Healthcare costs associated with rivaroxaban or warfarin use for the treatment of venous thromboembolism

Abstract

Introduction: Rivaroxaban has been shown to have similar efficacy but less major bleeding than warfarin in randomized trials of patients experiencing venous thromboembolism (VTE). This report sought to assess healthcare costs up to 12-months following an index VTE in patients prescribed either rivaroxaban or warfarin.

Materials and methods: This study analyzed claims from the MarketScan Commercial Claims and Encounters Database from November 2011–July 2015. It selected adults newly-diagnosed with VTE (deep vein thrombosis [DVT] or pulmonary embolism [PE]) if they had an outpatient prescription claim for rivaroxaban or warfarin within 7-days of the index event. Warfarin users were 2:1 propensity-score matched to rivaroxaban users and followed until the end of insurance coverage, end of data availability or 12-months of follow-up. Total per patient healthcare costs, including inpatient, outpatient, and overall pharmacy costs, were compared using a multivariable generalized linear model.

Results: In total, 10,929 rivaroxaban patients were matched to 21,858 warfarin patients. Mean follow-up for rivaroxaban and warfarin patients was 317- and 321-days for those experiencing an index DVT, and 313- and 318-days for those with PE. Mean overall treatment costs per patient were lower for rivaroxaban vs warfarin users (−$1,116, p = .0016). This cost difference was driven by lower inpatient (−$622) and outpatient (−$1,156) treatment costs, and the higher pharmacy costs ($661) were, therefore, fully offset. Results were similar when analysis was restricted to DVT patients. No significant difference in total costs was observed in patients experiencing an index PE.

Limitations: Claims databases are subject to inaccuracies and missing data. Prescription claims may not fully reflect actual medication utilization. Despite propensity-score matching and regression, residual confounding cannot be excluded.

Conclusions: Rivaroxaban was associated with significantly lower total per patient VTE treatment costs, despite higher pharmacy costs. These savings are the result of decreased inpatient and outpatient healthcare utilization costs associated with rivaroxaban.

Keywords:

• Healthcare costs
• venous thromboembolism
• rivaroxaban
• warfarin
• deep vein thrombosis
• pulmonary embolism

Introduction

Venous thromboembolism (VTE), defined as deep vein thrombosis (DVT) or pulmonary embolism (PE), is a major health problem in the US, with over 900,000 VTE events recorded annually, including 600,000 occurring in a hospital setting and ∼100,000 deaths¹. Despite adequate anticoagulation, studies estimate the cumulative incidence of recurrent VTE to be 4.9% at 3-months, 8.6% at 6-months and as high as 17.5% 2-years after the index event^2–4.

Randomized controlled trials of rivaroxaban have shown it to prevent recurrent VTE to a similar extent as warfarin^5,6, while reducing patients’ risk of major bleeding⁷. At present, little is known about the overall cost of VTE treatment in the first 12 months after the index event in patients receiving different oral anticoagulants.

The current study seeks to estimate the overall per patient healthcare costs during the 12 months following an index event in patients prescribed either rivaroxaban or warfarin.

Methods

We analyzed healthcare claims from the Truven Health MarketScan Commercial Claims and Encounters Database⁸ from November 2011 through July 2015. This database included active employees, early retirees, COBRA continuers, and their dependents insured by employer-sponsored health plans. Truven Health Analytics MarketScan databases are de-identified and fully compliant with all HIPAA privacy and security requirements to protect patient anonymity and confidentiality.

We used a retrospective longitudinal cohort study design to quantify total per patient healthcare costs including inpatient, outpatient, and all-cause pharmacy costs during the 12 months after an index VTE event among patients newly initiated on anticoagulation with either rivaroxaban or warfarin. Patients included in the study had ≥1 primary diagnosis of VTE (International Classification of Disease, 9th-Revision, Clinical Modification [ICD-9-CM]: DVT = 451.1, 451.2, 453.4, 453.8, 453.9, or PE = 415.1x) beginning November 2012 (date rivaroxaban was approved for the DVT/PE indications) during an outpatient, emergency department, or inpatient visit. In addition, they had to have at least 12-months of continuous eligibility prior to the first VTE event (baseline period), be ≥18-years of age at the first VTE event (index date), and have had ≥1 dispensing for either rivaroxaban or warfarin within 7-days of the index VTE event. The study observation period spanned up until 12-months, a patient’s end of continuous health plan enrollment (defined as a 2-month gap), end of data availability, or death, whichever occurred first. In order to identify a cohort of patients who had not recently experienced a VTE and who were treatment naïve, we excluded patients with a VTE diagnosis code in any position or receiving rivaroxaban, warfarin, or low-molecular weight heparin prescriptions during the 12-months prior to, but not during hospitalization for the index VTE event. In addition, we excluded patients with apixaban or dabigatran prescriptions at any time during the study or those initiating both warfarin and rivaroxaban on the same day.

