Interventions:
The rationale for the selection of the comparators was clear, as they were the treatments examined in the LEAD-1 trial, but the authors stated that rosiglitazone was no longer used in the authors setting due to its increased risk of congestive heart failure.
Effectiveness/benefits:
The treatment effect was appropriately based on a head-to-head randomised controlled trial of a large number of patients, which should have provided valid estimates. Typical sources and a validated model were used to extrapolate the short-term data to the long term. Extensive sensitivity analysis was conducted on all the model parameters. QALYs were a valid benefit measure, given the impact of diabetes and its complications on mortality and morbidity, and they allow comparisons with other disease areas. No information on the sources for the utility weights was provided.
Costs:
The cost items were consistent with the perspective of the health care payer. The unit costs were clearly presented for the drugs. Other medical costs were reported as category totals and were from publications, the methods of which were not reported. The discounting was appropriately reported as was the price year, allowing reflation exercises. No variation in the economic inputs was considered in the sensitivity analyses, but bootstrapping was applied to calculate standard deviations for the total costs.
Analysis and results:
The projected costs and benefits were clearly presented and were appropriately combined, using an incremental approach. The conventional threshold of $50,000 per QALY was used to identify the best strategy, but the authors pointed out that higher benchmarks could also be used. Both probabilistic and deterministic analyses were used to assess uncertainty, and the methods and results were clearly reported. The results were highly dependent on the inclusion or exclusion of the risk of congestive heart failure with rosiglitazone. The findings might be transferred to settings with similar relative drug prices.
Concluding remarks:
The analysis had a robust cost-effectiveness framework and the uncertainty was satisfactorily investigated, using a comprehensive approach. The authors’ conclusions appear to be valid.