Comparison of different methods of risk stratification in urgent and emergency surgery

Abstract

Abstract Background The aim was to compare a number of risk scoring systems prospectively in a cohort of patients who underwent non-elective surgery. Methods This was a cohort study of 2349 consecutive patients who had urgent or emergency surgery in a district general hospital in the UK. All patients were scored prospectively using the Revised Goldman Cardiac Risk Index (RGCRI), Portsmouth modification of the Physiological and Operative Severity Score for the enUmeration of Mortality and morbidity (P-POSSUM), Surgical Risk Score (SRS) and Biochemistry and Haematology Outcome Models (BHOM). Actual 30-day and 1-year survival rates were compared with the predicted outcomes using receiver–operator characteristic (ROC) curves and Hosmer–Lemeshow analysis. Results Some 141 patients (6·0 per cent) died within 30 days of operation. This increased to 254 (10·8 per cent) by 1 year. The area under the ROC curve for death within 30 days was 0·90 for P-POSSUM, 0·85 for SRS, 0·84 for BHOM and 0·73 for RGCRI. Only the first three risk scores were able to discriminate accurately within the groups (area under ROC curve over 0·8), with no significant variation between expected and observed mortality rates confirmed by Hosmer–Lemeshow analysis. Similar results were found for the ability of each score to predict outcome at 1 year. Conclusion P-POSSUM, SRS and BHOM scoring systems were all able to predict outcome after emergency and urgent surgery, but the SRS had the advantage of ease of calculation. BHOM requires only the most commonly available blood test data and the computer holding these data can easily perform the calculation.d.