When you go to your GP, you can rest assured that any drugs you are given have been tested to make sure they work and are safe. Or can you?

Hidden in medical journals are many reports about myriad drugs that are prescribed with no scientific evidence to show they work for the problem you've got.

Last year, the British medic Professor Michael Baum used this fact to mount an attack on CAM (complementary and alternative medicine). He urged Britain's National Health Service trusts to fund only treatments "based on solid evidence", and won a lot of support.

After all, drugs have to be rigorously tested, so why shouldn't the alternatives?

Prescribing a drug for a condition it has not been tested to treat is known as "off-label" prescribing. When a drug gets a licence, it is for a particular use - to treat asthma, lower cholesterol - and the drug company has to provide evidence that it does this better than a placebo.

The scale of the "off-label" problem was revealed in a paper published in the US Archives of Internal Medicine recently. It showed that 21 per cent of the 160 most commonly prescribed drugs in the US were given to people on an off-label basis.

For instance, 46 per cent of heart drugs (except those to lower cholesterol or reduce blood pressure) were prescribed off-label, while in the case of Neurontin, licensed for epilepsy and nerve pain, 98 per cent of its uses were for unrelated conditions such as bi-polar disorder and depression. In the US the Food and Drug Administration is under pressure to recognise the possibility that Neurontin increases suicide risk.

Of course doctors should be allowed to use their experience and judgment to prescribe unlicensed drugs where other treatments have failed. But the study - conducted at Stanford University - showed that for nearly three-quarters of the off-label uses, there was "little or no scientific support".

Off-label prescribing has a long history. In the 1980s, people with a mildly irregular heartbeat were given Tambocor or Enkaid, which reduced the risk of sudden death in people with severe arrhythmia. Around 200,000 Americans with a mild version were given the drugs on the assumption that it would help them, too. However, when a proper trial was finally done it showed that it doubled or tripled their risk of death.

What also needs to be examined is the role the drug companies play. Encouraging off-label prescribing, for instance, is a cheaper way of increasing drug sales than conducting large trials for every new use.

Reports in both the British Medical Journal and the Lancet have shown that sleeping pills are neither very safe nor very effective, but they still feature regularly in the top 20 most prescribed drugs in both Britain and the US. An extensive review recently found that counselling and psychotherapy were not only more effective but also without side-effects.