The US requirement to deposit trial data within a year is unworkable

A provision of the US Food and Drug Administration Amendments Act of 2007 (section 801) requires that sponsors or investigators summarise results on ClinicalTrials.gov for drug and device trials within one year of the date of completion.1 The “date of completion” is defined as “the date that the final subject was examined or received an intervention for the purposes of final collection of data for the primary outcome, whether the clinical trial concluded according to the prespecified protocol or was terminated.”

The act requires results to be summarised in tabular format, without discussion or conclusion. Sponsors or investigators may be granted extensions to the deadline if they can give reasons and the dates when they will provide results. Sponsors or investigators who fail to comply are subject to substantial fines.

The effect of the amendment is to force investigators to make data available before they have been finalised, analysed, and published. Surprisingly, the amendment does not include pending publication as a valid reason for extension of the deadline.

Deposit before publication exposes investigators to attack if data deposited do not match with what is later published. You need only review the troubles that VIGOR (VIoxx Gastrointestinal Outcomes Research) investigators had dealing with “expressions of concern” from editors of the New England Journal of Medicine to know the problems that befall investigators when there are discrepancies.2 3 4

A shortcoming of the amendment is the time limit for compliance. The one year time limit is unrealistic and sets researchers up to fail. It has the effect of eroding public trust in researchers and provides journalists and pundits with the opportunity to label results as withheld or delayed, implying lethargy on the part of investigators in making results available.5 6

It usually takes months to complete data harvests after the close of data collection and several more months before data are edited, cleaned, and ready for assembly into a dataset suitable for analysis. It was about a year after the decision to stop use of tolbutamide in the University Group Diabetes Program, for example, before a final count of deaths was possible because of the efforts needed after termination of the study to determine vital status for people who had dropped out.7 8

It is hard to see how the public benefits from forcing deposits before data have been properly assembled, edited, checked, and analysed. Ross and colleagues found the median time to publication from completion for 1336 clinical trials published in 2009 was 21 months, with the 25th centile being 13 months and the 75th centile being 32 months.9

The focus should be on publication in peer reviewed, indexed medical journals. The FDA amendment shifts the focus to a shoddy alternative driven by a contrived timetable. It forces investigators to devote scarce resources to an activity that siphons away energies that would be better spent on publishing. Beyond that it has the potential to lull sponsors and investigators into believing that publication is not necessary because they have already deposited the data.

Section 801 of the FDA amendment should be revised to:

1. Allow publication to substitute for deposits on ClinicalTrials.gov.

2. Extend the time limit for deposit of data from one year to three years to allow time for publication.

3. Require sponsors or investigators to deposit data at the end of three years after completion of the trial if results have not been published; require explanation as to why results were not published; and require submission history and reasons for rejection if results have been submitted for publication but were rejected. Require the same for studies terminated because of safety or efficacy.

Allowing publications to serve as surrogates for deposit removes the disconnect between deposit and publication and eliminates the problems when sponsors make deposits independent of investigators merely to meet the deadline.

As it stands, the provision for deposit independent of publication may deter publication—it may reduce drives to publish and give sponsors reasons not to fund paper writing. The suggested change would remove the onus from investigators and sponsors having to apologise for not being able to achieve an unrealistic time goal of one year.

Making deposit of data mandatory after three years from the date of completion, or termination for reasons of safety or efficacy, satisfies the need to make data available from trials not published and would have the advantage of providing reasons why results were not published. As the amendment stands, there is no way of knowing whether results were not written up at all or written up but rejected by journal editors and so no way to know the portion of publication bias due to investigators compared with that due to editors.