Skip main navigation

Comparing Efficacy and Safety Between Patients With Atrial Fibrillation Taking Direct Oral Anticoagulants or Warfarin After Direct Oral Anticoagulant Failure

Originally publishedhttps://doi.org/10.1161/JAHA.123.029979Journal of the American Heart Association. 2023;12:e029979

Abstract

Background

An increased risk of recurrent stroke is noted in patients with atrial fibrillation despite direct oral anticoagulant (DOAC) use. We investigated the efficacy and safety of treatment with each of 4 different DOACs or warfarin after DOAC failure.

Methods and Results

We retrospectively analyzed patients with atrial fibrillation with ischemic stroke despite DOAC treatment between January 2002 and December 2016. The different outcomes of patients with DOAC failure were compared, including recurrent ischemic stroke, major cardiovascular events, intracranial hemorrhage and subarachnoid hemorrhage, mortality, and net composite outcomes according to switching to different DOACs or vitamin K antagonist after index ischemic stroke. We identified 3759 patients with DOAC failure. A total of 84 patients experienced recurrent ischemic stroke after switching to different oral anticoagulants, with a total follow‐up time of 14 years. Using the vitamin K antagonist group as a reference, switching to any of the 4 DOACs was associated with a 69% to 77% reduced risk of major cardiovascular events (adjusted hazard ratio [aHR], 0.25 [95% CI, 0.16–0.39] for apixaban, 0.23 [95% CI, 0.14–0.37] for dabigatran, 0.23 [95% CI, 0.09–0.60] for edoxaban, and 0.31 [95% CI, 0.21–0.45] for rivaroxaban), and a 69% to 83% reduced risk of net composite outcomes (aHR, 0.25 [95% CI, 0.18–0.35] for apixaban, 0.17 [95% CI, 0.11–0.25] for dabigatran, 0.31 [95% CI, 0.17–0.56] for edoxaban, and 0.31 [95% CI, 0.23–0.41] for rivaroxaban).

Conclusions

In Asian patients with DOAC failure, continuing DOACs after index stroke was associated with fewer undesirable outcomes than switching to a vitamin K antagonist. Alternative pharmacologic and nonpharmacologic strategies warrant investigation.

Nonstandard Abbreviations and Acronyms

DOAC

direct oral anticoagulant

IS

ischemic stroke

ICH

intracranial hemorrhage

IPTW

inverse probability of treatment weighting

MACE

major cardiovascular event

NHIRD

National Health Insurance Research Database

OAC

oral anticoagulant

VKA

vitamin K antagonist

Clinical Perspective

What Is New?

  • After direct oral anticoagulant (DOAC) failure, switching to another DOAC was associated with fewer undesirable outcomes than switching to a vitamin K antagonist.

What Are the Clinical Implications?

  • A relatively high residual risk of recurrent stroke was noted in patients with atrial fibrillation with DOAC failure, and further treatment algorithms may be developed.

  • Until trials specifically addressing DOAC failure are available, continuing with another DOAC may be better than switching to a vitamin K antagonist after the index event in Asian populations.

With increased life expectancy and an aging society, the prevalence of atrial fibrillation (AF) has increased 3‐fold in the past 5 decades.1 AF is a major cause in cardioembolic stroke, which comprises 20% of ischemic strokes (ISs) and is associated with high severity and morbidity.2 The mainstay of preventive therapy for cardioembolic stroke in patients with AF is anticoagulation. In patients with AF, oral anticoagulants (OACs) can reduce the risk of IS compared with antiplatelet agents.3 In the past decade, direct oral anticoagulants (DOACs) have shown similar or better efficacy and fewer safety end points than vitamin K antagonists (VKAs); therefore, the current updated guideline recommends DOACs as alternatives to VKA for patients with AF with CHA2DS2‐VASc scores of ≥2 points for the prevention of thromboembolic events.4, 5, 6 However, there was an estimated risk of 1.4% per year of IS occurrence in patients with AF despite DOAC use.7 Potential treatment strategies in patients with DOAC failure may be switching to different DOACs or switching to a VKA, but recent clinical cohort studies revealed that switching to another anticoagulant does not appear to reduce the risk of recurrent stroke.8, 9, 10, 11, 12, 13 Most of the studies have focused on the included patients who had prior treatments that combined DOAC and VKA, not purely on patients with DOAC failure, except for the RENO‐EXTEND (Causes and Risk Factors of Cerebral Ischemic Events in Patients With Nonvalvular AF Treated with NOACs for Stroke Prevention‐Extend) study, which focused on this population.14 The end points in previous studies were not extended to include major cardiovascular events or net clinical risk and benefits.

Before the development of precise strategies or algorithms for DOAC failure, we hope to provide data from a nationwide database on direct comparison of the net clinical risk and benefits between switching to any of 4 different DOACs or to a VKA in these patients with DOAC failure.

Methods

Data are available from the National Health Insurance Research Database (NHIRD) published by Taiwan National Health Insurance Bureau. Due to legal restrictions imposed by the government of Taiwan in relation to the Personal Data Protection Act, data cannot be made publicly available. Requests for data can be sent as a formal proposal to the NHIRD (http://nhird.nhri.org.tw).

Study Design and Setting

This study was a population‐based cohort study that used the NHIRD released by the Health and Welfare Data Science Center in Taiwan. The NHIRD covers the full population data set in Taiwan and is a computerized database derived from the National Health Insurance program started in March 1995. The NHIRD uses claims data for reimbursement of medical care at outpatient, inpatient, and emergency departments and contains deidentified registration files. The National Health Insurance Administration performs cross‐checks for regular samples, so the diagnostic, prescription, and procedure codes in the NHIRD are widely and precisely recognized.15 This study was approved by the institutional review board of National Cheng Kung University Hospital (approval number A‐ER‐110‐097). The requirement for participant consent was waived.

Enrolled Population of DOAC Failure

We identified patients with DOAC failure by defining patients with AF who suffered index stroke despite regular DOAC use (Figure S1).

