A new study of cancer research finds almost nine out of 10 studies could not be reproduced.
Scientists publish their research so other scientists can use the findings in their research. The research papers should include the details of the experiment so others can confirm the original findings.
C. Glenn Begley was vice president at California biotech company, Amgen, when researchers there tried to confirm published findings related to their work.
Out of 53 important papers describing research done elsewhere, the Amgen researchers could only duplicate results in six of them.
"This really was a shocking finding, frankly," says Begley, who published his team's findings in Nature. "I certainly had not expected the ability to reproduce the studies would be as low as that."
In many cases, the Amgen scientists contacted the original researchers to make sure they were following the same procedures, but still mostly got different results.
Begley isn't saying the original results were wrong, just that they are not reproducible, as they should be.
"To be clear, these are all good people that are all trying to do the right thing, that are trying to improve the outcomes for patients," he says. "But there are inherent pressures within the system."
The main pressure for researchers is to find positive results, according to Begley, which offer new directions for further research, while negative results are seen as a dead end.
"And so they don't get the same sort of attention at scientific conferences," he says. "They don't get the same sort of attention within the scientific literature."
To avoid negative results, Begley says, some researchers may use only a selected part of the experimental data to produce more clear-cut findings, or don't take the time to repeat critical experiments to validate the results.
But Begley says negative findings - for example, that a potential drug just doesn't work - can be very valuable, by directing scientists to more productive research.
He stresses that getting the early stage laboratory research - long before potential drugs are tested on patients - is critically important.
"What we do in the laboratory is fundamentally important in terms of setting the stage for what will ultimately be new therapies that find their way to patients," he says.