Comparison of measured versus predicted energy requirements in critically ill cancer patients.

Background: Accurate determination of caloric requirements is essential to avoid feeding-associated complications in critically ill patients.

Methods: In critically ill cancer patients we compared the measured and estimated resting energy expenditures. All patients admitted to the oncology intensive care unit between March 2004 and July 2005 were considered for inclusion. For those patients enrolled (n = 34) we measured resting energy expenditure via indirect calorimetry, and estimated resting energy expenditure in 2 ways: clinically estimated resting energy expenditure; and the Harris-Benedict basal energy expenditure equation.

Results: Clinically estimated resting energy expenditure was associated with underfeeding, appropriate feeding, and overfeeding in approximately 15%, 15%, and 71% of the patients, respectively. The Harris-Benedict basal energy expenditure was associated with underfeeding, appropriate feeding, and overfeeding in approximately 29%, 41%, and 29% of the patients, respectively. The mean measured resting energy expenditure (1,623 +/- 384 kcal/d) was similar to the mean Harris-Benedict basal energy expenditure without the addition of stress or activity factors (1,613 +/- 382 kcal/d, P = .87), and both were significantly lower than the mean clinically estimated resting energy expenditure (1,862 +/- 330 kcal/d, P < or = .003 for both). There was a significant correlation only between mean measured resting energy expenditure and mean Harris-Benedict basal energy expenditure (P < .001), but the correlation coefficient between those values was low (r = 0.587).

Conclusions: Underfeeding and overfeeding were common in our critically ill cancer patients when resting energy expenditure was estimated rather than measured. Indirect calorimetry is the method of choice for determining caloric need in critically ill cancer patients, but if indirect calorimetry is not available or feasible, the Harris-Benedict equation without added stress and activity factors is more accurate than the clinically estimated resting energy expenditure.