Design a site like this with
Get started

Correspondence with Circulation about retraction of the MIST Trial publications

Below is the letter sent previously to Dr Hill, the editor-in-chief of Circulation, requesting retraction of both versions of the MIST Trial paper.

The response of the American Heart Association’s senior attorney said:

Dr. Wilmshurst

The American Heart Association (“AHA”) received the attached correspondence requesting a retraction of the article, Migraine Intervention With STARFlex Technology (MIST) Trial (“the MIST Trial”), which was published in the AHA’s Circulation journal on March 18, 2008, as well as a correction to the article published in September 2009.

AHA respectfully declines any further involvement. We consider this matter closed and we will not pursue additional comment or review.

Thank you,

Ms. Thuy P. Huyn

I leave it to readers to judge whether the matter should be closed after they have read my letter to Dr Hill, the editor-in-chief:

Dear Dr Hill

I request that Circulation retract Dowson A, Mullen MJ, Peatfield R, et al. Migraine intervention with STARFlex Technology (MIST) Trial: a prospective, multicenter, double-blind, sham-controlled trial to evaluate the effectiveness of patent foramen ovale closure with STARFlex septal repair implant to resolve refractory migraine headache. 2008;117:1397-1404. Circulation published an amended version, correction and data supplement in 2009. I request that both of the published versions of the paper (2008 and 2009), the correction and the data supplement all be retracted because they are all false.

The 2009 correction was published because, a few days after Circulation published the paper in March 2008, I wrote to Dr Lozcalzo, the Editor of Circulation at that time. I said that I was the MIST Trial’s principal cardiologist. I sent to Circulation documents which showed that Dr Simon Nightingale (another member of the trial steering committee) and I had informed the three other steering committee members (Dr Dowson, Dr Mullen and Dr Muir) that we refused to be authors because NMT Medical, the MIST Trial’s sponsor, had refused to allow the clinicians to see the complete trial dataset but, even without access to the full data, it was clear that the published data were false.

I provided documentary evidence that concerns about the original 2008 version of the paper included misrepresentation of the methods, falsification of data, concealment of adverse events, inappropriate authorship, failure to properly acknowledge the work of contributors and failure to disclose financial conflicts of interest.

I told Circulation that in March 2004 the Northern and Yorkshire Region Multicentre Research Ethics Committee (MREC) had removed Dr Dowson from his role as the principal investigator in an unrelated multicentre migraine trial sponsored by Allergan, because he had entered false data for vital signs (such as blood pressure and pulse) on clinical record forms of trial patients that he had not seen and signed the forms to certify that they were a true record. That MREC said that Dr Dowson was not fit to be a trial principal investigator and they reported Dr Dowson to the General Medical Council (GMC). The GMC is the statutory body that licences doctors to practise in the UK.

I told Circulation that in 2006 a Fitness to Practise Tribunal of the GMC found Dr Dowson guilty of misconduct in the Allergan trial and placed conditions on his practise as a doctor. This evidence of Dr Dowson’s misconduct is a matter of public record. It is available on the GMC website now, as it was in 2008 when I informed Circulation.

Even before publication of the 2008 version of the MIST Trial paper, Circulation should have been alerted to the possibility of misconduct because the journal had been sent four drafts of the paper. As Circulation admitted on 18 April 2008, Dr Nightingale and I were named as authors on only the second draft of the paper. Some of those named as authors of the 2008 paper saw only a draft that named Dr Nightingale and me as co-authors. They later said that if they had known that Dr Nightingale and I had refused to be authors of the paper, they also would have refused to be co-authors.

Circulation published a new version of the paper, a long correction and a data supplement in 2009. Dr Nightingale and I would have told Circulation that they are also false, but Circulation refused to allow us to see them before publication. Much of the dishonesty in those publications should have been apparent from the documents that I had sent previously to Circulation after the 2008 version was published. In addition I had warned Circulation on 15 February 2009 that Dr Dowson could not be trusted to produce a correction that was honest and accurate. I also warned Circulation of the involvement of a vice president of NMT in writing the paper, as described below.

I asked how much NMT had paid Circulation for reprints of the MIST Trial paper and another paper reporting a different trial of one of NMT’s devices, the BEST Trial (Mullen MJ et al. Circulation 2006;114:1962-7), in which I was involved but not an author. Circulation did not answer those questions.

Karen Barry, the Managing Editor of Circulation, sent me an email on 18 May 2009 that said “All documents (which I had sent to Circulation) have been forwarded to the American Heart Association (AHA) and the Scientific Publishing Committee”. On 15 June 2009, Karen Barry sent me an email that said “The journal is still working with the AHA and Dr Dowson on the text of the correction.” Because the correction is false, as I will describe below, and because the documents, which Circulation and AHA possessed at the time of publication of the 2009 version, showed it was false, it follows that Circulation and AHA must share responsibility for publishing a fraudulent correction.

