How Accurate Are Resting Energy Expenditure Prediction Equations in Obese Trauma and Burn Patients?From 1 University of Kansas School of Medicine, Departments of 2 Surgery,3 Dietetics and Nutrition, and4 Respiratory Therapy, University of Kansas Hospital, Kansas City, Kansas. Address correspondence to: Chee-Chee H. Stucky, BS, University of Kansas School of Medicine, 3901 Rainbow Boulevard, Kansas City, KS 66160; e-mail: cstucky{at}kumc.edu.
Background: While the prevalence of obesity continues to increase in our society, outdated resting energy expenditure (REE) prediction equations may overpredict energy requirements in obese patients. Accurate feeding is essential since overfeeding has been demonstrated to adversely affect outcomes. Objectives: The first objective was to compare REE calculated by prediction equations to the measured REE in obese trauma and burn patients. Our hypothesis was that an equation using fat-free mass would give a more accurate prediction. The second objective was to consider the effect of a commonly used injury factor on the predicted REE. Methods: A retrospective chart review was performed on 28 patients. REE was measured using indirect calorimetry and compared with the Harris-Benedict and Cunningham equations, and an equation using type II diabetes as a factor. Statistical analyses used were paired t test, ±95% confidence interval, and the Bland-Altman method. Results: Measured average REE in trauma and burn patients was 21.37 ± 5.26 and 21.81 ± 3.35 kcal/kg/d, respectively. Harris-Benedict underpredicted REE in trauma and burn patients to the least extent, while the Cunningham equation underpredicted REE in both populations to the greatest extent. Using an injury factor of 1.2, Cunningham continued to underestimate REE in both populations, while the Harris-Benedict and Diabetic equations overpredicted REE in both populations. Conclusions: The measured average REE is significantly less than current guidelines. This finding suggests that a hypocaloric regimen is worth considering for ICU patients. Also, if an injury factor of 1.2 is incorporated in certain equations, patients may be given too many calories.
Key Words: resting energy expenditure • obesity • trauma • burn • Cunningham Resting energy expenditure (REE) is used in trauma and burn settings to determine the nutrition needs of patients under extremely stressed biological conditions. It is measured using different techniques (direct and indirect calorimetry, double-labeled water, and so on), but these techniques are expensive in both financial cost and staff training requirements. A faster, less expensive way of determining REE is calculating it using prediction equations. The disadvantage of using prediction equations is that they may not be accurate. When determining REE for a trauma or burn patient, the accuracy is very important. Many studies agree that strict glycemic control is optimal for appropriate healing of the critically ill patient.1,2 Especially in the obese population, hyperglycemia has been shown to increase morbidity and mortality. Specifically, these patients experience prolonged mechanical ventilation,3-5 increased susceptibility to infection and therefore extended antibiotic treatment,6,7 and multiple metabolic complications leading to increased cost to the hospital.2,8 Other studies have shown that giving inadequate nutrition to the critically ill may prolong the rate of wound healing mainly due to an insufficient amount of nitrogen for proper protein anabolism.9,10 The first and most popular of the prediction equations is that developed by Harris and Benedict in the early 1900s.11 Unfortunately, this group studied normal, healthy patients as their population to develop this equation. In the trauma and burn settings, the stress of the injury will significantly alter the REE, thereby making an equation developed for healthy individuals inaccurate under these circumstances. One way to correct for this alteration is by multiplying a stress factor based on the extent of the injury to the REE total. Another major difference from the population today is that the average body weight in the early 1900s was much less than the current average weight. Today, the population is experiencing what is called an "obesity epidemic" in which 66.2% of adults age 20-74 years have a body mass index (BMI) >25 (considered overweight) and 32.9% have a BMI >30 (considered obese).12 The accuracy of the prediction equations is jeopardized because lean body mass is considered the site of metabolism, but most of the equations are based on total weight. This means that an obese individual will have a large REE as calculated when using total weight, even though his lean body mass is only slightly greater than a person of the same height with a healthy weight.13 The Cunningham prediction equation potentially alleviates this error by taking lean body weight, rather than total weight, into consideration.14 Along with obesity often comes the problem of type II diabetes mellitus. In some studies, this has been shown to increase the REE.15 Therefore, using a standard prediction equation for a diabetic patient may lead to inaccuracy of REE. A group from Royal Prince Alfred Hospital in Australia has determined a REE prediction equation that may predict REE more accurately in patients with type II diabetes.15 Taking these concepts into account, 1 group from Rhode Island compared the calculated REE with that measured in obese hospitalized patients. The equations used in this particular study included variations of the Harris-Benedict equation and the Ireton-Jones equation. Fat-free mass and stress factors were also used in the calculations. However, the reasons for hospitalization as well as comorbidities of the patients were variable, and the effects of these factors were not specifically evaluated.16 To our knowledge, there has not been a published study evaluating REE in critical care obese trauma and burn patients specifically, nor has there been a study assessing the effect of diabetes mellitus on the REE in these 2 populations. The first goal of this study was to evaluate the accuracy of the Harris-Benedict, Cunningham, and Diabetic prediction equations compared with the REE measured by using indirect calorimetry in populations consisting of obese trauma and burn patients only. The second goal was to determine the need for and accuracy of including a stress factor to correct for the trauma or burn injury.
This was a retrospective study that included a population of adult obese trauma patients and a population of obese burn patients both with BMI  30
kg/m2. These patients were treated in the Trauma and Burn Center of
the University of Kansas Hospital from January 2003 to May 2005. Approval for
analyzing the demographic and biological data in these patients was obtained
from the University of Kansas Hospital Institutional Review Board. At the time
of presentation to the Trauma facility, the patients' height and weight were
measured to the nearest centimeter and kilogram, respectively. In the
instances where the trauma or burn was too extensive to measure immediately,
the patient's reported height and weight were recorded. This occurred in only
25% of the patients studied. Presence or absence of comorbidity with type II
diabetes mellitus was reported by the patient, family, or medical records. The REE was measured for each patient by indirect calorimetry using a metabolic cart. Values were obtained between the third and fifth postinjury day and weekly thereafter throughout the duration of the patient's stay in the Intensive Care Unit (ICU). All metabolic cart measurements were discontinued once patients were transferred out of the ICU. However, multiple measurements were used in this analysis for 2 patients in the trauma population and 1 patient in the burn population. All procedures were conducted by trained respiratory therapists at the University of Kansas Hospital. The patient was placed in a supine position and allowed to rest while remaining in an awake state. Therapists initially excluded patients with air leaks in the system and those receiving inspired oxygen greater than FiO2 of 0.6 (equal to 60%). The calorimeter was allowed to calibrate to a steady state condition. After achieving steady state condition for 5 minutes, the REE was measured. The following equations were used to predict REE.
Harris-Benedict:
Cunningham:
Diabetic:
Fat-free mass was calculated by first defining the ideal BMI for each patient to be 25 and then back-calculating the ideal weight with the patient's height. The excess weight was the difference between the actual weight and the ideal weight. Fat-free mass was then defined as the sum of the ideal weight plus 25% of the excess weight.13 The injury factor was set at 20% as an estimate for the increase in energy expenditure seen in the critically injured by means of either trauma or burn.17 The data are reported as mean ± SD as well as ± 95% confidence interval (CI). While many studies, including the Rhode Island study mentioned above, use a correlation coefficient to compare the calculated vs the measured data, this was not done in the present study.16 The danger of using a correlation coefficient in this situation is that the linear relationship between the 2 data sets is emphasized, rather than the accuracy of prediction. Therefore, comparison between the measured REE and that predicted with each equation was done using the Bland and Altman analysis.18 The results of this analysis are reported by plotting numerical bias vs mean REE [(measured + predicted)/2]. The numerical bias is defined as the difference between the measured and predicted REE values. The paired t test was used to determine significant differences between the measured and predicted REE.
A total of 28 patients were included in the study: 19 trauma patients and 9 burn patients. The average BMI for the trauma and burn populations was 35.4 and 33.9, respectively. Patient demographics for both populations are reported in Table 1.
The first analysis was conducted without the inclusion of an injury factor. The ±95% CI of the measured REE (kcal/d) was 2518.37/2061.15 in the trauma population and 2475.75/1920.59 in the burn population. The REE predicted by the Harris and Benedict equation was 2170.16/1902.2 and 2129.62/1780.47 for trauma and burn, respectively. The REE predicted by the Cunningham equation was 1817.58/1646.78 and 1831.8/1538.2 for trauma and burn, respectively. The REE predicted by the Diabetic equation was 2095.23/1862.51 and 2052.2/1720.36 for trauma and burn, respectively. Tables 2 and 3 give comprehensive reports of the REE measured and predicted for both populations. When compared with the measured values, all 3 equations underpredicted REE in both populations. Harris-Benedict underpredicted to the least extent, while Cunningham underpredicted by the largest bias at around 31% for both trauma and burn. Predictions from the Diabetic equation fell in between the other 2. Figure 1 shows Bland and Altman plots for each of the 3 prediction equations. One point worth mentioning from these results is that the average REE measured for both populations was approximately 21 kcal/kg/d.
Upon further investigation of these data sets, the question arose as to whether using multiple metabolic cart measurements per patient would effect the results of this study. While multiple measurements only existed for 2 patients in the trauma population and 1 patient in the burn population, we chose to remove the measurements taken in the nonacute injury phase (postdays 3–5) and re-analyze the data for both populations. Table 4 shows the measured REE in the acute injury phase only compared with that predicted by each equation.
The next analysis was done using an injury factor of 1.2. Comprehensive reports of REE predicted by all 3 equations multiplied by the injury factor are reported in Tables 5 and 6. According to the Bland and Altman analysis of each equation, the Harris-Benedict and the Diabetic prediction equations now overestimate REE when the injury factor is included. The Cunningham equation continues to underpredict REE with a 9% bias for both populations. Figure 2 shows the Bland and Altman plots for each of the 3 prediction equations with an injury factor included. Table 7 shows the acute injury phase results using the same factor of 1.2.
The present study was focused on populations of obese patients seen in either a trauma or burn setting. The main objective of this study was to compare the REE of each patient as calculated by various prediction equations with that measured using indirect calorimetry. After the initial REE values were predicted, an injury factor of 20% was included in the equations. Our results indicate that the most accurate equation for these populations is the Harris-Benedict equation when the injury factor is not included in the calculations. When the injury factor is taken into account, the Cunningham equation predicts the REE most accurately without overestimating the required calories for both trauma and burn populations. After removing the multiple measurements and looking only at the acute-injury phase measurements, we still see the same trends. The Harris-Benedict equation is still most accurate before adding the injury factor, while the Cunningham equation is most accurate after adding the injury factor. In the trauma population, removing the multiple measurements results in REE predictions closer to the measured results but never by more than 1% bias.
While indirect calorimetry remains the most accurate means of determining energy expenditure, it may not always be practical or even present in all ICUs (especially in the smaller institutions). In these situations, our findings may be used as a guide to estimate REE, while bearing in mind the determined accuracy of each equation. These results may also be used in large volume facilities needing a quick estimate of REE to begin a nutrition regimen before a proper nutrition consult is available. One weakness to this study is that in a few instances the patient's actual height and weight were not always available, and, therefore, reported heights and weights were used instead. If the reported data were not accurate, the measured and predicted REE may not be accurate as well. However, when these patients are excluded from the data analyses, the overall conclusions and the statistical significance of the study are not affected. Even so, since the data collection for this study, a new policy has been established so that every patient admitted to our hospital must have his/her height and weight measured. This policy was established as a result of the multiple studies showing the importance of adequate nutrition in the healing trauma/burn patient as well as the need for accurate weights used in weight-based medication dosing. Future studies on this topic done at our institution will have the benefit of consistently measured heights and weights to use in the analysis. In both situations, the Cunningham equation underestimates REE while the Harris-Benedict and the Diabetic equations both overestimate REE when the 20% injury factor is included. This form of analysis differs from the study conducted in Rhode Island where only the absolute value was used when the bias of each equation was calculated.16 Newest trends in obese critically ill patients indicate that a hypocaloric, high protein nutrition regimen is more beneficial to overfeeding by reducing the days spent in the ICU, decreased duration of antibiotic days, and a decreased number of ventilation days as well as fat weight loss as an added bonus.19 Therefore, with a hypocaloric nutrition regimen as the treatment goal, using the Cunningham equation (when indirect calorimetry is unavailable) is indeed indicated. Along this same notion, our results showed an average measured REE of 21 kcal/kg/d for both populations, which is less than the 25 kcal/kg/d recommended by the American College of Chest Physicians (ACCP) ICU guidelines.20 While it is not the intention of the authors to suggest a new guideline be set based on the results of this study, 21 kcal/kg/d would be an appropriate amount to give for adequate nutrition while preventing overfeeding and resultant hyperglycemia. On a related note, many medications are dosed based on a milligram per kilogram basis, but our study indicates that this form of dosing may be inaccurate considering the metabolic profile of the obese population. Given the small sample size in this study and the inherent heterogeneity of both trauma and burn patients, we propose future studies with a larger population to determine the best feeding regimen for obese trauma and burn patients.
Financial disclosure: none declared. Received for publication May 11, 2007. Accepted for publication March 7, 2008.
Journal of Parenteral and Enteral Nutrition, Vol. 32, No. 4,
420-426 (2008)
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
|
|
 |
|
Top
Methods
Results
