Indirect comparison and cost-utility of dabigatran etexilate and rivaroxaban in the treatment and extended anticoagulation of venous thromboembolism in a UK setting

Abstract

Background:

Acute venous thromboembolism (VTE), including deep vein thrombosis (DVT) and pulmonary embolism (PE), is traditionally managed with a short course of parenteral anticoagulation followed by 3–6 months of a vitamin-K antagonist. Non-vitamin K oral anticoagulants (NOACs) do not require routine monitoring and dose adjustment, thus potentially provide an alternative treatment option.

Methods and results:

Because of the lack of head-to-head clinical studies, an indirect comparison was conducted of dabigatran etexilate and rivaroxaban based on the respective phase III clinical trial. The derived relative safety and efficacy estimates were used to evaluate the cost-utility of dabigatran compared with rivaroxaban in the treatment and secondary prevention of VTE. The results of the indirect comparison showed no significant difference between dabigatran and rivaroxaban in avoiding recurrent VTE following index PE, index DVT, or DVT/PE combined, in treatment and extended anticoagulation. Dabigatran has significantly less major or clinically relevant bleeds (MCRBE) compared to rivaroxaban in treatment after index DVT and treatment after DVT or PE combined, but was not significantly different from rivaroxaban after index PE or in extended anticoagulation. In cost-utility deterministic analyses, dabigatran was projected dominant in all analyzed settings, given its marginally lower total cost and marginally higher QALYs gained compared to rivaroxaban. Probabilistic analyses results showed a high likelihood of dabigatran being considered good value for money in the UK, in treatment and in secondary prevention of VTE.

Conclusion:

The cost-effectiveness evaluations showed that dabigatran can be considered the dominant treatment strategy compared to rivaroxaban in the patients’ sub-groups considered, given the projected marginally higher clinical benefits and lower treatment costs.

Keywords::

• Venous thrombosis
• Dabigatran
• Rivaroxaban
• Indirect comparison
• Cost-effectiveness

Introduction

Venous thromboembolism (VTE) is associated with a high disease burden, due to mortality risks following acute pulmonary embolism (PE) and life-long complications such as chronic thromboembolic pulmonary hypertension (CTEPH), and post-thrombotic syndrome (PTS)1–3. Traditionally, VTE has been managed with short course heparin, followed by 3–6 months dose-adjusted vitamin K antagonists (VKAs)4. Unlike VKAs, non-vitamin K oral anticoagulants (NOACs) have the benefits of a predictable pharmacokinetic and pharmacodynamic profile, and allow fixed dose administration. Hence, they may provide an opportunity for longer and safer anticoagulation, and cost-savings from reduction in medical resources involved in anticoagulation monitoring. Due to the increasing availability of NOACs in the treatment of acute VTE, it is paramount to understand their comparability in terms of relative safety, efficacy, and cost-effectiveness.

In the absence of head-to-head randomized trials (RCTs) of dabigatran etexilate, hereafter referred to as dabigatran, and rivaroxaban, the most appropriate method to assess their relative efficacy and safety is by means of an indirect comparison (IC). Several ICs of NOACs have already been published in treatment of acute VTE5–7, and in extended anticoagulation8,9. Published ICs of dabigatran and rivaroxaban were conducted on efficacy endpoints: recurrent VTE (rVTE) and death5, rVTE and VTE-related death7, rVTE, recurrent pulmonary embolism (rPE), and recurrent DVT (rDVT)6. IC were conducted on safety end-points: major bleeding events (MBE)5–7, and major and clinical relevant bleeding events (MCRBE)5,7, as well as on death due to any cause5,7. MBE included intracranial hemorrhage and major extra-cranial hemorrhage, whereas MCBE also included clinically relevant non-major bleeds in addition to MBE.

The objectives of our analyses were to conduct additional ICs in patient’s sub-groups respective of the type of index event: DVT alone, PE with or without DVT, and respective of the duration of active anticoagulation, 6 months treatment with dabigatran vs 6 months treatment with rivaroxaban. Next, we used the IC estimates of rVTE and VTE-related death, and MCRBE to evaluate the cost-utility of dabigatran, as an alternative to rivaroxaban in the treatment and in extended anticoagulation of VTE (DVT and/or PE), from an NHS and PSS perspective in England and Wales.

