Rapid Responses to:
|
|
Rapid Responses: Rapid Responses published:
-
![[Read Rapid Response]](/icons/misc/sectionDOWN.gif)
Best practice for publishing trial data in peer-reviewed journals
- I Hrynaszkiewicz (26 March 2009)
-
Medical Observational Studies
- S. Stanley Young (28 March 2009)
-
More transparency in medical journals.
- Roger K.A. Allen (29 March 2009)
-
BMJ opens can of worms
- Trisha Greenhalgh (30 March 2009)
-
Ethical considerations to data sharing within a postmodernist paradigm.
- Sophie E Park (2 April 2009)
-
The challenges of data sharing for qualitative health research
- Michael D Gallagher, Allison Worth, Aziz Sheikh (2 April 2009)
-
"Anonymised Raw Trial Data and the Unnecessary Need for Consent"
- Neil McIntosh, Elaine M Boyle (21 April 2009)
-
Lack of Online Financial Transparency is a Potential Financial Conflict
- M Felix Freshwater (22 April 2009)
-
Re: Lack of Online Financial Transparency is a Potential Financial Conflict
- Trish Groves (22 April 2009)
|
|
|||
|
I Hrynaszkiewicz, Managing Editor BioMed Central Ltd, Gray’s Inn Road, London, WC1X 8HL, UK
Send response to journal:
|
Our thanks to the BMJ for highlighting the data sharing debate in the clinical research community. Indeed, journal policies have been associated with the increased prevalence of sharing of certain types of research data, but the overall prevalence of clinical data sharing remains low [1]. In order to substantially influence cultural changes and facilitate data sharing in clinical research however, mandates will probably be needed at the funder, institution, and/or journal level. Researchers may yet need to see more evidence of the benefits of data sharing, and codes of conduct [2] for analysis of published raw data should be agreed. Further infrastructure may also be required, such as the Medical Research Council's Data Support Service and the UK Data Archive [3]. As well as proposing the solutions that have been reproduced in the current article (box) [4], a further action to emerge from the discussions Dr Groves refers to was to produce widely-agreeable guidance on preparing raw data for publication in peer-reviewed journals. How we handle trial data retrospectively (from previously published trials) and prospectively (from proposed or ongoing research) also requires consideration. As the journal Trials encourages the inclusion of supplementary material and large data sets with manuscripts submitted to the journal, its editors are currently undertaking the challenge of producing this 'best practice' guidance. Any members of the clinical trials community interested in participating in this initiative and/or would be interested in reviewing draft guidance are encouraged to contact the editorial office at: editorial@trialsjournal.com and to visit the website for more information: http://www.trialsjournal.com/info/about/ References 1. Piwowar HA, Chapman WW: A review of journal policies for sharing research data. Proceedings of the ELPUB 2008 Conference on Electronic Publishing. Toronto, ON: 25-27 June 2008 [http://elpub.scix.net/data/works/att/001_elpub2008.content.pdf]. 2. Vickers AJ: Whose data set is it anyway? Sharing raw data from randomized trials. Trials 2006, 7:15. 3. UK Data Archive. [http://www.data-archive.ac.uk/sharing/] 4. Hrynaszkiewicz I, Altman DG: Towards agreement on best practice for publishing raw clinical trial data. Trials 2009, 10:17 Competing interests: IH is employed by BioMed Central Ltd, which publishes the journal, Trials. |
|||
|
|
|||
|
S. Stanley Young, Statistician NISS, PO Box 14006, RTP, NC 27709, USA
Send response to journal:
|
The reasons for making research data and materials are well documented(1). One purpose of making data available is so that claims can be subjected to verification. In RCT it is common to adjust the statistical testing for multiple comparisons. In observational studies there is most typically no adjustment for multiple comparisons. In the case of RCT the data is often submitted to regulatory agencies so in that case there is verification. There is largely no independent verification for medical observational studies; it is largely “trust me” science. The large credibility gap for medical observational studies could usefully be filled if editors and funding agencies pressed the case for data sharing. The decision of BMJ to have authors state the degree of data availability is useful sunshine. 1. Board on Life Sciences, Sharing Publication-Related Data and Materials: Responsibilities of Authorship in the Life Sciences (National Academies Press, Washington, DC, 2003) (see www.nap.edu/books/0309088593/html/). Competing interests: None declared |
|||
|
|
|||
|
Roger K.A. Allen, Senior Consultant Thoracic and Sleep Physician Wesley Medical Centre, Auchenflower, Brisbane, Qld 4066, Australia
Send response to journal:
|
