Study of NHS errors unreliable, says official

A study of mistakes and accidents in the NHS is unreliable because of computer and administrative errors, Sir Liam Donaldson, the chief medical officer, said yesterday.

The National Patient Safety Agency (NPSA) recorded 27,110 "adverse incidents" in 28 health trusts within nine months, which equates to about one million a year nationally.

But, presenting the data to the NPSA conference yesterday, Sir Liam said that in up to 60 per of cases there was no indication of the severity of the error.

He also said that the NPSA computer system "did not talk well" with computers in hospitals, doctors' surgeries and mental health services that supplied the data.

Sir Liam said the findings of the study were "good news and bad news. The very, very good news is that NHS staff of all grades were making reports and were interested in participating in the scheme," he said. The downside was the data collection problem, but he said a pilot study was identifying teething problems.

The Government's treatment of the survey has been controversial. Earlier this week Alan Milburn, the Secretary of State for Health, reversed a decision not to publish its findings, which ministers feared would increase public anxiety about the NHS.

According to the study, there were about 135 deaths among the 27,110 incidents that were confidentially reported by health service staff to a government agency for the study.

Half of the deaths have been put down to natural causes but about 50 of the deaths cannot be easily explained and are the subject of inquiries at hospitals across England.

Three of the deaths were caused after staff made a mix-up on wards over a solution, potassium chloride, which is often given to keep the heart working in seriously ill patients but is sometimes confused with other solutions or administered incorrectly.

Sir Liam issued a full alert over the solution and it has been withdrawn from general wards in its present form. Instead these wards will have access only to diluted forms of the drug to try to prevent more blunders.

The report also revealed 542 reported cases of "moderate" errors, which resulted in no long-term harm and 4,600 cases leading to no harmful effects. There were 5,100 minor mistakes, such as patients falling over or minor drug reactions.