Energy Expenditure in Children With Severe Head Injury: Lack of Agreement Between Measured and Estimated Energy ExpenditureDivision of Pediatric Critical Care Medicine, Department of Pediatrics, Advocate Lutheran General Children's Hospital, Park Ridge, Illinois Correspondence: Suresh Havalad, MD, Division of Pediatric Critical Care Medicine, Department of Pediatrics, Advocate Lutheran General Children's Hospital, 1775 Dempster St, Park Ridge, IL 60068. Electronic mail may be sent to Suresh.Havalad-MD{at}advocatehealth.com. Background: The purpose of this study was to test the hypotheses that estimates of resting energy expenditure (REE) vary significantly from measured energy expenditure in a population of head-injured children and are not accurate for use in determining nutrition needs in this population. Methods: This is a retrospective study of 30 children with severe head injury, with Glasgow Coma Scale (GCS) score of <8 and needing mechanical ventilation. Measured REE was obtained using indirect calorimetry. Estimated REEs were calculated using Harris-Benedict, World Health Organization (WHO), Schofield, and White formulas. Severity of illness was calculated using Pediatric Risk of Mortality (PRISM) score. Agreement between measured REE and estimated REE was tested using the Bland-Altman method. Correlation coefficient between PRISM score and measured REE was calculated using Spearman test. Results: More than half of the estimates of REE differed from measured REE by >10%. Significant disagreement between estimated REE and measured REE was demonstrated using the Bland-Altman method. There was no correlation between severity of illness and measured REE to explain the inaccuracies of REE estimates. Conclusion: Energy expenditure in critically ill children cannot be estimated accurately; hence, nutrition for critically ill children with head injury should be provided according to measurement of REE to avoid the consequences of overfeeding or malnutrition. Traumatic brain injury is a major cause of morbidity and mortality in the pediatric age group1; thus, children with head injuries represent a significant percentage of pediatric intensive care patients. These critically ill children require nutrition support as part of their daily therapeutic interventions, which can be provided either enterally or parenterally. Nutrition support must be adequate to meet the metabolic needs of the patient.2–5 The causal effect of malnutrition on immune dysfunction in adults and children has been well documented, and malnutrition sufficient to impair immune function occurs in nearly half of the adult patients admitted to hospitals in this country.3 Malnutrition adversely affects both humoral and cellular immunity.4,5 Likewise, overfeeding may result in other significant complications, such as liver and respiratory dysfunction from increased substrate load.6,7 Hence, caloric intake should also match energy expenditure to avoid overfeeding. Overfeeding with carbohydrate calories leads to lipogenesis, with a respiratory quotient (RQ) >1.8,9 A high RQ is associated with increased carbon dioxide production and can lead to increased ventilatory needs in critically ill patients. In addition, overfeeding leads to increased deposition of fat and glycogen in the liver, and hyperglycemia.7 This may compound the degree of hyperglycemia associated with critical illness.10,11 In addition, it has been demonstrated that uncontrolled hyperglycemia is associated with increased mortality in the critically ill patient.12 Resting energy expenditure (REE) is the major component of the daily energy requirement of critically ill children13 and can be estimated by using various equations. The Harris-Benedict (HB) equation has been used to estimate REE in adults to provide a baseline from which caloric requirements can be calculated.14 In 1985, Schofield,15 using a data set of 7549 subjects, published an equation that was age-specific and better suited to estimating REE in children. In 1985, the World Health Organization (WHO) published recommendations for feeding children according to predictive equations of energy expenditure.16 More recently, White et al17 evaluated the current methods of predicting energy requirements and found them to be inaccurate when applied to critically ill children who were mechanically ventilated. It is uncertain whether any of the previously described equations can accurately estimate energy expenditure in critically ill children so that optimal nutrition support can be provided. In addition, children in the pediatric intensive care unit (PICU) may have various comorbidities or conditions that affect REE, and thus, these patients do not represent a homogenous population for estimating REE. For example, sepsis, surgery, and trauma have varying effects on REE; hence, equations based solely on weight, height, age, and sex are likely to be unreliable for critically ill pediatric patients. In recent years, the bedside measurement of REE has become technically easier to perform and allows clinicians to accurately measure the basal energy needs of critically ill children. However, it is not widely used, and most clinicians continue to use the Harris-Benedict, WHO, Schofield, or White equations referenced above to estimate REE. Given the significant risks related to overfeeding or underfeeding in this population, accurate estimates of energy expenditure are needed to determine optimal nutrition support. The purpose of this study was to compare measured REE (MREE) with estimated REE using the Harris-Benedict, WHO, Schofield, and White et al17 equations in critically ill children with severe head injury and to evaluate the accuracy of these estimates.
