Tattered Threads

Abstract

Eugene Viscusi gave his typically superb talk on perioperative analgesic management last December at the Post Graduate Assembly in New York. Midway through his talk, he presented a series of articles from Anesthesia & Analgesia by Dr. Scott Reuben showing the long-term beneficial effects of perioperative nonsteroidal antiinflammatory drug administration. I listened in silent horror, knowing that the articles in Anesthesia & Analgesia were likely fraudulent. I had known for months of an ongoing investigation at Baystate Medical Center. However, because of the confidentiality of the proceedings, and the lack of a formal statement of findings, I could not say anything. I sat silently as Dr. Viscusi misinformed his audience based on fraudulent work published in Anesthesia & Analgesia. Should I have spoken up, violating the confidentiality of Baystate’s investigation, and besmirching the reputation of a prominent investigator based upon unsubstantiated allegations? Or did I do the right thing keeping quiet, knowing that physicians in the audience might alter their therapy and expose patients to unnecessary risks based on fabricated data? Richard Feynman1 eloquently observed that “nature uses only the longest threads to weave her patterns, so that each small piece of her fabric reveals the organization of the entire tapestry.” Scientists and clinicians are weavers as well, examining the incremental bits of knowledge published in the peer-reviewed literature for trends and patterns. We weave these bits of evidence into our knowledge of physical reality. It does not matter whether the goal is to understand planetary motion, evolution, global warming, or perioperative analgesia; the interlocking of many threads of evidence gives beauty, elegance, and intellectual strength to the interwoven tapestry of human understanding. With this issue of Anesthesia & Analgesia, and concurrent issues of other journals in our specialty, a thread has been ripped out of that tapestry. Twenty-one articles and abstracts spanning 15 years of investigation have been identified as fraudulent (Table 1). The fabric of our understanding of perioperative analgesia has been badly damaged. In an editorial in this issue of Anesthesia & Analgesia, Drs. White, Kehlet, and Liu have attempted (on very short notice) to summarize what we still know, and what we do not know, about perioperative analgesia.2 This is merely our initial effort to repair the tapestry. The broken thread is deeply woven into many review articles, meta-analyses, lecture summaries, and the memories of every individual who attended Dr. Viscusi’s outstanding Post Graduate Assembly presentation, as well as many other continuing medical education programs on the effect of perioperative and multimodal analgesia on patient outcome.Table 1: Articles with Fabricated Data Identified by Baystate Medical CenterScientists make mistakes. Mistakes are not only tolerated, but are a necessary part of science. The perfect study cannot be performed. Even the most meticulous article is only approximately correct. Our measurements are imprecise. Our models are crude. We do not measure every variable that affects the outcome of an experiment. That is why we are forever seeking better experiments, and results and conclusions seem to change as additional experiments are performed. We are forever weaving new bits of information into our understanding. The evolution of the tapestry slowly, inexorably, corrects for scientific error. Intentional fraud is not part of the scientific process. However, the same self-correcting mechanism that slowly eliminates experimental error will remove fraudulent results, whether or not the fraud is exposed. The complex ethical issues regarding scientific fraud are magnified by the scale of this case, permitting closer examination. Baystate Medical Center released to Anesthesia & Analgesia a list of articles based on fabricated data. For ethical and legal reasons, Baystate was able to release to Anesthesia & Analgesia only the articles we had published. It was by comparing notes that the Editors-in-Chief of the affected journals were able to construct a complete list of 19 articles and two abstracts, and to start to heal the tapestry. Baystate has confirmed that the list in Table 1 is complete and has included the same list in reports to the Office for Human Research Protections, and the Food and Drug Administration. Dr. Reuben was a prolific writer for many years. Baystate Medical Center has indicated that the documentation for many published investigations was often incomplete. This made validation of data impossible, particularly for some older studies. Despite its best efforts, including two complete reviews of all articles, and the full cooperation but limited recall of Dr. Reuben, Baystate was not able to reconstruct the scientific trail to make conclusive findings on any of the balance of Dr. Reuben’s published writings. Our position at Anesthesia & Analgesia is that those articles remain part of the unimpeached literature, at least for now. However, since Baystate could not reconstruct the scientific trail supporting these studies, the extensive history of data fabrication casts a shadow over all of Dr. Reuben’s work. These studies may also need to be retracted. We are presently attempting to independently verify the integrity of Dr. Reuben’s nonretracted publications through discussions with his collaborators. Fortunately, his collaborators and coauthors have cooperated fully with the investigation, and all have been exonerated of any involvement in the data fabrication. What about the peer review process—how did it fail to miss 21 fraudulent submissions over 15 years? The process failed because science is based on trust. We trust authors to honestly report their experiments and their findings. Reviewers rarely consider the possibility of outright fabrication. For example, the original figures submitted by Dr. Reuben with his article on long-term patient outcomes for outpatient anterior cruciate ligament reconstruction (Table 1, Ref. 20) were of poor quality. As Dr. Reuben was leaving on vacation, and I had a deadline to meet, I asked him to send me his data. Using his “raw data” spreadsheet, I created several of the figures that appeared in the final article. Despite working with his data, the possibility of fabrication never entered my mind. Data fabrication by a sophisticated cheat is almost impossible to detect. Science is difficult. Society places enormous financial, regulatory, and ethical demands upon scientists. Despite the profound failure of the peer-review system to identify 15 years of fraudulent research, in my view it would not be appropriate to impose additional regulatory and institutional burden on scientists. There is no reason to think that additional levels of oversight would be effective. For 15 years Dr. Reuben’s data fabrication was missed by his coinvestigators (who were exonerated in Baystate‘s investigation), his reviewers, and journal editors (including me). What magical powers would an oversight committee need to detect this fraud? Although I am unwilling to recommend fundamental changes in how science is conducted, the Editorial Board of Anesthesia & Analgesia will consider what steps we might take as a journal to prevent this in the future. I am not willing to take steps that further chill scientific progress or impose additional burdens on investigators. However, the journal can always do better. We will engage in some introspection on how we can improve the review process at Anesthesia & Analgesia. On behalf of the Editorial Board of Anesthesia & Analgesia, I deeply apologize to those clinicians who were misled by the fraudulent articles published in Anesthesia & Analgesia, and to patients receiving inappropriate treatment as a result. This incident serves as a somber reminder of our responsibility to contribute accurate threads to the tapestry of knowledge that guides patient care. The tapestry will survive and eventually heal. I hope that is also the fate of any and all patients injured by the tattered thread.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 64 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.