Current Medical Research and Opinion

Effectiveness and Safety of Rivaroxaban Versus Warfarin in Obese Nonvalvular Atrial Fibrillation Patients: Analysis of Electronic Health Record Data

Olivia S. Costa, Jan Beyer-Westendorf, Veronica Ashton, Dejan Milentijevic, Kenneth Todd Moore, Thomas J. Bunz & Craig I. Coleman

To cite this article: Olivia S. Costa, Jan Beyer-Westendorf, Veronica Ashton, Dejan Milentijevic, Kenneth Todd Moore, Thomas J. Bunz & Craig I. Coleman (2020): Effectiveness and
Safety of Rivaroxaban Versus Warfarin in Obese Nonvalvular Atrial Fibrillation Patients: Analysis of Electronic Health Record Data, Current Medical Research and Opinion, DOI: 10.1080/03007995.2020.1762554
To link to this article: https://doi.org/10.1080/03007995.2020.1762554

© 2020 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis

Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=icmo20

Effectiveness and Safety of Rivaroxaban Versus Warfarin in Obese Nonvalvular Atrial Fibrillation Patients: Analysis of Electronic Health Record Data

Olivia S. Costaa,b, Jan Beyer-Westendorfc, Veronica Ashtond, Dejan Milentijevicd, Kenneth Todd Mooree,
Thomas J. Bunzf, Craig I. Colemana,b

a Department of Pharmacy Practice, University of Connecticut School of Pharmacy, Storrs, CT, USA
b Evidence-Based Practice Center, Hartford Hospital, Hartford, CT, USA
c Thrombosis Unit, University Hospital “Carl Gustav Carus”, Technical University Dresden, Dresden,
Germany
d Janssen Scientific Affairs, LLC, Titusville, NJ, USA
e Janssen Pharmaceuticals, Inc, Titusville, NJ, USA
f New England Health Analytics, LLC, Granby, CT, USA

Correspondence: Craig I. Coleman, University of Connecticut, School of Pharmacy, 69 North Eagleville Road, Unit 3092, Storrs, CT 06269, USA. Email: [email protected]

Transparency
Declaration of funding
This study was funded by Janssen Scientific Affairs LLC.

Declaration of financial/other relationships
CIC has received grant funding and consultancy fees from Janssen Scientific Affairs LLC, Titusville, NJ, Bayer AG, Berlin, Germany and Portola Pharmaceuticals, South San Francisco, CA, as well as speaker fees from Medscape Inc. JB-W has received grant funding and consultancy fees from Bayer AG, Berlin, Germany; Boehringer Ingelheim Pharmaceuticals, Ridgefield, CT; Pfizer New York, NY; Daiichi Sankyo, Basking Ridge, NJ; and Portola Pharmaceuticals, South San Francisco, CA. VA and DM are employees of Janssen Scientific Affairs LLC, Titusville, NJ. KTM is an employee of Janssen Pharmaceuticals Inc., Titusville, NJ. OSC and TJB have no declarations-of-interest. Peer reviewers on this manuscript have received an honorarium from CMRO for their review work but have no other relevant financial relationships to disclose.

Acknowledgements
None.

Abstract
Background: Although rivaroxaban has demonstrated consistent drug levels in normal weight and obese patients, sufficient confirmation of equal clinical effectiveness and safety is currently lacking.

Purpose: To evaluate the effectiveness and safety of rivaroxaban versus warfarin for prevention of stroke and systemic embolism (SSE) in obese nonvalvular atrial fibrillation (NVAF) patients.

Methods: Using Optum de-identified Electronic Health Record (EHR) data from 11/2011-9/2018,we evaluated NVAF patients with a body mass index (BMI)≥30 kg/m2 newly initiated on rivaroxaban or warfarin (index date), with ≥12-months of EHR activity and ≥1 encounter before the index date. We excluded patients with valvular disease or evidence of oral anticoagulant (OAC) use at baseline. Patients who were prescribed rivaroxaban were 1:1 propensity-score matched to patients who were prescribed warfarin (standard differences <0.10 achieved for all covariates). Outcomes included SSE and major bleeding using an intent-to-treat approach. Subanalyses stratified by BMI (30.0-34.9, 35.0-39.9 and ≥40 kg/m2) were performed. Cox regression was performed and reported as hazard ratios (HRs) and 95% confidence intervals (CIs).

