Effective audit: reporting to the National Confidential Enquiry into Perioperative Deaths

To investigate the effectiveness of computer based and manual district and unit information systems for identifying hospital deaths eligible for reporting to the National Confidential Enquiry into Perioperative Deaths (NCEPOD).

Prospective six to 10 week study of hospital (death register, immediate coding of medical records) and district information systems followed by six month assessment after modification to entry of data.

Acute units within Lewisham and North Southwark Health District.

All 290 patients dying in hospital during the six weeks, for whom the medical records were obtainable in 231.

Sensitivity and specificity of the information systems in ascertaining eligible surgical deaths (patients dying in hospital who had during 30 days previously had a surgical procedure while under the care of a consultant in a surgical specialty) tested against validated list of screened medical records.

Of 231 medical records, 30 (12 from Lewisham, 18 from North Southwark) met the national inquiry's criteria. The computer based systems of both units detected less than 60% of eligible deaths (sensitivity 53%, specificity 83%); the death register detected about 60% (sensitivity 61%, specificity 89%); manual systems detected all eligible deaths. Subsequent modification to ensure immediate coding of records into the computerised systems during follow up failed to show any improvement.

Routine hospital information systems may miss up to half the deaths eligible for NCEPOD.