Validation of a predictive method for an accurate assessment of resting energy expenditure in medical mechanically ventilated patients
Résumé
Objective: Use comparison with indirect calorimetry to confirm the ability of our previously described equation to predict resting energy expenditure in mechanically ventilated patients.Design: Prospective, validation study. Setting: Eighteen-bed, medical intensive care unit at a teaching hospital. Patients: All adult patients intubated >24 hrs were assessed for eligibility. Exclusion criteria were clinical situations that could contribute to erroneous calorimetric measurements. Interventions: Resting energy expenditure was calculated using the original Harris-Benedict equations and those corrected for usual stress factors, the Swinamer equation, the Fusco equation, the Ireton-Jones equation, and our equation: resting energy expenditure (kcal/day) = 8 × weight (kg) + 14 × height (cm) + 32 × minute ventilation (L/min) + 94 × temperature (°C) − 4834. Measurements and Main Results: Resting energy expenditure was measured by indirect calorimetry for the 45 included patients. Resting energy expenditure calculated with our predictive model correlated with the measured resting energy expenditure (r2 = .62, p < .0001), and Bland-Altman analysis showed a mean bias of −192 ± 277 kcal/day, with limits of agreement ranging from −735 to 351 kcal/day. Resting energy expenditure calculated with the Harris-Benedict equations was more weakly correlated with measured resting energy expenditure (r2 = .41, p < .0001), with Bland-Altman analysis showing a mean bias of 279 ± 346 kcal/day between them and the limits of agreement ranging from −399 to 957 kcal/day. Applying usual stress-correction factors to the Harris-Benedict equations generated wide variability, and the correlation with measured resting energy expenditure was poorer (r2 = .18, p < .0001), with Bland-Altman analysis showing a mean bias of −357 ± 750 kcal/day and limits of agreement ranging from −1827 to 1113 kcal/day. The use of the Swinamer, Fusco, or Ireton-Jones predictive methods yielded weaker correlation between calculated and measured resting energy expenditure (r2 = .41, p < .0001; r2 = .38, p < .0001; r2 = .39, p < .0001, respectively) than our equation, and Bland-Altman analysis showed no improvement in agreement and variability between methods. Conclusions: The Faisy equation, based on static (height), less stable (weight), and dynamic biometric variables (temperature and minute ventilation), provided precise and unbiased resting energy expenditure estimations in mechanically ventilated patients.