More than one patient in every 10 admitted to hospital for treatment will suffer some sort of harm while they are there, according to the first study of its kind in the UK.

The toll of what are known as "adverse events", ranging from picking up a minor infection on the ward to the recent tragedies when patients have died through drugs being injected in the wrong part of the body, costs the NHS around £1bn a yearin extra treatment, says the study, published today in the British Medical Journal.

That is besides the high cost of litigation - £77m compensation was paid to patients last year - and the human cost in suffering for patients and their relatives.

Although there have been studies in the US and Australia to establish how many patients suffer unexpected harm in hospital, the exercise by Charles Vincent, professor of psychology, and colleagues in the clinical risk unit of University College London has never been undertaken before in the UK.

They found that the rate of adverse events in their pilot study of 1,014 patients in two London hospitals was 10.8%. Since some patients suffered more than one event, the percentage of medical misadventures was still higher, at 11.7%.

This is considerably higher than the 3.7% recorded by a Harvard study in the US, but lower than the Australian figure of 16.6%. The team found that half the adverse events in the UK were preventable. "I strongly suspected there was a problem of this magnitude," said Professor Vincent. "I don't think we have faced up to the scale of it or the economic consequences of it."

Prof Vincent was a member of the panel put together by the chief medical officer, Liam Donaldson, which last year launched a blueprint for the NHS to report mistakes and learn from them.

The document, An Organisation With a Memory, promised confidential phone lines for medical staff to report mistakes anonymously and central databases which would reveal common and preventable mistakes.

But little appears to have happened so far, and without precise data for the rates and numbers of mistakes being made by doctors and nurses, it will be impossible to tell whether the confidential reporting system is aware of more than the tip of the iceberg.

"Reporting is not going to tell you what is going on where, and how much it is costing," Prof Vincent said.

He and his team want to carry out a large-scale study to provide a picture of the rate and cost of adverse events nationwide.

In an editorial in the BMJ, Sir George Alberti, president of the Royal College of Physicians, concurs. "Only a full-scale study can substantiate this estimate, and if the NHS is serious about learning about and reducing errors it should fund such a study," he writes.

But Prof Vincent has tried without success to get financial backing for the work, which raises questions as to how eager the NHS is to know the truth about medical error.

The mistakes that are made in hospital were "much more often an accumulation of small things rather than one big blunder, such as problems in record-keeping and delays", he said. Some were barely preventable, such as bedsores, but others could certainly have been avoided, such as infections picked up after surgery by patients who should have been given antibiotics as a preventative measure.

In his editorial, Sir George points out that there are things that can be done to prevent errors happening, once it is clear what is going on. Better surgical training is needed.

"Fewer operations and procedures during the night may also help," he says.

Drugs errors remain a problem, but Birmingham has a computer linked pharmacology system, which sends out warnings when drugs that could be dangerous in combination are prescribed.

"The introduction of such a system nationwide could prevent hundreds, indeed thousands, of errors," he writes.

There may be more mistakes made by doctors than even this study suggests, Sir George says, because not every error leads to harm to the patient. "A pilot study by the Royal College of Physicians into deaths after admission for medical emergencies suggests that some error occurred in as many as one in five cases.. These data should be interpreted cautiously, but do suggest that actual recorded adverse events are the tip of the iceberg."