New system developed to monitor deaths in general practice

Researchers from Imperial College London have developed a system using statistical control charts to help monitor mortality rates in general practice.

Although the system, details of which are published online today in The Lancet, was developed as a practical response to monitoring issues raised by the Harold Shipman case, it can also be used to monitor variations in the quality of care between GPs and practices.

Dr Paul Aylin, one of the paper’s authors, from Imperial College London comments: “Following the Shipman affair it became clear that no method existed to monitor mortality rates in general practice.

“One use of this system would be to help stop a repeat of the Shipman affair, but tools such as this could also help in monitoring the performance of GPs and their practices by continuously keeping track of mortality rates.

“We envisage these methods being used as a governance tool for monitoring performance since they enable a first-pass analysis of the data and can highlight units with an unusual outcome. We caution however, that the charts themselves cannot shed light on the reasons for apparent poor performance.

“Excess mortality will not necessarily mean bad practice or even criminal behaviour. Excess mortality could result from many different situations. For example, practices involved in terminal care for cancer patients or treating patients in a number of nursing homes. The system can also be used to help spot GPs or practices with particularly low mortality rates, which may be indicative of good practice.
“Any GPs or practices that are seen as having unusual mortality patterns could be investigated further through audit. This process could be improved by the collection of additional information on the death certificate.”

Dr Nicky Best, an author of the paper, from Imperial College London comments: “If the NHS is to deliver high quality cost effective care leading to improved health through guidance, audit and best practice, it needs high quality and timely information. Any method, for analysing and comparing performance, no matter how sophisticated, will founder if this is unavailable.”

The researchers collected data from more than 1000 GPs (including Shipman) or practices over five health authority areas between 1993 and 1999, linking information from death certificates and patient lists. From this they were able to work out how many patients died per GP or practice.

In order to calculate a figure from which abnormal mortality rates could be established, local reference rates were derived from the age and year specific mortality rates for the relevant Health Authority.

The researchers then plotted the mortality figures by year on to a graph using a CUSUM (Cumulative Sum) method. This allowed the researchers to monitor the mortality figure year on year, highlighting any unusual trends in mortality. If the cumulative difference between the observed mortality and the reference rates exceeded a pre-defined threshold, this signalled a warning that the mortality rates for the GP or practice in question warranted further investigation.