Due to the non-experimental nature of the study, it was anticipated that patients in the rivaroxaban and warfarin cohorts would vary in important characteristics. In order to minimize the impact of these confounding factors on study results, rivaroxaban patients were “nearest neighbor” matched 1:2 to the warfarin patients (stratified by index DVT only or PE) based upon propensity score⁹. We generated propensity scores using probability estimates from a logistic regression model including demographics (e.g. age, gender, insurance type), clinical characteristics (e.g. Quan-Charlson comorbidities¹⁰, VTE risk factors including arrhythmia, chronic obstructive pulmonary disease, major surgery, fracture, or use of oral contraceptives within 3 months, diabetes, hyperlipidemia, hypertension, multiple trauma, obesity, pneumonia, other serious infection, and baseline resource utilization) and hospital characteristics (e.g. region) to control for confounding factors such as hospital protocols and formularies. The adequacy of the matching procedure was assessed via absolute standardized differences comparing baseline characteristics between cohorts⁹. We considered an absolute standardized difference <0.1 well-balanced.

We summarized baseline demographics using percentages for categorical and means ± standard deviation for continuous variables. We compared total per patient healthcare costs including inpatient, outpatient, and any cause pharmacy costs using a generalized linear model (log-link and gamma distribution) adjusting for any characteristics not well balanced after propensity score matching. All analyses were performed in SAS version 9.3 (SAS Institute Inc., Cary, NC) with a p < .05 considered significant.

Results

In total, 10,929 rivaroxaban patients were 1:2 propensity score matched to 21,858 warfarin patients and most of the characteristics (except region and insurance type) were well balanced (Table 1). Mean follow-up for rivaroxaban and warfarin patients experiencing an index DVT was 317 and 321 days, respectively, and 313 and 318 days for those experiencing a PE.

For all patients experiencing VTE (DVT and/or PE), mean overall per patient treatment costs were lower for rivaroxaban compared to warfarin users (−$1,116, p = .0016) (Table 2). This cost difference was driven by decreased inpatient (−$622) and outpatient (−$1,156) treatment costs, and, therefore, the higher pharmacy costs ($661) were fully offset.

When evaluating DVT patients alone, mean overall per patient treatment costs were lower for rivaroxaban compared to warfarin users (−$1,370, p = .0002), and, again, were driven by both lower inpatient (−$875) and outpatient (−$1,160) treatment costs (Table 3). Total prescription drug costs were $664 higher in the rivaroxaban compared to warfarin cohort.

In the sub-group analysis of PE patients, no significant difference in mean total per patient treatment costs was observed (p = 0.7795) (Table 4). While a similar numerical reduction in outpatient costs was observed (−$1,148) with rivaroxaban, the difference in inpatients costs between rivaroxaban and warfarin-anticoagulated PE patients was attenuated (−$183) compared to DVT patients.

Discussion

In this present study we found mean overall per patient treatment costs following VTE to be significantly lower for rivaroxaban vs warfarin users. This cost difference was driven by both lower inpatient and outpatient treatment costs. Although total pharmacy costs were higher in the rivaroxaban-treated patients, this did not completely offset the economic benefits of rivaroxaban use. Upon analyzing DVT and PE patients separately, we found statistically significant economic benefit associated with using rivaroxaban among DVT patients. While PE patients who were anticoagulated with rivaroxaban appeared to realize similar decreases in outpatient costs vs warfarin users, economic benefits in regard to inpatient costs were also less.