With regards the above topic may I squeeze out some more pus from the carbuncle of medical research as was posted in my blog recently (www.sarcoidosis.com.au) and also on the BMJ site doc2doc where my comments did not trigger even 1 on the Richter Scale (not surprisingly). As one who has done “research” for several decades, I have long thought that articles in journals are often a sad reflector of the truth. Papers conform to a convention steeped in time although medical journals have only been commonplace for about one hundred years. We all know that often the co-authors do very little and get on the paper by dint of politics, power, diplomacy or just keeping the peace. How many professors become co- authors when they did stuff all. They join the hay ride so to swell their curriculum vitae or to lure precious grants, the life-blood of the “department”. Yes I have worked in such and have their recipe card in my head. I also believe that most research encounters significant logistic or methodological problems which are iced over on the final cake presented with “hundreds and thousands” (sugary coloured sprinkles) with a name/title on top for the unwitting and naive consumers like you and me. What I propose is a section called “Authors’ contribution” and “Methological and logistic problems”. I want researchers to bare all; the full Monty, no bull, the verum corpus. This does not have to be long; no magna carta. At present one guesses what was done and by whom by looking at their names and asterixes showing their rank and institution. It may be a lie; a deceit. On the submission of a paper one would be asked to sign a statutory declaration along with conflict of interest which incidentally is always in micrographic fonts invisible for the presbyopic reader. For example, most of the work may have been done by a registrar or even a data manager who is either not listed or is the third author (haven’t we all seen this?). The project may have been bogged down by problems that others wanting to corroborate the work need to know. Some bits may be "fudged", nay data "massaged". That is why Scot’s and Amundsen’s journals are still read by those wanting to go for a stroll across Antarctica. How many times have we heard some luminary expounding the virtue of some drug in a drug trial usually called something like the SMUG 2 trial, where the said drug "nearly reached significance". Editors would regard this with dismay as it would mean that articles may end up being half a page longer and we all know that words cost cash. However, we need to rise above this and demand a warts and all article where the authors can “fess up” as the former American president and literary genius used to say. As an aside, his lasting legacy I believe was the fiscal mutilation of both his and other nations, not to mention his mother tongue. These suggestions may sound revolutionary to those still eating brioche viz. cake but the empty tumbrels are waiting in the courtyard. Competing interests: None declared |
|||
|
|
|||
|
Trisha Greenhalgh, Professor of Primary Health Care University College London
Send response to journal:
|
The moral tone of Trish Groves’ editorial – research is publicly funded; fellow scientists and citizens have a “right” to access “raw numbers, analyses, facts, ideas, and images”; some naughty researchers are colluding with industry to “keep the data hidden away” – was inappropriate. An increasing number of journal editors are making naïve assumptions about the nature of research knowledge. A proportion of such knowledge (e.g. in basic sciences) can be legitimately be treated as “facts” or “raw numbers” that can be extracted from their context and analysed unproblematically by others at some future date. But the data behind most BMJ papers (broadly, health services research) is not so cleanly cleaved from the context in which it was collected or the people who supplied it and interpreted it. There is an important literature on the lack of transferability of knowledge between communities even in the basic sciences (1,2).The problem increases as the unit of analysis moves from the base pair to the healthcare organisation (3,4). Second, the delicate trust relationship between researchers and research participants is under threat. It is one thing to agree to give an interview (or a tissue sample) to a researcher on the assurance that the raw data will be carefully protected and destroyed once the analysis is complete. It is quite another thing to agree to the passing of one’s data to unknown others for unknown future purposes. Neither scientists nor the general public have a “right” to access these data just because the work was publicly funded. Research ethics committees typically require the destruction of data such as audiotapes within three years of collection. Third, I thought we had moved on from the GOBSAT (good old boys sat around a table) method of developing guidelines, especially when the good old boys and girls were all selected from the ‘ayes’ rather than the ‘nays’. BMJ editors have a responsibility to familiarise themselves with both sides of the argument before placing moral pressure on all of us to “share” research data. Just because the American medical journals and the geneticists are doing it doesn’t mean it’s a sensible course of action for the BMJ. (1) Knorr Cetina KD. Epistemic Cultures: How the Sciences Make Knowledge. Boston: Harvard University Press; 1999. (2) Latour B, Woolgar S. Laboratory life: the social construction of scientific facts. London: Sage; 1979. (3) Leonard-Barton D. Wellsprings of knowledge. Boston: Harvard Business School Press.; 1995. (4) Østerlund C, Carlile P. Relations in practice: Sorting through practice theories on knowledge sharing in complex organizations. The Information Society 2005; 21:91-107. Competing interests: I am a researcher who has ethical obligations to people who supplied data to me |
