Current Medical Research & Opinion 0300-7995 doi:10.1185/03007995.2014.907140
Vol. 30, No. 7, 2014, 1317–1325
Article FT-0045.R1/907140 All rights reserved: reproduction in whole or part not permitted
Curr Med Res Opin Downloaded from informahealthcare.com by University of Laval on 07/06/14 For personal use only.
Original article Real-world comparative effectiveness and safety of rivaroxaban and warfarin in nonvalvular atrial fibrillation patients Franc¸ois Laliberte´ Michel Cloutier Groupe d’analyse, Lte´e, Montre´al, QC, Canada
Winnie W. Nelson
Abstract Background: Rivaroxaban was shown to be effective in reducing the risk of stroke and systemic embolism in patients with nonvalvular atrial fibrillation (AF) in a randomized controlled trial setting.
Janssen Scientific Affairs, LLC, Raritan, NJ, USA
Craig I. Coleman Hartford Hospital, Hartford, CT, USA University of Connecticut School of Pharmacy, Storrs, CT, USA
Dominic Pilon Groupe d’analyse, Lte´e, Montre´al, QC, Canada
William H. Olson Janssen Scientific Affairs, LLC, Raritan, NJ, USA
C.V. Damaraju Janssen Research and Development, LLC, Raritan, NJ, USA
Jeffrey R. Schein Janssen Scientific Affairs, LLC, Raritan, NJ, USA
Patrick Lefebvre Groupe d’analyse, Lte´e, Montre´al, QC, Canada Address for correspondence: Franc¸ois Laliberte´ MA, Groupe d’analyse, Lte´e – NA 1000 De La Gauchetie`re Ouest, Montre´al, Quebec H3B4W5, Canada. [email protected] Keywords: Comparison of treatment – Nonvalvular atrial fibrillation – Rivaroxaban – Warfarin Accepted: 18 March 2014; published online: 2 April 2014 Citation: Curr Med Res Opin 2014; 30:1317–25
! 2014 Informa UK Ltd www.cmrojournal.com
Objective: To assess real-world safety, effectiveness, and persistence associated with rivaroxaban and warfarin in nonvalvular AF patients. Methods: Healthcare claims from Symphony Health Solutions’ Patient Transactional Datasets from May 2011 to July 2012 were analyzed. Adult patients newly initiated on rivaroxaban or warfarin, with 2 AF diagnoses (ICD9-CM: 427.31) and a CHADS2 score 1 during the 180 day baseline period were included. Cohorts were matched 1:4 using propensity score methods. Study outcomes were major bleeding, intracranial hemorrhage (ICH), gastrointestinal (GI) bleeding, composite stroke and systemic embolism, and venous thromboembolism (VTE) events. Cox proportional hazard models were used to compare event and persistence rates. Results: The matched sample included 3654 rivaroxaban and 14,616 warfarin patients. Matching was adequate, with all standardized differences in patient characteristics510%. No significant differences were observed for bleeding and composite stroke and systemic embolism outcomes, although rivaroxaban users were associated with significantly fewer VTE events (hazard ratio [HR] ¼ 0.36, 95% confidence interval [CI]: 0.24–0.54, p50.0001) compared to warfarin users. Rivaroxaban was also associated with a significantly lower risk of treatment non-persistence (HR ¼ 0.66; 95% CI: 0.60–0.72, p50.0001). Limitations: Claims data may have contained inaccuracies, and mortality and laboratory data were not available. Confounding may still have been possible even after propensity score matching. Early use pattern of medications may have changed over time. Conclusion: This analysis suggests that rivaroxaban and warfarin do not differ significantly in real-world rates of composite stroke and systemic embolism and major, intracranial, or GI bleeding. Rivaroxaban, however, was associated with significantly fewer VTE events and significantly better treatment persistence compared with warfarin.
Comparative rivaroxaban and warfarin in AF patients Laliberte´ et al.
1317
Current Medical Research & Opinion Volume 30, Number 7
July 2014
Curr Med Res Opin Downloaded from informahealthcare.com by University of Laval on 07/06/14 For personal use only.
Introduction Atrial fibrillation (AF) is the most common cardiac rhythm disturbance and the American Heart Association estimates its prevalence in the US to be 2.7 to 6.1 million persons1. Chronic anticoagulants are regularly used in the prevention of stroke in patients with nonvalvular AF2, and until recently vitamin K antagonists (VKAs) were the only available option for stroke prevention in patients with AF3. Newer target-specific oral anticoagulants, such as rivaroxaban, dabigatran, and apixaban, have been approved for the prevention of stroke and systemic embolism in patients with nonvalvular AF, and have the advantage of not requiring international normalized ratio (INR) monitoring compared to warfarin4. The Rivaroxaban Once Daily Oral Direct Factor Xa Inhibition Compared with Vitamin K Antagonism for Prevention of Stroke and Embolism Trial in Atrial Fibrillation (ROCKET-AF) trial demonstrated that rivaroxaban is effective in reducing the risk of stroke and systemic embolism5. Despite the well established efficacy of VKAs in the treatment of nonvalvular AF, their use is limited by the frequent dose adjustments (INR monitoring) based on a patient’s medications and diet, and variability in a patient’s responses6. These dosing characteristics may negatively impact treatment persistence to VKAs. Outcomes for rivaroxaban and warfarin users among patients with nonvalvular AF have not been compared in a real-world setting. Therefore, the objective of the current study was to assess the safety, effectiveness, and continued use (i.e. medication persistence) of rivaroxaban and warfarin in patients with nonvalvular AF, using health care claims data from the United States.
Methods Data source The analysis utilized data from Symphony Health Solutions’ (SHS) Patient Transactional Datasets from May 2011 through July 2012. The SHS database is a longitudinal patient data source which captures adjudicated prescription claims across the United States across all payment types, including commercial plans, Medicare Part D, cash, assistance programs, and Medicaid. The database contains approximately 4.8 billion prescription claims linked to over 190 million unique prescription patients, of whom approximately 70 million patients have two or more years of prescription drug history. Claims from hospital and physician practices include over 190 million patients with CPT/HCPCS medical procedure history as well as ICD-9 diagnosis history of which nearly 91 million prescription drug patients are linked to a diagnosis. 1318
Comparative rivaroxaban and warfarin in AF patients Laliberte´ et al.
The overall sample represents nearly 30,000 pharmacies, 1000 hospitals, 800 outpatient facilities, and 80,000 physician practices. All data collected from the database were de-identified in compliance with the patient confidentiality requirements of the Health Insurance Portability and Accountability Act (HIPAA). Institutional review board was not required for this study.
