Quantifying the Impact of Inconsistencies in Economic Modeling Assumptions: A Case Study in Metastatic Colorectal Cancer
Author(s)
Rachael Batteson, MSc1, Jaesh Naik, MSc1, Hannah Gillies, MSc2.
1Petauri Evidence, Nottingham, United Kingdom, 2Petauri Evidence, Bicester, United Kingdom.
1Petauri Evidence, Nottingham, United Kingdom, 2Petauri Evidence, Bicester, United Kingdom.
OBJECTIVES: As health economic modelling methods have developed over time, clear guidance (e.g. Technical Support Documents) has been published for specific topics; however, some structural assumptions are seen as lower priority and often not covered within the guidance.
To quantity the impact of methods discrepancies, we tested alternative modelling approaches for various settings using an oncology case study.
METHODS: A case study was identified by searching the National Institute for Health and Care Excellence website. TA405 (metastatic colorectal cancer) was selected based on the transparency of reporting, which allowed the model to be reconstructed. A three-state partitioned survival model was recreated using pseudo patient-level data, and consistent inputs with TA405 where possible.
The following alternative modelling approaches were tested: continuous versus annual discounting; dynamic versus constant general population sex distribution; adverse event (AE) per-cycle versus one-off implications; half-cycle correction (HCC) applied in the first and last cycles versus per cycle; normal versus gamma distribution for costs in sensitivity analysis; and ±10% of the mean versus an assumed distribution for one-way sensitivity analysis.
RESULTS: The assumption with the largest impact was applying AE costs per cycle, which resulted in a 2% decrease in the incremental cost-effectiveness ratio (ICER) compared with one-off costs. Applying HCC in the first and last cycles only had a large impact on the treatment costs in each arm (19% and 34% decrease for intervention and comparator), however the incremental impact was small. Alternative approaches for discounting and the general population sex distribution did not have a large impact on results (<1% change).
CONCLUSIONS: Overall, although changing assumptions appeared to have a small impact on the ICER, such changes could collectively have an impact on decision-making, especially when cost-effectiveness thresholds are applied. As such, clear methods guidance would be beneficial to improve consistency between evaluations.
To quantity the impact of methods discrepancies, we tested alternative modelling approaches for various settings using an oncology case study.
METHODS: A case study was identified by searching the National Institute for Health and Care Excellence website. TA405 (metastatic colorectal cancer) was selected based on the transparency of reporting, which allowed the model to be reconstructed. A three-state partitioned survival model was recreated using pseudo patient-level data, and consistent inputs with TA405 where possible.
The following alternative modelling approaches were tested: continuous versus annual discounting; dynamic versus constant general population sex distribution; adverse event (AE) per-cycle versus one-off implications; half-cycle correction (HCC) applied in the first and last cycles versus per cycle; normal versus gamma distribution for costs in sensitivity analysis; and ±10% of the mean versus an assumed distribution for one-way sensitivity analysis.
RESULTS: The assumption with the largest impact was applying AE costs per cycle, which resulted in a 2% decrease in the incremental cost-effectiveness ratio (ICER) compared with one-off costs. Applying HCC in the first and last cycles only had a large impact on the treatment costs in each arm (19% and 34% decrease for intervention and comparator), however the incremental impact was small. Alternative approaches for discounting and the general population sex distribution did not have a large impact on results (<1% change).
CONCLUSIONS: Overall, although changing assumptions appeared to have a small impact on the ICER, such changes could collectively have an impact on decision-making, especially when cost-effectiveness thresholds are applied. As such, clear methods guidance would be beneficial to improve consistency between evaluations.
Conference/Value in Health Info
2025-11, ISPOR Europe 2025, Glasgow, Scotland
Value in Health, Volume 28, Issue S2
Code
EE629
Topic
Economic Evaluation, Health Technology Assessment, Methodological & Statistical Research
Disease
Oncology