The reductions in outpatient treatment costs among rivaroxaban anticoagulated VTE patients may be, at least in part, due to reductions in anticoagulation monitoring costs^11–13. A study by Mercaldi et al.¹¹ estimated warfarin patients required, on average, 12.9 international normalized ratio (INR) measurements annually. As a result of the need for this frequent monitoring and related healthcare provider contact related to anticoagulation monitoring, excess warfarin monitoring costs have been estimated to be as high as $1,400 per year^11–13.

A second potential explanation for reduced VTE treatment costs associated with rivaroxaban vs warfarin use may be the reduced risk of admission resulting from major bleeding with rivaroxaban vs warfarin^5–7. In the US, hospitalizations for major bleeds such as gastrointestinal bleeds and intracranial hemorrhage are estimated to cost an average of $9,757 and $19,658 per admission (in 2013 dollars), respectively¹⁴. In an a priori pooled analysis of the EINSTEIN DVT and PE studies⁷, rivaroxaban was shown to reduce patients risk of major bleeding compared to enoxaparin bridging to a vitamin K antagonist by 46% (1.0% vs 1.7%; hazard ratio = 0.54, 95% CI =0.37–0.79). Less major bleeding would likely result in less healthcare utilization including fewer hospitalizations and outpatient office visits; and, subsequently, reduced treatment costs. In fact, the difference in major bleeding rates between rivaroxaban and warfarin was shown to be a substantial driver of rivaroxaban’s cost-effectiveness in an economic modeling study by Seaman et al.¹⁵. Whether or not rivaroxaban was associated with less major bleeding in our study and to what extent lower bleeding rates may have drove observed reductions in VTE treatment costs was not evaluated in our study.

The current study was subject to certain limitations. As with any claims database analysis, there may be errors or omissions in coded procedures, diagnoses, or pharmacy claims including possible information or classification bias (e.g. identification of false positive VTE events). It is also possible that VTE events were under-coded (e.g. false negatives). Second, pharmacy claims data may not fully reflect patients’ actual medication use. Next, we did not have access to laboratory data in order to determine whether patients were discharged with a therapeutic INR, nor could we estimate the time patients spent in the therapeutic INR range during follow-up. Finally, despite propensity score matching and statistical adjustment using multivariable regression, residual confounding due to differences in unmeasured variables cannot be ruled out.

Conclusion

The overall 12-month per patient cost impact of rivaroxaban for VTE treatment may be favorable when compared to warfarin due to lower expenditure on healthcare resources. Rivaroxaban was associated with a significant reduction in post-VTE inpatient and outpatient treatment costs, but higher pharmacy costs vs warfarin. The decrease in inpatient treatment costs associated with rivaroxaban appeared qualitatively greater in DVT than in PE patients; while outpatient treatment costs were reduced similarly, regardless of whether the index event was a DVT or PE.

Transparency

Declaration of funding

This study was funded by Janssen Scientific Affairs, LLLC, Raritan, NJ.

Declaration of financial/other interests

CIC has received research grant funding or honorarium from Janssen Scientific Affairs, LLC, Bayer Pharma AG, Boehringer Ingelheim Pharmaceuticals. CB is a Roche Diagnostics Advisory Board member and received consulting fees from Janssen Pharmaceuticals. He also reports research support from Janssen Pharmaceuticals and Boehringer Ingelheim. CC, DM, SWW, and LL are employees of Janssen Scientific Affairs, LLC, and WWN was an employee of Janssen Scientific Affairs, LLC at the timing the research was conducted. JME peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Table 1. Characteristics of propensity-score matched rivaroxaban and warfarin patients experiencing an index deep vein thrombosis of pulmonary embolism.