Study design A retrospective matched-cohort design was used to assess the safety and effectiveness of rivaroxaban and warfarin. Patients included in the study were newly initiated on rivaroxaban or warfarin after November 2011 (the time of rivaroxaban approval for non-valvular AF in the US), were 18 years of age, had a CHADS2 score 1 during the 180 day baseline period, and had 2 diagnoses of AF (International Classification of Disease, Ninth Revision, Clinical Modification [ICD-9-CM]: 427.31) during the baseline or follow-up period. The requirement of at least two claims with AF diagnosis reduced the chance of false positives. The first dispensing was defined as the index date, and the study patients were required to have at least 6 months of clinical activity (a variable included in the SHS data) prior to the index date (baseline period). Patients with prior use of warfarin, but who initiated rivaroxaban after its approval in November 2011, were classified in the rivaroxaban cohort, consistent with recent clinical trials studying the use of novel oral anticoagulants by AF patients that have combined VKAexperienced and -naı¨ve patients (ROCKET-AF, RE-LY, ARISTOTLE)7–9. A total of 50–62% of all patients enrolled in these phase III trials were warfarin-experienced at randomization7–9. Based on results from the ROCKETAF trial suggesting that rivaroxaban subjects naı¨ve to warfarin were associated with better primary efficacy and safety endpoints relative to warfarin-exposed subjects, not restricting our sample to naı¨ve patients only in the current study produced more conservative estimates of difference between groups. Furthermore, in RE-LY and ARISTOTLE, no significant differences between VKAexperienced and -naı¨ve patients were observed for stroke, systemic embolism, and major bleeding during the observation period7,9. In addition, patients diagnosed at baseline with valvular involvement (ICD-9-CM: 394.x-397.x, 424.x, 746.0x746.7x, V42.2, V42.3; CPT-4: 33400-33478), pregnancy (ICD-9-CM: V22, V23, V27, 630.x-676.x), malignant cancers (ICD-9-CM: 140.x-208.xx, 230.x-234.x), and transient causes of AF (ICD-9-CM: 415.x, 429.4; CPT-4: 33400-33999) were excluded from the study. Patient baseline characteristics were collected during the 6 months preceding the index date, and the observation period (follow-up) spanned from the index date until www.cmrojournal.com ! 2014 Informa UK Ltd
Current Medical Research & Opinion Volume 30, Number 7
Table 1. Demographic and clinical characteristics of rivaroxaban and warfarin users.
Curr Med Res Opin Downloaded from informahealthcare.com by University of Laval on 07/06/14 For personal use only.
Characteristics
Rivaroxaban Cohort (N ¼ 3654)
Warfarin Cohort (N ¼ 14,616)
Demographicsa Age, years, mean (SD) Gender, female, n (%)
73.3 (8.4) 1865 (51.0)
73.7 (8.3) 7530 (51.5)
4.7 1.0
Insurance type, n (%)a Cash Commercial Employer group Medicaid Medicare PBM Processors TPA Worker’s compensation Unknown
97 (2.7) 28 (0.8) 316 (8.6) 42 (1.1) 1747 (47.8) 813 (22.2) 8 (0.2) 4 (0.1) 0 (0) 599 (16.4)
395 (2.7) 101 (0.7) 1262 (8.6) 175 (1.2) 7167 (49.0) 3213 (22.0) 44 (0.3) 17 (0.1) 0 (0) 2242 (15.3)
0.3 0.9 0.0 0.4 2.5 0.6 1.6 0.2 – 2.9
1.2 (1.4)
3.9
5817 (39.8) 4341 (29.7) 2396 (16.4) 1059 (7.2) 536 (3.7) 467 (3.2)
3.4 0.9 0.7 1.4 0.9 3.0
2.0 (1.0)
2.0 (1.0)
3.6
0 (0) 1464 (40.1) 1304 (35.7) 578 (15.8) 229 (6.3) 65 (1.8) 14 (0.4)
0 (0) 5514 (37.7) 5409 (37.0) 2423 (16.6) 952 (6.5) 258 (1.8) 60 (0.4)
– 4.8 2.7 2.1 1.0 0.1 0.4
3.4 (1.4)
3.5 (1.4)
5.5
0 (0) 218 (6.0) 712 (19.5) 1116 (30.5) 884 (24.2) 451 (12.3) 273 (7.5)
0 (0) 677 (4.6) 2746 (18.8) 4502 (30.8) 3645 (24.9) 1847 (12.6) 1199 (8.2)
– 6.0 1.8 0.6 1.7 0.9 2.7
2.7 (1.9)
2.7 (1.9)
4.1
Quan–Charlson comorbidity indexb Mean (SD) 1.1 (1.3) N (%) 0 1515 (41.5) 1 1071 (29.3) 2 590 (16.1) 3 252 (6.9) 4 128 (3.5) 5 98 (2.7) CHADS2 scoreb Mean (SD) N (%) 0 1 2 3 4 5 6 CHA2DS2-VASc scoreb Mean (SD) N (%) 0 1 2 3 4 5 6 ATRIA scoreb Mean (SD) N (%) Low (0–3) Intermediate (4) High (5–10) HAS-BLED scoreb Mean (SD) N (%) 0 1 2 3 4 5 6
2902 (79.4) 222 (6.1) 530 (14.5)
11,458 (78.4) 912 (6.2) 2246 (15.4)
Standardized Difference in %
2.5 0.7 2.4
1.9 (0.8)
1.9 (0.8)
2.0
50 (1.4) 1139 (31.2) 1793 (49.1) 551 (15.1) 102 (2.8) 18 (0.5) 1 (0.0)
215 (1.5) 4399 (30.1) 7243 (49.6) 2244 (15.4) 428 (2.9) 84 (0.6) 3 (0.0)
0.9 2.3 1.0 0.8 0.8 1.1 0.4
SD: Standard deviation; VTE: Venous thromboembolism. a Evaluated at the index date. b Evaluated during the 6 month baseline period.
! 2014 Informa UK Ltd www.cmrojournal.com
July 2014
the earliest of end of clinical activity, a switch to warfarin (for rivaroxaban users), a switch to rivaroxaban (for warfarin users), or end of data availability (i.e., July 2012). In addition, for the safety and effectiveness outcomes, the end of the observation period was further restricted to 14 days after treatment non-persistence. This criterion increased the certainty that the clinical outcomes were evaluated during exposure to the medications of interest.
Study endpoints The study endpoints included safety and effectiveness outcomes of major bleeding, intracranial hemorrhage (ICH), gastrointestinal (GI) bleeding, composite stroke and systemic embolism (ischemic stroke, hemorrhagic stroke, systemic embolism), and venous thromboembolism (VTE) (deep vein thrombosis [DVT] and pulmonary embolism [PE]) events. The complete list of ICD-9-CM codes is included in Appendix A. The composite stroke and systemic embolism endpoints were required to be identified during a hospitalization or emergency department visit as a primary or a secondary diagnosis. VTE events were required to be identified during either (1) a hospitalization or emergency department visit or (2) during an outpatient visit with a 6 month washout (to ensure the identification of a new VTE event). Hemorrhagic stroke was defined as the occurrence of both a diagnosis of ICH and a diagnosis of late effects of cerebrovascular disease during the same hospitalization. Persistence with rivaroxaban and warfarin therapies was also evaluated and compared between the two groups. Treatment nonpersistence was defined as having a gap of greater than 60 days following the presumed exhaustion of a prescription fill without another refill.
Propensity score matching Propensity score matching was performed to minimize sample selection bias and the risk of confounding between rivaroxaban and warfarin users. Patients in the rivaroxaban cohort were matched 1:4 to patients in the warfarin cohort, using random selection among propensity score calipers of 25%. Propensity scores were calculated using a multivariate logistic regression model, incorporating the following baseline characteristics: demographics, insurance type comorbidities, and risk factors for bleeding, stroke and VTE events. The characteristics included in the propensity score model are listed in Tables 1 and 2.
