Lancet Article Retracted for Concerns over Data

You may have read my recent post in regards to The Lancet article entitled “Hydroxychloroquine Poses Significant Risk of Cardiac Arrythmia in COVID-19 patients,” however due to some concerns in the reproducibility of the data, the article has been retracted.

I have since edited my original post as retracted, and with this blog post I will describe the concerns of the article and the work leading up to the retraction of the paper.

Here is a statement written by three of the authors of the original article in their retraction statement:

After publication of our Lancet Article, several concerns were raised with respect to the veracity of the data and analyses conducted by Surgisphere Corporation and its founder and our co-author, Sapan Desai, in our publication. We launched an independent third-party peer review of Surgisphere with the consent of Sapan Desai to evaluate the origination of the database elements, to confirm the completeness of the database, and to replicate the analyses presented in the paper.

Our independent peer reviewers informed us that Surgisphere would not transfer the full dataset, client contracts, and the full ISO audit report to their servers for analysis as such transfer would violate client agreements and confidentiality requirements. As such, our reviewers were not able to conduct an independent and private peer review and therefore notified us of their withdrawal from the peer-review process.

Source: The Lancet

The immediate concern from this statement is that Surgisphere, the company that contains the patient data, refused to share the full data stating that it would violate client agreement and confidentiality.

It is true that it is the responsibility of the group that owns the data should not provide any information that could lead to identification of the individuals who participated in the study. However, it is extremely common to de-identify patient data, and should not be a problem for a company that routinely is sharing clinical data to scientists.

Not only is there an issue with the company not willing to share the full data set with either the co-authors or the third-party peer reviewers investigating the claims, but there also is a discrepancy in the reported numbers and their geographic locations.

Andrew Gelman, a statistician at Columbia University who reportedly started the discussion on the faulty data, wrote multiple blog posts and posted on PubPeer about the discrepancies in the data.

Gelman along with several other statisticians and medical professionals were concerned about the lack of information provided about the modification of the observational data during data analysis, and the surprisingly high mortality rate of patients who undergo hydroxychloroquine treatment. Another question arised from data collected out of Africa, where a uncommonly high percentage of COVID-19 positive patients participated in the study. The full concerns were addressed in an open letter to the authors of the original study and can be found here.

Furthermore, this is not the first article that Surgisphere has provided data for the risk of hydroxychloroquine and cardiac arrhythmia.

Surgisphere recently provided data for a New England Journal of Medicine article at the beginning of May, and is supposedly working on another article with Lee Wallace regarding COVID-19.

Wallace apparently feels confident about the dataset provided for his future article.

Wallis tells The Scientist that he has seen aggregated rather than hospital-level data, but that he is satisfied by Desai’s detailed descriptions of the dataset, and that all the necessary ethical and data-ownership requirements have been met.

Source: The Scientist

The founder of Surgisphere and a co-author on the original Lancet manuscript, Sapan Desai, reported this to The Scientist.

Desai says he can understand people’s concerns and that the burden of proof rests with Surgisphere. “We want to prove this to the world,” he tells The Scientist. “One thing that we might be able to do is get what we’ve done audited. That will be external, third party, independent of who we are, and can help validate all of this.”

Source: The Scientist

Surgisphere has also issued a response on their company website regarding their contribution to The Lancet article.

We thank the scientific community for their commitment to data integrity and for pointing out opportunities to revise and clarify our paper.

Since original publication in The Lancet, we discovered that a hospital that joined the Surgisphere registry on April 1, 2020 (in between our quarterly audit periods) self-designated as belonging to the Australasia continental designation. In reviewing the data from each of the hospitals in the registry, we noted that this hospital should have more appropriately been assigned to the Asian continental designation. This hospital was properly reclassified in our database. The findings of the paper are unaffected by this update.

We found that in the original publication, one of our data tables of propensity score matched and weighted data, was being misinterpreting as raw data, which would, if raw, make the data appear overly homogeneous. While the data were accurate, we are providing The Lancet and updated table with unadjusted data to help resolve the confusion. There was no error in analysis, and none of the results of the paper are affected.

Source: Surgisphere

I cannot find the updated table, and frankly, find it ridiculous to believe that the company will be able to solve all of the issues of this paper with one hospital incorrectly assigned and reporting unadjusted data. I have serious questions over Surgisphere and Desai as a medical proffesional and scientist. Until the dataset used for The Lancet is openly provided to the fellow authors and third party reviewers, we cannot trust the validity of the data. I would also suggest that any articles that use Surgisphere for their data analysis should also be questioned until the data can be validated for accuracy.

While it is now unclear as to the risks of hydroxychloroquine for COVID-19 positive patients, the evidence suggesting hydroxychloroquine is not effective for the prevention or treatment of COVID-19 is strengthening. I will be dedicating a blog post to this shortly, but you can read this randomized double blind placebo controlled study which will be the main focus of my next post.

As a final note, I would like to apologize to my readers for not catching this in my initial read and research of the article. I work diligently to provide scientifically accurate information, and will continue to do to the best of my ability.

One positive takeaway from this experience is that science at its core is a self-correcting process. Scientists around the world examine peer-reviewed articles every day and question articles that are not up to par. It is this fundamental drive to seek out the correct answer that drove me to a science career.

I will continue to post scientifically accurate articles, and if another post is found to be invalid, I will correct it in the same manner I have done with The Lancet post.