Sie sind auf Seite 1von 3


The benefits and risks of requiring researchers

toshare data
The views expressed in this editorial are those of the author and do not necessarily reflect the position of the publisher.

The International Committee of Medical Journal Editors has sponsored many
developments in medical research reporting that have improved the quality of
research. Their proposal to require a statement by researchers regarding the
availability of underlying data for sharing has the potential for a significant
advance in medical knowledge. The Canadian Journal of Surgery supports the
initiative and will develop mechanisms to support authors in meeting this new

or many young authors, the steps to publication are such as in certain types of survival curves, comparison
frustrating and annoying. Their older peers can glee between which may involve complex computer code. In
fully tell them that it was worse in the old days. Not January of this year, ICMJE posted a provisional policy to
too long ago, manuscripts were prepared by typewriter. If require reports to include a statement regarding the avail
a reviewer requested changes, great ingenuity was required ability of supporting data for sharing.4 This statement
to avoid disrupting the citation sequence too much. If a should include information regarding the trial protocol
paper was refused, submission to another journal usually (which may be available from; the sta
required complete retyping of the manuscript to meet its tistical code used and the data set. While the primary aim
different specifications. In 1978, a group of journal editors is to allow for the confirmation of research claims, other
met in Vancouver, BC, to address this latter problem by benefits will include better archiving of data, better meta-
developing uniform requirements for manuscripts submit analysis and the possibility of secondary analysis.
ted to biomedical journals. Not only did they give us the The strongest objections to the proposal have come
Vancouver citation style, they also continued as the Inter from members of the cardiovascular clinical trials com
national Committee of Medical Journal Editors (ICMJE) munity, who are concerned that compliant authors will
to promote quality in reporting of medical science. The be disadvantaged by sharing data with others who have
Canadian Journal of Surgery chose then, and continues now, neither paid nor worked for the results. They are prob
to follow ICMJE recommendations. ably expressing in polite terms what is much more
The scope of ICMJE broadened from requirements for acutely felt by commercial sponsors. They counter-
manuscript preparation to include conduct, reporting, propose a 24-month moratorium on data sharing to per
editing and publication of scholarly work in medical jour mit authors the opportunity for secondary analysis. 5
nals.1 While CJS highlighted issues, such as the min There are, however, other risks to data sharing. Patient
imum contribution for authorship and fraudulent report identification, imagined or real, may be possible using
ing, ICMJE also tried to improve the definition of the residual identifiers despite current de-identification
roles and responsibilities of reviewers, editors, publishers techniques. Hostile reviewers can use the data to dis
and owners with respect to bias, conflict of interest and credit researchers in a forum where rebuttal is difficult
editorial freedom.2,3 In 1997, a clinical trial registry, now or impossible. Archiving data and answering queries
known as, was established in the United have costs that may extend beyond the budget and
States to give the public access to research data, particu period of the research. The co-operation of patients and
larly about pharmaceutical therapies. ICMJE exploited the research ethics boards has not been established. Finally,
existence of this registry to require preregistration of clin methods to protect researchers from frivolous requests
ical trials submitted for publication to supporting journals have not been considered.
in order to avoid the suppression of trials with negative The current proposal applies only to data presented in
outcomes. CJS adopted this requirement. aggregate form in reports of prospective trials of an inter
Scientific reports traditionally present aggregate data. vention in humans, and it is permissible to state that data
Presentation may require intense statistical manipulation, are not available. Extension of the policy is inevitable. The

