A diabetes paper that received quite a bit of media attention when it was published in June 2013 was retracted and reissued to fix data errors shortly after publication.
The paper, which showed a steep decline in mortality rates for diabetics in Ontario, Canada, and the UK between 1996 and 2009, was republished in December 2013, with the same conclusion and the errors corrected.
Here’s the retraction notice for “Mortality trends in patients with and without diabetes in Ontario, Canada and the UK from 1996 to 2009: a population-based study”:
The authors wish to retract this article, published online on 22 June 2013, because a calculation error has become apparent. Some of the data in Table 2 and ESM Tables 4 and 5 are incorrect, as are some of the secondary conclusions. The authors regret this error and confirm that the main conclusion remains valid.
This retraction is intended to maintain the integrity of material published in Diabetologia and the journal acknowledges the full cooperation of the authors.
A corrected version of the paper is available at: DOI 10.1007/s00125-013-3063-1.
Here’s the abstract from the corrected paper, which has now been cited twice, according to Thomson Scientific’s Web of Knowledge:
AIMS/HYPOTHESIS:
The aim of this study was to determine the contemporary rate ratio of mortality and changes over time in individuals with vs without diabetes.
METHODS:
Annual age- and sex-adjusted mortality rates were compared for adults (>20 years) with and without diabetes in Ontario, Canada, and the UK from January 1996 to December 2009 using The Health Improvement Network (THIN) and Ontario databases. The total number of individuals evaluated increased from 8,757,772 in 1996 to 12,696,305 in 2009.
RESULTS:
The excess risk of mortality for individuals with diabetes in both cohorts was significantly lower during later vs earlier years of the follow-up period (1996-2009). In Ontario the diabetes mortality rate ratio decreased from 1.90 (95% CI 1.86, 1.94) in 1996 to 1.51 (1.48, 1.54) in 2009, and in THIN from 2.14 (1.97, 2.32) to 1.65 (1.57, 1.72), respectively. In Ontario and THIN, the mortality rate ratios among diabetic patients in 2009 were 1.67 (1.61, 1.72) and 1.81 (1.68, 1.94) for those aged 65-74 years and 1.11 (1.10, 1.13) and 1.19 (1.14, 1.24) for those aged over 74 years, respectively. Corresponding rate ratios in Ontario and THIN were 2.45 (2.36, 2.54) and 2.64 (2.39, 2.89) for individuals aged 45-64 years, and 4.89 (4.35, 5.45) and 5.18 (3.73, 6.69) for those aged 20-44 years.
CONCLUSIONS/INTERPRETATION:
The excess risk of mortality in individuals with vs without diabetes has decreased over time in both Canada and the UK. This may be in part due to earlier detection and higher prevalence of early diabetes, as well as to improvements in diabetes care.
The managing editor of the journal gave us a few more details:
What happened was basically as follows:
- After online (but not print) publication a reader contacted the authors with a question related to the paper. This led the authors to realise there was a calculation error which affected data in table 2 and the supplementary material, as well as the secondary conclusions of the paper
- The authors alerted us to this and asked to correct their paper. They sent a version of the paper in which the required corrections were highlighted.
- Our editor evaluated the requested changes and approved the corrections having verified that main conclusions were not affected.
- Given the extent of the changes required the authors requested permission to print a ‘clean’ version of their paper rather than an erratum. The only way to achieve this was for us to retract the paper with the incorrect data and then publish the corrected article in its entirety.
- This course of action was approved by our publisher, Springer, and their quality assurance team because the miscalculation directly affected the results of the paper.
