Rapid Responses to:

RESEARCH: Bodil Hammer Bech, Carsten Obel, Tine Brink Henriksen, and Jørn Olsen Effect of reducing caffeine intake on birth weight and length of gestation: randomised controlled trial BMJ 2007; 334: 409 [Abstract] [Full text]

Rapid Responses: Submit a response to this article

Rapid Responses published:

Randomisation method

David J Torgerson, Catherine E Hewitt   (26 February 2007)

Randomisation method 26 February 2007
David J Torgerson, Director, York Trials Unit University of York YO10 5DD, Catherine E Hewitt Send response to journal: Re: Randomisation method	Editor The trial by Bech and colleagues describe, in the methods section, that they used 'balanced' randomisation using balanced blocks of 6. In the discussion, however, they state that the randomisation schedule was changed part way through the trial to favour the caffeinated group. As the authors state this will have no impact on the scientific validity of the trial, however, if the randomisation ratio is changed part way through a trial this needs to be taken into account in the analysis otherwise an arithmetical artefact will bias the results. In addition, the description of the randomisation process in the methods section is no longer correct. They clearly did not use "balanced blocks of six" they must have randomised the first x number of patients this way and then the remaining patients using some other form of blocked allocation. As a clear and accurate description of the randomisation process is crucial to assessing a trial's quality the authors need to explain exactly how this was undertaken. In addition they need to report an analysis corrected for change in the randomisation ratio. Catherine Hewitt David Torgerson Competing interests: None declared