After many subsequent emails from me, on 20 September 2012, Rose Marie Robertson, Chief Scientific Officer of AHA, stated that the AHA had responded to questions from the GMC in March and April 2011. She said that the AHA would review the findings of the GMC when they were available.

On 21 January 2016, I wrote again to Karen Barry to inform Circulation that a further Fitness to Practise Tribunal lasting three months had found that Dr Dowson had committed multiple counts of misconduct including dishonesty in the MIST Trial. Dr Dowson appealed against the decision to the High Court of England and Wales, but in late 2015 his appeal had been rejected. The Court confirmed that Dr Dowson had been dishonest in the MIST Trial. In his judgement rejecting Dr Dowson’s appeal, Mr Justice Edis wrote “He (Dowson) now created a false document ……. which was part of the records of the MIST Trial.” As a result, Dr Dowson was suspended from medical practise in the UK.

The legal findings confirmed much of the misconduct I had notified to Circulation in 2008. Yet even after the GMC and High Court findings, Circulation has continued to standby the false 2009 publications in Circulation written by a doctor with a record of repeated misconduct in research aided by the vice president of the sponsoring corporation.

In December 2020 I discovered that Circulation had removed from its website the correction and data supplement, leaving only the false 2009 version of the paper. Removal of these documents had the effect of concealing concerns about the paper. Circulation should now retract all these false publications.

My claims can be verified from documents that I sent previously to Circulation and from publicly available legal documents, including the findings of the High Court of England and Wales and the findings and transcripts of the Fitness to Practise Tribunals. (An Act of Parliament governs Fitness to Practise Tribunals for doctors and they have the same legal powers as the High Court.) There are additional documents and emails that support my claims of misconduct.


I will provide some additional background for the benefit of those that I am copying this too, because they may not know the full facts, as Circulation does.

The MIST Trial was sponsored by NMT Medical, a Boston-based corporation that manufactured and marketed the STARFlex device used in the MIST trial under licence from the patent holder, Boston Children’s Hospital.

In late 2006, when Dr Nightingale and I raised concerns about the failure of NMT to provide the steering committee with the full data, NMT dissolved the steering committee and replaced it with a writing committee consisting of the remaining three members of the steering committee (Dr Dowson, Dr Mullen and Dr Muir), two of whom were NMT shareholders, plus an additional investigator, Dr Peatfield. Because I was named in the protocol as the trial’s principal cardiologist, it is clear that my removal from the steering committee would require a protocol amendment, and one was never obtained. In fact the West Midlands MREC, which approved the MIST trial, specifically instructed NMT that they could not exclude Dr Nightingale and me from the trial. NMT ignored the MREC’s instruction. The issues were taken up by the National Research Ethics Service (NRES), which has oversight of the Regional MRECs, and NMT threatened senior officials of NRES with legal action against them personally.

Around that time, the Chair of the West Midlands MREC reported Dr Dowson to the GMC because of concerns that Dr Dowson was conspiring with NMT to misrepresent the trial data.

NMT brought various libel actions against me between 2007 and 2011. NMT also instructed their lawyers to sue Dr Nightingale, but did not serve the claim on him. NMT’s legal actions started six days after Dr Nightingale and I said that we were not prepared to be authors of the paper in Circulation because it contained false statements. The reason given by NMT for the libel actions was that I publicly stated that the MIST Trial’s results were being misrepresented. In fact misrepresentation occurred during oral presentations at scientific meetings, in publications including in Circulation, in NMT’s annual report and on NMT’s website.

In December 2010, the Guardian Newspaper (a UK national newspaper) reported in its “Bad Science” column, written by Dr Ben Goldacre, that the claims about the MIST Trial on NMT’s website were false and testimonials from patients were misused to suggest that the MIST Trial showed that implanting NMT’s devices, not just STARFlex, cured migraine. (A link to the on-line version is at the end of this letter.) NMT’s website (and NMT’s annual report) featured testimonials from three named patients (with their photographs – in one case with her young children) who had a STARFlex implant and were pain free during the analysis period in the MIST Trial. There was no mention that 3 out of 73 in the sham-operated control group were also migraine free. NMT should not have known the identity of those patients because, outside the sites that treated them, they were identified only by an alpha-numeric code. They were all treated by doctors who owned NMT shares during the trial.

