Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2023 Dec 1.
Published in final edited form as: Circ Cardiovasc Qual Outcomes. 2022 Dec 1;15(12):e008951. doi: 10.1161/CIRCOUTCOMES.122.008951

Comparative Effectiveness and Safety of Direct Oral Anticoagulants Versus Warfarin Among Adults with Cancer and Atrial Fibrillation

Hemalkumar B Mehta 1,2, Huijun An 1,2, Shirin Ardeshirrouhanifard 1,2, Mukaila A Raji 3, G Caleb Alexander 1,2,4, Jodi B Segal 1,2,4
PMCID: PMC9772095  NIHMSID: NIHMS1846109  PMID: 36453260

Abstract

Background:

While clinical guidelines recommend direct-acting oral anticoagulants (DOAC) over warfarin to treat isolated non-valvular atrial fibrillation (NVAF), guidelines are silent regarding NVAF treatment among individuals with cancer, reflecting the paucity of evidence in this setting. We quantified relative risk of ischemic stroke or systemic embolism and major bleeding (primary outcomes), and all-cause and cardiovascular death (secondary outcomes) among older individuals with cancer and NVAF comparing DOACs and warfarin.

Methods:

This retrospective cohort study used Surveillance, Epidemiology, and End Results cancer registry and linked US Medicare data from 2010 to 2016, and included individuals diagnosed with cancer and NVAF who newly initiated DOAC or warfarin. We used inverse probability of treatment weighting (IPTW) to control confounding. We used competing risk regression for primary outcomes and cardiovascular death, and Cox proportional hazard regression for all-cause death.

Results:

Among 7,675 individuals included in the cohort, 4,244 (55.3%) received DOACs and 3,431 (44.7%) warfarin. In the IPTW analysis, there was no statistically significant difference among DOAC and warfarin users in the risk of ischemic stroke or systemic embolism (1.24 vs. 1.19 events per 100 person-years, adjusted hazard ratio [aHR] 1.41, 95% confidence intervals [CI] 0.92-2.14), major bleeding (3.08 vs. 4.49 events per 100 person-years, aHR 0.90, CI 0.70-1.17), and cardiovascular death (1.88 vs. 3.14 per 100 person-years, aHR 0.82, CI 0.59-0.1.13). DOAC users had significantly lower risk of all-cause death (7.09 vs. 13.3 per 100 person-years, aHR 0.81, CI 0.69-0.94) compared to warfarin users.

Conclusions:

Older adults with cancer and AF exposed to DOACs had similar risks of stroke and systemic embolism and major bleeding as those exposed to warfarin. Relative to warfarin, DOAC use was associated with a similar risk of cardiovascular death and a lower risk of all-cause death.

Keywords: Direct-acting oral anticoagulants, warfarin, atrial fibrillation, cancer, comparative effectiveness

INTRODUCTION

Atrial fibrillation affects between 2.7 and 6.1 million adults in the US and the prevalence is expected to double by 2030.1,2,3 Cancer is an independent risk factor for the development of atrial fibrillation.4,5,6,7 Due to increasing cancer incidence and rising life expectancy of individuals with cancer, the incidence of atrial fibrillation among individuals with cancer will continue to increase.8,9 Among individuals with atrial fibrillation, those with cancer are at two-fold increased risk of stroke than those without due to hypercoagulability and noninfective endocarditis.10,11 Also, individuals with cancer are at a higher risk of bleeding than the general population due to cancer- and treatment-related thrombocytopenia and disseminated intravascular coagulation.12,13

While clinical guidelines recommend direct-acting oral anticoagulants (DOAC) over warfarin to treat isolated non-valvular atrial fibrillation (NVAF), guidelines are silent regarding NVAF treatment among individuals with cancer, reflecting the paucity of evidence in this setting.14,15 Although exploratory analyses of clinical trials tried to determine the safety and effectiveness of DOAC vs. warfarin in this population, these trial findings were limited by small sample sizes, heterogenous cancer diagnoses, and an absence of key cancer-related characteristics such as stage, grade, size and time of cancer diagnosis.16,17,18 Three single-institutional observational studies found that DOAC are safe and effective in patients with cancer and NVAF, but lacked an active control group, preventing any direct comparisons.19,20,21 Results from two real-world studies have provided contradictory evidence. Shah et al. found similar risks of ischemic stroke and major bleeding comparing DOAC and warfarin users, although apixaban users had a lower risk of bleeding;22 whereas Kim et al. found lower rates of ischemic stroke or systemic embolism, major bleeding, and all-cause death among DOAC users compared to warfarin users.23

We used the Surveillance Epidemiology and End Results (SEER) cancer registry and linked Medicare claims data to compare the effectiveness and safety of DOAC versus warfarin for the risk of ischemic stroke or systemic embolism, major bleeding, all-cause death and cardiovascular death among a large, diverse population of older adults with cancer and NVAF. We hypothesized that individuals using DOACs have lower risk of ischemic stroke, systemic embolism and bleeding, and similar risk of all-cause and cardiovascular death, compared to warfarin.

METHODS

Data source

We used SEER linked to Medicare claims data from 2009 to 2016. The SEER-Medicare data is a linkage of two large population-based sources of data that provides detailed information about Medicare beneficiaries with cancer.24 The SEER data collects information on cancer incidence, treatment, and survival from 21 cancer registries and covers approximately 35% of the US population. This has been linked by the National Cancer Institute, the SEER registries, and the Centers for Medicare and Medicaid to the Medicare data, which are the healthcare claims for all Medicare beneficiaries.

The Johns Hopkins Bloomberg School of Public Health Institutional Review Board approved this study. The data that support the findings of this study are available from the National Cancer Institute, and additional materials including SAS codes can be provided by the corresponding author for the purposes of reproducible results upon reasonable request.

