Friday, February 04, 2011

Policing academic fraud

A big scandal is brewing in good ol Germany. A former hospital based clinical and researcher, Joachim Boldt, was fired by his employer some time last year when it came to light that a study he had published in a US medical journal was forcibly retracted because the journal had sufficiently serious doubts about the veracity of the data. 

The editor of the journal published this editorial today:

1. There are no original patient data or laboratory data to support the findings in the study.
2. According to the head of the perfusionist team, no albumin has been used as a priming solution since 1999.
3. According to the pharmacy, no albumin has been delivered to the cardiac operating rooms for many years.
4. All laboratory measurements, including IL-6, IL-10, intercellular adhesion molecule, neutrophil gelatinase-associated lipocalin, and alpha-glutathione-S-transferase, would have been performed in the clinical laboratory at the Klinikum Ludwigshafen. These assays have only been performed on patients receiving hydroxy-ethyl starch priming solutions. The laboratory could identify no assays from patients receiving albumin priming solutions.
5. Professor Boldt has admitted forging the signatures of the coauthors on the copyright transfer form submitted to Anesthesia & Analgesia.
6. The coauthors denied participation in the fabrication.
7. There is no convincing evidence that this study was performed at all.

The state medical board also investigated the good doctor. A press released issued today states: 

Based on today’s announcement, LÄK-RLP has reviewed 74 scientific articles describing
clinical trials subject to the requirements of the German Medicinal Act. This includes the article
by Professor Boldt recently retracted by Anesthesia & Analgesia and an article submitted by
Professor Boldt to Anaesthesia but not published. By law these studies required IRB approval.
Although the articles typically stated that IRB approval had been obtained, LÄK-RLP could not
find evidence of approval for 68 of these articles.
LÄK-RLP also identified 30 published articles that describe research that did not fall under the
jurisdiction of the German Medicinal Act but physicians performing such human research must
conform to the Code of Deontology, which includes a requirement for IRB review. Some articles
describing epidemiological studies with non-identifiable data did not require IRB review. For six
articles LÄK-RLP was able to document IRB approval. For the remaining articles LÄK-RLP has
not been able to document IRB approval.
LÄK-RLP has again contacted and asked the authors of published articles for which LÄK-RLP
cannot document IRB approval to provide evidence of such approval. LÄK-RLP is presently
awaiting documentation of IRB approval from the first/corresponding author of every questioned
article before making a final determination. LÄK-RLP will notify the respective journals of
articles describing clinical research for which there is no evidence of IRB approval.
Few comments:
1) it is odd indeed that co-authors on the initially published study claim to have not been 'involved' in the fabrication of relevant data. Well, does it mean, they did know about the fabcritation, but didn't technically contribute to the falsification? Or does it mean that he or she was co-author on a paper of which they had no idea how the basic data had come about? Or does it mean that because the good Prof Boldt forged their signature under the copyright release form, they had no idea they were even listed as co-authors? The latter is not very plausible a claim, because every researcher checks on google scholar or PubMed every now and then how their publications are doing by way of citations etc. Surely, a paper one is unaware to have contributed to would have popped out like a sore thumb then. Is it possible, in other words, that these folks really were unaware of their publication with Joachim Boldt. It is possible. Is it very likely? Not really. 
2) While this case is pretty terrible, it's a bit unclear what consequences should be drawn, procedurally, by journal editors or oversight bodies. Realistically journal editors are unable to investigate such fraud. How should we check whether or not eg a co-author signature is real or fraudulent? Similarly for the ethics committee (in the USA IRB) approval documentation. Some journals ask simply for confirmation from the principal author that ethics approval was given. If an author then lies, there is little we can do. Some journals ask for the approval letter from the IRB, but truth be told, anyone with even a rudimentary understanding of Photoshop and the necessary criminal energy, could forge such documents. And, as Boldt's case demonstrates, these sorts of folks are not exactly lacking criminal energy. This is not an easy one then to solve, because with the necessary infrastructure and resource deployment most such frauds could be caught, but realistically most publishers of academic journals are in it for the profits, and this would deplete such profits in a quite significant way. What should we do? We could unload the task to the employers of researchers. They could be required to keep tabs on their researchers' publications and cross-check them against their ethics approvals. If the necessary penalties were on the book for institutions failing to monitor their staff conduct sufficiently, I suspect we might be able to reduce unethical research. However, would this, or any other solution, completely eliminate academic misconduct of the type described? No chance. 

No comments:

Post a Comment