The Guardian also reported that the NHS website of the Royal Brompton Hospital misrepresented the results of the MIST Trial. The website said “As many as 40% of patients could have their migraine symptoms significantly relieved through interventions to close the holes in their hearts.” Dr Mullen, the second author of the paper in Circulation, worked in that NHS hospital and on the webpage he was quoted saying “This is significant news for migraine sufferers. For the first time this study has shown that closing a PFO can have a significant effect in reducing the symptoms for patients with severe migraine.” The website also had a testimonial from a patient (with her name and photograph). She was one of the three patients out of 74 who had a STARFlex implant in the MIST Trial and who was migraine free during the analysis period. The testimonial said “I’ve now been completely cured.” As the Guardian article showed, the Royal Brompton’s webpage and misuse of a testimonial from a patient mimicked NMT’s website. Dr Mullen also appeared on NMT’s website and provided a testimonial.  Dr Mullen disclosed share-holdings and consultancy payments from NMT in the paper in Circulation but not on the Royal Brompton Hospital’s webpages. The Royal Brompton Hospital’s webpages containing false claims about the MIST Trial were removed when the Guardian newspaper reported the dishonest claims.

Dr Mullen is the second author of the MIST Trial paper. He is the first author of the paper describing NMT’s BEST Trial, which was also published in Circulation. I wrote part of the protocol of the BEST Trial and was due to provide the echo core lab for the BEST Trial until I raised concerns about the MIST Trial.

NMT’s libel actions against me ended in 2011 when NMT went into liquidation.

Some additional effects of NMT’s libel actions were that Members of the UK Parliament were enraged by NMT’s attempts to silence medical researchers. The Secretary of State for Justice wrote in The Times (of London) that NMT’s actions convinced him of the need for reform of the defamation laws. NMT’s libel actions were part of the catalyst for change that resulted in the Defamation Reform Act of 2013, which makes it more difficult for corporations to use the libel laws to silence scientists. Another effect was that in 2012 the BMJ gave me its first ever Editor’s Award “For persistence and courage in speaking truth to power.” I mention these facts only to show that my views about the MIST Trial and the publications in Circulation may be trusted more than those of Dr Dowson or Dr Mullen.

It is relevant that NMT did not originally intend that Dr Dowson should be the principal investigator in the MIST Trial. He was appointed after Professor Goadsby, who was due to be principal investigator, withdrew from the trial at the start of July 2004. Goadsby and I were both members of the trial design committee. None of the 15 authors on the paper was on the design committee and there is no acknowledgement in the paper of those who were on the design committee. Neither are the members of the trial’s steering committee named, presumably because to do so would have drawn attention to the fact that two members were not authors of the paper. Similarly the four members of the “writing committee” are not named, which I presume is because only three were prepared to be “responsible authors”. Dr Muir was not a responsible author. I cannot say whether his reluctance to be held responsible for the paper has anything to do with his concerns about the true residual shunt rate (as described below). The paper does not mention that statistical analysis was performed by Mr Wilkinson, a statistician employed by NMT. He is not named in the paper. It is particularly relevant that the investigators were not permitted to see the data that he analysed. NMT gave the investigators his analysed results and, as described below, those analysed results changed repeatedly.

Failure to disclose previous misconduct

The GMC transcripts confirm that in July 2004, when NMT asked Dowson to be principal investigator in the MIST Trial, Dowson told NMT that earlier that year he had been found to have falsified data in a trial sponsored by Allergan and that the Northern and Yorkshire Regional MREC had removed him from being principal investigator in that trial. The MREC also said that Dr Dowson was not fit to be a trial principal investigator. Nevertheless NMT wanted Dr Dowson to be principal investigator in the MIST Trial.

It may seem strange that NMT knowingly recruited a doctor guilty of research misconduct as principal investigator in the MIST trial. NMT also employed as a senior investigator in two US clinical trials a doctor that NMT knew had received a written warning from the FDA that he might face criminal prosecution for misconduct in PFO research. (The FDA letter to the doctor is also a matter of public record.) The willingness of NMT to recruit doctors with a record of research misconduct raises questions about NMT’s motives.

When NMT appointed Dr Dowson to the MIST Trial in July 2004, Dr Dowson and NMT did not inform Dr Nightingale or me about his research misconduct or that a MREC had decided that they considered him unsuitable to be principal investigator in research. Neither did Dr Dowson inform the West Midlands MREC that approved the MIST Trial in September 2004. Later when Dowson had conditions imposed on his registration by the GMC in 2006, he did not inform the West Midlands MREC, Dr Nightingale or me. His 2015 Fitness to Practise Tribunal found that these were examples of misconduct by Dowson, because he should have informed the West Midlands MREC and his co-investigators. The High Court agreed.

Failure to disclose conflicts of interest

The 2015 Tribunal also found that Dr Dowson was dishonest in his conflicts of interest disclosures to the West Midlands MREC that granted ethics approval to the MIST Trial. Dr Dowson also failed to disclose any of his many conflicts of interest in the paper published in Circulation.

