Thrombosis Research 133 (2014) 1145–1151
Contents lists available at ScienceDirect
Thrombosis Research journal homepage: www.elsevier.com/locate/thromres
Regular Article
Indirect treatment comparison of new oral anticoagulants for the treatment of acute venous thromboembolism Nayon Kang, Diana M. Sobieraj ⁎ University of Connecticut School of Pharmacy, 69 N Eagleville Rd. Unit 3092, Storrs, CT 06269, United States
a r t i c l e
i n f o
Article history: Received 10 February 2014 Received in revised form 11 March 2014 Accepted 18 March 2014 Available online 24 March 2014 Keywords: Venous thromboembolism Anticoagulation Treatment
a b s t r a c t Background: Numerous new oral anticoagulants (NOACs) have been compared to a parenteral anticoagulant/oral vitamin K antagonist (VKA) for the treatment of acute venous thromboembolism (VTE). We aimed to conduct a systematic review and adjusted indirect comparison meta-analysis to compare the efficacy and safety of NOACs for this indication. Methods: We conducted a systematic literature search through November 2013 for randomized trials that evaluated treatment of acute VTE with a NOAC including rivaroxaban, apixaban, dabigatran and edoxaban. Trials had to report at least one of the following outcomes of interest: mortality, recurrent VTE, recurrent pulmonary embolism (PE), recurrent deep vein thrombosis (DVT), or major bleeding. Included trials were evaluated for quality using the Cochrane Risk of Bias tool. We performed an adjusted indirect comparison meta-analysis to evaluate the comparative efficacy and safety of NOACs, reporting relative risks (RRs) and 95% confidence intervals for each outcome. Results: Six trials (n = 27,069) met inclusion criteria, one each evaluating apixaban and edoxaban and two trials each evaluating rivaroxaban and dabigatran. Risk of bias was low for all trials. NOACS did not differ significantly in the risk of mortality, recurrent VTE, recurrent PE or recurrent DVT. Dabigatran increased major bleeding risk compared to apixaban [RR 2.69 (1.19 to 6.07)] as did edoxaban compared to apixaban [RR 2.74 (1.40 to 5.39)]. Conclusion: Although NOACs do not appear to differ in the efficacy of treating acute VTE, data suggests apixaban to be the safer than some of its competitors. © 2014 Elsevier Ltd. All rights reserved.
Introduction Acute venous thromboembolism (VTE) is a common disorder with an estimated annual incidence of 1.83 adults per 1000, with deep vein thrombosis (DVT) occurring more frequently than pulmonary embolism (PE) [1]. For decades, the standard treatment strategy has been bridging a parenteral heparin product and an oral vitamin-K antagonist (VKA) until a therapeutic international normalized ratio (INR) is achieved, with continuation of the VKA for a minimum of three months [2]. While effective, this treatment regimen has several limitations. Parenteral administration can be unfavorable to patients and requires additional nurse time in the inpatient setting and possibly on an outpatient basis. Routine INR monitoring requires transportation and often leads to dose-adjustments which presents an opportunity for medication errors. Lastly, there is a potential for significant drug and food interactions with
VKAs which can become important given their narrow therapeutic window. In recent years, several new oral anticoagulants (NOACs) have been under development for the treatment of acute VTE. NOACs that are currently approved by the Food and Drug Administration (FDA) or with phase III trial results include the factor Xa inhibitors apixaban, rivaroxaban, and edoxaban and the direct thrombin inhibitor dabigatran. For the acute treatment of VTE, NOAC have been compared to the gold standard regimen of a parenteral heparin product plus an oral VKA, although they have yet to be evaluated in head-to-head trials. In the absence of direct comparative evidence, indirect comparisons may provide information to aid in clinical decision making and therefore we aimed to conduct a systematic review and an adjusted indirect comparison meta-analysis to evaluate the efficacy and safety of NOACs for the treatment of acute VTE. Methods
Abbreviations: CI, confidence interval; DVT, deep vein thrombosis; FDA, Food and Drug Administration; LMWH, low-molecular weight heparin; NOAC, New oral anticoagulant; PE, pulmonary embolism; RR, relative risk; UFH, unfractionated heparin; VKA, vitamin K antagonist; VTE, venous thromboembolism. ⁎ Corresponding author. Tel.: +1 860 545 2429; fax: +1 860 5452277. E-mail address: [email protected] (D.M. Sobieraj).
http://dx.doi.org/10.1016/j.thromres.2014.03.035 0049-3848/© 2014 Elsevier Ltd. All rights reserved.
We conducted a systematic literature search in MEDLINE and Cochrane Central databases through November 2013 using the search strategy in e-Appendix A. A manual search was also performed using the references of clinical trials and review articles to identify additional
1146
N. Kang, D.M. Sobieraj / Thrombosis Research 133 (2014) 1145–1151
relevant articles. In order for a study to be included in the analysis, it had to be a randomized controlled trial that evaluated patients with acute VTE treated with a NOAC and reported at least one outcome of interest. NOACs included were those with current FDA approval or with published phase III trial results. Only studies evaluating the FDA approved dosing regimen for rivaroxaban were included and only studies using the same dosing regimen evaluated in phase III trials were included for the remaining NOACs. Outcomes of interest included mortality, recurrent VTE, recurrent DVT, recurrent PE and major bleeding. Two independent investigators separately reviewed all citations identified by the search for inclusion and abstracted data from included trials. Disagreements were resolved through discussion. The following data was collected from each trial: author identification, year of publication, funding source, report of conflicts of interest, study design
characteristics, study population (inclusion and exclusion criteria, geographic location, length of study, duration of patient follow-up), patient baseline characteristics, VTE treatment regimen (name, strength, frequency, dose, route of administration, duration of therapy, time in therapeutic range for VKA arms), and outcomes data (number of events, definitions, period of follow-up, and diagnostic tests for confirmation). To assess the methodological quality of the included trials, the Cochrane Collaboration’s risk of bias tool was used [3]. This tool evaluates seven domains including sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective outcome reporting and any other identifiable issues. Each domain is assessed as having low, high, or unclear risk of bias and then a summary assessment of each trial across domains is made as low, high or unclear risk of bias.
Abbreviations: RCT= randomized controlled trial; VTE=venous thromboembolism Fig. 1. Inclusion of studies. Abbreviations: RCT = randomized controlled trial; VTE = venous thromboembolism.
Table 1 Baseline characteristics. Trial
Randomized intervention (n)
Overall treatment duration d (SD)
RECOVER-II N = 2589
Dabigatran 150 mg BID × 6 m (n = 1279)
NR
EINSTEIN-DVT N = 3449
RECOVER N = 2564
Prior VTE N (%)
Recent Trauma/ Surgery N (%)
781 (61)
54.7 (16.2)
NR
247 (19.3)
NR
776 (60.2)
55.1 (16.3)
1569 (58.3)
57.2 (16.0)
1598 (59.1)
56.7 (16.0)
250 (111.8)
2360 (57.3)
55.7 (16.3)
Enoxaparin or heparin × 5d + warfarin daily for 3-12 m (n = 4122) Rivaroxaban 15 mg BID × 3w then 20 mg daily for 3, 6, or 12 m (n = 2419) Enoxaparin 1 mg/kg BID × 5d + VKA daily for 3, 6, or 12 m (n = 2413) Rivaroxaban 15 mg BID × 3w then 20 mg daily for 3, 6, or 12 m (n = 1731) Enoxaparin 1 mg/kg BID × 5d + VKA daily for 3, 6, or 12 m (n = 1718) Dabigatran 150 mg BID × 6 m (n = 1273)
248.4 (112.6)
2356 (57.2)
55.9 (16.2)
216 (NR)
1309 (54.1)
57.9 (7.3)
214 (NR)
1247 (51.7)
57.5 (7.2)
993 (57.4)
55.8 (16.4)
967 (56.3)
56.4 (16.3)
DVT: 1718 (100)
738 (58)
55.0 (15.8)
746 (58.9)
54.4 (16.2)
DVT: 880 (69.1) PE: 270 (21.2) Both: 121 (9.5) DVT: 869 (68.6) PE: 271 (21.4) Both: 124 (9.8)
NR
Enoxaparin 1 mg/kg BID × 5d + VKA daily for 6 m (n = 1266)
NR
NR
203 (15.8)
Active Cancer N (%) 50 (3.9)
Known Thrombophilia N (%)
Risk of Bias
NR
Low
50 (3.9)
TTR: 57% Above: 19% Below: 24%
2416 (89.8)
463 (17.2)
2429 (89.8)
409 (15.1)
2713 (65.9)
784 (19.0)⁎
2697 (65.4)
736 (17.9)⁎
1566 (64.7)
455 (18.8)
415 (17.2)⁎
114 (4.7)⁎
138 (5.7)
1551 (64.3)
489 (20.3)
398 (16.5)⁎
109 (4.5)⁎
121 (5.0)
1055 (60.9)
336 (19.4)
338 (19.5)⁎
118 (6.8)⁎
107 (6.2)
1083 (63.0)
330 (19.2)
335 (19.5)⁎
89 (5.2)⁎
116 (6.8)
NR
327 (25.7)
NR
64 (5.0)
NR
322 (25.4)
NR
NR
66 (2.5)
74 (2.8)
77 (2.8)
59 (2.2)
378 (9.2)⁎
NR
Low
TTR: 61 Above: 16 Below: 23 Low
393 (9.5)⁎
57 (4.5)
TTR %
TTR: 63.5 Above: 17.6 Below: 18.9 Low
TTR: 62.7 Above: 15.5 Below: 21.8 Low
N. Kang, D.M. Sobieraj / Thrombosis Research 133 (2014) 1145–1151
EINSTEIN-PE N = 4832
Un-provoked Index Event N (%)
Edoxaban 30 mg or 60 mg daily for 3-12 m (n = 4118)
Apixaban 10 mg BID × 7d then 5 mg BID × 6 m (n = 2691) Enoxaparin 1 mg/kg BID × 5d + VKA daily for 6 m (n = 2704)
Hokusai-VTE N = 8240
Index event N (%)
DVT: 877 (68.5) PE: 298 (23.3) Both: 104 (8.1) DVT: 873 (67.8) PE: 297 (23.1) Both: 117 (9.1) DVT: 1749 (65.0) PE: 678 (25.2) Both: 252 (9.4) DVT: 1783 (65.9) PE: 681 (25.2) Both: 225 (8.3) DVT:2468 (59.9) PE: 1240 (30.1) Both: 410 (10.0) DVT: 2453 (59.5) PE: 1265 (30.7) Both: 404 (9.8) DVT: 0 PE: 1813 (74.9) Both: 606 (25.1) DVT: 0 PE: 1823 (75.5) Both: 590 (24.5) DVT: 1731 (100)
Enoxaparin 1 mg/kg BID × 5d + VKA daily for 6 m (n = 1289) AMPLIFY N = 5395
Age y (SD)
Males N (%)
TTR: 57.7 Above: 16.2 Below: 24.4 Low
TTR: 60 Above: 19 Below: 21
Abbreviations: d = days; BID = twice a day; m = months; mg = milligrams; NR = not reported; SD = standard deviation; TTR = time in therapeutic range; VKA = vitamin k antagonist; VTE = venous thromboembolism; w = weeks. ⁎ Represents patients whose index event was attributed to given characteristic.
1147
1148
N. Kang, D.M. Sobieraj / Thrombosis Research 133 (2014) 1145–1151
Fig. 2. Network diagram. This diagram represents the network created by the included trials. Therapies that are directly compared in at least one trial are connected with a solid line. The actual number of trials for each comparison is depicted.
The risk of bias was evaluated for each trial by two separate investigators and conflicts were resolved through discussion. In the absence of direct comparative trials, methods for indirect comparison can be employed. For example, if one trial compares treatment A to treatment B and a second trial compares treatment C to treatment B, methods of indirect comparison can be used to estimate the relative effects of treatment A to treatment C through the common comparator treatment B. We carried out such analysis using the adjusted indirect comparison meta-analysis methods proposed by Bucher et al. [4] First, we calculated pairwise relative risks (RR) and corresponding 95% confidence intervals (CIs) for each comparison (each individual NOAC versus the common comparator of a parenteral anticoagulant plus oral VKA) and outcome using StatsDirect version 2.7.8. If only one trial representing a given comparison was available, we calculated a RR and 95% CI for each outcome using the trial data. If more than one trial representing the same comparison was available, traditional metaanalysis using a random effects model was used to generate a pooled RR and 95% CI. Using the pairwise data, we compared each NOAC to another using the publically available software for adjusted indirect metaanalysis [5]. RR and 95% CIs were generated for each indirect comparison and outcome. One sensitivity analyses was run to observe the effects of pooling the EINSTEIN trials using meta-analytic techniques instead of using the collective data published in Prins et al. Results Study selection and characteristics The results of our literature search are presented in Fig. 1. A total of 396 citations were identified through our literature search with 3 manually added citations. Of those, two were excluded as duplicate publications, and 326 and 65 were excluded at the abstract and full text level, respectively, leaving 6 unique trials (Table 1, Fig. 2) [6–11].
