University of Leicester
journal.pone.0161864.PDF (715.21 kB)
Download file

Exploratory Network Meta Regression Analysis of Stroke Prevention in Atrial Fibrillation Fails to Identify Any Interactions with Treatment Effect.

Download (715.21 kB)
journal contribution
posted on 2016-11-11, 11:59 authored by S. Batson, Alex Sutton, Keith Abrams
BACKGROUND: Patients with atrial fibrillation are at a greater risk of stroke and therefore the main goal for treatment of patients with atrial fibrillation is to prevent stroke from occurring. There are a number of different stroke prevention treatments available to include warfarin and novel oral anticoagulants. Previous network meta-analyses of novel oral anticoagulants for stroke prevention in atrial fibrillation acknowledge the limitation of heterogeneity across the included trials but have not explored the impact of potentially important treatment modifying covariates. OBJECTIVES: To explore potentially important treatment modifying covariates using network meta-regression analyses for stroke prevention in atrial fibrillation. METHODS: We performed a network meta-analysis for the outcome of ischaemic stroke and conducted an exploratory regression analysis considering potentially important treatment modifying covariates. These covariates included the proportion of patients with a previous stroke, proportion of males, mean age, the duration of study follow-up and the patients underlying risk of ischaemic stroke. RESULTS: None of the covariates explored impacted relative treatment effects relative to placebo. Notably, the exploration of 'study follow-up' as a covariate supported the assumption that difference in trial durations is unimportant in this indication despite the variation across trials in the network. CONCLUSION: This study is limited by the quantity of data available. Further investigation is warranted, and, as justifying further trials may be difficult, it would be desirable to obtain individual patient level data (IPD) to facilitate an effort to relate treatment effects to IPD covariates in order to investigate heterogeneity. Observational data could also be examined to establish if there are potential trends elsewhere. The approach and methods presented have potentially wide applications within any indication as to highlight the potential benefit of extending decision problems to include additional comparators outside of those of primary interest to allow for the exploration of heterogeneity.


This study was funded by DRG Abacus. DRG Abacus provided support in the form of salary for author SB, but did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. The specific roles of these authors are articulated in the 'author contributions' section.



PLoS ONE 11 (8): e0161864. 2016

Author affiliation

/Organisation/COLLEGE OF MEDICINE, BIOLOGICAL SCIENCES AND PSYCHOLOGY/School of Medicine/Department of Health Sciences


  • VoR (Version of Record)

Published in

PLoS ONE 11 (8): e0161864. 2016


Public Library of Science



Acceptance date


Copyright date


Available date


Publisher version


S1 File. Supporting Information. doi:10.1371/journal.pone.0161864.s001 (DOCX)



Usage metrics

    University of Leicester Publications