Authors: Nipun Atreja, Anandkumar Dubey, Amiee Kang, Jenny Jiang, Melissa Hagan, Abimbola Michael-Asalu, Dong Cheng, Steven Deitelzweig
Categories: Original Research, Apixaban, Atrial fibrillation, Dabigatran, Direct oral anticoagulants, Rivaroxaban, Switch, Stroke, Systemic embolism, Warfarin
Source: Advances in Therapy
Authors: Nipun Atreja, Anandkumar Dubey, Amiee Kang, Jenny Jiang, Melissa Hagan, Abimbola Michael-Asalu, Dong Cheng, Steven Deitelzweig
Atrial fibrillation (AF), a common heart rhythm abnormality, is linked to a higher risk of stroke. Traditionally, warfarin has been the primary anticoagulation treatment for reducing the stroke risk. The new standard of treatment by direct oral anticoagulants (DOACs) offers greater benefits including improved efficacy and fewer adverse effects with reduced monitoring. This study aims to evaluate the risk of stroke/systemic embolism (SE) and major bleeding (MB) among patients with AF who switched from warfarin to DOACs.
This study utilized Medicare data to conduct a retrospective analysis of patients with non-valvular atrial fibrillation (NVAF) who switched from warfarin to DOACs between January 1, 2012, and December 31, 2019. Patients with NVAF aged 65 and older who switched from warfarin and had continuous health plan enrollment were included. Descriptive statistics, propensity score matching (PSM), and Cox proportional hazard (PH) models were utilized to compare the outcomes and assess risks of SE and MB across the DOAC cohorts.
Among 1,843,495 patients with NVAF on warfarin, 171,700 switched to DOACs within 90 days of discontinuation (apixaban: 90,850; 67,698; 12,900). The mean follow-up period across DOAC cohorts ranged from 552 to 628 days. After PSM, apixaban showed significantly lower rates of stroke/SE compared to dabigatran (2.99% vs. 3.98%, p < 0.0001) and rivaroxaban (3.08% vs. 3.80%, p < 0.0001). MB rates were also lower with apixaban versus dabigatran (4.29% vs. 5.57%, p < 0.0001) and rivaroxaban (4.07% vs. 6.35%, p < 0.0001). Cox PH models confirmed these findings, with apixaban demonstrating lower risks of stroke/SE [hazard ratio (HR) 0.83, 95% confidence interval (CI) 0.72–0.96 vs. dabigatran; HR 0.91, 95% CI 0.85–0.96 vs. rivaroxaban] and MB (HR 0.79, 95% CI 0.71–0.89 vs. dabigatran; HR 0.68, 95% CI 0.65–0.72 vs. rivaroxaban).
The risk of stroke/SE and MB varies significantly among patients with NVAF switching from warfarin to different DOACs, with apixaban presenting the lowest risk compared to dabigatran and rivaroxaban.
The online version contains supplementary material available at 10.1007/s12325-024-03099-y.
***Why carry out this study?***Traditionally, warfarin has been the primary treatment for anticoagulation.However, warfarin has a narrow therapeutic window and requires regular monitoring and, thus, warrants use of alternatives like direct oral anticoagulants (DOACs).This study evaluated the risk of stroke/systemic embolism (SE) and major bleeding (MB) amongst patients with non-valvular atrial fibrillation (NVAF) who switched from warfarin to DOACs.***What was learned from the study?***The risk of stroke/SE and MB varies significantly among patients with NVAF switching from warfarin to different DOACs.Apixaban demonstrated lower risks of stroke/SE and MB versus dabigatran and rivaroxaban.
Atrial fibrillation (AF) [1] is the most common abnormality of heart rhythm, characterized by a fast and irregular heartbeat, with underlying etiology that includes medical conditions such as hypertension, sleep apnea, hyperthyroidism, alcohol abuse, etc. [1–4]. The incidence and prevalence of AF is increasing globally and approaching epidemic proportions, especially in western countries due to its association with increasing age and its propensity for causing stroke and myocardial infarction. In the United States, approximately 3–6 million patients have AF, a number projected to increase to 6–16 million by the year 2050. Non-valvular atrial fibrillation (NVAF) is the most prevalent type of AF, characterized as AF without moderate-to-severe mitral stenosis or a mechanical heart valve, accounting for up to 95% of all AFs diagnosed in the United States and associated with a 5 times higher risk of ischemic strokes [1].
Historically, vitamin K antagonists such as warfarin was the treatment of choice for anticoagulation, preventing thromboembolic events for patients with AF [5, 6]. However, the use of warfarin has suffered from limitations such as a narrow therapeutic window, numerous drug–drug and diet–drug interactions, and regular blood-level monitoring. Warfarin is thus associated with increased risks of thromboembolism or bleeding with either under- or over-treatment, respectively [5–7]. The need for regular monitoring, risk of hemorrhage/stroke, and low portions of time spent in the therapeutic international normalized ratio range have led to a low adherence to medication [8, 9]. Other factors that have also been associated with low adherence to warfarin include education level, mental health functioning/psychosocial factors (e.g., depressive symptoms, cognitive impairment, and health-related quality of life), and employment status [9].
Direct oral anticoagulants (DOACs), such as dabigatran, rivaroxaban, apixaban, and edoxaban, have become prominent alternatives to the long-standing warfarin standard of care in the prevention of thromboembolic events in NVAF. Critical registrational randomized clinical trials have demonstrated superiority or comparability to warfarin in reducing the risk of stroke/systemic embolism (SE) with reduced risk of bleeding among patients with NVAF leading to the approval of dabigatran in 2010, the first DOAC in the United States (US), and others in subsequent years [1, 10–12]. Additional advantages of DOACs include reduced monitoring requirements, fewer follow-ups, faster drug onset and offset effects (important for peri-operational procedures and acute bleeding management), and fewer drug–drug and diet–drug interactions [13–15].