Statistical analysis Descriptive statistics were generated to summarize the baseline characteristics of the study population. Frequencies and proportions were reported for categorical Comparative rivaroxaban and warfarin in AF patients Laliberte´ et al.
1319
Current Medical Research & Opinion Volume 30, Number 7
July 2014
Table 2. Baseline risk factors of rivaroxaban and warfarin usersa,b.
Curr Med Res Opin Downloaded from informahealthcare.com by University of Laval on 07/06/14 For personal use only.
Characteristics
Rivaroxaban Cohort (N ¼ 3654)
Warfarin Cohort (N ¼ 14,616)
Standardized Difference in %
2626 (71.9) 919 (25.2)
10,415 (71.3) 3861 (26.4)
1.3 2.9
3313 (90.7) 3018 (82.6) 2570 (70.3) 1983 (54.3) 464 (12.7) 446 (12.2) 274 (7.5) 392 (10.7) 376 (10.3) 357 (9.8) 285 (7.8) 202 (5.5) 89 (2.4) 17 (0.5)
13,344 (91.3) 12,365 (84.6) 10,589 (72.4) 8344 (57.1) 1739 (11.9) 1896 (13.0) 1200 (8.2) 1681 (11.5) 1522 (10.4) 1434 (9.8) 1176 (8.0) 831 (5.7) 358 (2.4) 78 (0.5)
2.2 5.4 4.7 5.7 2.4 2.3 2.6 2.5 0.4 0.1 0.9 0.7 0.1 1.0
3018 (82.6) 1983 (54.3) 1616 (44.2) 716 (19.6) 357 (9.8)
12,365 (84.6) 8344 (57.1) 6408 (43.8) 3033 (20.8) 1434 (9.8)
5.4 5.7 0.8 2.9 0.1
3313 (90.7) 1616 (44.2) 323 (8.8) 234 (6.4) 44 (1.2)
13,344 (91.3) 6408 (43.8) 1407 (9.6) 1028 (7.0) 177 (1.3)
2.2 0.8 2.7 2.5 0.6
Bleeding, stroke, and VTE risk factors, n (%) Hypertension Diabetes Bleeding risk factors, n (%) Age 460 Age 465 Age 470 Age 475 NSAID use Renal disease Chronic kidney disease Anemia Excessive fall risk (Parkinson’s disease, etc.) Cerebrovascular accident (stroke) Previous bleeding Non-major bleeding GI bleeding Major bleeding Stroke risk factors, n (%) Age 465 Age 475 Hyperlipidemia Heart failure Cerebrovascular accident (stroke) VTE risk factors, n (%) Age 460 Hyperlipidemia Multiple trauma Other serious infections Previous VTE
VTE: venous thromboembolism; GI: gastrointestinal; ETOH: ethanol, or drinking alcohol; COPD: chronic obstructive pulmonary disease; CVD: cardiovascular disease; SERM: selective estrogen receptor modulators. a Evaluated during the 6 month baseline period. b Table shows only factors observed in at least 5% of the sample with the addition of previous VTE. Additional propensity score matching factors not reported in this table include the following bleeding, stroke, and VTE risk factors observed in less than 5% of the study sample: use of drugs, myocardial infarction, left ventricular dysfunction, thrombocytopenia (low platelet count), hepatic disease, ETOH abuse, peptic ulcer, antiplatelet use, bleeding diathesis, depression, COPD, family history of CVD, obesity, pneumonia, abdominal surgery, coagulation defect, thrombophilia, rheumatoid arthritis, hip, pelvis, or leg fracture, varicose veins, inflammatory bowel disease, total knee replacement, total hip replacement, central venous catheter, surgical resection of abdominal or pelvic cancer, treatment with erythropoiesis stimulating agents, spinal cord injury, treatment with aromatase inhibitors, contraceptive pill, immobility, and treatment with SERMs.
variables and compared using a Pearson chi-squared test. Mean and standard deviation (SD) were reported for continuous variables and compared using Student’s t-tests. Standardized differences were used to compare the baseline characteristics between the two cohorts, where a standardized difference of less than 10% was considered not clinically relevant10–12. Cox proportional hazard regressions were used to calculate hazard ratios (HRs) of bleeding (major, ICH, and GI), the composite of stroke and systemic embolism, and VTE (DVT and PE) to compare the time to first occurrence of the study endpoints between rivaroxaban and warfarin users. Kaplan–Meier estimates and HRs were used to compare the time to non-persistence of therapy between the two cohorts. A sensitivity analysis was conducted for the analysis of persistence with therapy for rivaroxaban and warfarin users, where the use of other oral anticoagulants (i.e., dabigatran) during the follow-up was allowed (not considered 1320
Comparative rivaroxaban and warfarin in AF patients Laliberte´ et al.
as a gap in therapy). This sensitivity analysis was conducted to evaluate the potential impact on persistence of patients switching to another oral anticoagulant. Statistical significance was assessed with two-sided tests at a significant level of 0.05. All statistical analyses were conducted using SAS 9.3 (SAS Institute Inc., Cary, NC, USA).
Results Patient characteristics A total of 3654 rivaroxaban and 26,825 warfarin users with a CHADS2 score 1 were included (Figure 1). All rivaroxaban users were matched 1:4 with 14,616 warfarin users, respectively. The baseline characteristics of the matched cohorts are presented in Table 1. Between cohorts, all baseline characteristics were well balanced www.cmrojournal.com ! 2014 Informa UK Ltd
Current Medical Research & Opinion Volume 30, Number 7
July 2014
Curr Med Res Opin Downloaded from informahealthcare.com by University of Laval on 07/06/14 For personal use only.
Rivaroxaban/warfarin users after November 2011 N = 1,083,888
Rivaroxaban Cohort
Warfarin Cohort
Rivaroxaban users N = 25,789
Warfarin users N = 1,058,099
≥180 days of continuous activity N = 20,821
≥180 days of continuous activity N = 878,069
Newly initiated (180 day washout period) N = 277,252 ≥2 AF diagnoses N = 5882
≥2 AF diagnoses N = 44,260
≥18 years of age N = 5882
≥18 years of age N = 44,260
1378 were excluded 1008 had a valvular involvement 2 were pregnant 408 had a malignant cancer 73 had transient causes of AF
12,397 were excluded 8167 had a valvular involvement 19 were pregnant 3867 had a malignant cancer 1872 had transient causes of AF
Rivaroxaban users after exclusions N = 4504
Warfarin users after exclusions N = 31,863
Rivaroxaban users with CHADS2 score ≥1 at baseline
Warfarin users with CHADS2 score ≥1 at baseline
N = 3654
N = 26,825
AF: atrial fibrillation.