|||
|
|
|||
|
Sophie E Park, GP and Clinical Teaching Fellow in Primary Care UCL Medical School, Royal Free Campus, Rowland Hill Street, London, NW3 2PF
Send response to journal:
|
Making all BMJ research data available is an exciting and challenging ambition 1. It poses some interesting questions in the field of qualitative research, particularly relating to ethics and the questions ‘what is data’ and ‘to whom does it belong’? Quantitative and qualitative research have developed from very different ontological and epistemological perspectives 2. The value judgements we make about the validity and worth of a piece of research may vary considerably depending on its use of methodology 3. Within the positivist paradigm, originating from Descartes’ ‘I think therefore I am’, importance is attributed to objectivity 2. Qualitative methods generally value the individual nature of inquiry; interpretation; and crucially the ability of the researcher to be reflexive in their gathering and analysis of that data 4. In qualitative methods, data may include a vast array of representations, observations and experience including interviews, conversations, photographs and researcher reflective logs 5. Of great importance to the interpretation of that data is its context. Researchers working within this paradigm are searching to make meaning and interpretation from studying things in their ‘natural setting’ 5 within their individual context. The researcher’s interpretation and reflexive analysis are important contributors to that data. Ownership of data brings associated ethical responsibilities. The ethical considerations of the qualitative researcher are many. For example, during an interview, the researcher has a responsibility to clearly define their role 2 while exploring intimate and potentially unexpected and emotionally challenging areas for the participant 4. The greater the degree of rapport and trust between researcher and participant, the greater the potential depth and richness of the data produced. This relationship represents an ‘implicit contract’ 4 between the two parties requiring respect and confidentiality. Within, for example, narrative research participants might be invited to unveil their life stories before the researcher 6, revealing many unique identities and situations. A researcher who is embedded within the data will have an ethical obligation to anonymise wherever possible prior to publication 4 and, for example in ethnography, often involve participants in the review, construction and reporting of that data 2. We need to consider at what point within the process of analysis, interpretation and anonymisation, data is considered original and therefore appropriate for public viewing. Will participants be easily identified to those who know them, and if so might this have consequences for the individual involved and future recruitment to similar research? Once published, difficulties exist for the individual to withdraw that information and their original consent. How too will other researchers approaching that data resolve issues surrounding the original context and co-construction of that data 7? Also, who carries the responsibility previously contracted between researcher and participant regarding the use, interpretation and application of context to future analysis? These ethical responsibilities are great. The issues raised are not prohibitive, but need to be considered fully before data are shared. 1. Groves T. Managing research data for future use: The BMJ is now asking authors for data sharing statements. BMJ 2009;338:729-730. 2. Ritchie J, and Lewis, J., editor. Qualitative Research Practice: A Guide for Social Science Students and Researchers. London: SAGE Publications, 2003. 3. Greenhalgh T. How to read a paper: the basics of evidence based medicine. 2nd ed. London: BMJ Books, 2001. 4. Josselson R. The Ethical Attitude in Narrative Research: Principles and Practicalities. In: Clandinin J, editor. Handbook of Narrative Inquiry: Mapping a Methodology. London: Sage, 2007:537-566. 5. Denzin NK, Lincoln, Y.S., editor. Handbook of Qualitative Research. 2nd ed. Thousand Oaks, CA: Sage, 2000. 6. Bruner J. Acts of Meaning. Cambridge, Massachusetts: Harvard University Press, 1990. 7. Oakley A. Interviewing women - a contradiction in terms. In: Roberts H, editor. Doing Feminist Research. London: Routledge and Kegan Paul, 1981. Competing interests: I am studying for an EdD at the Institute of Education using qualitative methods. |
|||
|
|
|||
|
Michael D Gallagher, Research Fellow Centre for Population Health Sciences, University of Edinburgh, EH8 9DX, Allison Worth, Aziz Sheikh
Send response to journal:
|
