OBJECTIVES: NICE recently introduced a ‘severity modifier’, which alters the decision-making threshold based on the severity of a condition. This is based on the absolute and proportional quality-adjusted life-year (QALY) shortfall with standard-of-care relative to the general population. If considered, either a 1.7x or 1.2x weighting is applicable to the incremental QALYs of an intervention. There is a large difference in the QALY shortfall required to meet each threshold, and therefore some disease areas may fall just under the 1.7x weighting but only qualify for the 1.2x which may impact decisions on cost-effectiveness in England. This research looks at a framework using linear interpolation between the two weightings based on PSA iterations.

METHODS: A partitioned-survival model was produced using digitized overall survival and progression-free survival Kaplan-Meier curves from the EMILIA study (NCT00829166) of trastuzumab-emtansine (T-DM1) in 2L metastatic breast cancer. Pseudo patient-level data was created and parametric curves fitted to extrapolate outcomes. Age from the EMILIA study was used with health-state utility values from literature to estimate the QALY shortfall(s). The QALY weighting was calculated for 1,000 PSA iterations between 1.2x and 1.7x in increments of 0.1.

RESULTS: With a deterministic result of 2.21 QALYs, a 1.2x severity modifier weighting may be considered appropriate by NICE based on current thresholds. Analysis of the PSA indicated that in 70% of cases, the QALYs gained for T-DM1 would meet a 1.3x weighting with the remaining 30% meeting a 1.4x weighting.

CONCLUSIONS: This analysis presents an alternative novel approach to estimating severity weighting by combining PSA with an increased number of severity weights. This case study highlights that, in some instances, the current methodology may lack sensitivity to adequately capture the severity of a condition which lies between the two weights. A sliding scale approach may offer a solution to quantifying disease severity more accurately.