Results: We included 35,613 rivaroxaban and 35,613 warfarin users with NVAF. Patients were followed for a median of 2.6 years (25%-75% range=1.2–4.1). Rivaroxaban was associated with a reduced risk of SSE (HR=0.83, 95%CI=0.73-0.94) and major bleeding (HR=0.82, 95%CI=0.75-0.89) compared to warfarin. Subanalysis did not show a statistically significant interaction across BMI categories for SSE (p- interaction=0.58) or major bleeding (p-interaction=0.44) outcomes.

Conclusions: Among obese NVAF patients, prescription of rivaroxaban was associated with a reduced risk of SSE and major bleeding compared to warfarin, which remained consistent across BMI classes.

Keywords: obesity; atrial fibrillation; rivaroxaban; warfarin; anticoagulants

INTRODUCTION
Non-vitamin K oral anticoagulants (NOACs), including rivaroxaban, are now recommended as a first-line treatment to reduce the risk of stroke and systemic embolism in patients with nonvalvular atrial fibrillation (NVAF) [1]. While there are several comorbid conditions identified as risk factors for NVAF and stroke, obesity has been known to play a prominent independent role [2,3]. An important consideration when treating NVAF patients with obesity, is that the pharmacological characteristics of a drug may be affected by the patient’s body mass index (BMI) or overall weight. Therefore, during clinical drug development, phase 1 healthy volunteer studies assessing these potential effects are typically conducted, the results of which then inform the labeling of the drug. Very rarely are large Phase 3 randomized clinical trials conducted to assess the efficacy and safety of the drug solely in this population, unless the drug in question is meant to treat obesity.

The potential effects of obesity on the clinical pharmacology, efficacy and safety of rivaroxaban were assessed during the clinical development of the compound. Initially a phase 1 study was conducted in healthy volunteers and showed that high BMI or weight did not substantially affect the clinical pharmacology of the compound [4]. Furthermore, in the pivotal phase 3 safety and efficacy trial, ROCKET-AF, a sub-group analysis of BMI found comparable safety and efficacy [5]. Additional results from prospective and retrospective cohort studies have reported clinical outcomes for NOAC therapy and did not indicate an impact of obesity on NOAC effectiveness or safety. However, these studies were small and lacked statistical power, especially for patients with morbid obesity (i.e. BMI ≥ 40kg/m2 or weight >120 kg) [6-9].

Given that large enough prospective cohort studies are likely not feasible to compare the effectiveness and safety of NOACS in morbid obesity, retrospective claims data or electronic health record (EHR) analyses are helpful as these identify and compare treatment effects in very large cohorts. A recent real- world study provided data supporting the effectiveness and safety of rivaroxaban versus warfarin in morbidly obese patients with NVAF [10]. This study, however, was primarily focused on patients with a BMI ≥40 kg/m2 and depended on billing codes alone, which are prone to misclassification bias [11], to identify morbid obese patients. Thus, the objective of this study was to evaluate the effectiveness and safety of rivaroxaban compared to warfarin in obese patients with NVAF using patient-level BMI values.

METHODS
The report for this analysis was written to comply with the Reporting of Studies Conducted using Observational Routinely Collected Health Data (RECORD) statement [12].

We performed a cohort analysis using US Optum® de-Identified Electronic Health Record (EHR) data from November 1, 2010 through September 30, 2018 [13]. The Optum® EHR database includes longitudinal patient-level medical record data for 97 million patients seen at 700 hospitals and 7,000 clinics across the United States. The database includes records of prescriptions as prescribed and administered, laboratory results, vital signs, body measurements, diagnoses and procedures.

Patients with a NVAF diagnosis with BMI>30 kg/m2 initiated on rivaroxaban or warfarin, were included in the study. All patients needed to have ≥12-months of EHR activity prior to the index date and received care documented in the EHR database from at least one provider in the 12-months prior. We excluded patients with an index date prior to November 2011 (rivaroxaban’s US approval date), evidence of valvular heart disease (mitral stenosis or any valve replacement), an additional indication for OAC use (e.g., venous thromboembolism, hip or knee replacement surgery) or any prior OAC utilization per written prescription or patient self-report at baseline. Obesity was defined as having a BMI ≥30 kg/m2 and morbid obesity was defined as having a BMI ≥40 kg/m2 [14] using patient anthropometric measurements closest to the index date and within the 12-month baseline period. As Optum® EHR data [13] provides patient level BMI observations, the use of claims-based identification of obesity was not required. Obesity status was further sub-stratified using definitions provided by the National Heart, Lung and Blood Institute (NHLBI) (class 1 = 30-34.9, class 2 = 35-39.9, class 3 ≥40 kg/m2) using patient- level BMI data [14].

Patients prescribed rivaroxaban were 1:1 matched to warfarin patients based on propensity scores calculated via multivariable logistic regression [15]. Propensity score calculations were based on variables and risk factors affecting differential oral anticoagulation use or outcomes including demographics, comorbidities and medications which might serve as confounders for thrombosis (e.g., heart failure, diabetes, prior stroke, etc.) or bleeding (e.g., aspirin non-steroidal anti-inflammatory drugs, systemic corticosteroids, proton pump inhibitors, etc.) at baseline (Table 1, Supplementary Materials, eTables 1-3). Separate models were fit for the primary, secondary and exploratory (subgroup) endpoint analyses. Residual differences in covariates between the matched cohorts (using a caliper=0.2 standard deviations of the logit of the propensity score) were assessed by calculating standardized differences. Standardized differences <0.1 indicated cohorts were well-balanced for the corresponding covariates [15]. Plots for visual inspection of propensity score distributions after matching were also reviewed. Propensity score matching was performed using the ‘MatchIT’ package and R v3.4.3 (The R Project for Statistical Computing).

The primary effectiveness endpoint was the incidence of SSE defined by an appropriate inpatient discharge ICD-9 (433.01, 433.11, 433.21, 433.31, 433.81, 433.91, 434.01, 434.11, 434.91, 436.x) or -10
(I63) diagnosis code in the primary coding position. Our primary safety endpoint was major bleeding using the validated Cunningham algorithm [16]. Secondary endpoints included ischemic stroke, intracranial hemorrhage (ICH) and extracranial bleeding as separate endpoints. We utilized an intent-to- treat approach to analyze the data; in which, patients were followed until occurrence of endpoint, end- of-EHR activity or through the end of data availability (September 30, 2018).

Descriptive statistics were used to analyze baseline characteristics. Continuous data were reported as medians with interquartile ranges (or 25%, 75% range). Categorical data were reported as proportions. Outcome incidence rates were reported as events/100 person-years (or %/year). Cox regression models were fit to compare event rates for the matched rivaroxaban and warfarin cohorts and reported as hazard ratios (HRs) along with accompanying 95% confidence intervals (CIs). The proportional hazard assumption was tested based on Schoenfeld residuals (and found valid for all outcomes). All statistical analysis was performed using IBM SPSS v26.0 (IBM Corporation, Armonk, New York, NY). A p-value
<0.05 was considered statistically significant unless otherwise noted.

We performed exploratory subgroup analyses to determine whether patient characteristics including age (≥75 years or <75 years), sex, obesity class (I, II or III) [14], the presence or absence of comorbid diabetes, presence or absence of concomitant antiplatelet use and CHA2Ds2-VASc score (<4 or ≥4) had significant influence on our study results. The presence of statistical interactions on SSE and major bleeding across subgroup analyses were tested using the methods described by Altman and Bland [17]. To reduce the chances of obtaining false-positive results (Type I error) as a result of our multiple hypothesis testing/subgroup analyses, we utilized a Bonferroni corrected P-value of <0.0125 to suggest a statistically significant subgroup interaction for the SSE or major bleeding endpoints [18].