Materials and methods

The dabigatran phase III clinical development program in treatment and prevention of recurrent VTE included two double blind, double-dummy, head-to-head, non-inferiority studies conducted against warfarin: RE-COVER10 and RE-COVER II11, and two double-blind, randomized control studies of dabigatran 150 mg twice daily compared with warfarin (active-control study) or with placebo (placebo-control study), in patients who completed at least 3 months of anticoagulation prior to randomization12. The two acute treatment studies were identical in their design: inclusion of patients, definition of end-points, and timing for outcomes assessment. Both compared dabigatran 150 mg twice daily with warfarin, adjusted to maintain an international normalized ratio (INR) of 2.0–3.0 during 6 months ‘double-dummy phase’; prior to initiation of the ‘double-dummy phase’ patients in both arms received parenteral anticoagulation followed by warfarin or matched placebo for at least 5 days and until the INR had been 2.0 or higher. Hence, pooling of results from these two studies was straight-forward. The primary efficacy end-point was recurrent, symptomatic, objectively confirmed VTE and VTE-related deaths. The important safety end-point was MCRBE. We decided to include MCRBE which occurred from the start of any treatment rather than MBE only, because, even though clinically relevant non-major bleeds do not lead to treatment discontinuation, they play an important role in terms of additional costs and quality-of-life decrements.

The rivaroxaban phase III VTE-treatment development program, EINSTEIN, consisted of two acute treatment studies in patients with acute symptomatic PE (EINSTEIN-PE)13, and in acute symptomatic DVT with or without PE (EINSTEIN-DVT)14, and one extension study in patients treated for an additional 6 or 12 months in whom 6–12 months of treatment with anticoagulants had been completed (EINSTEIN-Ext)14. In the acute treatment studies, rixaroxaban was administered twice daily at 15 mg for 3 weeks, followed by 20 mg once daily in the acute treatment studies, and was compared with enoxaparin (for at least 5 days) followed by a VKA (warfarin or acenocoumarol) for 3, 6, or 12 months, at a target INR of 2–3. EINSTEIN-Ext was a double-blind, randomized, event-driven superiority study that compared rivaroxaban alone (20 mg once daily) with placebo.

Across the studies, the characteristics of patients at baseline (average age, proportion of Caucasian-origin, proportion with cancer and baseline, proportion with prior of VTE; Table 1) were comparable. In all studies, warfarin or VKA was titrated to a target INR level of 2–3. Patients in RE-COVER studies received low molecular weight or unfractioned heparin for at least 5 days before randomization, while in EINSTEIN studies patients were excluded if they had received therapeutic doses of low molecular weight heparin, fondaparinux, or unfractionated heparin for more than 48 h before randomization. All of the trials comparing rivaroxaban or dabigatran with VKAs were designed to assess for non-inferiority; efficacy analyses of RE-COVER were conducted on a modified intention-to-treat patients set, representing all randomized patients who took at least one dose of the study drug; in EINSTIN treatment studies, analyses were conducted on the intention-to-treat dataset. A total of 5132 patients were randomized to receive warfarin or dabigatran in the RE-COVER studies, and 8281 patients received rivaroxaban or VKA in the EINSTEIN treatment studies. The primary efficacy end-point in all acute VTE studies was VTE and VTE-related death, however, in EINSTEIN studies, PE was considered the cause of death if there was an objective documentation or if death could not be attributed to a documented cause and PE could not be confidently ruled out. This deviated from the definition of VTE-related death in the RE-COVER studies, where unexplained deaths were excluded from the primary efficacy assessment. All studies had a common safety end-point, MCRBE. In both the RE-COVER and EINSTEIN, a bleeding event was defined as MCRBE if it was clinically overt and associated with a fall in the hemoglobin level of 20 g/l or more, or if it led to transfusion of two or more units of red cells, or if it occurred in a critical site, or contributed to death. The EINSTEIN study definition also includes retroperitoneal and intracranial bleeding explicitly, in addition to critical site bleeding.

Table 1. Acute treatment studies characteristics.