First, we identified patients >18 years of age from 2002 to 2016. The history of AF was defined as an AF diagnosis code (International Classification of Diseases, Ninth Revision and Tenth Revision (ICD‐9 and ICD‐10): ICD‐9: 427.3 and ICD‐10: I48) at 2 outpatient visits or 1 inpatient visit without a previous AF diagnosis code. DOAC failure was defined as patients with emergency department visits or hospital admission records due to index IS after receiving regular DOACs. To further confirm the diagnosis of index IS, we only enrolled patients who had brain computed tomography or magnetic resonance imaging examination while their inpatient visit code was IS (ICD‐9: 434; ICD‐10: I63).

To confirm the diagnosis of DOAC failure, the subjects meeting any of the following criteria were excluded: (1) <10 cumulative days of taking the same DOAC before the index IS and (2) combined use of both a VKA and DOAC in the prior 10 days before the index IS. We also excluded patients with AF with end‐stage renal disease, because patients with end‐stage renal disease did not fulfill the reimbursement criteria for DOAC prescriptions in Taiwan.

Definitions of the Different Treatment Groups After DOAC Failure

After DOAC failure, different treatment strategies may be chosen by clinicians, including maintaining the same DOACs adjusting the dose of the same DOAC (titrating up or tapering down the original DOAC), switching to different DOACs, switching to a VKA, and discontinuing all OACs due to specific reasons such as uncontrolled bleeding risk or other contraindications for OAC use. This study only enrolled participants with DOAC failure who were treated by switching to different DOACs or a VKA after the index IS for outcome comparison.

Based on original DOAC prescriptions, the treatment strategies of different prescriptions other than original DOACs after index stroke included switching to dabigatran, rivaroxaban, apixaban, or edoxaban. Those patients with DOAC failure who switched to a VKA after the index stroke were defined as switching to a VKA.

The timing of resuming OACs or the choice of OAC dosing after the index may be different according to various clinical scenarios. To assure the persistence of the DOAC prescription after the index IS, the timing for determination of a DOAC prescription in this study was 30 days after the DOAC‐failure IS, and a cumulative duration of DOAC usage of at least 10 days was required. We used an intention‐to‐treat model for analysis in this study.

Covariates and Outcome Measurements

The covariates included demographic data (age at the index date, sex, living area, urbanization, enrollee category, monthly income) and related comorbidities, which included diabetes (ICD‐9: 250; ICD‐10: E08‐E13), hypertension (ICD‐9: 401–405; ICD‐10: I10‐I16), hyperlipidemia (ICD‐9: 272; ICD‐10: E78.1‐E78.5), coronary heart disease (ICD‐9: 410–414; ICD‐10: I20‐I25), heart failure (ICD‐9: 428; ICD‐10: I50), peripheral arterial occlusive disease (ICD‐9: 443.9, ICD‐10: I73.9), chronic kidney disease (ICD‐9: 585, ICD‐10: N18), chronic obstructive lung disease (ICD‐9: 491, 492, 494, 496; ICD‐10: J41, J42, J43, J44, J47), liver cirrhosis (ICD‐9: 571.2, 571.5, 571.6; ICD‐10: K74), malignancy (ICD‐9: 140–209; ICD‐10: C00‐C80, C7A), mental disorder (ICD‐9: 290–319; ICD‐10: F01‐F99), and IS history.

All of the patients in the study cohort were followed up after DOAC‐failure IS until the end of 2017. The starting follow‐up point was the time point after defining the subgroup into which the patient was categorized; this corresponded to the time point 30 days after the index stroke. The shortest follow‐up period was at least 1 year. The primary outcomes included recurrent IS and major cardiovascular events (MACEs). The secondary outcomes included recurrent fatal stroke, intracranial hemorrhage (ICH)/subarachnoid hemorrhage (SAH), death, and net composite end points comprising any IS/MACE/ICH/SAH/death. The diagnoses of IS, MACE, and ICH/SAH were retrieved from the diagnosis code of admission to the hospital or emergency department; these codes included IS (ICD‐9: 434; ICD‐10: I63), MACE (ICD‐9: 410–414, 430–438; ICD‐10: I21‐24, I60‐63), and ICH/SAH (ICD‐9: 430–432; ICD‐10: I60, I61, I62). Death was coded from the cause of death registry in the NHIRD.16 The stroke diagnosis was also validated with the examination code of brain computed tomography or magnetic resonance imaging during hospitalization or emergency department visits, with a positive predictive value of 88% to 94%.17 In the above subgroups, the incidence and hazard ratios of the above primary and secondary outcomes after DOAC‐failure IS were analyzed.

Statistical Analysis

Participants' characteristics and comorbidities were presented as the means and standard deviations for continuous variables and frequencies for categorical variables. The inverse probability of treatment weighting (IPTW) based on propensity scores was applied18 to balance the sample size and adjust for potential confounders. Propensity score was calculated from multinomial logistic regression, with treatment as the dependent variable and age group, sex, living area, urbanization, enrollee category, monthly income, comorbidity (diabetes, hypertension, hyperlipidemia, coronary artery disease, heart failure, peripheral arterial occlusive disease, chronic kidney disease, chronic obstructive lung disease, liver cirrhosis, malignancy, mental disorder, IS history, CHA2DS2‐VASc score, and HAS‐BLED score) as independent variables. We conducted Cox models with the IPTW method by using propensity score to assess the outcomes. Wald χ2 tests were conducted after the IPTW method to assess categorical covariates to evaluate the differences in age, sex, living area, urbanization, enrollee category, monthly income, multiple comorbidities, CHA2DS2‐VASc score, and HAS‐BLED scores between the groups. The event rates of the primary and secondary outcomes after IPTW were analyzed, and the Cox model was used to demonstrate the hazard ratios of the primary and secondary outcomes between different DOAC and VKA groups and among the 4 DOAC‐change groups for direct comparisons. The confounding factors adopted in the Cox model included the covariates listed in the above section. For checking proportional hazard assumption, we used the cumulative Martingale residuals plots to assess the functional form of continuous variables (CHA2DS2‐VASc score and HAS‐BLED score) and standardized score process plots (transformation of Martingale residuals plot) to assess the proportional hazard assumption for all variables with the raw data. We examined all outcomes for Cox model assumption test and adjusted the data after examination. Then, a competing risk model using a Fine‐Gray subdistribution hazard model was also applied to adjust the effect of death between these subgroups. The adjustments for competing risk model were made by age group, sex, living area, urbanization, enrollee category, monthly income, comorbidities, CHA2DS2‐VASc score, and HAS‐BLED scores. All statistical procedures were performed with the statistical software package SAS for Windows (version 9.2.; SAS Institute, Cary, NC). Data are expressed as the mean±SD or as a percentage. A P value of <0.05 was considered indicative of statistical significance.