Hat tip: Rolf Degen
When we publish in a journal, we expect to know those who are in charge of the journal quality control, namely the editors and the members that form the editor board. Similarly, when Diabetologia’s managing editor refers to Springer’s “quality assurance team”, why is it that we are never told who forms this secret council of experts? Why is it that a search on SpringerLink reveals zero hits for such a term, or for such a “team”? I call on Springer to show the “board” of experts who make up this “quality assurance team” because this forms an integral part of the trust that forms part of a scientist’s relationship with a publisher, based on intrinsic transparency about such issues. Trust can only be borne of transparency. As for the process, I give a thumbs up to the handling, simply because there might not have been any other (or better) way. However, I do not have access to this paper, and two lingering queries/concerns remain (for me):
a) in the republished paper, did the authors specifically indicate that the paper is a republished version of a retracted paper (e.g., as part of a disclaimer, or COI statement, or similar)? In other words, in 10 years’ time, will scientists who read the “new” paper know that it was based on flaws of an “old”, retracted paper? MY argument is that if the new paper has no such mention of correction and republishing, then the authors will appear to have conducted the experiment perfectly, which is of course a fallacy.
b) imagine that another reader finds another fault in the new paper. Will it be retracted once more and how many cycles of retractions and re-publications is considered to be the “limit” by Springer (or any other publisher)? What does COPE have to say about this issue (because Springer is a COPE member)? I feel that in this case, too, we are entering into dangerously unchartered waters, even though the managing editor (who is who, i.e., what’s his name?) has explained the decision rather carefully and logically.
If scientists do not start to demand of the publisher more explanations of the process and demand more transparency in the same way that honesty and transparency is demanded of them (by the publisher), then we may be heading down the wrong path to finding the reasons why retractions take place and how the literature has come to be so corrupted.
Very good points. There is also the possibility of republishing retracted papers under different titles and under different authors name variations.
I have not looked into this specific paper. You have some good points on transparency. However, your comment has such a negative tone on corrected papers, that I want to state some principles that seem to be overlooked in RW comments quite often:
1.) nobody is without fault. If you think you are, then that thought is your first fault.
2.) there probably are few scientific papers without any error.
3.) there is such a thing as honest error in science.
RW has dealt fine with fraudulent papers. There is no doubt that fraud needs to be retracted, period.
What seems to be a lot more difficult is the question of how to deal with honest error.
If honest error is correctable, why should it not be corrected? Why should authors not be allowed to republish their corrected findings, if they or the editor/journal decide to retract the original paper? What’s the alternative, dump correct scientific knowledge forever? Or should another group take the retracted work, repeat it and publish it as if it was their work, since the original work is retracted and technically non-existant? What solution do you have?
In the end, it is the task of the scientific community and every single scientist to critically evaluate a paper and not to take it as Gods given truth just because it’s a paper.
The original authors seem to have been pretty careful to use appropriate ‘people first’ language, talking about ‘individuals with diabetes” rather than “diabetics.’ However the RW entry says “which showed a steep decline in mortality rates for diabetics ..” This kind of language (referring to people as ‘diabetics” or “hypertensives”) is discouraged these days in favor of “people first” language.
Speaking as a ‘person with testicles’ in response to the ‘person without testicles and who originally might have had uterus, fallopian tubes and breasts’, I think that sometimes over the top concern with the political correctness becomes laughable in certain environments.
While I agree that concern over the correctness of every written sentence in a blog post can reach an absurd threshold, I also argue that the need to define the difference between a ‘diabetic’ and a person with said condition is less about political correctness and more about respect for the individual.
Being identified in terms of one’s condition is what leads to lazy overgeneralized stereotypes that further perpetuate the social stigmas that are felt by the affected persons.
I think that since the perceived problem was that of the authors of the blog, whom we all know as intelligent and sensible people, it was have been infinitely more polite – and probably more effective – to just send them a message pointing out the impolitic usage, as you see it, rather than jumping up and waving the ‘slippery slope’ flag.
I do not assert that there is anything “impolitic” about the post. I only chose to present another view in order to highlight how language has an immensely powerful role in certain instances and should not be overlooked and unduly dismissed, especially within the context of a strongly gendered reply.
As the author of the original post about ‘diabetics’ I wasn’t trying to ‘jump up and down’ or wave the flag of ‘political correctness’ or attack anyone. The idea of ‘people first’ language has been around for quite some time now, and I think it’s good if people are careful about not using labels like ‘hypertensives,’ plus a number of journals won’t allow such language.