The transcripts of the 2015 Tribunal show that Dowson admitted receiving about £2,000 per week from NMT for his role as chief investigator in the MIST Trial. The Tribunal was told that he started to receive payments in July 2004 and he was still being paid by NMT when the paper was published nearly four years later in March 2008 – he was principal investigator in the follow-up MIST 3 Trial at that time. These payments were not known to the MREC. In addition, Dowson’s contract with NMT had a clause for bonus payments depending on outcome. There is no doubt that NMT paid Dowson a large amount of money, but none of that is disclosed in either version of the paper.

In addition Dowson admitted to the Tribunal that he bought NMT shares one week after the steering committee heard preliminary encouraging results. That was six weeks before we released those data during a presentation I made at EuroPCR 2005. After my presentation, NMT’s share price increased. Dowson sold the shares before the final negative results of the MIST Trial were published and he made a profit from his share dealing.

Before the 2009 version and correction were published, I informed Circulation that Dr Dowson had not disclosed his conflicts of interest in the 2008 version. Despite that, no conflicts of interest are disclosed for Dr Dowson in the 2009 version and the correction contains the statement. “The authors confirm that they disclosed all relevant relationships and potential conflicts of interest that were present during the two years leading up to the manuscripts submission, as required by the American Heart Association.” As stated, Dr Dowson admitted at his 2015 Tribunal that he was still being paid by NMT when the paper was published in March 2008. His 2015 Tribunal also heard that he discussed his financial conflicts of interest with the editors of Circulation and they advised him to disclose no conflicts of interest on the MIST paper. The GMC had direct communications with Circulation and AHA (as confirmed by Rose Marie Robertson, Chief Scientific Officer of AHA).

Two authors of the MIST Trial paper in Circulation did disclose financial conflicts of interest in the paper, but seven others with financial conflicts did not disclose them. I told Circulation in 2008, that those conflicts of interest were revealed by NMT’s Chief Financial Officer in an interview to journalist, Shelley Wood. (For the record, Nightingale and I refused the offer of consultancy payments from NMT to avoid a financial conflict of interest.)

Failure to disclose ghost authorship

In 2008 I told Circulation that a vice president of NMT wrote important parts of the paper published in 2008 but it was not disclosed. His contribution was apparent from track changes on drafts of the paper that I sent to Circulation. Dr Dowson’s 2015 Tribunal found that the same track changes in the drafts of the paper proved the vice president wrote significant parts of the published text. His contribution to ghost authorship of the paper is confirmed in NMT’s legal pleadings in the libel cases. NMT’s pleadings are publicly available documents. Parts of the paper that NMT’s vice president wrote, as shown by track changes and accepted as his work by Dr Dowson’s 2015 Tribunal, remain in the 2009 version, without any disclosure of that conflict of interest, even though by 2009 the editorial staff of Circulation and AHA knew that they were written by a vice president of the sponsoring corporation.

Gift “ghost” authorship

Dr Anthony Rickards is the last author of both versions (2008 and 2009) of the MIST paper. Dr Rickards died in May 2004. The day before he died, Dr Rickards had attended a lunch-time meeting at the British Cardiovascular Society’s Annual Conference for cardiologists who we hoped would participate in the MIST trial. Dr Rickard was not on the trial design committee. (The members of the trial design committee are not named in either version of the paper. I was on that committee.)  Dr Dowson joined the MIST Trial in July 2004, two months after the death of Dr Rickards. The protocol was submitted for ethical approval even later in September 2004 and the trial started in January 2005. I therefore expressed concern to Circulation that someone who had not been involved in trial design, data collection, data analysis or writing the paper and had not agreed to be an author was named as an author. The death of Dr Rickards in 2004 did not prevent him moving up to fourth author of a letter in Circulation in 2009 (Circulation 2009;119:e194) replying to correspondence from Johansson et al.

I knew Dr Rickards, both as a UK cardiologist and because we were jointly involved in exposing an unrelated case of research misconduct. I believe that he would have been embarrassed by what has gone on to link him, without his agreement, to the flawed MIST Trial paper.

Incidentally in the letter replying to Johansson et al, under the heading “Disclosures”, the word “none” appears even though some authors had conflicts of interest. Some had even disclosed their conflicts of interest in both versions of the paper, but the correspondence denied those conflicts. Some of those named as authors of the letter have said that they had no knowledge of the letter or indeed what had gone on with the paper. The majority were site investigators who placed their trust in others for interpretation of the data and writing the paper.

False statement that authors had full access to the data

The paper contains the important statement “The authors had full access to and take full responsibility for the integrity of the data”. I told Circulation in 2008 that that statement is false and that NMT’s legal pleadings confirm that NMT did not allow Dr Dowson or any investigator to see the full data. The pleadings are in the public record. So, it is difficult to know how the authors could vouch for the integrity of the data when none of them had seen it. The false statement remains in the 2009 version

False claim that the protocol was followed

I informed Circulation that the 2008 version of the paper misreported the protocol. The 2009 version and data supplement provide some differences in the protocol, but they are also false. Neither version published in Circulation followed the protocol used in the MIST Trial. For example the procedure for contrast echocardiography was different in the 2008 and 2009 versions, but neither is the method used in the trial.