	RE-COVER		RE-COVER II		EINSTEIN-DVT		EINSTEIN-PE
	DBG	WRF	DBG	WRF	RIV	VKA	RIV	VKA
Number of patients	1273	1266	1280	1288	1731	1718	2419	2413
Average age (mean)	55.0	54.4	54.7	55.1	55.8	56.4	57.9	57.5
Caucasian origin	95.2%	94.4%	77.6%	77.6%	n/r	n/r	n/r	n/r
Cancer at baseline	5.0%	4.5%	3.9%	3.9%	n/r	n/r	n/r	n/r
Previous VTE	25.7%	25.4%	19.3%	15.8%	19.4%	19.2%	18.8%	20.3%

DBG, dabigatran; DVT, deep vein thrombosis; RIV, rivaroxaban; PE, pulmonary embolism; VKA, vitamin K antagonist; VTE, venous thromboembolism; WRF, warfarin.

A notable difference between the designs of the four acute treatment studies was the intended duration of treatment, which was 6 months in RE-COVER studies and 3, 6, or 12 months in EINSTEIN treatment studies. Potential bias due to differences in unobserved patient characteristics affecting the assignment to a treatment duration was mitigated by conducting an IC on the 6 months treatment EINSTEIN-PE and 6 months treatment EINSTEIN-DVT groups, vs 6 months of treatment in RE-COVER I, II.

We conducted a series of ICs using the adjusted indirect comparison methodology developed by Bucher et al.15. This method derives the indirect estimates by comparing the effects of each treatment vs the common comparator, and retains the benefits of randomization from the original trial data. This approach is in line with the conclusions from a comprehensive review of published indirect comparisons commissioned by the UK National Health Service (NHS) Health Technology Assessment Program, which highlighted the need to use an adjusted indirect comparison to minimize bias and the drawing of overly precise conclusions16. ICs were conducted to compare 6 months of treatment with dabigatran in the following groups: index VTE, with rivaroxaban pooled 3, 6, and 12 months treatment, index VTE, with rivaroxaban 6 months treatment group, index DVT, with rivaroxaban pooled 3, 6 and 12 months treatment, or rivaroxaban 6 months treatment group, index PE, with rivaroxaban pooled 3, 6, and 12 months treatment, or with rivaroxaban 6 months treatment group. Efficacy and safety end-points of extended anticoagulation were obtained from published ICs in secondary prevention5–7.

For the analyses of cost-effectiveness, we used a health-economic model in treatment of VTE published for the comparison between dabigatran and warfarin17. Briefly, this was a Markov cohort model with health states defined around the primary composite end-points: rVTE and VTE-related death, and MCRBE. Long-term consequences of VTE, CTEPH, and PTS were considered. Patients entered the model at the age of 55 years following index VTE (DVT/PE) and received 5-days low molecular weight heparin (LMWH) followed by 6 months treatment with dabigatran or 1-day LMWH followed by 6 months treatment with rivaroxaban. At any time, patients may discontinue treatment if experiencing a recurrent VTE, a MBE, or due to other reasons as observed in the clinical studies. After each rVTE, a new treatment course of 5 days LMWH followed by 6 months warfarin is reinstated, irrespective of the initial treatment (Figure 1). Data on the relative safety and efficacy of rivaroxaban vs dabigatran pertained to the conducted ICs presented herein. The baseline probabilities for rVTE were sourced from the pooled dabigatran arms of RE-COVER studies for the treatment period10,11, and from the dabigatran arm of RE-SONATE12 in extended anticoagulation. The 6 months probability of recurrent VTE and VTE-related death was 2.35% (60/2553) in index VTE, 2.39% (42/1755) in index DVT, and 2.26% (18/795) in PE; after treatment, in extended anticoagulation, the baseline probability applied over 6 months for rVTE and VTE-related death was 0.44% (3/681). The same sources were used to inform baseline probabilities of MCRBE during treatment and in extended anticoagulation. Six-months probabilities were 5.54% (134/2553) in index VTE, 5.02% (84/1757) in index DVT, 6.73% (50/793) in index PE, and 5.26% (36/684) in extended anticoagulation. Beyond active anticoagulation, the probability of recurrent VTE was sourced from the literature18. Dabigatran clinical trials did not assess the incidence of PTS and CPHT. The probabilities of CPHT2 and PTS3 were, therefore, obtained from the literature and assumed equal between dabigatran and rivaroxaban. Health-state utility values for recurrent events and bleeds were sourced from Euroqual 5 dimensions (EQ-5D) results of RE-COVER. Unit costs of drugs, medical contacts, diagnostic tests, and inpatient stays were sourced from fees published by the NHS in England and Wales, and the Personal Social Services Research Unit (PSSRU). Medical resource use data was obtained from different sources (literature or clinical expert advice) since the data collected in the clinical trials was limited and not representative of real world clinical practice (Table 2). Both deterministic and probabilistic analyses were conducted. For the probabilistic analysis, a distribution was assigned to each parameter of the model based on specific characteristics. A Gamma distribution was assigned to costs and to resource use, a Normal distribution was assigned to relative risks and hazard ratios a Beta distribution was assigned to probabilities of events and utility values. The analysis was performed with 1000 Monte Carlo simulations and the results were presented in a CE plane and through a cost-effectiveness acceptability curve, which together quantified the level of confidence that can be placed in the model results.