Figure 1. Patient disposition.
with standardized differences below 10%. Mean ages of the rivaroxaban and warfarin groups were 73.3 and 73.7 years (standardized difference ¼ 4.7%), respectively; 51.0% and 51.5% were female (standardized difference ¼ 1.0%). Rivaroxaban and warfarin users’ Charlson Comorbidity Index were 1.1 and 1.2, respectively (standardized difference ¼ 3.9%). Corresponding results for the CHADS2 score were 2.0 and 2.0 for rivaroxaban and warfarin users, respectively (standardized difference ¼ 3.6%). Notably, the most frequent (425%) baseline risk factors for bleeding, stroke, and VTE were diabetes, hypertension, age 60 years or older, and hyperlipidemia (Table 2). ! 2014 Informa UK Ltd www.cmrojournal.com
Dosing patterns and concomitant medications Table 3 presents patients’ dosing patterns and concomitant medications in the observation period. Rivaroxaban and warfarin users were observed for an average of 83 and 113 days, respectively (p50.0001). The mean observation was shorter for rivaroxaban users since patients started to use rivaroxaban gradually in the first months after its approval (in November 2011). The mean number of dispensing per patient received by warfarin users was higher than that of rivaroxaban users with 2.8 and 2.6, respectively (p50.0001). The mean days of supply per dispensing was also higher in the warfarin cohort with 51 days compared to 37 days for the rivaroxaban cohort (p50.0001). Comparative rivaroxaban and warfarin in AF patients Laliberte´ et al.
1321
Current Medical Research & Opinion Volume 30, Number 7
July 2014
Table 3. Dosing patterns and concomitant medications of rivaroxaban and warfarin users.
Curr Med Res Opin Downloaded from informahealthcare.com by University of Laval on 07/06/14 For personal use only.
Variables
Rivaroxaban Cohort (N ¼ 3654)
Observation perioda, days, mean (SD) Number of dispensings per patient, mean (SD) Days of supply per dispensing, mean (SD) Daily dosage, mg, mean (SD) Concomitant anticoagulant medication, n (%) Low molecular weight heparins Fondaparinux Dabigatran Persistenceb with index anticoagulant, % 3 months 6 months Persistenceb allowing switch to other oral anticoagulant, % 3 months 6 months
Warfarin Cohort (N ¼ 14,616)
p-Value
83 (58) 2.6 (1.8) 37.3 (19.7) 19.1 (2.1)
113 (70) 2.8 (2.2) 50.6 (28.6) 5.0 (4.1)
50.0001 50.0001 50.0001
11 (0.3) 0 (0.0) 72 (2.0)
865 (5.9) 17 (0.1) 341 (2.3)
50.0001 0.0392 0.1872
84.50 81.48
75.60 68.32
50.0001 50.0001
85.80 82.53
77.19 69.81
50.0001 50.0001
SD: Standard deviation. a Earliest date between 14 days after treatment non-persistence, end of data availability, end of clinical activity, or a switch to warfarin (rivaroxaban cohort) or rivaroxaban (warfarin cohort). b Treatment non-persistence was defined as a gap greater than 60 days between index therapy refills of prescription.
In addition, dabigatran was concomitantly dispensed to the same proportion of patients (2%; p ¼ 0.1872) in the two cohorts (the reason for this pattern is unknown), while low molecular weight heparins were concomitantly dispensed to 0.3% and 5.9% of rivaroxaban and warfarin users, respectively (p50.0001).
lower likelihood of non-persistence with the treatment compared to warfarin users after 6 months (HR ¼ 0.66, 95% CI: 0.60–0.72, p50.0001). For the sensitivity analysis allowing the use of other oral anticoagulants, rivaroxaban users still had a significantly higher persistence with therapy after 6 months (82.5% vs. 69.8%, p50.0001).
Hazard ratios of bleeding, composite stroke and systemic embolism, and VTE events
Discussion
The rates of major bleeding, ICH, GI bleeding, and composite stroke and systemic embolism events were not statistically different for rivaroxaban and warfarin users. The rates of major bleeding for rivaroxaban and warfarin users were 3.3 and 2.9 events per 100 person-years (HR ¼ 1.08, 95% CI: 0.71–1.64, p ¼ 0.7209; Figure 2), while the corresponding rates for ICH were 1.8 and 1.5 events per 100 person-years (HR ¼ 1.17, 95% CI: 0.66– 2.05, p ¼ 0.5905), and, respectively, 9.5 and 7 events per 100 person-years for GI bleeding (HR ¼ 1.27, 95% CI: 0.99–1.63, p ¼ 0.0590). Rates of composite stroke and systemic embolism for rivaroxaban and warfarin users were 4.6 events and 5.9 events per 100 person-years (HR ¼ 0.77, 95% CI: 0.55–1.09, p ¼ 0.1364). Rates of VTE were significantly lower for rivaroxaban compared to warfarin users, at 2.9 and 7.1 events per 100 person-years (HR ¼ 0.36, 95% CI: 0.24–0.54, p50.0001), respectively.
Treatment persistence Rivaroxaban users had a significantly higher persistence with therapy after 3 months (84.5% vs. 75.6%, p50.0001; Table 3) and after 6 months (81.5% vs. 68.3%, p50.0001). Figure 3 presents the Kaplan–Meier curves for the two cohorts. Rivaroxaban users had a 1322
Comparative rivaroxaban and warfarin in AF patients Laliberte´ et al.
This study used real-world claims data from a US population to compare the safety and effectiveness of rivaroxaban with warfarin therapy in an adult nonvalvular AF population at risk for stroke. No significant differences were observed between rivaroxaban and warfarin in terms of safety and effectiveness regarding major bleeding, ICH, GI bleeding, and composite stroke and systemic embolism outcomes, but rivaroxaban was associated with significantly lower rates of VTE and with a higher persistence with therapy. Anticoagulant persistence is of importance for nonvalvular AF patients, especially given the severity of potential adverse events such as stroke. The persistence rate below 70% the authors found at 6 months for newly initiated warfarin users is worrisome. Potential explanations for the lower persistence associated with warfarin relates to its numerous limitations for chronic use13. Complexity of warfarin chronic use is caused by its natural limitations of narrow therapeutic range and constant need for monitoring14. Also, pharmacology sensitivity to both dietary and physiologic changes, as well as drug–drug interactions, makes monitoring even more important and also a primal cause of the increased likelihood of non-adherence15. Simpler regimens have been shown to promote medication adherence and persistence with cardiovascular www.cmrojournal.com ! 2014 Informa UK Ltd
Current Medical Research & Opinion Volume 30, Number 7
July 2014
Major bleeding
1.08 (0.71 - 1.64)
ICH
1.17 (0.66 - 2.05)
GI bleeding Composite strike and systemic embolisma Ischemic strokea
1.27 (0.99 - 1.63)
Hemorrhagic strokea
1.11 (0.13 - 9.60)
Systemic embolisma
0.54 (0.13 - 2.33)
VTEb
0.36 (0.24 - 0.54)*
DVT onlyb PE with or without DVTb
0.33 (0.21 - 0.53)*
0.77 (0.55 - 1.09) 0.81 (0.57 - 1.14)
0.64 (0.35 - 1.18)
0
0.5
1.5
1
Rivaroxaban better
2
2.5
3
Warfarin better
ICH: intracranial hemorrhage; GI: gastrointestinal; VTE: venous thromboembolism; DVT: deep vein thrombosis; PE: pulmonary embolism. *Statistically significant at 0.05 level. aEvent identified during a hospitalization or emergency department visit. bEvent identified during a hospitalization, an emergency department visit, or an outpatient visit (with a 6 month washout).
Figure 2. Hazard ratios of bleeding, stroke and VTE events for rivaroxaban compared to warfarin users.
100%
Proportion of patients persistent with treatment
Curr Med Res Opin Downloaded from informahealthcare.com by University of Laval on 07/06/14 For personal use only.