RESULTS
A total of 35,613 rivaroxaban and 35,613 warfarin patients with NVAF and a BMI ≥30 kg/m2 were included in the study (Figure 1). Forty-eight percent of patients (n=34,174) had a BMI of 30.0-34.9 kg/m2, 27% (n=19,017) a BMI of 35.0-39.9 kg/m2 and 25% (n=18,034) a BMI≥ 40 kg/m2 (~28% had a body weight >120 kg). At baseline, the median (25%, 75% range) patient age was 68 (61, 75) years, median
CHA2DS2-VASc score was 3 (2, 4), 40% were female and 56% were receiving concomitant antiplatelet therapy. Of the patients prescribed rivaroxaban, 14.0% (n=4,978) were prescribed the low dose (15 mg) with the remainder prescribed the standard 20 mg dose. The two cohorts were well balanced based on propensity score distributions between the rivaroxaban and warfarin cohorts appeared similar (Supplementary Materials, eFigure 1). All baseline characteristics had an absolute standardized difference <0.1 between cohorts. Patients were followed for a median of 2.6 (1.2, 4.1) years (median rivaroxaban=2.1 years versus median warfarin=3.1 years, p<0.01).

Upon Cox regression, rivaroxaban was associated with a significantly reduced hazard of SSE (HR=0.83, 95%CI=0.73-0.94) and major bleeding (HR=0.82, 95%CI=0.75-0.89) compared to warfarin (Figure 2). Because the duration of available follow-up between cohorts was longer in the warfarin group, we performed a post-hoc sensitivity analysis whereby we capped the duration of follow-up at 2-years [median follow-up 2.0 (1.2, 2.0) years, rivaroxaban=2.0 versus warfarin=2.0 years]. Two-year capped results were consistent with the base case findings (HR=0.86, 95%CI=0.74-1.00 for SSE and HR=0.86, 95%CI=0.78-0.96 for major bleeding). Rivaroxaban was associated with a reduced risk of ischemic stroke alone versus warfarin (HR=0.89, 95%CI=0.78–1.01) but did not reach statistical significance. Both ICH (HR=0.62, 95%CI=0.47–0.81) and extracranial bleeding (HR=0.85, 95%CI=0.78–0.93) were associated with a significant reduction with rivaroxaban compared to warfarin (Table 2). Specifically, gastrointestinal bleeding was associated with a significant 30% reduction with rivaroxaban versus warfarin (1.83% vs. 2.62%, HR=0.70, 95%CI=0.63-0.77).

Exploratory analyses did not show a statistically significant interaction across BMI categories for the SSE (p-interaction=0.58) or major bleeding (p-interaction=0.44) outcomes (Figure 2). Restricting inclusion to patients with a BMI≥35 kg/m2 also did not impact our study’s overall conclusions regarding the comparative impact of rivaroxaban versus warfarin on SSE (HR=0.90, 95%CI=0.76–1.07) or major bleeding (HR=0.77, 95%CI=0.68–0.86) (Table 2). Other exploratory subgroup analyses stratified by age (≥75 years or <75 years), sex, presence or absence of comorbid diabetes, presence or absence of concomitant antiplatelet therapy and CHA2Ds2-VASc score (<4 or ≥4) had no significant influence on our study’s results (Tables 3 and 4).

DISCUSSION
In this large (71,000+ person), real-world EHR study with a mean of over 2 1/2 years of patient follow- up, we found prescription of rivaroxaban to be associated with a significant 17% reduction in the hazard of SSE and 18% reduction in major bleeding compared to warfarin in NVAF patients with a BMI≥30 kg/m2. There was no significant difference across BMI classes for these findings. Our findings also remained robust upon subanalyses stratifying patients by age, sex and the presence or absence of diabetes. Intra- and extra-cranial bleeding were associated with a significant risk reduction with rivaroxaban compared to warfarin by 38% and 15%, respectively. While rivaroxaban was associated with a reduced risk of ischemic stroke alone (1.12%) compared to warfarin (1.65%), this change was not considered statistically significant as the 95%CI crossed the line of unity (95%CI=0.78–1.01).