Figure 1. Model diagram. CRNMBE, clinically relevant non-major bleeds; CTEPH, chronic thromboembolic pulmonary hypertension; ICH, intracranial hemorrhage; iDVT, index DVT; iPE, index PE; iVTE, index VTE; MBE, major bleeds; PTS, post-thrombotic syndrome; rDVT, recurrent DVT; rPE, recurrent PE; rVTE, recurrent VTE.

Table 2. Unit costs and resource use inputs, £(2014).

	Unit cost	Resource	Applied in model
Pharmacy costs
RIV 20 mg/day	2.10	day 0–21	2.10
RIV 30 mg/day	4.20	day 22+	4.20
DBG 110/150 mg	1.10	220 mg/day	2.20
Administration costs
Visits - follow-up, DBG	27.99	1 visit	27.99
Visits - follow-up, RIV	27.99	1 visit	27.99
Recurrent events
Acute parenteral anticoagulation PE, outpatient
Pharmacy, average/day			7.78
Administration average/event	5 day	2.13
Acute parenteral anticoagulation DVT, outpatient
Pharmacy, average/day			9.63
Administration average/event	5 day	10.24
Re-treatment after rVTE with WRF
Pharmacy, WRF 4 mg	0.03	4 mg/day	0.07
Visits WRF - initial	62.56	1 visit	62.56
Visits WRF - titration	27.99	4 visits	111.97
Visits WRF - follow-up	27.99	1/month	27.99
Diagnosis recurrent PE event	/event	292
Diagnosis recurrent DVT event	/event	158
Inpatient cost/PE event	250	6.09 day	1369
Inpatient cost/DVT event	168	5.68 day	497
Inpatient cost/ICH event	10,182	/event	10,182
Disabled after ICH		/year	3011
Cost extra-cranial MBE	2357	/event	2357
Cost CRNMBE		/event	396
Cost severe PTS, yr1		/year	442
Cost severe PTS, yr2+	45	2 visits/year	90
Cost CTEPH
Pulmonary endarterectomy	7447	/event	3724
Follow-up cost	1240	/month	1240

CRNMBE, clinically relevant non-major bleeds; DGB, dabigatran; DVT, deep vein thrombosis; ICH, intracranial hemorrhage; IU, international units; kg, kilograms; LMWH, low-molecular weight heparin; MBE, major bleeds; mg, milligrams; PE, pulmonary embolism; PTS, post-thrombotic syndrome; RIV, rivaroxaban; UH, unfractioned heparin; yr, year; assuming a distribution of stroke severity of: 17% mild; 19% moderate; 64% severe.

