Interventions:
The authors justified their selection of the comparators: worldwide, timolol was the most commonly used first-line agent for the treatment of glaucoma; and latanoprost was the first available prostaglandin analogue.
Effectiveness/benefits:
: A systematic literature review was performed in order to obtain clinical data, but the methodology and conduct of this review were not reported. The basic characteristics of the primary data sources (the study population, design, follow-up and so on) were not given, although it appears from the references that the data on treatment effect were obtained from randomised controlled trials. The lack of explicit details on the sources for these data meant that an objective assessment of the validity of the clinical inputs was not possible. Similarly, little information on the derivation of the utility valuations was provided. QALYs are a validated benefit measure allowing cross-disease comparisons.
Costs:
The authors did not explicitly report the viewpoint of the economic analysis. However, the categories of costs suggested that the perspective of the health service payer was adopted. A breakdown of cost items was provided, but information on the resource consumption was not given, although resource use was likely to have reflected real-world patterns in each country. The authors did not describe the sources used to derive the unit costs. The price year was not explicitly reported. The use of discounting was not reported, although it could have been relevant as the long-term costs were calculated. In general, the economic analysis was not extensively described.
Analysis and results:
The synthesis of the costs and benefits was appropriately performed by means of an incremental analysis. The results of the base-case analysis were clearly presented. The issue of uncertainty was only partially addressed given that the sensitivity analysis considered variations in the key economic inputs individually. The authors noted that some assumptions were necessary in order to simplify the decision model, which may have reduced the accuracy of the simulation. The authors compared their results with those from other studies and highlighted the reasons for some differences.
Concluding remarks:
Overall, the study was based on a valid methodology, but was not extensively reported and the analysis of uncertainty was limited. Thus, the authors’ conclusions should be treated with some caution.