A survey reported on page 1070 of this issue suggests that pharmaceutical companies often have financial links with the people who write guidelines on how physicians prescribe drugs. More than a third of the authors of such guidelines declared a financial connection to the companies that produced the drugs.

No one is suggesting that the authors in question are deliberately adjusting guidelines to match these interests: most of them surely have their patients' interests at heart. But that doesn't mean that the potential conflicts don't matter. In other areas of medicine — including clinical trials and educational programmes — pharmaceutical money has been shown to exert a subtle influence on the choices made by physicians and researchers (see A. Wazana J. Am. Med. Assoc. 283, 373–380; 2000). So it is not unreasonable to assume that a similar effect is at work in the preparation of prescription guidelines.

Several simple steps could be taken to improve matters. The doctors' organizations that write most of the guidelines should require that authors fully disclose potential conflicts of interest and add their disclosure statements to the guidelines. This doesn't always happen now — despite widespread agreement that it's the right thing to do.

But disclosure is only a first step towards objectivity. The survey also revealed that about 10% of all guideline panels include an author who holds stock in a relevant company. These holdings should be restricted to modest levels, above which the authors should excuse themselves from writing guidelines. The professional societies that produce the guidelines should also consider introducing a ceiling on what authors can earn from relevant companies. Some organizations, such as the Institute for Clinical Systems Improvement, a Minnesota-based collaboration between US healthcare providers that produces advice on clinical practice, already have such criteria. There is no reason why others shouldn't follow suit.

But the time may also be ripe for a more radical overhaul of the system, particularly in the United States, where the financial stakes are highest and where the government currently plays almost no role. The organizations that produce the guidelines argue that it is impractical to exclude every individual with such a financial link from their preparation. But bodies in related spheres already do this. The Consensus Development Program at the National Institutes of Health, for example, which publishes advice on areas of clinical practice where there is debate about treatment options, operates a ‘jury and witnesses’ model, in which experts with financial links can only present evidence, not make judgements on it.

A ‘jury and witnesses’ mechanism could be set up at the US health department to steer the preparation of guidelines.

A similar mechanism could be set up at the US health department, for example, to steer the preparation of clinical guidelines. This would, of course, require modest public funding. Britain's National Institute for Health and Clinical Excellence, which receives government funding and devolves its decisions to external experts, provides a useful model.

But would such an organization be able to produce guidelines quickly enough? When a new drug becomes available, a team of physicians can review the evidence and make recommendations in a few months. More sophisticated, consensus-based approaches typically take longer. But if it is adequately funded, such a body could produce guidelines expeditiously. If US physicians really want to separate marketing from medicine, they should support its creation.