Results

The IC results on the primary efficacy end-point, rVTE and VTE-related death, showed treatment with rivaroxaban to be associated with a lower risk of rVTE, when compared with dabigatran. However, differences were not statistically significant, with relative risk for rivaroxaban vs dabigatran equal to 0.83 (95% CI = 0.46–1.49) in the combined index DVT/PE for the rivaroxaban 3, 6, 12 months treatment set and equal to 0.90 (95% CI = 0.53–1.52) for the 6 months treatment set. In the index DVT sub-group, relative risks for rivaroxaban vs dabigatran were 0.57 (95% CI = 0.31–1.05) and 0.70 (95% CI = 0.35–1.39) for the 3, 6, 12 months, and the 6 months treatment set, respectively. In contrast, the rate of recurrent VTE in the index PE population was higher with rivaroxaban than with dabigatran, in both 3, 6, 12 months (RR = 1.30, 95% CI = 0.62–2.70) and 6 months treatment groups (RR = 1.30, 95% CI = 0.57–2.94) (Table 3); however, also here the difference was non-statistically significant. Observed MCRBE rates were higher with rivaroxaban than with dabigatran in the combined DVT/PE treatment group with relative risks of 1.49 (95% CI = 1.16–1.92) and 1.61 (95% CI = 1.18–2.17) for the 3, 6, 12 months, and the 6 months treatment set, respectively. In the index DVT sub-group, the same relative risks were 1.75 (95% CI = 1.25–2.50) and 2.04 (95% CI = 1.39–3.03), respectively. Differences in MCRBE rates were statistically significant in the index DVT group, but were non-statistically significantly different in the index PE group, with relative risks of 1.22 (95% CI = 0.83–1.79) and 1.22 (95% CI = 0.80–1.85) for the 3, 6, 12 months, and the 6 months treatment set, respectively (Table 3). In extended anticoagulation, rivaroxaban vs dabigatran relative risk of rVTE was 2.25 (95% CI = 0.52–9.69) and MCRBE relative risk was 1.93 (95% CI = 0.69–5.41), however differences were non-significant.

Table 3. Indirect comparison results—RE-COVER I, II, and EINSTEIN studies.

Treatment group	RR (DBG vs WRF)	RR (RIV vs VKA*)	RR (RIV vs DBG)	Lower CI (RIV vs DBG)	Upper CI (RIV vs DBG)
Recurrent VTE and VTE-related death
Index VTE (DVT + PE)
RIV 3/6/12 months DVT + PE	1.09	0.90	0.83	0.46	1.49
RIV 6 months DVT + PE	1.09	0.98	0.90	0.53	1.52
Extended anticoagulation#	0.08±	0.18±	2.25	0.52	9.69
Index DVT
RIV 3/6/12 months DVT	1.23	0.70	0.57	0.31	1.05
RIV 6 months DVT	1.23	0.86	0.70	0.35	1.39
Index PE
RIV 3/6/12 months PE	0.87	1.13	1.30	0.62	2.70
RIV 6 months PE	0.87	1.13	1.30	0.57	2.94
MCRBE, from start of any treatment
Index VTE (DVT + PE)
RIV 3/6/12 months DVT + PE	0.63	0.94	1.49	1.16	1.92
RIV 6 months DVT + PE	0.63	1.01	1.61	1.18	2.17
Extended anticoagulation#	2.69±	5.19±	1.93	0.69	5.41
Index DVT
RIV 3/6/12 months DVT	0.57	1.00	1.75	1.25	2.50
RIV 6 months DVT	0.57	1.16	2.04	1.39	3.03
Index PE
RIV 3/6/12 months PE	0.75	0.91	1.22	0.83	1.79
RIV 6 months PE	0.75	0.91	1.22	0.80	1.85

CI, confidence interval; DBG, dabigatran; RIV, rivaroxaban; RR, relative risk; VTE, venous thromboembolism; WRF, warfarin.

±vs placebo; *warfarin or acenocoumarol; #based on published indirect comparisons5–7.

