The rationale for the choice of comparators was explicitly reported. The study was thorough in the coverage of available alternatives for the particular study population.
A systematic literature review was performed to obtain clinical data, and the methodology was reported. The basic characteristics of the primary data sources (the study population, design and so on) were not given, although it appeared from the references that the data were mainly obtained from two RCTs and one open-label non-randomised study. But the lack of explicit details on the sources for these data made it difficult to objectively assess the validity of the effectiveness data. The issue of heterogeneity among the sources of data was not addressed. The Delphi panel techniques were reported, but the use of such a method as a primary source of evidence was appropriately acknowledged by the authors as a limitation to their study. For example, one-month pregabalin effectiveness was determined by the Delphi panel, based on a study that reported three-month effectiveness. In addition, two different scales from different studies were combined to estimate outcomes. It seemed doubtful whether the method used could provide reliable estimates. Equally, little information on the derivation of the utility valuations was provided. The authors used a disease-specific benefit measure as well as QALYs, which represented a validated benefit measure that allowed cross-disease comparisons.
The costs included in the analysis reflected the perspective stated by the authors. A breakdown of costs items was provided, as well as information on the resource consumption. There may be a problem regarding the sources used to obtain resource use. For example, the number of plasters was based on US observational data, and it was unclear whether this level of resource use produced the same level of effectiveness assumed in the model. Secondly, the authors appropriately acknowledged that gabapentin efficacy might have been overestimated, as the dosage included in the model was less than that reported in clinical trials. The sources used to derive unit costs and the price year were reported. In the sensitivity analyses, where the impact of a longer time horizon was investigated, discounting was appropriate. In general, the economic analysis was well reported.
Analysis and results:
The synthesis of costs and benefits was appropriately performed using incremental analysis. The issue of uncertainty was extensively investigated using deterministic and probabilistic approach. The results of the base case and the sensitivity analysis were well reported. The results of the deterministic sensitivity analyses were only briefly discussed. Cost-effectiveness acceptability curves were generated. The authors compared their results with those from other studies and highlighted important similarities.
The methodology was characterised by several limitations, so the authors' conclusions should be treated with caution.