Comments on the ICMJE’s Proposals for Sharing Clinical Trial Data

Displaying 16 - 20 of 320 comments
  • P.J. Devereaux
    McMaster University, Hamilton, Ontario, Canada
    Role(s):
    • Researcher
    • Clinician
    • Clinical trialist
    • Clinical trial sponsor / funder
    Date Submitted: Monday, April 18, 2016 - 12:50

    Requirement To Share Data
    Comments: A total of 336 investigators from 28 countries have come together and signed onto a letter drafted by Drs. P.J. Devereaux, Salim Yusuf, Gordon Guyatt, Hertzel Gerstein, and Stuart Connolly of McMaster University, Canada, in response to ICMJE’s proposed data sharing plan. We have submitted this letter directly to the ICMJE and you can access the letter at http://www.phri.ca/wp-content/uploads/2016/04/Letter-from-336-investigators-from-28-countries-to-ICMJE-April-18-2016.pdf. We support data sharing in an appropriate and timely manner, but we do not support the proposal of the ICMJE, as it currently stands. We believe that the current proposed plan of forced data sharing will not result in a net benefit to patients, investigators, sponsors, or society. Our concern regarding possible net harm arises from likely unintended adverse consequences of the proposal, highlighted in Table 1 of our letter. These possible adverse consequences are based on several concerns that have lead us to propose alternative approaches to achieving the benefits of data sharing while minimizing the adverse consequences.

    6 Month Time Frame
    Comments: Major trials often take 5-10 years to conceive, fund, establish, recruit patients, complete follow-up, analyse, and report. In our experience, important findings often require a few years of analyses and drafting reports for publication. The current proposal of 6 months to release data deprives investigators who actually invested many years of efforts, the rightful opportunity to utilize the data for both fully understanding their results as well as reaping justifiable scientific credit for their many years of efforts. These investigators should be given a reasonable timeline to undertake the analyses of the relevant reliable data, submit the resulting papers, and get them published. Six months is not a realistic timeline, and it would result in situations where various groups simultaneously submit similar analyses to different journals, resulting in duplicate publications, some of which are likely to be incorrect and misleading. We propose instead that every study be provided a minimum of 2 years after the first publication of the results and up to an additional 6 months for every year that it took to complete the study (i.e., write the protocol, obtaining funding, obtain ethics and regulatory approvals, recruit the patients, complete the patient follow-up, and clean the database).

    Require a Data Sharing Plan

    Providing Credit
    Comments: The ICJME proposal for forced data sharing fails to consider fair alternatives that could achieve the goals without the adverse consequences outlined in Table 1 of our letter. Trial investigators themselves should be afforded the first opportunity to undertake analyses proposed by investigators who were not involved in a trial. Often it is only after the main trial results have emerged that follow-up questions are identified. This requires significant effort, money (which often requires additional grants), and time on the part of the original investigators. Only if the investigators cannot undertake the proposed analyses and submit the paper for publication within an 18 month time period, should the investigators who were not involved with the trial be given the opportunity to access the data for the proposed analyses. An independent entity like ClinicalTrials.gov could facilitate such a process.

    Other Comments: The ICMJE proposed plan on data sharing should be modified as follows: 1. To enhance readers’ confidence in published data, an independent statistician should be offered the opportunity to conduct confirmatory analyses before publication of a paper. 2. The timeline for providing de-identified IPD should be a minimum of 2 years after the first publication of the results and up to an additional 6 months for every year that it took to complete the study (i.e., write the protocol, obtaining funding, obtain ethics and regulatory approvals, recruit the patients, complete the patient follow-up, and clean the database). 3. Individuals who were not involved in an investigator-initiated trial but want access to the data should have to financially compensate the original investigators who conducted the trials, so they can recoup some of their costs of conducting the trial, for the efforts of making the data available, and to support further analyses of their data. 4. Trial investigators should be afforded 18 months to undertake any analyses and submit the resulting paper proposed by investigators who were not involved in a trial.