EINSTEIN-PE and EINSTEIN-DVT evaluated rivaroxaban (the pooled outcomes data published by Prins et al. [12] were used in place of the individual trials), RECOVER-I and RECOVER-II evaluated dabigatran, AMPLIFY evaluated apixaban, and Hokusai-VTE evaluated edoxaban. All trials used the common comparator of a parenteral anticoagulant plus an oral VKA. In RECOVER-I and RECOVER-II, unfractionated heparin (UFH), low-molecular weight heparin (LMWH) and fondaparinux were parenteral anticoagulant options although 89 to 90% of patients were treated with LMWH. In AMPLIFY either UFH or LMWH could have been used. In the remaining trials enoxaparin was the specified parenteral anticoagulant. All trials were published as full text articles were judged to have low risk of bias. The six included trials represent a total of 27,069 patients with acute VTE. With exception of EINSTEIN-DVT which enrolled patients with an index DVT and EINSTEIN-PE which enrolled patients with an index PE or both a PE and DVT, the index event was a DVT in 60 to 70% of patients, PE in 21 to 31% of patients and both in 8 to 10% of patients. Although in all trials patients could have been treated with therapeutic anticoagulant doses prior to randomization for a limited period of time (typically b48 hours), in the RECOVER-I and II and Hokusai-VTE trials initial treatment with a parenteral anticoagulant was required in all patients. The median treatment duration in the RECOVER-I trial was 9 days, in RECOVER-II was 9.4 to 9.6 days, and in Hokusai-VTE was 7 days. Randomized treatment duration was fixed in AMPLIFY and the RECOVER trials at 6 months while clinician discretion determined duration in the remaining trials, which could range from 3 to 12 months. The mean number of treatment days in Hokusai-VTE ranged from 248 to 250 days and 204 to 207 days for the EINSTEIN trials collectively. The time in therapeutic range was similar across trials at around 60%, while time spent outside of the range was slightly more below range (19 to 24%) than above range (16 to 19%). Patient characteristics were balanced across groups with the exception of cases of unprovoked VTE which were higher in the AMPLIFY trial.
Table 2 Pairwise relative risks for each new oral anticoagulant versus parenteral anticoagulant plus vitamin K antagonist.⁎ Base-case analyses
Rivaroxaban Dabigatran Apixaban Edoxaban
Morality
Recurrent VTE
Recurrent PE
Recurrent DVT
Major bleeding
0.97 (0.73 to 1.27) 1.00 (0.67 to 1.50) 0.79 (0.53 to 1.19) 1.05 (0.83 to 1.33)
0.90 (0.68 to 1.20) 1.12 (0.79 to 1.60) 0.84 (0.60 to 1.18) 0.89 (0.71 to 1.12)
1.08 (0.74 to 1.57) 1.04 (0.59 to 1.85) 1.06 (0.69 to 1.65) 0.88 (0.65 to 1.20)
0.70 (0.45 to 1.09) 1.17 (0.70 to 1.93) 0.61 (0.36 to 1.06) 0.91 (0.64 to 1.29)
0.55 (0.38 to 0.81) 0.76 (0.49 to 1.18) 0.31 (0.17 to 0.54) 0.85 (0.60 to 1.21)
0.95 (0.64 to 1.42)
0.90 (0.56 to 1.43)
1.07 (0.73 to 1.58)
0.71 (0.40 to 1.24)
0.56 (0.38 to 0.82)
Sensitivity analysis Rivaroxaban
⁎ Values represent relative risks and 95% confidence intervals.
N. Kang, D.M. Sobieraj / Thrombosis Research 133 (2014) 1145–1151
Indirect Comparison Results The relative risks generated for each NOAC versus parenteral anticoagulant plus VKA are presented in Table 2. Four NOACs were compared indirectly: dabigatran, apixaban, rivaroxaban and edoxaban. For the
1149
efficacy outcomes including mortality, recurrent symptomatic VTE, recurrent DVT and current PE there were no significant differences amongst all of the comparisons made (Fig. 3). The risk of major bleeding increased with dabigatran compared to apixaban [RR 2.69 (1.19 to 6.07)] and with edoxaban compared to apixaban [RR 2.74 (1.40 to
Abbreviations: DVT=deep vein thrombosis; PE=pulmonary embolism; VTE=venous thromboembolism Fig. 3. Results of the indirect treatment comparison. Abbreviations: DVT = deep vein thrombosis; PE = pulmonary embolism; VTE = venous thromboembolism.
1150
N. Kang, D.M. Sobieraj / Thrombosis Research 133 (2014) 1145–1151
Abbreviations: DVT=deep vein thrombosis; PE=pulmonary embolism; VTE=venous thromboembolism Fig. 4. Results of sensitivity analysis. Abbreviations: DVT = deep vein thrombosis; PE = pulmonary embolism; VTE = venous thromboembolism.
