Boston Globe: Under the microscope (fraud in scientific journals 1/22/06

[Guest guest]

THIS STORY HAS BEEN FORMATTED FOR EASY PRINTING

(http://www.boston.com/news/globe/)

Under the microscope

Recent revelations of fraud have caused some editors of scientific journals

to rethink their responsibilities. But can journal editors be muckrakers?

January 22, 2006

By Dizikes

EARLIER THIS month, the journal Science formally retracted two papers by

South Korean scientist Hwang Woo Suk and his colleagues, whose claims about

creating stem cell lines from cloned human embryos were revealed to be false.

In

December, The New England Journal of Medicine (NEJM) declared that a report

it published in 2000 on the painkiller Vioxx contained ''inaccuracies " due to

incomplete data on potential side effects. And on Friday, the NEJM issued an

''expression of concern " that two cancer-research studies it published in

2001 and 2004 appeared to contain misleading evidence-just days after the

British medical journal The Lancet made its own announcement that a 2005 study

from

the same Norwegian-based research team included fabricated data.

In the wake of these and other science scandals in the past several

years-ranging from fabricated findings to misleadingly incomplete data-some

editors

of science publications are rethinking their roles and asking themselves

whether they should act more like muckraking investigators than purveyors of

scientific discovery. Yet journal editors, even those associated with

successful

investigations into malfeasance, demur when asked if sleuthing is, or should

be, part of their jobs. ''Journals cannot be investigating prosecutors or

detectives, " says Campion, senior deputy editor at the NEJM, expressing

a

view common even among reform-minded science editors.

Campion has a point. The global growth of scientific research, often with

corporate backing, has placed some severe limitations on the ability of editors

to do investigative spadework. But there are also signs that editors are

beginning to develop new ideas about oversight in science, and some science

journals have recently performed innovative acts of detection to expose

falsified

results. Today's science editors may not have the power of a prosecutor or

the funding of an industry-sponsored study. But armed with a healthy

skepticism, some believe that they can, and must, play a greater role in the

fight

against scientific fraud.

.. . .

Exercising oversight in the scientific world is no simple matter.

''Sometimes people say, 'Why don't you check everything? " ' says Drummond

Rennie, deputy

editor of the Journal of the American Medical Association (JAMA). But to do

so, Rennie notes, journals would need ''a team of people who've got nothing

to do, but are brilliantly expert in forensic science, who are prepared to

spend a month, or six months " looking at the laboratory work behind every paper

they receive. JAMA receives about 7,000 submissions per year.

And even if journals did have a dedicated staff for detective work, they

would still lack the funding to reproduce most big-time science research.

Hwang's team received $65 million from the South Korean government, and global

clinical trials of pharmaceuticals can be even more expensive. ''When a single

trial costs $150 million, how is anyone ever going to replicate it? " asks

Rennie.

In lieu of lab work, science journals use peer review, in which

experts-typically academic scientists or physicians-read submissions and send

confidential reports on them to editors. But peer review does not involve the

scientific

process of replicating results. It is an inherently subjective estimation of

a project's value by reviewers who have no more lab access than editors.

''Peer review is not good at picking up serious research misconduct, and it

probably never will be, " says Sabine Kleinert, an executive editor at The

Lancet, who has studied patterns of malfeasance in science. For that matter,

peer reviewers are unpaid, and have an incentive to complete reviews quickly,

in

order to resume their own work.

Still, just because researchers have millions of dollars at their disposal

doesn't mean they spend the money, or much time, disguising their fraudulent

claims. Many of the recent attempts to alter data or disguise results have

been surprisingly unsophisticated-and detectable by a determined and creative

editorial staff.

