The rationale for the selection of the interventions was explicitly reported. The authors compared two branded treatments and excluded generic options.
The effectiveness data were obtained from randomised placebo-controlled trials. The basic characteristics of the primary sources (study population, design, and follow-up) were provided and the issue of heterogeneity among these sources was addressed. None of these trials compared the two treatments head-to-head, which was a limitation acknowledged by the authors. The utility values assigned to different health states were reported, but limited information on the derivation of these values was provided. QALWs, like quality-adjusted life-years, are a validated benefit measure allowing cross-disease comparisons to be made.
The costs appeared to reflect the perspective stated. An extensive breakdown of cost items was provided, but no information on resource consumption was given. The price year was not reported, which will hinder future reflation exercise. The uncertainty surrounding the cost estimates was evaluated by sensitivity analysis.
Analysis and results:
The synthesis of the costs and benefits was appropriately performed. The results of the base case and the sensitivity analysis were clearly reported using a tornado diagram and an incremental cost-effectiveness scatter plot. The authors acknowledged some of the potential limitations of their study, including how robust the utility estimates were and the nature of the primary studies from which the estimates of effectiveness were obtained.
The methodology appears to have been appropriate and, on the whole, was clearly and transparently reported. The conclusions reached by the authors appear to be appropriate.