Exploring model uncertainty in economic evaluation of health interventions: the example of rotavirus vaccination in Vietnam.

OBJECTIVE

Motivated by observed discrepancies between 2 published studies on the cost-effectiveness of rotavirus vaccination in Vietnam, the authors' objectives were to illustrate a specific, systematic approach to assessing model (structure and process) uncertainty and to quantify explicitly the contributions of different sources of variation in the outputs of different studies that share the same research question.

METHODS

On the basis of a series of working definitions of key model elements, the authors developed 5 alternative computer simulation (state-transition) models of rotavirus disease. They examined how epidemiological outcomes and cost-effectiveness ratios associated with rotavirus vaccination would change as elements of model structure and modeling process were progressively modified. They also explicitly decomposed the relative contributions of different modeling elements to differences in the cost-effectiveness results between the 2 previous analyses motivating the present study.

RESULTS

The findings suggest that within the category of a static, deterministic, aggregate-level model, different choices in model structure and process lead to relatively modest differences in the estimated cost-effectiveness of rotavirus vaccination, but that intermediate epidemiologic outcomes vary more substantially depending on the choice of model structure.

CONCLUSIONS

The authors caution against generalizing the quantitative results in this study beyond the present example but suggest that the approach presented here may serve as a template for other examinations of model uncertainty. As new research questions arise after the introduction of rotavirus vaccination programs, a reevaluation of model uncertainty is likely to be needed.