BMJ Ethics Committee Meeting, 7 October 2003

Present: Richard Smith (acting Chair)

Liz Wager

Ann Sommerville

Rachel Fetches

Peter Singer (via phone)

Jocalyn Clark

Giulio Bognolo

Apologies: Tom Wilkie

Derick Wade

Jeff Tobias

The Committee accepted the minutes of the last meeting with some minor amendments from LW.

RS updated the committee on the search for finding a new Chair. There had been four serious applicants, one of which was a current member of the committee. In addition there had been two additional queries from individuals who where interested in becoming members. The interviewing panel would consist of Richard Smith, Derick Wade, and Rachel Fetches. The committee discussed expanding the committee membership once the Chair had been appointed. AS noted that the BMA Ethics Committee has 25 members but not all of them always attend the meetings. The committee discussed that that would probably be too large a number of members but that the committee might benefit from slightly increasing its membership to allow for the fact that not all members could always attend.

RS noted that it was an outstanding action for the committee to write an account of its activities since its inception. RS felt it would be good to have such an account signed by all members. Sandy McCall-Smith had been intending to draft this but had obviously been too busy to manage this. LW offered to draft a piece and circulate it to the members. RS felt that an account of the committee’s activities would work best in the education and debate format which was 1500–2000 words long and could be slightly longer online. It was agreed that LW would put together a draft but that the committee would wait until the Chair had been appointed before publishing the article.

Background: BMJ had strong suspicions that a paper, which was a follow on to a paper published in the 1990s, was fabricated. A university investigation has found no evidence. The consensus in the committee was that there was inadequate reassurance that a proper investigation has been conducted, and there remains a significant possibility that the whole paper might be entirely fraudulent. RS had written a follow up letter to the University authorities expressing the concern of the committee that the raw data had not been examined during the investigation. The author had subsequently left the university’s employment without providing the university access to his/her raw data and records. The author had published the article under investigation in the BMJ in another journal. Also, the author had published another article in the other journal which published the original article. Following advice from the ethics committee the editor contacted the editors of the journal that published the original article in the early 1990s and also the journal which had published the article which was being investigated by the BMJ.

RS updated the committee on the case. The editor of the journal which had published the author’s December 2002 article wanted to publish a critical piece about the author in April 2003 but had been held up by legal concerns. He does plan on publishing the article soon. The other journal was also planning on publishing an article on the article (which was virtually identical to the one the BMJ had serious doubts about) they had published. RS had not heard back from the university but would chase the Vice-Chancellor’s office again. LW queried what action RS would take if the university did not take any further action? RS felt that he should pursue it further. PS suggested going to the country’s association of universities as a higher authority. PS also queried whether the author was registered with a national regulatory body? With academics there may also be recourse to the relevant national College.

Background: An author who has several published papers in a particular area submitted a number of new papers which gave cause for concern due to a lack of reference to the earlier published papers, overlap and similarity to published papers, and questions about the reliability of data. The authors were asked to clarify these issues. The authors then submitted further papers that were rejected with suspicions on the validity of the data. Unsolicited revisions were then submitted. Two separate groups raised concerns of the validity of the authors' work. A further paper was submitted and the editor requested the raw data on all 3 papers. The authors submitted the raw data in a format that was difficult to examine. Two of the papers under peer review were also submitted to other journals; reviewers alert the journals to this fact. Two further papers are submitted and one of the clinical referees raises the issue of potential fraudulent work. A statistician compiles a full report concluding that there is evidence that the data are fraudulent. After further papers are submitted the authors demand a decision on the outstanding papers. The editor rejects the papers and informs the authors that a full report will be forthcoming. The allegations of fraud were relayed to the authors. The authors' claim that the data has been destroyed by natural causes, blinding in the study was impossible, and the last available data for patients who died or were lost to follow up were used. The authors deny the claim that the data were generated to fit the values in the tables. In further correspondence, the authors claim that the persons raising concerns have a personal axe to grind and are not credible. The editor contacts the country's human rights organisation and they in turn forward the editor's complaint to the country's national medical research council. The human rights organisation's report concludes that despite errors and inaccuracies, the lack of information about the data source collection makes it difficult to comment on the authors' work. Despite a follow up letter, no further communication was received from the authors' national medical research council. The original published article has become a citation classic.

