OBJECTIVES:

Immature overall survival (OS) data is a key challenge in assessing the cost-effectiveness of new oncology drugs. Leveraging external data to inform OS extrapolation has become an area of research. This review surveyed data sources and methods used to incorporate external data in OS extrapolations in National Institute for Health and Care Excellence (NICE) oncology technology appraisals.

METHODS:

Among the 84 partitioned survival models used in NICE oncology technology appraisals published between December 2019 and December 2022, we identified 9 cases in which external data were incorporated to extrapolate the investigational drug’s OS. For these cases, information on external data sources, extrapolation methods, and Committee considerations were extracted.

RESULTS:

In most cases (8/9), extrapolation leveraging external data was used in companies’ base-cases, whereas it was initiated upon Committee’s request in one appraisal. Uncertainty due to extrapolation from the registrational trial with immature OS (3/8), and better reflection of clinical practice (2/8), were listed by companies as key reasons for incorporating external data. Data from other clinical trials (7/9) were more frequently used than real-world data (2/9). Data reconstruction from survival curves was used frequently (6/9). Methods to incorporate external data included: transfer of OS from external data sources (2/9), pooling external and registrational trial data using weighting/matching methods (2/9), and applying relative treatment effects to reference survival curves obtained primarily from external data (2/9). Committees generally accepted external data incorporation to inform OS extrapolation (7/9). Main reasons cited for non-acceptance were poor visual fit between extrapolated and observed registrational trial OS, and perceived sufficiency of the registrational OS data alone.

CONCLUSIONS:

External data to inform OS extrapolation is rarely used in NICE oncology appraisals. Companies mainly employed data from other trials using a variety of methods. Committees appear to recognize benefits of the additional evidence when registrational trial follow-up time is limited.