Are basal metabolic rate prediction equations appropriate for female children and adolescents?

Transcription

1 Are basal metabolic rate prediction equations appropriate for female children and adolescents? WILLIAM W. WONG, NANCY F. BUTTE, ALBERT C. HERGENROEDER, REBECCA B. HILL, JANICE E. STUFF, AND E. O BRIAN SMITH United States Department of Agriculture/Agricultural Research Service Children s Nutrition Research Center, Department of Pediatrics, Baylor College of Medicine, Houston, Texas Wong, William W., Nancy F. Butte, Albert C. Hergenroeder, Rebecca B. Hill, Janice E. Stuff, and E. O Brian Smith. Are basal metabolic rate prediction equations appropriate for female children and adolescents? J. Appl. Physiol. 81(6): , The basal metabolic rate (BMR), which accounts for 50 70% of total energy expenditure, is essential for estimation of patient and population energy needs. Numerous equations have been formulated for prediction of human BMR. Most equations in current use are based on measurements of Caucasians performed more than four decades ago. We evaluated 10 prediction equations commonly used for estimation of BMR in 76 Caucasian and 42 African- American girls between 8 and 17 yr of age against BMR measured by whole-body calorimetry. The majority of the prediction equations (9 of 10) overestimated BMR by kcal/day (range, kcal/day). This overestimation was found to be significantly greater (P, 0.05) for African- Americans ( kcal/day) than for Caucasians ( kcal/day) in six equations, controlling for age, weight, and sexual maturity. We conclude that ethnicity is an important factor in estimation of the BMR and that the current prediction equations are not appropriate for accurate estimation of the BMR of individual female children and adolescents. whole body calorimetry; energy metabolism; Caucasians; African-Americans THE BASAL METABOLIC RATE (BMR) is the minimal rate of energy consumption necessary to support all cellular functions and accounts for 50 70% of total energy expenditure in humans. BMR is used routinely by clinicians for estimation of energy requirements in patient care as well as by governmental agencies and health organizations in defining population energy requirements. BMR is measured by indirect calorimetry after the subject has fasted,12 h (usually overnight), then rested motionless in a supine position min in a thermally neutral environment. The subject is instructed to remain still for the next min while the indirect calorimetry measurements are taken. The procedure is not only time-consuming but requires extensive subject cooperation as well as accurate and precise flow and concentration measurements, using sophisticated flow and gas analyzers. Recognizing the significance of BMR in defining energy requirements in humans, several authors have generated simple equations for estimation of BMR based on age, body weight, height, and gender (3, 9, 22, 26). Although these equations were formulated based on BMR measurements performed between 1919 and 1952, many of these equations presently are employed by clinicians, governmental agencies, and health organizations to estimate human energy requirements. BMR measurements done.44 years ago were characterized by the use of inappropriate techniques; duplicate results and group means in the calculation; data obtained from nonfasted, sleeping, or agitated subjects; and data collected under nonstandardized conditions, such as temperature and altitude, without appropriate corrections. These concerns prompted the Food and Agriculture Organization of the United Nations (FAO)/ World Health Organization (WHO)/United Nations University (UNU) (30) and Schofield (23) to scrutinize the BMR results to exclude unacceptable data and derive more appropriate prediction equations. In a recent study, Dietz et al. (6) compared BMR values obtained for 54 adolescents with the use of the ventilation hood and indirect calorimetry with BMR values calculated from prediction equations. These authors concluded that the FAO/WHO/UNU equations (30) yielded average BMR values which did not differ significantly from the measured mean values. However, similar BMR measurements performed by Maffeis et al. (15) on 33 obese and 97 nonobese prepubertal children in Italy were found to be significantly lower than those calculated using the prediction equations, including those formulated by FAO/WHO/UNU. We describe here the use of BMR measured by whole body calorimetry to evaluate the applicability of 10 equations commonly used for prediction of BMR in Caucasian and African-American female children and adolescents of different body composition and stages of sexual maturation. We hypothesize that the prediction equations, which were derived primarily from BMR data collected in adults and in Caucasians, are not suitable for estimation of BMR in children and adolescents of different ethnic origins. METHODS Subjects. We studied 118 female children and adolescents (76 Caucasians, 42 African-Americans) between 8 and 