RS updated the committee that he had commissioned a freelance journalist to write up the extensive story on this author and his/her papers. The journal felt it had amassed a large body of evidence to show that the author’s work was seriously flawed and hoped to publish this in 2004. In addition, another medical editor had informed RS of the progress with another recently published paper of the author’s, where fraud was suspected. The author’s co-author was refusing to admit that the paper might be fraudulent. The co-author had received all the data from the author (working in another country) and had done the analysis in his home country. The co-author went to visit the author after the allegations were made and claims to believe the author’s assertion that the data were eaten by termites (as was the case in all the papers the BMJ questioned in the last 10 years). The freelance journalist had received a response from the author who threatened to sue for a large sum of money.

PS queried what course of action editors had where people threatened to sue. Where there guidelines or specific steps editors could/should take in cases of suspected fraud? RS noted that there weren’t any specific guidelines and that he often received letters from other editors on the issue. PS felt that it would be useful for editors to have a set of guidelines for such cases. This would help editors use good practice in such cases, hopefully expediting the process and additionally could provide some measure of legal protection through adhering to a recognised standard. RS noted that this could potentially be addressed through COPE or WAME perhaps through a code of practice for editors. PS felt that it would be quite useful to put in suggested timelines like three months for responses from institutions/organisations before chasing again or escalating to a higher authority. This might help avoid decade long correspondence with retrenchments and threats of legal action. RS agreed and noted that many other editors are "frightened off" from pursuing further investigation and publication of notices. Additionally, the process can be extremely time consuming and may take up significant resources.

RS updated the committee that he had written up the attempt to bribe him in his editor’s choice but could not locate the paper so could not contact the authors of the paper as suggested by the committee.

PS had circulated the article on the revision of the consent policy and invited any responses over the following two weeks. If not the piece would be submitted for publication.

RS asked the committee for any thoughts or lessons that could be drawn from the issues surrounding the obituary of Dr David Horrobin (BMJ 2003;327:229) and the PCC’s ruling (http://www.pcc.org.uk/reports/details.asp?id=354).

LW queried what in particular RS wanted comments on? RS stated that (1) he believed the contents of the obituary to be true; (2) the obituary should have contained more evidence to back up assertions; (3) usually the BMJ is pretty careful about the main obituary which is written by journalists. Generally the journalists don’t know the individually personally but here the journalist did know Dr David Horrobin.

PS suggested that the main issues had been how the family had been dealt with at the time of publishing. PS wondered if a letter to the family as a notice of the forthcoming publication would help? It would always be a tough position to be at the intersection between publishing a critical obituary and that person’s family. RS said that he would be reluctant to send the article in advance. Part of the problem had been the speed of publication after Dr Horrobin’s death, though the PCC felt that this was less of a problem. AS felt that a condolence letter after publication might be viewed as hypocritical. AS noted that in reading the obituary she had felt that it was more opinion than a critical piece and that it was written by someone who did not like Dr Horrobin. Overall the piece did not come over as balanced. AS also noted that, in her capacity as Head of BMA Ethics, she had received complaints after publication from people who knew Dr Horrobin. LW agreed that a letter would raise false expectations of being able to change an article. It may have been a problem in this instance that the journalist knew Dr Horrobin. Additionally, it was difficult to disentangle who said what. LW suggested that in future it would perhaps be better to avoid people who knew the person being written about as this can affect objectivity. Additionally, the fact that the BMJ got some important facts wrong undermined the BMJ’s case. RS agreed that for negative reviews it was much more critical to double check facts. PS noted that the PCC had focused much more on the communication with the family. PS noted a particular example in the correspondence where a family member had been reduced to tears. RS stated that the obituary had not been published to deliberately upset the family. Initially he had felt that it would be wrong to apologise for the article but had been subsequently convinced that the BMJ could apologise for the distress caused without apologising for the evidence. PS suggested perhaps a box acknowledging that the article may cause distress to the family. LW suggested that in future the BMJ could publish "for and against" obituaries. AS felt that the purpose of an obituary was to set a marker of someone outstanding in some way, and that this particular obituary was too opinionated. RS disagreed somewhat in that he did not feel that an obituary was a definitive statement of someone’s life, and that some obituaries would be more valid than others. LW wondered if because the journal was saying something about a person who could not respond because they were deceased if the "right of reply" could be extended by analogy to the deceased’s family? RS noted that the libel laws recognise death as an end of a person’s right of action so this analogy did not work. RS felt that in retrospect the main problem with the article had been that it should have been much more based on the evidence available. RS queried whether the BMJ should follow up the obituary with more of the evidence? AS & LW both felt that this would look like a very defensive action. AS differentiated this case from current ongoing cases where the editor has a duty to pursue the issue to stop individuals from perpetrating further frauds. AS felt that RS should let the issue drop. AS felt that there was a duty of care to the readership to get the facts right in an obituary but that there was no duty of care to the family.

