- Poster presentation
- Open Access
Estimating the effect of “treatment in the treated” - instrumental variable analysis vs conventional regression methods in the titre-2 trial in cardiac surgery
© Rogers et al. 2015
- Published: 16 November 2015
- Instrumental Variable
- Observational Analysis
- Ischaemic Event
- Liberal Group
- Conventional Regression
Perioperative anaemia is associated with adverse outcomes after cardiac surgery but, paradoxically, observational analyses have shown that red cell transfusion is associated with worse clinical outcomes. TITRe2 tested the hypothesis that a restrictive threshold for transfusion would reduce post-operative morbidity compared to a liberal threshold.
Adults undergoing cardiac surgery with post-operative haemoglobin <9g/dL were recruited. Participants were randomised to transfusion if haemoglobin <7.5g/dL (restrictive) or <9g/dL (liberal). The primary analyses were by intention-to-treat. A secondary analysis of a composite outcome (serious infection or ischaemic event or death in the 3-months after randomisation) to assess the effect of receiving a transfusion was pre-specified. Two methods for handling confounding were applied: adjustment conventionally for covariates (CA) or using randomised allocation as an instrumental variable (IV).
2003 patients were randomised (1000 restrictive group, 1003 liberal group. Transfusion rates were 53.4% and 92.2% in the restrictive and liberal groups, respectively. The primary intention-to-treat analysis suggested a similar outcome in the two groups (odds ratio=1.11, 95%CI 0.91-1.34, p=0.30). In the CA analysis the odds of morbidity/mortality increased with transfusion (odds ratio=1.28 95%CI 1.03 to 1.60, p=0.028), but the IV analysis was in the opposite direction (relative risk=0.78, 95%CI 0.53-1.14, p=0.20).
CA analysis supports previous observational analyses and contradicting the primary analysis. IV analysis suggested a marginally protective effect of RBC transfusion consistent with the ITT analysis of the RCT. We conclude that the CA results are explained by residual confounding.
Acknowledgement and disclaimer
This project was funded by the National Institute for Health Research Health Technology Assessment Programme (project number 06/402/94). The views and opinions expressed therein are those of the authors and do not necessarily reflect those of the HTA, NIHR, NHS or the Department of Health.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.