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

Displaying 61 - 65 of 320 comments
  • Neil Wright
    Queen Mary University of London
    Role(s): Researcher
    Date Submitted: Friday, April 15, 2016 - 11:10

    Requirement To Share Data
    • I agree with this general approach
    Comments: The IPD “underlying” the results presented in an article may not be a clear enough definition. Is it intended that the analysis, figure, or table could be reproduced from the shared deidentified IPD? However, data may need to aggregated, or variables removed, to deidentify the IPD. Then some or all of the presented results may not be reproducible at all or in part, exactly or approximately. Would this complete-as-possible IPD fulfil the requirements? How, and by who, is a judgement of utility of deidentified IPD to be made? Additionally, qualitative data, images or audio may not be able to be deidentified. And routine data may, legally, not be able to shared further.

    6 Month Time Frame
    Comments: Timely sharing of data is important, but I have several concerns about the proposed 6 month timeframe: The time frame may not appropriate if the paper concerns short-term data from a long-term follow up study, or where further secondary analysis is planned by the by research team. The burden of the processes involved (including: deidentification of a data set, preparing and completing data use agreements, fulfilling technical requirements for transferring data, and the administration of data sharing mechanism) are all unknown. Any imposed time frame should be supported by experience of these (time and resource) costs, and suitable for all types of study that the requirements are to be applied. If mechanism of data sharing is “after application to and approval by a learned intermediary”, where would the 6 month time frame fit with this process? Would applications for the data have to be accepted from at least 6 months post publication, or would the data itself need to be available to share at this point (having complete all other steps of the process).

    Require a Data Sharing Plan
    • I agree with this general approach
    Comments: A data sharing plan would be a reasonable requirement. The requirement that data can be shared “under less restrictive, but not more restrictive, conditions than were indicated in the registered data-sharing plan” is understandable, but could be inappropriate and self-defeating. There can be cases where more restrictive conditions are needed later in the process. For example, something that is expected to be anonymous is learned to not be, or other data is made publicly available which affects identification in the study data. Further, whether PID is deidentified depends on the data itself: when data is collected there may be unexpected patterns (e.g. outliers, rare combinations) which affect the identifiable nature of the data and in turn the plans for data sharing. If there is only flexibility in one direction this may encourage data sharing plans to be written as restrictive as possible, rather than be a fair reflection of the intentions of the study team to share data.

    Providing Credit

    Position:
    Comments:
  • dr vikas rajpurohit
    dr sn medical college jodhpur
    Role(s): Clinician
    Date Submitted: Friday, April 15, 2016 - 10:37

    Requirement To Share Data
    • I agree with this general approach

    6 Month Time Frame

    Require a Data Sharing Plan

    Providing Credit

    Position:
    Comments:
  • Christoph Gerlinger, EFSPI regulatroy chair
    European Federation of Statisticians in the Pharmaceutical Industry
    Role(s): Other
      • learned society
    Date Submitted: Friday, April 15, 2016 - 10:22

    Requirement To Share Data
    • I agree with this general approach
    Comments: The European Federation of Statisticians in the Pharmaceutical Industry (EFSPI) supports responsible data sharing. From a statistical point of view, responsible data sharing includes that patient data are handled in a strictly confidential manner to avoid misuse under all possible circumstances and that shared data will be used for further development of science as much as possible applying good statistical principles. In our view, patient data cannot be de-identified in such a way that re-identification is virtually impossible, given today’s technical possibilities, let alone future capabilities. Therefore patient data should only be shared using strict access controls to the data. Good Clinical Practice requires an independent review of the research protocol and qualified researchers to conduct and analyze a clinical trial. Similar standards should be required for secondary analyses of clinical trials. In our view, transparency is a two-way effort. Secondary researcher should be obliged to register their studies and to publish, or at least web-post, their results. This is the only way to assess the multiplicity problem raised by secondary research.

    6 Month Time Frame
    Comments: In our view only data that is not under regulatory review should be available for sharing. Hence, we believe that a general 6 month time frame is not appropriate.

    Require a Data Sharing Plan
    • I agree with this general approach
    Comments: The European Federation of Statisticians in the Pharmaceutical Industry (EFSPI) believes that a proper planning of the future data sharing is essential, including agreed principles of restricted access and pre-specified plans for secondary analyses as indicated above. However, if the sponsor of the trial has a general data sharing policy in place we do not see a benefit in requesting an additional study specific data sharing plan.

    Providing Credit
    • I agree with this general approach
    Comments: As a minimum any publication based on shared data should have a statement in the publication reflecting the research being presented is based on shared data and to cite where the data came from.

    Other Comments: While pharmaceutical companies already have systems in place that allow responsible data sharing (e.g., ClinicalStudyDataRequest.com) ICMJE could assist academia in setting up similar possibilities. Please see also our forthcoming papers on the statistical aspects of data sharing: Fletcher et al. Statistical guidance for responsible data sharing: An overview. Tucker et al. Ensuring patient confidentiality when sharing patient-level data from clinical trials. Hollis et al. Best Practice for Analysis of Shared Clinical Trial Data. Sudlow et al. EFSPI/PSI Working Group on Data Sharing: Accessing and Working with Clinical Trial Patient Level Datasets – a Primer for Researchers. BMC Medical Research Methods (in press).