Results

Demographic Data

The final cohort identified 3759 patients suffering from an index IS despite regular DOAC treatment, including 1105 patients who maintained the original DOACs, 369 patients whose doses of original DOACs were adjusted after index stroke, 878 patients who were switched to different DOACs or to a VKA, and 1407 patients who stopped all OACs after index stroke. Among the 878 patients who switched to different DOACs or a VKA after DOAC failure, a total of 208 patients took a VKA after the index stroke and 670 patients received different DOAC prescriptions, including 212, 159, 68, and 231 patients who were administered apixaban, dabigatran, edoxaban, and rivaroxaban, respectively (Table S1). The baseline distribution and the characteristics of the study population are shown in Table 1. Before IPTW, the distribution of the mean age, sex, living area, and comorbidity status, including the presence or absence of diabetes, hypertension, hyperlipidemia, heart failure, IS history, and CHA2DS2‐VASc score of the patients were significantly different among the groups; thus, IPTW based on propensity scores was applied to adjust and balance these potential confounders.

Table 1. Demographic Data of Those Who Switched to Different Direct Oral Anticoagulants and Vitamin K Antagonist After Direct Oral Anticoagulant Failure

Before IPTWAfter IPTW
Warfarin (N=208)Apixaban (N=212)Dabigatran (N=159)Edoxaban (N=68)Rivaroxaban (N=231)P valueWarfarinApixabanDabigatranEdoxabanRivaroxabanP value
Variablen (%)n (%)n (%)n (%)n (%)%%%%%
Age group, y0.4490.413
<6024 (11.54)20 (9.43)25 (15.72)9 (13.24)26 (11.26)14.3513.979.0313.7910.32
≥60184 (88.46)192 (90.57)134 (84.28)59 (86.76)205 (88.74)85.6586.0390.9786.2189.68
Age, mean±SD77.9±10.378.4±9.874.0±9.577.8±10.377.6±10.0<0.00177.6±10.877.1±10.477.0±9.278.0±10.277.5±10.00.943
Men93 (44.71)116 (54.72)96 (60.38)36 (52.94)120 (51.95)0.05051.4853.3151.4945.8051.740.904
Region of Taiwan0.0200.997
North44.1343.9541.1043.4346.46
Central20.2817.4520.6722.7518.52
South6.609.358.009.206.12
East4.505.134.582.284.92
Other or unknown24.4924.1125.6422.3423.98
Degree of urbanization0.2700.892
1 (most)52 (25.00)59 (27.83)33 (20.75)9 (13.24)49 (21.21)22.1723.3119.0818.3621.52
2 (medium)51 (24.52)47 (22.17)35 (22.01)22 (32.35)60 (25.97)26.8024.6622.5929.4327.27
3 (least)105 (50.48)106 (50.00)91 (57.23)37 (54.41)122 (52.81)51.0352.0358.3252.2151.20
Enrollee category0.6251.000
1*12 (5.77)11 (5.19)7 (4.40)7 (10.29)10 (4.33)6.336.054.216.585.92
2445 (21.63)43 (20.28)32 (20.13)14 (20.59)37 (16.02)18.9618.3618.9220.4818.65
395 (45.67)103 (48.58)79 (49.69)28 (41.18)107 (46.32)47.2245.5048.5141.8146.39
4§56 (26.92)55 (25.94)41 (25.79)19 (27.94)77 (33.33)27.4930.1028.3631.1329.04
Monthly income0.8250.994
≤NT$ 15 84043 (20.67)37 (17.45)31 (19.50)13 (19.12)51 (22.08)18.0820.6219.2522.2520.44
NT$ 15 841– 25 00066 (31.73)69 (32.55)59 (37.11)24 (35.29)81 (35.06)33.8833.5038.2529.2933.84
≥NT$ 25 00111 (5.29)11 (5.19)8 (5.03)7 (10.29)12 (5.19)4.935.266.125.935.11
Dependent88 (42.31)95 (44.81)61 (38.36)24 (35.29)87 (37.66)43.1140.6136.3842.5240.60
Comorbidity
Diabetes83 (39.90)44 (20.75)41 (25.79)13 (19.12)73 (31.60)<0.00129.2929.3532.1235.3428.160.808
Hypertension143 (68.75)135 (63.68)105 (66.04)34 (50.00)174 (75.32)0.00282.6185.0184.0786.5181.920.859
Hyperlipidemia46 (22.12)46 (21.70)47 (29.56)11 (16.18)70 (30.30)0.03722.8526.5824.1613.7925.580.335
Coronary heart disease67 (32.21)62 (29.25)41 (25.79)12 (17.65)73 (31.60)0.14429.1125.8629.3129.0429.170.928
Heart failure87 (41.83)51 (24.06)46 (28.93)9 (13.24)72 (31.17)<0.00130.1829.2936.1627.5329.180.556
PAOD0.8032.381.791.703.361.480.886
CKD/ESRD0.0806.806.8210.102.067.230.351
COPD36 (17.31)23 (10.85)23 (14.47)5 (7.3522 (9.52)0.06011.3910.4412.2017.7612.220.659
Cirrhosis0.9151.261.821.290.971.720.978
Malignancy19 (9.13)14 (6.60)6 (3.77)3 (4.41)19 (8.23)0.2576.056.8410.325.516.960.548
Mental disorder40 (19.23)40 (18.87)30 (18.87)5 (7.35)53 (22.94)0.08218.3819.9519.1717.8818.800.994
Ischemic stroke history92 (44.23)81 (38.21)48 (30.19)19 (27.94)108 (46.75)0.00240.7938.0942.6226.6239.320.287
CHA2DS2‐VASc score5.62±1.365.16±1.334.97±1.394.85±1.535.44±1.33<0.0015.34±1.395.27±1.395.41±1.425.38±1.645.28±1.360.857
HAS‐BLED score3.13±0.923.08±0.762.98±0.832.96±0.823.02±0.740.3113.06±0.813.02±0.793.12±0.912.97±0.833.04±0.770.700
Antiplatelet121 (58.17)54 (25.47)48 (30.19)20 (29.41)68 (29.44)<0.001

*Enrollee category 1 includes full‐time or regular personnel in public shocks and governmental agencies, including civil servants.