In all, treatment and secondary prevention settings, cost-effectiveness analyses results showed marginally higher clinical benefits with dabigatran than with rivaroxaban. These resulted from avoidance of bleeding events, except for other major bleeds which were higher in the dabigatran arm for the VTE and for index DVT, and less recurrent non-fatal PEs in all settings, as well as less recurrent all VTE in the PE sub-group. Results were consistent across the two rivaroxaban treatment sets: 3, 6, 12 months treatment (Table 4) and 6 months treatment (Table 5). Marginal cost savings were projected for dabigatran in all settings, dabigatran resulting dominant over treatment with rivaroxaban (higher net health benefits and lower costs). However, in all probabilistic sensitivity analyses conducted (PSA), the cloud of the scatter plots was located on both sides of the 0x and 0y axes, covering all possible cost-effectiveness scenarios: dominant, dominated, lower costs lower outcomes, higher costs higher outcomes (Figure 2). At a willingness-to-pay threshold of £20,000, the probability that dabigatran would be considered good value for money compared with 3, 6, 12 months rivaroxaban is 61% and 88%, in treatment and extended anticoagulation of VTE, respectively, and 62% and 62% in DVT and PE, respectively.

Figure 2. Probabilistic sensitivity analyses results. (1a, b) CE plane and CEAC VTE (DVT/PE) treatment; (2a, b) CE plane and CEAC VTE extended anticoagulation; (3a, b) CE plane and CEAC index DVT treatment; (4a, b) CE plane and CEAC index PE treatment; CE, cost-effectiveness; CEAC, cost-effectiveness acceptability curve; DVT, deep vein thrombosis; IC, incremental costs; Incr., incremental; PE, pulmonary embolism; QALY, quality-adjusted life years; VTE, venous thromboembolism.

Table 4. Deterministic analyses results: 6 months dabigatran vs 3-6-12 months rivaroxaban.

	VTE treatment			VTE treatment and extended anticoagulation
	DBG	RIV	Incr.	DBG	RIV	Incr.
Events per 10,000 patients
All recurrent VTE	12,683	12,641	42	12,542	12,542	1
Symptomatic recurrent non-fatal VTE events	11,248	11,208	40	11,128	11,130	−2
Non-fatal DVT	8254	8182	72	8162	8122	40
Non-fatal PE	2995	3026	−32	2965	3007	−42
VTE-related death	1435	1433	2	1414	1412	3
All major and clinically relevant bleeds	1037	1296	−260	1471	2189	−718
Non-fatal major and clinically relevant bleeds	1028	1286	−258	1462	2178	−716
Intracranial hemorrhage	19	22	−3	19	22	−3
Other major bleeds	235	168	68	259	267	−8
Clinically relevant non-major bleeds	774	1096	−322	1184	1890	−705
Deaths from bleeding	9	10	−2	9	10	−2
PTS	1252	1246	6	1244	1241	3
CTEPH	236	237	−1	234	236	−1
Discounted^a QALY (mean per patient) (CI)	12.406 (12.09–12.65)	12.404 (12.10–12.63)	0.002 (−0.02–0.03)	12.430 (12.17–12.67)	12.421 (12.15–12.64)	0.009 (−0.01–0.07)
Total costs, mean per patient (£) (CI)	7474 (4619–11,122)	7510 (4672–11,209)	−27 (−145–77)	7757 (4862–11,778)	7830 (4912–11,866)	−74 (−301–40)
Drug costs (£)	461	461	0	776	772	4
Monitoring costs (£)	370	364	6	363	357	6
Event costs (£)	6643	6686	−43	6617	6700	−83
ICER, costs/QALYs			dominant			dominant
	Index DVT			Index PE
Events per 10,000 patients
All recurrent VTE	12,676	12,573	103	12,696	12,755	−59
Symptomatic recurrent non-fatal VTE events	11,245	11,144	101	11,250	11,310	−60
Non-fatal DVT	8292	8167	125	8132	8209	−77
Non-fatal PE	2953	2976	−24	3118	3101	17
VTE-related death	1431	1429	1	1447	1445	2
All major and clinically relevant bleeds	984	1343	−359	1154	1297	−144
Non-fatal major and clinically relevant bleeds	975	1332	−357	1147	1287	−141
Intracranial hemorrhage	21	24	−3	15	18	−3
Other major bleeds	258	181	76	187	151	36
Clinically relevant non-major bleeds	696	1127	−431	945	1119	−174
Deaths from bleeding	9	11	−2	7	10	−3
PTS	1510	1499	10	686	693	−7
CTEPH	114	115	−1	501	500	1
Discounted^a QALY (mean per patient) (CI)	12.403 (12.14–12.66)	12.400 (12.14–12.65)	0.003 (−0.02–0.03)	12.409 (12.14–12.66)	12.404 (12.14–12.64)	0.005 (−0.04–0.06)
Total costs, mean per patient (£) (CI)	3798 (2855–5139)	3818 (2873–5186)	−20 (−166–89)	15,509 (8,353–25,834)	15,514 (8,369–26,021)	−5 (−253–179)
Drug costs (£)	465	463	2	453	460	−7
Monitoring costs (£)	374	365	9	361	361	0
Event costs (£)	2959	2990	−31	14695	14,693	2
ICER, costs/QALYs			dominant			dominant

