Introduction There has been much controversy over the use of hospital admission data to measure health outcomes. However with the current implementation of ‘Summary Hospital-level Mortality' measures, it is more important than ever to understand the quality of Hospital Episode Statistics (HES). Therefore the authors have compared the recent upper gastrointestinal bleeding national audit data with the appropriate data from HES.

Methods The NHS Blood and Transplant and British Society of Gastroenterology's 2007 audit of upper gastrointestinal bleeding (UGB) was a national web based audit that occurred between 1 May and 31 June 2007. Using anonymous HES data linked to the Office of National Statistics national death register from the same time period, the authors selected UGB records with provider codes and/or geographical details that matched the participating English hospitals in the audit. The authors compared recorded numbers of admissions, deaths and endoscopies.

Results The audit contained 7484 records of admission to hospital with UGB from England, and 5582 complete records that were included in the final analysis. Over the same time period HES had 8495 UGB admissions from audit participating hospitals (ie, the audit captured 88.1% of these). There were 5385 admissions primarily for UGB included in the audit compared to 5264 admissions in the HES data. Endoscopy was recorded in 55.6% of all records in the BSG audit compared to 46.3% of all HES data. More deaths within 28 days were identified using the ONS linked dataset than in the national audit (see table 1).

Conclusion During the recent national audit, HES recorded reassuringly similar numbers for UGB hospital admissions and procedures. This demonstrates the success of the web based system for national audits in picking up a large proportion of bleeds, and with comparable rates of endoscopy suggests HES are not incorrectly coding large numbers of cases as UGB. The value of HES data is its national coverage, linkage to the national death register, and accurate recording of hospitalisation rates. However the strength of the audit in providing data to permit detailed analysis of the predictors of mortality, procedures, blood transfusions, and the calculation of risk scores cannot be reproduced in the HES data. To gain a complete and accurate picture of the risks, burden, and outcome of UGB the results from both sources of data will need to be considered.