The results of the present study build upon and are supported by prior published literature [4-10]. Kubitza and colleagues [4] evaluated the effect of extreme elevated body weight (>120 kg) on pharmacokinetics (PK) and pharmacodynamics (PD) of rivaroxaban 10 mg compared to those of normal body weight. This randomized, single-blind, placebo-controlled, parallel-group study in 48 healthy subjects found rivaroxaban’s pharmacological profile was not affected by extremes in BMI or weight in a clinically relevant manner. Consequently, no dosage adjustment based on BMI or body weight is indicated for rivaroxaban in its product labeling [16]. The ROCKET subanlaysis [5] evaluated the impact of BMI (18.5-24.9 kg/m2, n=3,289; 25.0-29.9 kg/m2, n=5,535; ≥30 kg/m2, n=5,206) on the efficacy and safety of rivaroxaban versus warfarin as part of a post-hoc sub-analysis of 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). These authors reported incidences of SSE and major bleeding to be 1.88%/year and 3.33%/year in obese patients (not dissimilar to our own), with no statistical interaction across BMI classifications for any of the primary endpoints including SSE (p- interaction=0.40), ischemic stroke (p-interaction=0.11), major bleeding (p-interaction=0.54) or ICH (p- interaction=0.08). More recently, Peterson and colleagues [10] performed a retrospective cohort study using data from IBM Watson MarketScan Commercial Claims and Encounters and Medicare Supplemental databases from December 1, 2010 through December 31, 2016. After propensity-score matching 7,126 morbidly obese patients (per International Classification of Diseases-9th or 10th-Revision coding), they found the risks of ischemic stroke/systemic embolism and major bleeding to be similar for rivaroxaban and warfarin users (ischemic stroke/systemic embolism: 1.5% versus 1.7%; odds ratio [OR]=0.88; 95%CI=0.60-1.28; major bleeding: 2.2% versus 2.7%; OR=0.80, 95%CI=0.59-1.08). However, this study was limited by diagnosis codes for which obesity claims have exhibited a sensitivity of 7.75%, specificity of 98.98%, NPV of 80.84% and PPV of 65.94% [11]. Therefore, while patients identified as morbidly obese in the study by Peterson et al. likely were morbidly obese, their dependence of billing codes likely meant they failed to identify many other eligible but uncoded cases of morbid obesity which lowered the sample size (and statistical power) and may have impacted the generalizability of their findings.

While as of 2016, the ISTH [20] recommends NOACs (including rivaroxaban) not be used in morbidly obese patients, citing insufficient data; a breadth of new clinical data supporting rivaroxaban’s use in obese and morbidly obese patients from randomized [5], observational [6-9] claims database [10] and now EHR studies (such as this current study, which reports rivaroxaban to be no worse than warfarin across effectiveness and safety endpoints in 18,000+ NVAF patients with a BMI ≥40 kg/m2 or body weight >120 kg) have been published. It is uncertain if the outcomes observed with rivaroxaban in our study are representative of all NOACs (i.e., a class effect) as there is a relative paucity of published data evaluating other NOACs. Additionally, pharmacokinetic and pharmacodynamic data for other NOACs in obese patients, for example apixaban, has shown decreased maximum drug concentration, exposure and anti-factor Xa activity in patients with a body weight ≥ 120kg [21]. Whether these pharmacokinetic findings equate to differences in clinical outcomes among obese NVAF patients is unclear. Moreover, existing data suggests the use of warfarin in obese/morbidly obese NVAF patients is associated with its own challenges, including a greater difficulty obtaining the recommended international normalized range (INR), a need for higher warfarin doses and concerns that common weight loss/dietary change recommendations may interfere with patients’ ability to maintain target INR. [22-24]. The ease of NOAC use compared to warfarin and their consistent effectiveness and safety results (particularly rivaroxaban), places NOACs in an ideal position to support the care of this population. As with all research, there are limitations of this study that should be addressed. Non-randomized studies can be impacted by confounding, misclassification and sampling biases which have the potentialto effect studies’ internal validity [25.

