3 The manufacturer's submission
The Appraisal Committee (appendix A) considered evidence submitted by the manufacturer of dabigatran etexilate for the prevention of venous thromboembolism (VTE) after hip or knee replacement surgery in adults and a review of this submission by the Evidence Review Group (ERG; appendix B).
3.1 The manufacturer's submission compared dabigatran etexilate (at 150 mg and 220 mg daily doses) with enoxaparin, a low molecular weight heparin (LMWH), using direct evidence from randomised controlled trials (RCTs), and with fondaparinux, using RCT evidence which had been incorporated into a mixed-treatment comparison. Outcomes analysed included: mortality; incidence of deep vein thrombosis (DVT); incidence of pulmonary embolism (PE); adverse effects of treatment including bleeding events; post-DVT complications including post-thrombotic syndrome; length of hospital stay; and health-related quality of life. All of these outcomes were specified in the decision problem for this appraisal. The manufacturer's submission (MS) did not include analysis of outcomes at the site of the orthopaedic intervention, such as joint infection.
3.2 The manufacturer conducted a systematic review which included three randomised, active-controlled parallel-group, non-inferiority trials of dabigatran etexilate (each including two dosing regimes) versus enoxaparin. These trials were:
RE-NOVATE: a pivotal phase III, double-blind RCT of elective total hip replacement patients where dabigatran etexilate 150 mg or 220 mg once daily (started 1–4 hours after surgery with a half dose of 75 mg or 110 mg, respectively) was compared with 40 mg enoxaparin once daily (started the day before surgery). Both treatments were continued for 28–35 days (n = 3494 randomised)
RE-MODEL: a pivotal phase III, double-blind RCT of elective total knee replacement patients where dabigatran etexilate 150 mg or 220 mg once daily (started 1–4 hours after surgery with a half dose of 75 mg or 110 mg, respectively ) was compared with 40 mg enoxaparin once daily (started the day before surgery). Both treatments were continued for 6–10 days (n = 2101 randomised).
RE-MOBILIZE: described in the manufacturer's submission as a supporting North American, phase III, double-blind RCT of elective total knee replacement patients where dabigatran etexilate 150 mg or 220 mg once daily (started 6–12 hours after surgery with a half dose of 75 mg or 110 mg, respectively) was compared with 30 mg enoxaparin twice daily (started 12–24 hours after surgery). Both treatments were continued for 12–15 days (n = 2615 randomised).
The primary efficacy outcome for all three trials was a composite of total incidence of VTE (proximal and distal DVT based on venogram or objectively confirmed symptomatic DVT and PE) and all-cause mortality. Follow-up in all trials was 12–14 weeks after surgery. Participants were randomised to 150 or 220 mg doses irrespective of renal function status.
3.3 A range of meta-analyses were conducted which combined the two pivotal (RE-NOVATE, RE-MODEL), two knee (RE-MODEL, RE-MOBILIZE) and all three trials. Two phase II trials were excluded from meta-analysis. One trial was excluded because the dosing regimen was different to that specified in the marketing authorisation, and the other was excluded because only preliminary data were reported for dabigatran etexilate compared with placebo.
3.4 The results of primary efficacy and bleeding outcomes reported in individual RCTs are detailed below.
The total incidence of the composite outcome of VTE and all-cause mortality at 12–14 weeks after surgery in:
RE-NOVATE was 6% (n = 880), 8.6% (n = 874) and 6.7% (n = 897) for dabigatran etexilate 220 mg, dabigatran etexilate 150 mg and enoxaparin, respectively. Absolute risk differences were −0.7% (95% confidence interval [CI] −2.9 to 1.6) and 1.9% (95% CI −0.6 to 4.4) for 220 mg and150 mg respectively compared with enoxaparin. Relative risks were 0.9 (95% CI 0.63 to 1.29) and 1.28 (95% CI 0.93 to 1.78) for 220 mg and150 mg respectively compared with enoxaparin.
RE-MODEL was 36.4% (n = 503), 40.5% (n = 526) and 37.7% (n = 512) for dabigatran etexilate 220 mg, dabigatran etexilate 150 mg and enoxaparin, respectively. Absolute risk differences were –1.3% (95% CI –7.3 to 4.6) and 2.8% (95% CI –3.1 to 8.7) for 220 mg and150 mg respectively compared with enoxaparin. Relative risks were 0.97 (95% CI 0.82 to 1.13) and 1.07 (95% CI 0.92 to 1.25) for 220 mg and 150 mg respectively compared with enoxaparin.