At the NEJM, for instance, editors discovered that the drug maker Merck had

not reported all the relevant data about the risks of Vioxx by examining

previous versions of the study left on a computer diskette that had been

submitted

to the journal. As the Globe reported last week, the Journal of Cell Biology

now routinely checks the photographic evidence accompanying submitted papers

after managing editor Mike Rossner, by happenstance, discovered one

digitally altered photo. Beginning in late 2002, the journal instituted a

policy of

rigorously scrutinizing all the images it plans to publish and has concluded

that 1 percent of all papers tentatively accepted for publication contained

altered images, a figure that sobered Rossner and the editorial staff. Science

announced earlier this month that it will adopt a similar policy.

Undertakings like Rossner's remain, for now, rare. Journal editors have

never systematically studied the prevalence of fraud or other forms of dubious

research in science publications. ''Nobody really knows the incidence of these

things, " says Horace Freeland Judson, a longtime historian of science at

Washington University and author of ''The Great Betrayal: Fraud in

Science " (2004).

Rennie, a scientific fraud-fighter for more than two decades, has suggested

journals join forces for an audit of accuracy in their articles. Testifying

before Congress in 1989, he proposed allowing senior scientists to perform

confidential, hands-on investigations of published research, including basics

like verifying the existence of subjects of clinical trials. ''You would find

out whether the prevalence of fraud was 1 in 10, or 1 in 100, or 1 in 1,000. "

''If it's 1 in 10, " says Rennie, ''we completely revise the way we do

science. "

But the widespread cooperation needed for the kind of project Rennie

suggests-not to mention the massive amount of funding it would require-would be

hard

to find. Many scientists, and many editors, are wary of creating a public per

ception of rampant fraud in the lab, and no consensus exists that action is

needed.

Reformers like Rennie and Rossner have thus adopted what safeguards they can

on their own. In the late 1990s, Rennie implemented a policy at JAMA in

which coauthors of papers must disclose the nature of their participation in a

research project. This is a point of contention in the Hwang scandal, where the

precise role of Hwang's American coauthor, biologist Gerald Schatten of the

University of Pittsburgh, remains under investigation by the university. ''It

takes a tiny amount of space, " Rennie says. ''And it would say Dr. Schatten

analyzed the results, or whatever he did. " Science indicated this month that

it is considering the adoption of the format.

Other editors advocate open peer review, in which the names of the reviewing

scientists are disclosed, in order to impose more individual accountability

on reviewers. The British Medical Journal has used this system since 2000,

and even appends reader responses to the end of each journal article on its

website.

And there have been some signs of cooperation among publications with a

shared interest in transparency. Medical journals have banded together in an

effort to access the tightly guarded research data of clinical trials funded by

private drug companies, which yield overwhelmingly positive results. Although

the medical journals feel compelled to inform readers of new results from

such trials, in 2004 journals including NEJM and The Lancet declared they would

not publish the studies unless they were registered in a free online

database, _ClinicalTrials.gov_ (http://clinicaltrials.gov/) . As part of the

registration process, researchers must state the working hypothesis of a trial

at the

start. Otherwise, notes Campion, ''Investigators may, after a trial, decide

what their hypothesis should have been, and make it sound like it was that

from the outset. "

Scientific journals, like newspapers and magazines, will always be

susceptible to the deceptions of determined fakers. But a clutch of journal

editors

have put scientists on notice that they're on the lookout, and even the modest

steps taken so far may help preempt future malfeasance. As the case of Hwang

Woo Suk makes clear, the disappointment of a failed experiment is far easier

to stomach than the shame of being exposed as a fraud.

Dizikes is a writer living in Arlington. He frequently writes about

science and technology.

© _Copyright_ (http://www.boston.com/help/bostoncom_info/copyright) 2005 The

New York Times Company

[Guest guest]

I wonder why this is being posted, now?

(Paranoia?) I have heard that some big discoveries on mold may be

about to hit the scientific journals.. could that have anything to do

with it? (End paranoia/)

Sure, corporations.. (like drug companies) often, it seems, try to

create a fake buzz or influence scientific journals..

But CSM is generic now, so, to the best of my knowledge.. I don't

think that anyone of the big boys would make any money from a sudden

upswing in CSM usage, (perhaps because of some medical journal

article?)

...really..