OBJECTIVES: To assess the implications of using vignette studies to estimate health state preferences on the evaluation of new drugs by the Swedish (TLV) and the Norwegian (NoMA) agencies.

METHODS: The published TLV and NoMA assessments over the past five years were reviewed to identify instances where vignette-based utilities were used in cost-effectiveness analysis (CEA). Information on the drugs, diseases, and evaluations pertaining to health-related utilities were collected in the agencies assessments

RESULTS: Vignette studies were employed in the assessment of five drugs in Sweden and three drugs in Norway within the last five years. These drugs were classified as orphan drugs by the European Medicines Agency (EMA) and were indicated for children with rare diseases of varying severity. The TLV expressed considerable uncertainty about the use of vignette-based methods, casting doubt on the utilities used in the CEA, influencing the final decision. Conversely, while NoMA assessments acknowledged the uncertainty associated with vignettes, it had limited impact on the overall evaluation of cost-effectiveness.

CONCLUSIONS: Vignette studies serve as an alternative method for estimating health state preferences when direct patient measurement is not feasible. A vignette represents a description of a specific health state, valued through preference elicitation, to obtain a utility estimate. These estimates, not directly obtained from patients with the relevant condition, introduce significant uncertainty into CEA, thereby affecting the decision-making process. While the TLV lacks clear guidance on the use of vignette methods for drug submissions, Norwegian guidelines offer a more open stance toward utilizing vignette studies in drug reimbursement assessments. Given the challenges associated with severe and rare diseases and the scarcity of reliable data on health state utilities, it would be beneficial for the TLV to provide explicit recommendations on incorporating vignettes into CEAs when traditional data sources are insufficient