OBJECTIVES: Clinical research in oncology is progressively moving towards earlier treatment lines and curative settings. In early-stage (neoadjuvant, peri-operative and adjuvant) treatment settings the potential for patients to be ‘cured’ can create challenges in extrapolating survival outcomes in cost-effectiveness analyses. The study objective was to review how the modelling of cure was implemented across NICE technology appraisals (TAs) for pharmacological treatments in early-stage oncology settings.
METHODS: The NICE website was searched to identify relevant TAs published up to the end of May 2022. Information about the approaches used to model cure and critique was extracted from documents available on the NICE website.
RESULTS: The searches yielded 10 adjuvant TAs and 1 neoadjuvant TA in different oncology indications (4 breast, 3 melanoma, 3 gastrointestinal and 1 lung). Approaches used to model cure included: switching from standard parametric survival models to background mortality at a timepoint when patients are clinically considered cured (4 TAs); using both external registry data and background mortality over different time periods (3 TAs); and fitting ‘mixture-cure’ models that inherently separate cured and uncured patients (1 TA). Different assumptions were made about the timing of cure and proportion of cured patients, with 3 TAs modelling a time-varying cured proportion. Cure was not explicitly modelled in 3 TAs but extrapolated survival rates were bound by background mortality. Approaches used to model cure were generally accepted by the Evidence Review Groups but additional scenario analyses altering key parameters (including cure fraction and timepoint) were explored, with more pessimistic assumptions often preferred.
CONCLUSIONS: This review provides insights into the assumptions and critiques of approaches used to model cure within economic evaluations submitted to NICE in early-stage oncology. The results of this study may be used to inform the cure assumption methodology for future TAs.