Design/Procedures This retrospective, exploratory study was conducted in the PICU of our institution. Approval for the study was obtained from our institutional review board. The sample was composed of mechanically ventilated children with severe head injury who were admitted to the PICU between 1996 and 2001 and who had a metabolic study performed within the first 24 hours of admission. Severe head injury was defined by an initial Glasgow Coma Scale (GCS) score of 8 or less.18 Children who required inotropes to support blood pressure or pentobarbital to treat increased intracranial pressure were excluded as these interventions significantly affect energy expenditure.19–21 Demographic data and clinical information obtained included age, gender, height, weight, vital signs, intracranial pressures, laboratory data, and medications. Severity of illness was estimated using the Pediatric Risk of Mortality (PRISM) II score. The PRISM II score is calculated using vital signs and laboratory data from the first 24 hours of admission22 and has been validated extensively through studies in the United States and Europe. Our rationale for using PRISM II score to assess severity of illness instead of the Injury Severity Score (ISS) or the Pediatric Trauma Score was that our study population had single-organ involvement and hence the ISS or the pediatric trauma score would not be meaningful. Consistent with our standard practice for the provision of optimal nutrition support, REE was measured in all children admitted to the PICU who were mechanically ventilated and likely to need ventilator support for at least 3 days. Measurements of REE were obtained within the first 24 hours of admission to the PICU. Measurements of REE were performed only if there was no significant air leak (ie, <5%) around the endotracheal tube and if the FIO2 was <60%. Mechanical ventilation was provided by a Puritan Bennett 7200 ventilator (Puritan Bennett, Pleasanton, CA). Mechanical ventilation was needed for these children to maintain arterial carbon dioxide within a narrow range specifically for the management of their head injury. A Deltatrac II metabolic monitor (Datex-Ohmeda, Finland) was used to measure oxygen consumption (VO2) via a paramagnetic analyzer and carbon dioxide production (VCO2) via an infrared carbon dioxide analyzer. Measurements were obtained for at least 20 minutes after a steady state was reached. Steady state was defined when readings of VO2 and VCO2 did not fluctuate by >5%. No interventions were performed for 30 minutes before or during measurements to assure a resting state. Nitrogen excretion was not determined because it contributes only a small portion (ie,  4%) to the total energy
expenditure.23
During measurements, the mixing chamber of the metabolic monitor was connected
to the expiratory port of the ventilator, and the bias flow was turned off.
The inspired gas was sampled from the outlet of the humidifier. Measured
energy expenditure was calculated from the VO2 and VCO2
using the modified Weir
equation.24 Before
each measurement, the machine was calibrated using a reference gas of 96%
oxygen and 4% carbon dioxide. In addition, the machine was tested twice a year
for accuracy by using an alcohol burn kit supplied by the manufacturer.
Data Collection and Analysis
Using a second method, differences between MREE and estimated REE were evaluated using a technique recommended by Bland and Altman27 for assessing agreement between 2 methods. Bland and Altman advocate that a mean bias close to 0 with a small standard deviation is ideal in demonstrating agreement between different methods. Correlation between the PRISM II score and MREE was evaluated using the Spearman correlation coefficient. To assess for the effect of body composition on REE, weight for age "Z" (WAZ) score was correlated with MREE. "Z" Scores for weight were computed with the Centers for Disease Control and Prevention's Anthropometric software program. For correlational analysis, MREE was standardized to body weight. A significance level of p < .05 was considered statistically significant. Statistical analysis was performed using GraphPad Prism version 4.00 for Windows (GraphPad Software, San Diego, CA).
A total of 30 children met the eligibility criteria for the study. Subject characteristics are described in Table 2. Twenty-five of these children survived to be discharged from the hospital. Sixteen of the 30 children subsequently developed significant intracranial hypertension that required treatment. The frequency distribution of estimates that varied from MREE is shown in Figure 1. Less than half of the estimated REEs were within 10% of the MREE. Of the 120 estimates obtained using the 4 formulas, only 48 estimates varied by <10%. Thirty-five estimates varied by 10%–14%, 28 estimates varied by 15%–24%, and 9 estimates varied by >25%, as shown in Figure 2.
Comparisons of measured and estimated REEs using the method of Bland and Altman27 are shown in Table 3 and Figure 3. There was poor agreement between the MREE and estimated REE using all 4 equations. The mean bias ranged from –11.7 to 50.19, depending on the equation used. The limits of agreement were very broad for all 4 equations.
Using our 30 subjects, a table (Table 4) was developed to show the number of estimates using the 4 formulas that fall within error tolerance levels ranging from <10% to >25%.
The severity of illness, measured using PRISM II scores, did not correlate with MREE. The correlation coefficient between PRISM II scores and MREE showed an r = –0.119 with a p value of .5312. Similarly, there was no correlation between WAZ and MREE (r = –0.177 with a p value of .349).
