Trial By Error: My Exchange With Archives of Disease in Childhood
7 MARCH 2018
By David Tuller, DrPH
On January 30th, Professor Racaniello e-mailed
a letter of concern to
Archives of Disease in Childhood about
a clinical trial of the Lightning Process in children with CFS/ME (as the study called the illness.) The letter, signed by 21 experts and academics, documented the trial’s questionable methodological choices and the investigators’ failure to disclose exactly what they had done.
In particular, the letter noted that the investigators extended a feasibility trial into a full trial while at the same time swapping primary and secondary outcome measures—a recipe for bias. As a result, they reported much more impressive-sounding results than would have been possible without the outcome measure swap.
Archives of Disease in Childhood is part of the BMJ group. The journal’s editor, Dr. Nick Brown, assured Professor Racaniello that he took the matter seriously and that it would be reviewed. Since then, another BMJ journal,
BMJ Open, has published
a paper that touches directly on the issue. The new study examined whether studies submitted to
The BMJ and not accepted because they were not properly prospectively registered were subsequently published elsewhere, among other things. The study concluded that “improperly registered trials are almost always published, suggesting that medical journal editors may not actively enforce registration requirements.”
Last week, I sent the following e-mail to Dr. Brown:
Dear Dr. Brown–
I’m writing a post about BMJ Open’s new study of improperly registered trials that have nonetheless gotten published. The study is germane to the issue of the Lightning Process study, given that more than half the participants were enrolled before the trial registration date and that the outcome measures were swapped based on those first results–meaning that this was not a properly registered prospective trial. The registration itself claims that it is a “prospective” trial. However, given the circumstances, this is clearly not the case–especially since the trial registration also mentions that the participants include those from the feasibility trial.