Study design and population

In this new-user retrospective cohort study, we included adults aged 66 years or older diagnosed with cancer who used at least one oral anticoagulant from January 1, 2010 through December 31, 2016 (Figure 1 in Data Supplement). We included individuals with breast, bladder, colorectal, esophagus, lung, ovary, kidney, pancreas, prostate, stomach and uterus cancer because they are common and have a high prevalence of atrial fibrillation.6,25 The first prescription of an oral anticoagulant was considered the individual’s index date. We excluded individuals who: 1) were not continuously enrolled in Medicare parts A, B and D or enrolled in a health maintenance organization within the twelve months prior to the index date; 2) had no diagnosis code for atrial fibrillation during the twelve months prior to the index date; 3) had a diagnosis code for mitral or aortic disease, heart valve surgery or mitral/aortic valve surgery; 4) received oral anticoagulants within twelve months prior to the index date; 5) received both a DOAC and warfarin on the index date; and 6) were receiving hospice care on the index date. We identified atrial fibrillation based on the presence of at least one inpatient or two outpatient claims in 1 year prior to the index date, a definition with a sensitivity of 79% and positive predictive value of 89% (Table 1 in Data Supplement).26,27

Exposure to oral anticoagulants

We used prescription fill information from the Medicare Part D data to define exposure to DOACs (apixaban, dabigatran, rivaroxaban or edoxaban) and warfarin. We included all possible doses of DOAC and warfarin for defining drug exposure. Based on the first oral anticoagulant prescription, we considered patients exposed to DOAC or to warfarin. In the primary analysis, we defined drug exposure using an “as-treated” approach where individuals were censored when they were no longer exposed to their initial drug class.

Outcomes

The primary outcomes were ischemic stroke or systemic embolism, and major bleeding (Table 1 in Data Supplement). Secondary outcomes were all-cause death and deaths due to cardiovascular diseases. We identified patients hospitalized for ischemic stroke or systemic embolism as the primary discharge diagnosis in the inpatient claims. These codes have positive predictive value over 80% for ischemic stroke definition.28,29 Major bleeding included hospitalization due to intracranial bleeding, gastrointestinal bleeding or non-gastrointestinal bleeding based on the primary discharge diagnosis in any inpatient claims. The positive predictive values are 75.5% for intracranial bleeding and 91.8% for major gastrointestinal and non-gastrointestinal bleeding.30 We used SEER cause of death and identified cardiovascular death if individuals died from heart diseases, hypertension, cerebrovascular disease, atherosclerosis, aortic aneurysm or other diseases of arteries, arterioles, or capillaries. We included hospitalization due to sepsis as a negative control outcome to assess the influence of unmeasured confounding.31,32

Covariates

We identified covariates for adjustment using information from the twelve months prior to the index date (Table 2 in Data Supplement). We used SEER data to identify age, sex, race, median household income, Medicaid eligibility, residential census region, and cancer-related characteristics such as cancer site, stage, grade, tumor size and the use of chemotherapy. Active cancer was defined as receipt of chemotherapy, radiation, or cancer surgery in 12 months prior to the oral anticoagulation initiation. We operationalized 20 comorbidities, 18 measures of medication use, a set of frailty indicators, and prior healthcare use from the Medicare claims as potential confounders.14,22,23 Among these were the diagnoses required to calculate CHA2DS2-VASc and HAS-BLED scores, which we operationalized as individual covariates.33,34 Healthcare utilization was quantified by the count of hospital admissions and physician visits.

Statistical analyses

We calculated a propensity score with logistic regression to predict DOAC vs. warfarin use conditional on individual’s sociodemographic, cancer and clinical characteristics. We then generated stabilized inverse probability of treatment weights (IPTW), which we trimmed at the 1st and 99th percentiles.35 We evaluated the balance of covariates between groups in the original and IPTW weighted samples, where a standardized difference of <0.10 suggests acceptable balance in covariates across groups.36 We separately modeled the outcomes of ischemic stroke or systemic embolism, major bleeding, and cardiovascular death using Fine and Gray’s competing risk regression.37,38, 39 Individuals were censored when they discontinued their index drug class (gap of more than 30 days between two prescriptions), switched to a different anticoagulant drug class, lost fee-for-service Medicare coverage, were admitted to hospice, or concluded the study period. We considered all-cause death to be a competing risk for ischemic stroke or systemic embolism and major bleeding, and non-cardiovascular death to be a competing risk for cardiovascular death. We used a multivariable Cox proportional hazards regression model for all-cause death. The IPTW-weighted competing risk regression and Cox regression included calendar year of drug initiation, time from cancer diagnosis to oral anticoagulant initiation, and any unbalanced covariates. We also used Fine and Gray’s competing risk regression model for the negative control outcome of sepsis, expecting a null association. We reported adjusted hazard ratios (aHR) and 95% confidence intervals (95% CI) for each outcome of interest.

Subgroup and sensitivity analyses

We interacted treatment with age (<75 years, ≥ 75 years), sex, active cancer, and cancer type, each in separate regression models. In sensitivity analyses, we compared each unique DOAC to warfarin. Second, we performed intention-to-treat analysis in which individuals were analyzed in their index drug exposure group and not censored at the time of switching or discontinuation. Third, we used Cox proportional hazards regression model for ischemic stroke or systemic embolism, bleeding, and sepsis, with censoring upon death. Fourth, to evaluate for a dose-response relationship, we stratified by duration of oral anticoagulant use (<90 days, 91-180 days, >180 days) in the as-treated analysis. We performed quantitative bias assessment by calculating an E-value, which represents the strength of association that an unmeasured confounder needs to have with both the exposure and outcome variables, conditional on the measured covariates, to change the interpretation of our findings.40

We conducted all analyses using SAS version 9.4 (SAS Institute, Cary, NC).

RESULTS

Study cohort

We identified 105,923 individuals diagnosed with primary cancer who filled at least one prescription of oral anticoagulants from 2010 to 2016. After applying exclusion criteria, the final cohort included 7,675 individuals diagnosed with cancer and atrial fibrillation who newly initiated DOAC or warfarin (Figure 1). Overall, 4,244 (55.3%) individuals received DOAC and 3,431 (44.7%) warfarin. The use of DOAC increased over time, from 2% in 2010 to 76.2% in 2016 (Figure 2 in Data Supplement). The mean (standard deviation) age of the cohort was 76.6 (6.9) years, 48.1% were female, and 88.0% were non-Hispanic white. Individuals diagnosed with prostate (22.2%), breast (19.6%), lung (19.3%) and colorectal (14.5%) cancers accounted for approximately three-fourths of the cohort.

Figure 1. Cohort derivation.

Figure 1.

Open in a new tab

aFirst prescription of oral anticoagulants is considered as an index date.

Abbreviations: DOAC, direct-acting oral anticoagulants; HMO, health maintenance organization; NVAF, non-valvular atrial fibrillation

Balance of patient characteristics

DOAC users differed significantly from warfarin users by sociodemographic, cancer diagnoses, and clinical characteristics (Table 1, Table 3 in Data Supplement). After IPTW, both groups were well balanced, with all standardized differences below −0.022 (or 2.2%). There was a considerable overlap in the propensity score distribution among warfarin and DOAC users after IPTW (eFigure 3).

