After I blogged about disclosure of pharmaceutical clinical trial results last week, several people wrote me pooh-poohing the problem, arguing that since all clinical trials are pre-registered with the FDA, there is no way pharmaceutical companies can conceal unfavorable results.

Today’s New York Times, however, suggests otherwise . According to the article,

The issue of The American Journal of Psychiatry that hit the desks of its 37,000 readers this month reported test results for the antidepressant drug Celexa, indicating it could help children and teenagers.

Before publication, the article received the kind of scrutiny common among medical journals. The study’s authors had been asked to divulge their financial ties, if any, to the drug’s marketer, Forest Laboratories Inc., which sponsored the clinical trial. And the report was sent to reviewers who examined the trial methodology and checked to make sure that the article reflected other relevant research about the use of antidepressants in youngsters.

But neither the article nor the 27 scholarly footnotes that accompanied it mentioned another major drug-industry-sponsored trial completed in 2002, which found that Celexa did not help depressed adolescents any more than a placebo. Nor would the article’s reviewers have been likely to find any clues of that trial’s existence. The results of that trial were first noted last year on a single line of a chart that appeared on Page 96 of a textbook – one written in Danish.

The article goes on to say, “Editors say they must also be vigilant against companies’ cherry-picking favorable but limited data from a trial that had originally set out to test other aspects of a drug’s performance.”

This is exactly right. A few years ago, JAMA faced this problem with a widely-publicized Celebrex trial. Tom Perry, a Canadian physician, recounts the reaction to the revelation that the authors of the study had concealed some pertinent facts:

Boston gastroenterologists Drs. David R. Lichtenstein and M. Michael Wolfe had written a cautiously favourable editorial about the study in the same issue of JAMA. However, last February, when Dr. Wolfe was shown the full study as a member of the FDA Arthritis Advisory Committee, he saw that the complete trial data painted a different picture, and that celecoxib did not appear to offer a significant safety advantage over the older, less expensive medications.

“We were flabbergasted,” said Dr. Wolfe, after learning that what the authors had represented as a single 6-month trial was actually a combined analysis of the first 6 months of 2 separate 12-month trials. The authors had omitted the second 6-month data set, in which the apparent celecoxib advantage melted away. JAMA Editor Catherine D. DeAngelis said the journal was not informed about the missing data. “I am disheartened to hear that they had those data at the time that they submitted to us. We are functioning on a level of trust that was, perhaps, broken.”