Enrollee category 2 includes full‐time or regular employees of privately owned enterprises.

Enrollee category 3 includes other employees, members of farmers' or fishers' associations, and self‐employed individuals.

§Enrollee category 4 includes members of low‐income individuals, substitute service draftees, and veterans.

Warfarin: switching from original direct oral anticoagulant to warfarin after index stroke; apixaban: switching from original direct oral anticoagulant to apixaban after index stroke; dabigatran: switching from original direct oral anticoagulant to dabigatran after index stroke; edoxaban: switching from original direct oral anticoagulant to edoxaban after index stroke; rivaroxaban: switching from original direct oral anticoagulant to rivaroxaban after index stroke. Cells with ellipses (…) indicate that the exact number cannot be shown because the number is too small and is not permitted to be shown because of regulation of the National Health Insurance Research Database. CAD indicates coronary heart disease; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; ESRD, end‐stage renal disease; IPTW, inverse probability of treatment weighting; NT$, New Taiwan Dollor; and PAOD, peripheral arterial occlusive disease.

During follow‐up, a total of 313 patients had identified outcomes, including recurrent IS (84, 9.6%), ICH/SAH (51, 5.8%), recurrent fatal stroke (30, 3.4%), MACEs (164, 18.7%), and death (218, 24.8%) (Table S2). Compared with those patients who switched to a VKA, those who switched from original DOACs to apixaban, dabigatran, and rivaroxaban had lower rates of recurrent IS, MACEs, recurrent fatal stroke, death, and ICH/SAH (Table 2, Figure, Table S3). Using the VKA group as a reference, those who switched to apixaban significantly reduced the risk of recurrent IS by 78% (adjusted hazard ratio [aHR], 0.22 [95% CI, 0.11–0.43]), MACEs by 75% (aHR, 0.25 [95% CI, 0.16–0.39]), mortality by 68% (aHR, 0.32 [95% CI, 0.21–0.49]), ICH/SAH by 79% (aHR, 0.21 [95% CI, 0.09–0.53]), and net composite outcomes by 75% (aHR, 0.25 [95% CI, 0.18–0.35]). Dabigatran significantly reduced the risk of IS by 69% (aHR, 0.31 [95% CI, 0.16–0.60]), MACEs by 77% (aHR, 0.23 [95% CI, 0.14–0.37]), mortality by 83% (aHR, 0.17 [95% CI, 0.09–0.29]), ICH/SAH by 71% (aHR, 0.29 [95% CI, 0.12–0.72]), and net composite outcomes by 83% (aHR, 0.17 [95% CI, 0.11–0.25]). Rivaroxaban significantly reduced the risk of recurrent IS by 73% (aHR, 0.27 [95% CI, 0.15–0.49]), MACEs by 69% (aHR, 0.31 [95% CI, 0.21–0.45]), mortality by 55% (aHR, 0.45 [95% CI, 0.32–0.64]), ICH/SAH by 64% (aHR, 0.36 [95% CI, 0.18–0.72]), and net composite outcomes by 69% (aHR, 0.31 [95% CI, 0.23–0.41]). Edoxaban significantly reduced the risk of MACEs by 77% (aHR, 0.23 [95% CI, 0.09–0.60]) and net composite outcomes by 69% (aHR, 0.31 [95% CI, 0.17–0.56]). Therefore, compared with patients switching to a VKA, switching to another type of DOAC was associated with an overall reduction in the risks of MACEs and net composite outcomes (Table 2, Figure). Table S3 also shows the competing risk ratio in which the effect of death was adjusted with respect to the outcomes, including recurrent IS, MACEs, and ICH/SAH. Figure S2 shows the Kaplan‐Meier plots of the primary and secondary outcomes.

Table 2. Cox Model of Outcome After Inverse Probability of Treatment Weighting (Change Direct Oral Anticoagulant to Vitamin K Antagonist or Another Direct Oral Anticoagulant)

OutcomeWarfarin, N=208Apixaban, N=212Dabigatran, N=159Edoxaban, N=68Rivaroxaban, N=231
Recurrent ischemic strokeReference0.22 (0.11–0.43)0.31 (0.16–0.60)*0.57 (0.21–1.56)0.27 (0.15–0.49)
MACEReference0.25 (0.16–0.39)0.23 (0.14–0.37)0.23 (0.09–0.60)*0.31 (0.21–0.45)
Recurrent fatal strokeReference0.02 (0.00–0.33)*0.02 (0.00–0.33)*0.11 (0.04–0.35)
DeathReference0.32 (0.21–0.49)0.17 (0.09–0.29)0.55 (0.25–1.20)0.45 (0.32–0.64)
ICH/SAHReference0.21 (0.09–0.53)0.29 (0.12–0.72)*0.79 (0.24–2.54)0.36 (0.18–0.72)*
Net composite outcomesReference0.25 (0.18–0.35)0.17 (0.11–0.25)0.31 (0.17–0.56)0.31 (0.23–0.41)

Values are adjusted hazard ratio (95% CI). Net composite outcomes: any ischemic stroke/MACE/ICH/SAH/death. The cell with the ellipses (…) indicates that the exact number cannot be shown because the number is too small and is not permitted to be shown because of regulation of the National Health Insurance Research Database. ICH indicates intracerebral hemorrhage; MACE, major cardiovascular event; and SAH, subarachnoid hemorrhage.

*P<0.01.

P<0.001.

Figure 1. Cox model of different outcomes for patients who switched to different DOACs compared with those who switched to a VKA after DOAC failure.

Net composite outcomes: any ischemic stroke/MACE/ICH/SAH/death. DOAC indicates direct oral anticoagulant; HR, hazard ratio; ICH, intracerebral hemorrhage; MACE, major cardiovascular event; SAH, subarachnoid hemorrhage; and VKA, vitamin K antagonist.