Table 1.

Baseline patient characteristics among warfarin and direct-acting oral anticoagulant (DOAC) users, before and after inverse probability of treatment weighting (IPTW)a

Patient
characteristics		 Original Sample IPTW sample
Warfarin
(n=3,431)		 DOAC
(n=4,244)		 SMD Warfarin
(n=3,407)		 DOAC
(n=4,200)		 SMD
Sociodemographic
Age, mean ± sd 77.1 ± 7.1 76.5 ± 6.8 −0.091 76.7 ± 7.0 76.7 ± 6.9 −0.002
Gender
 Male 1,727 (50.3) 2,253 (53.1) 1,774 (52.1) 2,183 (52.0)
 Female 1,704 (49.7) 1,991 (46.9) −0.055 1,633 (47.9) 2,017 (48.0) 0.002
Race
 White 3,005 (87.6) 3,752 (88.4) 0.025 2,991 (87.8) 3,687 (87.8) 0.00004
 Black 238 (6.9) 234 (5.5) −0.059 217 (6.4) 267 (6.4) −0.0006
 Hispanic 51 (1.5) 41 (1.0) −0.047 44 (1.3) 53 (1.3) −0.002
 Other 137 (4.0) 217 (5.1) 0.054 155 (4.6) 193 (4.6) 0.002
Marital Status
 Married 1,742 (50.8) 2,286 (53.9) 0.062 1,779 (52.2) 2,209 (52.6) 0.008
 Other 1,689 (49.2) 1,958 (46.1) 1,628 (47.8) 1,991 (47.4)
Medicaid Eligibility 842 (24.5) 828 (19.5) −0.122 747 (21.9) 917 (21.8) 0.008
Census Region
 West 1,282 (37.4) 1,639 (38.6) 0.026 1,286 (37.7) 1,596 (38.0) 0.006
 Northeast 737 (21.5) 997 (23.5) 0.048 767 (22.5) 956 (22.8) 0.006
 Midwest 576 (16.8) 459 (10.8) −0.174 463 (13.6) 559 (13.3) −0.008
 South 814 (23.7) 1,104 (26.0) 0.053 863 (25.3) 1,053 (25.1) −0.006
 Missing 22 (0.6) 45 (1.1) 0.046 28 (0.8) 36 (0.9) 0.003
Median household income per capita
 1st Quartile 909 (26.5) 955 (22.5) −0.093 834 (24.5) 1,015 (24.2) −0.007
 2nd Quartile 930 (27.1) 936 (22.1) −0.118 839 (24.6) 1,026 (24.4) −0.005
 3rd Quartile 812 (23.7) 1,051 (24.8) 0.026 837 (24.6) 1,026 (24.4) −0.003
 4th Quartile 698 (20.3) 1,166 (27.5) 0.168 801 (23.5) 1,015 (24.2) 0.015
 Missing 82 (2.4) 136 (3.2) 0.049 96 (2.8) 119 (2.8) 0.0007
Cancer
Cancer Site
 Bladder 294 (8.6) 365 (8.6) 0.001 296 (8.7) 363 (8.7) −0.002
 Breast 600 (17.5) 906 (21.4) 0.098 660 (19.4) 831 (19.8) 0.010
 Colorectal 593 (17.3) 521 (12.3) −0.141 491 (14.4) 588 (14.0) −0.012
 Esophagus 53 (1.5) 58 (1.4) −0.015 46 (1.3) 56 (1.3) −0.0006
 Kidney 170 (5.0) 204 (4.8) −0.007 165 (4.9) 205 (4.9) 0.002
 Lung 770 (22.4) 712 (16.8) −0.143 667 (19.6) 811 (19.3) −0.007
 Ovary 79 (2.3) 57 (1.3) −0.072 58 (1.7) 70 (1.7) −0.003
 Pancreas 97 (2.8) 70 (1.7) −0.080 77 (2.3) 95 (2.3) 0.0008
 Prostate 591 (17.2) 1,112 (26.2) 0.219 752 (22.1) 939 (22.4) 0.007
 Stomach 53 (1.5) 68 (1.6) 0.005 53 (1.6) 66 (1.6) 0.002
 Uterus 131 (3.8) 171 (4.0) 0.011 142 (4.2) 175 (4.2) −0.00002
Cancer Stage
 Stage 0 295 (8.6) 416 (9.8) 0.042 314 (9.2) 387 (9.2) 0.00006
 Stage I 1,058 (30.8) 1,504 (35.4) 0.098 1,144 (33.6) 1,416 (33.7) 0.003
 Stage II 841 (24.5) 1,195 (28.2) 0.083 889 (26.1) 1,109 (26.4) 0.007
 Stage III 551 (16.1) 549 (12.9) −0.089 488 (14.3) 587 (14.0) −0.010
 Stage IV 502 (14.6) 383 (9.0) −0.174 402 (11.8) 489 (11.6) −0.005
 Unknown 184 (5.4) 197 (4.6) −0.033 171 (5.0) 212 (5.0) 0.001
Grade
 Grade I 386 (11.3) 553 (13.0) 0.055 411 (12.1) 510 (12.1) 0.002
 Grade II 1,192 (34.7) 1,604 (37.8) 0.064 1,250 (36.7) 1,539 (36.6) −0.001
 Grade III 949 (27.7) 1,184 (27.9) 0.005 945 (27.7) 1,163 (27.7) −0.001
 Grade IV 172 (5.0) 173 (4.1) −0.045 153 (4.5) 190 (4.5) 0.002
 Cell type not determined 732 (21.3) 730 (17.2) −0.105 648 (19.0) 798 (19.0) −0.001
Tumor size
 0cm 20 (0.68) 26 (0.6) 0.004 20 (0.6) 24.1 (0.6) −0.001
 0.1 - 1cm 239 (7.0) 402 (9.5) 0.091 272 (8.0) 351 (8.4) 0.014
 1.1 - 2cm 503 (14.7) 627 (14.8) 0.003 502 (14.8) 617 (14.7) −0.002
 2.1 - 3cm 427 (12.5) 492 (11.6) −0.026 417 (12.3) 509 (12.1) −0.004
 3.1 - 4cm 355 (10.4) 375 (8.8) −0.051 326 (9.6) 400 (9.5) −0.002
 4.1 - 5cm 263 (7.7) 279 (6.6) −0.042 238 (7.0) 289 (6.9) −0.004
 >5cm 592 (17,3) 526 (12.4) −0.137 502 (14.7) 603 (14.4) −0.011
 Not Available/Site-specific codes 1,032 (30.1) 1,517 (35.7) 0.121 1,129 (33.1) 1,407 (33.5) 0.007
Receipt of chemotherapy 1,000 (29.2) 1,193 (28.1) −0.023 981 (28.8) 1,203 (28.7) −0.003
Active cancer 1,272 (37.1) 1,476 (34.8) −0.048 1,253 (36.8) 1,484 (35.3) −0.030
Comorbiditiesa
 Anemia 2,030 (59.2) 1,969 (46.4) −0.258 1,779 (52.2) 2,183 (52.0) −0.005
 Dementia 274 (8.0) 260 (6.1) −0.073 235 (6.9) 288 (6.9) −0.002
 Liver disease 600 (17.5) 567 (13.4) −0.114 528 (15.5) 642 (15.3) −0.006
 Peptic ulcer disease 169 (4.9) 129 (3.0) −0.097 132 (3.9) 157 (3.7) −0.008
 Prior major bleeding or predisposition to bleeding 2,418 (70.5) 2,568 (60.5) −0.211 2,230 (65.5) 2,728 (65.0) −0.010
 Renal disease 1,284 (37.4) 1,243 (29.3) −0.173 1,136 (33.3) 1,388 (33.0) −0.006
Stroke, TIA, thromboembolism history 1,742 (50.8) 1,393 (32.8) −0.370 1,401 (41.1) 1,686 (40.2) −0.020
Vascular disease history 1,471 (42.9) 1,478 (34.8) −0.166 1,316 (38.6) 1,609 (38.3) −0.006
Co-medication usea
 Antiplatelets 850 (24.8) 1,031 (24.3) −0.011 853 (25.0) 1,038 (24.7) −0.008
 Heparin 936 (27.3) 428 (10.1) −0.452 609 (17.9) 715 (17.0) −0.022
Open in a new tab