Characteristic	Deep vein thrombosis			Pulmonary embolism
	Rivaroxaban % (n = 6,939)	Warfarin % (n = 13,878)	Standardized difference	Rivaroxaban % (n = 3,990)	Warfarin % (n = 7,980)	Standardized difference
Age, years (mean ± SD)	49.3 ± 10.6	49.7 ± 10.5	0.045	48.5 ± 11.5	49.3 ± 11.1	0.066
Male gender	56.4	55.8	0.012	50.6	50.6	0.001
Prior healthcare utilization			—			—
Hospitalizations (mean ± SD)	0.2 ± 0.6	0.3 ± 0.7	0.082	0.2 ± 0.7	0.3 ± 0.7	0.051
Emergency department visits (mean ± SD)	0.6 ± 1.5	0.7 ± 1.7	0.022	0.8 ± 2.2	0.7 ± 1.7	0.027
Outpatient contacts (mean ± SD)	43.1 ± 56.2	43.9 ± 61.2	0.014	44.2 ± 55.3	44.7 ± 61.6	0.010
Region			0.285			0.310
Northeast	15.7	18.2	—	15.1	18.6	—
North central	25.6	28.0	—	23.9	28.0	—
South	42.8	31.1	—	44.8	31.1	—
West	12.3	19.9	—	13.0	19.7	—
Unknown	3.6	2.8	—	3.3	2.6	—
Insurance type			0.101			0.103
Comprehensive insurance	2.8	3.0	—	3.0	3.3	—
EPO	0.8	0.8	—	1.0	0.7	—
HMO	9.3	11.7	—	9.8	12.1	—
POS	6.7	6.9	—	6.1	6.8	—
POS with capitation	62.9	61.4	—	62.4	60.3	—
CDHP	9.4	8.1	—	0.8	0.7	—
HDHP	6.0	5.5	—	9.7	8.5	—
Unknown	1.3	1.8	—	5.8	5.8	—
Comorbidities			—			—
Quan-Charlson score (mean ± SD)	0.9 ± 1.8	1.0 ± 1.9	0.052	1.0 ± 1.8	1.0 ± 1.9	0.019
Arrhythmia	6.2	7.6	0.052	7.6	8.3	0.024
Cancer	54.0	53.5	0.010	55.4	55.6	0.003
COPD	5.7	5.9	0.008	7.3	7.7	0.014
Oral contraception use	7.8	6.2	0.063	10.5	8.7	0.060
Diabetes	12.5	13.1	0.019	12.3	12.6	0.010
Fracture	6.3	5.7	0.025	3.3	3.6	0.014
Hyperlipidemia	29.1	28.2	0.020	29.6	28.0	0.036
Hypertension	32.7	32.7	0.001	34.7	33.4	0.027
Major surgery	50.6	52.6	0.041	52.6	53.7	0.022
Multiple trauma	7.3	6.6	0.027	4.3	4.7	0.021
Obesity	10.4	10.3	0.005	13.0	12.2	0.024
Pneumonia	4.8	4.6	0.011	8.4	7.5	0.036
Other serious infection	14.9	14.5	0.010	15.2	14.6	0.016

SD, standard deviation; EPO, Exclusive provider organization; HMO, Health maintenance organization; POS, Point of Service; PPO, Preferred provider organization; CDHP, Consumer-directed health plan; HDHP, High-deductible health plan; COPD, Chronic obstructive pulmonary disease.

Table 2. Cumulative 12-month per patient post-venous thromboembolism costs.

Mean costs	Rivaroxaban (n = 10,929)	Warfarin (n = 21,858)	p-value*
Total costs	$20,972	$22,088	.0016
Inpatient costs	$4,989	$5,611
Outpatient costs	$11,507	$12,663
Prescription costs	$4,476	$3,815

* p-value obtained by a general linear model with log link and gamma distribution, adjusting for any characteristics with standardized difference >10%.

Table 3. Cumulative 12-month per patient post-deep vein thrombosis costs.

Mean costs	Rivaroxaban (n = 6,939)	Warfarin (n = 13,878)	p-value*
Total costs	$19,487	$20,857	.0002
Inpatient costs	$4,209	$5,084
Outpatient costs	$11,015	$12,175
Prescription costs	$4,263	$3,599

* p-value obtained by a general linear model with log link and gamma distribution, adjusting for any characteristics with standardized difference >10%.

Table 4. Cumulative 12-month per patient post-pulmonary embolism costs.

Mean costs	Rivaroxaban (n = 3,990)	Warfarin (n = 7,980)	p-value*
Total costs	$23,554	$24,230	.7795
Inpatient costs	$6,345	$6,528
Outpatient costs	$12,363	$13,511
Prescription costs	$4,847	$4,190

* p-value obtained by a general linear model with log link and gamma distribution, adjusting for any characteristics with standardized difference >10%.