Table S4 shows the demographic data of those patients who switched to different DOACs after the index stroke. Before IPTW, switching to dabigatran was typically administered to patients who were younger in age, and the distribution of comorbidities of diabetes, hypertension, hyperlipidemia, heart failure, mental disorder, IS history, and CHA2DS2‐VASc score was different among the groups. In regard to the direct comparison among the 4 DOACs after DOAC failure, the occurrence of recurrent IS was increased when the patients were switched to edoxaban compared with those who were prescribed apixaban or rivaroxaban after DOAC failure (aHR, 4.13 [95% CI, 1.35–12.62] when the reference group was apixaban and aHR, 3.49 [95% CI, 1.17–10.41] when the reference group was rivaroxaban). There was no significant difference in the rate of MACEs among the 4 subgroups when using any of the 4 DOACs as the reference, but dabigatran significantly reduced the risk of death by 51% to 67% (aHR, 0.49 [95% CI, 0.25–0.97] compared with apixaban, aHR, 0.34 [95% CI, 0.12–0.94] compared with edoxaban, and aHR, 0.33 [95% CI, 0.18–0.63] compared with rivaroxaban) and the risk of net composite outcomes by 45% to 58% (aHR, 0.55 [95% CI, 0.34–0.89] compared with apixaban, aHR, 0.42 [95% CI, 0.20–0.87] compared with edoxaban, and aHR, 0.47 [95% CI, 0.30–0.74] compared with rivaroxaban) (Tables S5 and  S6‐1 to S6‐6).

Discussion

The results of this study, which was conducted in Asian patients with AF with IS despite regular DOAC treatment, showed that switching to a VKA was associated with higher risks of MACEs, and net composite outcomes than switching to any of the 4 DOACs after DOAC failure. Compared with switching to a VKA, switching to any of the DOACs was associated with a 69% to 77% reduced risk of MACEs, and a 69% to 83% reduced risk of net composite outcomes.

We noted a relatively high residual risk of recurrent stroke in patients with AF with DOAC failure in our study. In our DOAC‐change patients after DOAC failure, the total event rate of recurrent IS was 9.6%, and incidence rate of IS was 8.83% per year during a 14‐year follow‐up period. The short‐term event rate of recurrent IS was 3.15% within 3 months after the index stroke. These event rates were higher than the recurrent IS rates in previous DOAC trials (0.97% per year in Apixaban for Reduction in Stroke and Other Thromboembolic Events in Atrial Fibrillation,19 2.26% per year in the Randomized Evaluation of Long‐Term Anticoagulation Therapy,20 and 3.02% per year in the Effective Anticoagulation with Factor Xa Next Generation in Atrial Fibrillation–Thrombolysis in Myocardial Infarction 4821), but our data were comparable to the recurrent IS rates reported from previous observational cohort studies (4.6% within 3 months in a multicenter cohort study,12 4.7%/year in an individual patient data pooled analysis of 7 prospective real‐world cohort studies,8 6.9% within 3 months in the Initiation of Anticoagulation after Cardioembolic stroke study,10 and 7.2% per year in the RENO‐EXTEND study11). A previous study also demonstrated that those patients with stroke with AF and prior OAC use had a higher risk of recurrent IS than those without prior OAC use,8 indicating a relatively higher residual stroke risk for those patients who suffered from DOAC failure in the real world. Although we evaluate the stroke risk by risk scores commonly used for patients with AF, such as CHA2DS2‐VASc scores, the history of stroke/transient ischemic attack in OAC‐naïve patients, or in the patients with DOAC failure, both contributed 2 points, indicating a 2.2% annual risk of thromboembolism. However, the exact stroke recurrence rate is different and higher in patients with DOAC failure than in OAC‐naïve patients. Therefore, targeted treatment strategies are needed, but may be challenging.

Previous studies showed that the absolute rates or risk ratios of ischemic/safety end points were higher in patients who switched to a VKA than in those without DOAC regimen changes after DOAC failure.13, 14, 22 The direct drug–drug comparison of different outcomes in our study further demonstrated that patients who switched to any of the DOACs had better outcomes than those who switched to a VKA. Fewer safety end points have been well documented in the DOAC group than in the VKA group in previous large‐scale randomized control trials and meta‐analyses. The results of our study demonstrated that this effect is also consistent in patients with AF with DOAC failure. In addition, we noted that switching to VKA was associated with higher rates of recurrent IS and MACEs compared with switching to any type of the 4 DOACs. Several potential hypotheses may be considered. First, switching to a VKA instead of maintaining DOAC use may indicate that a patient did not fulfill the clinical criteria of DOAC use after index stroke; for example, the ineligibility of DOAC use may be due to their renal dysfunction during index stroke, implying that the increased risk of both ischemic and hemorrhagic stroke observed in these patients may be due to progressed renal dysfunction.23 Therefore, the increased thromboembolic risk in those patients who switch to a VKA may be potentially attributed to the patients' characteristics and is part of the consequence of there being hidden confounding factors. In addition, in these data sets, we could not confirm the amount of time that the VKA level was in the therapeutic range while the patients were taking VKAs, which is also one of the hidden confounding factors.

