Doctors relying on studies published in top journals for guidance about how to treat women with breast cancer may not be getting the most accurate information, according to a new analysis.
"Investigators want to go overboard to make their studies look positive," said Dr. Ian Tannock, the senior author of the new study in the Annals of Oncology.
In two-thirds of the 164 studies Tannock and his colleagues scrutinized, that meant not listing toxicities - in other words, serious side effects, whether of chemotherapy, radiation or surgery - in the paper's abstract. Such abstracts summarize the findings, and run a few hundred words.
That's important, said Tannock, of Princess Margaret Hospital in Toronto, because "most of us are so damn busy, we only read the abstract and skim the tables and figures."
In fact, a fifth of studies didn't include toxicities in results tables, and about a third failed to mention them in either the abstract or the discussion section.
Most surprising, said Tannock, was that in a third of studies, if the treatment didn't work as well as one might hope, researchers moved the goalposts, reporting results that weren't what the study was originally designed to test.
Often, those so-called "secondary endpoints" may be less important and meaningful. There is a difference, for example, between showing people lived longer overall, and simply lived longer without their cancers coming back.
Cancer research is not the only area where some researchers are concerned. In November, a group of cardiology journal editors urged authors to watch their language when describing their results. And two pediatrics researchers warned of "spin and boasting" in their field's journals in October.
Researchers "gain more influence with positive studies," said Tannock, whose team analyzed reports of late-stage trials of the kind used by the U.S. Food and Drug Administration to decide whether to approve drugs.
There are various pressures on researchers to make their results "look better than they really are," Tannock told Reuters Health, including drug companies, which often sponsor trials. However, in the new study, who paid for a study didn't have any relationship with how the results were presented.
Scientists may also spin their results to increase their chances of publishing in the top journals surveyed by the new study. Such marquee publications can improve the chances for tenure, promotion and grants.
One of the journals whose studies Tannock and his colleagues looked at, the New England Journal of Medicine, declined to comment, saying they don't typically comment on other studies. Another, the Journal of Clinical Oncology, could not provide a comment by deadline.
Journals can help, Tannock said, by insisting that authors include toxicities in abstracts. "Even in 250 words, everybody can get that in there," he said.
Still, Tannock said, oncologists - the intended audience for these papers - "have to be educated to be critical of what they're reading."