Interventions:
The authors described the available treatments and the selection of entecavir and lamivudine was based on a recent head-to-head RCT, which compared the efficacy and safety of the two drugs. At the time of the study, no clinical trial directly comparing entecavir with other treatments was available. Dosages and length of treatment were reported.
Effectiveness/benefits:
The sources of data appear to have been appropriate. RCT data were combined with long-term data from the cohort study to simulate the management of patients over time. This approach is quite common in studies with a long time horizon. The authors provided some key details of the two primary sources of data, such as the study sample, follow-up, missing data, and location. Both studies had specific strengths, which made them valid sources of data. The utility valuations, used to calculate the QALYs, were derived from a sample of individuals from the general population using a validated approach. Due to the uncertainty underlying the estimation of the utilities, an alternative source of data was also considered in the sensitivity analysis. QALYs are a valid measure of benefit, not only because they capture the impact of disease on a patient's health, but also because they are generalisable across diseases.
Costs:
The categories of costs were relevant to the perspective. The drug costs were presented in detail, while those of complications were reported as macro-categories and were not broken down into individual items. This was due to the fact that these costs were derived from previous studies, the methodologies of which were not described. Other details of the analysis such as the price year and use of discounting were reported. The uncertainty surrounding the cost estimates was investigated in the deterministic sensitivity analysis.
Analysis and results:
The costs and benefits were clearly presented and the use of an incremental analysis was appropriate. The issue of uncertainty was satisfactorily addressed using both a deterministic and a probabilistic approach. The authors justified their selection of a specific modelling framework, which allowed a better simulation of the disease management. Some potential methodological limitations were pointed out, for example: the possible demographic and clinical differences between the patient samples in the two studies; the paucity of reliable long-term data on the rebound rates after treatment discontinuation; or the appropriateness of extrapolating Chinese data to the US population. The authors pointed out that their analysis referred to patients without co-infections and these findings should be restricted to this specific sub-group of HBV (mono-infected) patients. The authors noted that their results were conservative.
Concluding remarks:
In general, the analysis appears to have been carried out using valid methodology, which makes the authors’ conclusions more robust.