In principle, we support the notion of data sharing as outlined by Trish Groves. (1) Yet as Trisha Greenhalgh’s (2) and Sophie Park’s (3) comments suggest, qualitative data raise particular difficulties for data sharing and secondary analysis (4,5). Both consent and anonymity raise ethical concerns for data sharing. Consent to share data could be sought at the outset of research, but consent can be compromised by various contextual factors (6), and in any case its viability over archival timescales is debateable. As Park notes, implicit trust between researchers and participants often plays a significant role here. The raw outputs of qualitative research are also not easily anonymised (7). Identifiers can be removed, but data are often sufficiently detailed and personal to enable identification by those who know the participants. Partial data sharing – archiving a selection of carefully edited data – could go some way to addressing these issues. However, the resulting incompleteness of the data sets would only exacerbate the difficulties for secondary analysis noted by Greenhalgh and Park. We do not think that these problems are wholly insurmountable. Indeed, contrary to Park, it could be argued that a postmodern perspective suggests a world of multiplicity, fluidity and flows more suited to the ideologies of open source and creative commons than notions of data ownership. As some possible ways forward, we therefore suggest the following: • Provisions for archiving should be built in at the proposal stage and consent (notwithstanding its inherent flaws) should be designed around these provisions. There may be situations where any form of archiving could be considered unethical, and this should be acknowledged. • Secondary analysis should where possible be undertaken collaboratively (5), in close liason with those who collected the data, to enable a degree of re-contextualisation. • Partial sharing could be used to provide other researchers with a ‘flavour’ of a study’s raw data. This would enable other researchers to decide whether secondary analysis of a fuller data set might be fruitful in their own work. In such cases, contact could be made, collaborative relationships developed and data shared more fully. 1. Groves T. Managing research data for future use: The BMJ is now asking authors for data sharing statements. BMJ 2009;338:b1252 2. Greenhalgh, T. BMJ opens can of worms. BMJ 2009;338:b1252 3. Park, SE. Ethical considerations to data sharing within a postmodernist paradigm. BMJ 2009;338:b1252 4. Parry O and Mauthner NS. Whose Data are They Anyway?: Practical, Legal and Ethical Issues in Archiving Qualitative Research Data. Sociology 2004; 38:139-152 5. Van den Berg, H. Reanalyzing Qualitative Interviews From Different Angles: The Risk of Decontextualization and Other Problems of Sharing Qualitative Data. Forum: Qualitative Social Research 2005; 6(1) Art. 30. Available at: http://www.qualitative-research.net/fqs-texte/1-05/05-1-30- e.htm [Date of access: April 2, 2009]. 6. Gallagher M. Ethics, in E.K.M. Tisdall, J. Davis and M. Gallagher, Researching with children and young people 2009. London : Sage 7. Van Den Hoonaard WC. Is anonymity an artifact in ethnographic research? Journal of Academic Ethics 2003;1:141–151 Competing interests: All the authors are researchers who have an interest in qualitative health research. |
|||
|
|
|||
|
Neil McIntosh, Emeritus Professor of Child Life and Health University of Edinburgh EH9 1UW, Elaine M Boyle
Send response to journal:
|
As researchers working with parents and children, it has always been of major importance to us that both understand why they are being asked to be partners in a research project. Like others, we have been confused over recent years as to why information sheets and consent forms have become longer and more complex. We believe this has been driven by purist ethicists who may know little about the practicalities of dealing with patients in a doctor-patient relationship, and by litigation-conscious health service management with limited understanding of research. For parents and children, long and involved information sheets and consent forms have been shown to be counter-productive for obtaining informed consent. The more information that is given, the less is likely to be understood or retained (Pediatrics 2009. In press) The intention, reported by Groves (March 28th), of making properly anonymised raw data from trials more widely available once the main papers have been published is welcomed, and it has been a long time coming. But is it really necessary to make the patients aware that this is happening by a tick box on each consent form? This is anonymised data: why should they want or need to prevent it? Our concerns may best be illustrated by an example: At present, parents being asked to allow their newborn infant to participate in the MRC funded UK BOOST II Trial are confronted by a wordy eight page leaflet. Sadly, in conforming to governance requirements, the reasons for the trial and the essence of the research are swamped by details of what parents should do if something goes wrong and by issues of ongoing data confidentiality. Such information goes over the heads of many parents, particularly when they have to consider this within a few hours of an extremely premature birth; and this leaflet is for a study simply comparing two marginally different doses of oxygen in the