Abbreviations: SMD standardized mean difference

a

Comorbidities, frailty indicator, co-medications and health service use were used as covariates in logistic regression model to derive IPTW weights. The balance of these covariates before and after IPTW weighting are reported in Table 3 in Data Supplement.

Risk of ischemic stroke or systemic embolism, major bleeding and mortality

The cumulative incidence of ischemic stroke or systemic embolism was not significantly different among DOAC and warfarin users (Figure 4A in Data Supplement). Over a median follow-up of 7.7 months (interquartile range [IQR], 3.1-16.7 months), the rate of ischemic stroke or systemic embolism per 100 person years among DOAC and warfarin users were 1.24 and 1.19, respectively. In the IPTW weighted analysis, DOAC use was not associated with a significantly greater risk of ischemic stroke or systemic embolism (aHR 1.41, CI 0.92-2.14) compared to warfarin use (Table 2). Over a median follow-up of 7.4 months (IQR, 3.0-16.2 months), the rate or major bleeding per 100 person years among DOAC users and warfarin users was 3.08 and 4.49, respectively. (Figure 4B, Table 4 in Data Supplement) In the IPTW weighted analysis, there was not a significant difference in rates of bleeding (aHR 0.90, CI 0.70-1.17) (Table 2). The E-values were 2.17 for ischemic stroke or systemic embolism and 1.46 for major bleeding.

Table 2.

Adjusted association of direct-acting oral anticoagulant (DOAC) with outcomes relative to warfarin

Warfarin DOAC
Number of patients 3431 4244
Ischemic stroke or systemic embolism (primary outcome)
 Events, n 42 52
 Total person time follow up 3525.39 4178.02
 Events per 100 patient years, n (95% CI) 1.19 (0.86, 1.61) 1.24 (0.93, 1.63)
 Unadjusted HR (95% CI) Reference 1.11 (0.74, 1.66)
 IPTW adjusted HR (95% CI) Reference 1.41 (0.92-2.14)
 E-value for IPTW adjusted HR - 2.17
Major bleeding (primary outcome)
 Events, n 153 126
 Total person time follow up 3406.55 4088.97
 Events per 100 patient years, n (95% CI) 4.49 (3.81, 5.26) 3.08 (2.57, 3.67)
 Unadjusted HR (95% CI) Reference 0.69 (0.55, 0.88)
 IPTW adjusted HR (95% CI) Reference 0.90 (0.70-1.17)
 E-value for IPTW adjusted HR - 1.46
All-cause death (secondary outcome)
 Events, n 471 298
 Total person time follow up 3539.03 4201.64
 Events per 100 patient years, n (95% CI) 13.3 (12.1, 14.6) 7.09 (6.31, 7.95)
 Unadjusted HR (95% CI) Reference 0.51 (0.44, 0.59)
 IPTW adjusted HR (95% CI) Reference 0.81 (0.69-0.94)
 E-value for IPTW adjusted HR - 1.77
Cardiovascular death (secondary outcome)
 Events, n 111 79
 Total person time follow up 3539.03 4201.64
 Events per 100 patient years, n (95% CI) 3.14 (2.18, 3.78) 1.88 (1.49, 2.34)
 Unadjusted HR (95% CI) Reference 0.61 (0.46, 0.81)
 IPTW adjusted HR (95% CI) Reference 0.82 (0.59-1.13)
 E-value for IPTW adjusted HR - 1.74
Sepsis (negative control outcome)
 Events, n 70 105
 Total person time follow up 3517.81 4171.78
 Events per 100 patient years, n (95% CI) 1.99 (1.55, 2.51) 2.52 (2.06, 3.05)
 Unadjusted HR (95% CI) Reference 1.34 (0.99, 1.81)
 IPTW adjusted HR (95% CI) Reference 1.07 (0.79-1.46)
 E-value for IPTW adjusted HR - 1.34
Open in a new tab

Abbreviations: CI confidence interval, HR hazard ratio, IPTW inverse probability of treatment weighting

In the IPTW weighted analysis, DOAC use was associated with fewer all-cause deaths relative to warfarin use (aHR 0.81, CI 0.69-0.94), with E-value of 1.77 (Figure 5 in Data Supplement and Table 2). Overall, 190 individuals died due to cardiovascular diseases (Table 5 in Data Supplement). In the IPTW weighted analysis, DOAC use was not associated with reduced risk of cardiovascular deaths relative to warfarin (aHR 0.82, CI 0.59-1.13), with E-value of 1.74 (Figure 6 in Data Supplement and Table 2). We did not observe any statistically significant difference between DOAC use and warfarin use and the risk of sepsis (aHR 1.07, CI 0.79-1.46).

