News

FDA disagrees with BMJ study that found clinical trials were not being reported

BMJ 2012; 344 doi: http://dx.doi.org/10.1136/bmj.e3277 (Published 08 May 2012) Cite this as: BMJ 2012;344:e3277
  1. Nigel Hawkes
  1. 1London

The US Food and Drug Administration has denied charges made in a study in the BMJ that rules making it mandatory to report clinical trial results are being flouted. The study, published in January by Andrew Prayle and colleagues at the University of Nottingham,1 calculated that only 22% of the trials that should have reported their results on the ClinicalTrials.gov website had actually done so.

In response to the study Henry Waxman, the Democrat Congressman for part of Los Angeles, wrote to the FDA demanding an explanation. Jeanne Ireland, the FDA’s assistant commissioner for legislation, replied, contesting the figures.

She said that the analysis included some trials that were completed before the law took effect and had not excluded all those, such as uncontrolled trials, that are exempt from the reporting requirements. The list also included some trials of products yet to be approved, which are exempt from the law, and included some for which data had been submitted but not yet vetted by the FDA. In addition, she said, the BMJ authors could not tell from publicly accessible data whether the deadline for reporting had been extended, as the law allows in some cases. She wrote that the FDA was pursuing “individual responsible parties in 15 cases for failure to report on time.”

The FDA’s response to Waxman has not been published but was obtained by Nature under freedom of information legislation.2 The BMJ was told by the FDA that it would have to submit a separate freedom of information request of its own to see the letter.

Prayle said that the Nottingham team stands by its findings as the best possible job that could have been done with publicly available information.

“We cross referenced two publically available datasets in order to try to work out which studies should have reported results and whether they had in fact done so” he told the BMJ. “We tried with the data available to exclude trials which were not required to report results. It is not surprising that they may differ from the absolute number of trials which should have reported but did not, as they have access to confidential data which we do not.

“Additionally, trials can request an extension, and we have no way of knowing if this has occurred. Furthermore, we can’t tell if trials had reported results but those results had not yet made it through quality control and on to the ClinicalTrials.gov website. All of these could change the denominator.”

The US National Institutes of Health (NIH), jointly responsible with the FDA for enforcing the law, produced an unofficial analysis for Nature of trials that have reported, agreeing with the BMJ analysis that the industry is doing better than government or academia in filing results on time. Industry funded trials were reported on time in 52% of cases, against 21% of those funded by NIH and 14% for NIH funded academic sponsors, the NIH found. The BMJ study found that 40% of industry funded studies reported on time, against 9% of those not solely industry funded.

“I think that although we may argue about the specific percentages, it appears clear that the majority of trials didn’t report on time, even with the unofficial NIH analysis,” said Prayle. “The conclusion to our BMJ paper was that ‘most trials subject to mandatory reporting did not report results within a year of completion.’ In my opinion the NIH analysis does not change this conclusion.”

Waxman’s office said that the Congressional Energy and Commerce Committee, which he chairs, was “continuing to evaluate” the NIH and FDA responses.

Prayle said that the motivation for writing the paper was to encourage greater reporting of trial results on ClinicalTrials.gov. “It would be great if all trials, not just those who are required to, reported results, as this would massively increase access to research results of unpublished studies,” he said.

Notes

Cite this as: BMJ 2012;344:e3277

References