The authors justified their selection of treatment sequences. A new British recommendation on primary open-angle glaucoma treatment was to offer patients a prostaglandin analogue, making monotherapy the first-line treatment, followed by fixed combination therapy of the prostaglandin analogue with timolol, after initial treatment failure.
The authors justified their use of clinical data from a physicians’ database, which offered real-world estimates of treatment efficacy, which should be useful for making health decisions. This administrative source required extensive extrapolation of data for the cost-effectiveness analysis. The authors acknowledged that a clinical trial would have provided higher internal validity, but stated would have been less representative of the real clinical situation. They performed statistical analyses to account for potential confounders and differences in the length of follow-up between treatment groups. These groups were well balanced at baseline in their clinical and demographic features, but the size of the groups was not balanced and power calculations were not carried out to ensure the significance of the comparison. The endpoints of the analysis were the direct outcomes of the glaucoma treatments and these cannot be compared with other disease outcomes.
The categories of costs reflected the perspective of the UK NHS as stated by the authors. The resource use was from the UK-GPRD, which should ensure that the data were detailed and representative of real life. The unit costs were from appropriate standard UK sources, but they were not presented separately from the resource quantities. No sensitivity analysis of the costs was conducted. These issues limit the ability to transfer and reproduce the analysis. The total costs were presented for subgroups of items that allowed the consideration of the relative impact of each category of cost. No discounting was applied and the costs might have occurred in different time periods over the three years.
Analysis and results:
The results were clearly presented. A synthesis of the costs and benefits was not performed as the analysis had a cost-consequences framework. The uncertainty was not investigated as the authors used real-world data that reflected the actual clinical and economic impact of treatments. Statistical analyses were carried out to consider the impact of confounders on both the costs and efficacy data. The generalisability of the results was not discussed and the findings should be considered to be specific to the UK.
The analysis had a simple framework that reported the real-world economic and clinical impact of the three treatments. The authors’ conclusions appear to be valid.