Rivaroxaban vs. Warfarin Hazard ratios [95% CI]
Hazard Ratio: 0.66 (0.60 - 0.72) P
Vol. 30, No. 7, 2014, 1317–1325
Article FT-0045.R1/907140 All rights reserved: reproduction in whole or part not permitted
Curr Med Res Opin Downloaded from informahealthcare.com by University of Laval on 07/06/14 For personal use only.
Original article Real-world comparative effectiveness and safety of rivaroxaban and warfarin in nonvalvular atrial fibrillation patients Franc¸ois Laliberte´ Michel Cloutier Groupe d’analyse, Lte´e, Montre´al, QC, Canada
Winnie W. Nelson
Abstract Background: Rivaroxaban was shown to be effective in reducing the risk of stroke and systemic embolism in patients with nonvalvular atrial fibrillation (AF) in a randomized controlled trial setting.
Janssen Scientific Affairs, LLC, Raritan, NJ, USA
Craig I. Coleman Hartford Hospital, Hartford, CT, USA University of Connecticut School of Pharmacy, Storrs, CT, USA
Dominic Pilon Groupe d’analyse, Lte´e, Montre´al, QC, Canada
William H. Olson Janssen Scientific Affairs, LLC, Raritan, NJ, USA
C.V. Damaraju Janssen Research and Development, LLC, Raritan, NJ, USA
Jeffrey R. Schein Janssen Scientific Affairs, LLC, Raritan, NJ, USA
Patrick Lefebvre Groupe d’analyse, Lte´e, Montre´al, QC, Canada Address for correspondence: Franc¸ois Laliberte´ MA, Groupe d’analyse, Lte´e – NA 1000 De La Gauchetie`re Ouest, Montre´al, Quebec H3B4W5, Canada. [email protected] Keywords: Comparison of treatment – Nonvalvular atrial fibrillation – Rivaroxaban – Warfarin Accepted: 18 March 2014; published online: 2 April 2014 Citation: Curr Med Res Opin 2014; 30:1317–25
! 2014 Informa UK Ltd www.cmrojournal.com
Objective: To assess real-world safety, effectiveness, and persistence associated with rivaroxaban and warfarin in nonvalvular AF patients. Methods: Healthcare claims from Symphony Health Solutions’ Patient Transactional Datasets from May 2011 to July 2012 were analyzed. Adult patients newly initiated on rivaroxaban or warfarin, with 2 AF diagnoses (ICD9-CM: 427.31) and a CHADS2 score 1 during the 180 day baseline period were included. Cohorts were matched 1:4 using propensity score methods. Study outcomes were major bleeding, intracranial hemorrhage (ICH), gastrointestinal (GI) bleeding, composite stroke and systemic embolism, and venous thromboembolism (VTE) events. Cox proportional hazard models were used to compare event and persistence rates. Results: The matched sample included 3654 rivaroxaban and 14,616 warfarin patients. Matching was adequate, with all standardized differences in patient characteristics510%. No significant differences were observed for bleeding and composite stroke and systemic embolism outcomes, although rivaroxaban users were associated with significantly fewer VTE events (hazard ratio [HR] ¼ 0.36, 95% confidence interval [CI]: 0.24–0.54, p50.0001) compared to warfarin users. Rivaroxaban was also associated with a significantly lower risk of treatment non-persistence (HR ¼ 0.66; 95% CI: 0.60–0.72, p50.0001). Limitations: Claims data may have contained inaccuracies, and mortality and laboratory data were not available. Confounding may still have been possible even after propensity score matching. Early use pattern of medications may have changed over time. Conclusion: This analysis suggests that rivaroxaban and warfarin do not differ significantly in real-world rates of composite stroke and systemic embolism and major, intracranial, or GI bleeding. Rivaroxaban, however, was associated with significantly fewer VTE events and significantly better treatment persistence compared with warfarin.
Comparative rivaroxaban and warfarin in AF patients Laliberte´ et al.
1317
Current Medical Research & Opinion Volume 30, Number 7
July 2014
Curr Med Res Opin Downloaded from informahealthcare.com by University of Laval on 07/06/14 For personal use only.
Introduction Atrial fibrillation (AF) is the most common cardiac rhythm disturbance and the American Heart Association estimates its prevalence in the US to be 2.7 to 6.1 million persons1. Chronic anticoagulants are regularly used in the prevention of stroke in patients with nonvalvular AF2, and until recently vitamin K antagonists (VKAs) were the only available option for stroke prevention in patients with AF3. Newer target-specific oral anticoagulants, such as rivaroxaban, dabigatran, and apixaban, have been approved for the prevention of stroke and systemic embolism in patients with nonvalvular AF, and have the advantage of not requiring international normalized ratio (INR) monitoring compared to warfarin4. The Rivaroxaban Once Daily Oral Direct Factor Xa Inhibition Compared with Vitamin K Antagonism for Prevention of Stroke and Embolism Trial in Atrial Fibrillation (ROCKET-AF) trial demonstrated that rivaroxaban is effective in reducing the risk of stroke and systemic embolism5. Despite the well established efficacy of VKAs in the treatment of nonvalvular AF, their use is limited by the frequent dose adjustments (INR monitoring) based on a patient’s medications and diet, and variability in a patient’s responses6. These dosing characteristics may negatively impact treatment persistence to VKAs. Outcomes for rivaroxaban and warfarin users among patients with nonvalvular AF have not been compared in a real-world setting. Therefore, the objective of the current study was to assess the safety, effectiveness, and continued use (i.e. medication persistence) of rivaroxaban and warfarin in patients with nonvalvular AF, using health care claims data from the United States.
Methods Data source The analysis utilized data from Symphony Health Solutions’ (SHS) Patient Transactional Datasets from May 2011 through July 2012. The SHS database is a longitudinal patient data source which captures adjudicated prescription claims across the United States across all payment types, including commercial plans, Medicare Part D, cash, assistance programs, and Medicaid. The database contains approximately 4.8 billion prescription claims linked to over 190 million unique prescription patients, of whom approximately 70 million patients have two or more years of prescription drug history. Claims from hospital and physician practices include over 190 million patients with CPT/HCPCS medical procedure history as well as ICD-9 diagnosis history of which nearly 91 million prescription drug patients are linked to a diagnosis. 1318
Comparative rivaroxaban and warfarin in AF patients Laliberte´ et al.
The overall sample represents nearly 30,000 pharmacies, 1000 hospitals, 800 outpatient facilities, and 80,000 physician practices. All data collected from the database were de-identified in compliance with the patient confidentiality requirements of the Health Insurance Portability and Accountability Act (HIPAA). Institutional review board was not required for this study.
