OBJECTIVES:

There is limited guidance about accounting for treatment effect waning in pharmacoeconomic models for non-oncology products in the absence of long-term data. This study evaluated approaches to waning in recent submissions to the National Institute for Health Care Excellence (NICE). In addition, we constructed a simple health economic model to illustrate how this can affect cost-effectiveness results.

METHODS:

We reviewed 18 NICE technology appraisals for non-oncology products undertaken over the past five years involving Markov cohort models. We also developed a model in which a generic treatment for atopic dermatitis was evaluated against standard of care (SOC) from the perspective of the NHS over patients’ lifetimes. Patients were assumed to accumulate benefits (increased utility and lower direct medical expenditures) and costs of treatment until experiencing loss of response, which was represented in a variety of ways, including first-cycle waning, no waning, linear waning over pre-specified time intervals, and waning based on parametric fits to hypothetical trial data.

RESULTS:

Four of the 18 submissions allowed for explicit reductions in treatment efficacy, while an additional seven proxied using treatment discontinuation or gradual decline in quality of life. NICE/Evidence Review Group committee responses were variable, ranging from satisfaction with the omission of waning, to requests to explicitly model loss of response (e.g., when proxies were applied) or to remove waning from consideration when it had been included. Results of the simple model varied broadly depending upon the waning assumptions used. In comparison with a no waning scenario (ICER = £16,712 for new treatment vs. SOC) the varying assumptions considering treatment waning resulted in ICERs ranging from being dominated by SOC to £14,368.

CONCLUSIONS:

Assumptions regarding treatment waning may significantly impact cost-effectiveness analyses, but these are applied inconsistently. Guidance is required to advise on when and how waning should be incorporated in pharmacoeconomic models accompanying NICE submissions.