A recent real-world data (RWD) study compared patients with NVAF who were on warfarin treatment but switched to DOACs (dabigatran, rivaroxaban, and apixaban) using an adjudicated claims database of commercially insured individuals from 2010 to 2015. The researchers found that switching from warfarin to DOAC was associated with fewer inpatient, emergency room (ER), and outpatient visits, and lower non-drug costs among patients with NVAF [4, 16]. In addition, lower or comparable healthcare expenditure and/or healthcare utilization for DOAC patients compared with patients on warfarin have been reported in other RWD studies [4, 17, 18]. Another recent RWD study deployed retrospective analyses of effectiveness and safety profile on patients with NVAF switched from warfarin to apixaban, rivaroxaban, and dabigatran using a US commercial claims database. This analysis revealed that apixaban was associated with better or similar effectiveness and reduced risk in major bleeding (MB) compared with rivaroxaban and dabigatran [19].
Given the exponential increased prevalence of AF with advancing age, and the fact that strokes from AF account for about 30% of ischemic strokes in adults older than 65 years [20], this study aimed to evaluate the risks of stroke/SE and MB between DOACs among Medicare patients who switched to DOACs.
This study utilized Centers for Medicare and Medicaid Service (CMS) Medicare data. This database contains medical and pharmacy claims from 100% national Medicare data, which includes Medicare Parts A–D and claim level information like beneficiary demographic information, International Classification of Diseases (ICD) diagnosis, procedure, CMS Healthcare Common Procedure Coding System (HCPCS) codes, diagnosis-related groups, dates of service, reimbursement amounts, provider numbers, etc. It amounts to approximately 38 million fee-for-service beneficiaries.
A retrospective study design was utilized to identify patients with NVAF that switched from warfarin to a DOAC (apixaban, dabigatran, rivaroxaban) from January 1, 2013 to December 31, 2019. The study designated a 12-month baseline period prior to the identification period (January 1, 2012 to December 31, 2019) while the follow-up period was designated from the day after the index date to death, end of study period, discontinuation, switch, or end of continuous medical and pharmacy enrollment, whichever occurred first. Index date was defined as the first DOAC prescription date within 90 days of warfarin discontinuation (Fig. 1).Fig. 1Study periods (for illustration purposes only, may not be proportional). Bars and lengths of bars are not proportional to time periods. OAC oral anticoagulant
Switching was defined as having ≥ 1 pharmacy claims for warfarin and ≥ 1 pharmacy claims for a DOAC within 90 days after warfarin treatment ended during the identification period. In the primary analysis, only the first switching episode from warfarin to a DOAC was considered for each patient to ensure consistency in evaluating treatment outcomes. Treatment end date was defined as end of days of supply. To ensure consistency in evaluating outcomes and to avoid potential biases from including multiple episodes, only the first episode of switching from warfarin to DOAC was considered for each patient in the analysis. Treatment episodes were defined as the time from the DOAC episode start date to the earliest of DOAC discontinuation, switch, end of study period, or death disenrollment (multiple episodes per patient were allowed, but only the first was considered). Discontinuation was defined as no evidence of index DOAC treatment use for 30 days from the last day of the days’ supply. The discontinuation date was the last day of the days’ supply of the last filled prescription. The follow-up period was censored at 30 days after the index drug discontinuation date.
Patient eligibility was based on selection from the CMS Medicare database according to the following criteria. Inclusion was based on patients having AF diagnosis (ICD-9-CM: 427.31; ICD-10-CM: I480-I482, I4891) with at least 1 pharmacy claim for a warfarin or DOAC during the identification period at any time during the identification period. Patients were allowed to have multiple medications. For patients and episodes where patients were ≥ 65 years of age on the initial oral anticoagulant (OAC) prescription date (treatment date), and for; patients and episodes where patients had a continuous health plan enrollment with medical and pharmacy benefits for ≥ 12 months pre-treatment date (baseline period), no gap was allowed. Patients and episodes where patients had ≥ 1 medical claim for AF prior to or on the treatment date were all included in the study population.
All pregnant patients or those with no follow-up period were excluded from the study. Also excluded patients who had episodes of medical claims indicating a diagnosis or procedure code indicative of rheumatic mitral valvular heart disease or valve replacement procedure during the 12 months prior to or on the treatment date; episodes of a diagnosis of venous thromboembolism during the 12 months prior to or on the treatment; episodes of medical claims indicating hip/knee replacement surgery within 6 weeks prior to the OAC prescription date; and episodes of medical claims indicating a diagnosis or procedure code of transient AF (heart valve replacement/transplant, pericarditis, hyperthyroidism, and thyrotoxicity) during the 12 months prior to or on the OAC prescription date. Episodes with > 1 OAC treatment on the index date were also excluded, while patients on edoxaban were excluded from analyses due to the small sample size.
To evaluate switching among patients with NVAF from warfarin, three cohorts were switch to apixaban, to dabigatran, and to rivaroxaban. Cohorts used to assess the risks of stroke/SE, MB between DOACs among patients who discontinued warfarin apixaban versus dabigatran, apixaban versus rivaroxaban, and dabigatran versus rivaroxaban.
Patient demographics included age, sex, US geographic region, Medicaid dual eligibility and part B low-income subsidy. The Deyo–Charlson–Quan Comorbidity Index [21], CHADS2 score, CHA2DS2-VASc score [22], and HAS-BLED scores [23] were calculated using standard methods. Additionally, several comorbidities beyond the CCI (Charlson Comorbidity Index) were included, such as liver disease, renal disease, prior bleeding history, hypertension, dyspepsia, non-stroke/SE peripheral vascular disease, stroke/SE, transient ischemic attack, anemia and coagulation defects, peripheral arterial disease, coronary artery disease, etc.
