Major heart journal retracts two papers from Oxford group for misconduct

The Journal of the American College of Cardiology (JACC) has retracted two 2018 papers out of the University of Oxford because of misconduct.

Both retraction notices blame first author Alexander Liu, a student in the lab at the time, who disputes the retractions. The studies were part of a larger effort to improve heart imaging that caught the attention of cardiologists and was highlighted by Oxford in 2015.

Here’s the notice for “Diagnosis of Microvascular Angina Using Cardiac Magnetic Resonance,” a paper that has been cited 59 times, according to Clarivate Analytics’ Web of Science, earning it a “highly cited paper” designation:

The JACC Journals Ethics Board has voted to retract this paper, relying on the findings of misconduct after an investigation by the University of Oxford (outlined below). The decision to retract the paper follows the conclusion of an investigation under the University of Oxford’s (“the University’s”) Code of Practice and Procedure on Academic Integrity in Research (“the Code”). The Registrar of the University convened a Panel under the Code. The Panel considered a number of issues, including in relation to this paper. The Panel concluded that the first author, Dr Alexander Liu, was responsible for misconduct in research. The Panel’s findings with regards to misconduct were limited to the actions of the first author. No other co-author was found to be involved in the misconduct. It is understood that the first author disagrees with the Panel’s findings. The first author has raised a complaint with the Office of the Independent Adjudicator (OIA) (The OIA reviews complaints from students about their higher education provider).

In relation to this paper, the Panel’s findings included that:

• rather than 50 individual patients in the study (28 CAD providing 35 vessels and 22 NOCAD providing 66), there were actually 26 CAD patients, providing 31 vessels and 25 NOCAD patients providing 60 vessels, but three patients each providing two vessels were double counted, providing 66 vessels in all. Only six patients met the stated definition of NOCAD;

• 5 controls with discrepant ages were included in the study;

• controls had been iteratively excluded from the analysis as the paper was revised, without explanation;

• that subjects used as controls in the study were patient volunteers and not healthy controls; and

• Figure 5 had been fabricated.

The Panel’s view was that this paper would likely need to be retracted from the literature as it had major irregularities and its conclusions were unsafe.

The following co-authors agree that a retraction is appropriate: Vanessa Ferreira, Rajesh Kharbanda, Stefan Neubauer, Stefan Piechnik, Michael Jerosch-Herold, Keith Channon, John C. Forfar, Joanna Liu and Rohan Wijesurendra.

The notice for “Gadolinium-Free Cardiac MR Stress T1-Mapping to Distinguish Epicardial From Microvascular Coronary Disease” has the same first paragraph, and then continues:

In relation to this paper, the Panel’s findings included that:

• certain data had been fabricated by the first author amending the actual study data so that the paper and the central illustration would show a compelling case that T1 mapping could distinguish between epicardial obstructive coronary artery disease and coronary microvascular dysfunction;

• the number of control subjects, their age, and the statistical test to calculate the significance of a difference between the patients in this paper were incorrect; and

• Figure 2 had been fabricated.

The Panel’s view was that this paper would likely need to be retracted from the literature as it had major irregularities and its conclusions were unsafe.

The following co-authors agree that a retraction is appropriate: Vanessa Ferreira, Rajesh Kharbanda, Stefan Neubauer, Stefan Piechnik, Keith Channon, John C. Forfar, Michael Jerosch-Herold, Andreas Greiser, Joanna Liu and Rohan Wijesurendra.

That paper has been cited 41 times.

‘I had made honest errors’

Liu disputes the retractions. In a statement to Retraction Watch, which we have published in full here, Liu said:

I had made honest errors which I was unaware of during the preparation of the papers as a student under supervision.  I feel very sorry to have made the errors which have led to much insight and self-reflection.  The errors were unfortunately not picked up during internal reviews; the papers were submitted after final approvals from the corresponding author.  The alleged data manipulation was not done by me.  My position is supported by third-party senior academic and barrister opinions.  

Li shared those opinions with Retraction Watch — one by a researcher he said volunteered to help his case — but under the condition that we not quote from or publish them, and that we keep the names of the writers confidential.

With regard to “Gadolinium-Free Cardiac MR Stress T1-Mapping to Distinguish Epicardial From Microvascular Coronary Disease,” Liu’s statement continued: 

• the T1 data in the central illustration distinguishing epicardial obstructive coronary artery disease and coronary microvascular dysfunction were not manipulated by me, but by a third party whereby spreadsheet formulae appear to have been overridden with different values in multiple spreadsheets at the very last stages of analysis which jeopardised the integrity of my work; 

• “the number of control subjects, their age and statistical tests” were miscalculated due to honest errors, including p-values between controls and patients which did not affect main study conclusions and were unfortunately not picked up during internal reviews; and 

• figure 2 is an illustrative figure created by a honest mix up of images from different subjects, when many single-subject options were available. 

With regard to “Diagnosis of Microvascular Angina Using Cardiac Magnetic Resonance,” Liu’s statement reads:

• I made an honest mistake in mixing up per-vessel and per-subject analyses resulting in an error in the number of subjects and vessels. I was trained by one of the senior authors to perform per-vessel data analysis during multiple iterations of an earlier paper, when later preparing this manuscript for the corresponding author which required per-subject analysis, in hindsight I became confused and performed the analysis on a per-vessel basis; the per-vessel analysis itself was performed accurately. I was highly disappointed by this error. Unfortunately it was not picked up during internal reviews. I did not know there was double counting of vessels at the time, which was an honest mistake: there were many other vessels I could have used; 

• 5 controls with discrepant ages were included by an honest mistake, which did not change study conclusion and were not picked up during internal reviews; · 25 controls were available in total, 20 controls were required for final analysis, therefore 5 controls were excluded, which did not affect study conclusions; 

• a small number of volunteers were not completely healthy and were included by honest error, which did not change the study conclusion and were not picked up during internal reviews; and 

• figure 5 was an illustrative figure created by a honest mix up of images from different subjects when many single-subject options were available. This error was made in the same manner as figure 2 in the other paper, likely due to needing to operate different software to extract different types of images.

We asked Liu who might have manipulated the data. He told Retraction Watch:

I don’t know who would have had access to the file that is not my role to investigate. 

Liu said that while he agreed “that manipulated data should not be in the publication,” his solution — and that of his outside expert and, according to him, the JACC editorial board — was to replace the paper with a version from which the manipulated data were removed.

That, however, seems closest to a retraction and replacement, rather than a correction, and journals typically reserve that approach for cases in which they can demonstrate that honest error led to the issues in a paper.

Like Retraction Watch? You can make a tax-deductible contribution to support our work, follow us on Twitter, like us on Facebook, add us to your RSS reader, or subscribe to our daily digest. If you find a retraction that’s not in our database, you can let us know here. For comments or feedback, email us at [email protected].