Too Good to be True11/08/2012
A Japanese scientist is believed to have fabricated up to 200 studies. Würzburg anesthesiologists have long been suspicious of him, having voiced concern about conspicuous data as early as twelve years ago. But their warnings fell on deaf ears. Now the issue has grown into a much bigger scandal.
"Stop your crusade against Prof. Fujii. You should re-focus on more productive work." This and a lot of similar advice was given by peer reviewers to Peter Kranke and his co-authors of the Department of Anesthesia and Critical Care at Würzburg University Hospital more than ten years ago when they critically examined the reported results of the Japanese researcher Yoshitaka Fujii in an article and were going to publish obvious discrepancies between those results and the findings of other study groups. This piece of advice was even one of the polite ones. For instance, another reviewer complained to the editor of the journal that he considered reviewing the article to be a waste of time. This was in 2001.
Biggest purge under way
What is known today: Kranke and his co-authors were right. The case of suspicion has now developed into one of the greatest medical research scandals in recent years. Up to 200 of Fujii's papers are believed to be fabricated; more and more journals are retracting his papers; the prestigious journal Nature speaks in a comment of "one of the biggest purges of the scientific literature in history", the record number of retracted papers rocking the scientific community.
"When I was collecting data for my doctoral thesis, I soon had the sobering experience that it is very difficult to find suitable patients for clinical studies," Kranke recalls, who is now professor of anesthesiology at the Department of Anesthesia and Critical Care of Würzburg University Hospital. Therefore, the members of the study group at Professor Norbert Roewer's Department inevitably started to wonder about the enormous number of Fujii's publications in various journals.
168 papers were published by Fujii between 1991 and 2011 – averaging eight per year. At peak times, the Japanese researcher had almost 30 publications in various journals without raising much suspicion. Only Kranke asked himself how this could be possible and therefore decided to examine Fujii's studies more carefully.
Large numbers of identical data
On closer inspection, a number of particularities soon caught his eye: In his studies, the Japanese researcher had often compared several drugs for preventing nausea or vomiting after general anesthesia in patients. Fujii found that a certain drug, namely granisetron, had significant advantages in comparison with similar drugs.
When Kranke statistically compared the results of 47 of Fujii's publications more than twelve years ago, a number of irritating similarities struck him as odd: The values of certain parameters and the reported frequencies of side effects were often absolutely identical in different tables. "For instance, the side effect 'headache' was identically distributed in independent patient groups," says Kranke.
Such congruence is usually not found in patient-related studies, even if you assume that the event rate is not influenced by the drugs under examination. "If this is reported in several studies, it becomes even less probable," Kranke explains.
Furthermore, the results on the efficacy of the examined drugs differed significantly from the findings of other study groups; they were simply too good to be true – not only with respect to the efficacy data.
Critical article without impact
The Würzburg researchers, however, were cautious about denouncing the research outright as fraudulent and fabricated. Instead, they published their results in the journal Anesthesia & Analgesia with the comparatively moderate headline "Reported Data on Granisetron and Postoperative Nausea and Vomiting by Fujii et al. Are Incredibly Nice!"
In the journal Acta Anaesthesiologica Scandinavica, they drew attention on the lacking consistency of Fujii's efficacy data with the findings of other authors (Kranke et al. "The influence of a dominating center on a quantitative systematic review of granisetron for preventing postoperative nausea and vomiting"). "We basically intended to point out that assessments of drug effectiveness should rather be based on the side effect and efficacy data of other studies," Kranke explains.
The reply was disappointing: Fujii merely declared in his response that he stood by his data, but without going into any detail; the scientific community chose to be satisfied with this. Kranke was advised to turn his attention to "more serious issues".
Recent studies set the ball rolling
Over the next decade, the Japanese researcher continued publishing a great number of papers in various journals without anybody taking issue. Only in March this year, the scandal began to unravel: At this point, the British anesthesiologist John Carlisle published an article in the journal Anaesthesia, demonstrating that 168 of Fujii's papers had results with near-zero likelihoods.
One month later, the editors of 23 journals in the field of anesthesiology reacted: They informed the heads of six universities and medical institutions with which Fujii had worked that altogether 193 papers would have to be retracted unless the respective institutions could guarantee the veracity of data. Five of them were not able to do so; the sixth one succeeded only in a fraction of cases.
These revelations brought to mind the early warnings from Würzburg. Steven Shafer, editor-in-chief of Anesthesia & Analgesia expressed his regret in a personal letter to all subscribers over the journal's failure to follow-up the issue after Kranke's article. The response was "inadequate". The subsequent submissions by Dr. Fujii should not have been published without first conducting a thorough investigation of the obvious indications of fraud.
According to Peter Kranke, there were several reasons why it was still possible for Fujii to continue publishing obviously invented data over a period of so many years: "He continually sought new institutions to work with, changed his co-authors and submitted his papers to many different journals," says Kranke. Furthermore, his results were never completely at odds with the prevailing opinion: "The drug examined by Fujii is good. It's just not quite as good as he claimed it to be," says the anesthesiologist. On top of that, the Japanese researcher was never available for discussions: "He never attended any conventions. All our efforts to get in contact with him proved unsuccessful," Kranke recalls.
A flaw in the system
There still remains the question how such a case of fraud could have happened. Kranke does not consider it appropriate to simply shrug it off as an "isolated case" and to resume day-to-day business. Certainly, it is presumptuous to lay the responsibility solely on faulty incentive systems. However, we need to think about "flaws in the system" as well. There is a lot of pressure in a system where scientific achievement, measured by the number of publications in scientific journals, plays such a decisive role with respect to funding, careers and livelihoods. Thus, scientists can easily get tempted to boost their career in a dishonest way, especially if they are in temporary employment with a follow-up job at stake.
In Kranke's opinion, as a lesson to be learned from this and similar cases of fraud, we should always pay proper attention to the assessment and critical review of scientific results. "Even the most obviously faulty system of incentives does not justify dishonest work," Kranke concludes.
Looking back, is there any bright side to the "story"? "Anyhow, the review system intrinsic to science basically worked out in the end," the medical scientist points out. Moreover, this case of fraud doesn't seem to have caused any significant harm to patients. Kranke is well aware, however, that this is only some small comfort, considering how long it took to finally confirm the initial doubts about the conspicuous data.
Prof. Dr. Peter Kranke, MBA, T: +49 (0)931 201-30050