17 yr of age. The subjects were recruited from schools in the greater Houston metropolitan area. Based on medical history, vital signs, standard clinical blood chemistries, and physical examination, all subjects were healthy at the time of the study. All subjects tested negative for pregnancy by using the QuickVue One-Step hcg-urine test (Quidel, San Diego, CA). The protocol met the Occupational Safety and Health Administration/Department of Health and Human Services guidelines for human immunodeficiency virus/hepatitis B virus occupational safety and was approved by the Baylor Affiliates Review Board for Human Subjects at Baylor College of Medicine. After thorough explanation of the procedures to the subjects and to their parents, all subjects and their parents gave written informed consent /96 $5.00 Copyright r 1996 the American Physiological Society 2407

2 2408 BMR EQUATIONS VS. INDIRECT CALORIMETRY Whole body calorimetry. Two large (34 m 3 ) and two small (19 m 3 ) room respiration calorimeters were used in the study (18). Each calorimeter is equipped with a bed, table, video and stereo equipment, exercise cycle, telephone, lavatory, commode, and intercom system for communication with the calorimetry and nursing staff. Microprocessor controllers (CMP 3244; Conviron, Winnipeg, Manitoba, Canada) maintained temperature and humidity to within 60.3 C and 65% relative humidity, respectively. Air inflow and exhaust flow rates of these chambers were controlled by thermal-mass flow controllers to within 62% full scale or 60.5% for air flows between 13 and 50 l/min (Sierra, Monterrey, CA). Concentrations of CO 2 in these chambers were maintained at 0.45%. The CO 2 concentrations in the inflow and exhaust air streams were measured with nondispersive infrared CO 2 analyzers (Ultramat 5E, Siemens, Karlsruhe, Germany). Oxygen concentrations in the inflow and exhaust air streams were measured by paramagnetic oxygen gas analyzers (Oxymat 5E, Siemens). Water content of air samples from the inflow and exhaust air streams was reduced to,0.01% with perfluorosilicate membrane dryers (PD SS; Perma-Pure, Toms River, NJ) before subjects entered the gas analyzers. Twentyfour-hour N 2 -CO 2 infusion tests indicated that errors for CO 2 production rate (V CO2 ) and oxygen consumption rate (V O2 ) using these respiration calorimeters were and %, respectively. Response times of these calorimeters were 2 6 min for V O2, which ranged from 100 to.4,000 ml/min. The gas analyzers and flow controllers were tested by N 2 -CO 2 infusion before each study. Physical movement and heart rate of each subject inside the calorimetric chamber were monitored continuously by Doppler microwave sensor (D9/50; Microwave Sensors, Ann Arbor, MI) and by telemetry (Dynascope 3300; Fukuda Denshi America, Redmond, WA), respectively. Each subject checked into the Metabolic Research Unit (MRU) of the Children s Nutrition Research Center 1 day before the calorimetric measurement and received oral and written instructions regarding the schedule, procedures, and operations of the chamber. After an overnight stay in one of the volunteer suites at the MRU, the subject was awakened at 7:00 A.M. and took a shower; a heart-rate monitor was then taped on the subject s chest above the heart. Each subject entered the chamber at 8:00 A.M. and ate breakfast at 8:30 A.M. Lunch was served at 12:00 P.M. and dinner at 5:30 P.M. All subjects remained awake until bedtime at 10:00 P.M. No food, other than caffeine-free beverages, was allowed after dinner. BMR. At 6:50 A.M. the next day, each subject was awakened, allowed to urinate, and then returned to bed. At 7:20 A.M., the subject was awakened if she was asleep and instructed to find a comfortable position in bed and remain awake and motionless for the next 40 min. The V O2 and V CO2 measurements per minute with the least movement (#50 counts) as indicated by the Doppler microwave sensor during the 40-min measurement period were converted to BMR using the Weir nonprotein equation (29) BMR(kcal/day) V O V CO 2 (1) Sound-sleep metabolic rate. To ensure that BMR was measured rather than sleeping metabolic rate, soundsleeping metabolic rate (SSMR) was extracted from total sleeping metabolic rate. SSMR was the sleeping energy expenditure with,10 counts of physical movement, as determined by the Doppler microwave sensor. A valid BMR measurement must have a BMR-to-SSMR ratio.1. Anthropometric measurements. The body weight of each subject, dressed in a hospital gown and without socks or shoes, was measured to the nearest 0.1 kg with a digital balance (Scale-Tronix, Dallas, TX). Height was measured to the nearest 1 mm with a stadiometer (Holtain, Crymmych, Pembs, UK). All measurements were performed by one investigator. Body mass index (BMI) was calculated from body weight (Wt, in kg) and height (Ht, in m) as BMI(kg/m 2 ) 5 Wt/Ht 2 (2) Sexual maturity. Tanner stages of breast and pubic hair development (27) were used to classify sexual maturity based on physical