Index for predicting mortality in elderly surgical patients

Background: The aim of the study was to validate a mortality index for use in patients over 65 years of age undergoing operative treatment or conservative management. Methods: From a study conducted more than 20 years ago of elderly patients admitted with a surgical condition, 11 variables were identified as being significant in predicting mortality: age, white cell count, serum urea concentration, pulse rate, mean blood pressure, type of admission, urgency of operation, complexity of surgery, conservative management, and diagnostic categories of vascular disease and malignancy. Using multivairiate logistic regression analysis, a mortality index was devised. This index was validated in a separate study of comparable patients treated in 1999-2001. Results: Comparison of the derivation and validation data sets showed similarity in the distribution of variables with no significant difference in mortality (overall mortality rate 9(.)3 per cent for the derivation set versus 9(.)6 per cent for the validation set; chi(2) = 0.084, 1 d.f., P = 0(.)771). Tests of calibration and discrimination showed no significant difference in predicted to actual deaths, and the receiver-operator characteristic plots of the two data sets showed good discrimination. Conclusion: This mortality index was derived specifically for patients over 65 years of age receiving either operative or conservative treatment. It could be of use when counselling patients and their relatives about the possible outcome of treatment.