TESTIMONY BEFORE THE DINGELL SUBCOMMITTEE 12 APRIL 1988

Statement of

Walter W. Stewart, J.F.
and
Ned Feder, M.D.

National Institute of
Diabetes and Digestive and Kidney Diseases
National Institutes of Health

Before the
Subcommittee on Oversight and Investigations
of the
Committee on Energy and Commerce

12 April 1988

Mr. Chairman, we are pleased to be appearing before the House Subcommittee on Oversight and Investigations. We are gratified that the National Institutes of Health has allowed us to carry out as an official duty much of the research we report here today and has allowed us to appear here as NIH scientists. We wish to emphasize, however, that in our testimony we are presenting our own views as scientists and are not acting as spokesmen for the NIH, PHS, or DHHS.

The two of us are scientists who work as full-time federal employees in a laboratory at NIH. We synthesize chemical compounds that are used to study the shape of nerve cells, and we carry out basic research on the genetic control of nerve cell shape. Over the past four years, we have also conducted basic research on the difficult subjects of the accuracy of the scientific literature and of scientific misconduct.

Why should anyone want to spend $5 billion on biomedical research? It is because of the extraordinary and unpredictable benefits that flow from basic research.

Basic research is the study of nature -- mathematics, physics, chemistry, biology, and their related fields. Judged purely as an intellectual discipline, science is one of the most exciting and successful ventures of mankind. But there is an additional remarkable feature that is neither simple nor obvious but has proved consistent: discoveries in basic science have led, often in completely unexpected ways, to improvements in human health and to material benefits. We have been relieved of some our burden of suffering, and our society is immensely richer, because of discoveries in basic science.

One example is basic research done years ago by Dr. Bruce Ames on the genetics of the synthesis of histidine in a bacterium. This research, widely respected at the time for its quality, appeared to have no practical value. Some ten years later it formed the basis of a remarkable advance: a cheap and rapid test to identify carcinogens. It is referred to as the Ames test in honor of its discoverer. This is a clear and simple example of the extraordinary practical benefits that steadily come from good basic research of no apparent practical value.

Mr. Chairman, we are pleased to be appearing before the House Subcommittee on Oversight and Investigations. We are gratified that the National Institutes of Health has allowed us to carry out as an official duty much of the research we report here today and has allowed us to appear here as NIH scientists. We wish to emphasize, however, that in our testimony we are presenting our own views as scientists and are not acting as spokesmen for the NIH, PHS, or DHHS.

The two of us are scientists who work as full-time federal employees in a laboratory at NIH. We synthesize chemical compounds that are used to study the shape of nerve cells, and we carry out basic research on the genetic control of nerve cell shape. Over the past four years, we have also conducted basic research on the difficult subjects of the accuracy of the scientific literature and of scientific misconduct.

Why should anyone want to spend $5 billion on biomedical research? It is because of the extraordinary and unpredictable benefits that flow from basic research.

Basic research is the study of nature -- mathematics, physics, chemistry, biology, and their related fields. Judged purely as an intellectual discipline, science is one of the most exciting and successful ventures of mankind. But there is an additional remarkable feature that is neither simple nor obvious but has proved consistent: discoveries in basic science have led, often in completely unexpected ways, to improvements in human health and to material benefits. We have been relieved of some our burden of suffering, and our society is immensely richer, because of discoveries in basic science.

One example is basic research done years ago by Dr. Bruce Ames on the genetics of the synthesis of histidine in a bacterium. This research, widely respected at the time for its quality, appeared to have no practical value. Some ten years later it formed the basis of a remarkable advance: a cheap and rapid test to identify carcinogens. It is referred to as the Ames test in honor of its discoverer. This is a clear and simple example of the extraordinary practical benefits that steadily come from good basic research of no apparent practical value.

Because of the deep connection between science and human progress, the U.S. government has wisely invested considerable sums in support of basic research. Science in the United States has succeeded remarkably. It is our clear impression that both scientists and nonscientists take pleasure in the success of modern American science.

It is precisely because of the trust placed in scientific research that reports of scientific misconduct are deeply disturbing. Until recently it was customary to assume that such cases of misconduct were extremely rare. However troubling each case might be, it seemed likely that the damage to science was small and that the only real damage was to the public perception of science.

The cases of scientific misconduct reported in the past few years in the popular press are one reason to question the accuracy of this view. Others are studies suggesting that misconduct is not as rare as previously thought.

We recently performed and published such a study. Despite its limitations, it is clearly relevant here. We studied the professional practices of a group of 47 scientists who were unwitting coauthors of a scientist later shown to have committed extensive scientific fraud. We reached the disturbing conclusion that about a third of them had published misleading statements or had engaged in similar departures from accepted standards of scientific research. Our sample was not chosen randomly, and because of this and limitations inherent in the way we did the study, it is not possible to generalize from this figure.

It is correct to say that neither we nor anyone else has accurate figures on the frequency of scientific misconduct. Considering the importance of the subject, this is regrettable.

The scientist who cheats in his research violates the basic precept of science, which is to find the truth and make it known. A scientist who cheats injures other scientists by misleading them, and he injures the public by wasting their money and sometimes by falsely claiming discoveries that directly affect the public health and welfare. Similar damage can be done by the scientist who, through irresponsibility or carelessness, publishes erroneous research or fails to correct it. Finally, there is the monetary cost of "junk research" that is known at the time it is done to be of little or no value.

Science has reliable and time-tested methods for dealing with error. These include replication and peer review. These mechanisms, however, were not designed to detect or correct deliberate misrepresentation. Indeed, conventional science has hardly considered this problem, so that individual cases have been handled with improvised methods. Often these methods have failed disastrously.

We discuss here five cases we studied in which aspects of the system did not work well.

Dr. John Darsee was a promising young researcher in cardiology at Emory and Harvard medical schools. In May 1981 he was caught fabricating data. Eventually almost all of Darsee's published work, much of it supported by NIH, was found to be grossly defective. The extent of Darsee's fraud was shown by an NIH-appointed committee headed by Dr. Howard E. Morgan. Based on the results of the committee's painstaking investigation, Darsee was barred for a period of ten years from receiving DHHS research funds.

There were two earlier and unsatisfactory investigations. The failure of these investigations, conducted entirely within Darsee's own institution, provides a valuable lesson. For a period of five months after Darsee confessed to a single act of fraud, his research was investigated secretly by his two supervisors, who continued to collaborate and publish with him during this period. These distinguished scientists found no further evidence of fraud during their investigation.

Soon after NIH learned of the problem, the dean of Harvard Medical School appointed the Ross committee, a group of eight distinguished university professors who conducted the second investigation. Its report indicated that none of the work Darsee published at Harvard contained fabricated data. The NIH committee headed by Dr. Morgan soon showed that these eminent scholars had reached a conclusion diametrically opposed to the facts.

How did this happen? First, most of the Ross committee worked at Harvard. Their close ties to the university may have affected their ability to pursue a fraud that would necessarily reflect poorly on Harvard. Second, although the committee members were distinguished scholars, they lacked experience in investigating fraud. Finally, the committee failed to look at any of the laboratory data. The evidence of fraud was all around them, but the Ross committee somehow failed to notice it.

We now consider the case brought to light by Dr. Robert Sprague, from whom the committee will hear today. In late 1983, he wrote a letter to NIMH describing in detail his evidence and conclusion that his associate, Dr. Stephen Breuning, had committed extensive scientific fraud in NIMH-supported research on the drug therapy of mental retardation. Eventually Dr. Sprague's allegations proved entirely valid; indeed the final committee report added surprisingly little to the evidence that he had already provided more than three years before. What is disturbing is that it required three and one-half years of intense work on Dr. Sprague's part, pressuring NIMH directly and through Congress and the press, before the NIMH completed its thorough but unduly prolonged investigation and made its findings public. During this period, patients were exposed unnecessarily to the questionable treatment advocated by Breuning on the basis of his fabricated research. The drugs involved in Breuning's research often have serious side effects. This is a graphic and compelling illustration of the direct danger to the public posed by fraudulent research. A disturbing point and one that is not well known was the chronic and growing friction between Dr. Sprague and NIMH as NIMH proceeded slowly to public disclosure of the problem.

We note here that when Dr. Sprague requested continuing support for his research from NIMH, his request was unanimously approved by a panel of his peers. Nonetheless, NIMH deferred approval of his request for further support, and by its action left him without NIMH funding for the first time in 17 years. Ultimately, after an outcry in the press, NIMH rereviewed his proposal and reinstated his funding at a considerably reduced level for a period of one year in order to allow him to wind down his research. The inadequate institutional performance contrasts with Dr. Sprague's persistent and principled conduct.

We wish to draw attention to the fact that Dr. Sprague was a Principal Investigator on the grant used by Breuning. He had supervisory responsibility for Breuning's actions and therefore for his fraud. By drawing attention to the fraudulent actions of a close colleague working on his own grant, Dr. Sprague was necessarily exposing his own actions to minute scrutiny, which in fact occurred. Clearly, Mr. Chairman, the actions taken by Dr. Sprague required integrity and unusual courage; he served the public well.

These are of course our own views, not necessarily those of NIH, PHS, or DHHS. There may be a satisfactory explanation for these disturbing events of which we are not aware. Representatives from NIMH are here today and may give their comments on these events.

We next describe the case of an important paper in molecular biology whose accuracy has been challenged by us and by Dr. Margot O'Toole, who is here today to give testimony. The paper was authored by six scientists and published in the scientific journal Cell. In May 1986 Dr. O'Toole, a senior postdoctoral fellow in the laboratory of Dr. Thereza Imanishi-Kari at MIT, while carrying out her assigned laboratory duties came across records suggesting to her that the paper published in Cell contained serious errors. Dr. O'Toole called the discrepancies between the laboratory records and the published paper to the attention of Dr. David Baltimore, Dr. Imanishi-Kari, and Dr. David Weaver, all of whom were authors of the paper. She also spoke to several officials at MIT and Tufts. She wrote a memorandum dated 6 June 1986 analyzing what she considered "serious weaknesses" in the published paper and presenting her scientific reasoning. She urged a published correction of the errors in the paper, but this was not done.

Two committees, one consisting of Dr. Eisen at MIT, and one headed by Dr. Wortis at Tufts, examined the allegations in private and concluded that no published correction of the questioned paper was necessary. Dr. O'Toole was told the final decision of the committees, but was not given a copy of the reports until recently. She was discouraged by Dr. Baltimore from pursuing the matter further.

We first heard about the dispute from Dr. Charles Maplethorpe, who is here today. Acting entirely on our own initiative, we contacted Dr. O'Toole and requested a copy of the original laboratory records. We emphasized our view that a scientist with private knowledge of serious defects in a published paper has a professional obligation to make those defects known. Indeed, this was listed as one of the ten generally accepted standards of scientific research in our manuscript on professional practices, since published. The manuscript had at that time been read by several hundred scientists and not one had raised an objection to this point. Although we recognized that she was in a difficult position, we urged that she had a obligation to comply with this standard, which had been implicitly approved by many scientists. After careful consideration, Dr. O'Toole sent us a copy of the laboratory records.

We analyzed them and concluded that the published paper contained a number of serious misrepresentations of scientific fact. The discrepancies between the laboratory records and the published paper were particularly disturbing for the following reason: the precise agreement of 14 numerical values showed that in certain cases the laboratory records contained the data that were actually published. In a number of crucial cases, the published paper gives an inaccurate and in fact misleading picture of the underlying data.

After carefully analyzing the evidence and reaching the provisional conclusion that there were serious problems, we contacted the authors of the paper and described our reservations to them. Dr. Imanishi-Kari and Dr. Weaver indicated that Dr. Baltimore would speak for them. Dr. Baltimore refused to discuss the substance of Dr. O'Toole's allegations of scientific error or of our analysis which confirmed and extended her conclusions. He instead attacked her character and motives, and he characterized us personally in various ways. He made threatening statements and suggested that he would resort to legal action. He stated that the Eisen and Wortis reports showed Dr. O'Toole's allegations to be false. Despite repeated attempts, we were unable for a long time to obtain these reports.

Dr. Baltimore proposed that yet another committee of immunologists examine whether the norms of science had been transgressed. We would have had to promise in advance that if the committee did not agree with us, we would never discuss the matter further in public or in private. We suggested instead a committee more consistent with the tradition of free and open debate, one where all parties would retain the right to discuss the issues. This was not accepted.

We drafted our scientific conclusions in a manuscript. We asked and obtained reviews of this manuscript from many scientists. None suggested that our manuscript contained errors of fact. We experienced difficulty in obtaining reviews and found that very few scientists were willing to express themselves publicly on the accuracy of our manuscript. We eventually obtained NIH permission to submit our manuscript to a scientific journal. We submitted it to Cell, the journal that published the original article coauthored by Dr. Baltimore. Cell rejected it without the customary review. We then submitted it to Science, which also rejected it without review. The main reasons given for rejection were that our submission was not a scientific manuscript and that it could not be subjected to peer review. We do not find either assertion credible. We have submitted our manuscript informally to several other journals, all of which rejected it without review.

Most recently, we have begun an analysis of the Eisen and Wortis reports. Our preliminary findings are that the reports contain facts which confirm some of Dr. O'Toole's assertions. The remainder of Dr. O'Toole's assertions are either ignored or treated evasively. We consider both reports so seriously defective that they are not useful analyses of the paper published in Cell. This is particularly disturbing in light of the scientific credentials and distinction of the two scientists who prepared the reports. It is also disturbing that Dr. O'Toole was not promptly given a copy of the reports.

Mr. Chairman, it is important to recognize that those whose views and actions we criticize are not here today to explain or defend their course of conduct. Thus it would not be possible at this time to reach a final evaluation of the events described in our testimony. Similarly, the issue of the scientific accuracy of the paper is contested and may well remain so until all relevant facts are submitted to the scientific public for analysis and for free and open debate on the merits of the issues.

Despite the fact that it is too early to reach a judgement on any of these events, there are three disturbing issues that must eventually be considered besides the accuracy of the Cell paper itself. The first is the issue of the Eisen and Wortis reports. If our scientific evaluation of these reports as seriously inadequate proves correct, this will be a matter of deep concern. The second issue is also a sensitive one: the response of Dr. Baltimore to the allegations of scientific error and his conduct toward the scientists who made the allegations. The third is the reluctance of so many scientists to express their views on the scientific accuracy of our manuscript and the difficulty we have had in finding a journal willing to consider our manuscript on the basis of the accuracy of its scientific assertions.

Our next case involves allegations about Dr. Hector DeLuca and some of his scientific associates at the University of Wisconsin. Dr. DeLuca has received extensive scientific recognition for his research on vitamin D. There were a number of allegations, but we discuss here only those involving his work on the synthesis of a medically important vitamin D derivative called 1-alpha-HCC, work supported in part by NIH. We emphasize that at present these are simply allegations and that the alleged misconduct has not yet been adjudicated.

Allegations surfaced because of litigation between two private parties. We served as expert witnesses in our private capacity for the party attacking the validity of Dr. DeLuca's patent. We were not paid for this work. Our efforts were focussed on discovering the truth rather than furthering the commercial interests of either party.

A Cambridge-based research institute, RIMAC, filed a lawsuit against the Wisconsin Alumni Research Foundation (WARF), which held patent rights on Dr. DeLuca's research on the synthesis of 1-alpha-HCC. During the litigation, RIMAC and some of its expert witnesses alleged that the synthesis did not work, that laboratory notebooks had in some cases been altered, and that the defects in the patent had been concealed by subsequent misrepresentations in the scientific literature. They further alleged that Dr. DeLuca and his associates had misappropriated a synthesis devised by Dr. Derek Barton and Dr. Robert Hesse and had misrepresented it as their own work.

The lawsuit was settled out of court in June 1987; in return for a sum of money, RIMAC relinquished its U.S. patent rights to WARF and agreed to join WARF in asking the court to dismiss the suit. The amount of money paid by WARF to RIMAC has been ordered sealed by the court. On 11 March 1988 an article on the allegations of scientific misconduct appeared in The Isthmus, a weekly Wisconsin newspaper. A few days before the article appeared and nine months after the case had been settled, the University of Wisconsin announced that it was about to appoint an investigating committee.

During the two-year period of litigation, the University of Wisconsin did not begin a misconduct investigation. The university may have refrained so as not to interfere with the judicial process. It is also true that the university had a financial stake in the outcome of the litigation. In any case there was clearly a tension between commercial interests and the university's responsibilities as an academic institution. A cause for concern is the nine-month delay between settlement of the litigation and the initiation of the investigation.

Another case we have recently begun to study involves an allegation by Dr. Steven Barlow. Dr. Barlow alleged that his former supervisor, Dr. James Abbs at the University of Wisconsin's Speech Motor Control Laboratories, had published fabricated data as part of his NIH-supported research. Dr. Barlow's allegations were investigated by a panel appointed by the University of Wisconsin, which found that the charges were without merit. The NIH has formally accepted the university's report and has closed the case. We have examined some of the data on which the allegation was based. After seeing the data, and a part of the response made to the allegation, we urged the NIH, thus far without success, to open its own investigation of the allegations.

We have discussed here many of the cases we have studied. They have often been chosen because they appeared to us important. These cases represent a small fraction of all cases of alleged misconduct, they are not a random sample, and they may not be representative. Despite the limitations of our experience, we offer below some observations based on our studies.

One of the patterns in the cases we have studied, Mr. Chairman, is that scientists alleging misconduct are often treated poorly. They have been subjected to criticism and to personal and professional retaliation, even in cases where their allegations have ultimately proved valid. There seems little doubt that the unfavorable reception affects scientists' willingness to report misconduct. We have talked with considerably more than 20 scientists who have alleged misconduct to us privately, but have been reluctant to make their allegations through NIH or institutional channels. Their reluctance strongly suggests to us that the reported cases of misconduct represent only a small fraction of the total. It is our considered estimate that we have learned of about 100 allegations of misconduct over the past year. Many of these allegations concern past episodes of misconduct, and we have no easy way of determining what fraction of these allegations are meritorious. It is our belief that we receive most of these allegations under the expectation that they will be kept strictly confidential.

Some have told us that their reluctance is due to fear of the professional consequences. Another reason often given is the belief that no effective action will result. We do not assert that either of these beliefs is always correct. We do think, however, that these beliefs are common and that they reflect a lack of confidence by many scientists about the efficacy of the current systems.

Because of the unjustified harm that has sometimes occurred to those alleging misconduct, it is understandable that consideration should be given to acting on anonymous allegations. Indeed NIH rules currently provide for the possibility of anonymous allegations. Though this rule is a commendable attempt to protect those alleging misconduct, we feel that the use of anonymous allegations runs counter to the scientific tradition of free and open debate. It may well encourage unwarranted allegations. It is a matter of considerable satisfaction that Dr. Sprague and Dr. O'Toole, both of whom are here today, have taken direct personal responsibility for every allegation they made. We suggest that the use of anonymous allegations may endanger the fabric of trust between scientists.

When we have seen failures in investigations, they have often involved one or more of the following four problems: secrecy, lack of experience in investigations, a conflict of interest, and a lack of speed. Here is an example: certain investigations of alleged misrepresentation have not even included an inspection of the underlying data on which the questioned publication was based. Secrecy is sometimes said to be necessary to protect the reputations of the accuser or of those unjustly accused. We question this argument: rumors travel quickly and can have profound effects. It is our impression that the present system for resolving allegations is sometimes unfair to the accused, the accuser, and the public.

The importance of expertise in the particular field of science in which the allegation was made has been, in our opinion, overstated by many. In our experience, many allegations can be technically evaluated without special competence in the field. Furthermore, those outside the field are less likely to have a conflict of interest. It is our opinion that such conflicts affect the outcome of investigations far more often than is generally realized.

Systems for dealing with allegations of scientific misconduct are relatively new and will undoubtedly change and improve in the future. What directions should these changes take? This is a difficult question to which we have no guaranteed solution. Certain approaches, however, clearly carry a great risk of doing more harm than good. Among them are direct regulation or policing of scientific research by authorities outside of science.

Research is a difficult, intricate, and important enterprise. The complex fabric of collegial interaction, which depends on trust, i8 necessary for scientific progress and is easily damaged. Great caution should be taken not to implement measures that will weaken instead of strengthen science.

We suggest that strong preference be given to measures that are fully consistent with accepted principles of science and particularly with the principle of free and open debate. Ideas that might be considered are:

bulletthe recognition that error is inevitable in the process of scientific research and should not be automatically criticized or condemned;
bulletincreased public support for those alleging or admitting scientific error;
bulletbetter forums for the discussion and evaluation of contested matters of scientific fact;
bulletcareful distinction between allegations of scientific error and allegations of scientific misconduct; this distinction is not always unambiguous, but it should be made whenever possible;
bulletprocedures for handling allegations of scientific misconduct that incorporate the scientific method; in particular, procedures which are designed to be fast, open, free from conflict of interest, and which employ some persons with experience in investigating misconduct;
bulletincreased public support for those who have made meritorious allegations of scientific misconduct;
bulletbasic research on the questions of misconduct in science and the accuracy of the scientific literature.

These are some ideas we think should be studied. We hope that scientists themselves will take the lead in analyzing and solving the problems. There are signs this is beginning to happen. We have received hundreds of constructive and thoughtful letters from scientists, many of whom are distinguished senior scientists; there were surprisingly few letters of vigorous opposition. We do not mean to imply that our colleagues would always agree with our specific findings, but by their actions they have endorsed the subject as important.

In our opinion, there are two powerful forces for improvements -- the impetus provided by the results of careful investigation, and the stimulus of free and open debate. These are the essential tools of progress in science. We are pleased that we can participate.

Back