examination. BMR prediction equations. The equations formulated by Harris and Benedict (9), Boothby et al. (3), Talbot (26), Robertson and Reid (22), FAO/WHO/UNU (30), Schofield (23), and Maffeis et al. (15) were used to predict BMR of our female subjects. These equations are summarized in Table 1. Statistical analyses. The Bland-Altman comparison technique (2) was used to compare BMR values predicted by equations with values measured by whole body calorimetry. Differences between predicted and measured BMR were plotted against the averages of predicted and measured BMR. Regression analysis was used to test the relationship between these differences and means. If the slope was not significant, the relative bias (mean difference between methods) and the 95% limits of agreement (mean difference 6 2 SD of the differences) were computed. If the slope relating the differences and means was significant, the 95% limits of agreement were estimated as 2 SE of the estimate around the regression line. Table 1. Equations for prediction of basal metabolic rate Researchers and Subjects Ref. Equations Harris and Benedict 9 Boothby et al. 3 Talbot 26 Robertson and Reid 22 BMR59.6Wt11.9Ht 24.7A1655 BMR524BMR BSA BSA* Taken from BMR tables based on Wt and Ht BMR524BSA[30130/ (11e u /10)] FAO/WHO/UNU to 10-yr old BMR522.5Wt to 18-yr old BMR512.1Wt to 18-yr old BMR57.4Wt1482Ht1217 Schofield to 10-yr old BMR520.3Wt1486 BMR517.0Wt11.6Ht to 18-yr old BMR513.4Wt1693 BMR58.4Wt14.7Ht1200 Maffeis et al. 15 BMR58.6Wt13.7Ht 28.7A1371 Subjects, by age; Ref., reference numbers; FAO/WHO/UNU, Food and Agricultural Organization/World Health Organization/United Nations University; BMR, basal metabolic rate expressed as kcal/ day; Wt, body weight in kg; Ht, height in cm; A, age in yr; BMR BSA, BMR estimated from body surface area in kcal/m 2 ; BSA, body surface area in m 2 ; u is a function calculated from age. *BSA is calculated, using the Du Bois equation (7): BSA (m 2 ) Wt Ht The function, u A A A A A 5, where A 5 age.

3 BMR EQUATIONS VS. INDIRECT CALORIMETRY 2409 Table 2. Age, physical characteristics, and sexual maturity of 76 Caucasian and 42 African- American subjects Student s t-test was used to test for differences in age, anthropometric characteristics, sexual maturity, and calorimetric data between the African-American and Caucasian girls. Because Tanner stages of sexual maturity are not continuous variables, differences in sexual maturity between the two ethnic groups at various stages of development were done again by frequency analyses, using x 2 testing (Minitab, State College, PA). Univariate analysis was used to test for significant effect of sexual maturation and ethnicity on the agreement between the predicted and the estimated BMR values. After identification of the best equations for prediction of BMR, analysis of covariance (ANCOVA; Minitab) was used to assess the effect of ethnicity on the magnitude of the difference between methods while controlling for age, anthropometric characteristics, and sexual maturation. RESULTS Caucasian African- American P Values Age, yr ,0.02 Weight, kg ,0.01 Height, cm ,0.01 BMI, kg/m ,0.01 Sexual maturity, % Stage (18.3) 0 (2.7),0.01 (,0.01)* Stage (18.3) 2.7 (0) Stage (18.3) 21.6 (16.2) Stage (29.6) 8.1 (27.0) Stage (15.5) 67.6 (54.1) Values are means 6 SD. BMI, body mass index. Sexual maturity according to Tanner stages of sexual maturation, shown as %subjects with breast development; %subjects with pubic hair development in parentheses. *P values by t-test and by x 2 testing. Age, physical characteristics, and sexual maturity. Age, body weight, height, BMI, and sexual maturity of our volunteers are summarized in Table 2. The African- American girls on average were older than the Caucasian girls. Twelve subjects (11 Caucasians, 1 African- American) were below the 15th percentile of BMI, 77 (48 Caucasians, 29 African-Americans) were between the 15th and 85th percentiles, and 29 (16 Caucasians, 13 African-Americans) were above the 85th percentile (19). The African-American girls also were heavier, taller, and had higher BMI than the Caucasian girls. After Table 3. Whole body calorimetric data of 76 Caucasian and 42 African-American subjects Caucasian African- American P Values Temperature, C Relative humidity, % BMR, kcal/day 1, * 1, *,0.02 Duration of BMR, min Sound-sleep metabolic rate (SSMR), kcal/day 1,183698* 1, *,0.01 BMR/SSMR Values are means 6 SD. *Adjusted for mean body weight of 2 ethnic groups. Table 4. Comparison of BMR predicted by equations and by whole body calorimetry of 76 Caucasian and 42 African-American subjects Source of Equations Ref. %BMR by Whole Body Calorimetry Caucasian African- American P Values Harris and Benedict ,0.02 Boothby et al ,0.01 Talbot, by Wt ,0.01 Talbot, by Ht Robertson and Reid ,0.01 FAO/WHO/UNU, by Wt ,0.01 FAO/WHO/UNU, by Wt and Ht ,0.03 Schofield, by Wt ,0.01 Schofield, by Wt and Ht ,0.01 Maffeis et al ,0.01 Values are means 6 SD. controlling for age, the African-American