Skip to content



Open Access

Efficiency gains resulting from the ordinal analysis of a functional outcome scale: a case study of a major phase III stroke trial

  • Gordon D Murray1,
  • Else Charlotte Sandset2,
  • Philip MW Bath3 and
  • Eivind Berge2
Trials201112(Suppl 1):A67

Published: 13 December 2011


Traumatic Brain InjuryAcute StrokePrimary AnalysisMajor PhaseEfficiency Gain


Phase III clinical trials in areas including acute stroke and traumatic brain injury commonly use ordinal functional outcome scales as their primary outcome measure. Conventionally these scales are analysed by dichotomising the ordinal scale into a binary scale:- ‘dead or dependent’ versus ‘independent’. This can potentially discard much relevant information, reducing both the clinical relevance of the results and the statistical efficiency of the analysis.


Methodological work in stroke (by the OAST Group) and in traumatic brain injury (by the IMPACT Investigators) has demonstrated that using more appropriate approaches to the analysis of ordinal outcome scales, such as proportional odds regression or the ‘sliding dichotomy’, can potentially lead to substantial efficiency gains relative to the conventional dichotomous analysis. However, to date relatively few trials have prospectively adopted ordinal techniques for their primary analysis. We report here how in SCAST [1], a major Phase III trial of blood pressure reduction in acute stroke, ordinal methods were adopted for the primary analysis of the modified Rankin Scale (mRS), an ordinal functional outcome scale.


Since ordinal methodology was evolving in parallel with the conduct of the trial, the Statistical Analysis Plan was not finalised until close to database lock. It was decided to use proportional odds regression for the primary analysis of the mRS with the sliding dichotomy as a sensitivity analysis. Relative to a conventional dichotomous analysis both of these approaches did indeed lead to substantial efficiency gains, equivalent to more than doubling the sample size.


SCAST shows that the potential efficiency gains demonstrated in basic methodological research can be realised in practice. This has major implications for the design and analysis of future trials based on ordinal outcome scales.



This work is presented on behalf of the SCAST Study Group.

Authors’ Affiliations

Edinburgh MRC Hub for Trials Methodology Research, University of Edinburgh, Edinburgh, UK
Department of Internal Medicine, Oslo University Hospital Ullevål, Oslo, Norway
Stroke Trials Unit, University of Nottingham, Nottingham, UK


  1. Sandset EC, Bath PMW, Boysen G, Jatuzis D, Korv J, Luders S, Murray GD, Richter PS, Roine RO, Terent A, Thijs V, Berge E, on behalf of the SCAST Study Group: The angiotensin-receptor blocker candesartan for treatment of acute stroke (SCAST): a randomised, placebo-controlled, double-blind trial. Lancet. 2011, 377: 741-750. 10.1016/S0140-6736(11)60104-9.View ArticlePubMedGoogle Scholar


© Murray et al; licensee BioMed Central Ltd. 2011

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 (, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.