Work

Causal modelling in stratified and personalised health : developing methodology for analysis of primary care databases in stratified medicine

Year: 2016
Type: dissertation
Abstract: Predictive markers are variables associated with differential response to treatment, and can be useful in guiding treatment decisions.Predictive markers are those found to moderate the relationship between treatment and an outcome.Primarily the interest is in variables that modify the absolute, as opposed to the relative, effect of treatment which corresponds to treatment effect modification on the additive scale.Treatment effect modification is typically identified through statistical interaction terms in regression models.However, in a multiplicative regression model, a statistical interaction corresponds to effect modification on the multiplicative sale.This paper presents and discusses various ways in which one can evaluate treatment effect modification on the additive scale in data where multiplicative regression models are most commonly used, focussing particularly on applications in observational studies.We propose the use of a novel measure, the ratio of absolute effects, which can be calculated from the output of some multiplicative regression models, including the Cox proportional hazards model.The methods discussed are demonstrated in a practical application involving the identification of predictors of type II diabetes as a side-effect of oral steroid therapy in patients with rheumatoid arthritis.
Language: English
Open Access status: green
OpenAlex ID: W2933307853
Cites: 88
Cited by:
Related to: 9