Document details for 'Cox regression survival analysis with compositional covariates: application to modelling mortality risk from 24-hour physical activity patterns'

Authors McGregor, D.E., Palarea Albaladejo, J., Dall, P.M., Hron, K. and Chastin, S.F.M.
Publication details In "Book of Abstracts of the 8th International Workshop on Compositional Data Analysis (CoDaWork2019)", 49. Eds. M.I. Ortego. Universitat Politecnica de Catalunya-BarcelonaTECH, Barcelona, Spain.
Publisher details Universitat Politecnica de Catalunya-BarcelonaTECH, Barcelona, Spain
Abstract The Cox's proportional hazards model is a popular approach to survival analysis, particularly in medical and public health research. It is used in assessing the association between a time-to-event outcome, such as mortality or incidence of a disease, and exposure variables or interventions of interest. When the exposure includes compositional covariables; that is, variables representing relative parts of a whole such as nutritional intake or time distribution across physical activity behaviours, conventional analyses which consider the variables in isolation can be misleading. This is due to the intrinsic interplay between compositional variables and the violation of basic assumptions of the model and its associated significance tests. By formulating the Cox regression model in terms of log-ratio coordinates however, the constraints of compositional covariates are suitably addressed, common statistical inference methods can be applied, and scientifically meaningful interpretations are obtained. We show that standard statistical tests associated with Cox regression remain meaningful and are invariant to orthogonal rotations of the isometric log-ratio coordinate representation of the data. The practical application of our approach is demonstrated through the estimation of the mortality hazard associated with the composition of daily activity behaviour (physical activity, sitting time and sleep) using data from the U.S. National Health and Nutrition Examination Survey (NHANES).
ISBN 978-84-947240-1-5
Last updated 2019-08-06

Unless explicitly stated otherwise, all material is copyright © Biomathematics and Statistics Scotland.

Biomathematics and Statistics Scotland (BioSS) is formally part of The James Hutton Institute (JHI), a registered Scottish charity No. SC041796 and a company limited by guarantee No. SC374831. Registered Office: JHI, Invergowrie, Dundee, DD2 5DA, Scotland