Kahan and Morris1 raised a fully justified concern about the methods employed for group allocation and the associated statistical analysis used in randomised controlled trials (RCT) published in four leading medical journals. The results were striking: for more than one-third (37%) of the RCTs analysed the method of randomisation was unclearly reported, and only 26% of the RCTs that used balancing factors for randomisation correctly adjusted their primary analysis for these factors. As a result, some of the trials may have (wrongly) reported no effect of the treatment due to a decrease in power (reduction in type I error rate) induced by the lack of a correct adjustment, as indicated by the authors1.

However, this paper has some limitations that should be taken into account: no scientific rationale was given for the choice of the journals analysed (BMJ, Journal of the American Medical Association, Lancet, and New England Journal of Medicine) and the year of article’s publication (2010). Moreover, whilst comparisons with previous reviews made in 1997 and 2002 provided an indication on the overtime variation regarding the quality of the reporting of the method of randomisation in RCTs, this variation was probably affected by subjectivity in data extraction (even if standardised forms were used), since reviews were made by different researchers. Therefore, concluding that “while reporting of the randomisation method may have improved between 1997 and 2002 there has been little improvement subsequently” seems to be precipitated. Furthermore, as a reader of all journals included in the review, I would like to know if there were differences in the quality of the reporting of the method of randomisation according to the journal where RCTs were published.

With this review and reanalysis study, Kahan and Morris evidenced large discrepancies in reporting randomisation procedures among RCTs, which renders the “work” of the readers (especially when they are not experts in statistics) difficulty to be made. Besides the suggestion of the authors that reports of RCTs should “explicitly state when randomisation was not stratified (…) and when an unadjusted analysis was done”, in my opinion, journals’ policy must systematically demand to authors of RCTs a complete description of randomisation methods to be downloaded as a supplementary material (on-line only). This would provide clarity for the reporting and empower readers by improving their ability to judge articles’ quality by themselves.

References
1. Kahan BC, Morris TP. Reporting and analysis of trials using stratified randomisation in leading medical journals: review and reanalysis. BMJ 2012;345:e5840.

e-mail: philipebarreto81@yahoo.com.br

Competing interests: None declared