Study design A retrospective matched-cohort design was used to assess the safety and effectiveness of rivaroxaban and warfarin. Patients included in the study were newly initiated on rivaroxaban or warfarin after November 2011 (the time of rivaroxaban approval for non-valvular AF in the US), were 18 years of age, had a CHADS2 score 1 during the 180 day baseline period, and had 2 diagnoses of AF (International Classification of Disease, Ninth Revision, Clinical Modification [ICD-9-CM]: 427.31) during the baseline or follow-up period. The requirement of at least two claims with AF diagnosis reduced the chance of false positives. The first dispensing was defined as the index date, and the study patients were required to have at least 6 months of clinical activity (a variable included in the SHS data) prior to the index date (baseline period). Patients with prior use of warfarin, but who initiated rivaroxaban after its approval in November 2011, were classified in the rivaroxaban cohort, consistent with recent clinical trials studying the use of novel oral anticoagulants by AF patients that have combined VKAexperienced and -naı¨ve patients (ROCKET-AF, RE-LY, ARISTOTLE)7–9. A total of 50–62% of all patients enrolled in these phase III trials were warfarin-experienced at randomization7–9. Based on results from the ROCKETAF trial suggesting that rivaroxaban subjects naı¨ve to warfarin were associated with better primary efficacy and safety endpoints relative to warfarin-exposed subjects, not restricting our sample to naı¨ve patients only in the current study produced more conservative estimates of difference between groups. Furthermore, in RE-LY and ARISTOTLE, no significant differences between VKAexperienced and -naı¨ve patients were observed for stroke, systemic embolism, and major bleeding during the observation period7,9. In addition, patients diagnosed at baseline with valvular involvement (ICD-9-CM: 394.x-397.x, 424.x, 746.0x746.7x, V42.2, V42.3; CPT-4: 33400-33478), pregnancy (ICD-9-CM: V22, V23, V27, 630.x-676.x), malignant cancers (ICD-9-CM: 140.x-208.xx, 230.x-234.x), and transient causes of AF (ICD-9-CM: 415.x, 429.4; CPT-4: 33400-33999) were excluded from the study. Patient baseline characteristics were collected during the 6 months preceding the index date, and the observation period (follow-up) spanned from the index date until www.cmrojournal.com ! 2014 Informa UK Ltd
Current Medical Research & Opinion Volume 30, Number 7
Table 1. Demographic and clinical characteristics of rivaroxaban and warfarin users.
Curr Med Res Opin Downloaded from informahealthcare.com by University of Laval on 07/06/14 For personal use only.
Characteristics
Rivaroxaban Cohort (N ¼ 3654)
Warfarin Cohort (N ¼ 14,616)
Demographicsa Age, years, mean (SD) Gender, female, n (%)
73.3 (8.4) 1865 (51.0)
73.7 (8.3) 7530 (51.5)
4.7 1.0
Insurance type, n (%)a Cash Commercial Employer group Medicaid Medicare PBM Processors TPA Worker’s compensation Unknown
97 (2.7) 28 (0.8) 316 (8.6) 42 (1.1) 1747 (47.8) 813 (22.2) 8 (0.2) 4 (0.1) 0 (0) 599 (16.4)
395 (2.7) 101 (0.7) 1262 (8.6) 175 (1.2) 7167 (49.0) 3213 (22.0) 44 (0.3) 17 (0.1) 0 (0) 2242 (15.3)
0.3 0.9 0.0 0.4 2.5 0.6 1.6 0.2 – 2.9
1.2 (1.4)
3.9
5817 (39.8) 4341 (29.7) 2396 (16.4) 1059 (7.2) 536 (3.7) 467 (3.2)
3.4 0.9 0.7 1.4 0.9 3.0
2.0 (1.0)
2.0 (1.0)
3.6
0 (0) 1464 (40.1) 1304 (35.7) 578 (15.8) 229 (6.3) 65 (1.8) 14 (0.4)
0 (0) 5514 (37.7) 5409 (37.0) 2423 (16.6) 952 (6.5) 258 (1.8) 60 (0.4)
– 4.8 2.7 2.1 1.0 0.1 0.4
3.4 (1.4)
3.5 (1.4)
5.5
0 (0) 218 (6.0) 712 (19.5) 1116 (30.5) 884 (24.2) 451 (12.3) 273 (7.5)
0 (0) 677 (4.6) 2746 (18.8) 4502 (30.8) 3645 (24.9) 1847 (12.6) 1199 (8.2)
– 6.0 1.8 0.6 1.7 0.9 2.7
2.7 (1.9)
2.7 (1.9)
4.1
Quan–Charlson comorbidity indexb Mean (SD) 1.1 (1.3) N (%) 0 1515 (41.5) 1 1071 (29.3) 2 590 (16.1) 3 252 (6.9) 4 128 (3.5) 5 98 (2.7) CHADS2 scoreb Mean (SD) N (%) 0 1 2 3 4 5 6 CHA2DS2-VASc scoreb Mean (SD) N (%) 0 1 2 3 4 5 6 ATRIA scoreb Mean (SD) N (%) Low (0–3) Intermediate (4) High (5–10) HAS-BLED scoreb Mean (SD) N (%) 0 1 2 3 4 5 6
2902 (79.4) 222 (6.1) 530 (14.5)
11,458 (78.4) 912 (6.2) 2246 (15.4)
Standardized Difference in %
2.5 0.7 2.4
1.9 (0.8)
1.9 (0.8)
2.0
50 (1.4) 1139 (31.2) 1793 (49.1) 551 (15.1) 102 (2.8) 18 (0.5) 1 (0.0)
215 (1.5) 4399 (30.1) 7243 (49.6) 2244 (15.4) 428 (2.9) 84 (0.6) 3 (0.0)
0.9 2.3 1.0 0.8 0.8 1.1 0.4
SD: Standard deviation; VTE: Venous thromboembolism. a Evaluated at the index date. b Evaluated during the 6 month baseline period.
! 2014 Informa UK Ltd www.cmrojournal.com
July 2014
the earliest of end of clinical activity, a switch to warfarin (for rivaroxaban users), a switch to rivaroxaban (for warfarin users), or end of data availability (i.e., July 2012). In addition, for the safety and effectiveness outcomes, the end of the observation period was further restricted to 14 days after treatment non-persistence. This criterion increased the certainty that the clinical outcomes were evaluated during exposure to the medications of interest.
Study endpoints The study endpoints included safety and effectiveness outcomes of major bleeding, intracranial hemorrhage (ICH), gastrointestinal (GI) bleeding, composite stroke and systemic embolism (ischemic stroke, hemorrhagic stroke, systemic embolism), and venous thromboembolism (VTE) (deep vein thrombosis [DVT] and pulmonary embolism [PE]) events. The complete list of ICD-9-CM codes is included in Appendix A. The composite stroke and systemic embolism endpoints were required to be identified during a hospitalization or emergency department visit as a primary or a secondary diagnosis. VTE events were required to be identified during either (1) a hospitalization or emergency department visit or (2) during an outpatient visit with a 6 month washout (to ensure the identification of a new VTE event). Hemorrhagic stroke was defined as the occurrence of both a diagnosis of ICH and a diagnosis of late effects of cerebrovascular disease during the same hospitalization. Persistence with rivaroxaban and warfarin therapies was also evaluated and compared between the two groups. Treatment nonpersistence was defined as having a gap of greater than 60 days following the presumed exhaustion of a prescription fill without another refill.
Propensity score matching Propensity score matching was performed to minimize sample selection bias and the risk of confounding between rivaroxaban and warfarin users. Patients in the rivaroxaban cohort were matched 1:4 to patients in the warfarin cohort, using random selection among propensity score calipers of 25%. Propensity scores were calculated using a multivariate logistic regression model, incorporating the following baseline characteristics: demographics, insurance type comorbidities, and risk factors for bleeding, stroke and VTE events. The characteristics included in the propensity score model are listed in Tables 1 and 2.
Statistical analysis Descriptive statistics were generated to summarize the baseline characteristics of the study population. Frequencies and proportions were reported for categorical Comparative rivaroxaban and warfarin in AF patients Laliberte´ et al.