    Position:
    Comments:
  • Tara Federici
    Advanced Medical Technology Association (AdvaMed)
    Role(s): Other
      • Medical Device Trade Association
    Date Submitted: Monday, April 18, 2016 - 12:17

    Requirement To Share Data
    Comments: The Advanced Medical Technology Association, AdvaMed, greatly appreciates the opportunity to provide feedback on the ICMJE's proposal to require authors to share de-identified patient data (IPD) underlying results presented in articles in member journals. However, we have a number of concerns with the ICMJE proposal to require clinical trial data sharing which are detailed in our comprehensive response to the ICMJE proposal located on AdvaMed's website at this link: http://advamed.org/res/1097/advamed-comments-on-icmje-clinical-trial-data-sharing-requirement

    6 Month Time Frame
    Comments: AdvaMed has a number of concerns with the ICMJE proposal to require clinical trial data sharing detailed in our comprehensive response to the ICMJE proposal located on AdvaMed's website at the link provided in our response to the first question in this series of ICMJE feedback questions.

    Require a Data Sharing Plan
    Comments: AdvaMed has a number of concerns with the ICMJE proposal to require clinical trial data sharing detailed in our comprehensive response to the ICMJE proposal located on AdvaMed's website at the link provided in our response to the first question in this series of ICMJE feedback questions.

    Providing Credit
    Comments: AdvaMed has a number of concerns with the ICMJE proposal to require clinical trial data sharing detailed in our comprehensive response to the ICMJE proposal located on AdvaMed's website at the link provided in our response to the first question in this series of ICMJE feedback questions.

    Position:
    Comments:
  • Scott Stroup
    Columbia University
    Role(s):
    • Researcher
    • Clinician
    • Clinical trialist
    Date Submitted: Monday, April 18, 2016 - 11:25

    Requirement To Share Data
    • I agree with this general approach
    Comments: This is okay but one size fits all may not be appropriate. Some small trials may need to be exempted to protect subject identity. NIH vs. industry should be considered too.

    6 Month Time Frame
    • I agree with this general approach
    Comments: 6 months after publication is too short. Would need to clarify on-line or in-print publication especially since on-line publication can be very quick. Suggest one year after print publication.

    Require a Data Sharing Plan
    • I agree with this general approach

    Providing Credit
    • I agree with this general approach
    Comments: Perhaps an author of corporate authorship to authors of the original paper should be required. More collaboration would warrant being a named author. If original authors disagree with new paper could turn down authorship. Perhaps original authors should be offered opportunity to comment/rebut on future publications if they don't want to be authors or authorship otherwise not suitable.

    Position:
    Comments:
  • Abigail Goben
    University of Illinois at Chicago
    Role(s):
    • Researcher
    • Other
      • Medical Librarian
    Date Submitted: Monday, April 18, 2016 - 11:13

    Requirement To Share Data
    • I agree with this general approach
    Comments: Though supporting the release of IPD in order to provide further information for verification, replication, and reuse of the data, several consideration warrant further attention. This includes direction on navigating international privacy issues, guidance for including or removing indirect or quasi-identifiers, and suggestions on preventing false positives with re-identification-- particularly for highly sensitive/risky data. Data hosting for sharing should not be a function of the publishers, who have not generated the data, who do not curate or archive the data, and who have a demonstrated conflict of interest. There are institutional considerations which must be addressed for sharing of data. Due to the sensitive nature of this data, security and data brokering must be an institutional obligation, not merely the manuscript PI or the authors. This will be particularly important as researchers move between institutions and research data “upon request” is frequently unavailable. While the risks are made clear to study participants, that they must not overstated to the point that participants refuse. Boilerplate for institutional review boards should lean towards opting out of data sharing rather than opting in. Finally it is important to identify the opportunities for libraries to collaborate and facilitate clinical data sharing.

    6 Month Time Frame
    • I agree with this general approach
    Comments: Clarification is needed as to the 6 month timeline--is that from the pre-print being available or the date of final publication. Also, it is not clear how long data should be retained. It is unlikely that all clinical trial data can be preserved indefinitely, so a path for selection, deaccession, and destruction without loss of reputation to institutions or researchers is encouraged. Data librarians can assist here in helping clinical trial researchers create data sharing plans and maintaining these as living documents with appropriate documentation and metadata over time. Libraries can also provide guidance on best practices, de-identification and anonymization applications, and software and tools for data management and sharing. Additional clarification on the costs related to this and how that will be borne--included in grant funding, a sole obligation of the institution, etc, is necessary.