Lifestyle factors or exposures such as alcohol use disorder were captured and included in the comorbidities. All comorbidities and lifestyle factors were captured using ICD-9-CM or ICD-10-CM codes on at least one inpatient or outpatient claim during the 12-month pre-index period.
Stroke/SE was identified during the follow-up period using hospital claims that included a stroke/SE diagnosis code as the primary listed ICD-9-CM or ICD-10-CM diagnosis code, and equaled 1 if there was ≥ 1 stroke event during the follow-up period. The first stroke/SE event was stratified into 3 ischemic stroke, hemorrhagic stroke, and SE.
MB was identified in the follow-up period using hospital claims that included a MB diagnosis code as the primary listed ICD-9-CM or ICD-10-CM diagnosis code, and equaled 1 if there was ≥ 1 MB event during the follow-up period. The first MB event was stratified into 3 gastrointestinal hemorrhage, intracranial hemorrhage, and other MB.
Time to stroke/SE or MB was calculated from the day after the index date to the date of stroke/SE or MB. The incidence rate of stroke/SE or MB was calculated per 100 person-years.
Descriptive analyses were conducted for baseline demographic and clinical factors, such as CHA2DS2-VASc and HAS-BLED scores, concomitant medications, and comorbidities across cohorts. Means, standard deviations, median, and inter-quartile range for continuous variables and differences across cohorts were calculated for continuous variables. Frequency and percentages were produced for categorical variables. Differences between groups were compared using t tests for continuous variables, and Chi-square tests for categorical variables. All variables were analyzed descriptively for each drug (apixaban, dabigatran, and rivaroxaban). Numbers and percentages were provided for dichotomous and polychotomous variables. Differences between cohorts (apixaban vs. dabigatran, apixaban vs. rivaroxaban, and dabigatran vs. rivaroxaban) were assessed using McNemar’s test for categorical variables and pairwise t tests for continuous variables.
The 1 propensity score matching (PSM) method was used to adjust the baseline patient differences in each pair. Propensity scores were estimated by unconditional logistic regression analyses which incorporated potential predictors of treatment as independent variables in the regression and group status. The propensity score was defined as the probability of being treated with each of the DOACs, based on a set of baseline characteristics in the DOAC cohort. All baseline variables were evaluated as covariates to be matched between treatments using the PSM technique and as covariates to be included in multivariate models.
Cox proportional hazards (PH) models were used to compare the time at risk for stroke/SE and MB between apixaban versus dabigatran, apixaban versus rivaroxaban, and dabigatran versus rivaroxaban cohorts. The proportional hazards assumption was met for most of the models. In instances where the assumption was violated, interaction terms with time were included in the model to account for non-proportional hazards. These adjustments were made in the analysis to ensure that the reported hazard ratios reflect the time-varying nature of the relationship between the covariates and the outcomes where necessary.
All statistical tests were two-sided, with α < 0.05 and performed using SAS software (version 9.4; SAS Institute, Cary, NC, USA).
Since this study did not involve the collection, use, or transmittal of individually identifiable data, Institutional Review Board approval was not required. This study only analyzed de-identified data which are a priori exempt from the Federal Policy for the Protection of Human Subjects (1991) and do not meet the identification criteria necessary to be privileged under the Health Insurance Portability and Accountability Act (HIPAA). Both the data and the security of the offices where the data were kept met HIPAA requirements. All authors had permission to access and use the database utilized in this study.