364 o
J can chir, Vol. 59, N 6, dcembre 2016 2016 Joule Inc. or its licensors

logic for data sharing applies to all biomedical research. Affiliations: Coeditors, Canadian Journal of Surgery (McAlister, Harvey);
Department of Surgery, Western University, London, Ont. (McAlister);
Funding agencies do not currently require data sharing, and the Division of Orthopaedic Surgery, Montreal General Hospital,
but most likely will in the future. Registration of a trial on Montreal, Que. (Harvey). now requires the disclosure of data shar Competing interests: E.J. Harvey is the Chief Medical Officer of Grey
ing arrangements. box Healthcare (Montreal) and Chairman of the Board of NXT-Sens Inc.
CJS supports the goals of the data sharing proposal. (Montreal). None declared for V.C. McAlister.
We will make available to authors a consensus regarding
adequate participant de-identification and safe storage of DOI: 10.1503/cjs.015116
experimental data. We will develop a model confidential
ity agreement that would allow bona-fide researchers References
access to data for the purpose of analyzing the published
1. International Committee of Medical Journal Editors. Recommendations
results, with a requirement that reports be submitted to for the conduct, reporting, editing, and publication of scholarly work in
the journal for publication so that the original author has medical journals. Available: (accessed
an opportunity to reply. All other agreements for data 2016 Nov. 2).
sharing will be between the researcher and the applicant 2. Meakins JL, Waddell JP. Authorship. Can J Surg 2002;45:84-7.
according to current practice. Data sharing, properly 3. Meakins JL. Authorship. Can J Surg 2001;44:405-6.
administered, represents a generational advance in the 4. Taichman DB, Backus J, Baethge C, et al. Sharing clinical trial data: a
development of medical knowledge. proposal from the International Committee of Medical Journal Editors.
CMAJ 2016;188:91-2.
5. Patel MR, Armstrong PW, Bhatt DL et al. Sharing data from cardio
Vivian C. McAlister, MB; Edward J. Harvey, MD vascular clinical trials a proposal. N Engl J Med 2016;375:407-9.

Les avantages et les risques dexiger que les

chercheurs partagent leurs donnes
Les opinions exprimes dans cet ditorial sont celles de lauteur et ne reprsentent pas ncessairement celles de lditeur.

Le Comit international des rdacteurs de revues mdicales a rendu possibles
de nombreuses avances dans la communication des rsultats de recherche
mdicale qui ont amlior la qualit de la recherche. Sa proposition de
demander aux chercheurs une dclaration sur laccessibilit de donnes sous-
jacentes partager pourrait faire grandement progresser les connaissances
mdicales. Le Journal canadien de chirurgie appuie linitiative et laborera des
mcanismes visant aider les auteurs respecter cette nouvelle exigence.

our beaucoup de jeunes auteurs, le processus de le style de citation Vancouver, mais ils ont aussi form le
publication est frustrant et irritant. Leurs pairs plus Comit international des rdacteurs de revues mdicales
gs leur diront volontiers que les choses taient (CIRRM), qui continuerait promouvoir la qualit des
bien pires autrefois. Il ny a pas si longtemps, les manus rapports de recherche mdicale. Le Journal canadien de
crits taient prpars la machine crire. Si un examina chirurgie (JCC) a alors choisi de suivre les recommanda
teur demandait des changements, il fallait faire preuve tions du CIRRM il le fait dailleurs toujours aujourdhui.
dune grande ingniosit pour viter de trop perturber la La porte des activits du CIRRM sest largie pour
squence des citations du texte. Si un article tait refus, il inclure en plus des exigences de prparation des manu
fallait gnralement pour le soumettre une autre revue le scrits la conduite, les rapports, ldition et la publication
retaper en entier pour respecter les diffrentes exigences. des travaux universitaires dans les revues mdicales 1.
En1978, des rdacteurs en chef se sont runis Vancouver Alors que le JCC mettait en vidence des problmes,
(C-B) pour rgler ce dernier problme : ils ont mis en comme la contribution minimale pour lattribution de la
place des exigences uniformes pour les manuscrits soumis paternit dune uvre et les rapports frauduleux, le
des revues biomdicales. Ils nous ont non seulement donn CIRRM tentait damliorer la dfinition des rles et

