| Study | Applicability | Limitations | Other comments | Incremental cost | Incremental effects | Cost effectiveness | Uncertainty |
|---|---|---|---|---|---|---|---|
| Morris 200793 (UK) | Directly applicablea | Potentially serious limitationsb | Lifetime model based on patient level data from two randomised placebo-controlled phase II trials16,16. Data was supplemented with additional UK data to estimate costs and benefits (mortality following the trial duration, and QoL). | £13,243 | 0.70 QALYs | £18,825 per QALY | 52% (61%) probability of being cost effective at a threshold of £20,000 (£30,000). |
| Rossaint 2007 121,122 (Germany) | Partially applicablec | Potentially serious limitations d | Lifetime model based on patient level data from two randomised placebo-controlled phase II trials16,16. Data was supplemented with additional German data to estimate costs and benefits (mortality following the trial duration, and QoL). | £14,831 | 0.69 | £21,613 per QALY | 48% (60%) probability of being cost effective at a threshold of £20,000 (£30,000)h |
| Pohar 2009 113,113 (Canada) | Partially applicablee | Potentially serious limitationsf | Decision tree model based on patient level data from two randomised placebo-controlled phase II trials16,16., supplemented by further sources for costing, utilities and in extrapolation technique to estimate long-term survival estimates. | £20,342g | 1.68 QALYsg | £12,108 per QALYg | 36% (52%) probability of being cost effective at a threshold of £20,000 (£30,000)h |
Abbreviations: NR, not reported; QoL, quality of life; QALYs, quality-adjusted life years
Appropriate population (blunt severe trauma who are bleeding or at risk of bleeding), intervention and comparison, in a UK setting with costs and benefits discounted at 3.5%.
Adverse events not included (from the interventions and consequences of blood transfusions). The trial that the economic evaluation is based on is stated to be underpowered to detect mortality and comprised of a sample of 143 patients. Potential conflict of interest from the authors and funders (study funded by drug manufacturer). First two authors have received fees from the company and third and fourth authors are employees of the company. No information given on the structure of the model. Extrapolation methods used to predict probability of survival post 30 days are not explained enough to identify whether there may be any issues such as the previous stage in the 3 stage process are having an impact on the probability derived for later stages. Also the populations compared within the TARN database are stated to be older and less severely injured than the patients in the Boffard trial.
Appropriate population (blunt severe trauma who are bleeding or at risk of bleeding), intervention and comparison. Costs from a German third party payer perspective (social insurance). Costs and effects discounted at 5%.
Adverse events not included (from the interventions and consequences of blood transfusions). The trial that the economic evaluation is based on is stated to be underpowered to detect mortality and comprised of a sample of 143 patients. Potential conflict of interest from the authors and funders (original trial funded by drug manufacturer and most of the authors have received fees from Novo Nordisk). Limitations in trial data used to estimate of mortality in first 30 days may carry through in limiting the estimation of longer term mortality, thus limiting the lifetime horizon estimates
Appropriate population (blunt severe trauma who are bleeding or at risk of bleeding), intervention and comparison. Costs from the Canadian perspective. Does not report the discounted QALYs or ICER despite reporting that discounted values were calculated. Benefits discounted at 5%, costs not discounted as only include first year costs.
No adverse events considered. Costs beyond one year were not considered. Not possible to work out a discounted ICER as mean discounted QALYs not reported. So ICER estimated in table above using the undiscounted QALYs reported. The trial that the economic evaluation is based on is stated to be underpowered to detect mortality and comprised of a sample of 143 patients. Estimation of mortality post 30 days used data from the Rossaint paper (please see limitations described in footnote (d) above).
Only undiscounted values reported. Mean discounted QALY not reported, however, confidence interval for discounted incremental QALY reported to be -1.50 to 2.95 ICER presented was calculated by NCGC using undiscounted mean values.
From inspection of cost-effectiveness acceptability curve
From: 10, Assessment and management of haemorrhage
NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.