OBJECTIVES: This study aimed to analyze how frequently TLV’s and NOMA’s base case analyses reflect the manufacturers' choice of using flexible survival models to extrapolate survival outcomes.

METHODS: The study used a bespoke database (NMAi) containing all the drug assessments published on TLV´s and NOMA´s websites. This database allows for the search of keywords within all the published assessments. NMAi was used to systematically identify assessments where the manufacturer had used flexible models via advanced keyword searches (combination of keywords and filtering). The identified assessments were then analyzed to determine if TLV and NOMA accepted the use of the flexible models in their base case analyses.

RESULTS: The search and following analyses identified 25 and 37 assessments using flexible survival models in Sweden and Norway, respectively. TLV used flexible models in their base case in 18 out of the 25 assessments. Out of the 18 assessments, eight used cure models, seven used spline models, two used piecewise models, and one used a landmark model. In six assessments, TLV did not present a base case ICER due to excessive uncertainties, and in one assessment, TLV opted for a standard parametric model in their base case. NOMA used flexible models in their base case in 28 out of the 37 identified assessments. Out of the 28 assessments, 12 used piecewise models, 11 used spline models, and five used cure models. NOMA opted for standard parametric models in the remaining nine assessments in their base case. More than 50% of the assessments were concerning monoclonal antibodies. Additionally, all but one of the assessments were in oncology.

CONCLUSIONS: TLV and NOMA accepted the use of flexible survival models in most of the identified assessments. The analysis identified a trend towards using flexible models in the field of immunotherapy.