Data Sharing

Implications for the Editor’s Network of a New Editorial Initiative of the International Committee of Medical Journal Editors 

This is an abridged version of a joint simultaneous publication initiative involving all interested National and Affiliated Cardiovascular Journals of the European Society of Cardiology

The Editor’s Network of the European Society of Cardiology (ESC) is committed to promote the implementation of high-quality editorial standards among ESC National Societies Cardiovascular Journals (NSCJ).^1–4 National Societies Cardiovascular Journals play a major role in disseminating scientific research but also in education and harmonization of clinical practice.³ National Societies Cardiovascular Journals complement ESC official journals and help to disseminate European cardiovascular research. Developing high-quality editorial policies remain of paramount importance to increase the prestige of NSCJ.^1–4 The editorś network advocates the adherence to the uniform recommendations of the International Committee of Medical Journal Editors (ICMJE).¹ Ethical issues play a major role in ensuring the credibility of the scientific process.^5–13 Novel ICMJE recommendations always appear as provocative and ambitious when developed. Recent examples include trial registration, conflicts of interest initiative and the new requirements for authorship.^9–13 However, as NSCJ are highly heterogeneous new recommendations should be embraced progressively. Herein, the novel ICMJE recommendations on data sharing¹⁴ are discussed with the aim to provide comprehensive editorial insights and, hopefully, to be progressively adopted and implemented by the NSCJ.

Sharing clinical trial data: the new International Committee of Medical Journal Editors proposal

The ICMJE considers that there is a moral obligation to responsible share the data generated by clinical trials.¹⁴ The rationale behind this initiative is that patients have assumed a risk by accepting to participate in a trial. Accordingly, making data publicly available represents a way to facilitate the advancement of science. Sharing the data would increase trust in the conclusions reach by trials as results may be confirmed by independent researches.¹⁴ Furthermore, new hypothesis may be pursuit to answer research questions not contemplated in the original study.

Governments, funding agencies, scientific societies, the industry and the lay society growingly demand sharing clinical trial data. Therefore, the ICMJE suggests that editors should also help to meet this ethical obligation by devising new editorial policies specifically addressing this issue.¹⁴ According to the ICMJE definition, a ‘clinical trial’ is a study that prospectively assigns people to an intervention to assess the cause-and-effect relationship between that intervention and the ensuing health outcome.⁵

The ICMJE considers that sharing ‘de-identified’ individual patient data should become part of the publication process.¹⁴ The requirement, however, is restricted to the individual-patient data underpinning the results presented in the published article. Importantly, a compromise and explicit plan for data sharing should be disclosed at the time of initial trial registration and should be also presented at the time of manuscript submission. It is important to keep in mind that clinical trial registration was a previous ICMJE editorial initiative aimed to address problems as publication bias, endpoints inconsistency, and redundant research.^9,10

Public repositories might provide an optimal tool for initial trial registration and also for individual-patient data sharing. In this regard, Clinicaltrials.com,^9,10 has already adapted its registration platform to specifically clarify data-sharing plans at the time of clinical trial registration. Details on whether the data would be freely available upon request or after a formal application, should be disclosed. The data should be made public no more than 6 months after publication of the original study in the journal.^9,10,14 The requirement would affect any clinical trial that begins patient enrolment 1 year after the official adoption of this policy by the journal.¹⁴

This editorial initiative may have profound consequences and may deeply influence publication strategies and the entire editorial process. Indeed, editors are supposed to monitor the data sharing process and address related potential irregularities. The ICJME acknowledges that the rights of the investigators and sponsors should be protected.¹⁴ Credit should be always given to the original investigators that posted the data after publication of their research. Investigators using these databases should request collaboration of the original investigators to ensure adequate data interpretation, management, and analysis.

Challenges of data sharing

There is inherent resistance to embrace open science initiatives from some academic institutions or investigators that prefer exploiting their ‘own’ data.^15,16 Until now trialists tended to see trial data as their personal property, refusing requests for data sharing. However, in other disciplines data sharing has been used for a long time.^15,16 Obtaining high-quality original data requires a major research effort. This new proposal will further increase the pressure on academic investigators that frequently have no resources to generate and publish subsequent analyses and require time to prepare the new manuscripts.¹⁴ The effort and resources required to organize the raw data in a way that would be comprehensible to others investigators remain a cause of concern.¹⁴ This would require adequate technical support and funding.