Adequate nutrition support of critically ill children is extremely important to avoid malnutrition effects such as delayed healing and recovery. In addition, critically ill and injured pediatric patients are at increased risk of developing malnutrition because of accelerated catabolism and hypermetabolism.2 Traditionally, nutrition requirements are determined by estimating energy expenditure using various equations. The most commonly used equations use age, weight, and height to estimate energy. The Harris-Benedict equation is based primarily on adult data. The Schofield equation was thought to be more accurate for use in children because Schofield15 used comprehensive statistical analysis to improve REE estimations according to the Talbot data for children. To compensate for the presence of a hypermetabolic state, White et al17 developed an equation using body temperature as a measure of metabolic stress. In 1985, the WHO published the revised energy and protein requirements for children, which is widely used by nutritionists and other clinicians today.28
Our study demonstrates that estimating energy expenditure using any of the above-mentioned equations results in a high proportion of inaccurate estimations. The Bland-Altman plot showed a large mean bias, with standard deviations ranging from 203.5 to 222.27, which amounts to about 15% of the average MREE. As shown in Table 4, using any of the formulas, a clinician would have to tolerate differences of at least 25% between MREE and the estimated REE in order for the estimates to be acceptable 90% of the time. This margin of error would be unacceptable for the provision of optimal nutrition support and could lead to significant underfeeding or overfeeding, with potential adverse effects.25,26,29
It is possible that our patients were hypermetabolic because of their injuries. In their study, Peerless et al30 conclude that seriously injured patients are hypermetabolic in the early postinjury period. Body temperature is a good indicator of the metabolic state.17 Evidence exists that an increase in body temperature will increase energy expenditure in both adults and children.31,32 In the 13 patients whose MREE was greater than the estimated REE, only 2 had a body temperature >38°C. Therefore, our data does not support the argument that the discrepancy between MREE and estimated REE is due to a hypermetabolic state. On the contrary, our data suggests that the degree of metabolic stress can be variable and, therefore, difficult to accurately predict. Severity of illness may affect the metabolic state and, therefore, influence REE. Although our group included only 1 disease type, it is possible that the severity of the head injury may have affected MREE. In our study, 16 of 30 children developed significant intracranial hypertension that required treatment, suggesting that their illness was of greater severity than the other 14 subjects. However, the MREE did not demonstrate a trend in a direction that would suggest that severity of the head injury had any influence on MREE. In 6/16 of these patients, estimated REE was above MREE by >10%. In 4/16 of these patients, MREE was below estimated REE by >10%. In addition, we were unable to correlate severity of illness as measured by PRISM II score and MREE. These results suggest that the severity of injury did not predictably influence the energy expenditure in our subjects.
In order to create a homogenous population and minimize confounding variables, we restricted our study population to 1 disease type and to an age range of 6–16 years. We also excluded patients who required inotropes because catecholamines are known to increase VO2 and REE.19 All of our patients were well-sedated, and neuromuscular blockade was used in all but 2 subjects. In spite of these criteria, we found significant discrepancies between the estimated energy expenditure and MREE in these patients. Of the estimates that varied from the MREE by >10%, approximately half were overestimates and half were underestimates. Hence, a correctional factor in the formulas to estimate REE will not correct the discrepancy between MREE and estimated REE. Although all 4 formulas for estimating REE had significant errors, the White formula had the most acceptable estimates (16/30) in our study population, followed by the Harris-Benedict formula (13/30). A potential limitation of this study is that measurements of REE were obtained during the first 24 hours of intensive care unit stay. Our rationale for using measurements from the first 24 hours was to avoid confounding variables such as the effects of feeding and the effects of secondary complications on energy expenditure. Most critically ill children are fed either enterally or parenterally after 24 hours. The specific dynamic action of enteral feeding, although small, adds to the energy expenditure. The metabolic stress of secondary complications of head injury such as infection also adds to energy expenditure. Parameters such as age, gender, weight, and height used to estimate energy expenditure using various formulas do not change significantly over the subsequent days. A second limitation of our study is that it is a retrospective study. A larger, prospective study would allow for additional analysis of the effects of overfeeding or underfeeding on this population. Measuring energy expenditure in mechanically ventilated children has become easier to perform than in the past. The instrumentation is simple and does not interfere with the patient's ventilation. Many of the ventilator manufacturers offer options for attachments that can perform measurements of REE. It is possible to accurately extrapolate 24-hour values of energy expenditure from a 20-minute measurement.33 Our results suggest that significant inaccuracies lie in estimating energy expenditure and that the calculation of optimal nutrition requirements for critically ill children should be based on the measurement, rather than estimation, of energy expenditure. We thank Denise Angst, DNSc, without whose help and guidance this project would not have been completed. We would also like to thank Anita Ross, PhD for help with statistical data analysis.
Nutrition in Clinical Practice, Vol. 21, No. 2,
175-181 (2006) This article has been cited by other articles:
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
|
|
 |
|
Top
Materials and Methods
Results