In this study, 171,700 patients had a DOAC treatment episode within 90 days of discontinuing warfarin (apixaban = 90,850, dabigatran = 12,900, and rivaroxaban = 67,698). (Supplementary Material Tables 1, 2). Following PSM, there were 12,900 apixaban–dabigatran, 67,578 apixaban–rivaroxaban, and 12,891 dabigatran–rivaroxaban matched pairs. Edoxaban was excluded from further analyses due to the small sample size (Fig. 2). Table 1Post-propensity score matching baseline characteristics of patients prescribed apixaban, dabigatran, or rivaroxaban among patients who switched from warfarinPatients who switch from warfarinApixaban cohort (reference) (n = 12,900)Dabigatran cohort (n = 12,900)STD^b^Apixaban cohort (reference) (n = 67,578)Rivaroxaban cohort (n = 67,578)STDRivaroxaban cohort (reference) (n = 12,891)Dabigatran cohort (n = 12,891)STDAge, mean (SD)^a^80.7 (7.1)79.3 (7.2)19.5480.6 (7.16)80.1 (7.2)7.1979.8 (7.18)79.3 (7.2)6.69Age categories, n (%) 65–742823 (21.88)3724 (28.87)16.115,529 (22.98)17,000 (25.16)5.093429 (26.6)3724 (28.89)5.11 75–792916 (22.6)3039 (23.56)2.2615,538 (22.99)15,514 (22.96)0.082958 (22.95)3036 (23.55)1.43 ≥ 807161 (55.51)6137 (47.57)15.9336,511 (54.03)35,064 (51.89)4.296504 (50.45)6131 (47.56)5.79Sex, n (%) Male5659 (43.87)5533 (42.89)1.9729,282 (43.33)29,391 (43.49)0.335637 (43.73)5527 (42.87)1.72 Female7241 (56.13)7367 (57.11)1.9738,296 (56.67)38,187 (56.51)0.337254 (56.27)7364 (57.13)1.72US geographic region, n (%) Northeast2924 (22.67)2781 (21.56)2.6714,101 (20.87)13,714 (20.29)1.422667 (20.69)2779 (21.56)2.13 Midwest2832 (21.95)2837 (21.99)0.0915,246 (22.56)15,201 (22.49)0.162883 (22.36)2836 (22)0.88 South4793 (37.16)4717 (36.57)1.2225,331 (37.48)25,407 (37.6)0.234862 (37.72)4715 (36.58)2.36 West2344 (18.17)2544 (19.72)3.9612,817 (18.97)13,153 (19.46)1.262451 (19.01)2540 (19.7)1.75 Other7 (0.05)21 (0.16)3.383 (0.12)103 (0.15)0.828 (0.22)21 (0.16)1.25Medicaid dual eligibility, n (%)2389 (18.52)3383 (26.22)18.5712,976 (19.2)17,680 (26.16)16.683345 (25.95)3380 (26.22)0.62Part D low income subsidy, n (%)3218 (24.95)4363 (33.82)19.5817,266 (25.55)22,241 (32.91)16.244234 (32.84)4360 (33.82)2.07Race, n (%) Non-Hispanic white11,544 (89.49)11,114 (86.16)10.2160,339 (89.29)59,092 (87.44)5.7611,325 (87.85)11,107 (86.16)5.03 Non-Hispanic Black652 (5.05)788 (6.11)4.593451 (5.11)3847 (5.69)2.59712 (5.52)787 (6.11)2.49 Other704 (5.46)998 (7.74)9.193788 (5.61)4639 (6.86)5.21854 (6.62)997 (7.73)4.3 Deyo–Charlson comorbidity index, mean (SD)3.1 (2.8)2.8 (2.6)13.172.9 (2.7)2.8 (2.6)3.92.8 (2.6)2.7 (2.6)0.58 Deyo–Charlson comorbidity score categories, n (%) 02626 (20.36)2608 (20.22)0.3514,707 (21.76)14,305 (21.17)1.452731 (21.19)2607 (20.22)2.37 11899 (14.72)2357 (18.27)9.5810,950 (16.2)11,839 (17.52)3.512309 (17.91)2356 (18.28)0.95 21966 (15.24)2122 (16.45)3.3110,543 (15.6)10,886 (16.11)1.392107 (16.34)2120 (16.45)0.27 3 + 6409 (49.68)5813 (45.06)9.2631,378 (46.43)30,548 (45.2)2.475744 (44.56)5808 (45.05)1 CHA2DS2-VASc Score, mean (SD)4.6 (1.8)4.5 (1.8)4.214.5 (1.77)4.5 (1.8)1.464.5 (1.78)4.5 (1.8)2.35 CHA2DS2-VASc score categories, n (%) 00 (0)0 (0)00 (0)0 (0)00 1248 (1.92)333 (2.58)4.441420 (2.1)1496 (2.21)0.77296 (2.3)333 (2.58)1.86 21231 (9.54)1262 (9.78)0.816823 (10.1)6825 (10.1)0.011343 (10.4)1262 (9.79)2.08 32364 (18.33)2427 (18.81)1.2613,133 (19.4)13,238 (19.59)0.392541 (19.7)2423 (18.8)2.32 4 + 9057 (70.21)8878 (68.82)3.0146,202 (68.4)46,019 (68.1)0.588711 (67.6)8873 (68.83)2.7 HAS-BLED score, mean (SD)3 (1.4)2.9 (1.3)5.852.85 (1.3)2.8 (1.3)1.462.84 (1.3)2.9 (1.3)3.1 HAS-BLED score categories, n (%) 00 (0)0 (0)00 (0)0 (0)00 (0)0 (0)0 12077 (16.1)1984 (15.38)1.9811,466 (17)11,137 (16.48)1.32113 (16.4)1982 (15.38)2.78 23138 (24.33)3473 (26.92)5.9517,915 (26.5)18,563 (27.47)2.163558 (27.6)3472 (26.93)1.5 3+ 7685 (59.57)7443 (57.7)3.8138,197 (56.5)37,878 (56.05)0.957220 (56)7437 (57.69)3.4 Bleeding history, n (%)3172 (24.59)2861 (22.18)5.715,009 (22.2)14,294 (21.15)2.572729 (21.2)2857 (22.16)2.41 Major bleeding hospitalization (primary), n (%)516 (4)445 (3.45)2.911958 (2.9)1680 (2.49)2.54314 (2.4)444 (3.44)5.97 Bleeding within 90 days before the index date, n (%)1679 (13.02)1483 (11.5)4.637423 (11)6747 (9.98)3.271295 (10)1481 (11.49)4.66 Comorbidities, n (%) Congestive heart failure4777 (37.03)4409 (34.18)5.9623,641 (35)22,803 (33.74)2.614335 (33.6)4405 (34.17)1.15 Diabetes mMellitus4494 (34.84)4846 (37.57)5.6823,739 (35.1)24,420 (36.14)2.14708 (36.5)4842 (37.56)2.15 Hypertension9255 (71.74)9213 (71.42)0.7247,646 (70.5)47,574 (70.4)0.239047 (70.2)9206 (71.41)2.71 Renal disease3471 (26.91)2426 (18.81)19.3815,216 (22.5)13,507 (19.99)6.182497 (19.4)2423 (18.8)1.46 Liver disease495 (3.84)478 (3.71)0.692501 (3.7)2441 (3.61)0.47457 (3.5)478 (3.71)0.87 Myocardial infarction1371 (10.63)1121 (8.69)6.566219 (9.2)5887 (8.71)1.721179 (9.1)1120 (8.69)1.61 Dyspepsia or stomach discomfort2355 (18.26)2385 (18.49)0.611,971 (17.7)12,183 (18.03)0.822381 (18.5)2382 (18.48)0.02 Non-stroke/ SE peripheral vascular disease2751 (21.33)2593 (20.1)3.0214,399 (21.3)14,049 (20.79)1.272629 (20.4)2591 (20.1)0.73 Stroke/SE3419 (26.5)3486 (27.02)1.1716,608 (24.6)16,783 (24.84)0.63237 (25.1)3484 (27.03)4.37 Stroke/SE hospitalization789 (6.12)901 (6.98)3.513782 (5.6)3480 (5.15)1.98671 (5.2)901 (6.99)7.46 Non-stroke/SE hospitalization during baseline953 (7.39)824 (6.39)3.954559 (6.7)4400 (6.51)0.95860 (6.7)822 (6.38)1.19 Stroke within 90 days before the index date2072 (16.06)2064 (16)09770 (14.5)9241 (13.67)2.251763 (13.7)2063 (16)6.55 Transient ischemic attack1454 (11.27)1211 (9.39)6.197007 (10.4)5667 (8.39)6.811075 (8.3)1210 (9.39)3.69 Anemia and coagulation Defects4558 (35.33)4170 (32.33)6.3622,279 (33)21,835 (32.31)1.44185 (32.5)4164 (32.3)0.35 Alcoholism213 (1.65)205 (1.59)0.49917 (1.4)893 (1.32)0.31178 (1.4)205 (1.59)1.73 Peripheral artery disease2668 (20.68)2583 (20.02)1.6413,993 (20.7)13,982 (20.69)0.042635 (20.4)2581 (20.02)1.04 Coronary artery disease5117 (39.67)4953 (38.4)2.6125,658 (38)25,449 (37.66)0.645010 (38.9)4948 (38.38)0.99Baseline medication use, n (%) ACE^d^/ARB^e^6361 (49.31)7027 (54.47)10.3533,478 (49.54)35,004 (51.8)4.526741 (52.29)7024 (54.49)4.4 Amiodarone1629 (12.63)1431 (11.09)4.758394 (12.42)7595 (11.24)3.661484 (11.51)1429 (11.09)1.35 Beta blockers9946 (77.1)9775 (75.78)3.1251,779 (76.62)50,249 (74.36)5.279634 (74.73)9767 (75.77)2.39 H2-receptor antagonist^f^1208 (9.36)1266 (9.81)1.536298 (9.32)6297 (9.32)0.011230 (9.54)1266 (9.82)0.94 Proton pump inhibitor4600 (35.66)4662 (36.14)123,915 (35.39)24,071 (35.62)0.484635 (35.96)4658 (36.13)0.37 Statins8381 (64.97)8413 (65.22)0.5244,593 (65.99)43,160 (63.87)4.448205 (63.65)8405 (65.2)3.24 Antiplatelets1208 (9.36)1192 (9.24)0.436277 (9.29)5860 (8.67)2.161166 (9.05)1192 (9.25)0.7 NSAIDs^g^1603 (12.43)1647 (12.77)1.038409 (12.44)8277 (12.25)0.591610 (12.49)1646 (12.77)0.84Dose of the index prescription, n (%) Standard dose (5 mg apixaban, 150 mg dabigatran, 20 mg Rivaroxaban)8466 (65.63)9714 (75.3)21.3345,060 (66.68)43,791 (64.8)3.968487 (65.84)9707 (75.3)20.88 Low dose (2.5 mg apixaban, 75 mg dabigatran, 15 mg rivaroxaban)4450 (34.5)3083 (23.9)23.4722,608 (33.45)20,554 (30.42)6.523795 (29.44)3082 (23.91)12.53 Other dose (rivaroxaban 10 mg, dabigatran 110 mg)0 (0)115 (0.89)13.410 (0)3,594 (5.32)33.52691 (5.36)114 (0.88)25.95Events during the baseline, n (%) Stroke/SE hospitalization (primary discharge)789 (6.12)901 (6.98)3.513782 (5.6)3480 (5.15)1.98671 (5.21)901 (6.99)7.46 Stroke/SE hospitalization during the 90 days before the index date (primary discharge)480 (3.72)510 (3.95)1.212292 (3.39)1894 (2.8)3.4364 (2.82)510 (3.96)6.26 Time from stroke-to-index date103 (96.82)118 (100.38)15.47106 (98.94)120 (103.5)13.32123 (105.88)118 (100.38)4.83 Major bleed hospitalization (primary discharge)516 (4)445 (3.45)2.911958 (2.9)1680 (2.49)2.54314 (2.44)444 (3.44)5.97 Major bleed hospitalization during the 90 days before the index date (primary discharge)312 (2.42)281 (2.18)1.61145 (1.69)874 (1.29)3.31164 (1.27)280 (2.17)6.92 Time from MB^h^-to-index date110 (98.01)106 (100.34)4.34111 (100.1)127 (106.82)15.39126 (105.71)106 (100.44)19.79Gap length between warfarin discontinuation to NOAC^I^ initiation, mean (SD)7.0 (18)5.0 (15)12.116.0 (17)5.0 (15)6.565.0 (15)5.0 (15)1.65 Min111111 Quartile 1111111 Median111111 Quartile 3111111 Max909090909090Length of warfarin therapy, mean (SD)394.7 (424.3)326.7 (393)16.63443.7 (499.9)419.3 (483.6)4.96333.0 (373)325.5 (390.3)1.91^a^SD standard deviation^b^STD standardized mean difference = 100 times the actual standardized difference; STD greater than 10 is considered significant^c^Systemic embolism^d^Angiotensin-converting enzyme inhibitors^e^Angiotensin receptor blockers^f^H2-receptor histamine 2 receptor^g^NSAIDs nonsteroidal anti-inflammatory drugs^h^Major bleeding^I^Non-vitamin K antagonist oral anticoagulantsTable 2Post-propensity score matching outcome descriptive table of patients prescribed apixaban, dabigatran, or rivaroxaban who switched from warfarinPatients who switched from warfarinApixaban Cohort (Reference) n = 12,900dabigatran Cohort n = 12,900P valueSTD^b^Apixaban Cohort (Reference) n = 67,578Rivaroxaban Cohort n = 67,578P valueSTDRivaroxaban Cohort (Reference) n = 12,891Dabigatran Cohort n = 12,891P valueSTD^c^Follow-up time (in days), mean (SD^a^)551.6 (522.9)581.6 (616.48) < 0.00015.25550.2 (520.52)613.7 (607.25) < 0.000111.23628.2 (621.87)581.7 (616.61) < 0.00017.51 Minimum111111 Quartile 11529015213290132 Median379352379411353415 Quartile 3793881788903881931 Maximum2,5112,5482,5112,5542,5482,542Reasons