Data-access to non-trial researches will increase transparency and trust in trial results but might also generate confusion and undue scientific controversies. Reliable assessment of the data requires a deep knowledge on the study background and additional technical considerations. Incorrect analyses would result in inaccurate results and erroneous interpretations, eventually damaging science.¹⁴ An undue scientific noise, with contradictory results and rectifications, will generate confusion and frustration in the scientific community. As many related issues should still clarified the ICMJE asked for feedback on its proposal on data sharing.¹⁴ The initiative should be enriched from the experience gained during its adoption and implementation.

Previous initiatives on data sharing

Several leading academic entities have paved the way on this field. The British Medical Journal pioneered data sharing.¹⁷ In 2012, the policy only took effect for trials on drug and devices but, in 2015, the requirement was extended to all clinical trials.¹⁷ Individual patient data could be also of value during the peer review process by permitting independent verification of the results before manuscript acceptance.¹⁸ However, reviewers are already deluged with work and many of them do not have the expertise required to manage data and to perform confirmatory statistical analyses.¹⁸ Likewise, JAMA, requested for statistical analyses of industry-sponsored trials by an independent academic statistician.¹⁹

The World Health Organization (WHO) and the Institute of Medicine (IOM) announced declarations on clinical trial transparency including data sharing.^20–23 Finally, the National Health, Lung, and Blood Institute (NHLBI) described detailed data-sharing practices allowing public access to trial raw data and subsequently developed a data repository.²⁴ In 2015, the NHLBI disclosed its intent to make public the data from its funded trials.²⁴

Repositories

A high number of clinical trials are conducted annually worldwide generating a huge volume of patient-level raw data.²⁵ Currently, however, repositories for data sharing are still not adequate. Most of them require a time-consuming request, including a detailed research proposal and the statistical plan.²⁵ The submitted proposal is then reviewed by an independent research panel.^21,25–27 Currently, this process usually takes too long. Some systems provide an electronic form or template²¹ but oftentimes a de novo proposal is required. Finally, data can be shared through a public website or by direct communication between the data holder and the new researcher. The data holder should be available to provide the required support should questions arise. A legal agreement may be requested.²⁷

Major care should be taken to prevent the perils that may undermine the value of data sharing.^14,28 A recent UK survey disclosed potential risks associated with data sharing, including, (i) data misuse, (ii) incorrect secondary analyses, (iii) resource requirements, and (iv) patients’ identification.^29,30 Developing and adopting standard approaches to protecting patient privacy are urgently required.¹⁴

Adequate infrastructure should be organized to support effective data sharing. The growing role of the industry in this regard is demonstrated by joint initiatives, as the Yale University Open Data (YODA) project.^16,31 In addition, academic research organization consortiums focused on cardiovascular diseases have developed interesting data sharing tools.³²

Statistical issues

Statisticians play a key role in data sharing strategies.³³ Statisticians should move from their classical role as data ‘gate-keepers’ to that of data ‘facilitators’.³³ A data sharing working group of medical research statisticians, from the pharmaceutical and biotechnological industry and from the academia, has been recently created. The idea is to facilitate that the data initially collected and entered into the data base will be reusable. Converting row data in standardized formats may be challenging. Independent statisticians should help during these re-analyses and protect from misleading conclusions.

They should be prepared for different results but also to face criticisms.³³ Moreover, they should alert of the risk of over-interpretation of results from multiple subgroups analyses.³⁴ Best practices in data anonymization should be followed.³⁵ Finally, high security levels should be implemented for data transferring. Resources, costs and effort required to make patient-level data available for third party research require adequate funding.³⁵

Credit to the original authors

A clear motivation for researches is the opportunity to publish different studies complementing the results of the main manuscript.^36–37 Accordingly, many have proposed that the time for data sharing should be >1 year. This will allow a precious additional extra time for original investigators to analyse in depth their data and finish additional analyses. Otherwise, original investigators may be tempted to consider ‘research parasites’ those performing secondary analyses of their data. A close collaboration between primary and secondary researches is needed to ensure adequate data analyses.^14,15 Scientific credit should be always given to the original investigators that designed and conducted the trial.²⁸

Conclusions

The data transparency revolution is just another step ahead into an ‘open science’ culture.^38,39,40 Major hurdles in the adoption and implementation of the new data sharing ICMJE initiative should still be overcome.⁴¹ Experience gained by leading journals will be of major help to pave the way during this process. National Societies Cardiovascular Journals should progressively adapt their policies to promote transparency in biomedical research.