The exclusion criteria did not alter from 2008 to 2009, but do not correspond with the approved protocol.

Certain procedures performed during the MIST trial were not in the protocol approved by the MREC but investigators were led to believe that they were. That happened because the steering committee asked for protocol amendments at the start of the trial. The printed Clinical Record Forms that we used had the changes to the protocol that the steering committee had requested. Therefore most investigators, including Dr Nightingale and I, believed that those protocol amendments had been submitted and approved by the MREC. In reality Dr Dowson and NMT knew that no application had been submitted for those amendments.

One of the amendments recommended by the steering committee was an alteration for the arrangements for the last two visits of each patient in the trial. Another concerned the performance of contrast echocardiograms at the end of the trial. Because Dr Dowson and NMT had not applied for protocol amendment before the patients were subjected to the altered procedures, the patients had procedures that were not part of the approved protocol at the time that the patients had them.

Six months after the results of the trial were known to NMT, Dr Dowson wrote to MREC to make some retrospective protocol amendments. Dr Dowson and NMT requested only the amendments that were beneficial to NMT for device marketing. They did not request amendment to cover other procedures performed, which had raised concerns about performance of the STARFlex device. The MREC did not respond to Dr Dowson to confirm that they approved the retrospective protocol amendments. Therefore, there must be concern that patients and investigators were deceived about the protocol that was followed and so have the readers of both versions of the paper.

Inaccuracy of reported data

The data is not reported accurately. For example, patients had their screening contrast echocardiograms by two teams – I performed just over half of the tests and Dr Mullen’s team at the Royal Brompton Hospital did the rest. The paper makes no mention of the fact that many patients who had their screening contrast echocardiograms at the Royal Brompton Hospital (Dr Mullen’s hospital) had pulmonary shunts misdiagnosed as PFOs. That was not surprising because Dr Mullen insisted that a research fellow with no prior experience of performing contrast echocardiography should do the tests at his hospital and Dr Mullen reviewed the recordings of the research fellow’s echocardiograms later. Therefore in many cases a pulmonary shunt was misdiagnosed as a PFO. As a result, many patients assessed at the Royal Brompton Hospital and randomised to an implant procedure had no PFO to close and were subjected to a general anaesthetic and a fruitless and sometimes dangerous attempt at PFO closure. In those cases the migraine was recorded as unchanged by the procedure. One of them even had a transseptal puncture to insert a STARFlex device when there was no atrial shunt to close. A second patient had an attempted transseptal puncture, which resulted in a pericardial effusion and had to be withdrawn from the trial: the paper mentions the withdrawal because of a pericardial effusion but fails to mention that the pericardial effusion was caused when a transseptal puncture was attempted in that patient.

We must presume that a similar number of patients who were assessed at the Royal Brompton Hospital and who had pulmonary shunts rather than a PFO were randomised to a sham procedure. (I have documents which confirm that I strongly advised against allowing someone inexperienced at contrast echocardiography to do the screening tests.)

The discussion also refers to five patients in whom a PFO was not crossed. (In fact there were seven patients if one adds the one who had a device inserted after a transseptal puncture and the one withdrawn because of a pericardial effusion caused by an attempted transseptal puncture.) Referring to those patients, the statement “the screening echocardiograms of the patients in whom a PFO was not found were reviewed again, and the conclusion was consistent with the original assessment” appears designed to mislead. I reviewed the contrast echocardiograms of those patients and I decided that in some of them, if not all, the shunt was pulmonary and not across an atrial shunt. Therefore my review was at variance with the original assessment. If the original assessment at the Royal Brompton Hospital had been that those patients had pulmonary shunts, the patients would not have gone on to randomization to PFO closure.

In many of the other patients, who had their contrast echocardiogram at the Royal Brompton Hospital, the implanting cardiologist experienced difficulty crossing their PFO when one was present. For example, one could only be crossed with a coronary angioplasty wire. Six others were only a little larger (2 to 4 mm diameter) on balloon-sizing at the time of the closure procedure and not typical of the size of PFOs that usually require closure, as I have reported in an article in another journal. The question that arises is whether these patients had a small PFO that was incorrectly graded as larger or whether they had a coincidental pulmonary shunt so that the combined small atrial shunt and a pulmonary shunt produced significant shunting. In that case, even if implantation of a STARFlex produced complete sealing of the PFO, there would be residual shunting through the pulmonary shunt.

There is no mention of balloon-sizing in the methods section of either version of the MIST paper in Circulation and no balloon-size data are presented. That omission of data led Dr Carroll to mistakenly conclude in his editorial, which accompanied the 2008 version of the paper, that balloon-sizing was not part of the protocol. In fact the balloon-sizing data were collected and were presented at both EuroPCR in 2006 and the American College of Cardiology Conference in 2006.

