Guardian | September 23, 2008
By James Randerson

The results of more than half of all clinical trials that demonstrate the safety and effectiveness of new drugs are not published within five years of the drug going on the market, according to an analysis of 90 drugs approved by US regulators between 1998 and 2000.

The researchers, who traced the publication or otherwise of 909 separate clinical trials in the scientific literature, wrote that the failure of drug companies to publish the evidence relating to new medicines amounted to "scientific misconduct". They said it "harms the public good" by preventing informed decisions by doctors and patients about new medicines and by hampering future scientific work.

Sir Iain Chalmers, who is director of the James Lind Library in Oxford and a founder of the Cochrane Collaboration, a respected organisation that reviews medical evidence, said that it was vital that all data on new medicines be made public. 

"Patients may otherwise suffer or die unnecessarily," said Chalmers, who was not involved in the work. "The people who participate in a trial have a right to expect that their participation and their data will be made available publicly so that people can take whatever decisions seem appropriate in the light of that information." 

The US researchers who carried out the study searched the academic literature for publication of the trials that drug companies relied on to convince the US Food and Drug Administration that their new products were safe and effective and so worthy of market approval. 

Information that is used to convince the regulators is not necessarily subsequently published for public and scientific scrutiny, but the scale of the missing information was found to be vast.

Five years after each of the 90 drugs was first available for patients, only 43% of the studies supporting the drugs' use had been published, with most publication happening in the first one or two years. In the case of one product – an antibiotic – the researchers could not find a single supporting trial in the scientific literature, while five trials were published twice and one was published three times. 

The team also found evidence for a "publication bias". Trials with statistically significant results were more likely to be published than those with non-significant results, as were those with larger sample sizes.

"In the years immediately following FDA approval that are most relevant to public health, there exists incomplete and selective publication of trials supporting approved new drugs," Prof Ida Sim and her colleagues at the University of California, San Francisco, wrote in the journal PLoS Medicine.

One possible explanation for the scientific data not being published is that drug companies hold back publication of the results that are least flattering to their new drugs. Another possibility is that academic journal editors are less inclined to publish papers on trials that have negative or ambiguous results.

"Regardless of the cause, publication bias harms the public good by impairing the ability of clinicians and patients to make informed clinical decisions, and the ability of scientists to design safer and more efficient trials based on past findings," the authors wrote. "Publication bias can thus be considered a form of scientific misconduct."

The reporting of clinical trial results should have improved since the period analysed by the researchers, because the 2007 FDA Amendments Act mandated basic results reporting for all trials supporting FDA-approved drugs and devices. However, the researchers said it remained to be seen whether clinical reporting would improve.

The new law could even have the opposite effect. "Might sponsors feel less compelled to publish equivocal trials because the basic results will already be in the public domain?" they speculated.