Can J Surg, Vol. 59, No. 6, December 2016 365


responsabilits des examinateurs, des rdacteurs en chef, La proposition actuelle ne sapplique quaux donnes
des diteurs et des propritaires en ce qui concerne la prsentes sous forme agrge dans les rapports dtudes
partialit, les conflits dintrts et la libert ditoriale2,3. prospectives portant sur une intervention chez lhumain, et
En 1997, un registre des essais cliniques, appel il est permis de dclarer que les donnes ne sont pas acces
aujourdhui, a t tabli aux tats-Unis sibles. Llargissement de la politique est invitable. La
pour donner au public accs aux donnes de recherche, logique derrire le partage des donnes sapplique toute
plus particulirement au sujet des traitements pharmaceu la recherche biomdicale. Les organismes de financement
tiques. Le CIRRM a tir parti de lexistence de ce registre nexigent actuellement pas le partage des donnes, mais ils
et demand la prinscription des essais cliniques soumis le feront probablement lavenir. Dailleurs, linscription
pour publication aux revues qui lappuient afin dviter la dun essai sur ncessite maintenant la
suppression des essais aux rsultats ngatifs, exigence que divulgation des ententes de partage des donnes.
le JCC a adopte. Le JCC appuie les objectifs de la proposition sur le
Les rapports scientifiques prsentent habituellement partage des donnes. Nous mettrons la disposition des
des donnes agrges, ce qui peut ncessiter dintenses auteurs un consensus sur lanonymisation adquate des
manipulations statistiques, par exemple, dans certains participants et le stockage scuritaire des donnes expri
types de courbes de survie, des comparaisons utilisant un mentales. Nous crerons galement un modle dentente
code informatique complexe. En janvier dernier, le de confidentialit qui permettra aux chercheurs srieux
CIRRM a publi une politique provisoire demandant que daccder aux donnes dans le but danalyser les rsultats
les rapports contiennent une dclaration sur laccessibilit publis, la condition quils prsentent leurs rapports la
de donnes dappui partager4. Cette dclaration devrait revue pour publication afin que lauteur original ait
comprendre des renseignements sur le protocole dessai loccasion de rpondre. Toutes les autres ententes de
(qui pourrait tre accessible sur, le partage des donnes seront conclues entre le chercheur et
code statistique utilis et lensemble de donnes. Bien que le demandeur selon la pratique actuelle. Le partage des
le but principal soit de permettre la confirmation de la donnes, bien administr, reprsente une avance gnra
proprit des recherches, cette exigence entranera tionnelle dans lavancement des connaissances mdicales.
dautres avantages, comme lamlioration de larchivage
des donnes et des mta-analyses, et la possibilit VivianC. McAlister, MB; EdwardJ. Harvey, MD
deffectuer une analyse secondaire.
Les principales objections la proposition viennent de Affiliations : Cordacteurs en chef, Journal canadien de chirurgie
membres de la communaut des essais cliniques cardiovas (McAlister, Harvey); Dpartement de chirurgie, Universit Western,
culaires, qui craignent que les auteurs sy conformant London, Ont. (McAlister); et Service de chirurgie orthopdique,
Hpital gnral de Montral, Montral, Qc (Harvey).
soient dsavantags de devoir partager leurs donnes avec
des personnes qui nont rien pay et nont pas travaill Intrts concurrents : E.J. Harvey est mdecin hyginiste en chef
de Greybox Healthcare (Montral) et prsident du Conseil
pour obtenir les rsultats. Ils expriment probablement en dadministration de NXTSens inc. (Montral). Aucun dclar pour
termes polis ce que les commanditaires commerciaux res V.C.McAlister.
sentent beaucoup plus intensment. Ces opposants propo
sent plutt un moratoire de 24 mois sur le partage des DOI: 10.1503/cjs.015416
donnes, qui donnerait aux auteurs loccasion deffectuer
une analyse secondaire5. Le partage des donnes comporte
cependant dautres risques. Lidentification des patients, Rfrences
hypothtiques ou rels, pourrait tre possible laide
1. International Committee of Medical Journal Editors. Recommendations
didentifiants rsiduels, malgr les techniques for the conduct, reporting, editing, and publication of scholarly work in
danonymisation employes. En outre, des examinateurs medical journals. Available: (accessed
hostiles pourraient utiliser les donnes pour discrditer les 2016 Nov. 2).
chercheurs dans un forum o la rfutation est difficile ou 2. Meakins JL, Waddell JP. Authorship. Can J Surg 2002;45:84-7.
impossible. Larchivage des donnes et la rponse aux 3. Meakins JL. Authorship. Can J Surg 2001;44:405-6.
questions ont des cots qui pourraient dpasser le budget 4. Taichman DB, Backus J, Baethge C, et al. Sharing clinical trial data: a
et la priode de la recherche. La coopration des patients proposal from the International Committee of Medical Journal Editors.
et des conseils dthique de la recherche na pas t CMAJ 2016;188:91-2.
tablie. Enfin, aucune mthode visant protger les cher 5. Patel MR, Armstrong PW, Bhatt DL et al. Sharing data from cardio
cheurs des demandes futiles na t envisage. vascular clinical trials a proposal. N Engl J Med 2016;375:407-9.

366 o
J can chir, Vol. 59, N 6, dcembre 2016