Parametric Piecewise Models for Survival Extrapolations: A Validation Study

Author(s)

ABSTRACT WITHDRAWN

OBJECTIVES : Although standard parametric models (SPM) are most frequently applied for survival analyses, this method may not sufficiently capture sudden fluctuations in the underlying hazard function. Parametric piecewise models (PPWM) for survival analyses allow more flexibility and are also used in health technology assessment (HTA) submissions or by evidence research groups evaluating the submission. This study aims to assess the performance of parametric piecewise models when there is a sudden change in hazards observed.

METHODS : Immature data of three trials in melanoma, breast cancer (BC) and multiple myeloma (MM) were extrapolated and compared with mature data of the same trials using SPM and PPWM methods. The immature survival data indicated fluctuating hazards in all trials, especially in melanoma and BC. The exponential, Gompertz, loglogistic, lognormal and Weibull PPWMs were compared with SPMs. The best fitting models were selected based on LOOIC. The extrapolation on immature data was compared to the mature trial data in terms of ΔMean-Absolute-Deviation (ΔMAD) (lower ΔMAD implies a better prediction) in months for all arms in the trial.

RESULTS : For the melanoma dataset, where the RMST of the mature dataset was 67.88 months, ΔMAD was 23.73 and 15.40 months for the SPM and PPWMs, respectively. For the BC dataset (RMST=123.42 months) ΔMAD was 44.07 and 8.45 months, and for the MM dataset (RMST=92.55 months) ΔMAD was 7.21 and 7.70, respectively. The best fitting distribution on immature data did not always give the best predictions.

CONCLUSIONS : For two out of three trials, the PPWM fitted on the immature data had better predictions than the SPM when compared to the mature survival data. This can be explained by the observed fluctuating hazards in the underlying survival data in these trials. Based on this study, PPWMs should be considered for survival extrapolations, especially when there are sudden fluctuations in hazards.

Conference/Value in Health Info

2020-11, ISPOR Europe 2020, Milan, Italy

Value in Health, Volume 23, Issue S2 (December 2020)

Code

PCN290

Topic

Clinical Outcomes, Methodological & Statistical Research

Topic Subcategory

Clinical Outcomes Assessment, Missing Data

Disease

Multiple Diseases, No Specific Disease, Oncology

Explore Related HEOR by Topic


Your browser is out-of-date

ISPOR recommends that you update your browser for more security, speed and the best experience on ispor.org. Update my browser now

×