The surgical risk scale as an improved tool for risk-adjusted analysis in comparative surgical audit
Article first published online: 5 NOV 2002
© 2002 British Journal of Surgery Society Ltd
British Journal of Surgery
Volume 89, Issue 6, pages 763–768, June 2002
How to Cite
Sutton, R., Bann, S., Brooks, M. and Sarin, S. (2002), The surgical risk scale as an improved tool for risk-adjusted analysis in comparative surgical audit. Br J Surg, 89: 763–768. doi: 10.1046/j.1365-2168.2002.02080.x
- Issue published online: 5 NOV 2002
- Article first published online: 5 NOV 2002
- Manuscript Accepted: 16 JAN 2002
Comparative surgical audit is becoming increasingly important although it is fraught with difficulties due to risk-adjusted analysis. Methods have been proposed to solve this problem and allow meaningful comparison of patient outcome. None has been described without faults, making such comparison flawed or overtly complicated. An alternative risk scoring system incorporating the Confidential Enquiry into Perioperative Deaths (CEPOD) grade, the American Society of Anesthesiologists (ASA) grade and the British United Provident Association (BUPA) operative grade was formulated and assessed.
This was a prospective audit of 4308 patients admitted under the care of three surgeons between May 1997 and October 1999, creating an initial data set of 3144 procedures with 134 deaths. Each procedure was allocated a score on the basis of the CEPOD, BUPA and ASA grade. The Surgical Risk Scale (SRS) was devised by adding together the values of the three variables, which generated a scale ranging from 3 to 14. Multivariate logistic regression analysis involving the three variables and univariate analysis of the SRS score were undertaken. Receiver–operator characteristic and calibration curves were formulated. This process was validated on another data set (2780 patients) derived from all admissions to the same surgeons between November 1999 and December 2000.
Univariate logistic analysis of the SRS score revealed it to be significantly predictive of death (β = 0·84, P < 0·001); it did not overpredict mortality for low-risk procedures.
The SRS is easy to use, formulate and interpret, and provides an accurate prediction of death in general surgical patients across the entire risk spectrum. © 2002 British Journal of Surgery Society Ltd