for censoring, in follow-upDiscontinuation, n (%)3497 (27.11)3828 (29.67) < 0.00015.6917,874 (26.45)18,973 (28.08) < 0.00013.653766 (29.21)3828 (29.7)0.3971.06 Time-to-discontinuation among patients that discontinued, mean (SD)424.7 (374)447.9 (420.2)0.01255.83433.4 (379.4)472.1 (420.4) < 0.00019.65485.8 (432.5)447.9 (420.2)0.00018.9Switch, n (%)1735 (13.45)4451 (34.5) < 0.000150.889446 (13.98)15,764 (23.33) < 0.000124.183008 (23.33)4449 (34.51) < 0.000124.84 Apixaban0 (0)1202 (9.32) < 0.000145.330 (0)3922 (5.8) < 0.000135.1722 (5.6)1201 (9.32) < 0.000114.18 Dabigatran58 (0.45)0 (0) < 0.00019.5353 (0.52)572 (0.85) < 0.00013.93120 (0.93)0 (0) < 0.000113.71 Edoxaban5 (0.04)4 (0.03)0.73880.4215 (0.02)23 (0.03)0.19430.717 (0.05)4 (0.03)0.36561.13 Rivaroxaban302 (2.34)839 (6.5) < 0.000120.351657 (2.45)0 (0) < 0.000122.420 (0)839 (6.51) < 0.000137.31 Warfarin1372 (10.64)2408 (18.67) < 0.000122.867425 (10.99)11,250 (16.65) < 0.000116.462161 (16.76)2407 (18.67) < 0.00015 Time-to-switch among patients that switched, mean (SD)156.4 (256.84)265.9 (416.69) < 0.000131.64153.0 (249.12)213.2 (343.45) < 0.000120.05213.4 (337.46)266.0 (416.77) < 0.000113.87Disenrollment, n (%)1972 (15.29)2341 (18.15) < 0.00017.679945 (14.72)13,291 (19.67) < 0.000113.152516 (19.52)2340 (18.15)0.00513.49Death, n (%)133 (1.03)278 (2.16) < 0.00018.99660 (0.98)1194 (1.77) < 0.00016.8264 (2.05)278 (2.16)0.54330.76Stroke/SE^d^ (primary discharge), n (%)332 (2.57)420 (3.26)0.00114.11816 (2.69)2193 (3.25) < 0.00013.3384 (2.98)420 (3.26)0.19711.61 Time to stroke (in days), mean (SD)361.02 (470.12)427.33 (521.87)0.043613.4351.6 (452.5)457.9 (491.1)0.000322.5477.9 (508.3)427.3 (521.9)0.16429.8 Hemorrhagic stroke39 (0.3)43 (0.33)0.65820.55219 (0.32)291 (0.43)0.00141.7442 (0.33)43 (0.33)0.91350.14 Ischemic stroke278 (2.16)338 (2.62)0.01443.051491 (2.21)1751 (2.59) < 0.00012.51317 (2.46)338 (2.62)0.40591.04 Systemic embolism21 (0.16)48 (0.37)0.00114.05142 (0.21)188 (0.28)0.01121.3830 (0.23)48 (0.37)0.04122.54Major bleeding (primary discharge), n (%)554 (4.29)719 (5.57) < 0.00015.912748 (4.07)4292 (6.35) < 0.000110.3756 (5.86)719 (5.58)0.32111.24 Gastrointestinal bleeding352 (2.73)546 (4.23) < 0.00018.211784 (2.64)3011 (4.46) < 0.00019.83537 (4.17)546 (4.24)0.77990.35 Intracranial hemorrhage138 (1.07)111 (0.86)0.08552.14641 (0.95)785 (1.16)0.00012.09128 (0.99)111 (0.86)0.26931.38 Other sites72 (0.56)75 (0.58)0.8040.31391 (0.58)596 (0.88) < 0.00013.56109 (0.85)75 (0.58)0.01193.13Stroke/SE time at risk (years)1.481.551.471.641.681.55 Hemorrhagic stroke1.511.591.51.681.721.59 Ischemic stroke1.491.571.481.651.71.57 Systemic embolism1.511.591.51.681.721.59Major bleeding time at risk (years)1.471.531.461.61.651.53 Gastrointestinal bleeding1.481.541.481.631.671.54 Intracranial hemorrhage1.51.581.51.671.711.58 Other sites1.511.591.51.671.711.59Stroke/SE incidence rate (per 100 person-years)1.742.091.822.01.762.09 Hemorrhagic stroke0.20.210.220.260.190.21 Ischemic stroke1.451.671.491.571.451.67 Systemic embolism0.110.230.140.170.140.23Major bleeding incidence rate (per 100 person-years)2.933.662.783.973.573.66 Gastrointestinal bleeding1.852.751.792.742.512.75 Intracranial hemorrhage0.710.540.630.70.580.54 Other sites0.370.370.390.530.490.37^a^SD standard deviation^b^STD standardized mean difference = 100 times the actual standardized difference; STD greater than 10 is considered significant^c^Some patients were counted more than once as they were under different oral anticoagulants (OACs) and regimens; therefore, the totals for number of patients will not add up^d^Systemic embolismFig. 2Patient selection criteria. AF atrial fibrillation, DOAC direct-acting oral anticoagulant, ICD-10-CM International Classification of Diseases, Tenth Revision, Clinical Modification, ICD-9-CM International Classification of Diseases, Ninth Revision, Clinical Modification, NVAF non-valvular atrial fibrillation, OAC oral anticoagulant, PSM propensity score matched, VTE venous thromboembolism
After applying PSM, patient demographics were similar across matched cohorts (Table 1). The average age for the matched cohorts was approximately 80 years of age. All matched cohorts were more likely to be female (> 56% for all cohorts) and reside in the South US region. The most prevalent baseline comorbidities among all matched cohorts were hypertension, diabetes mellitus, coronary artery disease, congestive heart failure, and anemia and coagulation defects. At baseline, the most common medications taken were ACEs/ARBs, beta blockers, and statins.