I can also prove that at the outset I strongly advised NMT that an echocardiography core lab was required. NMT refused to put it in the protocol. The steering committee initially wanted the post implantation contrast echocardiograms performed at the two screening centres and requested a protocol amendment. The minutes of the steering committee recorded our reasons. The steering committee were concerned that by allowing implanting cardiologists to do the final contrast echocardiograms, they would in effect by assessing their own work and, in addition, many of them had no experience of performing contrast echocardiography. Also performing initial and final contrast echocardiograms at the same centres would result in consistency of techniques and facilitate comparisons in each patient.

Dr Dowson did not apply for the protocol amendment requested by the steering committee because NMT decided it did not want it. So the steering committee then decided that all the post implantation contrast echocardiograms should be reviewed by someone not involved in the device implantations. Dr Mullen emailed that I should do it, which is ironic because when questioned about the review of the echocardiograms by journalist, Shelley Wood, he claimed that it was inappropriate for me to conduct the review. Documents prove that NMT arranged for the recordings of the final contrast echocardiograms performed by the implanting cardiologists to be sent to me for review in accordance with the wishes of the steering committee. Later NMT claimed that my review was unauthorised because they did not like what I reported.

Because my review of the contrast echocardiograms showed a high rate of significant residual atrial shunts and some pulmonary shunts, NMT requested a further review by Dr Missault, the trial monitor. The steering committee did not authorise Dr Missault’s review. In many cases, his review and mine differed about the site of residual shunting (i.e. atrial v pulmonary), but as Dr Muir wrote in an email there was moderately good agreement between the reviews of the contrast echocardiograms by Dr Missault and I on whether there was significant residual shunting, but the level of agreement of the residual shunt data reported by the implanting cardiologists with either Dr Missault’s review or mine was at a chance level.

On 26 August 2006, when commented by email on the similarities in the findings in the reviews of the final contrast echocardiograms by Dr Missault and I, Dr Muir wrote “It looks as if the message is similar, in the surprisingly high rate of residual shunts and a fairly high rate of sub-optimal recordings (of the contrast echocardiograms by the implanting cardiologists)…..The “headline” interpretation is perhaps views through rose-tinted spectacles since the rate of 94% closure success for atrial shunts assumes that all the unattributable shunts are pulmonary.” Dr Dowson responded “I agree”. Yet, despite the recognition by members of the steering committee that the 94% PFO closure success rate was wrong and that, in addition, other patients had significant residual shunts that were pulmonary, neither Dr Missault’s nor my review were reported in the MIST paper. As stated above, Dr Muir was on the writing committee but not a responsible author.

It is notable that Dr Dowson’s Fitness to Practise Tribunal in 2015 found that the reported PFO closure success rate of 94% was unreliable given the findings of the reviews by Dr Missault and me.

As an aside, I was invited to give a teaching session at the European Society of Cardiology Annual Congress in 2007. I presented one of the MIST patients whose migraine was not cured by implantation of a STARFlex device in her large PFO. Contrast echocardiography after the procedure showed a residual large pulmonary shunt. Review of the initial contrast echocardiogram showed early shunting of a large amount of bubble contrast consistent with an atrial shunt, so that coexistence of two sites of shunting had not been appreciated. A CT chest showed a single large pulmonary arteriovenous malformation, which I subsequently occluded with symptomatic improvement in her migraine.

I have no doubt that data, which showed a high residual shunt rate in patients randomised to STARFlex implantation, were concealed in the reports of the MIST Trial, including the publications in Circulation, in order to protect the commercial interests of NMT, the trial sponsor. I believe that publication of those data has the potential to increase the understanding of the link between right-to-left shunts and migraine with aura.

Problems with run in period

In order that the trial could be performed quickly, NMT wanted only a one month run in period for collection of headache data before patients were randomized to PFO closure or sham procedure. It should be clear from tables 2, 3 and 4 in the paper that during that run in month, many patients did not have 5 or more migraine headache days per month (some had none). Some others had less than 7 migraine-free days per month. If the entry criteria were based on headache frequency during that run in month rather than the patients’ self-reported headache frequency, many (probably more than half) of the patients in the study had a headache frequency which would have excluded them from the study. It should also be clear from the tables that some patients did not provide any headache data during the one month run in period. There was therefore no way of assessing their individual response to STARFlex implantation, yet they were allowed to continue in the study.