JC noted that in publishing a news item there was no duty of care to individuals who were written about. JC also pointed out that Caroline Richmond who had written the piece had explicitly acknowledged the fact that she had known Dr Horrobin. JC also felt that Caroline Richmond had been vilified and had not been given the opportunity to respond to comments. RS noted that Caroline Richmond intended to write a response but was currently ill. GB noted that in other European countries there is a tendency to sanctify a deceased person and negative elements of a person’s life were often ignored. LW agreed that often newspapers use code words rather than openly criticise a person. REF felt that the family had been more inflamed after their initial complaint by a slightly flippant response. RS agreed that communication with the family could be handled better in future. PS suggested discussing the use and manner of obituaries in the next BMJ theme issue on the concept of a good death.

LW declared an interest as she had previously worked for a pharmaceutical company and had also written Good Publication Practice for Pharmaceutical Companies. PS declared an interest in that he knew one of the authors.

RS informed the committee that the case had been discussed extensively with COPE, the journal’s co-owners, and the Publishing Group executive.

A BMJ specialist journal had published a paper and had subsequently been alerted to problems when a letter to the editor highlighted the fact that there was another published account of one of the trials. In addition to the overlapping publication, a national drug regulatory website showed negative endpoints for more clinically important outcomes. When the paper had been published the journal had fulfilled a large order for reprints, however, once the problems were highlighted the publishing group froze a second, substantial, reprint order. There were three options discussed: (1) do nothing—this has been dismissed at all discussion forums; (2) publish a notice of duplicate publication—this has been dismissed as an inadequate response; (3) ask the institution to investigate—this is difficult as authors are from different institutions and some are employees of the drug company; (4) to tell the factual story in the journal and ask the various players to comment on the facts. The journal decided to follow the final option and have asked a senior independent party to oversee the gathering of all the information, put forward a version of the story, collate the various parties’ responses, and reach a conclusion.

RS asked the committee to comment on the financial issue facing the BMJPG in that thousands of pounds worth of reprints had already been sold prior to the problems being known and there is currently a reprint order worth thousands of pounds which has been suspended by the publishing group pending the outcome of the issue. How should the publishing group approach the issue of reprints? PS enquired about the relationship between the publishing group and the co-owning society? RS indicated that both the co-owning society and the publishing group derive income from the journal’s profits.

LW agreed with the previous discussions at COPE that sending back the money from the initial reprints would achieve little. The journal has incurred costs in producing the reprints and it would be unduly hard on the publishing group to not recover the cost of this. It was an option to donate the profit to charity if the conclusion of the report was that there was an intention to mislead and deceive. AS suggested that if the second reprint order was fulfilled inserting a box with a factual statement detailing the issues with the paper. LW felt that the journal had a right to refuse to fulfil the second reprint order if it was shown that the author/company had engaged in selective publication. Selective publication is a very poor practice and editors need to take a clear stance against such practice. LW also queried whether the article would be retracted as publishing the facts about the paper may not be linked on Medline. RS felt that this was certainly something to consider after the issue had been debated in the public domain.

RS also informed the committee that the BMJ and The Lancet were planning on holding a one day conference on the relationship between science and the pharmaceutical industry and planned to use this case as the central case study for discussion. LW noted that there would be someone responsible for the development and running of the trial and that they should be canvassed for a view. Additionally the Association of the British Pharmaceutical Industry (ABPI) could be asked for a view. LW also suggested that there could be link to the Good Publication Practice for Pharmaceutical Companies. AS felt that before deciding what to do with the money already received for the first set of reprints, the publishing group should wait for the outcome of the investigation. PS noted that the articles should all be open access and the original article should also be opened up for publication as well. Additionally, if the publishing group were to fulfil the reprint order they should only supply it with the whole later discussion.

A news article had appeared in the BMJ which played on the name of a hospital to suggest that the hospital was a failing hospital. The Chairman and Medical Director wrote a letter of complaint to the journal pointing out the hospital’s improving record and the fact that the headline denigrated the efforts of the hospital’s staff.

The editor had written to the Chairman and Medical Director offering an unqualified apology and noting that the letter had been discussed with the staff member involved in writing the headline. The committee endorsed RS’s letter to the Chairman and Medical Director of the hospital as the appropriate course of action. RS informed the committee that the editor involved has admitted that the headline was a foolish one. In future the BMJ would be more careful.