Incr., Incremental; CRNMBE, clinically relevant non-major bleeds; CTEPH, chronic thromboembolic pulmonary hypertension; DBG, dabigatran; DVT, deep vein thrombosis; ICER, incremental cost-effectiveness ratio; ICH, intracranial hemorrhage; ICUR, incremental cost-utility ratio; MBE, major bleeds; PE, pulmonary embolism; QALYs, quality-adjusted life years; VTE, venous thromboembolism; WRF, warfarin; ^adiscounted at 3.5% annual rate; CI, 95% confidence interval (obtained from probabilistic sensitivity analysis).

Table 5. Deterministic analyses results: 6 months dabigatran vs 6 months rivaroxaban.

	VTE treatment			VTE treatment and extended anticoagulation
	DBG	RIV	Incr.	DBG	RIV	Incr.
Events per 10,000 patients
All recurrent VTE	12,683	12,656	27	12,542	12,556	−14
Symptomatic recurrent non-fatal VTE events	11,248	11,223	25	11,128	11,145	−17
Non-fatal DVT	8254	8194	60	8162	8134	28
Non-fatal PE	2995	3029	−35	2965	3010	−45
VTE-related death	1435	1433	2	1414	1412	3
All major and clinically relevant bleeds	1037	1360	−323	1471	2255	−785
Non-fatal major and clinically relevant bleeds	1028	1349	−320	1462	2244	−782
Intracranial hemorrhage	19	21	−3	19	21	−3
Other major bleeds	235	176	60	259	275	−16
Clinically relevant non-major bleeds	774	1151	−377	1184	1948	−763
Deaths from bleeding	9	11	−3	9	11	−3
PTS	1252	1247	5	1244	1242	2
CTEPH	236	237	−1	234	236	−2
Discounted^a QALY (mean per patient) (CI)	12.406 (12.12–12.65)	12.402 (12.12–12.64)	0.005 (−0.02–0.03)	12.430 (12.15–12.68)	12.419 (12.12–12.66)	0.012 (−0.01–0.08)
Total costs, mean per patient (£) (CI)	7474 (4888–12,104)	7519 (4740–11,610)	−45 (−187–53)	7757 (4940–12,026)	7847 (5007–12,178)	−91 (−330–29)
Drug costs (£)	461	461	0	776	774	2
Monitoring costs (£)	370	364	6	363	358	6
Event costs (£)	6643	6694	−51	6617	6716	−99
ICER, costs/QALYs			dominant			dominant
	Index DVT			Index PE
Events per 10,000 patients
All recurrent VTE	12,676	12,601	75	12,696	12,755	−59
Symptomatic recurrent non-fatal VTE events	11,245	11,171	74	11,250	11,310	−60
Non-fatal DVT	8292	8178	115	8132	8209	−77
Non-fatal PE	2953	2993	−41	3118	3101	17
VTE-related death	1431	1430	1	1447	1445	2
All major and clinically relevant bleeds	984	1478	−494	1154	1297	−144
Non-fatal major and clinically relevant bleeds	975	1466	−491	1147	1287	−141
Intracranial hemorrhage	21	25	−5	15	18	−3
Other major bleeds	258	192	65	187	151	36
Clinically relevant non-major bleeds	696	1249	−552	945	1119	−174
Deaths from bleeding	9	12	−3	7	10	−3
PTS	1510	1500	10	686	693	−7
CTEPH	114	116	−2	501	500	1
Discounted^a QALY (mean per patient) (CI)	12.403 (12.12–12.64)	12.394 (12.12–12.64)	0.009 (−0.02–0.05)	12.409 (12.13–12.67)	12.404 (12.11–12.67)	0.005 (−0.05–0.05)
Total costs, mean per patient (£) (CI)	3798 (2857–5236)	3854 (2898–5350)	−57 (−235–67)	15,509 (8145–25,571)	15,514 (8166–25,527)	−5 (−290–188)
Drug costs (£)	465	461	3	453	460	−7
Monitoring costs (£)	374	365	9	361	361	0
Event costs (£)	2959	3028	−69	14,695	14,693	2
ICER, costs/QALYs			dominant			dominant

