Drug Data Shouldn’t Be Secret

By Peter Doshi and Tom Jefferson
The New York Times - Opinion
Originally published April 10, 2012

IN the fall of 2009, at the height of fears over swine flu, our research group discovered that a majority of clinical trial data for the anti-influenza drug Tamiflu — data that proved, according to its manufacturer, that the drug reduced the risk of hospitalization, serious complications and transmission — were missing, unpublished and inaccessible to the research community. From what we could tell from the limited clinical data that had been published in medical journals, the country’s most widely used and heavily stockpiled influenza drug appeared no more effective than aspirin.

After we published this finding in the British Medical Journal at the end of that year, Tamiflu’s manufacturer, Roche, announced that it would release internal reports to back up its claims that the drug was effective in reducing the complications of influenza. Roche promised access to data from 10 clinical trials, 8 of which had not been published a decade after completion, representing more than 4,000 patients from every continent except Antarctica.

(cut)

In response to our conclusions, which we published in January, the C.D.C. defended its stance by once again pointing to Roche’s analyses. This is not the way medical science should progress. Data secrecy is a disservice to those who volunteer their bodies for clinical trials, and is dangerous to those being asked to swallow approved medicines. Governments need to become better stewards of the scientific process. 

The entire story is here.

File drawer effect: Science studies neglecting negative results

By Dan Vergano

USA Today: Science Fair

Some scientific disciplines are reporting far fewer experiments that didn't work out than they did twenty years ago, suggests an analysis of the scientific literature.

In particular, economists, business school researchers and other social scientists, as well as some biomedical fields, appear increasingly susceptible to the "file-drawer" effect -- letting experiments that fail to prove an idea go unpublished -- suggests the Scientometrics journal study by Daniele Fanelli of Scotland's University of Edinburgh.

"Positive results in research studies overall, became 22% more likely to appear in scientific journals from 1990 to 2007," says the study, which looked at a sample of 4,656 papers over this time period, looking for trends in science journals.

"One of the most worrying distortions that scientific knowledge might endure is the loss of negative data. Results that do not confirm expectations—because they yield an effect that is either not statistically significant or just contradicts an hypothesis—are crucial to scientific progress, because this latter is only made possible by a collective self-correcting process. Yet, a lack of null and negative results has been noticed in innumerable fields. Their absence from the literature not only inflates effect size estimates in meta-analyses, thus exaggerating the importance of phenomena, but can also cause a waste of resources replicating research that has already failed, and might even create fields based on completely non-existent phenomena," says the analysis.

The analysis looked at studies where authors proposed a hypothesis and then sought to test it, either confirming it for a positive result, or not. Overall, 70.2% of papers were positive in 1990–1991 and 85.9% were positive in 2007. "On average, the odds or reporting a positive result have increased by around 6% every year, showing a statistically highly significant trend," says the study.

Japan produced the highest rate of positive results, followed by U.S. results. Some regions, such as Europe, and disciplines, such as Geosciences and Space Science, didn't show the positive-result increases:

"The average frequency of positive results was significantly higher when moving from the physical, to the biological to the social sciences, and in applied versus pure disciplines, all of which confirms previous findings. Space science had not only the lowest frequency of positive results overall, it was also the only discipline to show a slight decline in positive results over the years, together with Neuroscience & Behaviour," says the study.

Geosciences and Plant & Animal sciences showed a flat rate of positive vs. negative results since 1990.

The entire piece can be read here.