Plan to monitor patient death rates to stop a second Shipman

Trials to monitor and compare GPs' patient death rates to try to prevent another murder spree like that of serial killer Harold Shipman might begin next year.

Efforts have begun to find reliable ways of measuring and matching the data of local NHS primary care trusts and GPs. Last night the professor who did the official audit into Shipman's murderous career called for a public debate on how pilot schemes might work.

Richard Baker, of Leicester University, who recommended routine monitoring two years ago, said last night there might be unavoidable "false alarms" over doctors with suspiciously high death rates, but there could also be important public health benefits in being able to spot GPs' everyday successes and failures.

Monitoring, which might include causes of deaths, each patient's age and sex, time of death and whether people were present, would have to be practical and inexpensive.

Some GPs already use their own patients' health records to see what lessons they can learn about treatment methods. But formal checks on success rates are nowhere near as advanced as those in hospitals or for individual surgeons.

Professor Baker and his colleagues hope there will soon be a public debate on monitoring. Writing in the British Medical Journal, they argue that checks would help maintain public trust in doctors, and better recording of non-fatal heart attacks, strokes or other illnesses, would probably bring public health benefits as well as early identification of the rare criminal.

Prof Baker investigated Shipman's horrific record, stretching over 24 years, and estimated that the GP killed at least 236 patients. One of his co-authors on the project was Peter Goldblatt, the chief medical statistician of the office for national statistics, who has been matching data systems across England and Wales.

The researchers put detection of serial killing as a priority, though they think it unlikely that any future cases will match Shipman's. Signs of "aberrant behaviour" by a GP should be investigated independently and perhaps without the doctor knowing, they say. There might be other times, too, when slightly above average death rates should be investigated in collaboration with the doctor so that clinical lessons can be passed on.

The Department of Health confirmed that it hoped to finalise work this year on a study linking data. It may then set up a "pilot database". Information would be used to identify negligence.