    Position:
    Comments:
  • Professor Leena Bruckner-Tuderman, Vice President of the German Research Foundation (DFG)
    on behalf of the DFG, the MFT, the KKS-Network and the TMF
    Role(s):
    • Researcher
    • Clinician
    • Clinical trialist
    • Clinical trial sponsor / funder
    Date Submitted: Friday, April 15, 2016 - 08:34

    Requirement To Share Data
    • I agree with this general approach
    Comments: The German Research Foundation (DFG), the Association of Medical Faculties in Germany (MFT), the Network of Coordinating Centres for Clinical Trials (KKS-Network) and the TMF - Technology, Methods and Infrastructure for Networked Medical Research welcome the ICMJE’s proposal as an important step to making the best use of clinical trial data. The intention of this proposal is in large parts in line with the DFG principle of open access to data from publicly funded research. Both, data sharing and reanalysis of clinical trial data need to be recognized as important scientific contributions. Especially in the context of clinical trials, however, competing interests between researchers, the public, and trial participants need to be carefully balanced. Any condition regarding data sharing needs to find broad acceptance within the scientific community and must comply with data protection regulations. Therefore, unrestricted public access to deidentified clinical trial data is most likely not appropriate in many cases. Authors are not to be barred from publication in ICMJE member journals if data sharing is not possible, e.g. in trials with very small numbers of patients (as in studies of rare diseases, safeguarding the confidentiality of trial participants), for ethical reasons or data protection regulation.

    6 Month Time Frame
    Comments: We support early data sharing and a timeframe of 6 months after publication is most likely suitable for the majority of published clinical trials. It offers the opportunity of independently validating the published results and generating additional knowledge from secondary analysis. However, a general 6 month timeframe does not adequately take into account individual circumstances that may preclude early data sharing. The original investigators should have the opportunity to extensively analyze and publish the data that sometimes took many years to generate. We propose that standards should be developed that take into account individual specificities of clinical trials as well as scientific, ethical, clinical, public health, intellectual property rights or policy implications.

    Require a Data Sharing Plan
    • I agree with this general approach
    Comments: To ensure that data collected in clinical trials are useful for the wider research community and suitable for replication and secondary analysis, a data sharing plan needs to be developed before data are collected. The data sharing concept will have consequences on the ethical evaluation, on the informed consent process and on how clinical trials are conducted. For reasons of transparency, the data sharing plan should be published as a component of the clinical trial registration. Sufficient time needs to be given to clinical trial registries to implement this functionality as part of the registration.

    Providing Credit
    • I agree with this general approach
    Comments: Conducting clinical trials requires an enormous effort in the preparation, conduct, and data analysis. Sharing clinical trial raw data should be based on a formal, contractual agreement of collaboration between those who generated the data and scientists wishing to reanalyze the data. A reanalysis plan clearly defining the use of data should be agreed on as the involvement of those who generated clinical trial raw data is crucial for the quality of data analysis. Even if a secondary data analysis is not based on a formal agreement, the work of the trialists who generated the data needs to be adequately acknowledged. It should be discussed whether the intellectual contribution of those who generated the data would justify co-authorship or whether a novel way of authorship needs to be established.

    Other Comments: Sharing clinical trial data with a wider scientific community has enormous potentials but also comes with great responsibilities. The scientific community has to develop policies to ensure that the best use is made of shared data and that data is not used inappropriately. For the benefit of clinical care and the advancement of knowledge, sharing and reanalysis of data need to follow the highest quality standards. As a funder, the German Research Foundation acknowledges that costs are associated with data sharing and that appropriate technologies and infrastructures need to be developed. Therefore, the DFG provides funds for data processing necessary for data sharing and for the use and establishment of data sharing infrastructures. The need for additional resources to implement such measures should also be acknowledged by other funders and institutions.

    Position:
    Comments:
  • david ansell
    IMS Health
    Role(s):
    • Researcher
    • Clinician
    Date Submitted: Friday, April 15, 2016 - 05:46

    Requirement To Share Data
    • I agree with this general approach
    Comments: Issues: It is deidentified data but pseudonymised (ie person can be reidentified if enough is known). We receive pseudonymised hospital data from NHS on conditions of : secure storage, logged named access, secure hard disk destruction to ISO 27001:2005 (disk disintegration), encrypted backups, auditor access to check compliance. This is not easy to enforce in a data share with a 3rd party

    6 Month Time Frame
    • I agree with this general approach
    Comments: see above data concerns

    Require a Data Sharing Plan
    • I agree with this general approach
    Comments: see above data concerns

    Providing Credit
    • I agree with this general approach

    Other Comments: Controls on access to shared data remains a concern. Data reuse Legal ownership Contractual responsibilities Storage Destruction Patient consent - EU law changes . While given consent for study, this is not a carte blanche consent. A patient may need to be given the opportunity to opt out of having his annonymised data sent "off site" from study

    Position:
    Comments:

Pages