On behalf of the Editors’ Network European Society of Cardiology Task Force.

Karlen Adamyan¹, Jean-Yves Artigou², Michael Aschermann, Michael Boehm, Lino Goncalves, Alfonso Buendia, Pao-Hsien Chu, Ariel Cohen, Livio Dei Cas, Mirza Dilic, Anton Doubell, Dario Echeverri, Nuray Enç, Ignacio Ferreira-González, Krzysztof J. Filipiak, Andreas Flammer, Eckart Fleck, Plamen Gatzov, Carmen Ginghina, Habib Haouala, Mahmoud Hassanein, Gerd Heusch, Kurt Huber, Ivan Hulín, Mario Ivanusa, Rungroj Krittayaphong, Chu-Pak Lau, Germanas Marinskis, François Mach, Luiz Felipe Moreira, Tuomo Nieminen, Latifa Oukerraj, Stefan Perings, Luc Pierard, Tatjana Potpara, Walter Reyes-Caorsi, Se-Joong Rim, Olaf Rødevand, Georges Saade, Mikael Sander, Evgeny Shlyakhto, Bilgin Timuralp, Dimitris Tousoulis, Dilek Ural, J. J. Piek, Albert Varga, Thomas F. Lüscher

¹Editor in Chief Armenian Journal of Cardiology, ²Editor in Chief Archives des maladies du cœur et des vaisseaux Pratique, ³Editor in Chief Cor et Vasa, ⁴Editor in Chief Clinical Research in Cardiology, ⁵Editor in Chief Archivos de Cardiologia de Mexico, ⁶Editor in Chief Acta Cardiologica Sinica, ⁷Editor in Chief Archives of Cardiovascular Diseases, ⁸Editor in Chief Journal of Cardiovascular Medicine, ⁹Editor in Chief Medicinski Zurnal, ¹⁰Editor in Chief SAHeart, ¹¹Editor in Chief Revista Colombiana de Cardiologia, ¹²Editor in Chief Kardiyovaskuler Hemsirelik Dergisi, ¹³Editor in Chief Revista Española de Cardiología, ¹⁴Editor in Chief Kardiologia Polska, ¹⁵Editor in Chief Cardiovascular Medecine, ¹⁶Editor in Chief Cardio News, ¹⁷Editor in Chief Bulgarian Journal of Cardiology, ¹⁸Editor in Chief Romanian Journal of Cardiology, ¹⁹Editor in Chief Revista Portuguesa de Cardiologia, ²⁰Editor in Chief Cardiologie Tunisienne, ²¹Editor in Chief The Egyptian Heart Journal, ²²Editor in Chief Basic Research in Cardiology, ²³Editor in Chief Austrain Journal fo Cardiology, ²⁴Editor in Chief Cardiology Letters, ²⁵Editor in Chief Cardiologia Croatica, ²⁶Editor in Chief Thai Heart Journal, ²⁷Editor in Chief Journal of the Hong Kong Colleage of Cardiology, ²⁸Editor in Chief Seminars in Cardiovascular Medicine, ²⁹Editor in Chief Cardiovascular Medecine, ³⁰Editor in Chief Arquivos Brasileiros de Cardiologia, ³¹Editor in Chief Sydänääni (Heart Beat), ³²Editor in Chief La Revue Marocaine de Cardiologie, ³³Editor in Chief Der Kardiologie, ³⁴Editor in Chief Acta Cardiologica, ³⁵Editor in Chief Heart and Blood Vessels, ³⁶Editor in Chief Revista Uruguaya de Cardiologia, ³⁷Editor in Chief Korean Circulation Journal, ³⁸Editor in Chief Hjerteforum, ³⁹Editor in Chief Heart News, ⁴⁰Editor in Chief Cardiologisk Forum, ⁴¹Editor in Chief Russian Journal of Cardiology, ⁴²Editor in Chief Anatolian Journal of Cardiology, ⁴³Editor in Chief Hellenic Journal of Cardiology, ⁴⁴Editor in Chief Archives of the Turkish Society of Cardiology, ⁴⁵Editor in Chief Netherlands Heart Journal, ⁴⁶Editor in Chief Cardiologia Hungarica, ⁴⁷Editor in Chief European Heart Journal

Acknowledgements

We are grateful for the support and assistance of Ismahem Ouertani and Michael Alexander from the ESC National Cardiac Societies Relations Department and the Publication Department at the Heart House.

Conflict of Interest: none declared.

References

References are available as supplementary material at European Heart Journal online.