BACKGROUND: Specific morbidity related to underfeeding and overfeeding necessitates the design of nutrition support regimens that provide calories equal to those required on the basis of energy expenditure. This prospective multicenter trial was designed to determine what percent of patients in long-term acute care facilities receive feeding appropriate to their needs and whether accuracy of feeding has an impact on patient clinical status. METHODS: Patients on mechanical ventilation who were hospitalized at 32 Vencor Hospitals over a 9-week period and who were receiving only enteral nutrition by continuous infusion at a presumed goal rate were evaluated once by indirect calorimetry (IC) while on feeding. Caloric intake over the preceding 24 hours was determined by physician orders and by patient intake/output (I/O) record. Caloric requirements were defined by measured resting energy expenditure (REE) + 10% for activity. Degree of metabolism was defined by the ratio: (measured REE/Harris-Benedict predicted REE) x 100, and the degree of feeding by the ratio: (calories provided/calories required) x 100. RESULTS: IC was performed on 335 patients (mean, 11.2 patients per center; range, 1 to 32), of which 72 were excluded for nonphysiological results or failure to achieve steady state, 21 for receiving parenteral nutrition, and 29 for not being on mechanical ventilation at time of testing. The 213 study patients were 58.7% male with mean age 70.1 years (range, 20 to 90 years). Measured REE was <25 kcal/kg in 66.2% of patients and 25 to 35 kcal/kg in 28.6%. Barely half (48.4%) of this patient population was hypermetabolic. Based on physician orders, the majority of patients (58.2%) were overfed, receiving >110% of required calories, and 12.2% were underfed, receiving <90% of requirements. Discrepancies based on I/O records, however, suggested that 36.1% of patients received <90% of those calories ordered. By either basis, only about 25% of patients received feeding within 10% of required calories. The percent of patients being overfed varied between centers, ranging from 32.2% to 92.8%, and was not affected by years of facility IC experience or volume of IC studies per month. The pattern of caloric provision as measured by degree of feeding correlated inversely to degree of metabolism (p < .0001, R2 = .24). Accuracy of feeding had an impact on ventilatory status, as degree of feeding correlated inversely with minute ventilation (p = .001, R2 = .05). Degree of overfeeding also led to significant increases in azotemia (p = .033, R2 = .02). Extrapolating study data over 1 year, reduction in excess volume of enteral formula would have resulted in a cost savings of up to $1.3 million for the Vencor system. CONCLUSIONS: Because energy expenditure is difficult to predict on the basis of conventional equations, patients in long-term acute care facilities routinely are overfed and underfed, with only 25% receiving calories within 10% of required needs. Measuring a patient's energy requirement at least once by IC is important, because the degree of metabolism predicts how easily a patient will be underfed or overfed. The amount of infused calories should be compared with caloric requirements measured by IC, because the accuracy or degree of underfeeding or overfeeding has an impact on ventilatory status and the likelihood for developing azotemia. Although physician practice or bias may reduce the optimal clinical effect, the use of IC to determine caloric requirements may result in significant cost savings.