PHILADELPHIA — PHILADELPHIA - In criminal proceedings in the United States, the prosecution is legally required to provide the defense with any information that might be beneficial to the defendant - even if it may hurt the prosecution's case. Scientists are bound to a similar type of disclosure through an unwritten code of ethics. It's part of what separates real science from pseudo-science or folk wisdom.
Real science requires "a kind of utter honesty," the late physicist Richard Feynman said in a 1974 address titled "Cargo Cult Science."
"The idea is to try to give all of the information to help others to judge the value of your contribution; not just the information that leads to judgment in one particular direction or another."
Society expects the research backing the drugs we take and the medical procedures we undergo to abide by such high scientific standards and ethics.
But a number of critics say that, in the world of medical research, important results are going unnoticed or unpublished - especially results of clinical trials where drugs don't perform as hoped or expected.
Two cases have brought the issue to light recently. First, New York Attorney General Eliot Spitzer filed a civil suit against giant GlaxoSmithKline for failing to adequately publicize studies indicating that the antidepressant Paxil may lead to severe emotional problems and possibly suicide when given to children. Then The New York Times and other publications exposed a clinical trial that was never published, showing that a similar drug, called Celexa, worked no better than a placebo for relieving depression in adolescents, while a much-publicized trial reached the opposite conclusion.
These are not isolated cases, said Kay Dickersin, a professor at Brown University's Center for Clinical Trials and Evidence-Based Healthcare. Over the years, she has collected data that she says illustrate a systematic bias in the medical literature. Studies showing that drugs work are three times as likely to be published as those showing they don't work or do more harm than good, she said.
That doesn't mean there's no safety net out there to protect society from bad drugs. The Food and Drug Administration won't approve a new prescription drug until the manufacturers demonstrate its safety and efficacy in specified applications in a regimented series of experimental trials. But in recent years, the rules have been relaxed and the number of required trials decreased in an attempt to speed drug approvals.
"I wish consumers knew about this," Dickersin said.
And once a drug is approved, the law doesn't stop doctors from prescribing it for other "off-label" uses or, in the case of antidepressants, for children and teenagers when the drugs were approved for adults. Yet negative results of trials related to those additional uses of a drug may not have been available to prescribing doctors or follow-up researchers because they were never published. The result is that decisions are made based on incomplete information.
In many cases, it's not that the journals refuse to publish negative results, Dickersin said, but that researchers never write them up or send them to the journals.
Sometimes, when researchers try to publish their results, the companies that sponsor them try to intervene. That happened in 2000 when Immune Response Corp. sued James Kahn, a University of California, San Francisco AIDS researcher, for defamation after he and a colleague published results showing that the company's experimental AIDS treatment did not help stave off AIDS symptoms or bolster the immune systems of the subjects. The company argued that Kahn's team failed to highlight some positive effects seen in a few of the research subjects. The lawsuit was later dropped.
Private companies have a tradition of keeping trade secrets, and for drug companies, some secrecy may be essential to compete in the marketplace. But critics say secrecy is inappropriate when drugs are being tested on human volunteers or the information withheld could have helped patients or doctors make better decisions.
Brown's Dickersin decided to trace all the drug trials conducted at the Johns Hopkins University as well as those conducted by the National Institutes of Health. Many of the ones that reflected badly on a new drug or treatment were left unpublished, she said, though the science was sound.
That bias can become amplified when researchers comb through the published literature and combine known studies in a meta-analysis. These collections of studies often try to re-examine the risks and benefits of common health measures - such as hormone replacement, cholesterol drugs or vitamin pills.
"If you only publish good news, the whole system breaks down," said Drummond Rennie, a deputy editor at the Journal of the American Medical Association. "I've been fighting for this for decades."
He attributed the shortage of negative results to researchers either bowing to pressure from the drug companies that fund them or trying too hard to please their sponsors.
That can lead to such situations as that recently seen with hormone-replacement therapy, Dickersin said. It was approved for hot flashes and other symptoms but was being prescribed for thousands of women under the assumption that it prevented heart disease and other chronic illnesses.
In the mid-1990s a group of European researchers did a meta-analysis that included unpublished work and found that, for many women, hormone replacement's risks may outweigh its potential benefits. Several years later a large clinical trial came to the same conclusion.
Why would different trials of the same drug give different results? Don't drugs either work or not?
It used to be that drugs that worked had a clear and powerful effect - penicillin clearly killed bacteria, and insulin clearly helped diabetics control their blood sugar. Today, even a blockbuster drug may ease pain or depression only a little better than a placebo and may not work at all in some patients.
And there are concerns. Say there are two identical pills, called A and B. In theory, both should be equally effective because they are identical. But given enough trials, Rennie said, a few will by chance show that A is better than B, and a few will show that B is better than A. If researchers want to sell B, they could publish only the results that make it look superior, ignoring the ones in which A comes out ahead.
No one is asking the drug companies to publish every scrap of information on every chemical they test for drug action. But a number of groups have begun to propose ways to ensure that all clinical trials are made public.
GlaxoSmithKline volunteered last spring to post all clinical trial results on the company Web site. The trade group Pharmaceutical Research and Manufacturers of America recently published "principles of conduct" for clinical trials that encourage full disclosure of important results.
Critics such as Rennie argue that such voluntary measures won't go far enough. Companies retain the power to stop posting all results once the controversy dies down.
He is in agreement with the American Medical Association, which in June urged the federal government to create a registry of all clinical trials and their results. Around the same time, a group of twelve editors from major journals proposed requiring all trials to go into a publicly accessible database in order to be considered for publication.
Only a legally binding obligation to disclose all trials will force all critical medical and drug data out into the open, Rennie said, and keep it there even when the heat is off.