36 months after article publication. The proposal may be reviewed by a committee led by Daiichi Sankyo. For any purpose, requests must be in writing and should be sent to yamt-tky@umin.ac.jp. To gain access, those requesting the data will need to sign a data access agreement.

Author Contributions

M.A., H.I., T. Yamashita, H.A., T.I., K.O., Y.K., S.S., H.T., K.T., A.H., M.Y., T. Yamaguchi, and W.S. designed and conducted the study. M.A. and W.S. interpreted the data analysis; S.T. carried out statistical analyses. M.A., T.K., Y.M., A.T., and W.S. wrote and reviewed the manuscript. All authors revised and commented on the manuscript and approved the final version.

References

Corresponding author

Register with J-STAGE for free!