Determination of malnutrition in hospitalized patients with the use of a group-based reference.

We clustered a selected population of patients at Bridgeport (Conn) Hospital into distinct nutritional classes using critical decisions points for the serum concentrations of aspartate aminotransferase activity, cholesterol, total protein, albumin, and prealbumin. The decision points that divided the populations with the highest efficiency were delineated by the information in the data set. At these decision values the hospital population adjusted malnutrition rate was just more than 30%. This was lower than the reported prevalence of malnutrition in the hospital population and was better in agreement with global estimates of malnutrition. The study identified a problem in the assignment of decision values that are used for nutritional support which has implications for nutritional interventions.