The mean CHA2DS2-VASc score was ~ 4.5 and the mean HAS-BLED score was ~ 3 for all cohorts. Baseline stroke/SE hospitalization events ranged from 5% (apixaban–rivaroxaban) to 7% (apixaban–dabigatran and rivaroxaban–dabigatran). Baseline MB hospitalization events ranged from 2% (rivaroxaban–dabigatran) to 4% (apixaban–dabigatran and apixaban–rivaroxaban). Patients generally initiated DOAC within 7 days of discontinuing warfarin (Table 1).
After PSM, patients on apixaban were less likely to have a stroke/SE overall compared to patients on dabigatran (2.99% vs. 3.98%, p < 0.0001; Table 2) and were less likely to have an ischemic stroke or SE. Their time at risk for stroke was also shorter compared to the dabigatran cohort.
Patients on apixaban were less likely to experience MB overall compared to patients on dabigatran (4.29% vs. 5.57%, p < 0.0001) and were significantly less likely to experience gastrointestinal (GI) bleeding. Both cohorts had similar likelihood to experience bleeding at other sites. Time at risk for MB was shorter for those in the apixaban cohort versus the dabigatran cohort. Additionally, the overall MB incidence rate was lower for the apixaban cohort compared to the dabigatran cohort. The incidence rate of apixaban for GI bleeding was lower but the incidence rate for intracranial hemorrhage (ICH) was higher compared with dabigatran. The incidence rate for bleeding from other sites was the same.
Cox PH model analysis revealed that apixaban was associated with lower risks of stroke/SE and MB compared with dabigatran after ensuring that the PH assumption was met. Within the stroke/SE category, apixaban was associated with a lower risk of SE and similar risk of hemorrhagic stroke, and ischemic stroke compared with dabigatran. Within the MB category, apixaban was associated with a lower risk of GI bleeding, similar risk of ICH, and similar risk of bleeding at other sites compared with dabigatran (Fig. 3).Fig. 3Hazard ratios of stroke/SE and major bleeding for PSM patients. CI confidence interval, PSM propensity score matching, IR incidence rate (per 100 person-years), Ref reference group, SE systemic embolism
Following PSM, patients on apixaban were significantly less likely to have a stroke/SE overall compared to patients on rivaroxaban (3.08% vs. 3.80%, p < 0.0001; Table 2). They were also significantly less likely to have the different types of strokes and SE. Time at risk for stroke was shorter for those in the apixaban cohort versus the rivaroxaban cohort. The apixaban overall stroke incidence rate was lower compared to the rivaroxaban cohort (2.09 vs. 2.33 per 100 person-years), with similar trends observed for individual strokes and SE.
Patients on apixaban were significantly less likely to have MB compared to patients on rivaroxaban (4.07% vs. 6.35%, p < 0.0001; Table 2). They were also significantly less likely to experience GI bleeding, ICH, and MB at other sites. Time at risk for MB was shorter for those in the apixaban cohort versus the rivaroxaban cohort. Overall MB incidence rate was lower for the apixaban cohort compared to the rivaroxaban cohort.
When compared with rivaroxaban using Cox PH model analysis where the proportional hazards assumption was diligently accounted for, apixaban was associated with a lower risk of stroke/SE and a lower risk of MB. Within the stroke/SE category, apixaban was associated with lower risks of hemorrhagic stroke. Similar risks were observed in both ischemic stroke and SE compared to rivaroxaban. Within the MB category, apixaban was associated with lower risks of GI, bleeding at other sites, and ICH compared with rivaroxaban (Fig. 3).
Post-PSM, patients on rivaroxaban were significantly less likely to have a stroke/SE overall (3.50% vs. 3.98%, p = 0.0418). Patients on rivaroxaban were significantly less likely to have SE but had similar likelihood of ischemic stroke and hemorrhagic stroke compared to patients on dabigatran (Table 2). Time at risk for stroke was longer for those in the rivaroxaban cohort versus the dabigatran cohort and the overall stroke incidence rate was the same for the rivaroxaban cohort compared to the dabigatran cohort. When assessing specific types of stroke/SE, the incidence rate for hemorrhagic stroke, ischemic stroke, and SE was lower for the rivaroxaban cohort compared to the dabigatran cohort.
Patients on rivaroxaban had similar likelihood of MB overall (5.86% vs. 5.58%, p = 0.3211), GI bleeding, and ICH compared to patients on dabigatran, but with more likely to have MB at other sites. Time at risk for MB was longer for those in the rivaroxaban cohort vs. the dabigatran cohort and the overall MB incidence rate was lower for the rivaroxaban cohort compared to the dabigatran cohort. This also held for GI bleeding incidence rates for rivaroxaban and dabigatran. The incidence rate was higher ICH and bleeding at other sites for rivaroxaban than dabigatran.
Cox PH model analysis was performed, and proportional hazards assumption was ensured. It indicates that patients on dabigatran had similar risks of stroke/SE, MB and in all categories of stroke or SE and MB except for bleeding at other sites compared to patients on rivaroxaban (Fig. 3).
