Nutrition xxx (2012) 16

Contents lists available at ScienceDirect

Nutrition
journal homepage: www.nutritionjrnl.com

Applied nutritional investigation

Description and prediction of resting metabolic rate after stroke and traumatic brain injury
David C. Frankeneld M.S., R.D. *, Christine M. Ashcraft R.D.
Department of Clinical Nutrition and Department of Nursing, Penn State Milton S. Hershey Medical Center, Hershey, Pennsylvania, USA

a r t i c l e i n f o
Article history: Received 6 June 2011 Accepted 15 December 2011 Keywords: Brain injury Stroke Critical illness Energy metabolism

a b s t r a c t
Objective: To compare the effect of stroke on the metabolic rate compared with the effect of traumatic brain injury and to determine whether the metabolic rate is predictable in both types of brain injury. Methods: Indirect calorimetry was conducted prospectively in mechanically ventilated patients within the rst 6 d of admission to a critical care unit owing to ischemic stroke, hemorrhagic stroke, isolated traumatic brain injury, or traumatic brain injury with collateral injuries. Clinical data were collected simultaneously and a predicted value of the resting metabolic rate was calculated using the Penn State equation (using body size, body temperature, and minute ventilation). Results: One hundred thirty patients were measured. Ischemic stroke showed a lower incidence of fever, a lower body temperature, and a lower resting metabolic rate than the other groups; whereas in hemorrhagic stroke, these variables were similar to the trauma groups. Sedation decreased the resting metabolic rate, but this effect seemed particular to the trauma patients. The Penn State equation predicted the resting metabolic rate accurately 72% of the time, and when its component variables of body temperature and minute ventilation were controlled in an analysis of variance, all the differences among the brain injury and sedation groups were eliminated. Conclusion: Stroke is a hypermetabolic event most of the time. Body size, temperature, and minute ventilation explain most of the variation in the resting metabolic rate after traumatic and nontraumatic brain injuries. The Penn State equation therefore predicts the resting metabolic rate in brain-injured patients no matter the mechanism of injury. 2012 Elsevier Inc. All rights reserved.

Introduction The metabolic rate after traumatic brain injury has been studied extensively, sufciently so that large systematic reviews have been published on the subject [1,2] spanning up to 24 individual studies as of 2006. In contrast, metabolic rate studies in patients with stroke are limited [38]. These six studies focused on patient groups with a variety of possibly confounding variables including type of stroke, respiratory status, sedation, curarization, and hypothermia. Four of these studies found that the resting metabolic rate was not different from that predicted for healthy people, whereas two found that the resting metabolic rate of mechanically ventilated patients with stroke was increased to a degree similar to critically ill patients with traumatic brain injury. None of these studies accounted for the
* Corresponding author. Tel.: 717-531-6042; fax: 717-531-7995. E-mail address: Dfrankeneld@hmc.psu.edu (D. C. Frankeneld). 0899-9007/$ - see front matter 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.nut.2011.12.008

potential effect of fever or sedation to explain the results, as has been done in the literature on traumatic brain injury, and none provided a clear indication of how the resting metabolic rate should be assessed clinically. The purpose of the present study was three-fold: 1) to determine whether stroke is a hypermetabolic event in mechanically ventilated patients, 2) to determine what clinical factors explain the variation in the resting metabolic rate after brain injury, and 3) to test the use of an established equation to predict the resting metabolic rate in mechanically ventilated, critically ill, brain-injured patients.
Materials and methods The protocol was approved by the investigational review board of the Penn State Milton S. Hershey Medical Center. Informed consent was waived because the study was of minimal risk, using measurements commonly performed in clinical settings. Individuals were eligible for measurement if they were admitted to a critical care unit with a traumatic or non-traumatic (stroke) brain injury, had

Please cite this article in press as: Frankeneld DC, Ashcraft CM, Description and prediction of resting metabolic rate after stroke and traumatic brain injury, Nutrition (2012), doi:10.1016/j.nut.2011.12.008

D. C. Frankeneld, C. M. Ashcraft / Nutrition xxx (2012) 16 Data scaling Given the demographics of stroke versus traumatic brain injury, it was anticipated that the stroke groups would be older, have a higher body mass index, and be more predominantly female than the trauma groups. Age, body size and composition, and sex are known determinants of the resting metabolic rate and therefore need to be controlled for. Most studies use body weight or a predicted healthy metabolic rate to scale the measured metabolic rate in illness by calculating a simple ratio of the two and comparing the ratio among patient groups. A potential problem of bias occurs with this method of scaling [18,19]. Therefore, in the present study, the resting metabolic rate data were scaled by adding the predicted healthy metabolic rate (MifinSt. Jeor) to the ANOVA as a covariate. Ratio scaling was used for only one descriptive reason and that was to compare the percentage of patients in each brain injury group who were hypermetabolic. Most studies dene hypermetabolism as a measured rate that is at least 10% above the predicted healthy rate. This guideline was used, although an additional threshold of 15% was also included because the Mifin equation sometimes misses the true metabolic rate in normal people by as much as 15% [20].

been in the critical care unit for no longer than 6 d, were mechanically ventilated, and were receiving nutritional support or had written orders to initiate nutritional support. Those who met the criteria were assessed as appropriate for measurements if they were lying quietly in bed, had no air leaks from their ventilator circuit or chest tubes, and had a fractional inspired oxygen concentration of 60% or less. All data were collected prospectively. Indirect calorimetry was performed in the early morning (05:30 to 07:30) according to established best practice methods [9]. The Deltatrac MB-100 (Sensormedics, Yorba Linda, CA) open-circuit indirect calorimeter was used for all studies [10,11]. The device was calibrated against standard gas before each use (96% oxygen, balance carbon dioxide; Viasys, Puritan Medical Products, Overland Park, KS, USA). Measurements were conducted for 30 min and resting steady-state conditions had to be met within that time. Rest was dened as a period in which the patient was not disturbed by family or caregivers, movement was minimal, and a steady state was achieved. Patients were clothed in hospital gowns and usually light blankets. Continuous feeding infusions were not interrupted for the study [9], but if the feeding was interrupted as part of the clinical care, the measurement proceeded anyway. A steady state was dened as a coefcient of variation of 10% or less for oxygen consumption and carbon dioxide production. If the measurement was interrupted or the patient started to move, two other steady-state denitions were allowed (10 min of measurement with a coefcient of variation within 10%, or 5 min with a coefcient of variation within 5%). The rst 5 min of each test was automatically discarded in all cases [9]. The resting metabolic rate was calculated from the oxygen consumption and the carbon dioxide production using the equation of Weir [12]. The clinical data were collected at the time of the indirect calorimetric measurement. These data included body temperature, minute ventilation, heart rate, mean arterial pressure, laboratory data, sedation use (continuous infusions only), inotrope use, current Glasgow Coma Scale score, height, weight, age, and reason for admission. Some of these data were used to score the sepsis-related organ failure assessment [13]. Other data were used to calculate the predicted resting metabolic rate. This calculation occurred in two steps. First, the theoretical resting metabolic rate (kilocalories per day) of the individual, if healthy, was calculated using the MifinSt. Jeor equation [14]: body weightkg 10 heightcm 6:25 age 5 female1 yes 166 5 The Mifin equation value was then inserted into the Penn State equation to predict the metabolic rate of the critically ill, mechanically ventilated individual (kilocalories per day) [15]: Mifflin 0:96 Tmax C 167 Ve L=min 31 6212 where Tmax is maximum recorded body temperature in the previous 24 h, and Ve is the minute ventilation read from the ventilator at the time of the indirect calorimetric study. For a subset of patients with body mass index of at least 30 kg/ m2 combined with an age 60 y or older, an alternative form of the Penn State equation was used [15,16]: Mifflin 0:71 Tmax C 85 Ve L=min 64 3085
 

Results One hundred thirty patients were successfully measured. The data collection period lasted from October 2006 through October 2010. Ischemic stroke (group 1) involved 30 patients, with 20% requiring decompressive craniotomy. There were 36 patients with hemorrhagic stroke (group 2). In this group, 58% had a craniotomy or craniectomy and 11% had a coil embolization of an aneurysm. Thirty-two patients had an isolated traumatic brain injury (group 3), 45% of whom had falls. The highest height was 12 feet (two instances), but in most cases the patient fell from standing height. The other 56% of patients in group 3 who did not fall sustained a brain injury in motor vehicle collisions. Group 4 consisted of 33 patients who developed a traumatic brain injury as part of a multiple-injury trauma. In this group, only 12% of the patients were injured in falls, these being from 20 to 75 feet (plus one patient who was thrown from a horse and then trampled). The remaining 88% of patients in this group were injured in motor vehicle crashes. These subject numbers had the ability to detect differences of 260 to 286 kcal/d with a power of 0.8. Thirty-seven of the 130 patients were included in a previous validation study of the Penn State equation that was not conned to brain injury [15]. At the time of the study, 10 patients had been without feeding in the 24 h before measurement. In addition, at the conclusion of the measurement, six patients were found to have been overfed. An analysis of the resting metabolic rate by the feeding level showed 2002 425 kcal/d for the unfed group, 1810 418 kcal/ d for the overfed group, and 1982 249 kcal/d for all other patients (rates adjusted using the MifinSt. Jeor value as a covariate). All patients were ventilated with a Drager Evita XL (Dragerwerk AG & Co. KGaA, Lubeck, Germany) or a Servo-I (Maquet GmbH & Co. KG, Rastatt, Germany). In 4% of cases, the fraction of inspired oxygen was 0.60; in the remaining cases, it was lower than 0.60. In 5% of cases, the positive end-expiratory pressure was 14 cm of water, with the remaining cases showing less than 12 cm of water. Forty-four percent of patients had an arterial blood gas measurement within 3 h of the indirect calorimetric measurement. Of the patients with data, the partial pressure of carbon dioxide was 37.6 6.4 mmHg for ischemic stroke, 34.7 7.8 mmHg for hemorrhagic stroke, 37.2 7.0 mmHg for isolated traumatic brain injury, and 42.3 4.1 mmHg for multiple traumas (P < 0.05, hemorrhagic stroke versus multiple traumas). Further descriptive information is provided in Tables 1 and 2. As anticipated, there were statistically signicant differences among the groups in terms of sex distribution and the incidence of obesity (body mass index !30.0 kg/m2). The traumatic brain

Statistical methods An analysis of proportions was used to analyze all frequency data (P < 0.05). The general linear model of analysis of variance (ANOVA) was used for the analysis of the resting metabolic rate. Two xed factors were of interest in this study: type of brain injury (ischemic stroke, hemorrhagic stroke, isolated traumatic brain injury, or multiple traumas including brain injury) and the use or absence of continuous sedation. Variations from body temperature and minute ventilation were controlled by entering these variables into the ANOVA as covariates. If the ANOVA indicated that a statistically signicant difference occurred among the groups, the Fisher method was used to determine in which groups that difference existed. Two possibly confounding events were tested with ANOVA before the analysis of the brain injury types. These were the mechanism of traumatic injury (fall versus non-fall) and the day of the study (<4 d versus 46 d after admission to the intensive care unit). Variability in the data was expressed as a standard deviation in all cases. The bias between the predicted and the measured metabolic rates was measured by examining the 95% condence interval of the mean differences between the two values. An interval that excluded zero indicated bias. An agreement between the predicted and measured values was determined by the percentage of the estimated metabolic rates occurring within 10% of the measured value. Large errors were dened as differences between the measured and predicted values that were 15% or higher. Minitab Release 14 (State College, PA, USA) was used for all analyses. Limits of agreement were calculated according to the BlandAltman method, being 1.96 standard deviations from the mean difference between the measured and calculated resting metabolic rates [17].

Please cite this article in press as: Frankeneld DC, Ashcraft CM, Description and prediction of resting metabolic rate after stroke and traumatic brain injury, Nutrition (2012), doi:10.1016/j.nut.2011.12.008

D. C. Frankeneld, C. M. Ashcraft / Nutrition xxx (2012) 16 Table 1 Descriptive data for the four brain injury groups Group 1: ischemic stroke Men Obese (BMI !30 kg/m2) Age !65 y Sedated (continuous infusion) Febrile in previous 24 h (temperature !38 C) Febrile at time of study (temperature !38 C) Hypermetabolism (>15% above Mifin) Hypermetabolism (>10% above Mifin) 53 47 60 43 33 7 60 73 Group 2: hemorrhagic stroke 58 50 28 64 61 31 86 94 Group 3: isolated traumatic brain injury 87 23 39 71 58 16 74 87

Group 4: multiple traumas with brain injury 67 27 24 88 61 24 79 94

BMI, body mass index Data are presented as percentages. All variables show at least one statistically signicant difference among the groups.

injury groups contained more men and fewer obese patients than either stroke group. The ischemic stroke group contained more people 65 y or older than did any other group. In general, the traumatic brain injury groups were younger and contained fewer obese individuals and more male patients than the stroke groups. The incidence of fever (body temperature !38 C) during the metabolic rate measurement and at any time in the 24-h period before measurement was lower in the ischemic stroke group, and this translated to signicantly lower mean body temperatures for the patients with ischemic stroke (Table 2). In the febrile members of each brain injury group, there were no differences in body temperature, indicating that the lower temperature in the ischemic stroke group was due to the incidence and not to the intensity of fever. Hypermetabolism by either denition (10% or 15% increase) was common in all four brain injury groups, but, as was the case with fever, the incidence of hypermetabolism was lower in patients with ischemic stroke than in the other groups (Table 1). Patients with ischemic stroke also had a lower mean heart rate. The mean minute ventilation was not signicantly different among the brain injury types (Table 2). There were signicant differences among the groups in the percentage of patients receiving continuous sedation at the time of measurement (Table 1). Fewer patients with ischemic stroke were sedated at the time of study than those with hemorrhagic stroke or traumatic brain injury (although, by denition, all subjects were in a resting condition to be included in the data

analysis). Dosages of the various medications in the individuals of each group who received those medications are listed in Table 3. The propofol dose was signicantly higher in the isolated traumatic brain injury group compared with the hemorrhagic stroke group. No patient with ischemic stroke was sedated with propofol. The doses of fentanyl and midazolam were not significantly different among the groups, but the doses tended to increase from ischemic stroke to hemorrhagic stroke to isolated traumatic brain injury to multiple traumas with brain injury. These relations were not altered when the drug doses were scaled to body weight (data not shown). The resting metabolic rate was not signicantly different in those patients measured in the rst 3 d of admission to the critical care unit compared with the fourth through sixth days of admission (1976 283 versus 2008 264 kcal/d, P 0.544, controlled for MifinSt. Jeor equation value). In patients with traumatic brain injury, resting metabolic rate in those who had falls did not differ from those had a blunt injury during a vehicular crash (2046 224 versus 2024 226 kcal/d, P 0.738). Having conrmed that these confounding variables were not factors in the data, the analysis of stroke versus traumatic brain injury proceeded. The results of the ANOVA are presented in Tables 4 and 5. Controlling for body size, age, and sex with the MifinSt. Jeor equation value (Table 4), the resting metabolic rate of patients with ischemic stroke was signicantly lower than in the other groups, but there were no differences among the other three groups. Sedated patients had a signicantly lower resting metabolic rate than non-sedated patients. A signicant

Table 2 Demographic and descriptive data Variable Height (cm) Weight (kg) Body mass index (kg/m2) Age (y) Current body temperature ( C) Maximum body temperature ( C) Heart rate (beats/min) Minute ventilation (L/min) ICU (d) Best GCS score (day of study) SOFA score (day of study) Group 1: ischemic stroke 169 89 31.4 67 36.8 37.6 78 8.8 4 7 7 10 24 7.7* 15z 0.8z 0.9z 16x,{ 3.1 2 2 2 Group 2: hemorrhagic stroke 170 87 30.0 55 37.5 38.1 87 9.9 4 8 6 10 23 6.9y 17x 0.9 0.8 20 3.5 2 2 2 Group 3: isolated traumatic brain injury 173 80 26.9 52 37.4 38.2 85 9.6 4 7 6 8 15 4.3 23 0.7 0.9 19 3.8 1 2 2 Group 4: multiple traumas with brain injury 174 85 28.0 44 37.5 38.0 92 9.3 4 8 6 7 19 6.0 19 0.7 0.9 15 2.2 1 3 2

GCS, Glasgow Coma Scale; ICU, intensive care unit; SOFA, sepsis-related organ failure assessment Data are presented as mean SD. * P < 0.05 versus groups 3 and 4. y P < 0.05 versus group 3. z P < 0.05 versus all groups. x P < 0.05 versus group 4. { P < 0.05 versus group 2.

Please cite this article in press as: Frankeneld DC, Ashcraft CM, Description and prediction of resting metabolic rate after stroke and traumatic brain injury, Nutrition (2012), doi:10.1016/j.nut.2011.12.008

4 Table 3 Sedation data in patients receiving these drugs

D. C. Frankeneld, C. M. Ashcraft / Nutrition xxx (2012) 16

Group 1: ischemic stroke Fentanyl dose (mg/h) Midazolam dose (mg/h) Propofol dose (mg $ kg1 $ min1) 61 67 (25200) 1.7 0.8 (13) d

Group 2: hemorrhagic stroke 71 55 (50250) 3.0 1.7 (26) 35 13 (50150)

Group 3: isolated traumatic brain injury 96 61 (25200) 4.3 3.8 (115) 100 41* (50150)

Group 4: multiple traumas with brain injury 127 78 (50400) 4.4 2.8 (110) 33 15 (2050)

Data are presented as mean SD (range). * P 0.001 versus groups 2 and 4 (group 1 received no propofol).

injury by sedation interaction occurred: sedated patients with ischemic stroke had a lower resting metabolic rate than sedated patients with hemorrhagic stroke, and unsedated patients with ischemic stroke had a lower metabolic rate than unsedated patients with an isolated traumatic brain injury. There were no differences between the sedated and non-sedated patients within any other brain injury group. The addition of body temperature and minute ventilation to the ANOVA as covariates (Table 5) eliminated all the differences in the resting metabolic rate among the brain injury types and between the sedated and non-sedated groups. However, a brain injury by sedation status interaction persisted, localized to the isolated traumatic brain injury group. There was no difference in the rate of an accurate prediction by the Penn State equation among the groups. For ischemic stroke, the predicted resting metabolic rate was within 10% of the measured metabolic rate 73% of the time, for hemorrhagic stroke 72%, for isolated traumatic brain injury 71%, and for multiple traumas with brain injury 73% (overall 72%). The mean difference between the predicted and measured resting metabolic rates was 39 224 kcal/d (absolute difference 170 149). The predicted value was unbiased, with a 95% condence interval of the mean difference just barely encompassing zero (77 to 1 kcal/d). Large errors (!15%) occurred in 15% of the patients in all the groups aggregated. The limit of agreement by the Bland Altman method for the Penn State equation compared the measured resting metabolic rate was 23% to 21%. Discussion Studies of the resting metabolic rate in traumatic brain injury date to the 1970s. Since that time, many studies have been published [1,2] and reviews of nutrition in brain-injured patients continue to appear [21]. Traumatic brain injury causes

Table 4 Analysis of variance of metabolic rate among the brain groups Brain injury type Resting metabolic rate (kcal/d) Sedated Ischemic stroke Hemorrhagic stroke Isolated traumatic brain injury Multiple traumas with brain injury All 1807 2087 1890 2034 254* 253 252x 254 Not sedated 1952 2007 2307 2220 258y 254 252 252 All 1879 2047 2099 2127
k

hypermetabolism. The hypermetabolism is attenuated by continuous sedation, and it is associated with body temperature. In fact, when the body temperature is controlled for statistically, the sedation effect disappears [22]. In contrast to traumatic injury, the resting metabolic rate after stroke has not been extensively studied. Two studies, one a small pilot study, have suggested that hemorrhagic stroke causes hypermetabolism to a degree similar to traumatic brain injury in mechanically ventilated patients [4,8]. In contrast, a larger study of patients with ischemic and hemorrhagic stroke requiring mechanical ventilation found that the resting metabolic rate was not increased above the predicted healthy resting rate for either type of stroke [7]. The same result was found for patients with stroke who did not require mechanical ventilation [3,5]. In the present study, hypermetabolism was found to be commonplace in patients with stroke. However, there was a distinction between ischemic and hemorrhagic stroke, with ischemic stroke showing a lower incidence of fever, a lower body temperature, a lower heart rate, less frequent hypermetabolism, and a lower resting metabolic rate, whereas in hemorrhagic stroke, these variables seemed to be more at the level found for traumatic brain injury. All differences in the resting metabolic rate among the brain injury groups were eliminated when the effect of the maximum body temperature and minute ventilation was accounted for. The inuence of the body temperature on the increasing metabolic rate has been reported for general critical care patients [23,24] and traumatic brain injury [22]. The temperature effect is generally seen as a reection of the inammatory response on the metabolic rate. In patients with stroke, there is another potential link among body temperature, inammation, and the metabolic rate. In the rst week after stroke, the body temperature correlates with the lesion volume (r 0.44 on day 3) and stroke severity (r 0.36 on day 3) [25,26]. Thus the febrile patients in the present study may have had more signicant injury, leading to a higher temperature and a higher metabolic rate. The association between the minute ventilation and the metabolic rate may be a cause (increased work of breathing leading to increased metabolic rate) or an effect (the higher the resting metabolic rate, the more fuel must be
Table 5 Analysis of variance of metabolic rate among the brain groups Brain injury type Resting metabolic rate (kcal/d) Sedated Ischemic stroke Hemorrhagic stroke Isolated traumatic brain injury Multiple traumas with brain injury All 1903 2076 1915 2062 205* 198 199y 200 Not sedated 1954 1937 2187 2120 203 201 202 199 All 1929 2007 2051 2091
z

255z 263 278 387

1955 266{

2121 293

The Mifin equation value held as a covariate. Data are presented as mean SD. * P 0.037 versus sedated hemorrhagic stroke, 0.003 versus non-sedated isolated traumatic brain injury. y P 0.021 versus non-sedated isolated brain injury. z P 0.048 versus hemorrhagic stroke, 0.009 versus isolated traumatic brain injury, and 0.016 versus multiple traumas with brain injury. x P 0.025 versus non-sedated isolated traumatic brain injury. { P 0.002, all sedated versus all non-sedated. k P 0.004, brain injury type by sedation interaction.

205 208 220 304

1989 211

2050 236

The Mifin equation value, body temperature, and minute ventilation held as covariates. Data are presented as mean SD. * P 0.038, y P 0.020 versus non-sedated isolated traumatic brain injury. z P 0.002, brain by sedation interaction.

Please cite this article in press as: Frankeneld DC, Ashcraft CM, Description and prediction of resting metabolic rate after stroke and traumatic brain injury, Nutrition (2012), doi:10.1016/j.nut.2011.12.008

D. C. Frankeneld, C. M. Ashcraft / Nutrition xxx (2012) 16

oxidized, creating more carbon dioxide that must be excreted, thus increasing the minute ventilation). The non-sedated patients had a higher resting metabolic rate than the sedated patients, but this was due largely to an effect that was conned to the traumatic injury groups. Others have made this same observation in patients with traumatic brain injury [22]. It is important to emphasize that all patients in the present study were observed to be at rest and met a priori criteria for a steady state, whether they were sedated or not. In other words, the effect of sedation in this study was not due to spontaneous motion. As with the differences among the brain injury groups, the body temperature and minute ventilation explain the differences in the resting metabolic rate in sedated and non-sedated patients. The mechanism of trauma (a fall versus a motor vehicle accident) among the isolated and multiple trauma groups was analyzed because differences in the amount of energy transferred during the injury might lead to different levels of brain damage and thus different levels of hypermetabolism in the aftermath of the traumatic injury. The partitioning of subjects by measurement the day after injury was undertaken to determine if the potential presence of infection might alter the metabolic rates of the patients. Infection data were not collected in the subjects, but the authors reasoned that more infections would have been encountered from day 4 to day 6 than from day 1 to day 3, and that infection could result in a higher metabolic rate. Neither of these potential confounding variables (mechanism of trauma or day of measurement) was found to have affected the data. As in nearly all studies of the resting metabolic rate in critically ill patients, feedings were not interrupted during the measurements. Therefore, there is the potential for a thermogenic effect of feeding to have increased the metabolic rate in the present study. However, several factors mitigate this. The thermogenic effect of feeding is minimized by tube feeding versus eating [27], and it is further minimized by continuous rather than bolus feeding [28]. The thermogenic effect of feeding is less pronounced if overfeeding is avoided [28]. In the present study, all patients were intubated and receiving feeding through an enteral tube, all feedings were delivered continuously, and overfeeding occurred in fewer than 5% of patients. The metabolic rate was not different between the patients who received feeding and those few who did not. The rationale for the Penn State equation is that no matter why a patient is admitted to a critical care unit, the metabolic rate will be determined mostly by the metabolic body size and the degree to which illness alters the body temperature and minute ventilation. The relations may be causal or purely associational, but to the extent that the body temperature and minute ventilation are controlled by the inammatory state, the degree of inammatory response is a major predictor of hypermetabolism in critically ill patients, so much so that the reason for admission is not even a variable in the equation. This rationale seems to be conrmed in the present study. In a group of critically ill patients who had in common the presence of brain damage, but with different mechanisms (ischemic stroke, hemorrhagic stroke, falls, motor vehicle crashes, pedestrians struck by motor vehicles, isolated brain injury, presence of collateral injury, surgery), the Penn State equation was able to predict the metabolic rate accurately more than 70% of the time without considering the injury mechanism. Furthermore, the equation overcame the metabolic rate differences by sedation status. Large errors were distributed similarly among the groups (15% incidence). These results indicate that the Penn State

equation is a reasonable method for predicting the metabolic rate in brain-injured patients when indirect calorimetry is not available or not indicated. Conclusion Stroke is usually a hypermetabolic event severe enough to require mechanical ventilation in a critical care unit. Compared with traumatic brain injury, ischemic stroke (but not hemorrhagic stroke) results in a smaller increase in the resting metabolic rate, but the differences in the resting metabolic rate among brain injury types are eliminated when the body temperature is considered. Sedation decreases the resting metabolic rate, particularly in traumatic brain injury, but a consideration of the body temperature and minute ventilation corrects this effect for the most part. All of these inter-relations are accounted for by using the Penn State equation to predict the resting metabolic rate, with more than 70% of all predictions occurring within 10% of the actual metabolic rate. References
[1] Foley N, Marshall S, Pikul J, Salter K, Teasell R. Hypermetabolism following moderate to severe traumatic acute brain injury: a systematic review. J Neurotrauma 2008;25:141531. [2] Krakau K, Omne-Ponten M, Karlsson T, Borg J. Metabolism and nutrition in patients with moderate and severe traumatic brain injury: a systematic review. Brain Inj 2006;20:34567. [3] Weekes E, Elia M. Resting energy expenditure and body composition following cerebro-vascular accident. Clin Nutr 1992;11:1822. [4] Kasuya H, Kawashima A, Namiki K, Shimizu T, Takakura K. Metabolic proles of patients with subarachnoid hemorrhage treated by early surgery. Neurosurgery 1998;42. 126847. [5] Finestone HM, Greene-Finestone LS, Foley NC, Woodbury G. Measuring longitudinally the metabolic demands of stroke patients. Resting energy expenditure is not elevated. Stroke 2003;34:5027. [6] Bardutzky J, Georgiadins D, Kollmar R, Schwab S. Energy expenditure in ischemic stroke patients treated with moderate hypothermia. Intensive Care Med 2004;30:1514. [7] Bardutzky J, Georgiadis D, Kollmar R, Schwarz S, Schwab S. Energy demand in patients with stoke who are sedated and receiving mechanical ventilation. J Neurosurg 2004;100:26671. [8] Esper DH, Coplin WM, Carhuapoma JR. Energy expenditure in patients with nontraumatic intracranial hemorrhage. JPEN J Parenter Enteral Nutr 2006;30:715. [9] Compher C, Frankeneld DC, Keim N, Roth-Yousey L. Best practice methods to apply to measurement of resting metabolic rate in adults: a systematic review. J Am Diet Assoc 2006;106:881903. [10] Phang PT, Rich T, Ronco J. A validation and comparison study of two metabolic monitors. JPEN J Parenter Enteral Nutr 1990;14:25961. [11] Chiolero RL, Bracco D, Revelly JP. Does indirect calorimetry reect energy expenditure in the critically ill patient? In: Wilmore DW, Carpentier YA, editors. Metabolic support of the critically ill patient. Berlin: SpringerVerlag; 1993. p 95. [12] Weir JBV. New methods for calculating metabolic rate with special reference to protein metabolism. J Physiol 1949;109:19. [13] Vincent JL, Moreno R, Takala J, et al. The SOFA (sepsis-related organ failure assessment) score to describe organ dysfunction/failure. Intensive Care Med 1996;22:70710. [14] Mifin MD, St Jeor ST, Hill LA, Scott BJ, Daugherty SA, Koh YO. A new predictive equation for resting energy expenditure in healthy individuals. Am J Clin Nutr 1990;51:2417. [15] Frankeneld 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;33:2736. [16] Frankeneld DC. Validation of an equation for resting metabolic rate in older obese, critically ill patients. JPEN J Parenter Enteral Nutr 2011;35:2649. [17] Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:30710. [18] Raubenheimer D. Problems with ratio analysis in nutritional studies. Funct Ecol 1995;9:219. [19] Kronmal RA. Spurious correlation and the fallacy of the ratio standard revisited. J R Stat Soc 1993;156:37992. [20] Frankeneld DC, Rowe WA, Smith JS, Cooney RN. Validation of several established equations for resting metabolic rate in obese and nonobese people. J Am Diet Assoc 2003;103:11529.

Please cite this article in press as: Frankeneld DC, Ashcraft CM, Description and prediction of resting metabolic rate after stroke and traumatic brain injury, Nutrition (2012), doi:10.1016/j.nut.2011.12.008

D. C. Frankeneld, C. M. Ashcraft / Nutrition xxx (2012) 16 [25] Audebert HJ, Rott MM, Eck T, Haberl RL. Systemic inammatory response depends on initial stroke severity but is attenuated by successful thrombolysis. Stroke 2004;35:212833. [26] Boysen G, Christensen H. Stroke severity determines body temperature in acute stroke. Stroke 2001;32:4137. [27] Garrel DR, deJonge L. Intragastric vs. oral feeding: effect on the thermogenic response to feeding in lean and obese subjects. Am J Clin Nutr 1994;59:9714. [28] Heymseld SB, Hill JO, Evert M, Casper K, Digirolamo M. Energy expenditure during continuous intragastric infusion of fuel. Am J Clin Nutr 1987;45:52633.

[21] Vizzini A, Aranda-Michel J. Nutritional support in head injury. Nutrition 2011;27:12932. [22] Bruder N, Raynal M, Pellissier D, Courtinat C, Francois G. Inuence of body temperature, with or without sedation, on energy expenditure in severe head-injured patients. Crit Care Med 1998;26:56872. [23] Raurich JM, Ibanez J, Marse P, Riera M, Homar X. Resting energy expenditure during mechanical ventilation and its relationship with type of lesion. JPEN J Parenter Enteral Nutr 2007;31:5862. [24] Frankeneld DC, Smith JS, Cooney RN, Blosser SA. Relative association of fever and injury with hypermetabolism in critically ill patients. Injury 1997;28:61721.

Please cite this article in press as: Frankeneld DC, Ashcraft CM, Description and prediction of resting metabolic rate after stroke and traumatic brain injury, Nutrition (2012), doi:10.1016/j.nut.2011.12.008