Changing claims about efficacy

At the American College of Cardiology conference in 2006, Dr Dowson reported the outcome data for the MIST Trial and reported “The primary endpoint for the complete elimination of headache was not achieved. However using more conventional migraine trial endpoints significant differences were found.” He reported that 42% of patients with a STARFlex implant had a reduction in migraine headache of 50% or more compared with 23% of those who had the sham procedure – the difference was significant (p = 0.038). Dr Dowson also reported that headache burden (frequency x duration) was reduced significantly more in the implant group (37%) than in the sham procedure group (17%) (p = 0.033).

The slides showing these data were completed only a few hours before the presentation because it was only the day before the presentation that the statistician paid by NMT (but not named in the paper in Circulation) produced the “results analysis”. In the subsequent weeks the members of the steering committee were told that the claims presented at the American College of Cardiology were not correct.  However, more than 4 years after NMT knew that the claims were false NMT continued to show on the corporation’s website the false claims from the American College of Cardiology presentation that significantly more patients with an implant than sham controls had a 50% or more migraine headache.

In his editorial, Dr Carroll was disturbed that the positive secondary outcomes reported at the American College of Cardiology had disappeared by the time that the paper was published in Circulation.  Dr Nightingale and I were also disturbed, and probably more that Dr Carroll, when we discovered that we had been misinformed about the data analysis and that a presentation with incorrect information had been presented in our names as steering committee members. This made us insist that we should see the full data before we put our names on any further publication or presentation, but NMT refused to allow us to see it.

Inappropriate exclusion of outliers

In the paper in Circulation, much emphasis has been given to the exploratory analysis, which is reported as follows – “excluding 2 outliers, the implant group demonstrated a greater reduction in total migraine headache days.” Dr Nightingale and I had concerns about the exclusion of those two individuals, particularly because the explanation for why they were being excluded changed a number of times.

The results section of the 2008 version, under the sub-heading “Efficacy” states “Two patients in the implant arm were noted to account for more than one third of migraine headache days (Figure 3) throughout the entire study period and differed from the rest of the population.” Figure 3 did not support the statement in the 2008 version of the paper because the figure only showed data for the sub-group of patients in the per-protocol analysis and only during the analysis period rather than the “entire study period”.

More important, in 2008 I informed Circulation that earlier the steering committee had been told that two patients were excluded because they accounted for more than half the migraine headache days throughout the entire study period. I also told Circulation that it was not plausible to Dr Nightingale and me that two patients could account for either half or one-third of the headache days in the entire study period. The changing explanation for excluding two patients who had a critical effect on the so-called exploratory analysis increased our resolve to see the full data.

The explanation was changed again in the 2009 version to “Two patients in the implant arm were noted to account for 20% of all headache days in the implant group during the analysis period.” I cannot see how that new claim can be true. Even if it were true, it is a concern that the statistical justification for this finding which NMT tried to put to commercial use, as judged by claims on NMT’s website and the Royal Brompton Hospital website, kept shifting. (Incidentally, though this explanation for the exclusion of two patients from the exploratory analysis changed in the results section of the 2009 version of the paper, it remained unchanged in the discussion section of the 2009 version, which is careless considering that Circulation, Dr Dowson and NMT spent 18 months producing the amended version.)

Furthermore we were concerned about using the Shapiro-Wilks test to see whether data was normally distributed, then using that analysis to justify exclusion of two patients, but then performing the subsequent comparison of the two groups using a non-parametric test to find the reduction in median total headache days. If one is going to use non-parametric statistics that does not depend on the data being normally distributed, using a parametric test to exclude the “inconvenient” patients with bad outcomes from STARFlex implantation seems contrived. That is particularly so when multiple analyses were performed by the statistician employed by NMT and there is no correction for multiple comparisons.

False claim about academic affiliation

In both versions of the paper and in the subsequent correspondence, Dr Dowson states that his contribution to the research was from King’s College Hospital, London. As the transcripts show, the GMC was informed by the hospital management that none of the research was performed there. It is clear that Dr Dowson performed the research in his private clinic in Guildford. That makes it impossible for institutional oversight or investigation of Dr Dowson’s research. It is the same place that Dr Dowson worked when he falsified the data in the Allergan trial.

Problems with the on-line Correction

The on-line Correction contains a series of statement that Circulation knew or should have known were false when the journal published them in 2009.

The second paragraph states “Investigators Dr Peter Wilmshurst and Dr Simon Nightingale did not sign the Copyright Transfer Agreement because of an internal disagreement about the conduct of the study.” That statement is false. Our reason was not about the conduct of the study. Dr Nightingale and I sent emails to the other steering committee members explaining our refusal to be authors. I wrote “I cannot in conscience agree to be an author of an article, which states that “THE AUTHORS HAD FULL ACCESS TO THE DATA AND TAKE FULL RESPONSIBILITY FOR ITS INTEGRITY”. We did not have full access and I do not accept responsibility for the integrity of the article.” I sent Circulation copies of the emails in 2008, which had the sentence in capital letters as shown. So Circulation knew the real reason was that we refused to be authors of false research.