5.39)]. Results of the sensitivity analysis are presented in Fig. 4. Overall our original results remained robust and did not change as a result of pooling the EINSTEIN trials. Discussion This analysis is based on over 25,000 patients treated with a NOAC for acute VTE for up to 12 months. Although there does not appear to be any significant difference in terms of efficacy suggested by the results, the NOACs do not appear to be similar in terms of bleeding risk. The risk of major bleeding was increased with dabigatran and with edoxaban in comparison to apixaban. Although the results for rivaroxaban versus other NOAC did not reach statistical significance, when compared to apixaban there was a trend in the direction of increased risk of major bleeding with rivaroxaban. These data suggest that apixaban may have the safest bleeding risk profile among the NOACs studied. There are several potential factors which could lead to this observation. Required use of parenteral anticoagulant treatment in the RECOVER and Hokusai-VTE trials could impact early bleeding rates favoring the other NOACs apixaban or rivaroxaban. Although both apixaban and rivaroxaban require a higher initial dose followed by a lower maintenance dose, the initial high-dose period is shorter for apixaban compared to rivaroxaban (7 versus 21 days) which could favor lower bleeding risk with apixaban. The mean number of treatment days in Hokusai-VTE were 248 to 250 days and in the EINSTEIN trials collectively were 204 to 207 days. Naturally with the option of longer treatment duration in these trials there was a longer period for events to accumulate in comparison to AMPLIFY and RECOVER, which could favor apixaban and dabigatran. One other published indirect treatment comparison by Fox et al. analyzes NOACs for acute VTE treatment [13]. The authors concluded that
there were no significant differences in mortality, recurrent VTE, or major bleeding when rivaroxaban and dabigatran were compared, which is similar to the results of our analysis. However, they excluded apixaban because there was “only 1 small trial” and their search preceded the publication of AMPLIFY which we have included. Their analysis also differed from ours in that dose-finding trials were included in their analysis and how different doses were handled or included was not specified. This limits the applicability of their findings since the doses available to prescribers are not purely represented and this methodology could favor less bleeding or lower efficacy if lower doses were pooled with standard, higher doses. Our analysis reflects the most current literature and also includes additional NOAC which Fox and colleagues were unable to evaluate. Our literature search was also more recent therefore we were able to include the results of RECORD-II published in full text in our analysis. There are limitations to our analysis that should be noted. Adjusted indirect treatment comparison meta-analysis is an established statistical technique in the scientific literature [14,15]. While seen as an advantage in situations where no direct comparative data exists, this can also be considered a limitation as data is purely generated using indirect evidence. Therefore, results are not as precise as what is possible with direct evidence. Since our network was without direct evidence, we are unable to evaluate consistency which is one factor considered important to the internal validity of indirect comparisons [14]. However, we believe our analysis is robust since in our judgment, trials met the similarity assumption and the individual trials were of high quality, which are two additional factors important to the internal validity of indirect comparisons [14]. We chose to use the data published in Prins et al. to represent the EINSTEIN trials collectively in an effort to reduce heterogeneity of populations across studies since all other trials included patients with an index DVT or PE, not just one VTE type. Through
N. Kang, D.M. Sobieraj / Thrombosis Research 133 (2014) 1145–1151
sensitivity analysis we evaluated if pooling the EINSTEIN trials using meta-analytic techniques would vary results rather than using the collective data, and again the results did not change overall. In summary, NOAC are similar in terms of efficacy for the acute treatment of VTE, although apixaban appears to have an advantageous safety profile as indirect estimates suggest the other evaluated NOAC have a higher risk of bleeding. Conflicts of interest The authors have no conflicts to disclose. Funding This work was unfunded. Acknowledgements None Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.thromres.2014.03.035. References [1] Oger E. Incidence of venous thromboembolism: a community-based study in Western France. EPI-GETBP Study Group. Thromb Haemost 2000;83:657–60. [2] Kearon C, Akle EA, Comerota AJ, Prandoni P, Bounameaux H, Goldhaber SZ, et al. Antithrombotic therapy for VTE disease: antithrombotic therapy and prevention of thrombosis. American College of Chest Physicians Evidence-based Clinical Practice Guidelines, 141, Chest, 9th ed. 2012. p. e419S–94S.
1151
[3] Higgins JPT, Altman DG, Gotzsche PC, Juni P, Moher D, Oxman AD, et al. The Cochrane Collaboration’s tool for assessing risk of bias in randomised trials. BMJ 2011;343:d5928. [4] Bucher HC, Guyatt GH, Griffith LE, Walter SD. The results of direct and indirect treatment comparisons in meta-analysis of randomized controlled trials. J Clin Epidemiol 1997;50:683–91. [5] Wells GA, Sultan SA, Chen L, Khan M, Coyle D. Indirect treatment comparison [computer program]. Version 1.0. Ottawa: Canadian Agency for Drugs and Technologies in Health; 2009. [6] Schulman S, Kearon C, Kakkar A, Mismetti P, Schellong S, Eriksson H, et al. Dabigatran versus warfarin in the treatment of acute venous thromboembolism. N Engl J Med 2009;361:2342–52. [7] Schulman S, Kakkar AK, Goldhaber SZ, Schellong S, Eriksson H, Mismetti P, et al. Treatment of acute venous thromboembolism with dabigatran or warfarin and pooled analysis. Circulation 2013. http://dx.doi.org/10.1161/CIRCULATIONAHA.113.004450. [8] The EINSTEIN-Investigators. Oral rivaroxaban for the treatment of symptomatic venous thromboembolism. N Engl J Med 2010;363:2499–510. [9] The EINSTEIN-PE Investigators. Oral rivaroxaban for the treatment of symptomatic pulmonary embolism. N Engl J Med 2012;366:1287–97. [10] Agnelli G, Buller HR, Cohen A, Curto M, Gallus AS, Johnson M, et al. Oral apixaban for the treatment of acute venous thromboembolism. N Engl J Med 2013;369:799–808. [11] The Hokusai-VTE Investigators. Edoxaban versus warfarin for the treatment of symptomatic venous thromboembolism. N Engl J Med 2013;369:1406–15. [12] Prins MH, Lensing AWA, Bauersachs R, van Bellen B, Bounameaux H, Brighton TA, et al. Oral rivaroxaban versus standard therapy for the treatment of symptomatic venous thromboembolism: a pooled-analysis of the EINSTEIN-DVT and PE randomized studies. Thromb J 2013;11:21. [13] Fox BD, Kahn SR, Langleben D, Eisenberg MJ, Shimony A. Efficacy and safety of novel oral anticoagulants for treatment of acute venous thromboembolism: direct and adjusted indirect meta-analysis of randomized controlled trials. BMJ 2012;345:e7498. [14] Jansen JP, Fleurence R, Devine B, Itzler R, Barrett A, Hawkins N, et al. Interpreting indirect treatment comparisons and network meta-analysis for health-care decision making: report of the ISPOR Task Force on Indirect Treatment Comparisons Good Research Practices: part 1. Value Health 2011;14:417–28. [15] Hoaglin DC, Hawkins N, Jansen JP, Scott DA, Itzler R, Cappelleri JC, et al. Conducting indirect-treatment-comparison and network-meta-analysis studies: report of the ISPOR Task Force on Indirect Treatment Comparisons Good Research Practices: part 2. Value Health 2011;14:429–37.
Contents lists available at ScienceDirect
Thrombosis Research journal homepage: www.elsevier.com/locate/thromres
Regular Article
Indirect treatment comparison of new oral anticoagulants for the treatment of acute venous thromboembolism Nayon Kang, Diana M. Sobieraj ⁎ University of Connecticut School of Pharmacy, 69 N Eagleville Rd. Unit 3092, Storrs, CT 06269, United States
a r t i c l e
i n f o
Article history: Received 10 February 2014 Received in revised form 11 March 2014 Accepted 18 March 2014 Available online 24 March 2014 Keywords: Venous thromboembolism Anticoagulation Treatment
a b s t r a c t Background: Numerous new oral anticoagulants (NOACs) have been compared to a parenteral anticoagulant/oral vitamin K antagonist (VKA) for the treatment of acute venous thromboembolism (VTE). We aimed to conduct a systematic review and adjusted indirect comparison meta-analysis to compare the efficacy and safety of NOACs for this indication. Methods: We conducted a systematic literature search through November 2013 for randomized trials that evaluated treatment of acute VTE with a NOAC including rivaroxaban, apixaban, dabigatran and edoxaban. Trials had to report at least one of the following outcomes of interest: mortality, recurrent VTE, recurrent pulmonary embolism (PE), recurrent deep vein thrombosis (DVT), or major bleeding. Included trials were evaluated for quality using the Cochrane Risk of Bias tool. We performed an adjusted indirect comparison meta-analysis to evaluate the comparative efficacy and safety of NOACs, reporting relative risks (RRs) and 95% confidence intervals for each outcome. Results: Six trials (n = 27,069) met inclusion criteria, one each evaluating apixaban and edoxaban and two trials each evaluating rivaroxaban and dabigatran. Risk of bias was low for all trials. NOACS did not differ significantly in the risk of mortality, recurrent VTE, recurrent PE or recurrent DVT. Dabigatran increased major bleeding risk compared to apixaban [RR 2.69 (1.19 to 6.07)] as did edoxaban compared to apixaban [RR 2.74 (1.40 to 5.39)]. Conclusion: Although NOACs do not appear to differ in the efficacy of treating acute VTE, data suggests apixaban to be the safer than some of its competitors. © 2014 Elsevier Ltd. All rights reserved.
Introduction Acute venous thromboembolism (VTE) is a common disorder with an estimated annual incidence of 1.83 adults per 1000, with deep vein thrombosis (DVT) occurring more frequently than pulmonary embolism (PE) [1]. For decades, the standard treatment strategy has been bridging a parenteral heparin product and an oral vitamin-K antagonist (VKA) until a therapeutic international normalized ratio (INR) is achieved, with continuation of the VKA for a minimum of three months [2]. While effective, this treatment regimen has several limitations. Parenteral administration can be unfavorable to patients and requires additional nurse time in the inpatient setting and possibly on an outpatient basis. Routine INR monitoring requires transportation and often leads to dose-adjustments which presents an opportunity for medication errors. Lastly, there is a potential for significant drug and food interactions with
VKAs which can become important given their narrow therapeutic window. In recent years, several new oral anticoagulants (NOACs) have been under development for the treatment of acute VTE. NOACs that are currently approved by the Food and Drug Administration (FDA) or with phase III trial results include the factor Xa inhibitors apixaban, rivaroxaban, and edoxaban and the direct thrombin inhibitor dabigatran. For the acute treatment of VTE, NOAC have been compared to the gold standard regimen of a parenteral heparin product plus an oral VKA, although they have yet to be evaluated in head-to-head trials. In the absence of direct comparative evidence, indirect comparisons may provide information to aid in clinical decision making and therefore we aimed to conduct a systematic review and an adjusted indirect comparison meta-analysis to evaluate the efficacy and safety of NOACs for the treatment of acute VTE. Methods
Abbreviations: CI, confidence interval; DVT, deep vein thrombosis; FDA, Food and Drug Administration; LMWH, low-molecular weight heparin; NOAC, New oral anticoagulant; PE, pulmonary embolism; RR, relative risk; UFH, unfractionated heparin; VKA, vitamin K antagonist; VTE, venous thromboembolism. ⁎ Corresponding author. Tel.: +1 860 545 2429; fax: +1 860 5452277. E-mail address: [email protected] (D.M. Sobieraj).
http://dx.doi.org/10.1016/j.thromres.2014.03.035 0049-3848/© 2014 Elsevier Ltd. All rights reserved.
We conducted a systematic literature search in MEDLINE and Cochrane Central databases through November 2013 using the search strategy in e-Appendix A. A manual search was also performed using the references of clinical trials and review articles to identify additional
1146
N. Kang, D.M. Sobieraj / Thrombosis Research 133 (2014) 1145–1151
relevant articles. In order for a study to be included in the analysis, it had to be a randomized controlled trial that evaluated patients with acute VTE treated with a NOAC and reported at least one outcome of interest. NOACs included were those with current FDA approval or with published phase III trial results. Only studies evaluating the FDA approved dosing regimen for rivaroxaban were included and only studies using the same dosing regimen evaluated in phase III trials were included for the remaining NOACs. Outcomes of interest included mortality, recurrent VTE, recurrent DVT, recurrent PE and major bleeding. Two independent investigators separately reviewed all citations identified by the search for inclusion and abstracted data from included trials. Disagreements were resolved through discussion. The following data was collected from each trial: author identification, year of publication, funding source, report of conflicts of interest, study design
characteristics, study population (inclusion and exclusion criteria, geographic location, length of study, duration of patient follow-up), patient baseline characteristics, VTE treatment regimen (name, strength, frequency, dose, route of administration, duration of therapy, time in therapeutic range for VKA arms), and outcomes data (number of events, definitions, period of follow-up, and diagnostic tests for confirmation). To assess the methodological quality of the included trials, the Cochrane Collaboration’s risk of bias tool was used [3]. This tool evaluates seven domains including sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective outcome reporting and any other identifiable issues. Each domain is assessed as having low, high, or unclear risk of bias and then a summary assessment of each trial across domains is made as low, high or unclear risk of bias.
Abbreviations: RCT= randomized controlled trial; VTE=venous thromboembolism Fig. 1. Inclusion of studies. Abbreviations: RCT = randomized controlled trial; VTE = venous thromboembolism.
Table 1 Baseline characteristics. Trial
Randomized intervention (n)
Overall treatment duration d (SD)
RECOVER-II N = 2589
Dabigatran 150 mg BID × 6 m (n = 1279)
NR
EINSTEIN-DVT N = 3449
RECOVER N = 2564
Prior VTE N (%)
Recent Trauma/ Surgery N (%)
781 (61)
54.7 (16.2)
NR
247 (19.3)
NR
776 (60.2)
55.1 (16.3)
1569 (58.3)
57.2 (16.0)
1598 (59.1)
56.7 (16.0)
250 (111.8)
2360 (57.3)
55.7 (16.3)
Enoxaparin or heparin × 5d + warfarin daily for 3-12 m (n = 4122) Rivaroxaban 15 mg BID × 3w then 20 mg daily for 3, 6, or 12 m (n = 2419) Enoxaparin 1 mg/kg BID × 5d + VKA daily for 3, 6, or 12 m (n = 2413) Rivaroxaban 15 mg BID × 3w then 20 mg daily for 3, 6, or 12 m (n = 1731) Enoxaparin 1 mg/kg BID × 5d + VKA daily for 3, 6, or 12 m (n = 1718) Dabigatran 150 mg BID × 6 m (n = 1273)
248.4 (112.6)
2356 (57.2)
55.9 (16.2)
216 (NR)
1309 (54.1)
57.9 (7.3)
214 (NR)
1247 (51.7)
57.5 (7.2)
993 (57.4)
55.8 (16.4)
967 (56.3)
56.4 (16.3)
DVT: 1718 (100)
738 (58)
55.0 (15.8)
746 (58.9)
54.4 (16.2)
DVT: 880 (69.1) PE: 270 (21.2) Both: 121 (9.5) DVT: 869 (68.6) PE: 271 (21.4) Both: 124 (9.8)
NR
Enoxaparin 1 mg/kg BID × 5d + VKA daily for 6 m (n = 1266)
NR
NR
203 (15.8)
Active Cancer N (%) 50 (3.9)
Known Thrombophilia N (%)
Risk of Bias
NR
Low
50 (3.9)
TTR: 57% Above: 19% Below: 24%
2416 (89.8)
463 (17.2)
2429 (89.8)
409 (15.1)
2713 (65.9)
784 (19.0)⁎
2697 (65.4)
736 (17.9)⁎
1566 (64.7)
455 (18.8)
415 (17.2)⁎
114 (4.7)⁎
138 (5.7)
1551 (64.3)
489 (20.3)
398 (16.5)⁎
109 (4.5)⁎
121 (5.0)
1055 (60.9)
336 (19.4)
338 (19.5)⁎
118 (6.8)⁎
107 (6.2)
1083 (63.0)
330 (19.2)
335 (19.5)⁎
89 (5.2)⁎
116 (6.8)
NR
327 (25.7)
NR
64 (5.0)
NR
322 (25.4)
NR
NR
66 (2.5)
74 (2.8)
77 (2.8)
59 (2.2)
378 (9.2)⁎
NR
Low
TTR: 61 Above: 16 Below: 23 Low
393 (9.5)⁎
57 (4.5)
TTR %
TTR: 63.5 Above: 17.6 Below: 18.9 Low
TTR: 62.7 Above: 15.5 Below: 21.8 Low
N. Kang, D.M. Sobieraj / Thrombosis Research 133 (2014) 1145–1151
EINSTEIN-PE N = 4832
Un-provoked Index Event N (%)
Edoxaban 30 mg or 60 mg daily for 3-12 m (n = 4118)
Apixaban 10 mg BID × 7d then 5 mg BID × 6 m (n = 2691) Enoxaparin 1 mg/kg BID × 5d + VKA daily for 6 m (n = 2704)
Hokusai-VTE N = 8240
Index event N (%)
DVT: 877 (68.5) PE: 298 (23.3) Both: 104 (8.1) DVT: 873 (67.8) PE: 297 (23.1) Both: 117 (9.1) DVT: 1749 (65.0) PE: 678 (25.2) Both: 252 (9.4) DVT: 1783 (65.9) PE: 681 (25.2) Both: 225 (8.3) DVT:2468 (59.9) PE: 1240 (30.1) Both: 410 (10.0) DVT: 2453 (59.5) PE: 1265 (30.7) Both: 404 (9.8) DVT: 0 PE: 1813 (74.9) Both: 606 (25.1) DVT: 0 PE: 1823 (75.5) Both: 590 (24.5) DVT: 1731 (100)
Enoxaparin 1 mg/kg BID × 5d + VKA daily for 6 m (n = 1289) AMPLIFY N = 5395
Age y (SD)
Males N (%)
TTR: 57.7 Above: 16.2 Below: 24.4 Low
TTR: 60 Above: 19 Below: 21
Abbreviations: d = days; BID = twice a day; m = months; mg = milligrams; NR = not reported; SD = standard deviation; TTR = time in therapeutic range; VKA = vitamin k antagonist; VTE = venous thromboembolism; w = weeks. ⁎ Represents patients whose index event was attributed to given characteristic.
1147
1148
N. Kang, D.M. Sobieraj / Thrombosis Research 133 (2014) 1145–1151
Fig. 2. Network diagram. This diagram represents the network created by the included trials. Therapies that are directly compared in at least one trial are connected with a solid line. The actual number of trials for each comparison is depicted.
The risk of bias was evaluated for each trial by two separate investigators and conflicts were resolved through discussion. In the absence of direct comparative trials, methods for indirect comparison can be employed. For example, if one trial compares treatment A to treatment B and a second trial compares treatment C to treatment B, methods of indirect comparison can be used to estimate the relative effects of treatment A to treatment C through the common comparator treatment B. We carried out such analysis using the adjusted indirect comparison meta-analysis methods proposed by Bucher et al. [4] First, we calculated pairwise relative risks (RR) and corresponding 95% confidence intervals (CIs) for each comparison (each individual NOAC versus the common comparator of a parenteral anticoagulant plus oral VKA) and outcome using StatsDirect version 2.7.8. If only one trial representing a given comparison was available, we calculated a RR and 95% CI for each outcome using the trial data. If more than one trial representing the same comparison was available, traditional metaanalysis using a random effects model was used to generate a pooled RR and 95% CI. Using the pairwise data, we compared each NOAC to another using the publically available software for adjusted indirect metaanalysis [5]. RR and 95% CIs were generated for each indirect comparison and outcome. One sensitivity analyses was run to observe the effects of pooling the EINSTEIN trials using meta-analytic techniques instead of using the collective data published in Prins et al. Results Study selection and characteristics The results of our literature search are presented in Fig. 1. A total of 396 citations were identified through our literature search with 3 manually added citations. Of those, two were excluded as duplicate publications, and 326 and 65 were excluded at the abstract and full text level, respectively, leaving 6 unique trials (Table 1, Fig. 2) [6–11].
EINSTEIN-PE and EINSTEIN-DVT evaluated rivaroxaban (the pooled outcomes data published by Prins et al. [12] were used in place of the individual trials), RECOVER-I and RECOVER-II evaluated dabigatran, AMPLIFY evaluated apixaban, and Hokusai-VTE evaluated edoxaban. All trials used the common comparator of a parenteral anticoagulant plus an oral VKA. In RECOVER-I and RECOVER-II, unfractionated heparin (UFH), low-molecular weight heparin (LMWH) and fondaparinux were parenteral anticoagulant options although 89 to 90% of patients were treated with LMWH. In AMPLIFY either UFH or LMWH could have been used. In the remaining trials enoxaparin was the specified parenteral anticoagulant. All trials were published as full text articles were judged to have low risk of bias. The six included trials represent a total of 27,069 patients with acute VTE. With exception of EINSTEIN-DVT which enrolled patients with an index DVT and EINSTEIN-PE which enrolled patients with an index PE or both a PE and DVT, the index event was a DVT in 60 to 70% of patients, PE in 21 to 31% of patients and both in 8 to 10% of patients. Although in all trials patients could have been treated with therapeutic anticoagulant doses prior to randomization for a limited period of time (typically b48 hours), in the RECOVER-I and II and Hokusai-VTE trials initial treatment with a parenteral anticoagulant was required in all patients. The median treatment duration in the RECOVER-I trial was 9 days, in RECOVER-II was 9.4 to 9.6 days, and in Hokusai-VTE was 7 days. Randomized treatment duration was fixed in AMPLIFY and the RECOVER trials at 6 months while clinician discretion determined duration in the remaining trials, which could range from 3 to 12 months. The mean number of treatment days in Hokusai-VTE ranged from 248 to 250 days and 204 to 207 days for the EINSTEIN trials collectively. The time in therapeutic range was similar across trials at around 60%, while time spent outside of the range was slightly more below range (19 to 24%) than above range (16 to 19%). Patient characteristics were balanced across groups with the exception of cases of unprovoked VTE which were higher in the AMPLIFY trial.
Table 2 Pairwise relative risks for each new oral anticoagulant versus parenteral anticoagulant plus vitamin K antagonist.⁎ Base-case analyses
Rivaroxaban Dabigatran Apixaban Edoxaban
Morality
Recurrent VTE
Recurrent PE
Recurrent DVT
Major bleeding
0.97 (0.73 to 1.27) 1.00 (0.67 to 1.50) 0.79 (0.53 to 1.19) 1.05 (0.83 to 1.33)
0.90 (0.68 to 1.20) 1.12 (0.79 to 1.60) 0.84 (0.60 to 1.18) 0.89 (0.71 to 1.12)
1.08 (0.74 to 1.57) 1.04 (0.59 to 1.85) 1.06 (0.69 to 1.65) 0.88 (0.65 to 1.20)
0.70 (0.45 to 1.09) 1.17 (0.70 to 1.93) 0.61 (0.36 to 1.06) 0.91 (0.64 to 1.29)
0.55 (0.38 to 0.81) 0.76 (0.49 to 1.18) 0.31 (0.17 to 0.54) 0.85 (0.60 to 1.21)
0.95 (0.64 to 1.42)
0.90 (0.56 to 1.43)
1.07 (0.73 to 1.58)
0.71 (0.40 to 1.24)
0.56 (0.38 to 0.82)
Sensitivity analysis Rivaroxaban
⁎ Values represent relative risks and 95% confidence intervals.
N. Kang, D.M. Sobieraj / Thrombosis Research 133 (2014) 1145–1151
Indirect Comparison Results The relative risks generated for each NOAC versus parenteral anticoagulant plus VKA are presented in Table 2. Four NOACs were compared indirectly: dabigatran, apixaban, rivaroxaban and edoxaban. For the
1149
efficacy outcomes including mortality, recurrent symptomatic VTE, recurrent DVT and current PE there were no significant differences amongst all of the comparisons made (Fig. 3). The risk of major bleeding increased with dabigatran compared to apixaban [RR 2.69 (1.19 to 6.07)] and with edoxaban compared to apixaban [RR 2.74 (1.40 to
Abbreviations: DVT=deep vein thrombosis; PE=pulmonary embolism; VTE=venous thromboembolism Fig. 3. Results of the indirect treatment comparison. Abbreviations: DVT = deep vein thrombosis; PE = pulmonary embolism; VTE = venous thromboembolism.
1150
N. Kang, D.M. Sobieraj / Thrombosis Research 133 (2014) 1145–1151
Abbreviations: DVT=deep vein thrombosis; PE=pulmonary embolism; VTE=venous thromboembolism Fig. 4. Results of sensitivity analysis. Abbreviations: DVT = deep vein thrombosis; PE = pulmonary embolism; VTE = venous thromboembolism.
5.39)]. Results of the sensitivity analysis are presented in Fig. 4. Overall our original results remained robust and did not change as a result of pooling the EINSTEIN trials. Discussion This analysis is based on over 25,000 patients treated with a NOAC for acute VTE for up to 12 months. Although there does not appear to be any significant difference in terms of efficacy suggested by the results, the NOACs do not appear to be similar in terms of bleeding risk. The risk of major bleeding was increased with dabigatran and with edoxaban in comparison to apixaban. Although the results for rivaroxaban versus other NOAC did not reach statistical significance, when compared to apixaban there was a trend in the direction of increased risk of major bleeding with rivaroxaban. These data suggest that apixaban may have the safest bleeding risk profile among the NOACs studied. There are several potential factors which could lead to this observation. Required use of parenteral anticoagulant treatment in the RECOVER and Hokusai-VTE trials could impact early bleeding rates favoring the other NOACs apixaban or rivaroxaban. Although both apixaban and rivaroxaban require a higher initial dose followed by a lower maintenance dose, the initial high-dose period is shorter for apixaban compared to rivaroxaban (7 versus 21 days) which could favor lower bleeding risk with apixaban. The mean number of treatment days in Hokusai-VTE were 248 to 250 days and in the EINSTEIN trials collectively were 204 to 207 days. Naturally with the option of longer treatment duration in these trials there was a longer period for events to accumulate in comparison to AMPLIFY and RECOVER, which could favor apixaban and dabigatran. One other published indirect treatment comparison by Fox et al. analyzes NOACs for acute VTE treatment [13]. The authors concluded that
there were no significant differences in mortality, recurrent VTE, or major bleeding when rivaroxaban and dabigatran were compared, which is similar to the results of our analysis. However, they excluded apixaban because there was “only 1 small trial” and their search preceded the publication of AMPLIFY which we have included. Their analysis also differed from ours in that dose-finding trials were included in their analysis and how different doses were handled or included was not specified. This limits the applicability of their findings since the doses available to prescribers are not purely represented and this methodology could favor less bleeding or lower efficacy if lower doses were pooled with standard, higher doses. Our analysis reflects the most current literature and also includes additional NOAC which Fox and colleagues were unable to evaluate. Our literature search was also more recent therefore we were able to include the results of RECORD-II published in full text in our analysis. There are limitations to our analysis that should be noted. Adjusted indirect treatment comparison meta-analysis is an established statistical technique in the scientific literature [14,15]. While seen as an advantage in situations where no direct comparative data exists, this can also be considered a limitation as data is purely generated using indirect evidence. Therefore, results are not as precise as what is possible with direct evidence. Since our network was without direct evidence, we are unable to evaluate consistency which is one factor considered important to the internal validity of indirect comparisons [14]. However, we believe our analysis is robust since in our judgment, trials met the similarity assumption and the individual trials were of high quality, which are two additional factors important to the internal validity of indirect comparisons [14]. We chose to use the data published in Prins et al. to represent the EINSTEIN trials collectively in an effort to reduce heterogeneity of populations across studies since all other trials included patients with an index DVT or PE, not just one VTE type. Through
N. Kang, D.M. Sobieraj / Thrombosis Research 133 (2014) 1145–1151
sensitivity analysis we evaluated if pooling the EINSTEIN trials using meta-analytic techniques would vary results rather than using the collective data, and again the results did not change overall. In summary, NOAC are similar in terms of efficacy for the acute treatment of VTE, although apixaban appears to have an advantageous safety profile as indirect estimates suggest the other evaluated NOAC have a higher risk of bleeding. Conflicts of interest The authors have no conflicts to disclose. Funding This work was unfunded. Acknowledgements None Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.thromres.2014.03.035. References [1] Oger E. Incidence of venous thromboembolism: a community-based study in Western France. EPI-GETBP Study Group. Thromb Haemost 2000;83:657–60. [2] Kearon C, Akle EA, Comerota AJ, Prandoni P, Bounameaux H, Goldhaber SZ, et al. Antithrombotic therapy for VTE disease: antithrombotic therapy and prevention of thrombosis. American College of Chest Physicians Evidence-based Clinical Practice Guidelines, 141, Chest, 9th ed. 2012. p. e419S–94S.
1151
[3] Higgins JPT, Altman DG, Gotzsche PC, Juni P, Moher D, Oxman AD, et al. The Cochrane Collaboration’s tool for assessing risk of bias in randomised trials. BMJ 2011;343:d5928. [4] Bucher HC, Guyatt GH, Griffith LE, Walter SD. The results of direct and indirect treatment comparisons in meta-analysis of randomized controlled trials. J Clin Epidemiol 1997;50:683–91. [5] Wells GA, Sultan SA, Chen L, Khan M, Coyle D. Indirect treatment comparison [computer program]. Version 1.0. Ottawa: Canadian Agency for Drugs and Technologies in Health; 2009. [6] Schulman S, Kearon C, Kakkar A, Mismetti P, Schellong S, Eriksson H, et al. Dabigatran versus warfarin in the treatment of acute venous thromboembolism. N Engl J Med 2009;361:2342–52. [7] Schulman S, Kakkar AK, Goldhaber SZ, Schellong S, Eriksson H, Mismetti P, et al. Treatment of acute venous thromboembolism with dabigatran or warfarin and pooled analysis. Circulation 2013. http://dx.doi.org/10.1161/CIRCULATIONAHA.113.004450. [8] The EINSTEIN-Investigators. Oral rivaroxaban for the treatment of symptomatic venous thromboembolism. N Engl J Med 2010;363:2499–510. [9] The EINSTEIN-PE Investigators. Oral rivaroxaban for the treatment of symptomatic pulmonary embolism. N Engl J Med 2012;366:1287–97. [10] Agnelli G, Buller HR, Cohen A, Curto M, Gallus AS, Johnson M, et al. Oral apixaban for the treatment of acute venous thromboembolism. N Engl J Med 2013;369:799–808. [11] The Hokusai-VTE Investigators. Edoxaban versus warfarin for the treatment of symptomatic venous thromboembolism. N Engl J Med 2013;369:1406–15. [12] Prins MH, Lensing AWA, Bauersachs R, van Bellen B, Bounameaux H, Brighton TA, et al. Oral rivaroxaban versus standard therapy for the treatment of symptomatic venous thromboembolism: a pooled-analysis of the EINSTEIN-DVT and PE randomized studies. Thromb J 2013;11:21. [13] Fox BD, Kahn SR, Langleben D, Eisenberg MJ, Shimony A. Efficacy and safety of novel oral anticoagulants for treatment of acute venous thromboembolism: direct and adjusted indirect meta-analysis of randomized controlled trials. BMJ 2012;345:e7498. [14] Jansen JP, Fleurence R, Devine B, Itzler R, Barrett A, Hawkins N, et al. Interpreting indirect treatment comparisons and network meta-analysis for health-care decision making: report of the ISPOR Task Force on Indirect Treatment Comparisons Good Research Practices: part 1. Value Health 2011;14:417–28. [15] Hoaglin DC, Hawkins N, Jansen JP, Scott DA, Itzler R, Cappelleri JC, et al. Conducting indirect-treatment-comparison and network-meta-analysis studies: report of the ISPOR Task Force on Indirect Treatment Comparisons Good Research Practices: part 2. Value Health 2011;14:429–37.
