In critically ill children, optimal energy and protein intakes are associated with improved prognoses. However, their protein and energy needs have been determined on the basis of insufficient data. In clinical practice, resting energy expenditure (REE) is frequently estimated from prédictive équations, despite their probable lack of précision, instead of using indirect calorimetry measurements. Thus, the aims of this PhD thesis were twofold: to determine the energy and protein needs in ventilated critically ill children, and to evaluate available prédictive équations for estimating REE in this population. Our first study was conducted to investigate how much protein and energy are needed to equilibrate the nitrogen and energy balances of ventilated critically ill children on sedation and analgesia. From 402 daily measurements of total urinary nitrogen (TUN) and REE in 74 children, we found that TUN was elevated and REE was reduced during mechanical ventilation. Minimum intakes of 1.5 g/kg/d of protein and 58 kcal/kg/d were required in children up to 4 years old, whereas older children needed more protein. Our second work aimed to determine, based on our indirect calorimetry measurements, the biases of 15 prédictive équations, including recent équations specifically developed for estimating REE in ventilated critically ill children. Only the six équations of Talbot, Schofield, Henry and Mehta had a bias < 10% between 200 and 1000 kcal/d, but their précision was still unsatisfactory in young children.
Following this work, we made a systematic review of the literature to document validation studies of prédictive équations in critically ill children. Of the 993 studies identified, 22 studies were included and involved 998 critically ill children and 2076 indirect calorimetry measurements. They assessed 21 prédictive équations, but only 6 équations were evaluated by at least three studies. The équation of Harris-Benedict clearly overestimated REE in a large majority of patients and should not be used in this population. The équations of Schofield and the tables of Talbot were the most accurate, but predicted REE within ± 15% of measured energy expenditure in only 53% of patients. Other analysed équations were either more imprecise or not assessed in a sufficient number of studies. In conclusion, this project precisely assessed the energy and protein needs of ventilated critically ill children aged 0 to 8 years and showed that most prédictive équations of REE are inaccurate, especially in the youngest children.