    Require a Data Sharing Plan
    • I agree with this general approach
    Comments: Libraries and data librarians have extensive experience in collaborating with researchers to develop data sharing plans to meet a variety of needs. Structured data sharing plans should contain, minimally, accessibility components such file formats and naming conventions; discoverability components; well-documented code; metadata; and retention/archive issues such as duration, backups, etc. Libraries can also assist with identification of an appropriate repository and clinical trial registry. The use of a standard metadata format, such as BioCaddie, is encouraged in addition to a ReadMe or other descriptive file.

    Providing Credit
    • I agree with this general approach
    Comments: There are a variety of needs in terms of providing credit for clinical trial data. One of the primary issues is that citation practices and research metrics will need to be widely adopted. Libraries already instruct in citation practices and can readily assist researchers and students in learning how to cite data. We also regularly assist in identifying impact of research and can expand this knowledge to determining the impact of clinical trial data. While credit and citation will be important, another piece of this will be helping researchers and students find data for reuse. This will have particular impact for educational reuse. This will also have impact on promotion and tenure norms at universities. Guidance on evaluating the impact of this clinical data and it’s value as a unique scholarly object will be immediately necessary. Adherence to data sharing norms will also impact grant applications and there will need to be seamless ways for researchers to report compliance to funders.

    Other Comments: Librarians have a unique position to act as collaborators in many data management, sharing, and curation activities. Many libraries already provide services such as these in support of funder and journal editor requirements, the addition of clinical trial data is a natural progression that we are well equipped to support in many ways. Libraries also value the broad sharing of scholarly outputs. We’ve been integral players of institutional open-access policies nationwide and are often the home for campus-wide scholarly communications, intellectual property staff, and support. Given our position within the scholarly landscape, Libraries persist. While disruptions in the economics of publishing put publishers in jeopardy of going out of business, libraries are central fixture of the institutions they serve and as such, will persist as long as those institutions do. We also have the added benefit of being local - near where the research data was created.

    Position:
    Comments:
  • Clinical Trials Unit
    London School of Hygiene & Tropical Medicine
    Role(s):
    • Researcher
    • Clinician
    • Clinical trialist
    Date Submitted: Monday, April 18, 2016 - 11:09

    Requirement To Share Data
    • I agree with this general approach
    Comments: How should IPD data be cited in publications in cases where data won't be made available at the time of publication? Do the ICMJE recommend that a metadata record describing the data is published in an appropriate registry? As an addition to the statement that deposition of data in a registry will not constitute prior publication, can it also be confirmed that creation of descriptive metadata in an appropriate registry is also not considered prior publication? Can the ICMJE provide examples of the 'necessary metadata' (mentioned in paragraph 3) that should be provided? Can the ICMJE clarify their position on the correct location to publish IPD data, can it be held & made available through a department website, for example, or is it preferable to publish it in registry (e.g. a digital repository, data enclave, CRIS, etc.)?

    6 Month Time Frame
    • I agree with this general approach
    Comments: Where the trial teams have secondary analyses planned which will take over 6 months, could the data sharing plan pre-specify some secondary analyses for which the data could not be used for a longer period?

    Require a Data Sharing Plan
    • I agree with this general approach
    Comments: Will a Data Sharing Plan template be made available and which organisation will be responsible for maintaining this? Theoretically, could a Data Sharing Plan be hosted in an institutional repository, CRIS, or other registry if it was assigned a persistent ID? In order to exercise control over who can access the data and in light of the high risks involved if people are able to publish incorrect results due to lack of oversight, it is important that the statement, '...or only after application to and approval by a learned intermediary...' remains in the final version. Could the data sharing plan require researchers to submit the specific research question(s) they want to address with a statistical analysis plan for approval? Will it be possible to exercise any control over research questions that the data can be used to address?

    Providing Credit
    • I agree with this general approach
    Comments: While this work is taking place, what type of credit would be appropriate? Is it sufficient to cite the data source in a Reference list/Data Access statement? Can the ICMJE's provide an example of the type of collaboration that could be established between data creators and data re-users?

    Other Comments: Could further guidance be provided on whether it will be necessary for secondary analyses to differ from the original research? How should non digital data be handled and will there be an exception in small trials where it might not be possible to keep identities anonymised? Should authors/institutions be encouraged to notify journals when data is available?

    Position:
    Comments:

Pages