For patients with DOAC failure, the benefit is uncertain when switching a DOAC to another DOAC after the index stroke. Previous studies showed that the risk of recurrent IS was similar or controversial in those patients who switched to another DOAC compared with those who maintained the same DOAC after DOAC failure.12, 13, 22 This may be due to the heterogeneity of DOAC failure causes. In most current studies, it was concluded that the factors contributing to DOAC failure included patient factors (poor drug compliance or genetic variants contributing to different pharmacokinetic/pharmacodynamic data),24, 25 prescription errors (inappropriate underdose or concomitant drug–drug interaction26), other competing stroke mechanisms other than AF‐related cardioembolic risk,12, 26 or excessive cardioembolic stroke burden related to AF itself despite sufficient anticoagulation. Among these factors, 32% of the causes of IS despite OAC use were classified as insufficient anticoagulation, and 24% of them were related to competing stroke mechanisms.12 The most common competing stroke mechanism other than AF‐related cardioembolic risk is large artery atherosclerosis, followed by small vessel diseases, and other less common factors such as coagulopathy and periprocedural stroke.12 Therefore, complete investigation to better ascertain the nonembolic causes in these patients with OAC failure or DOAC failure is important. For patients with AF already receiving adequate or sufficient doses of DOACs, direct drug–drug comparison may provide useful information to help clinicians choose the next DOAC strategy. In this study, lower mortality and net composite outcomes were noted in those patients who switched to dabigatran after the index stroke. In our study population, we focused on the DOAC change to DOAC group. All of the patients in the dabigatran group were subjected to class switching (from factor Xa inhibitor to direct thrombin inhibitor), but the patients in the apixaban/rivaroxaban/edoxaban groups may have been subjected to class switching (from direct thrombin inhibitor to factor Xa inhibitor) or within‐Xa‐class switching. A class‐switching effect could be a potential explanation for our results. To date, there are no direct head‐to‐head randomized controlled trials to determine the effect of mortality reduction between different DOAC classes. One nationwide cohort study from Sweden showed reduced mortality in direct thrombin inhibitor users compared with factor Xa inhibitor users among older male patients.27 Medicare data in patients with AF showed reduced mortality in dabigatran users who were >75 years of age and whose CHADS2 score was >2 points compared with rivaroxaban users.28 Less frequent vascular death was also noted in dabigatran 110 mg twice daily users than in rivaroxaban users in indirect comparison analysis.29 Therefore, possible effects of mortality reduction from a class‐switching effect in patients with DOAC failure merit further study. In addition, dabigatran use is less suitable in patients with worse renal function. Although the frequencies of chronic kidney disease in the 4 DOAC groups were similar after IPTW, the true glomerular rate of enrolled patients was unavailable in this real‐world study. It is uncertain whether patients switching to dabigatran may have had relatively better renal function, and this could be another possible source of confounding. Of note, patients who received dabigatran after DOAC failure should be free from nasogastric tube feeding; this suggests that dabigatran users may be in better general condition after index stroke than users of other DOACs, which represents another possible cause of better outcomes in the dabigatran group in our study.

A higher odds ratio of developing recurrent IS was noted in patients who switched to edoxaban after DOAC failure than in patients who switched to apixaban or rivaroxaban. A cautious interpretation is warranted on this result, because a potential limitation was that only 7% of the patients switched their DOACs to edoxaban after the index stroke. This small number may lead to unstable statistics with a wide range of 95% CIs.

Except for pharmacologic therapy, nonpharmacologic treatment may be considered in this high‐risk population after DOAC failure. The LAAO III (Left Atrial Appendage Occlusion Study III)30 showed a significant 33% reduction in the rate of stroke and systemic embolization in patients with AF who received left atrial appendage occlusion compared with the no‐occlusion group. The Carotid Artery Implant for Trapping Upstream Emboli for Preventing Stroke in Atrial Fibrillation Patients trial31 also showed the feasibility and safety of permanent carotid filter in patients with AF with high stroke risk. Future studies on these nonpharmacologic therapies in patients with DOAC failure warrant further investigation to test the clinical benefit in this high‐risk population.

Other limitations exist. First, the study was an observational cohort study, reflecting real‐world conditions, and it was not designed as a randomized study. The risk of type I and type II errors may occur and potentially relegate the findings, thereby hindering hypothesis generation. Cautious interpretation and generalization of the study results are needed. Second, the database could not provide details about real stroke severity, real physical condition (such as the fragility status and body weight), and absolute liver and renal functionality. These factors may interfere with the physicians' decision to change to different DOACs, and hidden confounding factors may exist due to the unknown off‐label low‐dose use of the 4 DOACs in the patients after the index stroke. In addition, there were no detailed data on the cause of the index stroke, precluding further insight into its pathogenesis of competing mechanisms. We also lack detailed information on brain conditions, such as the infarct volume, the volume of hemorrhagic transformation, and the small vessel disease burden; thus, we could not completely understand the underlying reasons and potential impacts of those treatment choices based on individual brain conditions. This study only reflected the different outcomes after those decisions were implemented following DOAC failure. Then, payment of idarucizumab, a reversal agent for dabigatran, from our National Health Insurance program was initiated in March 2017. Therefore, the influence of reversal agents on the safety outcomes or bleeding complications cannot be obtained in our study. However, the strength of this study also comes from its database, which covered the whole population; thus, we could perform patient selection to overcome potential confounding and enroll enough study participants for analysis.

In conclusion, in Asian patients with AF with DOAC failure, continuing another DOAC after the index stroke was associated with fewer undesirable outcomes than switching to a VKA. Future studies may be warranted to compare whether a particular DOAC has a better net clinical benefit. Novel nonpharmacologic or pharmacologic treatment options need to be tested for this high‐risk population.

Sources of Funding

The study was supported by a research grant from the National Cheng Kung University Hospital (NCKUH‐11202059, NCKUH‐11203026). The funding source was not involved in collection, analysis, interpretation of data, or in drafting the article. This research was also supported in part by the Higher Education Sprout Project, Ministry of Education to the Headquarters of University Advancement at National Cheng Kung University (NCKU) and Ministry of Science and Technology (NSTC 112‐2321‐B‐006‐016) (NSTC 112‐2627‐M‐006‐005).

Disclosures

None.

Acknowledgments

The authors are grateful to C.‐H. Hsu for providing the statistical consulting services from the Biostatistics Consulting Center, National Cheng Kung University Hospital. The authors are also grateful to the Health Data Science Center, National Cheng Kung University Hospital for providing administrative and technical support.

Footnotes

* Correspondence to: Pi‐Shan Sung, MD, PhD, Department of Neurology, National Cheng Kung University Hospital, College of Medicine, National Cheng Kung University, No. 138, Sheng Li Road, Tainan City 704, Taiwan. Email:

This article was sent to Luciano A. Sposato, MD, MBA, FRCPC, Associate Editor, for review by expert referees, editorial decision, and final disposition.

Supplemental Material is available at https://www.ahajournals.org/doi/suppl/10.1161/JAHA.123.029979

For Sources of Funding and Disclosures, see page 9.