Subgroup and sensitivity analyses

In subgroup analyses by age, sex, active cancer and cancer sites, the results for ischemic stroke or systemic embolism and major bleeding were similar to the main analysis indicating little heterogeneity in the association between the exposure and the outcome across subsets of patients. We noted all-cause survival benefit in some subgroups. However, when we evaluated cardiovascular death, benefit favoring DOAC was only seen among individuals with no active cancer (Table 6 in Data Supplement), which was not seen in the comparison group.

In analyses by individual DOAC; apixaban, dabigatran and rivaroxaban demonstrated no significantly increased risk for ischemic stroke or systemic embolism compared to warfarin. The risk of bleeding was similar for dabigatran and rivaroxaban, but lower for apixaban, compared to warfarin. All three products were associated with a lower risk of all-cause death compared to warfarin, but only rivaroxaban was associated with a lower risk of cardiovascular death (Table 3). Our findings were consistent in three sensitivity analyses where we evaluated intent-to-treat drug exposure, employed Cox regression models, and stratified drug exposure by duration of oral anticoagulant use (Table 7 in Data Supplement).

Table 3.

Adjusted association of individual direct-acting oral anticoagulant (DOAC) drug with outcomes relative to warfarina

Warfarin Apixaban Dabigatran Rivaroxaban
Number of patients 3431 1830 526 1874
Ischemic stroke or systemic embolism (primary outcome)
 Total person time follow up 3525.39 1357.01 730.25 2081.99
 Events per 100 patient years, n (95% CI) 1.19 (0.86, 1.61) 1.25 (0.73, 2.01) 1.23 (0.56, 2.34) 1.25 (0.82, 1.83)
 Unadjusted HR (95% CI) Reference 1.09 (0.62, 1.91) 1.10 (0.54, 2.25) 1.14 (0.70, 1.86)
 IPTW adjusted HR (95% CI) Reference 1.22 (0.65, 2.28) 1.19 (0.56, 2.51) 1.14 (0.66, 1.98)
 E-value for IPTW adjusted HR - 1.74 1.67 1.54
Major bleeding (primary outcome)
 Total person time follow up 3406.55 1346.42 697.19 2036.6
 Events per 100 patient years, n (95% CI) 4.49 (3.81, 5.26) 2.45 (1.69, 3.44) 3.87 (2.55, 5.64) 3.19 (2.46, 4.07)
 Unadjusted HR (95% CI) Reference 0.48 (0.33, 0.70) 1.03 (0.68, 1.55) 0.76 (0.57, 1.02)
 IPTW adjusted HR (95% CI) Reference 0.57 (0.38, 0.86) 0.84 (0.52, 1.36) 0.82 (0.60, 1.12)
 E-value for IPTW adjusted HR - 2.90 1.67 1.74
All-cause death (secondary outcome)
 Total person time follow up 3539.03 1360.79 738.83 2093.24
 Events per 100 patient years, n (95% CI) 13.3 (12.1, 14.6) 8.23 (6.78, 9.90) 5.41 (3.87, 7.37) 6.93 (5.85, 8.15)
 Unadjusted HR (95% CI) Reference 0.52 (0.42, 0.64) 0.45 (0.33, 0.62) 0.52 (0.43, 0.63)
 IPTW adjusted HR (95% CI) Reference 0.70 (0.56, 0.87) 0.68 (0.51, 0.90) 0.67 (0.55, 0.81)
 E-value for IPTW adjusted HR - 2.21 2.30 2.35
Cardiovascular death (secondary outcome)
 Total person time follow up 3539.03 1360.79 738.83 2093.24
 Events per 100 patient years, n (95% CI) 3.14 (2.58, 3.78) 2.35 (1.61, 3.32) 2.17 (1.24, 3.52) 1.48 (1.01, 2.10)
 Unadjusted HR (95% CI) Reference 0.69 (0.46, 1.03) 0.80 (0.47, 1.34) 0.49 (0.33, 0.73)
 IPTW adjusted HR (95% CI) Reference 0.86 (0.53, 1.40) 1.10 (0.68, 1.77) 0.62 (0.40, 0.95)
 E-value for IPTW adjusted HR - 1.60 1.43 2.61
Open in a new tab

Abbreviations: CI confidence interval, HR hazard ratio, IPTW inverse probability of treatment weighting

a

Very few people received edoxaban. Therefore, we did not compare edoxaban vs. warfarin.

DISCUSSION

In this retrospective cohort study of a large, diverse group of U.S. Medicare beneficiaries with cancer and non-valvular atrial fibrillation, although we hypothesized lower risk of stroke or systemic embolism, we did not find statistically significant difference among DOAC versus warfarin users in the risk of ischemic stroke or systemic embolism. We also did not find significant differences for major bleeding among DOAC versus warfarin users, except for apixaban which showed reduced risk. DOACs were associated with lower risk of all-cause mortality compared to warfarin. These findings persisted in subgroup and several sensitivity analyses.

Our findings add to a growing evidence base regarding the safety and effectiveness of DOACs among individuals with cancer and NVAF. Analyses of ROCKET AF (comparing rivaroxaban vs. warfarin),16 ARISTOTLE (apixaban vs. warfarin)17 and ENGAGE AF-TIMI (edoxaban vs. warfarin) trials,18 and an observational study using Marketscan claims data found that DOACs were as effective as warfarin to prevent stroke and major bleeding among cancer patients with NVAF.22 Data from our real-world study complements trial findings and a previous real-word study. Although a study from Korea reported protective effect of DOACs compared with warfarin on clinical outcomes, the study did not adequately control for confounding.23

Among individual DOACs, we found that apixaban compared to warfarin reduced the risk of major bleeding by 43%. The ARISTOTLE trial did not find any association of apixaban with major bleeding, whereas Shah et al. found that apixaban reduced the risk of major bleeding by 63% compared to warfarin. As noted by the authors, the ARISTOTLE trial may be underpowered to detect any significant differences. Our estimates may be more precise given we had large cohort and we better controlled confounding by including cancer characteristics such as stage, grade and tumor size which were available in the SEER cancer registry.

