Wednesday, February 6, 2013

Academic Fraud

Thirty years ago I wrote an unpublished medical novel, White Squirrel, some of which is in the process of being reworked into my third novel, The Dana Twins and Other Lies. One story line from White Squirrel which will not be worked into The Dana Twins and Other Lies involved academic fraud. When I set out to write fiction after a thirty year hiatus, I wondered whether or not this theme would resonate with readers of a novel, and decided that it might not.
The story of a drug company or a manufacturer falsifying research to make millions has become a staple of modern crime shows. Typically, in those dramas, a whistle blower is murdered to conceal the truth and the investigator-protagonist steps in at that point. Since the fraud involves huge amounts of money, the dishonesty is easily understood by even the most unsophisticated viewer.
Academic fraud, on the other hand, occurs on a much smaller scale. Real science is built in small pieces, one brick at a time, and the stakes in academic fraud are usually much smaller. Academic fraud often involves someone falsifying science in order to generate a series of publications, or to advance their career – gain tenure, or to please a mentor. To my knowledge, no one has ever been killed to conceal academic fraud, though lives and careers have been destroyed. The following tales happened approximately thirty years ago, and while they were known in the academic community at that time, they have faded into obscurity and a retelling seems appropriate, certainly more appropriate than creating a fictional fraud just to write a novel about it.
In the early 1980’s, Dr. John Darsee published an impressive series of papers while a medical resident at Emory. After becoming a research fellow in cardiology at a Harvard teaching hospital, Darsee was observed running a rhythm strip (EKG recording) on an animal, then breaking the recording into pieces and labeling them with different dates and times, making it seem that they were parts of different experiments. As reported in TIME magazine, and elsewhere, several hours of work appeared to be several weeks of data.
A preliminary investigation at Harvard found nothing out of the ordinary, except for what was apparently a single aberrant action. However, when the overall body of Darsee’s data was found to be inconsistent with results obtained by other investigators working on the same project, Harvard was criticized for claiming to have completely reviewed data which were – in essence – non-existent.
By the time the entire sham was discovered, Dr. Darsee had been offered a faculty position at Harvard based on his productivity. That offer was rescinded. Darsee was banned from receiving NIH research funding for a period of ten years. He left research, took a fellowship in critical care medicine, and, as best I could learn, is in practice as an internist in Indianapolis, Indiana.
One of the interesting themes of the Darsee story is that internal investigations are often flawed by the fact that the people who do the investigating are the people at the institution who are most familiar with the research issue in question, i.e. the people who should have been supervising the study. Finding a major problem in such a review would mean finding oneself guilty of not providing adequate supervision. Hence, since one is, in essence, investigating oneself, there is a vested interest in not finding academic fraud. That theme repeats itself in other stories of scientific fraud. Ironically, a NY Times article from December 15, 1981 praised Dr. Eugene Braunwald, Dr. Darsee’s supervisor at Harvard, for acting with speed and responsibility to perform a thorough investigation. That, of course, was before all the facts were known. At that time, the investigation had uncovered only the single aberrant act by Darsee at Harvard and had reached the conclusion that Darsee was a prolific and brilliant researcher whose work at Emory was not in question. That latter statement was true only until someone thought to look at Darsee’s previous work at Emory. Overall, it seems that Darsee learned at an early point in his career that it was easier to publish research data if one omitted the tedious step of actually doing the research. 
At the end of the Darsee affair, numerous papers from Harvard and Emory were retracted. Darsee’s supervisors were criticized for providing lax supervision, and the prestigious New England Journal of Medicine took the opportunity to criticize Darsee’s co-authors for attaching their names to work with which they were unfamiliar. While Darsee’s chicanery was not the only force behind a greater concern for assigning authorship on scientific papers, it was one of them.  
 Perhaps the most prominent fraud in medical science involved Dr. William Summerlin, working in the lab of the noted immunologist Dr. Robert Good, a scientist who had been on the cover of TIME magazine in March of 1973.
In 1973, Dr. Summerlin claimed that by growing cells in tissue culture he could graft incompatible cells into mice, the most dramatic example being a graft of cells from a black mouse into a white mouse. The work appeared to revolutionize transplant immunology. I remember being at the national presentation of Dr. Summerlin’s work in Atlantic City when my program director said, “That’s the most amazing thing I ever heard – if it’s true.”
The house of cards collapsed when an animal caretaker examining one of Summerlin’s mice found that the black color could be washed off with alcohol.  
As documented in the book The Patchwork Mouse, Summerlin was involved in a series of “interesting” experiments. For example, corneas can be transplanted from one individual to another because the site is protected from the immune system. This is not true when the corneal transplant involves the limbus, the area where the cornea joins the sclera. In addition to the black/white mouse experiments, Summerlin was running a study in which one eye of a dog allegedly had a corneal transplant involving the limbus without the tissue culture step. As expected that eye became cloudy. The other eye, the control, had a limbus type corneal transplant, allegedly with the tissue culture step. That eye was clear. There was just one problem. The clear eye was actually an eye that had not had any surgery and the cloudy eye was an eye that had undergone a transplant. In essence, Summerlin’s entire body of work was apparently fraudulent.
Summerlin’s background was in dermatology, not immunology, and he claimed that he was “under pressure” to produce good results. One is left to wonder why such an important group of experiments was entrusted to a dermatologist without formal training in immunology.
What I have found most interesting is the scene described by the authors of The Patchwork Mouse in which one of the animal caretakers brought the mouse to a young surgeon at Sloan-Kettering and showed him how the color “washed off.” Without hesitation, the surgeon, one of my medical school classmates, Dr. John Raaf, brought the animal to Dr. Good’s office. Informed by a secretary that Dr. Good was meeting with the Board of Directors, or some equally important group of functionaries, Dr. Raaf allegedly responded, “Get him out of there; this is important.”
The Summerlin story illustrates the critical point that if you give too much responsibility to an untrained individual, he may have to choose between failing or fabricating data. He may choose the latter option. That theme is echoed in the story below.
The last tale of academic fraud that I wish to share is my favorite because, as the story has been reported, no one listened to the female whistleblower. Thus, the tale is not only one of academic fraud, but also a case of the “old boy’s network” versus a woman who was intellectually wronged. I personally think that this tale has potential as a stage play. Unfortunately, as the case played out in real life, the ending was too abrupt to be optimally dramatic, but, with a little literary license, it might work.
In essence, the story is this. Dr. Helena Washslicht-Rodbard (hereafter referred to as Dr. Rodbard) and her supervisor, Dr. Jesse Roth, at the NIH performed a study measuring insulin receptors in women with anorexia. The women were restudied after they had been treated with behavioral therapy and had clinically improved. The details and the results of the study are not critical to this discussion.
The Rodbard and Roth paper was submitted to the New England Journal of Medicine (NEJM). As is the policy of all journals, the NEJM sent the paper to reviewers, in this case three reviewers. One of the reviewers was Dr. Phil Felig of Yale, who, as per relatively standard practice, gave the paper to Dr. Vijay Soman, an investigator in his lab, to review.
Dr. Soman recommended rejection of the Rodbard and Roth paper; the other two reviewers recommended acceptance after revision. Shortly thereafter, Soman and Felig submitted a similar paper to the American Journal of Medicine and, since the work was in a somewhat esoteric field, it was sent to Dr. Roth to review. As might be expected, he shared the manuscript with Dr. Rodbard, and a major scandal was brewing.
Dr. Rodbard immediately recognized that the Methods Section of the paper seemed to be taken word for word from her manuscript. Obviously, there was an issue of plagiarism. Since the methods section tells how something was done, the excuse was offered at some point, that Dr. Soman “borrowed” Dr. Rodbard’s description of her techniques since they had done the same thing and her description was “perfect.” The second issue was one of propriety. Dr. Rodbard guessed, correctly, that since Soman and Felig had “borrowed” her Methods Section, they had, obviously, seen her paper. Clearly, they had been one of the reviewers. In fact, they had given the review which recommended rejection to the Rodbard and Roth paper which presented work similar to their own work. 
In science, primary discovery, even in a very limited field, has intellectual if not monetary value. By recommending rejection of the Rodbard and Roth paper, Soman and Felig increased the chance that their work would be published first. The question was raised as to whether Soman and Felig should have notified the editors at NEJM of this potential conflict. Since Felig (at Yale) and Roth (at the NIH) were long-time friends, (not only professional acquaintances, but actually boyhood friends), they agreed to meet and review the two papers. When that was done, the plagiarism was obvious. To resolve the problem, an agreement was worked out whereby priority would be resolved in favor of Rodbard and Roth. The Rodbard and Roth paper would be published; the Soman and Felig paper would be published after Rodbard and Roth had published, and it would cite the work by Rodbard and Roth.
Things might have ended simply at this point, except for one thing. On further review, Dr. Rodbard became suspicious that Dr. Soman had done more than “borrow” her methods section. Her suspicions had a major impact on the careers of several individuals and on two major academic institutions.
Both the paper by Rodbard and Roth and the paper by Felig and Soman studied anorexic women at baseline and after their anorexia had improved. The subjects at the NIH had been part of an intense psychotherapy program which was responsible for their clinical improvement. The therapist was a co-author on Rodbard and Roth’s paper. Soman and Felig’s subjects had “improved” to a similar degree. However no statement was made as to what had been done from a psychiatric point of view to lead to the improvement. Also, there was no co-author from the Department of Psychiatry on their paper. Dr. Rodbard brought her concerns to a Dean at Yale and also to the New England Journal of Medicine.
The Dean at Yale (Dr. Robert Berliner) accepted reassurances from Dr. Felig that all was well with the work published under his name and Dr. Soman’s name. The Dean informed Dr. Rodbard of that fact. Not surprisingly, Dr. Rodbard was less than pleased with the results of the investigation. 
Somewhere during the process, Dr. Rodbard’s supervisor (and co-author), Dr. Roth, essentially told her to cease and desist, and not to use NIH stationery in promulgating her complaints – which he seemed to regard as a personal vendetta. There is of course no way to know how his opinion was biased by his long term friendship with Dr. Felig at Yale. 
When I write fiction, I tend to see scenes, and I imagine Dr. Roth telling Dr. Rodbard to calm down, and Dr. Rodbard replying that important principles were involved and that instead of her calming down, he should be getting excited. Considering that he is in a senior position, and male, and she is the junior person and female, one can imagine the rest of the dialogue on one’s own.   
In any case, with a controversy smoldering, and legitimate questions about the adequacy of the Yale internal review, the New England Journal of Medicine sent an associate editor (Dr. Jeffrey Flier) to Yale to interview Dr. Soman, the junior faculty member who had seemingly duplicated the work of Rodbard and Roth. Almost as soon as Dr. Flier began his review, he became suspicious of Dr. Soman’s work. Data appeared to be missing, and where there was data, it looked too good to be true. Stated simply, in laboratory studies, points may lie near a straight line, but not exactly "on the line."
Dr. Flier challenged Dr. Soman about the data and Dr. Soman quickly confessed that he had “made up” some of the results due to a tremendous pressure to produce results and publish quickly. Flier had uncovered in a few hours what an internal Yale review had failed to find in a year. 
The fallout from the fraud brought the scandal to the public eye, leading to the telling of the tale in the prestigious journal Science. Prior to Soman’s confessing to the fraud, Dr. Felig, who was Chief of Endocrinology at Yale, had been hired as Chief of Medicine at Columbia. Felig had recommended that Dr. Soman be appointed a member of the faculty at Columbia. With the fraud revealed, Felig informed Columbia about what had happened and told Columbia that Dr. Soman could not be offered a position. He did not mention, at that time, the failure of the previous internal Yale investigation to uncover the fraud.
When all the facts came to light, Columbia became aware that an outside auditor quickly found what the internal Yale review (by Dr. Felig and one of the Yale Deans) had not uncovered. Dr. Felig was asked to resign as Chair of Medicine at Columbia since it was felt that the entire affair tarnished his ability to lead by example. Dr. Soman had, by this time, resigned from his position at Yale and returned to his native India. Twelve papers that Dr. Soman had published, eight of which had been co-authored by Dr. Felig, were retracted. Dr. Felig returned to Yale as a tenured Professor but without a named chair. (Since I am a professor, but do not have a named chair and do not aspire to one, I cannot say how much of a penalty that represents.)
Dr. Rodbard left the NIH. However, she was, as best as I can determine, in the midst of her training and not in a permanent position. She went into private practice in endocrinology. When I taught about academic fraud in the History of Medicine course at Vanderbilt, Dr. Rodbard was one of the heroes and I was happy to learn that in the intervening decades she went on to serve as president of both the American College of Endocrinology and the American Association of Clinical Endocrinologists.
Dr. Jeffrey Flier, the external investigator who uncovered the fraud, also has had a distinguished career. He went on the become Dean of the Faculty at Harvard Medical School.
The cases of Dr. Summerlin and Dr. Soman are examples of a junior person functioning without adequate supervision and illustrate why internal reviews often fail to detect obvious problems. Whether academic fraud is more common or less common now as compared to thirty years ago is unknown. Academic fraud is hard to measure.
People have attempted to measure academic fraud by counting "results which cannot be replicated." However, as scientists know, we generally take a probability level of 0.05 (1 in 20) as a level of statistical significance. This means that if a finding is positive at a level of 0.05, there is a 1 in 20 chance that the finding is random. Failure to replicate a result does not mean that something untoward was involved. It may simply mean that the initial observation was random.
We do know that academic fraud has not disappeared. Fifteen years ago, the most common condition for which a stem cell transplant was done was breast cancer. The value of the procedure was unclear as all of the studies showing a benefit for high dose chemotherapy in conjunction with stem cell transplant involved historical controls which were not truly similar to the patients undergoing the transplant procedure. Since there was legitimate concern over the value of the procedure, several randomized concurrent studies were performed to demonstrate whether or not stem cell transplant in breast cancer was of clinical value. All but one of the studies showed no benefit; the doctor conducting the study showing a positive result confessed to manipulating his data.