A paper described an incident where a doctor had been discriminated against in an interview situation in favour of an internal candidate in the author’s institution. The doctor in question was unaware of the situation until the author of the article had notified the doctor. The doctor wanted to complain to the appropriate national medical body but the author of the article had declined to write a detailed statement in support of the doctor. The doctor wrote to the editor about the author’s actions indicating that he felt the doctor was unethical and that the article should be pulled.

RS asked the committee for their opinion on whether his response to the complainant physician was reasonable or not? PS felt that the response was entirely appropriate as the author did not name the other doctor who may or may not be identifiable from the piece.

RS informed the committee that a serious case of plagiarism of a BMJ paper had been drawn to his attention and he had asked a BMJ student to go through the two papers and identify the degree of overlap between the BMJ paper and the later paper. When questioned, the authors of the second paper claimed that there was a trivial level of overlap. Their excuse was that they had used the original paper in the writing up of the grant proposal and subsequently based the paper on the grant application. RS had contacted the editor of the second journal as well as the authors’ university and copied in the authors. RS had discussed the matter in confidence with a university Dean who indicated that universities treat plagiarism very seriously. Though this was a serious case it was probably more foolishness rather than maliciousness.

LW queried whether the authors were native English speakers as there was a greater likelihood for authors for whom English was a second language to lift chunks of other article. JC felt that this was not an excuse as even paraphrased ideas should be attributed to their original source. AS noted that the authors should have acknowledged the original source in the grant application as well which would have meant there were no problems later. GB felt that having English as a second language was not a justification and asked how senior the authors were. RS indicated that the authors seemed quite junior. RS was worried that there would be no proper investigation and that the junior authors would be severely dealt with and anyone more senior would escape censure.

JC tabled a summary of the events in this case. After provisional acceptance pending minor revisions and a request for information on funding and ethics information a revised version was received. The revised version did not have the requested funding and ethics information and it was sent back with an explicit request for this information. There were concerns about how the records were chosen and if the patient’s details were identifiable. JC asked the authors if the patients and doctors had consented to their letters being studied and if research ethics approval had been sought. The authors wrote back to say that no particular ethical issues were raised and that the study had been audit. After further correspondence the lead author provided a letter to the Chairman of the research ethics committee in the two study sites. JC felt it was inadequate to send a letter, which called the study a "personal project" and did not provide detailed information. In the letter the author implies that the project is not part of the audit programme but has now recast the work as audit.

PS felt that JC was right to raise the question and to pursue the matter. The case was in the grey zone and what was needed was confirmation from the two research ethics committee Chairmen that this research was audit and didn’t require review. LW queried what the position would be if the approval was given verbally? PS felt that what was needed was a letter from the Chairmen confirming that this was the case. PS suggested writing directly to the Chairmen. RS agreed that this was a grey area but that this particular piece did look like research. PS noted that without national guidelines it is hard to argue with any research ethics committee’s decision. AS felt that where an author was doing research using identifiable information on patients there was a clear need to go to the local research ethics committee. If, as the researchers suggest, they were using anonymised data to draw out general points it is still unclear who anonymised the data—the junior doctors? Generally, the BMA advises researchers to go the research ethics committee Chair. Approval could be via Chairman’s action or it could be a verbal reply. PS noted that in such grey areas there is often variability between research ethics committee’s decisions.

RS queried when the committee felt that ethics committee approval should be sought? Obviously it is important for research, however, with audit there is a case for sometimes getting approval and sometimes not. Factors in deciding whether or not to audit should be reviewed by a research ethics committee were: the level of automation, the nature of the data source, and the level of patient detail. AS noted that the purpose of audit was to identify best practice by measuring ordinary practice against the gold standard. This isn’t problematic unless there are identifiable patients. JC stated that the letters in this case were not anonymised before being provided to the content analysers and patients did not give their consent for this. RS felt that this case was research as the authors had gone beyond their own practice catchment. RS felt that just because the research ethics committee approved the study does not always mean that it was in fact correct. AS felt that it would be difficult to state that the research ethics committee is wrong. PS stated that this was a similar debate to the line between research and innovation which is difficult to pin down properly.

An author wrote to the editor of the BMJ to ask his advice on a paper which had been accepted for publication in another journal. The authors felt that the study described had been an audit, however, the journal disagreed and felt it was research. The authors had contacted their local research ethics committee who had stated that they too felt it was a research study. The local research ethics committee declined to give retrospective approval for the study.