newborn! Greenhalgh (BMJ April 18 2009) raises concerns about passing of “data to unknown others for unknown future purposes” and indeed, the details of how data can and should be shared appropriately remain open for discussion. Yet the judicious practice of making anonymised data available for scrutiny after completion of a randomized controlled trial must surely be ethically correct. Is it not secrecy that is ethically dubious, not the free availability of anonymous data to others in the research community? To "obtain explicit consent for publication of suitably anonymised raw data as part of patient recruitment procedures" by adding yet another 'box' on the consent form is, we believe, both unnecessary and likely to confuse patient understanding of the primary reason for the research. Competing interests: None declared |
|||
|
|
|||
|
M Felix Freshwater, Voluntary Professor of Surgery University of Miami 9100 S Dadeland Blvd Miami FL 33156-7815
Send response to journal:
|
I applaud Dr. Groves’ plea for “more transparency about the quality of research”, but there should be not only data transparency, but also financial transparency. Most major medical journals have made financial disclosure mandatory. Yet, now that the Internet allows free access to any biomedical abstract, it is possible for an abstract's reader to be blinded to its paper’s relevant financial disclosure unless the reader has a paid subscription to that journal. As Stanton wrote in the BMJ: “We all know that students, journalists, or others should read the full article, but the fact is that many people only read the abstract of many papers. Conflict of interest should always be included—including in an abstract.(Stanton R. Put declaration in abstracts. BMJ 2008; 336:345.) I reviewed the ICMJE Uniform Requirements, the author instructions for 20 journals including BMJ, JAMA, the nine current Archives, NEJM, and the Lancet journals and found that none of them required, recommended or even mentioned financial disclosure for either structured or unstructured abstracts. Thus, the reader might have the false impression that there was no potential financial conflict of interest when one exists. This is particularly important when the reader is a layperson attempting self- education on the Internet. A recent example highlights the problem. The website for the Brava® external tissue expander prominently states: “Read about a NEW STUDY IN A LEADING MEDICAL JOURNAL CONFIRMS AND EXCEEDS BRAVA'S ORIGINAL STUDY RESULTS”.(http://www.mybrava.com/home.asp) The underlined phrase links to an abstract on the website of the peer-reviewed journal Plastic and Reconstructive Surgery.(Schlenz I, Kaider A. The Brava external tissue expander: is breast enlargement without surgery a reality? Plast Reconstr Surg 2007;120:1680- 9. ) The abstract concludes that: “Long-term breast enlargement without surgery is possible with an external tissue expander". Thus the reader looking solely at the commercial website and abstract would be impressed with the fact that peer reviewed paper had found merit to the commercial device. Although the abstract lacks financial disclosure, the cited paper reveals that the senior author had a financial relationship with the parent company and patent holder of Brava's manufacturer. This lack of disclosure in the abstract misleads the reader. Of course, a reader could attempt to access a paper by paying a fee. The possibility exists that the lack of financial disclosure is itself a financial conflict. One time access to NEJM costs $10, one time access to JAMA costs $15 and one time access to Plastic and Reconstructive Surgery costs $30. Thus, it appears that many peer-reviewed journals have a potential financial conflict by not providing free access to the financial disclosure portion of the paper. BMJ should lead the way among major medical journals by mandating that abstracts have financial disclosure. Requiring a yes/no financial conflict statement in the structured abstract and allowing a free preview of the financial disclosure part of the paper online will give the abstract reader valid information. Based upon their access to financial disclosures, online readers will be able to make more reasoned decisions about the validity of abstracts’ conclusions. Competing interests: None declared |
|||
|
|
|||
|
Trish Groves, deputy editor BMJ
Send response to journal:
|
M Felix Freshwater calls for abstracts of research articles to include declarations of competing interests. The BMJ's new way of abridging original research articles for the print BMJ is essentially an extended abstract. It's called BMJ pico and it gives the research question, study design, and findings, along with details of funding and competing interests. [1] Furthermore, all BMJ research articles are published on bmj.com with open access, full text, no word limits, and full declarations of competing interests along with several other statements regarding transparency [2] 1. Groves T, Godlee F. Innovations in publishing BMJ research. BMJ 2008;337:a3123 2. Resources for authors: Research. http://resources.bmj.com/bmj/authors/types-of-article/research Competing interests: I'm the BMJ's senior research editor |
|||