Our study also utilizes EHR data from US patients only [9] making our findings most applicable to this population. Time in therapeutic international normalized ratio (INR) was not calculated for warfarin patients (though previous studies suggest US NVAF patients spend only about 55% of time in their therapeutic INR range)[26], and perhaps more importantly, we could not determine clinician’s intended target INR (although often assumed to be 2.0-3.0).. Due to the lack of INR data (and confirmation of target range) and the relatively small proportion of rivaroxaban patients receiving a low dose, we did not attempt to perform subgroup analysis by rivaroxaban dose. Although Optum® EHR covers insured and uninsured patients, it does not cover all institutions and therefore follow-up events may be missed. The EHR only reports estimated patient income values based on patient’s zip code (as opposed to actual patient-specific values) and does not provide any additional data regarding soci0economic status; therefore we could not investigate the potential impact of such factors on our results. Reliable mortality data was not available from the EHR database used. However, a previous study demonstrated that mortality in NVAF patients is generally not associated with the anticoagulant (either stroke or bleeding), but is typically a result of heart failure, cancer or pneumonia [27]. Lastly an EHR entry indicating initiation/use of an OAC does not guarantee a patient took it or allow for assessment of adherence or persistence due to the lack of prescription claims data capture. .

Subsequently, we only performed analyses using an intention-to-treat approach.
Despite these limitations, our study has several strengths. By using EHR data which included patients from different geographical areas of the US and encompassing commercial, Medicare and uninsured patients, we were likely able to identify an obese NVAF population that resembles that seen in the entire US. Additionally, use of EHR data allowed us to directly determine BMI rather than depend solely on the presence of billing codes (which have been shown to underreport obesity in administrative data) [11,13]. EHR data also allowed us to take into account patient and physician reported medications that may not be captured in claims data and include these in our propensity score matching algorithm. The use of EHR data likely resulted in more accurate (with a greater proportion of patients) reporting use of over-the-counter medications such as aspirin. Notably, however, the present study and previous studies suggest the relative safety of rivaroxaban versus warfarin is not impacted by antiplatelet use [28].
When dependence on billing codes was required to identify covariates or endpoints, we utilized validated algorithms whenever possible. Lastly, through implementation of propensity-score matching, we were able to assure the rivaroxaban and warfarin cohorts were well-balanced on many important covariates known to be a risk factors of SSE and/or major bleeding.

CONCLUSION
Among obese NVAF patients, prescription of rivaroxaban was associated with a significantly reduced risk of SSE and major bleeding compared to warfarin, including both intra- and extracranial bleeding. These results remained consistent across BMI classes, including the morbidly obese; as well as across subgroups stratified by age, sex and diabetes. The results reported here along with additional, previously published real-world data [6-10] and subpopulation analysis of ROCKET-AF [5] are consistent with the ROCKET-AF pivotal trial and, should provide clinicians with greater comfort that there is no
detrimental impact of BMI on rivaroxaban’s effectiveness or safety.

