You are searching the health shop for something to banish your cold, cystitis or irritable bowel. The assistant is vague. "This is supposed to be good," she says, proffering a pounds 10 tub of pills. You faintly remember reading something about this remedy in the paper. But how do you know if it works or whether you'll be wasting your money?

"The fact is, the public can't tell," says Dr Adrian White, senior lecturer in the department of complementary medicine at Exeter University. "They might have heard about an individual research trial from the press. But it is only when you weigh up all the evidence that you get an overall view."

St John's wort is an example. Early in the summer, American research was published showing that the herb, widely sold for the treatment of depression, had little effect on severely depressed people.

The trial, which received a good deal of publicity, created a stink. As manufacturers of herbal products pointed out, St John's wort has never been advocated for the treatment of severe depression, and the American trial was funded by Pfizer, a leading maker of anti- depressant drugs.

"I would agree that St John's wort is inappropriate for people suffering from severe depression but, in some cases, completely appropriate for those suffering mild to moderate symptoms," says consultant psychiatrist Dr David Wheatley.

"More than 50 clinical trials have been undertaken successfully comparing its effectiveness with placebos and a variety of active drugs in relieving the symptoms of mild to moderate depression and mild depressive symptoms, such as seasonal affective disorder."

Manufacturers are usually willing to cite clinical trials as evidence that their product works. But some studies are not worth the paper they are printed on. If the numbers are small, the trial short and there is no control group for comparison, the results are likely to be meaningless. There may also be bias.

"The whole process of scientific evaluation is meant to be unbiased," says Dr White. "But it is actually rare for a group of people - and that includes clinicians - to have no bias at all."

Trials can be influenced in various ways. Patients who are more likely to recover can be allocated to the group receiving the treatment under test; patients who drop out along the way can be excluded from the final result; and the clinician assessing the patients may look for a more positive result in the treatment group.

The "gold standard" of research is the randomised, double blind, placebo-controlled clinical trial. In this, half the subjects receive an active treatment and the rest an identical-looking dummy or placebo. Patients are allocated randomly to the two groups and neither doctor nor patient knows who has received what. The seal of approval is given when the research is accepted for publication in a prestige "peer-reviewed" journal, such as the New England Journal of Medicine or The Lancet.

"But not everything published in journals like these is of high quality," warns Douglas Altman, professor of statistics in medicine at Oxford University. "Many studies, even in good journals, are poor or even wrong."

Bias can also apply here. Positive studies are more likely to be accepted by journal editors than negative ones and - according to Dr White - those involving orthodox medicine are more likely to be published than those examining complementary products and therapies.

So where does this leave the bewildered patient? The systematic review, which identifies and evaluates all the available evidence, is probably the best guide to the effectiveness and safety of a product.

Members of the Department of Complementary Medicine at Exeter University are about to publish the results of their trawl through the research, including systematic reviews, on a huge range of complementary and alternative treatments from acupuncture to yoga (The Desktop Guide to Complementary and Alternative Medicine, Mosby pounds 24.95 - from October).

It is intended for health professionals, but is accessible to anyone interested in the subject and comes with a searchable CD. More lightweight, but also thorough, is the complementary health information produced by the health food chain GNC. You can find this at GNC shops (phone 0845 601 3248 for your nearest branch) or online at www.gnc.co.uk. Meanwhile, perhaps you should take clinical trials with a pinch of salt.

From Healthy.net