Clinical audit is never going to be popular with doctors. But in the wake of the Bristol fiasco, it's not going to go away.

Public confidence in doctors has seldom been lower - far too many worrying stories have emerged over the past year. Not only have we read of James Wisheart and Janardan Dhasmana at Bristol, blithely continuing to work with mortality rates four times the average, but there was Dr Sanjay Ingley, a surgeon with hepatitis B, who fatally passed the infection to an 83- year-old woman while replacing her hip.

Three people contracted malaria at Nottingham City Hospital, apparently from contaminated saline drips. Meanwhile, Manchester GP Harold Shipman stands accused of 15 murders.

Iatrogenic illnesses - those induced by doctors - are not new. The word 'iatrogenic' may have been coined by Ivan Illich in the 1960s, but the idea is an ancient one. Hippocrates recognised the problem, and made it clear that a doctor's first duty is not to make the patient worse. Hippocrates' successors have not always followed his dictum.

In 1963 the first study of iatrogenic illness was conducted by Yale University. The Shimmel report showed that 19 per cent of the hospital patients studied experienced complications due to their treatment. More than 1 per cent died.

Twenty-seven years later, Dr Patrick Pietroni, senior lecturer in general practice at St Mary's Hospital medical school, estimated that one in six hospital patients were there because of side-effects from their medication.

Part of the problem is that patients may often simply be being treated for the wrong condition. A study of 400 post-mortems by Dr Jane Mercer and Dr Ian Talbot in 1985 showed that the presumed cause of death had been wrong in more than 50 per cent of cases.

And private patients apparently face even greater risks than NHS ones. A 1994 study of maternity cases showed that in the UK, private maternity patients were, inexplicably, more than twice as likely to have caesarean deliveries than those in NHS beds.

In 1991, the British Medical Journal noted that only 15 per cent of medical interventions were actually fully supported by clinical evidence. Two years later it was still being estimated that 85 per cent of surgical and medical treatments were never fully evaluated. Of course, that's all in the past - isn't it?

Unsurprisingly, the absence of medical attention has on occasion apparently saved lives: during doctors' strikes in Los Angeles, Colombia and Israel in the 1970s , local death rates reportedly fell by 18, 35 and 50 per cent respectively.

While surgery has obvious and dramatic dangers, doctors' prescribing habits may be no less lethal. Even something as beneficial as penicillin kills more than 1,000 people each year throughout the world.

More than 30,000 branded drugs are now available on the world market, of which the World Health Organisation has said only about 200 are really needed. How, one wonders, can so many drugs be effectively monitored?

Drug monitoring relies on doctors' co-operation. But cases continue to emerge of doctors falsifying drug-trial data for gain. In 1975, London GP Dr JP Sedgwick simply invented 100 results and claimed a£10 fee for each from Bayer. The General Medical Council erased his name from the medical register, but this apparently did not deter others.

In 1993, 26 doctors were referred to the GMC for allegedly falsifying drug-trial data. Most recently, in November 1998, south-east London GP Dr James Bochsler was struck off for faking clinical drug-trial results and defrauding drug companies.

The government has rightly placed considerable emphasis on pursuing quality through the National Institute for Clinical Excellence and its enforcing body, the Commission for Health Improvement. But the key question remains: how can clinical audit be made genuinely effective? Two excellent examples of auditing, which illustrate how real clinical audit might be conducted, already exist in the NHS. Both are from the family health services.

One is the now discontinued auditing of prescription drugs. It is probably not well known that, until the early 1990s, family health services authorities and their predecessors, in co-operation with the Pharmaceutical Society, regularly collected random samples of dispensed medicines awaiting collection at community pharmacies and had them analysed for quality and quantity.

An equally robust system of clinical audit continues to operate for general dental services. Dental reference officers from the dental practices board have, since time immemorial, randomly selected NHS dental patients and invited them to attend for a second check on the work carried out, or proposed, by their dentists.

Surely it is sensible and reasonable to extend this kind of random audit to cover the complete range of healthcare provided under the NHS. Why shouldn't a small percentage of every doctor's caseload be subject to an unannounced second check? A recent suggestion that health authority medical advisers might lurk outside GPs' surgeries and invite emerging patients to a mobile surgery nearby - to confirm any diagnosis and verify the appropriateness of the prescription - might be a little over-dramatic. But couldn't something similar be devised without too much difficulty?

Only a crank would seriously suggest that doctors do no good at all, but the public's scepticism will not be reduced nor its confidence restored without a clear demonstration that standards are to be rigorously audited in future. NICE promises that this will be achieved - it remains to be seen whether that promise will be fulfilled.