Interventions:
The intervention was clearly described, along with the treatment strategy that followed. An appropriate comparator, standard care, was used. The different stages of standard care were clearly described, and were stated to be based on the established treatment pattern in France.
Effectiveness/benefits:
The effectiveness measure was clearly stated, but the description of the data was generally poor. No details of the retrospective survey were reported; the values for the effectiveness parameters were not reported; and the standard deviations used to derive probability distributions were not reported. It was not clear if the survey was the best available evidence; the authors stated that no head-to-head trials were available, but they did not report a comprehensive search for evidence. Due to these reporting limitations, it is not clear if the effectiveness estimates were appropriate, so the validity of the effectiveness outcomes is unclear.
Costs:
The costs indicated that a health service provider perspective was adopted. The cost categories were clearly stated and seem to have been appropriate for this perspective. Few details of the source used to derive costs were reported. Only the aggregate costs for each treatment were given, with no resource use and no individual unit costs. This reduces the transparency of the cost data. No discounting of those costs and benefits accrued beyond the first year appears to have been conducted.
Analysis and results:
The decision trees were clearly described, with a diagram. A full incremental analysis was not conducted – mean cost-effectiveness ratios were presented; these are less informative than incremental cost-effectiveness ratios. Due to a lack of data, the analysis had to use retrospective observational data, rather than randomised patient data. This increased the risk of selection bias. No attempt to control for any confounding variables appears to have been conducted, which further increased the risk of bias. Appropriate sensitivity analysis was conducted, but the distributions were not described. The cost and effectiveness estimates were from French sources, so care should be taken when attempting to generalise the results to other settings.
Concluding remarks:
The cost and effectiveness estimates were poorly reported, making the validity of the results unclear. No attempt was made to control for confounding variables, and the data were from limited observational evidence. The results should be used with caution.