Journal of Parenteral and Enteral Nutrition

http://pen.sagepub.com/ Prediction of Resting Metabolic Rate in Critically Ill Patients at the Extremes of Body Mass Index
David C. Frankenfield, Christine M. Ashcraft and Dan A. Galvan JPEN J Parenter Enteral Nutr 2013 37: 361 originally published online 16 August 2012 DOI: 10.1177/0148607112457423 The online version of this article can be found at: http://pen.sagepub.com/content/37/3/361

Published by:
http://www.sagepublications.com

On behalf of:

The American Society for Parenteral & Enteral Nutrition

Additional services and information for Journal of Parenteral and Enteral Nutrition can be found at: Email Alerts: http://pen.sagepub.com/cgi/alerts Subscriptions: http://pen.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav

>> Version of Record - Apr 22, 2013 OnlineFirst Version of Record - Aug 16, 2012 What is This?

Downloaded from pen.sagepub.com by ivan osuna on October 22, 2013

457423
Metabolic Rate at Extremes of BMI / Frankenfield et al 2012

PENXXX10.1177/0148607112457423Journal of Parenteral and Enteral Nutrition / Vol. XX, No. X, Month XXXXPredicting Resting

Original Communication

Prediction of Resting Metabolic Rate in Critically Ill Patients at the Extremes of Body Mass Index
David C. Frankenfield, MS, RD1; Christine M. Ashcraft, RD1; and Dan A. Galvan, MD, FACS2

Journal of Parenteral and Enteral Nutrition Volume 37 Number 3 May 2013 361-367 2012 American Society for Parenteral and Enteral Nutrition DOI: 10.1177/0148607112457423 jpen.sagepub.com hosted at online.sagepub.com

Abstract
Background: Although estimation of energy needs by mathematical equation is common in practice, there is relatively little validation data for the equations. This is especially true at the upper and lower extremes of body size. The purpose of the current study was to provide validation data for several common equations in underweight and morbidly obese critically ill patients. Methods: In mechanically ventilated, critical care patients with body mass index 21.0 or 45.0 kg/m2, indirect calorimetry was used to measure resting metabolic rate. Several equation methods were then compared with these measurements, including the Penn State equation, Faisy equation, IretonJones equation, MifflinSt Jeor equation, Harris-Benedict equation, and American College of Chest Physicians (ACCP) standard using ideal, actual, or metabolically active body weight. Results: Accuracy (percentage of estimates falling within 10% of measured) in the morbidly obese group was highest for the Penn State equation (76%) and lowest for the ACCP standard using actual body weight (0%). For the underweight group, the Penn State equation was accurate 63% of the time, but below a body mass index of 20.5, the accuracy rate dropped to 58%. No other equation was more accurate than this in the underweight patients. Conclusion: The Penn State equation is valid in morbid obesity, but the accuracy rate is much lower in underweight critically ill patients. A modification to the equation is suggested to improve accuracy in this group. (JPEN J Parenter Enteral Nutr. 2013;37:361-367)

Keywords
critical care; energy expenditure; indirect calorimetry

Clinical Relevancy Statement

Morbid obesity and underweight present real assessment challenges to the nutrition support practitioner. In the current study, accuracy of predicting resting metabolic rate at these extremes of body mass index was examined. The Penn State equation was found to be an acceptable method of estimation in morbid obesity but not in underweight individuals.

Methods
The current study was approved by our institutional review board, and all rules of ethics and confidentiality were followed in the conduct of the study. Participants had to be mechanically ventilated and at least 18 years old with body mass index 21.0 kg/m2 or 45.0 kg/m2. Patients were recruited from trauma, surgical, medical, neuroscience, and heart and vascular critical care units. Exclusion criteria included air leaks in the ventilator circuit (chest tubes, endotracheal cuffs, etc) and obvious agitation prior to conducting the measurement.