Incr., Incremental; CRNMBE, clinically relevant non-major bleeds; CTEPH, chronic thromboembolic pulmonary hypertension; DBG, dabigatran; DVT, deep vein thrombosis; ICER, incremental cost-effectiveness ratio; ICH, intracranial hemorrhage; ICUR, incremental cost-utility ratio; MBE, major bleeds; PE, pulmonary embolism; QALYs, quality-adjusted life years; VTE, venous thromboembolism; WRF, warfarin; ^adiscounted at 3.5% annual rate; CI, 95% confidence interval (obtained from probabilistic sensitivity analysis).

Discussion

Based on the ICs, conducted dabigatran and rivaroxaban appear to have a comparable efficacy profile in both patients with index VTE, and individually in index PE and index DVT; the risk of recurrent events was higher with dabigatran in index DVT and in combined DVT/PE, but lower in the index PE group, yet differences were not statistically significant in either of the analyzed groups. This was true for both the comparisons with rivaroxaban 3, 6, 12 months, and rivaroxaban 6 months treatment. Dabigatran was associated with a significantly lower risk of MCRBE in index DVT and in combined DVT/PE, both against 6 months rivaroxaban or the 3, 6, 12 months treatment set. Dabigatran had non-significantly lower rates of MCRBE in the index PE group.

Several other ICs of dabigatran and rivaroxaban have been published in treatment and extended anticoagulation5–9. Yet, the analyzed safety and efficacy end-points deviated from the ones needed to populate our cost-effectiveness model. Specifically, Mantha and Ansell5 compared rates of rVTE and death due to any cause in acute treatment studies, differently from VTE and VTE related deaths used in our model. Kang and Sobieraj6 looked at rVTE, rPE, and rDVT and MBE, whereas we conducted our analysis looking at MCRBE. Hence, our results are not directly comparable with those already reported. Despite inherent limitations of IC methods, these remain the most appropriate means of substantiating relative safety and efficacy in lack of head-to-head trials, and are useful to reimbursement and formulary decision-makers to assist in informing analyses of value for money.

Indeed, our cost-effectiveness evaluations showed that dabigatran can be considered the dominant treatment strategy compared to rivaroxaban in the patients’ sub-groups considered, given the projected marginally higher clinical benefits and lower treatment costs.

The present analyses were intended to offer an indication of the relative clinical and cost benefits of dabigatran compared with rivaroxaban, also available in England and Wales, and did not attempt to compare dabigatran with other NOACs currently in development, such as apixaban and edoxaban. This is a recognized limitation of the present study and analyses will need to be updated as these new molecules will become available to patients and prescribers.

The ICs and cost-effectiveness analyses conducted showed that dabigatran and rivaroxaban are comparable in terms of their net clinical benefits and net costs. In deterministic analyses, dabigatran was projected dominant in all analyzed settings. Probabilistic analyses results showed a high likelihood of dabigatran being considered good value for money in the UK, in treatment, and in extended anticoagulation of VTE.

Transparency

Declaration of funding

This study was funded by Boehringer Ingelheim International GmbH.

Declaration of financial/other relationships

Boehringer Ingelheim International GmbH employed IMS Health Consulting to conduct the present study and write the manuscript, of which MDF and ML are employees. TS and AU are employees of and VH is a consultant for Boehringer Ingelheim. JME peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Acknowledgments

The authors of this paper acknowledge the work of Sorrel E Wolowacz (RTI) and James Brockbank (RTI) who supported the development of the cost-effectiveness model.