References

  • 1 Schnabel RB, Yin X, Gona P, Larson MG, Beiser AS, McManus DD, Newton‐Cheh C, Lubitz SA, Magnani JW, Ellinor PT, et al. 50 year trends in atrial fibrillation prevalence, incidence, risk factors, and mortality in the Framingham Heart Study: a cohort study. Lancet. 2015; 386:154–162. doi: 10.1016/s0140-6736(14)61774-8CrossrefMedlineGoogle Scholar
  • 2 Lin HJ, Wolf PA, Kelly‐Hayes M, Beiser AS, Kase CS, Benjamin EJ, D'Agostino RB. Stroke severity in atrial fibrillation. The Framingham Study. Stroke. 1996; 27:1760–1764. doi: 10.1161/01.str.27.10.1760LinkGoogle Scholar
  • 3 Aguilar MI, Hart R, Pearce LA. Oral anticoagulants versus antiplatelet therapy for preventing stroke in patients with non‐valvular atrial fibrillation and no history of stroke or transient ischemic attacks. Cochrane Database Syst Rev. 2007; 18:Cd006186. doi: 10.1002/14651858.CD006186.pub2CrossrefGoogle Scholar
  • 4 Kleindorfer DO, Towfighi A, Chaturvedi S, Cockroft KM, Gutierrez J, Lombardi‐Hill D, Kamel H, Kernan WN, Kittner SJ, Leira EC, et al. 2021 Guideline for the prevention of stroke in patients with stroke and transient ischemic attack: a guideline from the American Heart Association/American Stroke Association. Stroke. 2021; 52:e364–e467. doi: 10.1161/str.0000000000000375LinkGoogle Scholar
  • 5 Cheung CC, Nattel S, Macle L, Andrade JG. Management of atrial fibrillation in 2021: an updated comparison of the current CCS/CHRS, ESC, and AHA/ACC/HRS guidelines. Can J Cardiol. 2021; 37:1607–1618. doi: 10.1016/j.cjca.2021.06.011CrossrefMedlineGoogle Scholar
  • 6 Perry M, Kemmis Betty S, Downes N, Andrews N, Mackenzie S. Atrial fibrillation: diagnosis and management‐summary of NICE guidance. BMJ. 2021; 373:n1150. doi: 10.1136/bmj.n1150CrossrefMedlineGoogle Scholar
  • 7 Ruff CT, Giugliano RP, Braunwald E, Hoffman EB, Deenadayalu N, Ezekowitz MD, Camm AJ, Weitz JI, Lewis BS, Parkhomenko A. Comparison of the efficacy and safety of new oral anticoagulants with warfarin in patients with atrial fibrillation: a meta‐analysis of randomised trials. Lancet. 2014; 383:955–962. doi: 10.1016/S0140-6736(13)62343-0CrossrefMedlineGoogle Scholar
  • 8 Seiffge DJ, De Marchis GM, Koga M, Paciaroni M, Wilson D, Cappellari M, Macha M, Kosmas TG, Ambler G, Arihiro S. Ischemic stroke despite oral anticoagulant therapy in patients with atrial fibrillation. Ann Neurol. 2020; 87:677–687. doi: 10.1002/ana.25700CrossrefMedlineGoogle Scholar
  • 9 Stretz C, Wu TY, Wilson D, Seiffge DJ, Smith EE, Gurol ME, Yaghi S. Ischaemic stroke in anticoagulated patients with atrial fibrillation. J Neurol Neurosurg Psychiatry. 2021; 92:1164–1172. doi: 10.1136/jnnp-2020-323963CrossrefMedlineGoogle Scholar
  • 10 Yaghi S, Henninger N, Giles JA, Guerrero CL, Mistry E, Liberman AL, Asad D, Liu A, Nagy M, Kaushal A. Ischaemic stroke on anticoagulation therapy and early recurrence in acute cardioembolic stroke: the IAC study. J Neurol Neurosurg Psychiatry. 2021; 92:1062–1067. doi: 10.1136/jnnp-2021-326166CrossrefMedlineGoogle Scholar
  • 11 Paciaroni M, Caso V, Agnelli G, Mosconi MG, Giustozzi M, Seiffge DJ, Engelter ST, Lyrer P, Polymeris AA, Kriemler L. Recurrent ischemic stroke and bleeding in patients with atrial fibrillation who suffered an acute stroke while on treatment with nonvitamin K antagonist oral anticoagulants: the RENO‐EXTEND study. Stroke. 2022; 53:2620–2627. doi: 10.1161/STROKEAHA.1121.038239LinkGoogle Scholar
  • 12 Polymeris AA, Meinel TR, Oehler H, Hölscher K, Zietz A, Scheitz JF, Nolte CH, Stretz C, Yaghi S, Stoll S, et al. Aetiology, secondary prevention strategies and outcomes of ischaemic stroke despite oral anticoagulant therapy in patients with atrial fibrillation. J Neurol Neurosurg Psychiatry. 2022; 93:588–598. doi: 10.1136/jnnp-2021-328391CrossrefMedlineGoogle Scholar
  • 13 Bonaventure Ip YM, Lau KK, Ko H, Lau L, Yao A, Lai‐Hung Wong G, Cheuk‐Fung Yip T, Leng X, Chan H, Chan H, et al. Association of alternative anticoagulation strategies and outcomes in patients with ischemic stroke while taking a direct oral anticoagulant. Neurology. 2023; 101:e358–e369. doi: 10.1212/wnl.0000000000207422CrossrefMedlineGoogle Scholar
  • 14 Paciaroni M, Caso V, Agnelli G, Mosconi MG, Giustozzi M, Seiffge DJ, Engelter ST, Lyrer P, Polymeris AA, Kriemler L, et al. Recurrent ischemic stroke and bleeding in patients with atrial fibrillation who suffered an acute stroke while on treatment with nonvitamin K antagonist oral anticoagulants: the RENO‐EXTEND study. Stroke. 2022; 53:2620–2627. doi: 10.1161/strokeaha.121.038239LinkGoogle Scholar
  • 15 Wu TY, Majeed A, Kuo KN. An overview of the healthcare system in Taiwan. London J Prim Care (Abingdon). 2010; 3:115–119. doi: 10.1080/17571472.2010.11493315CrossrefMedlineGoogle Scholar
  • 16 Cheng CL, Chien HC, Lee CH, Lin SJ, Yang YH. Validity of in‐hospital mortality data among patients with acute myocardial infarction or stroke in national health insurance research database in Taiwan. Int J Cardiol. 2015; 201:96–101. doi: 10.1016/j.ijcard.2015.07.075CrossrefMedlineGoogle Scholar
