The US Government’s top biomedical scientists have called for an overhaul of clinical trials in order to protect the public and restore trust in medicine. by Michael Day

Following a summit at the National Institutes of Health in Maryland, NIH director Elias Zerhouni warned that failure to spot serious drug side-effects quickly enough was “seeing a gradual erosion of public trust." And he announced an eight-point plan designed to improve clinical trials and reduce the likelihood of side-effects being swept under the carpet.

A key recommendation is for equal weight to be given to determining the safety as well as the efficacy of medicines. The NIH committee cited the scandal surrounding Merck & Co's painkiller Vioxx (rofecoxib) as evidence that safety needs to be stepped up.

Merck withdrew the product in 2004 after it emerged that the drug had increased users’ risk of heart disease and stroke. Earlier this month, a jury in New Jersey ordered the company to pay $9 million in punitive damages to 77-year-old man a US man after Vioxx was found to have contributed to his heart attack - and that Merck had failed to warn about the dangers of taking the drug. It is facing thousands of other lawsuits in the USA.

The NIH committee called for longer-term follow-up, increased powers for the FDA to conduct post-marketing surveillance and more effective dissemination of information relating to adverse events.

The NIH recommendations, published in the Public Library of Science (PLoS) Medicine (June 2006, volume 3, issue 6, e144), also call for researchers to treat prior observational evidence with greater scepticism. It pointed to the case of combined oestrogen/progestogen hormone replacement therapy: for years doctors were convinced that the therapy protected women from heart disease and stroke, based on limited observational studies.

However, in 2002, the Women’s Health Initiative trial showed that it actually increased the risk of circulatory disease. Selection bias in the original observation studies – women on HRT were likely to be fitter, more educated, and less likely to smoke – was to blame, says the NIH.

A tendency to mislead both patients and practitioners with statistics is another area singled out. In particular, the NIH notes that presenting drug efficacy in terms of relative risk alone, can be very misleading. It cites publicity for an anti-osteoporosis drug that simply said over 12 months the treatment cut vertebral fractures by 68% compared to placebo. However, in absolute terms that meant a reduction from 0.738% to 0.238% for those on placebo - i.e. a 0.5% difference.