REFERENCES
1. January CT, Wann LS, Calkin H, et al. 2019 AHA/ACC/HRS focused update of the 2014 AHA/ACC/HRS guideline for the management of patients With atrial fibrillation: A report of the American College of Cardiology/American Heart Association Task Force on clinical practice guidelines and the Heart Rhythm Society in collaboration with the Society of Thoracic Surgeons. J Am Coll Cardiol. 2019; 9: 104-132.
2. Vyas V, Lambiase P. Obesity and atrial fibrillation: Epidemiology, pathophysiology and novel therapeutic opportunities. Arrhythm Electrophysiol Rev. 2019; 8: 28-36.
3. Kernan WN, Inzucchi SE, Sawan C,et al.. A stubbornly obvious target for stroke prevention. Stroke. 2013; 44: 278-286.
4. Kubitza D, Becka M, Zuehlsdorf M, et al. Body weight has limited influence on the safety, tolerability, pharmacokinetics, or pharmacodynamics of rivaroxaban (BAY 59-7939) in healthy subjects. J Clin Pharmacol 2007;47:218-26.
5. Balla SR, Cyr DD, Lokhnygina Y, et al. Relation of risk of stroke in patients with atrial fibrillation to body mass index (from patients treated with rivaroxaban and warfarin in the rivaroxaban once daily oral direct factor Xa inhibition compared with vitamin K antagonism for prevention of stroke and embolism trial in atrial fibrillation trial). Am J Cardiol 2017;119:1989-96.
6. Kalani C, Awudi E, Alexander T, et al. Evaluation of the efficacy of direct oral anticoagulants (DOACs) in comparison to warfarin in morbidly obese patients. Hosp Pract(1995); 2019; 44:181-185
7. Kushnir M, Choi Y, Eisenberg R, et al. Efficacy and safety of direct oral factor Xa inhibitors compared with warfarin in patients with morbid obesity: a single-centre retrospective analysis of chart data. Lancet Haematol. 2019; 6: e359-e365.
8. Kido K, Ngorsuraches S. Comparing the efficacy and safety of direct oral anticoagulants with warfarin in the morbidly obese population with atrial fibrillation. Ann Pharmacother. 2019; 53:165- 170.
9. Tittl L, Endig S, Marten S, et al. Impact of BMI on clinical outcomes of NOAC therapy in daily care- results of the prospective Dresden NOAC registry (NCT01588119). Int J Cardiol. 2018; 262:85-91
10. Peterson ED, Ashton V, Chen YW, et al.. Comparative effectiveness, safety, and costs of rivaroxaban and warfarin among morbidly obese patients with atrial fibrillation. Am Heart J. 2019;212:113-119.
11. Martin BJ, Chen G, Graham M, et al. Coding of obesity in administrative hospital discharge abstract data: accuracy and impact for future research studies. BMC Health Serv Res. 2014;14:70.
12. Benchimol EI, Smeeth L, Guttmann A, et al. RECORD Working Committee. The REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) statement. PLoS Med. 2015; 12:e1001885.
13. Optum. Optum EHR Offering. Optum Inc, 2018. Available at: https://www.optum.com/campaign/ls/data-new-era-of-visibility/download.html (Last accessed on July 28, 2019).
14. National Heart, Lung and Blood Institute. Classification of Overweight and Obesity by BMI, Waist Circumference, and Associated Disease Risks. Available at: https://www.nhlbi.nih.gov/ health/educational/lose_wt/BMI/bmi_dis.htm (Last accessed on September 25, 2019).
15. Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res 2011;46:399–424.
16. Cunningham A, Stein CM, Chung CP, et al. An automated database case definition for serious bleeding related to oral anticoagulant use. Pharmacoepidemiol Drug Saf. 2011;20:560-566.
17. Altman DG, Bland JM. Interaction revisited: the difference between two estimates. BMJ 2003;326:219.
18. Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Society Ser B 1995;57:289–300.

19. XARELTO® (rivaroxaban) tablets, for oral use. Janssen Pharmaceuticals, Inc. Titusville, NJ, 2019
20. Martin K, Beyer-Westendorf J, Davidson BL, et al. Use of the direct oral anticoagulants in obese patients: guidance from the SSC of the ISTH. J Thromb Haemost 2016;14:1308-13.
21. Upreti VV, Wang J, Barrett YC, et al. Effect of extremes of body weight on the pharmacokinetics, pharmacodynamics, safety and tolerability of apixaban in healthy subjects. Br J Clin Pharmacol. 2013; 76:908-916.
22. Wallace JL, Reaves AB, Tolley EA, et al. Comparison of initial warfarin response in obese patients versus non obese patients. J Thromb Thrombolysis. 2013:36;96 101.
23. Jaffer A, Bragg L. Practical tips for warfarin dosing and monitoring. Cleve Clin J Med. 2003;70:361 371.
24. Moll S, Crona DJ, Martin K. Direct oral anticoagulants in extremely obese patients: OK to use? Res Pract Thromb Haemost. 2018;3:152-155.
25. Gandhi SK, Salmon W, Kong SX, et al. Administrative databases and outcomes assessment: an overview of issues and potential utility. J Manag Care Spec Pharm. 1999;5:215-222.
26. Baker WL, Cios DA, Sander SD, et al. Meta-analysis to assess the Rivaroxaban quality of warfarin control in atrial fibrillation patients in the United States. J Manag Care Pharm 2009;15:244-252.
27. Camm AJ, Amarenco P, Haas S, et al. XANTUS: A real-world prospective, observational study of patients treated with rivaroxaban for stroke prevention in atrial fibrillation. Eur Heart J. 2016; 37: 1145-153.
28. Shah R, Helkamp A, Lokhnygina Y, et al. Use of concomitant aspirin in patients with atrial fibrillation: findings from the ROCKET AF trial. Am Heart J. 2016; 179: 77-86.