[Comparison of prognostic nutrition indices in preoperative detection of risk patients. A prospective trial]

Malnutrition must be considered as a factor of risk in surgery and therefore it has to be taken into account in surgical planning. Many authors aggregated several measurements into an index or another mathematical model by stepwise regression or discriminant analysis. Hitherto none of these approaches has been subjected to a critical analysis designed to determine whether the information gained differentiates patients with increased operative risk from those without, to a degree that is clinically relevant. In a prospective study the predictive values of nutritional assessment techniques of various authors were examined in 246 surgical patients undergoing a major surgical procedure. The specificity, sensitivity, and validity of each assessment technique were determined. The statistical analysis showed that none of the assessment techniques separated patients who were at high risk from those who were at low risk in a statistically significant predictive power. Serum albumin level was a quite accurate prognostic indicator of postoperative morbidity and mortality. The mean complication rate in this study was 26.8%. Concerning the specificity, sensitivity, and validity the single measurement of the serum albumin had a predictive value as high as all other determined assessment techniques in this study. We contend that combining measurements into a statistically derived index is time-consuming and expensive and does not produce an assessment technique with sufficient predictive power to identity high risk patients in a clinically relevant fashion.