To our knowledge, this is the first study using RWD to compare the risk of stroke/SE and MB between DOACs among Medicare patients with NVAF who switched from warfarin to DOACs. Using data from the Medicare population, a robust nationally representative database for over 27 million enrollees [24], this study found that apixaban was associated with a significantly lower risk of stroke/SE compared to either dabigatran or rivaroxaban. Also, apixaban was associated with a lower risk of MB compared to either dabigatran or rivaroxaban. Dabigatran and rivaroxaban were found to have nonsignificant differences in risk of stroke/SE and MB.
PSM was performed for balanced comparison between cohorts. While PSM reduced the standardized differences in demographic and clinical characteristics between cohorts, as the baseline characteristics and the apixaban, dabigatran, and rivaroxaban cohorts are very different, some categories of the characteristics remained unbalanced with the post-PSM cohorts with standardized difference (STD) > 10. As shown in the Supplementary Material Table 1, the patients on apixaban were substantially older than the patients on dabigatran with an STD of 25.56. PSM did not remove the difference completely. The STD was 19.54 in the apixaban versus dabigatran cohorts, indicating that patients on apixaban were older when the outcomes were analyzed (Table 1). Other unbalanced categories were Medicaid dual eligibility, Part D low-income subsidy (for apixaban vs. dabigatran cohorts and apixaban vs. rivaroxaban cohorts), where more patients were qualified for Medicaid dual eligibility and Part D low-income subsidy for dabigatran and rivaroxaban (Table 1).
Previous RWD studies have mainly focused on comparing the effectiveness and safety of warfarin with DOACs or among different DOACs for patients who newly initiated DOACs. A systemic and meta-analysis by Li et al. found no significant differences in the risk of stroke/SE between rivaroxaban and apixaban or between apixaban and dabigatran, but found that apixaban was associated with having a lower risk of MB compared to dabigatran and rivaroxaban [25]. Our study also found a lower risk of MB associated with apixaban compared to dabigatran and rivaroxaban, but differs in that we observed that apixaban is associated with significantly lower risks of stroke/SE compared with rivaroxaban and dabigatran.
In comparing the risks of stroke/SE and MB among patients who switched from warfarin to apixaban, dabigatran, or rivaroxaban in four large US commercial databases, Lip et al. found that, after switching, patients on apixaban had a significantly lower risk of stroke/SE compared to dabigatran and a significantly lower risk of MB compared to both rivaroxaban and dabigatran. The findings in our study consistently align with Lip et al., but add further support of the effectiveness and safety of apixaban compared to rivaroxaban and dabigatran among Medicare beneficiaries, a completely non-overlapping population [19, 26].
Now that the American College of Cardiology and American Heart Association, and European Society of Cardiology recommend DOACs instead of warfarin to reduce the risk of stroke/SE for patients with NVAF, it is important to assess stroke/SE and MB risk in patients with AF who switched from warfarin to DOACs as this is becoming more common [27, 28]. The findings suggest that there are differences in stroke/SE and MB risks depending on which DOAC is prescribed after warfarin wich could be of importance in clinical practice.
Current results provide insights into the comparative risks of stroke/SE and MB between DOACs among patients with NVAF who switched from warfarin. Although a recent RWD study found lower inpatient, ER, and outpatient visits, and lower non-drug costs among patients with NVAF who switched from warfarin to DOAC [13, 14], and others observed comparable or lower healthcare expenditure and/or healthcare utilization among DOAC patients compared with patients on warfarin [4, 17, 18], none of the studies examined clinical outcomes comparing these DOACs. To the best of our knowledge, this is the first study to examine comparative effectiveness and safety profiles of DOACs for patients with NVAF who switched from warfarin within the Medicare population. Our findings suggest that, among all the DOAC, apixaban is associated with reduced risk of stroke/SE and MB compared to dabigatran and rivaroxaban among who were switched from warfarin.
Since the Medicare databases which were used in this study are largely representative of the US Medicare population, these findings may be generalizable to patients in similar populations with comparable characteristics. PSM was used between cohorts at baseline to adjust for heterogeneity in the population, therefore reducing bias in the results. Furthermore, for each cohort, there was adequate sample size for assessing the effectiveness outcomes. Lastly, this study used an adjusted Cox PH model to quantify the risk of stroke/SE and MB across different DOACs.
Our findings should be interpreted in the context of some limitations. As with other retrospective observational studies, causal relationships cannot be determined between the study variables and outcomes of interest. Potential residual confounders, such as over-the-counter aspirin use, serum creatinine/creatinine clearance, and laboratory values, were unavailable in our data and potentially could have resulted in bias. Also, medications were based on pharmacy claims for fills and do not represent if a patient took the medication as prescribed. Additionally, since ICD, CPT, and HCPCS codes were used to identify the diagnoses and procedures, human data entry errors are a possibility, and a potential for misclassification bias may exist. Further, the results are confined to the US healthcare system and may not be generalizable to other countries. Finally, unobserved heterogeneity may exist across the four datasets used in this analysis. However, the likelihood of duplicate observations is relatively low, researched to be 0.5%, and is not likely to have a significant impact on the study results. This study focused on warfarin–DOAC switchers to explore clinical outcomes across cohorts. Literature focusing on non-switchers versus switchers already exists; however, not much is known about switching across different DOAC’s, specifically among the Medicare population. We designed this study to fill this gap in the literature by understanding clinical outcomes in a real-world population.
To our knowledge, this is the first RWD study assessing stroke/SE and MB outcomes among Medicare patients with NVAF who switched from warfarin to a DOAC. Risk of stroke/SE and MB varied depending on which DOAC a patient with NVAF switched to after initially being prescribed warfarin. In general, apixaban had the lowest risk of stroke/SE and MB compared to other DOACs switched to after warfarin. These results may inform clinical decisions on which DOAC should be considered if a patient with NVAF was previously using warfarin, but further research is needed before providing any treatment-specific recommendations.
Below is the link to the electronic supplementary material.Supplementary file1 (PDF 234 KB)