OBJECTIVES: When appraising new healthcare technologies, the UK National Institute for Health and Care Excellence (NICE) considers the severity of the treated condition. Quantitatively, this is achieved using disease severity modifiers which adjust quality-adjusted life expectancy calculations for technologies that treat severe disease. The modifier is determined based on either an absolute or proportional quality-adjusted life year (QALY) shortfall, calculated using the lifetime difference in QALYs between patients with the condition and the general population. In the present study, a framework for rapidly estimating the QALY shortfall based on minimal study summary data was developed using a combination of published tools and methodologies.

METHODS: The framework was tested in the context of estimating a QALY shortfall using published Kaplan-Meier curves of overall survival (OS) and a published mean health state utility value (HSUV) for patients living with metastatic colorectal cancer. Kaplan-Meier curves were digitised and the Guyot et al. methodology employed to reconstruct the Kaplan-Meier data. Parametric survival models were fitted by maximum likelihood estimation and ranked based on goodness-of-fit criteria. Regularisation methods were then employed to overcome limitations of survival distributions with long tails/divergent integrals, and HSUVs (and discounting) were applied to cohort partitions derived from the survival models. Results were compared with McNamara et al. general population QALY estimates.

RESULTS: The framework facilitated the rapid estimation of QALY shortfalls in the UK based on published Kaplan-Meier curves and HSUVs.

CONCLUSIONS: Calculations supporting the use of a severity modifier must ultimately be as robust as those underpinning the health economic model submitted to NICE; however, in many circumstances, it is useful to rapidly understand the likelihood of a severity modifier being accepted by NICE based on early clinical data. The present framework takes an important step towards accelerating this process for health technologies whose cost-effectiveness can be evaluated using partitioned survival models.