One of the papers from a massive heart disease study in China, published in the Lancet, has been retracted and republished after the authors noticed a statistical error.
The article, by authors from Peking Union Medical College in China, Yale University, and elsewhere, presented the results of the China PEACE-Retrospective Acute Myocardial Infarction Study, part of a national initiative to study and improve care for cardiac problems. After being posted online on June 24, 2014, the authors noticed that they’d incorrectly weighed one of the cities in their calculations, which threw off a number of national estimates.
After the corrections were made, the paper was peer-reviewed again, and reviewers stated that despite the mistakes, the original conclusions were sound.
Today is a banner day on Retraction Watch: This is our second excellent example of transparency in 24 hours, and therefore the second entry in our “doing the right thing” category. An editorial lays out exactly what happened, including a timeline, allowing scientists to feel confident they’re basing the next research step on solid and accurate data. (We also appreciate the hat tip to the Committee on Publication Ethics retraction guidelines, which we often send out to editors of bad notices as a gentle reminder.)
Here’s the notice for “ST-segment elevation myocardial infarction in China from 2001 to 2011 (the China PEACE-Retrospective Acute Myocardial Infarction Study): a retrospective analysis of hospital data”:
On June 24, 2014, the China PEACE-Retrospective Acute Myocardial Infarction Study was published online by The Lancet. On Aug 14, the authors drew our attention to a miscalculation in the weight of one of the urban areas in the study, affecting the national estimates for rates of hospital admission, use of aspirin, clopidogrel, and primary percutaneous coronary intervention, proportion of patients not receiving reperfusion, median length of hospital stay, in-hospital mortality, outcomes, and characteristics of patients with ST-segment elevation myocardial infarction. Although the overall message of the study did not change, corrections were necessary in the findings section of the summary, the methods and results sections of the Article, figures 2, 3, and 4, both tables, and the appendix.
Regrettably, mistakes in calculations can happen. When they do it is important to correct the scientific record promptly, fully, and transparently. To ensure accuracy and validity of the changes as well as a correct and robust manuscript that can inform future clinical care, we carried out further peer review of the corrected paper. The clinical and statistical reviewers confirmed the appropriateness of the changes, and stated a belief that the changes, albeit extensive, made no difference to the interpretation of the findings.
Because of the extent of the changes and based on the Committee on Publication Ethics’ guidelines, which state that “Journal editors should consider retracting a publication if: they have clear evidence that the findings are unreliable, either as a result of misconduct (eg, data fabrication) or honest error”, we now retract the previous version and republish the corrected version of the paper in this issue. The old version will be added to a webappendix attached to the new version and will be marked as retracted. Additionally, the Comment discussing the study that was published online has been corrected to reflect the changes in the Article, in the online and accompanying print version.
References
Li, J, Li, X, Wang, Q…, and for the China PEACE Collaborative Group. ST-segment elevation myocardial infarction in China from 2001 to 2011 (the China PEACE-Retrospective Acute Myocardial Infarction Study): a retrospective analysis of hospital data. Lancet. 2014; (published online June 24.)http://dx.doi.org/10.1016/S0140-6736(14)60921-
Li, J, Li, X, Wang, Q…, and for the China PEACE Collaborative Group. ST-segment elevation myocardial infarction in China from 2001 to 2011 (the China PEACE-Retrospective Acute Myocardial Infarction Study): a retrospective analysis of hospital data. Lancet. 2015; 385: 441–451
Kirtane, AJ and Stone, GW. STEMI care in China: a world opportunity. Lancet. 2015; 385: 400–401
The journal also published an editorial on when to retract and when to republish, which reads in part:
Retractions are never easy and journals and editors are still all too often reluctant to take this step. However, it is important to reiterate that the purpose of retractions is the correction of the scientific literature, if the findings as presented are invalid or unreliable. Retraction is not a punishment or tainting of the reputation of one or more authors. When a retraction is due to serious misconduct rather than honest error further appropriate actions against the researchers responsible must be taken by their employers, such as academic institutions or pharmaceutical companies. By contrast, a retraction due to an honest error in the form of a miscalculation or misclassification can be followed by republication of a corrected paper, as in this case.
So where do we draw the line between a correction and a retraction followed by republication? The Committee on Publication Ethics states in its retraction guidelines that “journal editors should consider issuing a correction if a small portion of an otherwise reliable publication proves to be misleading (especially because of an honest error)”. So what should happen if a large portion is misleading? We believe that if many of the numerical findings in the results section change or the interpretation of the work is altered following a miscalculation or misclassification due to an honest error, republication should be considered. The corrected paper should pass peer review and editorial scrutiny once again and when republished the changes should be made transparent. Retraction and republication is a further example of correcting the scientific literature. In our opinion, it should be considered by journal editors in the interests of readers, research users, and the scientific community.
Update, 12:15 p.m. Eastern, 2/1/15: Please see a guest post by one of this paper’s authors, and an editor’s note explaining a change to this post.
Hat tip: Kerry Grens
Here’s to good, honest people everywhere. Kudos to all the authors and editors who have the courage to do the right thing.
And I think it could have been considered a correction – and it was a minor error in weighting – I would not consider it a major error.
“corrections were necessary in the findings section of the summary, the methods and results sections of the Article, figures 2, 3, and 4, both tables, and the appendix.”
It might be a minor error, but it affected a whole lot of the paper. This retraction-republication is much better than the megacorrections we have seen, since you’d have to read the original and the correction at the same time.
Perhaps we should try to find better terms to distinguish the various forms of retractions, error vs miconduct, but at the same time this may be a problem because so many of the retractions due to supposed errors (as per the notice) were in reality due to misconduct.
One of the positive outcomes of this case is that it avoids the “old” and incorrect version from being cited. So, in that sense, Lancet also deserves some praise here for accomodating the actual possibility of a republication. Most likely publishing in a journal like The Lancet carries its fair share of costs, so taking the risk of republishing a paper that had issues, is admirable. I am sure that many other journals (or their editors) in the same situation might have been reticent about republishing a paper that was retracted from the same journal. One of the serious problems with retractions at the moment – as I see it – is that there are few efforts – by authors, editors, journals and publishers – to correct the downstream literature. Consequently, one of the real risks is that papers that have been retracted may continue to be cited, or newly retracted papers that were already cited have not been corrected in the already published literature. The latter case is a monumental task that still needs to be addressed, but I personally see no pro-active efforts to address this point, by any publisher (at least on a journal-wide scale). So, kudos to both authors and The Lancet for regaining a fraction of trust in what is rapidly turning out to be most likely science’s largest crisis in trust.