Background: There is increasing pressure to monitor surgical performance. In the UK, the Department of Health has produced clinical indicators based on routine data to monitor performance. This study analysed whether such data could measure performance in aortic surgery.
Methods: Routine hospital data on postoperative mortality were collected for 1995-1997 in the Trent region. Procedural and diagnostic codes, modes of admission, districts of residence, treatment and specialty data were compared with audit data and the Operating Theatre Information System.
Results: Inaccuracies in the Health Resource Group (HRG) codes meant that 21.4 per cent of elective aortic cases (HRG Q02) were probably emergencies and 26 per cent of probable ruptured aneurysms were not coded as a vascular emergency. Case mix and patient selection introduced a bias, apparent between tertiary and district general hospitals. For patients aged over 80 years, two district hospitals undertook no elective aortic surgery; the rate for emergency aortic surgery varied between 16 and 25 per cent in the district hospitals, and was 77 per cent in the tertiary centre.
Conclusion: Crude mortality rates used as an indicator of performance are subject to bias and distortion owing to the collection of incorrect information, variation in patient selection between hospitals and case-mix differences. There was a considerable variation in selection and outcomes of patients undergoing aortic surgery in this study.