RE-MOBILIZE was 33.1% (n = 604), 33.7% (n = 649) and 25.3% (n = 643) for dabigatran etexilate 220 mg, dabigatran etexilate 150 mg and enoxaparin (30 mg twice daily), respectively. Absolute risk differences were 5.8% (95% CI 0.8 to 10.8) and 8.4% (95% CI 3.4 to 13.3) for 220 mg and150 mg respectively compared with enoxaparin. Relative risks were 1.23 (95%CI 1.03 to 1.47) and 1.33 (95% CI 1.12 to 1.58) for 220 mg and 150 mg respectively compared with enoxaparin (30 mg twice daily).
3.5 The incidence of the composite outcome of major bleeding and clinically relevant bleeding in:
RE-NOVATE was 6.2% (n = 1146), 6.0% (n = 1163) and 5.0% (n = 1154) for dabigatran etexilate 220 mg, dabigatran etexilate 150 mg and enoxaparin, respectively. Absolute risk differences were 1.2% (95% CI –0.7 to 3.1) and 1.0% (95% CI –0.9 to 2.9) for dabigatran etexilate 220 mg and dabigatran etexilate 150 mg respectively compared with enoxaparin.
RE-MODEL was 7.4% (n = 679), 8.1% (n = 703) and 6.6% (n = 694) for dabigatran etexilate 220 mg, dabigatran etexilate 150 mg and enoxaparin, respectively. Absolute risk differences were 0.7% (95% CI –2.0 to 3.4) and 1.5% (95% CI –1.3 to 4.2) for 220 mg and 150 mg respectively compared with enoxaparin.
RE-MOBILIZE was 3.3% (n = 857), 3.1% (n = 871) and 3.8% (n = 868) for dabigatran etexilate 220 mg, dabigatran etexilate 150 mg and enoxaparin (30 mg twice daily), respectively. Absolute risk differences were –0.5% (95% CI –2.3 to 1.2) and –0.7% (95% CI –2.4 to 1.0) for 220 mg and 150 mg respectively compared with enoxaparin (30 mg twice daily).
3.6 In the absence of direct evidence comparing dabigatran etexilate with fondaparinux, the manufacturer presented results of a mixed-treatment comparison based on a recent meta-analysis performed for the NICE clinical guideline 'Venous thromboembolism' (NICE clinical guideline 46 [Replaced by NICE clinical guideline 92]). Results were expressed as relative risks for each intervention versus no intervention. The analysis included trial data for dabigatran etexilate and fondaparinux as well as for other VTE prevention strategies (such as aspirin and non-pharmacological prevention), although this was not part of the decision problem specified in the scope. The manufacturer reported that fondaparinux and LMWH (given for an 'extended duration' of more than 14 days) were associated with lower relative risk for DVT than other existing interventions (compared to no intervention). The manufacturer also reported that fondaparinux had the highest relative risk for bleeding, but noted that confidence intervals for all the interventions overlapped. The manufacturer stated that dabigatran compared favourably with existing alternatives in terms of both efficacy and safety.
3.7 The manufacturer submitted an economic model assessing the impact of dabigatran etexilate for VTE prevention after hip and knee replacement compared with LMWH and fondaparinux. The model included an acute-phase decision-tree model to 10 weeks after surgery and a chronic-phase Markov model with a lifetime (60-year) time horizon.
3.8 The acute-phase model predicted transition between health states based on evidence from the two pivotal phase III trials of dabigatran etexilate (RE-NOVATE and RE-MODEL) compared with enoxaparin and the mixed-treatment comparison for dabigatran etexilate and fondaparinux. At 10 weeks, patients entered the chronic-phase Markov model in the same health state in which they ended the decision-tree model. No further treatment effect was applied in the chronic-phase model. Transitions between states in the chronic-phase model were dependent on recurrence rates for VTE from the literature. The health states in the chronic-phase model were: well; asymptomatic untreated VTE (proximal DVT, distal DVT and PE); treated VTE for people surviving after symptomatic VTE (proximal DVT, distal DVT and PE); recurrent DVT or PE; mild to moderate post-thrombotic syndrome; severe post-thrombotic syndrome; disabled (owing to an intracranial bleed); or dead.
3.9 Key assumptions in the economic evaluation are detailed in the manufacturer's submission. Among these, the manufacturer assumed that all LMWHs are bioequivalent, because literature on dalteparin, tinzaparin and enoxaparin and the NICE clinical guideline 'Venous thromboembolism' (NICE clinical guideline 46 [Replaced by NICE clinical guideline 92]) recommendations did not distinguish between LMWHs. Furthermore, a zero cost for administration was assumed for dabigatran etexilate, whereas LMWH and fondaparinux were assumed to require resources for administration (including provision for a proportion of people who were unable or unwilling to self-inject). These administration costs were determined to be £100.00 and £6.00 for LMWH and £83.00 and £6.00 for fondaparinux after hip or knee replacement, respectively.
