ENG I DEU
The International Committee of Medical Journal Editors’ (ICMJE) plan for mandatory sharing of clinical trial data as prerequisite for publication in medical journals1 triggered mixed responses from the scientific community. At first glance, data sharing has the potential to advance medical sciences and promote transparency. However, not all clinical trial investigators are looking forward to the prospect of sharing their data. ^2This article reviews some of the concerns regarding data access and ideas about its regulation that have been voiced since the ICMJE proposal in January 2016.
Benefits of trial data sharing:
• Independent verification of results
• Opportunities for new discoveries from secondary analyses
• Data for developing and testing of new hypotheses
• Avoidance of unnecessary trial repetition
• High quality systematic meta-analyses from original data of related studies
Potential problems with data sharing:
• Curating and presenting the raw data of a complex clinical trial in a way that is understandable for researchers who were not involved in the trial design process requires time and money. This may divert investigator’s resources from further research.3
• Publication of secondary analyses of trial results based on misunderstood parameters and data structure may reduce confidence in the primary results.4
• A large number of analyses on a given data set may result in finding associations by chance, possibly blurring the real evidence.4
• Inadequate data de-individualisation may jeopardise the study participants’ privacy.4
• Data sharing may create a niche for “research parasites” which use data generated by others to their own ends such as publication of sub-analyses ahead of the group who conducted the study, or trying to discredit the original findings and conclusions.2
These concerns may be best addressed by carefully regulating access to the original trial data. Some of the proposed systems for data management and access regulation are discussed in the following paragraph.
Storing and sharing of trial data managed by independent organisation
• The idea of data sharing is not new. The Yale Open Data Access YODA project is in operation since 2011 and acts as a “trusted intermediary” to which original authors entrust their data and where scientists can apply for data access.5
Committee decisions on sharing application
• At YODA, researchers interested in trial data submit an application and a decision is made by YODA scientists based on a blinded review process. Some of the criteria for a positive decision are a clearly described scientific purpose, use of the data to inform science and public health, and feasibility of the proposed research.5
• The Academic Research Organisation Consortium for Continuing Evaluation of Scientific Studies–Cardiovascular (ACCESS CV) suggests a committee formed by ACCESS CV scientists not involved in the original trial, the trial's principal investigator, a trial statistician and a member of the data and safety monitoring board to be in charge of reviewing applications and granting access if the secondary analysis is deemed “feasible, hypothesis-based, non-duplicative, and guided by investigators with technical capability and a plan for publication.”4
Embargo period after publication before data is shared to allow original authors time for further analyses and publication of sub-studies
• The ICMJE proposed a period of six months after publication before trial data is shared.1
• The International Consortium of Investigators for Fairness in Trial Data Sharing advocates a two-year baseline period which is extended by six months for each year the study took to complete, up to a maximum of five years.3
• ACCESS CV suggests a two years moratorium before data are shared.4
• The International Consortium of Investigators for Fairness in Trial Data Sharing additionally suggest data analysis by an independent statistician before publication to increase confidence in the results, and compensation for the effort and cost of making the data available to original study authors by researchers not involved in the primary study.3
Where are we now?
Many of the perceived problems with data sharing can be overcome by high standards for shared data sets and secondary analyses. However, this will require considerable efforts by primary and secondary study authors and reviewers, which has to be accounted for in planning future clinical trials.
Most scientists involved in the topic seem to agree that trial data sharing is a good thing in theory, but there is little common understanding about the practical terms of data access. Considering the different interests and attitudes of scientists who have invested years of work in generating the data, government institutions and the public who paid their salaries in tax money, and trial sponsors who may have invested millions of euros, it is unlikely that there will be a decision for a uniform data sharing system anytime soon.
What are the possible outcomes?
A unified sharing system for clinical trial data would certainly be the fairest outcome for authors and publishers. However, the current debate may result in a patchwork of sharing systems across the medical publication landscape. In this case, future study authors may base their decision for a journal on ownership of their data in addition to scope and impact, thus weakening the position of journals with strict sharing rules. Nevertheless, the NEJM is committed to sharing: “when appropriate systems are in place, we will require a commitment from authors to make available the data that underlie the reported results of their work within six months after we publish them.”6
Post scriptum: the potential of shared trial data is currently being demonstrated in the Systolic Blood Pressure Intervention Trial (SPRINT) data analysis challenge SPRINT Challenge, ending in April 2017. It will be interesting to see what can be achieved with such a data set in addition to the original study findings.
Author: Dr. Mario Pahl
1. Taichman DB et al. Sharing Clinical Trial Data--A Proposal from the International Committee of Medical Journal Editors. N Engl J Med 2016; 374: 384–386
2. Longo DL et al. Data Sharing. N Engl J Med 2016; 374: 276–277
3. Devereaux PJ et al. Toward Fairness in Data Sharing. N Engl J Med 2016; 375: 405–407
4. Patel MR et al. Sharing Data from Cardiovascular Clinical Trials--A Proposal. N Engl J Med 2016; 375: 407–409
5. Krumholz HM et al. The Yale Open Data Access (YODA) Project--A Mechanism for Data Sharing. N Engl J Med 2016; 375: 403–405
6. Drazen JM. Data Sharing and the Journal. N Engl J Med 2016; 374: e24