The choice to treat all, some, or no patients undergoing gastrointestinal surgery with nutritional support: a...

We used decision analysis to compare three strategies for reducing the incidence of nutrition-associated complications in patients undergoing gastrointestinal surgery: treat all patients with nutritional support, treat no patients, or perform a test which categorizes patients as high or low risk (treating only the high risk patients). We use the model to make management recommendations based on current knowledge and to set priorities for future research. The "test strategy" is optimal in the majority of cases which we believe are clinically encountered when the test involves subjective global assessment of nutritional status on admission (sensitivity 0.82, specificity 0.72) and nutritional support must be delivered by a central venous catheter. However, the result is particularly sensitive to the predictive properties associated with the test and is less sensitive to the effectiveness rate of nutritional support in reducing the incidence of postoperative complications. We conclude that in future research priority should be given to studying and improving the predictive properties of nutritional assessment techniques.