Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.784757
Title: Addressing treatment switching in Health Technology Assessment
Author: Boucher, Rebecca H.
ISNI:       0000 0004 7970 3031
Awarding Body: University of Leicester
Current Institution: University of Leicester
Date of Award: 2019
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
Background: Sometimes individual patient level (IPD) must be reconstructed data from summary information when treatment switching has occurred (i.e. proportion of patients changed treatment arms during the course of a randomised control trial) to facilitate re-analysis when the IPD is unavailable. However, to re-analyse overall survival (OS), information is needed on the time to treatment switching; this can usually be approximated by time to progression (TTP). Therefore, the reconstructed data must include TTP and time to death for patients, estimated using an illness-death modelling framework. Methods: Here it is assumed only summary information of Progression-free survival (PFS) and OS are available. Using coordinates extracted from the Kaplan-Meier curves, the survival distributions are modelled. These are then combined with the PFS and OS risk tables, models for TTP, and estimates of the censoring distributions and post-progression survival (PPS) The data are then simulated and combined to obtain the underlying survival data. The correct proportion of treatment switchers is selected from those experiencing disease progression and the dataset analysed using a Rank Preserving Structural Failure Time Model (RPSFTM) to account for treatment switching. Multiple datasets are created from these models; each is analysed separately and the results averaged over to obtain a final point estimate. Results: The simulated data are, on average, broadly representative of the original IPD, both in terms of the reported summary statistics and the RPSFTM analysis. Conclusions: This application demonstrates the success with which this method can be used to reconstruct the data, and achieve an appropriate re-analysis for treatment switching, fulfilling a fundamental gap in the research.
Supervisor: Abrams, Keith ; Lambert, Paul Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.784757  DOI: Not available
Share: