Practical methods to pool multi-study joint longitudinal and time to event data

Sudell, ME ORCID: 0000-0002-7919-4981, Kolamunnage-Dona, Ruwanthi ORCID: 0000-0003-3886-6208 and Tudur Smith, Catrin ORCID: 0000-0003-3051-1445 (2017) Practical methods to pool multi-study joint longitudinal and time to event data. In: ISCB 2017, 2017-7-9 - 2017-7-13, Vigo.

Text ISCB2017 - Maria Sudell.pdf - Unspecified Download (2MB)

Abstract

Background: Joint longitudinal and time-to-event data models have been established in a single study case as beneficial compared to separate longitudinal or time-to-event analyses in a range of cases, including data with study dropout, time-to-event models with longitudinal covariates measured with error, or cases when the relationship between longitudinal and time-to-event outcomes is of interest. However the methodology available for multi-study cases such as meta-analyses is limited. Aims: To investigate different approaches of modelling of multi-study joint longitudinal and time-to-event outcome data. Methods: Several methods are examined to account for between study heterogeneity, including as one stage methods that can include random effects at the study level, stratification of baseline hazard by study and use of fixed study indicator terms and their interactions with treatment assignment, or approaches for two stage pooling of joint model fits. These methods are applied to a real data example and further investigated in a simulation study. Software have been developed in R to allow these methods to be easily applied in future investigations, which will be available in a package alongside joineR collaboration. Results: The results from the real data example and simulation study will be presented at conference.

Item Type:	Conference or Workshop Item (Unspecified)
Depositing User:	Symplectic Admin
Date Deposited:	18 Dec 2017 10:18
Last Modified:	19 Jan 2023 06:46
URI:	https://livrepository.liverpool.ac.uk/id/eprint/3014436