We evaluated both all-cause and cardiovascular deaths. Contrary to our hypothesis of similar risk of death among DOAC and warfarin users, we noted lower risk of all-cause death among DOAC users. This was consistent with a recent Cochrane review of 10 RCTs (n=65,624), which showed a significant all-cause mortality reduction effect of factor Xa inhibitors compared with warfarin (OR 0.89, 95% 0.83 to 0.95) among individuals with atrial fibrillation.41 Similar results of lower all-cause mortality were seen among Medicare beneficiaries with atrial fibrillation who were using dabigatran,42 and commercially insured patients with atrial fibrillation using rivaroxaban.43 Our results of no benefits of DOACs, except for rivaroxaban, for cardiovascular deaths were not in agreement with an analysis of 4 trials (n=30,907) that showed a significant reduction in cardiovascular mortality among individual with atrial fibrillation who received DOACs (risk ratio 0.87, 95% CI 0.80 to 0.95).44 Although we controlled for baseline measured confounders, it is possible that unmeasured confounding or differences in cancer characteristics and treatment received after the oral anticoagulation initiation may impact survival. Further work might explore the basis for this association and disentangle causes of deaths among cancer patients with atrial fibrillation.

Our study had some strengths. This was a new user active comparator study, where we included eleven cancers and controlled confounding due to cancer-related and clinical factors. We used propensity score method to control selection bias and confounding. To assess the role of unmeasured confouding, we used negative control outcomes and E-values. Finally, we performed several subgroup and sensitity analyses to assess the robustness of study findings.

Our study had some limitations. First, our study findings may not be generalizable to patients with other cancers and those who are not enrolled in Medicare fee-for-service. Second, we cannot rule out the possibility of unmeasured confounding given relatively low E-value for some outcomes. Third, some subgroup analysis maybe underpowered because of low number of events. Fourth, due to small number of patients in some subgroups such as those with upper gastrointestinal cancers, we did not evaluate outcomes in such population. Fifth, nearly one-third of individuals discontinued DOACs or warfarin and this may introduce bias if the reason for discontinuation differed between these two groups. However, our results were consistent in an intention-to-treat analysis where we did not censor individuals upon switching or discontinuation. Sixth , we relied on administrative Medicare claims data to identify drug exposure and clinical outcomes and we cannot rule out the possibility of coding error as these data are not collected for research purposes. Seventh, because we looked at the class effect of DOACs in the primary analysis, we did not evaluate the effect of different dosing on clinical outcomes. Finally, we could not control for some factors that may play role in the prescription of DOACs such as frailty.

In conclusion, among individuals diagnosed with cancer and NVAF, DOACs demonstrated similar effectiveness in preventing stroke and systemic embolism without increasing the risk of major bleeding. Moreover, DOACs were associated with similar risk of cardiovascular death and lower risk of all-cause death. DOACs may be safe and effective alternatives to warfarin among individuals with cancer and NVAF.

Supplementary Material

Supplemental Publication Material File
STROBE Checklist

What is Known

  • Individuals with cancer and atrial fibrillation are at greater risk of ischemic stroke or systemic embolism and major bleeding than the general population with atrial fibrillation.

  • Exploratory analyses of three clinical trials were limited in their ability to determine the safety and effectiveness of direct-acting oral anticoagulants vs. warfarin among older adults with cancer and non-valvular atrial fibrillation.

What the Study Adds

  • Among individuals with cancer and non-valvular atrial fibrillation, direct-acting oral anticoagulants had similar risks of ischemic stroke or systemic embolism and major bleeding compared to warfarin.

  • Direct-acting oral anticoagulants were associated with lower risk of all-cause death compared to warfarin.

ACKNOWLEDGMENTS

We thank the Applied Research Program of the National Cancer Institute; the Office of Research, Development and Information of the Centers for Medicaid Services; Information Management Services, Inc; and the Surveillance, Epidemiology, and End Results (SEER) program tumor registries for their work in creating the SEER database.

Funding Support

This research was funded in part by an investigator-initiated research award from Bristol-Myers Squibb and Pfizer, as part of the American Thrombosis Investigator Initiated Research Program (ARISTA-USA). Dr. Ardeshirrouhanifard is supported by National, Heart, Lung, and Blood Institute (5T32HL139426-03). Dr. Mehta is supported by the National Institute on Aging (1K01AG070329). Dr. Segal is supported by the National Institute for Health (K24 AG049036/NIA).

List of Abbreviations

CI

Confidence interval

DOAC

Direct-acting oral anticoagulants

IPTW

Inverse probability of treatment weights

aHR

Adjusted hazard ratio

NVAF

Non-valvular atrial fibrillation

SEER

Surveillance Epidemiology and End Results

Footnotes

Disclosure

Dr. Alexander is a current member and past Chair of FDA’s Peripheral and Central Nervous System Advisory Committee, is a co-founding Principal and equity holder in Monument Analytics, a health care consultancy whose clients include the life sciences industry as well as plaintiffs in opioid litigation, and is a past member of OptumRx’s National P&T Committee. These arrangements have been reviewed and approved by Johns Hopkins University in accordance with its conflict of interest policies. All other authors do not have declarations of interest.

REFERENCES

  • 1.January CT, Wann LS, Alpert JS, et al. 2014 AHA/ACC/HRS guideline for the management of patients with atrial fibrillation: executive summary: a report of the American College of Cardiology/American Heart Association Task Force on practice guidelines and the Heart Rhythm Society [published correction appears in Circulation. 2014 Dec 2;130(23):e270-1]. Circulation. 2014;130(23):2071–2104. doi: 10.1161/CIR.0000000000000040 [DOI] [PubMed] [Google Scholar]
  • 2.Colilla S, Crow A, Petkun W, Singer DE, Simon T, Liu X. Estimates of current and future incidence and prevalence of atrial fibrillation in the U.S. adult population. Am J Cardiol. 2013;112(8):1142–1147. doi: 10.1016/j.amjcard.2013.05.063 [DOI] [PubMed] [Google Scholar]
  • 3.Miyasaka Y, Barnes ME, Gersh BJ, et al. Secular trends in incidence of atrial fibrillation in Olmsted County, Minnesota, 1980 to 2000, and implications on the projections for future prevalence [published correction appears in Circulation. 2006 Sep 12;114(11):e498]. Circulation. 2006;114(2):119–125. doi: 10.1161/CIRCULATIONAHA.105.595140 [DOI] [PubMed] [Google Scholar]
  • 4.Yun JP, Choi EK, Han KD, et al. Risk of Atrial Fibrillation According to Cancer Type: A Nationwide Population-Based Study. JACC CardioOncol. 2021;3(2):221–232. Published 2021 Jun 15. doi: 10.1016/j.jaccao.2021.03.006 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Yuan M, Zhang Z, Tse G, et al. Association of Cancer and the Risk of Developing Atrial Fibrillation: A Systematic Review and Meta-Analysis. Cardiol Res Pract. 2019;2019:8985273. Published 2019 Apr 14. doi: 10.1155/2019/8985273 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.O'Neal WT, Lakoski SG, Qureshi W, et al. Relation between cancer and atrial fibrillation (from the REasons for Geographic And Racial Differences in Stroke Study). Am J Cardiol. 2015;115(8):1090–1094. doi: 10.1016/j.amjcard.2015.01.540 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Farmakis D, Parissis J, Filippatos G. Insights into onco-cardiology: atrial fibrillation in cancer. J Am Coll Cardiol. 2014;63(10):945–953. doi: 10.1016/j.jacc.2013.11.026 [DOI] [PubMed] [Google Scholar]
  • 8.Siegel RL, Miller KD, Jemal A. Cancer statistics, 2020. CA Cancer J Clin. 2020;70(1):7–30. doi: 10.3322/caac.21590 [DOI] [PubMed] [Google Scholar]
  • 9.Kornej J, Börschel CS, Benjamin EJ, Schnabel RB. Epidemiology of Atrial Fibrillation in the 21st Century: Novel Methods and New Insights. Circ Res. 2020;127(1):4–20. doi: 10.1161/CIRCRESAHA.120.316340 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Koene RJ, Prizment AE, Blaes A, Konety SH. Shared Risk Factors in Cardiovascular Disease and Cancer. Circulation. 2016;133(11):1104–1114. doi: 10.1161/CIRCULATIONAHA.115.020406 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Navi BB, Reiner AS, Kamel H, et al. Risk of Arterial Thromboembolism in Patients With Cancer. J Am Coll Cardiol. 2017;70(8):926–938. doi: 10.1016/j.jacc.2017.06.047 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Melloni C, Shrader P, Carver J, et al. Management and outcomes of patients with atrial fibrillation and a history of cancer: the ORBIT-AF registry. Eur Heart J Qual Care Clin Outcomes. 2017;3(3):192–197. doi: 10.1093/ehjqcco/qcx004 [DOI] [PubMed] [Google Scholar]
  • 13.Angelini DE, Radivoyevitch T, McCrae KR, Khorana AA. Bleeding incidence and risk factors among cancer patients treated with anticoagulation. Am J Hematol. 2019;94(7):780–785. doi: 10.1002/ajh.25494 [DOI] [PubMed] [Google Scholar]
  • 14.January CT, Wann LS, Calkins H, et al. 2019 AHA/ACC/HRS Focused Update of the 2014 AHA/ACC/HRS Guideline for the Management of Patients With Atrial Fibrillation: A Report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines and the Heart Rhythm Society in Collaboration With the Society of Thoracic Surgeons [published correction appears in Circulation. 2019 Aug 6;140(6):e285]. Circulation. 2019;140(2):e125–e151. doi: 10.1161/CIR.0000000000000665 [DOI] [PubMed] [Google Scholar]
  • 15.Sorigue M, Miljkovic MD. Atrial Fibrillation and Stroke Risk in Patients With Cancer: A Primer for Oncologists. J Oncol Pract. 2019;15(12):641–650. doi: 10.1200/JOP.18.00592 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Chen ST, Hellkamp AS, Becker RC, et al. Efficacy and safety of rivaroxaban vs. warfarin in patients with non-valvular atrial fibrillation and a history of cancer: observations from ROCKET AF. Eur Heart J Qual Care Clin Outcomes. 2019;5(2):145–152. doi: 10.1093/ehjqcco/qcy040 [DOI] [PubMed] [Google Scholar]
  • 17.Melloni C, Dunning A, Granger CB, et al. Efficacy and Safety of Apixaban Versus Warfarin in Patients with Atrial Fibrillation and a History of Cancer: Insights from the ARISTOTLE Trial. Am J Med. 2017;130(12):1440–1448.e1. doi: 10.1016/j.amjmed.2017.06.026 [DOI] [PubMed] [Google Scholar]
  • 18.Fanola CL, Ruff CT, Murphy SA, et al. Efficacy and Safety of Edoxaban in Patients With Active Malignancy and Atrial Fibrillation: Analysis of the ENGAGE AF - TIMI 48 Trial. J Am Heart Assoc. 2018;7(16):e008987. doi: 10.1161/JAHA.118.008987 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Laube ES, Yu A, Gupta D, et al. Rivaroxaban for Stroke Prevention in Patients With Nonvalvular Atrial Fibrillation and Active Cancer. Am J Cardiol. 2017;120(2):213–217. doi: 10.1016/j.amjcard.2017.04.009 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Russo V, Rago A, Papa AA, et al. Use of Non-Vitamin K Antagonist Oral Anticoagulants in Atrial Fibrillation Patients with Malignancy: Clinical Practice Experience in a Single Institution and Literature Review. Semin Thromb Hemost. 2018;44(4):370–376. doi: 10.1055/s-0037-1607436 [DOI] [PubMed] [Google Scholar]
  • 21.Ianotto JC, Couturier MA, Galinat H, et al. Administration of direct oral anticoagulants in patients with myeloproliferative neoplasms. Int J Hematol. 2017;106(4):517–521. doi: 10.1007/s12185-017-2282-5 [DOI] [PubMed] [Google Scholar]