PS felt that this was a similar issue as to Case 03/010 and that the authors had been unlucky with the decision of the Chairman of the research ethics committee. RS felt that this case was a clearer cut example of audit: the paper detailed something that ought to be part of routine practice and the authors had not involved people outside of their practice. PS queried if the nature of the two data sources were really that different as both the patient referral letters and the patient database contained relevant details about patients. AS felt that it depended on where the information was going—audit was important to people involved in the work being audited and in this case had not gone beyond the individuals involved in the unit’s work. RS queried how can a health care provider decide the appropriate balance? Units carry out audit when they are getting poor results and they may try to modify their practice to make their results better. AS noted that audit involves making changes to achieve an established gold standard; making a change to determine what the standard should be is research. Groups do research to establish best practice; audit ensures units achieve the established best practice. In an audit there the people looking at the data are already entitled to access to that data through their own work. This differs to sending information to someone who would not normally have access to that information. RS felt that there was a distinction where data were anonymised and as this caused no harm. AS felt that if there was any potential for harm to patients then the people looking at the data should contact their research ethics committee for an opinion on whether the review of the data needed ethics approval. RS summarised that this work did seem to be an example of audit as these authors were analysing their own work data.

RS felt that in Case 03/010 the authors had gone out beyond their own practice and it had moved to being research. AS felt that if the two local research ethics committees knew about it then this was acceptable; it doesn’t seem that anyone would be hurt and is a borderline case of audit and research. JC felt that the authors had treated the article as research as the research tools and aim of the study were not to implement change at a local level as per audit. AS felt that PS was correct to state that this was a grey area and it would be best to get confirmation from the Chairman. PS felt that a verbal okay would be acceptable so long as it was confirmed in writing to the editorial staff. PS felt that the overall goal of obtaining research approval was to protect patient data and to protect them from harm. PS stated that there was a need for the BMJ to have a clear policy on this issue. RS summarised that the point to clarify is: when should ethics committee approval for audit be required? RS commented that AS has some guidelines from the BMA’s work in this area and additionally there are Canadian guidelines. RS asked REF to collate the available information in this area and to highlight areas of agreement and disagreement in this area.

An editor had asked the committee to consider whether it was acceptable to publish an article about adverse drug reactions without patient consent. The editor felt that there was a strong case for the publication without consent because of the overriding public interest and benefit from the publication of the article.

The committee agreed with the editor’s assertion that such cases could be published without consent because of the overriding public interest in publishing these cases. Though the committee felt that where possible the editors should always seek to get consent.

An editor asked for the committee’s view on the publication of photographs of children who had been the victims of child abuse in a book on the topic. Normally parents give consent for the publication of any photograph of their children but this presented a problem where the parents themselves had been involved in the abuse of the child. In earlier editions pictures had been used without consent because of this inherent difficulty in obtaining consent. At the time, the thinking was that bands across the eyes were sufficient to anonymise the photos. It was now clear that this is an inadequate protection of a person’s identity and there is a danger of someone recognising the child or, indeed, an adult recognising pictures of him or herself as a child.

AS noted that illustrations rather than photographs could be used to demonstrate the point at hand. RS felt that in the current environment the BMJ would either need consent or can’t use. Patients may well recognise even a body part. PS suggested that the publication could use photos already be in the public domain, for example, a press photo. AS agreed and suggested that the editors could use material provided by the NSPCC. REF queried whether in fact a patient would be able to identify a body part if it were used to illustrate a general point and not used in the context of describing a patient’s story. RS thought there was still a potential for identification. AS thought that in such a scenario it may be possible to completely anonymise the photo. RS summarised the discussion in that generally the editors couldn’t use photos without consent with the proviso that they could if they were very confident that it was completely unidentifiable.

RS updated the committee that the British government had withdrawn funding for the claimants and so he was not required to give evidence on peer review. RS thanked the committee for their comments via email and he had decided that he would have given expert evidence on peer review and would have declared it in the BMJ.

No further information has been received on the papers on consent for genetic pedigrees. Ask TW if he had been able to contact Angus Clarke/Peter Harper about this issue.

PS noted that with the busy schedules of the committee members discussion papers stemming from ideas often take some time to be put forward. PS wondered if perhaps the committee should pursue an alternative model such as funding an ethics fellow? RS thought that it would be a financial challenge to hire a fellow. AS suggested that the committee may be able to involve some of the BMA ethics staff in research. RS also suggested contacting the ethics programmes at Imperial College London and Kings College London.