Scientists Under the Microscope

Making up results is the worst sin a researcher can commit. But a new report questions whether smaller, more common misdeeds could be corroding the integrity of scientific endeavors

At its 2000 annual meeting, the North American Association for the Study of Obesity in Silver Spring, Maryland, honored physiologist Eric Poehlman of the University of Vermont with the Lilly Scientific Achievement Award. During the ceremony, the tenured professor presented his data that detailed how metabolism tanks, muscles dwindle, and body fat plumps as men and women age.

But Poehlman’s talk, it turns out, was laden with lies. Last March, the Office of Research Integrity (ORI), part of the U.S. Department of Health and Human Services, announced that he had cooked his spreadsheets, the worst kind of crime in a profession that relies on honesty in its mission to uncover truths. Over a decade, the researcher fudged or invented data that he published in 10 articles in respected journals and cited in 17 applications for federal grants, which ultimately netted him roughly $2.9 million in funding.

“It’s probably the most egregious case that we’ve had,” says lawyer Christopher Pascal, ORI’s director. “He was basically stealing funds from the government based on a misrepresentation.” Poehlman was fined $180,000, agreed to retract or correct the faked articles, and faces possible jail time. He is the first scientist ORI has barred for life from receiving federal money.

The Poehlman case isn’t the only scandal that recently has put the scientific profession under the microscope. Over the past 2 years, news reports have revealed that several dozen scientists at the National Institutes of Health (NIH) failed to properly disclose lucrative consulting arrangements with drug and biotech firms. And last fall brought startling headlines that the blockbuster painkiller Vioxx boosts the odds of heart attack, triggering a furor over whether the drug’s maker and the Food and Drug Administration kept the pill on the market despite having evidence of its dangers.

These headline grabbers raise the question of how trustworthy the scientific enterprise is. Although outright fraud such as Poehlman’s seems to be rare, a new Minneapolis study--the first of its kind--suggests that other, lesser transgressions that involve cutting corners or stretching the facts aren’t unusual among the lab-coat crowd. Such habits might undermine the integrity of science, warn the survey’s authors. Although the study drew criticism for certain shortcomings, it highlights a potential problem that some say the profession needs to address to maintain the confidence of its members, its funders, and the public.

Federal law narrowly defines scientific “misconduct” as fabrication (making up data), falsification (manipulating data), or plagiarism (using another’s ideas without credit). According to Pascal, ORI fields on average 150 to 200 allegations of misconduct per year, but since 1992 it has confirmed fraud in only 162 cases--a “tiny fraction,” he says, of the hordes laboring at the bench. By one estimate, the numbers add up to about one case of misconduct per year for every 100,000 working scientists.

But the new survey, published last June in Nature, indicates that the incidence of misconduct is higher. Querying 3200 postdocs and young professors who received NIH funding, sociologist Brian Martinson of the nonprofit HealthPartners Research Foundation and his colleagues at the University of Minnesota found that 3 in 1000 researchers admitted to the big no-no of falsifying data, and about 1 in 100 said they had plagiarized ideas.

Those numbers are still low. Yet the misdeeds didn’t stop there. Survey takers anonymously confessed to a broad range of questionable practices. For instance, Martinson notes, more than 10% said they used study methods that they knew were inappropriate or turned a blind eye to the failings of colleagues who used flawed data. These mundane misbehaviors present a greater threat to the scientific enterprise than high-profile fraud does because they can make research “a lot less efficient, a lot less trustworthy,” Martinson says. If incorrect findings creep into the published literature, for example, they could mislead the public with, say, the wrong advice on what to eat--or spur other labs to waste time trying to confirm something that isn’t real.

But critics counter that the Minnesota report itself bends the facts. Several questions were vague, making responses hard to interpret, says cell biologist Kerry Bloom of the University of North Carolina, Chapel Hill. For instance, about 15% of the participants acknowledged “changing the design, methodology, or results of a study in response to pressure from a funding source.” That behavior could be bad if the study were underwritten by a pharmaceutical company that pushed the investigator to fix the methods or results to exaggerate a drug’s benefits. But when the funder is NIH, Bloom says, grantees are expected to tweak their research strategies in response to reviewers' critiques.

Another finding whose significance is debatable concerns the handling of scientific information: 15% of survey takers reported dropping numbers from analyses “based on a gut feeling” that the data points were off. These responses raised a flag for ORI’s Pascal, who says that scientists can fall too easily into the trap of trying to clean up results to make them look better.

But cell biologist Frederick Grinnell, who founded the Ethics in Science and Medicine Program at the University of Texas Southwestern Medical Center in Dallas, points out that data dropping is a gray area, because different fields vary in their accepted practices. For instance, in clinical trials testing a drug’s safety or effectiveness, each patient’s response could be crucial, so investigators aren’t allowed to ditch any data on a hunch. But not so in cutting-edge laboratory research, Grinnell says, where information from one experiment suggests the direction of the next. When a result seems suspect, “very often your intuition turns out to be the very best way of deciding whether [it] counts or not,” he says. And in Grinnell’s lab, he says, “we throw out data all the time”--and redo the experiment.

Martinson stands by his interpretation; he says survey takers understood that answering “yes” generally meant admitting misbehavior. And facing up to these minor sins, he says, will ultimately benefit the research enterprise. Scientific organizations have long emphasized the fact that flagrant misconduct in the lab is rare, Martinson says. But the focus on the few “bad apples,” he says, has prevented discussion of how recent changes in the world of research--with intense competition, a thicket of regulations, and greater emphasis on commercialization--might contribute to a decline in scientific integrity. Researchers have told Martinson that they’ve made ethical compromises under the pressures of juggling experiments, teaching, and compliance with regulatory rules, all while scrambling for grants. “If you’re working as hard as you can, and working as smart as you can, what are the options left to you to get ahead?” he says. A junior professor might, for instance, cherry-pick his data to get the positive outcomes he needs to win his next research grant.

No easy solution exists. But when it comes to proper scientific conduct, education and self-scrutiny are crucial, says Susan Ehringhaus, deputy counsel of the Association of American Medical Colleges. Institutions and professional groups--not ORI--should define what types of minor transgressions are unacceptable and teach those standards, she says. Such courses on ethical conduct are now common in many universities. These programs are an important part of the answer, says Martinson, but they aren’t enough: He would also like to see the scientific community as a whole openly discuss the systemwide forces at play, perhaps enabling researchers to address the problem at its roots--identifying the pressures that drive investigators to cut corners.

In the meantime, researchers need to continue keeping an eye on one another: Poehlman was caught when a conscientious lab technician realized that the professor had produced data from nonexistent patients. Such deception can be nearly impossible to catch. But one saving grace, Bloom says, is that even egregious misconduct--or milder offenses such as those ferreted out by Martinson--probably won’t hinder scientific progress in the long run. Scientific exploration is by nature self-correcting. If an important discovery is real, other labs will quickly reproduce it and build on the finding, Bloom says, so the truth will ultimately prevail.

Ingfei Chen is a freelancer in Santa Cruz, California. She drops punctuation based on a gut feeling.