A journal posted an abstract online suggesting a link between vaccines and autism. After a firestorm of criticism, it removed the abstract, saying it was going to be re-reviewed. Now, the journal has decided to formally reject it.
As we reported last month, Frontiers in Public Health removed the abstract after it sparked criticism on social media. After doing so, the journal released a public statement claiming that the paper was “provisionally accepted but not published,” noting that the journal had reverted it to peer review to ensure it was re-reviewed.
Now, Gearóid Ó Faoleán, ethics and integrity manager at Frontiers (the journal’s publisher), told Retraction Watch that after consultation with an external expert, the journal has rejected the paper, adding:
The rejection was merited due to severe limitations in the validity of the results.
Faoleán noted:
…this article was provisionally accepted but not published. In response to concerns raised regarding the abstract – which was made provisionally available online – Frontiers reopened its review. Following further manuscript assessment by the Field Chief Editor of Frontiers in Public Health, in consultation with an external expert, the manuscript was subsequently rejected.
According to Faoleán, the reason for rejection has been communicated to the study’s authors in more detail. Faoleán declined to provide the rejection letter, noting that
the external posting of discussions from the review process is not allowed.
The study included findings from online questionnaires by 415 mothers of home-schooled children aged between six and 12. Almost 40% of the children in the study’s sample were not vaccinated. The authors reported that those that had been vaccinated were three times more likely to develop autism and other neurodevelopmental disorders.
According to Motherboard, the research in question was mostly funded by Generation Rescue, a non-profit led by actor Jenny McCarthy, who has claimed that vaccines can cause autism.
Earlier this week, the study’s first author, Andrew Mawson, a visiting professor at Jackson State University, told Motherboard:
Generation Rescue provided most of the funds for the study, but of course had no role in the study itself.
He added:
I can assure you that we fully acknowledge the importance of vaccinations in public health.
As we previously reported, in 2011, Mawson filed a lawsuit against the Mississippi State Department of Health, claiming his contract at the University of Mississippi Medical Center was not renewed after he advocated the need for more studies on vaccine safety. The suit was dismissed in 2012.
Frontiers is listed on librarian Jeffrey Beall’s list of “potential, possible, or probable predatory” scholarly publishers — a designation which divides researchers.
Like Retraction Watch? Consider making a tax-deductible contribution to support our growth. You can also follow us on Twitter, like us on Facebook, add us to your RSS reader, sign up on our homepage for an emailevery time there’s a new post, or subscribe to our daily digest. Click here to review our Comments Policy. For a sneak peek at what we’re working on, click here.
A journal posted an abstract online
Is this normal practice for Frontiers, to post an Abstract upon “provisional acceptance”, and then the full paper when that acceptance becomes definite? For the time that the Abstract was visible (before silently vanishing again), there was no special indication that it was incomplete… the Frontiers website simply presented it as a very short, detail-deficient but self-sufficient paper.
They still seem to be insisting that they never published the paper, as if the Abstract that everyone saw never existed.
The most interesting information in this story seems to be lacking: What was actually the critisism against the research paper? Was that critisism valid, or was the paper redrawn just because its findings were contrary to dominant opinion?
More on that would be interesting.
I agree that Frontiers should have been clear about what is wrong with this study. They say that “the external posting of discussions from the review process is not allowed”, but this is not really a peer review matter, it’s a retraction. Retractions ought to come with a retraction notice explaining what is wrong with the paper.
But the flaws with the paper were pretty obvious. The biggest, fatal flaw in my view was that this was an online survey study and so subject to selection bias i.e. parents whose children developed autism “following vaccination” may have been more likely to want to participate. This is especially true because the study was advertised in a community, homeschoolers, with lots of believers in a vaccine-autism link. Put simply, all this study shows is that in this community, parents *think* vaccination causes autism.
Clink on the link in these words in the above article: “As we reported last month,” it will take to you some of the criticisms.
Also, I saw the abstract through the Google cache, and there several questions about its basic design. If you have taken some basic statistics you would know that the goal is to randomly select your subjects. This was impossible in that it was an online survey targeted to specific populations that do not represent the total population of American children.
Thank you for clarification. I had not tried to check the critisism rised, just read the Retraction Watch comment above.
However, when doing so, I discover that while the lack of random sampling implies that there is no evidence for a vaccine x autism link, it is within the realm of possibility, and the authors do in fact conclude that there should be followups where the sampling is improved.
So, as I see it, the paper just says, in my wording: “in this obviously skewed sample, we found tendencies that are so worrying they should be taken seriously and checked out in better ways”. Which must be a valid statement.
One may certainly claim that such a result is not worth publishing, or that the authors should be clearer about these limitations, but I think I have often seen research conclusions where such severe limitations are better hidden, as this conclusion does not even claim the results to be generally valid, only that they are what characterizes this specific, skewed sample.
EF
“do in fact conclude that there should be followups where the sampling is improved. ”
They always say that. It is even used in studies on homeopathy, it is just your basic weasel words.
The problem will always be that they are starting with a conclusion and trying to get to that, which is perhaps the reason for the extremely biased sampling. Many who commented on the twitter stream and on the actual abstract (before it disappeared) really wanted to see the full paper. That way they can see all the methods and the data. It would also have been nice to see an explanation from the authors as to why did not get data from the CDC Vaccine Safety Datalink (which uses computerized medical data from several health management organizations).
Either way, it does not replace the several large epidemiological studies done over more than a decade that used large medical databases in several countries. The following is a meta-study of several of those papers covering over a million children:
Vaccines are not associated with autism: An evidence-based meta-analysis of case-control and cohort studies (a pdf of the uncorrected proof)
It must also be noted that no evidence was supplied of actual genuine diagnoses of autism, ADHD and the rest. Without that everything else crumbles…
We could easily agree that better validity and reliability would be better. But does that qualify for a retraction? I am still bewildered about that.
I suppose they had some counts of cases in their sample?
Do you mean that just taking people’s answers for face value is not good enough in an enquete?
One could easily agree that letting people just fill in a form on screen opens for skewed data, but do you mean a set of doctors’ diagnoses in written should be handed in by the respondents? That is normally not the case.
Still, I do not have any opinion on vaccine and autism. I am just trying to find the motives for the retraction. I am still bewildered, but this is not so important to me…
I have no sympathy for the whole anti-vaccine movement. However, this is at least the 2nd time that the Frontiers team has taken what I think is the wrong approach to problematic (or at least unpopular) papers. In my view:
* The ideal approach is better peer review and editorial control up front.
* Retraction is appropriate for fraud.
* When reviewers and editors miss errors (at sometimes they must), and those errors do not constitute fraud, the preferred approach in my mind is post-publication commentary. In the old days, that was letters to the editor; now Frontiers provides the opportunity for comments below the articles; some other journals provide a more formal voting system, e.g. http://www.cureus.com/siq; and there are systems outside the journal, as well, like PubMed Commons or PubPeer.
In other words, let the community point out the flaws of the articles; don’t change the publication after it appears.
Thank you for your wise comment! Best must be, as you write, som version of Letters to the editor or comments “in the open”, where there is discussion under real names, not pseudonyms and voting results.
Indeed, retractions done by an editor under pressure, could work the same way as the Cureus approach. or the “innate “wisdom of the crowd”” – as Cureus calls it: They could both very well lend characteristics from the dictatorship of the People’s tribunals, the hegemony of the fools, or the rule of the traditionalists.