1319
Current Medical Research & Opinion Volume 30, Number 7
July 2014
Table 2. Baseline risk factors of rivaroxaban and warfarin usersa,b.
Curr Med Res Opin Downloaded from informahealthcare.com by University of Laval on 07/06/14 For personal use only.
Characteristics
Rivaroxaban Cohort (N ¼ 3654)
Warfarin Cohort (N ¼ 14,616)
Standardized Difference in %
2626 (71.9) 919 (25.2)
10,415 (71.3) 3861 (26.4)
1.3 2.9
3313 (90.7) 3018 (82.6) 2570 (70.3) 1983 (54.3) 464 (12.7) 446 (12.2) 274 (7.5) 392 (10.7) 376 (10.3) 357 (9.8) 285 (7.8) 202 (5.5) 89 (2.4) 17 (0.5)
13,344 (91.3) 12,365 (84.6) 10,589 (72.4) 8344 (57.1) 1739 (11.9) 1896 (13.0) 1200 (8.2) 1681 (11.5) 1522 (10.4) 1434 (9.8) 1176 (8.0) 831 (5.7) 358 (2.4) 78 (0.5)
2.2 5.4 4.7 5.7 2.4 2.3 2.6 2.5 0.4 0.1 0.9 0.7 0.1 1.0
3018 (82.6) 1983 (54.3) 1616 (44.2) 716 (19.6) 357 (9.8)
12,365 (84.6) 8344 (57.1) 6408 (43.8) 3033 (20.8) 1434 (9.8)
5.4 5.7 0.8 2.9 0.1
3313 (90.7) 1616 (44.2) 323 (8.8) 234 (6.4) 44 (1.2)
13,344 (91.3) 6408 (43.8) 1407 (9.6) 1028 (7.0) 177 (1.3)
2.2 0.8 2.7 2.5 0.6
Bleeding, stroke, and VTE risk factors, n (%) Hypertension Diabetes Bleeding risk factors, n (%) Age 460 Age 465 Age 470 Age 475 NSAID use Renal disease Chronic kidney disease Anemia Excessive fall risk (Parkinson’s disease, etc.) Cerebrovascular accident (stroke) Previous bleeding Non-major bleeding GI bleeding Major bleeding Stroke risk factors, n (%) Age 465 Age 475 Hyperlipidemia Heart failure Cerebrovascular accident (stroke) VTE risk factors, n (%) Age 460 Hyperlipidemia Multiple trauma Other serious infections Previous VTE
VTE: venous thromboembolism; GI: gastrointestinal; ETOH: ethanol, or drinking alcohol; COPD: chronic obstructive pulmonary disease; CVD: cardiovascular disease; SERM: selective estrogen receptor modulators. a Evaluated during the 6 month baseline period. b Table shows only factors observed in at least 5% of the sample with the addition of previous VTE. Additional propensity score matching factors not reported in this table include the following bleeding, stroke, and VTE risk factors observed in less than 5% of the study sample: use of drugs, myocardial infarction, left ventricular dysfunction, thrombocytopenia (low platelet count), hepatic disease, ETOH abuse, peptic ulcer, antiplatelet use, bleeding diathesis, depression, COPD, family history of CVD, obesity, pneumonia, abdominal surgery, coagulation defect, thrombophilia, rheumatoid arthritis, hip, pelvis, or leg fracture, varicose veins, inflammatory bowel disease, total knee replacement, total hip replacement, central venous catheter, surgical resection of abdominal or pelvic cancer, treatment with erythropoiesis stimulating agents, spinal cord injury, treatment with aromatase inhibitors, contraceptive pill, immobility, and treatment with SERMs.
variables and compared using a Pearson chi-squared test. Mean and standard deviation (SD) were reported for continuous variables and compared using Student’s t-tests. Standardized differences were used to compare the baseline characteristics between the two cohorts, where a standardized difference of less than 10% was considered not clinically relevant10–12. Cox proportional hazard regressions were used to calculate hazard ratios (HRs) of bleeding (major, ICH, and GI), the composite of stroke and systemic embolism, and VTE (DVT and PE) to compare the time to first occurrence of the study endpoints between rivaroxaban and warfarin users. Kaplan–Meier estimates and HRs were used to compare the time to non-persistence of therapy between the two cohorts. A sensitivity analysis was conducted for the analysis of persistence with therapy for rivaroxaban and warfarin users, where the use of other oral anticoagulants (i.e., dabigatran) during the follow-up was allowed (not considered 1320
Comparative rivaroxaban and warfarin in AF patients Laliberte´ et al.
as a gap in therapy). This sensitivity analysis was conducted to evaluate the potential impact on persistence of patients switching to another oral anticoagulant. Statistical significance was assessed with two-sided tests at a significant level of 0.05. All statistical analyses were conducted using SAS 9.3 (SAS Institute Inc., Cary, NC, USA).
Results Patient characteristics A total of 3654 rivaroxaban and 26,825 warfarin users with a CHADS2 score 1 were included (Figure 1). All rivaroxaban users were matched 1:4 with 14,616 warfarin users, respectively. The baseline characteristics of the matched cohorts are presented in Table 1. Between cohorts, all baseline characteristics were well balanced www.cmrojournal.com ! 2014 Informa UK Ltd
Current Medical Research & Opinion Volume 30, Number 7
July 2014
Curr Med Res Opin Downloaded from informahealthcare.com by University of Laval on 07/06/14 For personal use only.
Rivaroxaban/warfarin users after November 2011 N = 1,083,888
Rivaroxaban Cohort
Warfarin Cohort
Rivaroxaban users N = 25,789
Warfarin users N = 1,058,099
≥180 days of continuous activity N = 20,821
≥180 days of continuous activity N = 878,069
Newly initiated (180 day washout period) N = 277,252 ≥2 AF diagnoses N = 5882
≥2 AF diagnoses N = 44,260
≥18 years of age N = 5882
≥18 years of age N = 44,260
1378 were excluded 1008 had a valvular involvement 2 were pregnant 408 had a malignant cancer 73 had transient causes of AF
12,397 were excluded 8167 had a valvular involvement 19 were pregnant 3867 had a malignant cancer 1872 had transient causes of AF
Rivaroxaban users after exclusions N = 4504
Warfarin users after exclusions N = 31,863
Rivaroxaban users with CHADS2 score ≥1 at baseline
Warfarin users with CHADS2 score ≥1 at baseline
N = 3654
N = 26,825
AF: atrial fibrillation.
Figure 1. Patient disposition.
with standardized differences below 10%. Mean ages of the rivaroxaban and warfarin groups were 73.3 and 73.7 years (standardized difference ¼ 4.7%), respectively; 51.0% and 51.5% were female (standardized difference ¼ 1.0%). Rivaroxaban and warfarin users’ Charlson Comorbidity Index were 1.1 and 1.2, respectively (standardized difference ¼ 3.9%). Corresponding results for the CHADS2 score were 2.0 and 2.0 for rivaroxaban and warfarin users, respectively (standardized difference ¼ 3.6%). Notably, the most frequent (425%) baseline risk factors for bleeding, stroke, and VTE were diabetes, hypertension, age 60 years or older, and hyperlipidemia (Table 2). ! 2014 Informa UK Ltd www.cmrojournal.com
Dosing patterns and concomitant medications Table 3 presents patients’ dosing patterns and concomitant medications in the observation period. Rivaroxaban and warfarin users were observed for an average of 83 and 113 days, respectively (p50.0001). The mean observation was shorter for rivaroxaban users since patients started to use rivaroxaban gradually in the first months after its approval (in November 2011). The mean number of dispensing per patient received by warfarin users was higher than that of rivaroxaban users with 2.8 and 2.6, respectively (p50.0001). The mean days of supply per dispensing was also higher in the warfarin cohort with 51 days compared to 37 days for the rivaroxaban cohort (p50.0001). Comparative rivaroxaban and warfarin in AF patients Laliberte´ et al.