Introduction
The Penn State equations for predicting resting metabolic rate in mechanically ventilated, critically ill patients have been validated cross-sectionally and longitudinally, across a wide range of body mass index.1-3 However, as in any normal distribution, the largest and smallest body mass indices are underrepresented in the existing data. These cut points occur at 21.0 kg/m2 and 45.0 kg/m2. Furthermore, although many studies exist examining the relationship between measured and estimated resting metabolic rate,4 very few of them focus on morbid obesity or underweight.5-8 Therefore, the goals of the current study were to validate the Penn State equation in morbidly obese and underweight critically ill patients and to also test the validity of several other equations in the same patients.

From 1The Pennsylvania State Milton S. Hershey Medical Center, Hershey, Pennsylvania, and 2Department of Surgery, Pennsylvania State University College of Medicine, Hershey, Pennsylvania. Financial disclosure: none declared. Received for publication June 6, 2012; accepted for publication July 17, 2012. This article originally appeared online on August 16, 2012. Corresponding Author: David C. Frankenfield, MS, RD, Department of Clinical Nutrition and Department of Nursing, The Pennsylvania State Milton S. Hershey Medical Center, 500 University Drive, Hershey, Pennsylvania, 17033, USA. Email: Dfrankenfield@hmc.psu.edu.

362

Journal of Parenteral and Enteral Nutrition 37(3)

Patients with amputations or spinal cord injury were excluded. Fraction of inspired oxygen had to be 60% or less.

Measurement protocol. Indirect calorimetry was conducted in the morning, between 5:00 and 7:30 AM using a standard protocol.9 The indirect calorimeter (Deltatrac MB-101; SensorMedics, Yorba Linda, CA) was warmed up and calibrated according to manufacturer instructions. Testing periods lasted for 30 minutes, with the first 5 minutes automatically discarded. The coefficient of variation for the remaining 25 minutes of data for oxygen consumption and carbon dioxide production had to be 10% or less.10 Patients had to be observed in a resting state during the test, without being disturbed by caregivers and with minimal movement. If resting steady-state conditions were interrupted, shortened periods were accepted (10 minutes of data with a coefficient of variation of 10%10 or 5 minutes of data with a coefficient of variation of 5%11). Equations. Hundreds of equations and permutations of equations exist. A selected group of these equations was examined in the current study. Among these were the MifflinSt Jeor equation,12 the Harris-Benedict equation,13 the American College of Chest Physicians (ACCP) recommendation based on kcal/kg body weight,14 the Faisy equation,15 the Ireton-Jones equations,16,17 and the Penn State equations.1,2 The MifflinSt Jeor equation is calculated as follows:
RMR (kcal/d) = Weight(10) + Height(6.25) Age(5) Female(166) + 5,

was used with ideal body weight and actual body weight in the underweight patients and actual, ideal, and calculated metabolically active body weight (MAW) in the morbidly obese patients. The Hamwi rule of thumb was used for ideal body weight calculation,19 and MAW was calculated as 25% of excess weight (actual weight ideal weight) added to the ideal body weight. For continuity with earlier studies,8 another variation of the ACCP standard was also tested (21 kcal/kg actual body weight). The Ireton-Jones equations include one specifically for obese patients and one for general critical care populations16,17: Obesity: RMR (kcal/d) = Wt(9) + Male(606) Age(12) + 1844. Nonobese: RMR (kcal/d) = Wt(5) Age(10) + Male(281) + Trauma(292) + 1925.

(4)

(5)

Two versions of the Penn State equation have been validated.1,2 The standard equation was used in all underweight patients and all obese patients younger than 60 years. However, the standard equation has been found previously to be inaccurate in patients 60 years or older with a body mass index 30 kg/ m2 or higher.1 Thus, a modified equation was developed and validated.2 Standard equation: RMR (kcal/d) = Mifflin(0.96) + Tmax(167) + Ve(31) 6212,

(6)

(1) Modified equation: RMR (kcal/d) = Mifflin(0.71) + Tmax(85) + Ve(64) 3085, (7)

where RMR is resting metabolic rate, weight is in kilograms, height is in centimeters, age is in years, and female = 1. Because the Mifflin equation was designed for healthy people, it was tested alone and in a modified version by multiplying it by 1.25 to account for illness hypermetabolism. Actual body weight was used in all Mifflin equation calculations. The Harris-Benedict was calculated with the same variables as the MifflinSt Jeor and with the same 1.25 multiplier: Men: RMR (kcal/d) = 67 + Weight(13.75) + Height(5) Age(6.8). Women: RMR (kcal/d) = 655 + Weight(9.6) + Height(1.85) Age(4.7).

(2)

where Mifflin is the MifflinSt Jeor equation calculated from actual body weight, height, age, and sex12; Tmax is maximum body temperature in degrees centigrade; and Ve is expired minute ventilation in L/min recorded from the mechanical ventilator at the time of the indirect calorimetry measurement. The Faisy equation15 is structured similarly to the Penn State equation, but it uses body weight and height instead of the Mifflin to account for body size and composition, as well as current body temperature instead of maximum body temperature. RMR (kcal/d): Wt(8) + Ht(14) + Ve(32) + Temp(94) 4834.

(3)

(8)

The ACCP equation was designed for critically ill patients, so no multiplier was applied. The standard is 25 kcal/kg body weight, with a 10%20% increase for patients with systemic inflammatory response syndrome (SIRS). When SIRS was identified in the current study, the energy calculation was increased to 28 kcal/kg body weight. SIRS was defined by the presence of at least 2 factors (temperature >38C or <36C, heart rate >90 beats/min, PaCO2 <32 mm Hg, and leukocytes >12,000 cells/mL).18 The ACCP equation is often used with modified body weights, so in the current study, the equation

Other data. At the conclusion of a successful indirect calorimetry study, clinical data were collected. These data included the variables necessary to calculate the metabolic rate equations, ventilator settings, and laboratory, culture, and radiologic data for calculating sequential organ failure assessment (SOFA) score and determining SIRS status. Statistics. Bias between the estimated and measured value for
resting metabolic rate was determined by calculating the 95%

Frankenfield et al
Table 1. Descriptive Data. Parameter ICU day of study, d Height, cm Weight, kg Body mass index, kg/m2 Age, y Heart rate, beats/min Maximum body temperature, C Body temperature during calorimeter study, C Minute ventilation, L/min Partial pressure CO2, torr End-tidal CO2 PaO2/FIO2 SOFA score % Sedated % Receiving inotropes/pressor agents % Febrile % with SIRS % with sepsis Morbidly Obese 5.6 3.9 (220) 167 11 (142183) 162 43 (100334) 57.7 12.2 (45.0112.0) 53 14 (2279) 84 16 (56125) 37.5 0.9 (35.339.7) 36.8 0.08 (35.139.5) 10.0 2.9 (4.917.0) 44.3 9.9 (24.573.0) 39 9 (1861) 215 85 (77440) 8 3 (318) 77 28 31 60 24 Underweight

363

5.4 3.9 (118) 168 10 (145188) 53 8 (3471)a 18.8 1.9 (12.220.9)a 55 23 (1894) 88 22 (44128) 37.5 0.9 (34.640.0) 36.7 0.9 (33.138.0) 7.4 2.6 (3.514.8)a 38.3 11.8 (22.974.5)a 34 8 (2160)a 295 132 (70625)a 7 3 (215) 63 23 25 63 21

Values are presented as mean SD (range), except where otherwise indicated. ICU, intensive care unit; PaO2/FIO r, ratio of partial pressure of oxygen and fraction of inspired oxygen; SOFA, sequential organ failure assessment; SIRS, systemic inflammatory response. a P < .05.

confidence interval of the difference (estimated measured). If the interval excluded zero, the estimate was considered biased. If the interval was positive, it represented an overestimation and, if negative, represented an underestimation of the true resting metabolic rate. Accuracy rate was calculated as the percentage of individual estimates falling within 10% of the measured value. For added clarity of the accuracy of the equations, the percentage of individual estimates falling within 5% of measured was also reported.

Results
A total of 55 participants with body mass index 45.0 kg/m2 were studied. At least 7 additional patients with body mass index in excess of 50.0 kg/m2 were encountered but not measured because their fraction of inspired oxygen was >60% the entire time they were on mechanical ventilation (because they either died or were rapidly weaned and extubated after the oxygen concentration was reduced). Thirteen of the 55 patients existed in previously reported validation data (ie, none of these patients were used in the development of the Penn State equations). Fifty-six patients with body mass index 21.0 kg/m2 were studied (7 already existing and not used in the development of the equations). Descriptive data for both groups are in Table 1. The resting metabolic rate of the underweight group was significantly lower than in the morbid obesity group (1499 335 vs 2503 561 kcal/d, P < .0001), but when body weight was controlled for, the rates equalized (2058 544 vs 1933 541 kcal/d, P = .342). The same effect occurred when metabolically active body weight was substi-

tuted for body weight in morbid obesity (2008 426 kcal/d for underweight vs 2002 423 kcal/d for morbid obesity, P = .955) or if metabolic rate was normalized by the calculated MifflinSt Jeor value (1949 372 vs 2045 371 kcal/d, P = .261). On a ratio basis, resting metabolic rate per kilogram of body weight in underweight patients (28.5 6.4 kcal/kg) was similar to the resting metabolic rate per kilogram of metabolically active body weight in the morbidly obese (27.6 3.4 kcal/kg). The morbidly obese sample was similar in age to the underweight but had more women (62% vs 46%) and fewer trauma patients (11% vs 22%). The underweight group included 4 patients with cystic fibrosis. The incidence of fever some time during the 24 hours prior to study was similar between groups (25% in underweight vs 31% in morbid obesity), and maximum body temperatures and temperatures during the resting metabolic rate measurement were similar between the groups. Minute ventilation was significantly lower in the underweight group than in the morbidly obese group. Controlling for body size (actual weight in underweight patients and metabolically active body weight in morbidly obese patients), the difference in minute ventilation became insignificant (underweight 8.5 3.9 vs morbidly obese 8.9 3.8 L/min, P = .656). Despite equivalent minute ventilation, both PaCO2 and end-tidal CO2 were higher in the morbidly obese than in the underweight patients. Oxygenation as determined by the ratio between PaO2 and FIO2 was lower in the morbidly obese group than in the underweight group. A summary of the validation statistics is shown in Tables 2 and 3. The Penn State equations had the highest percentage of

364

Journal of Parenteral and Enteral Nutrition 37(3)

Table 2. Validation Statistics for Tested Equations in the Morbidly Obese Group. Accuracy Rate Accuracy Rate (10% of Measured (5% of Measured RMR) RMR) 76 80 69 29 0 58 27 7 55 25 60 13 51 38 38 16 0 25 15 2 38 11 36 9 Largest Underestimation (% of Measured RMR) 18 18 21 23 27 27 43 11 37 21 36 20 Largest Overestimation (% of Measured RMR) 34 34 23 61 192 31 10 119 34 68 67 109

Equation Penn State (1 and 2)b Penn State (1 only)c Faisy Ireton-Jones ACCP (25 kcal/kg wt) ACCP (25 kcal/kg MAW) ACCP (25 kcal/kg IBW) ACCP (21 kcal/kg wt) MifflinSt Jeor MifflinSt Jeor 1.25 Harris-Benedict Harris-Benedict 1.25

Biasa 33 to 97 9 to 137 6 to 143 283 to 509 2078 to 2575 157 to 12 616 to 403 738 to 1063 299 to 82 267 to 487 105 to 149 487 to 819

ACCP, American College of Chest Physicians; IBW, ideal body weight; MAW, metabolically active body weight; RMR, resting metabolic rate. a Bias is the 95% confidence interval of the difference between estimated and measured metabolic rate. An interval that excludes zero is biased (positive values are bias toward overestimation and negative values are bias toward underestimation). b Equation 1 for patients younger than 60 years, equation 2 for patients 60 years or older. c Equation 1 for all patients.

Table 3. Validation Statistics for Tested Equations in the Underweight Group. Accuracy Accuracy Rate rate (10% of (5% of Measured Measured RMR) RMR) 63 52 25 53 43 23 43 19 58 25 52 38 31 15 30 20 13 24 8 26 16 27 Largest Underestimation (% of Measured RMR) 27 15 5 26 52 12 64 40 24 38 23 Largest Overestimation (% of Measured RMR) 44 52 95 49 57 88 18 23 54 28 60

Equation Penn State equation Penn State equation (using IBW) Ireton-Jones equation Faisy equation ACCP (2528 kcal/wt) ACCP (2528 kcal/kg IBW) ACCP (21 kcal/kg wt) Mifflin StJeor equation Mifflin StJeor 1.25 Harris-Benedict Harris-Benedict 1.25

Biasa 95 to 13 48 to 157 292 to 450 78 to 188 165 to 15 237 to 373 464 to 310 317 to 196 13 to 112 297 to 171 19 to 145

ACCP, American College of Chest Physicians; IBW, ideal body weight; RMR, resting metabolic rate. a Bias is the 95% confidence interval of the difference between estimated and measured metabolic rate. An interval that excludes zero is biased (positive values are bias toward overestimation and negative values are bias toward underestimation).

estimates falling within 5% and 10% of measured. In the morbidly obese group, the Penn State equations predicted the resting metabolic rate to within 10% in 76% of the cases, with 67% of these accurate predictions falling within 5% of measured. Errors (predictions falling outside of 10% of the measured value) were evenly divided between overestimations and underestimations (46% of errors were overestimates). The equation was unbiased. Of the 55 morbidly obese patients, 19 were older than 60 years, and thus the modified Penn State equation was used. In

these 19 patients, the modified equation was accurate 78% of the time, but the standard equation was accurate to within 10% of measured 89% of the time. Use of the standard equation in all morbidly obese patients regardless of age increased the overall accuracy in the group to 80%. On the other hand, the standard equation predicted the true resting metabolic rate to within 5% less frequently than did the modified equation in this subset of patients (38% vs 51% of all patients). Table 4 shows the change in predicted value in these 19 older obese patients when the standard vs modified equation was used.

Frankenfield et al
Table 4. Differences in Predicted Resting Metabolic Rate With the Standard and Modified Penn State Equations in Older Obese Participants. Participant 1 2a 3 4a 5a 6 7 8a 9 10 11 12 13 14 15 16 17 18 19 Measured RMR, Modified PSU, kcal/d kcal/d 1920 2335 2562 1816 1588 1962 2213 2252 1946 2517 2011 2273 1736 2471 2116 2160 1802 2759 1557 2099 2042 2429 1614 1729 1893 2198 1944 1829 2527 1978 2115 1489 2533 2063 2233 1817 2628 1500 Difference, kcal/d 179 293 133 202 140 69 15 309 117 10 33 158 247 62 53 73 15 131 57 Difference, % of Standard PSU, Measured kcal/d 0.09 0.13 0.05 0.11 0.09 0.04 0.01 0.14 0.06 0.00 0.02 0.07 0.14 0.02 0.03 0.03 0.01 0.05 0.04 2022 2214 2717 1715 1861 1975 2352 2038 2092 2597 2083 2124 1494 2683 1981 2362 1966 2987 1553 Difference, kcal/d 102 121 155 102 272 13 138 215 146 80 72 149 243 212 135 202 164 228 4

365

Difference, % of Measured 0.05 0.05 0.06 0.06 0.17 0.01 0.06 0.10 0.07 0.03 0.04 0.07 0.14 0.09 0.06 0.09 0.09 0.08 0.00

PSU, Penn State University; RMR, resting metabolic rate. a Patient in which the 1 equation predicted within 10% of measured and the other did not.

Error (percentage of measured value)

0.5 0.4 0.3 0.2 0.1 0.0 -0.1 -0.2 -0.3 10 20 30 40 50 60 70 80 90 100 110 120

Group Underweight Morbidly obese

Error (percentage of measured value)

0.5 0.4 0.3 0.2 0.1 0.0 -0.1 -0.2 -0.3 12 13 14 15 16 17 18 19 20 21

Body Mass Index (kg/meter squared)

Figure 1. Errors in estimated values for the Penn State equation in the morbidly obese and underweight groups (as a percentage of measured value).

Body Mass Index (kg/meter squared)

Figure 2. Errors in estimated values for the Penn State equation, with isolation on the underweight group.

In the 56 underweight patients, the Penn State equation predicted the true resting metabolic rate to within 10% in 63% of the cases, with 60% of these accurate predictions falling within 5% of measured. The most common error in the underweight patients was underestimation (71% of all errors were underestimations), but the 95% confidence interval of differences indicated an

unbiased estimate (95 to 13 kcal/d). Figures 1 and 2 show the spread of differences between the estimated and measured resting metabolic rates as a function of body mass index for morbidly obese and underweight patients for the Penn State equations. The ACCP equation result was sensitive to the body weight used in the equation. For underweight patients, use of actual

366

Journal of Parenteral and Enteral Nutrition 37(3)

body weight increased the accuracy rate compared with use of ideal body weight, although neither rate exceeded 50% (Table 3). Errors were evenly distributed (53% underestimates, 47% overestimates). In the morbidly obese patients, use of actual weight and ideal weight yielded much lower accuracy rates than using calculated MAW. All permutations of ACCP were biased in one direction or another. Use of actual body weight caused 100% overestimation. Calculated MAW increased overall accuracy but created bias toward underestimation. Ideal body weight accentuated this trend while reducing overall accuracy. Changing the kcal/kg actual body weight from 25 to 21 further decreased the accuracy rate and caused bias toward overestimation. As with ACCP, the performance of the MifflinSt Jeor and Harris-Benedict equations showed a different pattern of accuracy in morbidly obese compared with underweight patients. The MifflinSt Jeor and Harris-Benedict equations without stress multipliers were more accurate in morbidly obese patients than when multiplied by 1.25, with the multiplied equation being biased toward overestimation (91% of Mifflin St Jeor errors and 96% of Harris-Benedict errors were overestimates and the 95% confidence intervals of differences excluded zero). On the other hand, multiplication of the MifflinSt Jeor and Harris-Benedict equations by a 1.25 stress factor in the underweight group markedly improved the accuracy rate compared with the unmultiplied value (from below 30% to above 50%). The Ireton-Jones equations were accurate in <30% of individuals, and the estimated resting metabolic rate was biased toward overestimation in both groups. In the morbidly obese group, 90% of errors were overestimates, and in the underweight group, 100% of errors were overestimates. The Faisy equation was more accurate in the morbidly obese group than in the underweight group, but accuracy rates were not as high as in the Penn State equation. The equation was biased toward overestimation in both the morbidly obese and the underweight groups. Among the errors, overestimation predominated (76% in the morbidly obese group, 92% in the underweight group). A threshold for accurate prediction seemed to exist in underweight patients at a body mass index of 20.5 kg/m2. In the small segment of body mass index between 20.5 and 21.0 kg/m2, the Penn State equation accurately predicted resting metabolic rate in 9 of 11 patients, but below a body mass index of 20.5 kg/m2, the accuracy rate fell to 58% (Figure 2). In the group of patients with body mass <20.5 kg/m2, the 95% confidence interval of differences between estimated and measured resting metabolic rate was 99 to 29 kcal/d. A similar observation was made for the Faisy equation (82% accurate when body mass index was 20.521.0 kg/m2 but only 44% below body mass index 20.5 kg/m2), but the difference was less pronounced for the other equations.

With accuracy no greater than 60% for any equation in the group of patients with body mass index <20.5 kg/m2, it was determined that another modification of the Penn State equation might be necessary for accurate prediction of resting metabolic rate in patients in this group: Resting metabolic rate (kcal/d) = Mifflin (0.78) + Tmax (50) + Ve (58) 1762. (9) The R2 for this equation was 0.77. Addition of heart rate (HR) at the time of study (beats/min) and current body temperature (Tcurr) marginally improved the equation (ie, increased the R2 to 0.82 and reduced the sum of squares for the predicted values from 1,456,030 to 1,298,873): Resting metabolic rate (kcal/d) = Mifflin(0.76) + Tcurr(86) + HR(2) + Ve(45)3102.

(10)

Discussion
The principal finding in the current study is that the Penn State equations perform in morbidly obese patients as well as they do in nonobese and moderately obese patients, but that performance is poor in patients with body mass index <20.5 kg/m2. A secondary finding is that the modified version of the Penn State equation used in older obese patients may not offer an advantage over the standard version in terms of accuracy rate. No study published for underweight patients has demonstrated a high rate of accurate prediction,5-7 and the current study is no exception. In prior studies, the best accuracy rate was achieved using the Ireton-Jones equation (approximately 45% of estimates falling within 10% of the measured value). In the current study using the Penn State equation, the accuracy rate was better but still only 58% in patients with body mass index <20.5 kg/m2. No clear reason for difficulty in prediction is apparent in the data, but a few minor differences are suggested. First, when the equation is in error, that error is usually an underestimation (ie, actual resting metabolic rate is higher than expected based on body size, temperature, and minute ventilation). Second, compared with the morbidly obese group, heart rate and current body temperature are more predictive of resting metabolic rate than is maximum body temperature in the underweight group. It is not known whether simply recalculating the predictive equation in underweight patients using maximum body temperature and minute ventilation (equation 9) is sufficient to improve accuracy or whether it is necessary to replace maximum body temperature with heart rate and current body temperature (equation 10). Neither of these equations should be employed without first being validated. Recalculation of the Penn State equation for underweight patients was done with some trepidation as a previous modification for older obese patients, although validated in an independent sample, has since given cause for uncertainty as to its necessity. In the current study, the accuracy of the standard

Frankenfield et al

367

version of the equation to within 10% of measured in older obese patients was slightly higher than the accuracy of the modified equation, but on the other hand, accuracy to within 5% of measured was diminished. In a previous study of longitudinal accuracy,3 the standard version of the Penn State equation again seemed to perform better than the modified version in the older obese patient, even though the original validation study2 indicated that the modification did correct a deficiency in the standard equation. Despite uncertainty as to the need for the modified Penn State equation for older obese patients, both versions of the equation performed well in the morbidly obese group in the current study. In fact, the highest accuracy rate of any of our previous validation studies was achieved in this group of patients, and the rate was much higher than the rates of the other tested equations. Most other studies of metabolic rate prediction in obese critically ill have not focused exclusively on the morbidly obese. Not all of the equations presented in the current study were examined previously, nor were the same statistical procedures used as in the current study. Therefore, comparison of the current data for morbidly obese patients with previous work is not possible. In general, however, the authors of other studies have concluded that equations such as the ACCP standard, Harris-Benedict equation, and Ireton-Jones equations are not sufficiently accurate for use in critically ill obese patients, and the current study confirms this observation in morbidly obese patients.

Conclusion
In morbidly obese patients, the standard and modified Penn State equations have shown acceptable levels of accuracy. The modified equation for older obese patients did not offer a clear advantage in the current study in terms of overall accuracy rate. Prediction of resting metabolic rate in patients with body mass index <20.5 kg/m2 was problematic with all tested equations. A modification of the Penn State equation is suggested but has not been validated.

References
1. Frankenfield DC, Coleman A, Alam S, Cooney RN. Analysis of estimation methods for resting metabolic rate in critically ill adults. JPEN J Parenter Enteral Nutr. 2009;33:27-36.

2. Frankenfield DC. Validation of an equation for resting metabolic rate in older obese, critically ill patients. JPEN J Parenter Enteral Nutr. 2011;35:264-269. 3. Frankenfield DC, Ashcraft CM, Galvan DA. Longitudinal prediction of metabolic rate in critically ill patients. JPEN J Parenter Enteral Nutr. 2012;36:700-712. 4. Frankenfield DC, Hise M, Malone A, Russell M, Gradwell E, Compher C. Prediction of resting metabolic rate in critically ill adult patients: results of a systematic review of the evidence. J Am Diet Assoc. 2007;107:1552-1561. 5. Campbell CG, Zander E, Thorland W. Predicted vs. measured energy expenditure in critically ill, underweight patients. Nutr Clin Pract. 2005;20:276-280. 6. Alberda C, Snowden L, McGafgar L, Gramlich L. Energy requirements in critically ill patients: how close are our estimates? Nutr Clin Pract. 2002;17:38-42. 7. Ahmad A, Duerksen DR, Munroe S, Bistrian BR. An evaluation of resting energy expenditure in hospitalized, severely underweight patients. Nutrition. 1999;15:384-388. 8. Amato P, Keating KP, Quercia RA, Karbonic J. Formulaic methods of estimating calorie requirements in mechanically ventilated obese patients: a reappraisal. Nutr Clin Pract. 1995;10:229-232. 9. Compher CW, Frankenfield DC, Keim N, Roth-Yousey L. Best practice methods to apply to measurement of resting metabolic rate. J Am Diet Assoc. 2006;106:881-903. 10. Petros S, Engelmann L. Validity of an abbreviated indirect calorimetry protocol for measurement of resting energy expenditure in mechanically ventilated and spontaneously breathing critically ill patients. Intensive Care Med. 2001;27:1164-1168. 11. Frankenfield DC, Sarson GY, Blosser SA, Cooney RN, Smith JS. Validation of a 5-minute steady state indirect calorimetry protocol for resting energy expenditure in critically ill patients. J Am Coll Nutr. 1996;15: 397-402. 12. Mifflin MD, St Jeor ST, Hill LA, et al. A new predictive equation for resting energy expenditure in healthy individuals. Am J Clin Nutr. 1990;51:241-247. 13. Harris JA, Benedict FG. A Biometric Study of Basal Metabolism in Man. Washington, DC: Carnegie Institute; 1919. Publication No. 279. 14. Cerra FB, Benitez MR, Blackburn GL, et al. Applied nutrition in ICU patients: a consensus statement of the American College of Chest Physicians. Chest. 1997;111:769-778. 15. Faisy C, Guerot E, Diehl JL, Labrousse J, Fagon JY. Assessment of resting energy expenditure in mechanically ventilated patients. Am J Clin Nutr. 2003;78:241-249. 16. Ireton-Jones CS, Turner WW, Liepa GU, et al. Equations for estimation of energy expenditures in patients with burns with special reference to ventilatory status. J Burn Care Rehab. 1992;13:330-333. 17. Ireton-Jones CS, Turner WW. Actual or ideal body weight: which should be used to predict energy expenditure? J Am Diet Assoc. 1991;91:193-195. 18. Silva E, Passos RDH, Ferri MB, Poli de Figueiredo LF. Sepsis: from bench to bedside. Clinics. 2008;63:109-120. 19. Hamwi GL. Therapy: changing dietary concepts. In: Danowski TS, ed. Diabetes Mellitus: Diagnosis and Treatment. New York, NY: American Diabetes Association; 1964.