  • 17 Cheng CL, Kao YH, Lin SJ, Lee CH, Lai ML. Validation of the national health insurance research database with ischemic stroke cases in Taiwan. Pharmacoepidemiol Drug Saf. 2011; 20:236–242. doi: 10.1002/pds.2087CrossrefMedlineGoogle Scholar
  • 18 Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res. 2011; 46:399–424. doi: 10.1080/00273171.2011.568786CrossrefMedlineGoogle Scholar
  • 19 Granger CB, Alexander JH, McMurray JJ, Lopes RD, Hylek EM, Hanna M, Al‐Khalidi HR, Ansell J, Atar D, Avezum A, et al. Apixaban versus warfarin in patients with atrial fibrillation. N Engl J Med. 2011; 365:981–992. doi: 10.1056/NEJMoa1107039CrossrefMedlineGoogle Scholar
  • 20 Houston DS, Zarychanski R, Connolly S, Ezekowitz M, Yusuf S, Fihn S, McDonell M, Martin D, Casais P, Luceros A. Dabigatran versus warfarin in patients with atrial fibrillation. N Engl J Med. 2009; 31:2671–2675.Google Scholar
  • 21 Giugliano RP, Ruff CT, Braunwald E, Murphy SA, Wiviott SD, Halperin JL, Waldo AL, Ezekowitz MD, Weitz JI, Špinar J. Edoxaban versus warfarin in patients with atrial fibrillation. N Engl J Med. 2013; 369:2093–2104.CrossrefMedlineGoogle Scholar
  • 22 Hsieh MT, Liu CH, Lin SH, Lin PY, Chang YM, Wang CM, Chen CH, Sung PS. Recurrent ischemic stroke and cardiovascular outcomes in patients with atrial fibrillation with ischemic stroke despite direct oral anticoagulants. Stroke. 2023; 54:e145–e146. doi: 10.1161/strokeaha.122.041197LinkGoogle Scholar
  • 23 Kumar S, Lim E, Covic A, Verhamme P, Gale CP, Camm AJ, Goldsmith D. Anticoagulation in concomitant chronic kidney disease and atrial fibrillation: JACC review topic of the week. J Am Coll Cardiol. 2019; 74:2204–2215. doi: 10.1016/j.jacc.2019.08.1031CrossrefMedlineGoogle Scholar
  • 24 Kanuri SH, Kreutz RP. Pharmacogenomics of novel direct oral anticoagulants: newly identified genes and genetic variants. J Pers Med. 2019; 9:9. doi: 10.3390/jpm9010007CrossrefMedlineGoogle Scholar
  • 25 Raymond J, Imbert L, Cousin T, Duflot T, Varin R, Wils J, Lamoureux F. Pharmacogenetics of direct oral anticoagulants: a systematic review. J Pers Med. 2021; 11:37. doi: 10.3390/jpm11010037CrossrefMedlineGoogle Scholar
  • 26 Lin S‐Y, Tang S‐C, Tsai L‐K, Yeh S‐J, Huang C‐F, Jeng J‐S. Factors for recurrent stroke among Asian patients with non‐valvular atrial fibrillation under non‐vitamin K antagonist oral anticoagulant therapy. J Formos Med Assoc. 2020; 119:1799–1806. doi:10.1016/j.jfma.2020.02.003CrossrefMedlineGoogle Scholar
  • 27 Linder M, Iliadou Nyman A, Kieler H, Danielsson B, Borg N, Gry M, Collin J. Assessing safety of direct thrombin inhibitors, direct factor Xa inhibitors and vitamin K antagonists in patients with atrial fibrillation: a nation‐wide propensity score matched cohort from Sweden. Clin Epidemiol. 2020; 12:1029–1038. doi: 10.2147/clep.S258373CrossrefMedlineGoogle Scholar
  • 28 Graham DJ, Reichman ME, Wernecke M, Hsueh YH, Izem R, Southworth MR, Wei Y, Liao J, Goulding MR, Mott K, et al. Stroke, bleeding, and mortality risks in elderly Medicare beneficiaries treated with dabigatran or rivaroxaban for nonvalvular atrial fibrillation. JAMA Intern Med. 2016; 176:1662–1671. doi: 10.1001/jamainternmed.2016.5954CrossrefMedlineGoogle Scholar
  • 29 Diener HC, Hankey GJ, Easton JD, Lip GYH, Hart RG, Caso V. Non‐vitamin K oral anticoagulants for secondary stroke prevention in patients with atrial fibrillation. Eur Heart J Suppl. 2020; 22:I13–i21. doi: 10.1093/eurheartj/suaa104CrossrefMedlineGoogle Scholar
  • 30 Whitlock RP, Belley‐Cote EP, Paparella D, Healey JS, Brady K, Sharma M, Reents W, Budera P, Baddour AJ, Fila P, et al. Left atrial appendage occlusion during cardiac surgery to prevent stroke. N Engl J Med. 2021; 384:2081–2091. doi: 10.1056/NEJMoa2101897CrossrefMedlineGoogle Scholar
  • 31 Reddy VY, Neuzil P, de Potter T, van der Heyden J, Tromp SC, Rensing B, Jiresova E, Dujka L, Lekesova V. Permanent percutaneous carotid artery filter to prevent stroke in atrial fibrillation patients: the CAPTURE trial. J Am Coll Cardiol. 2019; 74:829–839. doi: 10.1016/j.jacc.2019.04.035CrossrefMedlineGoogle Scholar

eLetters(0)

eLetters should relate to an article recently published in the journal and are not a forum for providing unpublished data. Comments are reviewed for appropriate use of tone and language. Comments are not peer-reviewed. Acceptable comments are posted to the journal website only. Comments are not published in an issue and are not indexed in PubMed. Comments should be no longer than 500 words and will only be posted online. References are limited to 10. Authors of the article cited in the comment will be invited to reply, as appropriate.

Comments and feedback on AHA/ASA Scientific Statements and Guidelines should be directed to the AHA/ASA Manuscript Oversight Committee via its Correspondence page.