1321
Current Medical Research & Opinion Volume 30, Number 7
July 2014
Table 3. Dosing patterns and concomitant medications of rivaroxaban and warfarin users.
Curr Med Res Opin Downloaded from informahealthcare.com by University of Laval on 07/06/14 For personal use only.
Variables
Rivaroxaban Cohort (N ¼ 3654)
Observation perioda, days, mean (SD) Number of dispensings per patient, mean (SD) Days of supply per dispensing, mean (SD) Daily dosage, mg, mean (SD) Concomitant anticoagulant medication, n (%) Low molecular weight heparins Fondaparinux Dabigatran Persistenceb with index anticoagulant, % 3 months 6 months Persistenceb allowing switch to other oral anticoagulant, % 3 months 6 months
Warfarin Cohort (N ¼ 14,616)
p-Value
83 (58) 2.6 (1.8) 37.3 (19.7) 19.1 (2.1)
113 (70) 2.8 (2.2) 50.6 (28.6) 5.0 (4.1)
50.0001 50.0001 50.0001
11 (0.3) 0 (0.0) 72 (2.0)
865 (5.9) 17 (0.1) 341 (2.3)
50.0001 0.0392 0.1872
84.50 81.48
75.60 68.32
50.0001 50.0001
85.80 82.53
77.19 69.81
50.0001 50.0001
SD: Standard deviation. a Earliest date between 14 days after treatment non-persistence, end of data availability, end of clinical activity, or a switch to warfarin (rivaroxaban cohort) or rivaroxaban (warfarin cohort). b Treatment non-persistence was defined as a gap greater than 60 days between index therapy refills of prescription.
In addition, dabigatran was concomitantly dispensed to the same proportion of patients (2%; p ¼ 0.1872) in the two cohorts (the reason for this pattern is unknown), while low molecular weight heparins were concomitantly dispensed to 0.3% and 5.9% of rivaroxaban and warfarin users, respectively (p50.0001).
lower likelihood of non-persistence with the treatment compared to warfarin users after 6 months (HR ¼ 0.66, 95% CI: 0.60–0.72, p50.0001). For the sensitivity analysis allowing the use of other oral anticoagulants, rivaroxaban users still had a significantly higher persistence with therapy after 6 months (82.5% vs. 69.8%, p50.0001).
Hazard ratios of bleeding, composite stroke and systemic embolism, and VTE events
Discussion
The rates of major bleeding, ICH, GI bleeding, and composite stroke and systemic embolism events were not statistically different for rivaroxaban and warfarin users. The rates of major bleeding for rivaroxaban and warfarin users were 3.3 and 2.9 events per 100 person-years (HR ¼ 1.08, 95% CI: 0.71–1.64, p ¼ 0.7209; Figure 2), while the corresponding rates for ICH were 1.8 and 1.5 events per 100 person-years (HR ¼ 1.17, 95% CI: 0.66– 2.05, p ¼ 0.5905), and, respectively, 9.5 and 7 events per 100 person-years for GI bleeding (HR ¼ 1.27, 95% CI: 0.99–1.63, p ¼ 0.0590). Rates of composite stroke and systemic embolism for rivaroxaban and warfarin users were 4.6 events and 5.9 events per 100 person-years (HR ¼ 0.77, 95% CI: 0.55–1.09, p ¼ 0.1364). Rates of VTE were significantly lower for rivaroxaban compared to warfarin users, at 2.9 and 7.1 events per 100 person-years (HR ¼ 0.36, 95% CI: 0.24–0.54, p50.0001), respectively.
Treatment persistence Rivaroxaban users had a significantly higher persistence with therapy after 3 months (84.5% vs. 75.6%, p50.0001; Table 3) and after 6 months (81.5% vs. 68.3%, p50.0001). Figure 3 presents the Kaplan–Meier curves for the two cohorts. Rivaroxaban users had a 1322
Comparative rivaroxaban and warfarin in AF patients Laliberte´ et al.
This study used real-world claims data from a US population to compare the safety and effectiveness of rivaroxaban with warfarin therapy in an adult nonvalvular AF population at risk for stroke. No significant differences were observed between rivaroxaban and warfarin in terms of safety and effectiveness regarding major bleeding, ICH, GI bleeding, and composite stroke and systemic embolism outcomes, but rivaroxaban was associated with significantly lower rates of VTE and with a higher persistence with therapy. Anticoagulant persistence is of importance for nonvalvular AF patients, especially given the severity of potential adverse events such as stroke. The persistence rate below 70% the authors found at 6 months for newly initiated warfarin users is worrisome. Potential explanations for the lower persistence associated with warfarin relates to its numerous limitations for chronic use13. Complexity of warfarin chronic use is caused by its natural limitations of narrow therapeutic range and constant need for monitoring14. Also, pharmacology sensitivity to both dietary and physiologic changes, as well as drug–drug interactions, makes monitoring even more important and also a primal cause of the increased likelihood of non-adherence15. Simpler regimens have been shown to promote medication adherence and persistence with cardiovascular www.cmrojournal.com ! 2014 Informa UK Ltd
Current Medical Research & Opinion Volume 30, Number 7
July 2014
Major bleeding
1.08 (0.71 - 1.64)
ICH
1.17 (0.66 - 2.05)
GI bleeding Composite strike and systemic embolisma Ischemic strokea
1.27 (0.99 - 1.63)
Hemorrhagic strokea
1.11 (0.13 - 9.60)
Systemic embolisma
0.54 (0.13 - 2.33)
VTEb
0.36 (0.24 - 0.54)*
DVT onlyb PE with or without DVTb
0.33 (0.21 - 0.53)*
0.77 (0.55 - 1.09) 0.81 (0.57 - 1.14)
0.64 (0.35 - 1.18)
0
0.5
1.5
1
Rivaroxaban better
2
2.5
3
Warfarin better
ICH: intracranial hemorrhage; GI: gastrointestinal; VTE: venous thromboembolism; DVT: deep vein thrombosis; PE: pulmonary embolism. *Statistically significant at 0.05 level. aEvent identified during a hospitalization or emergency department visit. bEvent identified during a hospitalization, an emergency department visit, or an outpatient visit (with a 6 month washout).
Figure 2. Hazard ratios of bleeding, stroke and VTE events for rivaroxaban compared to warfarin users.
100%
Proportion of patients persistent with treatment
Curr Med Res Opin Downloaded from informahealthcare.com by University of Laval on 07/06/14 For personal use only.
Rivaroxaban vs. Warfarin Hazard ratios [95% CI]
Hazard Ratio: 0.66 (0.60 - 0.72) P