  • 22.Shah S, Norby FL, Datta YH, et al. Comparative effectiveness of direct oral anticoagulants and warfarin in patients with cancer and atrial fibrillation. Blood Adv. 2018;2(3):200–209. doi: 10.1182/bloodadvances.2017010694 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Kim K, Lee YJ, Kim TH, et al. Effect of Non-vitamin K Antagonist Oral Anticoagulants in Atrial Fibrillation Patients with Newly Diagnosed Cancer. Korean Circ J. 2018;48(5):406–417. doi: 10.4070/kcj.2017.0328 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.SEER-Medicare: Brief Description of the SEER-Medicare Database. Available at https://healthcaredelivery.cancer.gov/seermedicare/overview/ Accessed February 20, 2021
  • 25.Gross CP, Galusha DH, Krumholz HM. The impact of venous thromboembolism on risk of death or hemorrhage in older cancer patients. J Gen Intern Med. 2007;22(3):321–326. doi: 10.1007/s11606-006-0019-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Jensen PN, Johnson K, Floyd J, Heckbert SR, Carnahan R, Dublin S. A systematic review of validated methods for identifying atrial fibrillation using administrative data. Pharmacoepidemiol Drug Saf. 2012;21 Suppl 1(0 1):141–147. doi: 10.1002/pds.2317 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Yao RJR, Andrade JG, Deyell MW, Jackson H, McAlister FA, Hawkins NM. Sensitivity, specificity, positive and negative predictive values of identifying atrial fibrillation using administrative data: a systematic review and meta-analysis. Clin Epidemiol. 2019;11:753–767. Published 2019 Aug 23. doi: 10.2147/CLEP.S206267 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Andrade SE, Harrold LR, Tjia J, et al. A systematic review of validated methods for identifying cerebrovascular accident or transient ischemic attack using administrative data. Pharmacoepidemiol Drug Saf. 2012;21 Suppl 1(Suppl 1):100–128. doi: 10.1002/pds.2312 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Coding Trend Analyses: Ischemic Stroke. Available at https://www.sentinelinitiative.org/sites/default/files/surveillance-tools/validations-literature/Ischemic_Stroke_Final_Trend_Report.pdf Accessed February 20, 2021
  • 30.MINI-SENTINEL MEDICAL PRODUCT ASSESSMENT. A PROTOCOL FOR ASSESSMENT OF DABIGATRAN Available at https://www.sentinelinitiative.org/sites/default/files/Drugs/Assessments/Mini-Sentinel_Protocol-for-Assessment-of-Dabigatran_0.pdf Accessed February 20, 2021
  • 31.Lipsitch M, Tchetgen E, Cohen T. Negative controls: a tool for detecting confounding and bias in observational studies. Epidemiology. 2010;21(3):383–388. doi: 10.1097/EDE.0b013e3181d61eeb [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Xian Y, Xu H, O'Brien EC, et al. Clinical Effectiveness of Direct Oral Anticoagulants vs Warfarin in Older Patients With Atrial Fibrillation and Ischemic Stroke: Findings From the Patient-Centered Research Into Outcomes Stroke Patients Prefer and Effectiveness Research (PROSPER) Study. JAMA Neurol. 2019;76(10):1192–1202. doi: 10.1001/jamaneurol.2019.2099 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Lip GY, Nieuwlaat R, Pisters R, Lane DA, Crijns HJ. Refining clinical risk stratification for predicting stroke and thromboembolism in atrial fibrillation using a novel risk factor-based approach: the euro heart survey on atrial fibrillation. Chest. 2010;137(2):263–272. doi: 10.1378/chest.09-1584 [DOI] [PubMed] [Google Scholar]
  • 34.Pisters R, Lane DA, Nieuwlaat R, de Vos CB, Crijns HJ, Lip GY. A novel user-friendly score (HAS-BLED) to assess 1-year risk of major bleeding in patients with atrial fibrillation: the Euro Heart Survey. Chest. 2010;138(5):1093–1100. doi: 10.1378/chest.10-0134 [DOI] [PubMed] [Google Scholar]
  • 35.Austin PC, Stuart EA. Moving towards best practice when using inverse probability of treatment weighting (IPTW) using the propensity score to estimate causal treatment effects in observational studies. Stat Med. 2015;34(28):3661–3679. doi: 10.1002/sim.6607 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Austin PC. Balance diagnostics for comparing the distribution of baseline covariates between treatment groups in propensity-score matched samples. Stat Med. 2009;28(25):3083–3107. doi: 10.1002/sim.3697 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Fine JP, Gray RJ. A Proportional Hazards Model for the Subdistribution of a Competing Risk. Journal of the American Statistical Association. 1999;94:446:496–509. DOI: 10.1080/01621459.1999.10474144 [DOI] [Google Scholar]
  • 38.Rossello X, González-Del-Hoyo M. Survival analyses in cardiovascular research, part I: the essentials. Rev Esp Cardiol (Engl Ed). 2022;75(1):67–76. doi: 10.1016/j.rec.2021.06.003 [DOI] [PubMed] [Google Scholar]
  • 39.Rossello X, González-Del-Hoyo M. Survival analyses in cardiovascular research, part II: statistical methods in challenging situations. Rev Esp Cardiol (Engl Ed). 2022;75(1):77–85. doi: 10.1016/j.rec.2021.07.001 [DOI] [PubMed] [Google Scholar]
  • 40.VanderWeele TJ, Ding P. Sensitivity Analysis in Observational Research: Introducing the E-Value. Ann Intern Med. 2017;167(4):268–274. doi: 10.7326/M16-2607 Chai-Adisaksopha C, Hillis C, Isayama T, Lim W, Iorio A, Crowther M. [DOI] [PubMed] [Google Scholar]
  • 41.Bruins Slot KM, Berge E. Factor Xa inhibitors versus vitamin K antagonists for preventing cerebral or systemic embolism in patients with atrial fibrillation. Cochrane Database Syst Rev. 2018;3(3):CD008980. Published 2018 Mar 6. doi: 10.1002/14651858.CD008980.pub3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Graham DJ, Reichman ME, Wernecke M, et al. Cardiovascular, bleeding, and mortality risks in elderly Medicare patients treated with dabigatran or warfarin for nonvalvular atrial fibrillation. Circulation. 2015;131(2):157–164. doi: 10.1161/CIRCULATIONAHA.114.012061 [DOI] [PubMed] [Google Scholar]
  • 43.Alberts M, Chen YW, Lin JH, Kogan E, Twyman K, Milentijevic D. Risks of Stroke and Mortality in Atrial Fibrillation Patients Treated With Rivaroxaban and Warfarin. Stroke. 2020;51(2):549–555. doi: 10.1161/STROKEAHA.119.025554 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Chai-Adisaksopha C, Hillis C, Isayama T, Lim W, Iorio A, Crowther M. Mortality outcomes in patients receiving direct oral anticoagulants: a systematic review and meta-analysis of randomized controlled trials. J Thromb Haemost. 2015;13(11):2012–2020. doi: 10.1111/jth.13139 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplemental Publication Material File
NIHMS1846109-supplement-Supplemental_Publication_Material_File.pdf (591.6KB, pdf)
STROBE Checklist
NIHMS1846109-supplement-STROBE_Checklist.pdf (102.6KB, pdf)

RESOURCES