The third paragraph refers to the method of assessing cardiac shunts and refers readers to the data supplement. The method described was not the one used during the trial.

The suggestion, in the fourth paragraph, that a device lying in the right atrium or embolizing to the main pulmonary artery is an “unsatisfactory implant position” and need not be disclosed in the paper because an “unsatisfactory implant position was not considered a serious adverse event per protocol” is deliberate obfuscation. In addition, the fact that some patients had transseptal punctures when a PFO could not be found or crossed and that one had a pericardial effusion as a result should not have been concealed.

It is difficult to comment on the claims in paragraphs six and seven because no investigator was allowed to see the full data.

Paragraph eight says “The authors confirm that they disclosed all relevant relationships and potential conflicts of interest that were present during the two years leading up to the manuscript submission, as required by the American Heart Association.” As previously stated, at his Fitness to Practise Tribunal, Dr Dowson admitted that he was being paid by NMT Medical at the time the paper was published Other authors received payments from NMT, as confirmed by the Chief Financial Officer of NMT. The majority of them were investigators in the MIST 3 follow up study. So it is likely that they also were being paid by NMT at the time that the paper was published. In addition, parts of the paper were written by a Vice President of NMT, though his role in authorship was concealed.

Circulation may be interested to know, the GMC is now considering the allegation that during his 2015 hearing Dr Dowson misled the Tribunal. So one must ask why Circulation worked to produce a correction of the paper with such a doctor.

Conflicts of interest at Circulation when dealing with concerns about the paper

Dr Loscalzo was editor of Circulation at the time the MIST Trial paper was published. He dealt with my complaint about the paper and published the 2009 version. Dr Loscalzo is now and was then Hersey Professor of the Theory and Practice of Medicine at Harvard Medical School, Chairman of the Department of Medicine, and Physician-in-Chief at Brigham and Women’s Hospital. NMT was a corporation set up in Boston, Massachusetts, to manufacture and market the STARFlex and related devices under licence from the patent owner, Boston Children’s Hospital. The Boston Children’s Hospital is another hospital affiliated to Harvard Medical School and is adjacent to and structurally linked to the Brigham and Women’s Hospital. It is a matter of public record that some Harvard cardiologists received royalties from sale of STARFlex devices, include the prototype’s inventor, Professor James Lock, Cardiologist-in-Chief at Boston Children’s Hospital until 2016 and Board member of NMT. In addition, during the time that Professor Loscalzo was dealing with my complaint about the 2008 paper, his department was receiving payments from NMT for involvement in CLOSURE 1. A member of Loscalzo’s department was on the CLOSURE 1 steering committee. This raises the concern that some of those dealing with my complaint about the MIST Trial paper were conflicted.

What Circulation should do

  1. Circulation should retract both versions of the MIST Trial paper, the correction and the data supplement.
  2. Circulation should explain the actions of those at the journal who published the 2008 version of the paper, when the authorship of four submitted versions kept changing.
  3. Circulation should also explain why Circulation and AHA agreed to publish the 2009 version when the Journal’s staff knew that the first / corresponding author was guilty of research misconduct in unrelated research, when he had submitted a false 2008 version of the paper and when Circulation had received from me documents, which showed that claims in the then unpublished 2009 version and the correction were false.  

Yours sincerely

Peter Wilmshurst

Consultant cardiologist

Link mentioned in text:

NMT are suing Dr Wilmshurst. So how trustworthy are this company? Let’s look at their website… – Bad Science


2 responses to “Correspondence with Circulation about retraction of the MIST Trial publications”

  1. Dear Peter,
    thank you for fighting this fight.
    It is appalling that AHA and Circulation sent a lawyer to threaten you. But then again, they don’t know even how to spell the word “ethics” there.
    The former EiC of Circulation whom you refer to, Joseph Loscalzo, is a toxic character. He is best described as a member of the Piero Anversa gang which was responsible for one of the biggest biomedical research fraud scandals, that of non-existent heart stem cells.

    Harvard sacked Anversa and ordered oodles of retractions, one of them with Loscalzo in Circulation:
    Loscalzo published lots of fake science:
    And not all of it with Anversa, for example this Kuo, Schroder & Loscalzo 1997:

    Elsewhere at AHA, the EiC of Circulation Research, Roberto Bolli, was sacked. Not because he was of course also an active member of the Anversa’s fraudster gang, not at all. AHA does not mind research fraud and patient abuse, as you explain above.
    Bolli was sacked for being a homophobic bigot which AHA never minded for years up until it became public and caused an uproar.
    The only thing these people are afraid of, is publicity. Thank you for exposing them.


  2. […] letter was quoted by Wilmshurst in his blog post, where he also shares his original letter to Circulation‘s Editor-in-Chief Joseph A Hill, […]


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: