White coats, black deeds; the new scientific method: lie, cheat, and get good PR.

White Coats, Black Deeds

Late one afternoon in 1983, a senior mental health researcher named Robert Sprague was beginning to suspect that something was rong. Steven Breuning, a young doctor at the University of Pittsburgh, was studying drug treatments for hyperactive retarded children that contradicted the field's conventional wisdom. The established treatment called for using strong tranquilizers, but Breuning had concluded that taking patients off the tranquilizers could double their IQ scores. The state of Connecticut had already amended its treatment practices to conform to Breuning's groundbreaking research.

There was only one problem: Breuning's results were too good. In his studies, teams of nurses had subjectively ranked the violence of arm and leg seizures to gauge drug effectiveness. Improbably, Breuning's laboratory notebooks showed perfect agreement among different nurses' rankings. Worse yet, the notebooks recorded studies in one year that would have taken 270 working days--but Sprague knew there had been only 260 workings days and that Breuning was doing other time-consuming studies at the same time. The data had been faked.

Shaken, Sprague wrote a letter to the National Institute of Mental Health (NIMH), the federal agency that funded the fraudulent research. He returned to his own work, confident that the matter was settled, only to find months later that the NIMH was investigating his research, not Breuning's. Meanwhile, the university had learned of the possible fraud and launched its own, superficial investigation. Three months later, Breuning confessed and quietly resigned; the university then dropped its investigation--leaving the bad research unretracted. After another 14 months, the NIMH finally turned to investigating Breuning, taking a year and a half to conclude that the bulk of seven studies was faked. It cut off his research funding and recommended him to the Justice Department for prosecution. Until his conviction on federal fraud charges, however, Breuning continued ton conduct research for another mental health center. Sprague, meanwhile, found his own NIMH research funding cut, despite favorable peer reviews. A NIMH official described the cut as a "coincidence."

A disheartening experience? Absolutely. A typical one? Maybe not, but did you know that investigators from the National Institutes of Health (NIH) are checking into suspicions that Robert Gallo, one of the nation's premier biomedical scientists, stole credit for discovering the AIDS virus? The question of how much biomedical research is contaminated by misconduct--fraud, "nonstandard" research practices, commercial conflicts of interest, overstated conclusions, and careless error--has no concrete answer. When it occurs, however, misconduct often has concrete results. Consider the following cases:

* Under pressure to bring its blood clot-dissolving drug (known as tissue plasminogen activator, or TPA) to market before a competitor released a cheaper product, the biotechnology firm Genentech funded a major effort in conjunction with the NIH in which several laboratories studied the drug. Victor Marder, a member of the study's safety monitoring board, testified before Congress that when he and another board member raised questions about the dosages used in the TPA studies, the NIH dismissed his board and appointed a new one. Researchers twice increased the dosage without safety board approval, and several patients on the highest dose suffered brain hemorrhages. Five out of 311 died--3 within 24 hours of treatment. Years later, congressional investigators discovered that 14 scientists in the study or their close relatives owned Genentech stock.

* In 1986, Betty Eldreth, a 37-year-old woman with a history of breast cancer, underwent surgery at the Duke University Medical Center after a lump in her abdomen was diagnosed as cancerous. Upon cutting her open, Eldreth's doctors found no trace of malignancy. Eldreth had previously gone through two months of needless radiation treatment for cancer diagnosed in one of her ribs. Later, a doctor checking into Eldreth's case discovered that the misdiagnoses resulted from an experimental test known as B72.3, which William Johnston, a Duke pathologist, was researching and wrongly using as a final arbiter of cancer. Johnston, who is chief of the cytopathology and cytogenetics division at the Duke University Hospital, began using the test in late 1984; within a year, he was performing a hundred tests a month and receiving $15,000 a month in extra fees, despite instructions from the Pathology Department chairman to use the method "sparingly" (Johnston was later barred from taking the fees). In March 1987, Johnston signed a contract in which he agreed to supply a Massachusetts firm with B72.3 in return for a 20 percent commission on sales--money credited to his research account at Duke.

Duke has done its best to hush up the case, posting armed guards at the medical school to keep out a "20/20" camera crew and appointing largely ineffective committees to investigate the test's use. The university still has not explained how Eldreth's case occurred (she settled her lawsuit out of court). Meanwhile, if Eldreth develops cancer again (a distinct possibility), she will be unable to undergo any more radiation therapy, having already received maximum exposure.

* Scheffer Tseng, a Harvard Medical School fellow from 1984 to 1986, ran a clinical trial of a vitamin A lotion on "dry eye"--a condition caused by several diseases that can destroy the cornea and lead to blindness. Tseng never published his results. He had good reason not to, since they demonstrated that the vitamin A ointment he was using didn't work. Tseng and his clinical supervisor Kenneth Kenyon owned large blocks of stock in Spectra Pharmaceutical Services, a New Hampshire company formed to purchase patent rights to the vitamin A treatment. When Spectra went public in the middle of 1985, Tseng and Kenyon suddenly had strong incentives to cover up bad news. As the tests continued to show inconclusive results, Tseng's behavior became increasingly bizarre. He changed drug dosages and sometimes treated patients with additional drugs without their consent. There's no evidence that patients were harmed by any of this, but all of these practices were clearly unscientific and completely unethical. Today, Tseng, who says he "just followed orders" from Kenyon, conducts research at the University of Miami Medical School. Although he denies the story, The Boston Globe reported that Tseng and his relatives made more than $1 million from selling their Spectra stock.

* At the University of Pittsburgh, pediatric researcher Charles Bluestone accepted $17.5 million from the NIH to run clinical trials of amoxicillin, an antibiotic that Bluestone claimed was effective in treating children's ear infections. Meanwhile, Bluestone also received $3.5 million in research funds and more than $250,000 in honoraria and travel fees from pharmaceutical companies with interests in the study. When Erdem Cantekin, another researcher in the study, charged that Bluestone had wittheld data showing that amoxicillin was no more effective than other drugs, the university fired Cantekin and initiated a spurious misconduct investigation against him.

Drug money

As these examples show, people can get hurt when science goes bad. That's why it's crucial to identify and eliminate the cultural and commercial forces that foster scientific misconduct. "What is surprising . . . is that the system of science works sluggishly or actually works against a speedy, appropriate adjudication of suspected misconduct," Robert Sprague told a convention of the American Association for the Advancement of Science. "It is not only surprising, it is intolerable in a civilized society."

Study several cases of misconduct, and common threads begin to form ominous patterns. Government agencies fail to exercise proper oversight--like the NIMH in the Breuning case--while universities cover up well-supported allegations of fraud or bungle investigations. Committees write reports soft-pedaling conclusions that reflect badly on important scientists. Whistleblowers are blacklisted as troublemakers or shut out of scientific research--like Bruce Hollis, who as a junior professor at Case Western University was fired and pilloried in the local medica after he publicly accused his former supervisor of inept research practices. Prominent scientists take the Ed Meese line--like Duke's William Johnston, who argues his test wasn't experimental despite evidence to the contrary, and who claims exoneration because he didn't commit any technical violations of FDA or university regulations. Charges are proven true only after whistleblowers devote years of effort and make enormous sacrifices--as in the cases of Hollis, Sprague, and many, many others. It's hard to escape the impression that high stakes prompt prominent scientists and their institutions--what might fairly be called the "scientific establishment"--to forgive and dismiss instances of bad science far too easily.

Too often, misconduct is explained away as the result of individual character flaws. These explanations neglect how the scientific culture has been shaped by competition and money. Over the past 10 years, biomedical science has become big business. The biotechnology industry, which started up only in 1980, now includes more than 500 companies with annual sales of $1 billion. Last year, the NIH distributed nearly $7 billion in 5,300 new and continuing research grants--an average of $1.3 million per grant. The infusion of money and public attention brought on by the AIDS epidemic and advances in molecular biology, immunology, and medical research have changed the traditional incentives for scientific research. New tensions tug at the laboratories of major universities: competition fro grants, tenured positions, and high-profile committee assignments, as well as commercial incentives offered by biotechnology firms, such as consultation contracts, honoraria, and patent fees.

Rather than a pure end in itself, science is considered more like a commodity that can enhance a university's reputation, enrollment, or faculty. As Cornell sociologist Dorothy Nelkin documents in her book, Selling Science, scientists and university administrators now try to draw media attention for their discoveries--a direct result of increased competition for government and industry funding. Most major universities and hospitals employ full-time public relations staffs for this purpose. The commercial value of their scientific efforts makes it hard for universities to admit when something goes wrong. Scientists have internalized this defensive ethic, often refusing to admit even natural and entirely forgivable errors in scientific journals. "The literature is full of bits and pieces now known to be wrong, but it is not the tradition to point out each one publicly," Nobel laureate David Baltimore once wrote to a colleague. "The custom is that errors are buried and forgotten," says Charles McCutcheon, an NIH scientist with a long-standing interest in misconduct. Unfortunately, as the Genentech episode with TPA showed, in contemporary science such burials aren't always figurative.

Ties that blind

These pressures affect research in insidious ways. A typical biomedical laboratory costs between $500,000 and $2 million to run--and research funding isn't easy to get these days. Several years ago, NIH began to fund longer-term grants, which recently led to a shortage of money for new research projects and a decline in award rates to 29 percent. Scientists are feeling the pinch--and the temptation to exaggerate the significance of their results. "if your funding is cut off, you can teach, but you can't do any research," says Jonathan King, an MIT professor of molecular biology. "Scientific gossip these days is concerned with how to get funding and whether results are 'sexy' enough [for the NIH]."

In some universities, as many as 30 percent of biomedical researchers have consulting or other business relationships with biotechnology firms, according to Sheldon Krimsky, a Tufts University professor who maintains a database of such ties. Some researchers hold stock in companies whose products they evaluate. Despite the clear conflicts of interest posed by such ties, they are effectively unregulated. Last year, when the NIH proposed a rule which would have prohibited researchers with commercial ties from using federal funds for their work, it was withdrawn after intense lobbying by university and company officials and researchers. Carol Scheman, an analyst with the American Association of Universities, argues against such a ban on the grounds that university-industry ties are so close that the rule would have made research "nearly impossible to do." Well, if the tie is that close it deserves more, not less scrutiny. At the very least, full disclosure of such ties--something that is hardly the academic norm--seems called for.

Scientists have long claimed that extensive government regulation of their field is unnecessary because science contains "self-correcting" mechanisms. By publishing their work in peer-reviewed scientific journals, for instance, scientists make their results widely available to one another for comment and criticism. And because scientific experiments are replicable, error-ridden or fraudulent results should be quickly exposed.

In general, these scientists are right. Peer review and replication serve science well, particularly at the cutting edge of research where they efficiently filter out bad theories and experimental error. But especially when science is closer to the market place, these mechanisms sometimes work very slowly or break down altogether. Breuning, for example, was able to fake his research results for at least three years without detection--and even then, his fraud was discovered only after Sprague asked him for his patient records.

Dopes or dope pushers?

No one denies that scientific misconduct is a serious problem when it occurs. The disagreement is over how often it takes place, and consequently how serious a danger it poses. Most scientists hold that the problem is minor--like Daniel Koshland Jr., editor of the well-respected journal Science, who editorialized in 1988 that "99.9999 percent of reports are accurate and truthful." On the other hand, Harold Green, a law professor at George Washington University who represented a whistleblower in a science fraud case, recently told The Chronicle of Higher Education, "I hear from younger scientists that, in their opinion, there is an awful lot of fudging on research results."

Although no one really knows how much misconduct goes on, there are a couple of indirect estimates. Neither is particularly encouraging. One comes from the Food and Drug Administration, which randomly audits 5 to 10 percent of all clinical drug trials--often the last regulatory step before drugs are sold to the public. Given the immediate consequences for public health in these trials, you'd expect them to be exacting. Guess again.

Since 1977, FDA auditors found that at least a quarter of these studies were flawed by poor research procedures, failure to account for all of the drugs used in the trials, or inaccurate recordkeeping, according to Alan Lisook, who oversees the audits. Nearly 10 percent of the studies received official reprimands, a step which calls the research conclusions into question. In 12 years, the agency banned more than 50 scientists from testing experimental drugs because they submitted false technical reports or conducted careless research. Sixteen were criminally prosecuted, and at least four served jail time. Although these overall figures are slightly misleading--the actual number of reprimands has been dropping in recent years, a fact which FDA officials attribute to stricter supervision from the drug companies--the numbers imply that at least 200 drug trials since 1977 have been untrustworthy.

When NIH scientists Ned Feder and Walter Stewart studied the investigation of Harvard cardiologist James Darsee--who resigned in 1981 after admitting that he'd faked several heart studies--they found that 28 percent of his 47 coauthors had committed other serious errors, such as printing misleading statements, republishing old material under new titles, failing to acknowledge contributions from other researchers, or ignoring complaints that colleagues might be involved in "questionable data collection." In the most celebrated example, the published lineage of a family with a high incidence of heart disease included one 17-year-old man with four children aged four to eight--implying that he fathered his oldest child at age nine. Despite favorable comments from peer reviewers, Stewart's and Feder's manuscript languished for three years before being published, because journal editors who initially expressed interest in it received libel threats from some of the coauthors examined in the study.

The dweeb patrol

Once such cases began getting publicity, it was only a matter of time before Congress got involved. When Rep. John Dingell learned about the treatment of Robert Sprague, he turned the investigative powers of the House Energy and Commerce Subcommittee on Oversight and Investigations to probing cases of scientific misconduct. By now, probably no politician is despised as much within the scientific community as Dingell. Although Dingell has long supported funding for agencies like the NIH (and, in fact, has a brother who's a researcher in the Heart, Lung, and Blood Institute at the NIH), scientists widely believe that he is out to discredit the agency and the field of scientific research in general. "There's an internal struggle in Congress over which sectors of the budget are to be cut," said Jonathan King of MIT, who wrote a letter to the MIT faculty newsletter denouncing Dingell. "Dingell's investigation is the opening salvo in discrediting the NIH."

Dingell is used to provoking reactions like these. In the 14 years since he assumed the chairmanship of the Energy and Commerce committee, he has roasted General Dynamics for overcharging the Pentagon, survived FBI smears while investigating the nuclear power industry and Karen Silkwood's mysterious death, and forced accountants to regulate their profession when they were giving financially troubled businesses clean bills of health. Dingell's interest in scientific misconduct boils down to concern over how public money--like the $7 billion annually distributed by the NIH--is spent. He has never suggested a legislative remedy for misconduct, preferring to see universities and the NIH put their own safeguards in place--although he doesn't mind sowing a little constructive terror in the process. "If that's the only consequence--we're frightening them--I would say that's probably a desirable result," he once said.

Cell division

Uncovering misconduct is difficult enough, but a little media savvy can make a scientist's defenses even tougher to penetrate. When Dingell began checking into allegations of misconduct involving a paper published in Cell by MIT biomedical researcher David Baltimore and four co-authors, Baltimore's PR effort effectively cas the investigation as an assault on the basic integrity of science--a message easily picked up and amplified by journalists unequal to the technical complexity of the case. Baltimore's own work on the paper has never been questioned, but he was quick to assume the role of public spokesman for the co-authors once Dingell's investigation began in early 1988. As director of the MIT-affiliated Whitehead Institute for Biomedical Research, he had the resources and cachet to cover for Thereza Imanishi-Kari, the Tufts University scientist whose contributions to the paper were certainly erroneous and possibly fraudulent. Baltimore has lavishly praised Imanishi-Kari while doing his best to discredit Margot O'Toole, the postdoctoral student who first pointed out errors in the paper. Since research like Baltimore's could provide clues to combatting the AIDS virus, you'd think the integrity of his results would be his highest priority.

O'Toole grew suspicious after having trouble repeating one of Imanishi-Kari's experiments. After several months of frustrating attempts, she turned to the laboratory's old mouse breeding records, hoping to find an explanation for what she thought were her own errors. Instead she found 17 pages of data, many of them describing experiments that contradicted the results published in the Cell paper.

Initially, O'Toole took her concerns to her thesis advisor, but never got a fair hearing. Inquiries at MIT and Tufts dismissed her conclusions as "differences of scientific interpretation"--but no one at MIT ever examined the original data, and, according to O'Toole, the Tufts committee was more interested in protecting Imanishi-Kari's career than in learning the truth. O'Toole left Tufts after Imanishi-Kari asked her not to return, and has not worked in science since. O'Toole's steadiest job through the intervening years has been answering telephones at her brother's moving company.

The NIH's treatment of the case has been woeful. Its first investigative panel included two immunologists who had previously collaborated in research with Baltimore. A second NIH panel cleared the authors of fraud and misconduct while criticizing them for "serious errors and inadequacies" in their research--but it misconstrued a number of O'Toole's criticisms and ignored convincing evidence. For instance, Imanishi-Kari claimed that some of the data in the contested 17 pages, though mistakenly garbled, supported her published conclusions. But when a Dingell investigator unscrambled the data with a personal computer, they turned out to contradict her conclusions after all. The panel's report never mentioned that, and also unaccountably omitted words of praise for O'Toole which had existed in earlier drafts. Only after Dingell brought in the Secret Service to analyze Imanishi-Kari's notebooks was the NIH embarrassed enough to open a third investigation, this time with the intention of performing an extensive data audit.

These investigations may have had one dramatic if unintended effect--that of inducing Imanishi-Kari to cross the line from error into fraud. Allegedly, Imanishi-Kari provided the Tufts panel with unpublished data that, if true, would refute O'Toole's complaints. But forensic evidence provided by the Secret Service last year shows that these particular notebook pages, which bore dates from 1984 and 1985, were actually written later, at the time of the Tufts investigation--despite Imanishi-Kari's earlier assurance that all laboratory notes had been assembled while the research was going on.

Spin doctor

From the beginning, Baltimore's defense has been to cast Dingell as the Grand Inquisitor of Science. Shortly after Dingell's first hearings in April 1988, Baltimore launched a media offensive, circulating 400 copies of a "Dear Colleague" letter which supported Imanishi-Kari's story, reduced O'Toole's concerns to "alternate interpretations," and accused Dingell's investigators of using the case to "catalyze the introduction of new laws and regulations that could cripple American science." Philip Sharp, director of the MIT Cancer Center, took Baltimore's cue and distributed his own letter with the help of the Whitehead public relations staff. Sharp's packet included helpful "talking points" and asked scientists to write their newspapers and congressmen to protest Dingell's "witch hunt." These tactics succeeded famously. Dingell's subcommittee received hundreds of angry letters from scientists, few of them familiar with the facts, but all uniformly hostile to what they saw as a congressional attack on science. When Dingell called a second set of hearings on the case, the audience was filled with scientists friendly to Baltimore and hostile to O'Toole. Some snickered as O'Toole described her difficulty finding a job after leaving Tufts at Imanishi-Kari's behest.

Baltimore's PR didn't stop when Dingell called his co-authors to testify last May. Whitehead money paid for a prestigious Washington law firm to prep him for the hearings. In the last half-hour of a long day's session, Baltimore ambushed Dingell, saying the congressman had sprung forensic evidence on the authors (Dingell's staff actually presented the authors with the evidence in a five-hour meeting the day before) and unjustly accusing him of "fraud" (based on a misleading Boston Globe article in April 1988). As far as the press was concerned, this dramatic confrontation became the story, which suited Baltimore just fine. He'd spent much of the previous day criss-crossing the city to sell his story to reporters from The Washington Post, The New York Times, and The Wall Street Journal, and had spoken the night before to an uncritical audience at the D.C. Science Writer's Association.

Post-hearing coverage was overwhelmingly pro-Baltimore. The Washington Post wrote the case up as "Science Under Fire," emphasizing Baltimore's numerous scientific accomplishments and his victory over Dingell. The New York Times played the hearings straight during the week (which meant emphasizing the confrontation), but ran a "Week in Review" piece the following Sunday titled "Anxiety Over the Science Police" in which various scientists worried endlessly that Congress would "regiment" science and stifle creativity. The Wall Street Journal's Washington columnist Paul Gigot contributed the "Latest Chapter in the Fine Science of the Smear." Dingell's hometown newspaper, The Detroit News, castigated "Dingell's New Galileo Trial." Only The Boston Globe, which carefully examined O'Toole's allegations and evidence, avoided the stampede.

Like the scientific community itself, reporters have been reluctant to delve too deeply into technical issues of the case. But the two reporters who did attempt to penetrate the fog of scientific minutiae--Diana West of The Washington Times and Philip Weiss, writing in The New York Times Magazine--found a whole new layer to the story, one that led them to be critical of Baltimore. When Dingell revealed his forensic evidence during the hearings, for instance, most published accounts described it in point-counterpoint fashion: Imanishi-Kari's notes were out of order and had been redated, but she explained that she was a sloppy notekeeper who often recopied her work. Only West and Weiss realized that Imanishi-Kari's unpublished data--the linchpin of her defense--had been written up not only long after the paper was published, but while the Tufts panel was investigating it.

Journalists are susceptible to spin like Baltimore's because of the mystique associated with prominent scientists and the difficulty most reporters--even science reporters--have weighing the merits of scientific claims. Baltimore's defenders have been quick to emphasize his weighty scientific contributions, his "towering genius," and the complexity of the science--as if these factors had anything to do with whether Imanishi-Kari willfully misinterpreted her data.

In a fascinating postmortem twist, Baltimore was elected last June to chair the Scientist's Institute for Public Information, an organization which encourages the media to handle science issues in a "good and honest" fashion. Baltimore insists that the timing was a "sheet accident" (he previously sat on SIPI's board of directors) and that his experience with Congress won't in any way taint" his involvement with the media. Sure.

Lab test or self-examination?

While outright fraud is not difficult to recognize, most scientific misconduct is more complicated. Take David Baltimore. Can his eagerness to use his considerable influence to defend bad research be considered misconduct? Did his co-authorship of a paper make him responsible for the soundness of its research and conclusions? Amazingly, according to current NIH guidelines--the most widely applicable rules around--the answer to both questions is no. These rules condemn "fabrication, falsification, plagiarism, or other practices which seriously deviate from those that are commonly accepted within the scientific community for proposing, conducting, or reporting research." The catch lies in those practices which aren't commonly accepted," a squirmy phrase that reduces consideration of that vast gray continuum between fraud and honest error to a matter of "professional judgment," to borrow a term from Suzanne Hadley, acting director of the NIH's new Office of Scientific Integrity.

But determining misconduct requires more common sense than "professional judgment." Even a nonscientist knows it isn't fair to change the rules in the middle of the game--but that's what William Johnston did when he changed his experimental protocol while evaluating the B72.3 test. Everyone knows that your credibility suffers if you undertake an evaluation with a financial interest in a particular outcome--but the thought ever seems to have crossed the minds of Scheffer Tseng and Kenneth Kenyon as they tested their Vitamin A lotion, or of the scientists who evaluated TPA for Genentech while holding stock in the company. And it should be utterly obvious--so clear that it's almost embarrassing to make the point--that pointing out mistakes should get you reward, not blamed, blacklisted, or fired. The NIH guidelines clearly miss these points--as well as the larger one: Bad science can hurt people.

Two years ago, in response to several serious cases of fraud in the previous decade, the Harvard Medical School Distributed a set of recommendations specifying accepted research practices, so that research groups could develop their own written policies. As a result, when a Harvard visiting scientist faked the discovery of a new antibody called interluekin-4A in 1986, investigators wer equick to examine his raw adata and discover his fraud.

Federal funding agencies would be smart to follow Harvard's example. Here are more practices they should consider:

* Enunciate research and data-handling guidelines. Clear guidelines on laboratory procedures, statistical methods, and supervision of research trainees would make misconduct easier to identify. These agencies should also require researchers to hold on to their primary data for at least 10 years, making it available to other scientists upon request. Surprisingly, many scientists are hesitant to share their data. Baltimore, for instance, testified that he never asked Imanishi-Kari for her original data, because doing so would have signified a breach of trust.

* Enforce "primary authorship." Requiring that scientific papers bear a "primary author" responsible for the data and conclusions would not only add another layer of review, but also improve communication between authors, reducing the potential for misunderstanding and inadvertent error.

* Regulate conflicts of interest. Scientists and members of their immediate families should fully disclose any financial interests that might influence research. If misconduct as a result of commercial ties persists, researchers who maintain such ties should be barred from receiving federal funds.

* Make the bureaucracy work. More than a year ago, the NIH created the Office of Scientific Integrity as a clearinghouse for investigating allegations of scientific misconduct. It was a good idea, but OSI still has no permanent director and suffers from a lack of focus. (In this sense, OSI mirrors the generally disorgaized condition of the NIH, which also lacks a director.) The NIH needs to appoint an activist director who will exhort universities to adopt--and enforce--strict misconduct guidelines.

* Fund duplication of research. As things stand, federal agencies don't give grants for duplicating existing research, only for extending it. Changing this practice wouldn't be a bad idea. In fact, it would be easier in many situations to resolve allegations of misconduct by letting a whistleblower try to duplicate contested research than by forming traditional investigative committees.

* Hold a "scientific audit." Drummond Rennie, an editor of the Journal of the American Medical Association, has suggested that scientists conduct a confidential, "scientific" audit in order to determine the extent of research malpractice. In such an audit, retired scientists would exhaustively examine the work in a few hundred randomly selected laboratories. "No one has a clue how much fraud is out there," Rennie says. "We have to come up with a better answer than that we don't think it's very common." Although scientists haven't been particularly excited by the idea, Rennie thinks it would be a deterrent to further Congressional investigation. "I believe that Congress would be impressed by scientists getting scientific about science this way," he says.

None of this will work, of course, if institutions don't follow their guidelines. That's where the government comes in. Dingell's current bad-cop routine is probably the best way to get institutions to take the problem of scientific misconduct seriously. But if universities and hospitals continue to react like Duke or the University of Pittsburgh, more government intervention will follow. If it becomes necessary, the least disruptive course of action would be to institute random audits of clinical research, like the audits the FDA currently conducts.

"Science flourishes best if scientists are given broad latitude in ordering their own affairs," wrote Bernard Davis, a Harvard Medical School emeritus professor, last year in The Wall Street Journal. Maybe so. But the cases of Baltimore, Johnston, Tseng, and the rest suggest that they're not doing a very good job. Scientists owe it to themselves to address the issue of scientific misconduct. Nothing is clearer than that Congress will do it for them, given the threat to patients and the pressures to spend federal funds wisely. If researchers want to keep the "science police" out of their laboratories, they had better straighten them up themselves. And soon.

David P. Hamilton is a reporter-researcher at The New Republic.