BACKGROUND: The use of steady state as the endpoint for performance of indirect calorimetry (IC) is controversial. We designed this prospective study to evaluate the necessity and significance of achieving steady state. METHODS: Patients with respiratory failure placed on mechanical ventilation in a short- or long-term acute care unit at any 1 of 3 university-based urban hospitals were eligible for the study. The 24-hour total energy expenditure (TEE) was determined by a Nellcor Puritan Bennett 7250 continuous IC monitor. Measured gas exchange parameters were obtained and averaged every 1 minute for the initial hour and then every 15 minutes for the next 23 hours. Over the initial hour, resting energy expenditure (REE) was averaged for intervals over the first 20, 30, 40, and 60 minutes, and for various definitions of steady state where oxygen consumption (VO2) and carbon dioxide production (VCO2) changed by <10%, 15%, and 20%. Coefficient of variation (CV) was calculated for VO2 over the first 30 minutes of study. RESULTS: Twenty-two patients (mean age, 52.8 years, 59% male, mean Acute Physiology and Chronic Health Evaluation (APACHE III) score 42.0) were entered in the study. The best correlation between short-term "snapshot" REE and the 24-hour TEE was achieved by the steady-state period defined by the most stringent criteria (change in VO2 and VCO2 by <10%). The average REE for all steady-state and interval periods correlated significantly to TEE with no significant difference in the absolute values for REE and TEE. Adding 10% for an activity factor to the average REE for each steady-state and interval period again correlated to TEE in a similar fashion with the same R value, but the absolute values for REE + 10% for all steady-state and interval periods were significantly different than the corresponding TEE. In those patients with less variation (CV for VO2 < or = 9.0), the REE obtained for the steady-state period defined by the most stringent criteria still had the best correlation, but similar correlation could be obtained by interval testing of > or = 30-minute duration. In those patients with greater variation (CV for VO2 >9.0), interval testing of at least 60 minutes or more was required to attain levels of correlation similar to that achieved by the steady-state period defined by the most stringent criteria. CONCLUSIONS: These data support the use of steady state, best defined as an interval of 5 consecutive minutes whereby VO2 and VCO2 change by <10%. The mean REE from this period correlates best to the 24-hour TEE regardless of CV. IC testing can be completed after achievement of steady state. Activity factors of 10% to 15% should not be added to the steady-state REE, because this practice significantly decreases the accuracy. In patients who fail to achieve steady state, the CV helps to determine the appropriate duration of IC testing. In those patients with a low CV (< or = 9.0), 30-minute test duration is adequate. In patients with CV >9.0, test duration of at least 60 minutes may be required. These latter patients should be considered for 24-hour IC testing.

Steady state (best defined as an interval of 5 consecutive minutes whereby Vo ₂ and Vco ₂ change by ≤ 10%) should be used as the endpoint of testing by indirect calorimetry, because the mean REE from this period correlates best to 24-hour TEE. In patients who fail to achieve steady state, the coefficient of variation (CV) for Vo ₂ helps determine whether a 30-minute interval is sufficient (CV ≤ 9.0), or more prolonged testing is required (CV > 9.0).