The rationale for the selection of the comparators was clear since the current pattern of care (no vaccination) was compared with the proposed health technology. As the Helicobacter pylori
vaccine was a potential technology, no vaccination was the appropriate comparator in all settings.
Selecting the sources of clinical evidence can be a valid approach as long as they reflect the epidemiological setting of the study. In this study, Dutch databases and country-specific studies were used to derive the clinical inputs, and they should, therefore, be valid for the scope of the analysis. The key input was the vaccine efficacy, which was based on manufacturers’ recommendations and, as this was uncertain, it was tested in a specific sensitivity analysis. The use of only Dutch data limits the impact of potential differences from mixed sources. The benefit measure was appropriate for capturing the effects of the disease on the most relevant dimension of health, which was survival. Two discount rates were applied, as was appropriate for the objective of the analysis.
The economic analysis focused on the setting and the health care system, but this was not explicitly reported. The economic analysis was not presented in detail and the key cost category, which was the disease treatment costs, was not separated into individual items. These costs were from a previous study and the methods of this were not reported. Other aspects of the analysis, such as the price year, use of discounting, and statistical analyses, were reported and consistent with Dutch guidelines.
Analysis and results:
The expected costs and benefits were clearly presented and the incremental approach was appropriate for identifying the optimal strategy. The issue of uncertainty was satisfactorily investigated and was well discussed. The authors acknowledged some limitations of their analysis, which mainly related to the need for many assumptions, since the vaccine was not registered and clinical data were not available.
: The methods were valid and the authors’ conclusions highlighted the impact of model assumptions on the study results.