The authors justified their selection of the comparators, in that ENO was the most commonly used low molecular weight heparin, and FON was the new agent.
The approach used to identify the relevant studies was based on sources known to the authors rather than on a systematic search of the literature. Only a few details of the primary data were provided. The authors reported that most of the clinical data came from trials, which are usually considered to be a valid source of evidence due to their rigorous design. However, information on the sample size and other features of these trials was not provided. Also, data came from multiple international sources, which raises the issue of transferability to the Swedish setting. Both a disease-specific and a generalisable benefit measure were used. The authors acknowledged that the adjustment of the expected survival for quality of life would have been appropriate, but was not possible due to the limited published evidence. LYs are a validated and comparable benefit measure.
The analysis of costs was consistent with the study perspective. A breakdown of the cost items was not provided, with most of the costs being presented as macro-categories. The economic analysis relied on a previous cost study and little information on the methods used to derive the costs was given. Statistical analyses of costs were carried out to adjust the previous estimates for the current model. The price year, the use of discounting, and currency conversion rates were reported.
Analysis and results:
The findings were selectively presented given that the expected benefits were not reported. The issue of uncertainty was partially investigated in that the sensitivity analysis considered only variations in single model inputs which were not varied simultaneously. Alternative assumptions were made for the patterns of care, which may have reflected the Swedish setting. The lack of a comprehensive approach, such as a probabilistic analysis, was justified on the grounds of the scarcity of alternative estimates in the published literature. The authors provided extensive information on their decision model framework and discussed the issues related to model validation.
The quality of the study was satisfactory, although some sources of data were not fully reported. Overall, the authors’ conclusions appear to be valid.