Retraction (in all but name) of flu paper raises eyebrows
When is a retraction not a retraction? Why, when it’s a correction, of course — like the one the journal Emerging Infectious Diseases issued this month:
In the article Reassortment of Ancient Neuraminidase and Recent Hemagglutinin in Pandemic (H1N1) 2009 Virus (P. Bhoumik, A.L. Hughes), errors were made in selection of the hemagglutinin (HA) and neuraminidase (NA) sequences for the initial and subsequent data sets. As a result, the authors incorrectly concluded that the NA gene of the pandemic (H1N1) 2009 virus is of a more ancient lineage than the HA. Other researchers (and the authors) have not been able to reproduce the findings when using HA and NA matched pairs from viruses chosen on the basis of geography and time and correctly have pointed out errors in the data set that make the original conclusions invalid.
In other words, 1) the article was based largely on an error and 2) the central point could not be reproduced, two flaws that, at least in our book, usually constitute grounds for retraction.
The paper was written by Priyasma Bhoumik and Austin Hughes. Bhoumik, now a post-doc at Harvard, at the time was a PhD student at the University of South Carolina, where Hughes is a senior faculty member. Funding for the work came to Hughes from the National Institutes of Health, according to the original article.
We spoke with Hughes, who said that in this case, correction versus retraction is a distinction without a difference:
It’s equivalent to a retraction. I don’t think it matters which term you use.
The editors, he added
basically dictated the terms of the correction. It was really a retraction called a correction and at that point I didn’t really care. I wanted it to be a correction for the sake of the student. I was completely convinced her error was not due to malice, or an intent to deceive.
We tried unsuccessfully to reach the editor of the journal, a publication of the Centers for Disease Control and Prevention, to get their version of events. We hope to update this post when we hear back.
Initially, Hughes told us, he didn’t think the article was fatally flawed. The student had incorrectly included some gene sequences in her database analysis and failed to notice her mistake.
I thought she could take those out, redo the analysis and it would be a correction. For some reason went through a process that took almost a year, by reviewers who seemed very hostile too that idea
The error, according to the researchers, involved the unwitting inclusion of DNA from seasonal influenza with that of the pandemic strain, the result either of a blunder (in Hughes’ view) or a muddied database (the explanation of his co-author).
Hughes said he shoulders part of the blame for allowing the error to slip by him — but he also noted that if the reviewers had paid as much attention to the manuscript when it was submitted as they did once it was published, the problem would have been caught before it saw the light of day.
Hughes also took issue with the way the journal handled the situation. During the second review, he said, the reviewers claimed to have analyzed “thousands” of data points to support their contention that the original finding was unsupportable, but they did not allow the authors to see those data.
There was a certain asymmetry in the situation. I can understand a certain amount of that when you’ve messed up, but it’s at least possible that that wasn’t true; or that it didn’t show what they said it did.
Bhoumik, who performed the work for her doctoral dissertation, corroborated Hughes’ account, although she stood by the correction as a more appropriate way to handle the article.
Even though our results were not exactly the same initially, it still kind of supported what we wanted to show in the paper. … Other things we wanted to show are still true.
The handling of this paper reminds us of a similarly sweeping correction issued by Nature not long ago.
Bhoumik and Hughes are correct when they said that retraction in their case would have carried more stigma. But one might fairly argue that a correction that invalidates the central conclusion of an article could threaten the integrity of future research if that paper were to be cited, included in a meta-analysis or otherwise incorporated into someone else’s work.