3.10 The base-case analysis estimated that at 220 mg dabigatran etexilate was less costly and more effective than LMWH for both hip and knee replacement surgery. At the lower dose of 150 mg, dabigatran etexilate was less costly and more effective than LMWH for hip replacement surgery, but was more costly and less effective than LMWH for knee replacement surgery. In univariate sensitivity analyses none of the parameters were associated with a significant difference in the base-case results.
3.11 The economic evaluation estimated that at both doses dabigatran etexilate is less costly but also less effective than fondaparinux after hip replacement (ICERs were in the 'southwest' quadrant of the cost-effectiveness plane). After knee replacement, dabigatran etexilate at both doses was dominated by fondaparinux (that is, it was more costly and less effective than fondaparinux). In sensitivity analysis, increasing the relative risk of VTE for fondaparinux was associated with dabigatran etexilate dominating for hip replacement and being less costly, but being less effective in knee replacement.
3.12 Probabilistic sensitivity analysis and cost-effectiveness acceptability curves suggested probabilities of dabigatran etexilate being cost effective compared with LMWH (at a willingness-to-pay threshold range of £20,000 per additional QALY gained) of 99% for the 220-mg dose after hip replacement, 82% for the 220-mg dose after knee replacement, 76% for the 150-mg dose after hip replacement, and 38% for the 150-mg dose after knee replacement). The corresponding results for dabigatran compared with fondaparinux were 40% for the 220-mg dose and 32% for the 150-mg dose after hip replacement, and zero for both doses after knee replacement).
3.13 Following a request for clarification from the ERG, the manufacturer provided cost-effectiveness analyses with inputs from meta-analysis which included data from the RE-MOBILIZE trial. The revised economic evaluation estimated that dabigatran etexilate was dominated by LMWH (that is, it was more costly and less effective than LMWH) for knee replacement at both 220-mg and 150-mg doses.
3.14 The ERG confirmed that the decision problem matched the marketing authorisation for dabigatran etexilate and that LMWH and fondaparinux were appropriate comparators. The ERG acknowledged that clinical trials did not routinely report hip or knee joint outcomes, such as joint infection. The ERG also noted that there was limited clinical experience of using dabigatran etexilate in the special populations (including people with moderate renal impairment, those over 75 years and people who are receiving amiodarone) defined in the SPC.
3.15 The ERG identified evidence of variation in the dosing regimen for LMWH. In the RE-NOVATE and RE-MODEL studies the protocol was for enoxaparin to be started before surgery, but in some countries treatment was started post-operatively to reflect local practice. It was not clear from reporting of the studies what proportion of patients received which regimen. In the RE-MOBILIZE study, enoxaparin was started post-operatively. The ERG stated that the trials may not completely reflect the practice of administering LMWH preoperatively in England and Wales. This variation in practice could introduce confounding because in these situations LMWH would not be administered to people with uncontrolled bleeding following joint replacement. 'Venous thromboembolism' (NICE clinical guideline 46 [Replaced by NICE clinical guideline 92]) states that the currently available randomised evidence is too limited to determine whether giving LMWH can be safely delayed until after surgery, or whether it must be given preoperatively. The guideline identifies this as a major evidence gap and is therefore non-specific about timing.
3.16 The ERG commented that the mixed-treatment comparison did not provide indirect comparisons of fondaparinux and dabigatran etexilate, making it difficult to reach conclusions about their relative efficacy and safety. The ERG also noted that the outcome assessed in the mixed-treatment comparison was DVT (not the composite primary outcome of the dabigatran etexilate trials). It was also unclear how the trial data had been used to derive the mixed-treatment comparison of DVT outcome. The ERG suggested that results of the manufacturer's mixed-treatment comparison should be considered with caution.
3.17 The ERG commented that the model structure, health states and parameters were reasonable. The univariate sensitivity analysis was extensive and performed with appropriate parameters. The probabilistic sensitivity analysis was performed correctly. However, the ERG noted that previously published models included progression from distal to proximal DVT, which the manufacturer's model did not. The literature search for recurrence of VTE, rates of post-thrombotic syndrome and quality of life data used in the model was limited to published economic studies. The ERG suggested that it was therefore possible that the manufacturer had not considered all applicable data in the model.
3.18 The ERG highlighted that small numerical differences in data from pivotal trials were reproduced in the model in terms of small incremental costs and small incremental health benefits. A small change in the direction of these inputs resulted in a similar change in the direction of the model results. Inclusion of data from meta-analyses that included the RE-MOBILIZE trial produced such a change in direction of results, and dabigatran etexilate became dominated by LMWH (that is, it was more costly and less effective than LMWH).
3.19 Full details of all the evidence are in the manufacturer's submission and the ERG report.