girls remained heavier and had higher BMI than the Caucasian girls (P, 0.01). According to the Tanner stages of sexual maturation, the African-American girls were more mature for age than the Caucasian girls by Student s t-test and by x 2 testing (P, 0.01). Whole body calorimetry. Room indirect calorimetric results are summarized in Table 3. Minimal movement during SSMR measurements was detected by Doppler microwave sensor, with average physical movement of counts for the Caucasian girls and counts for the African-American girls. Physical movement as detected by the microwave sensor varied from zero to 1,600 counts for our subjects, with average counts of 10 during BMR measurements. Furthermore, the BMR values for both ethnic groups were 15% higher Table 5. Mean differences between predicted and measured BMR values of 76 Caucasian and 42 African-American subjects Source of Equations Ref. Caucasian Mean Difference, kcal/day P Values African-American Mean Difference, kcal/day P Values Harris and Benedict ,0.01 Boothby et al , ,0.01 Talbot, by Wt , ,0.01 Talbot, by Ht , Robertson and Reid , ,0.01 FAO/ WHO/UNU, by Wt ,0.01 FAO/ WHO/UNU, by Wt and Ht ,0.01 Schofield, by Wt ,0.01 Schofield, by Wt and Ht ,0.01 Maffeis et al , Values are means 6 SD.

4 2410 BMR EQUATIONS VS. INDIRECT CALORIMETRY Fig. 1.

5 BMR EQUATIONS VS. INDIRECT CALORIMETRY 2411 than the SSMR values, thus minimizing the possibility that sleeping metabolic rate was mistaken for BMR values. Among the 118 subjects, only one subject had a BMR/SSMR ratio of Because the African-American girls were heavier than the Caucasian girls, the BMR and SSMR presented in Table 3 were adjusted for the mean body weight of the two ethnic groups. The adjusted means of BMR and SSMR of the Caucasian girls were found to be significantly higher than those of the African-American girls (P, 0.02). BMR by prediction equations. The ratios of the predicted to the measured BMR expressed in percentages are shown in Table 4. With the exception of the Maffeis equation (15), the prediction equations on average overestimated ( %) the measured values of our subjects. With the exception of the BMR estimated according to height using Talbot s table (26) and the equation of Maffeis et al. (15), the overestimation expressed as percentages of the measured BMR values was significantly higher in the African-American girls than in the Caucasian girls (P, 0.03). Expressed in absolute differences between the predicted and the measured BMR values (Table 5), Student s t-test indicated that the average overestimation was statistically significant in five of the 10 equations among the Caucasian girls. Among the African-American girls, however, the overestimation of BMR by the prediction equations was statistically significant in nine of the 10 equations. Underestimation of BMR using the Maffeis et al. equation (15) was found to be statistically significant among the Caucasian girls but not among the African-American girls. Other than the BMR calculated using the equations of Robertson and Reid (22) and Maffeis et al. (15), the mean difference between the predicted and the measured BMR values was found to increase with sexual maturation by univariate analysis (P # 0.053). ANCOVA indicated that the overestimation remained significantly greater (P, 0.05) for the African-American girls ( kcal/day) than for the Caucasian girls ( kcal/day) in six of the 10 equations after controlling for differences in age, weight, and sexual maturity between the two ethnic groups. However, the effect of sexual maturation became insignificant in the ANCOVA. Among these prediction equations, the Maffeis equation (15) underestimated BMR of both the Caucasian and the African-American girls. Talbot s BMR table (26) based on height yielded BMR values with the largest SD among the 10 equations. The equation of Boothby et al. (3) overestimated BMR the most among the 10 prediction equations. According to the comparisons shown in Table 4, the equations proposed by FAO/WHO/ UNU (30) and Schofield (23) using both body weight and height in the calculation yielded the most accurate mean BMR compared with mean BMR by whole body calorimetry. The remaining equations by Harris and Benedict (9), by Talbot based on body weight (26), by Robertson and Reid (22), by FAO/WHO/UNU based on weight (30), and by Schofield based on weight (23) also yielded mean BMR similar to the mean BMR by whole body calorimetry. However, agreement between the mean BMR estimated by the prediction equations and the mean BMR by whole body calorimetry was consistently poorer with the African-American girls than with the Caucasian girls. Detailed comparisons (2) of predicted and measured BMR, as shown in Fig. 1, offer a different interpretation. With the exceptions of the BMR values derived from the table of Talbot based on height (26; (Fig. 1D), the equations of Robertson and Reid (22; Fig. 1E), FAO/WHO/UNU based on weight (30; Fig. 1F), and Schofield based on weight and height (23; Fig. 1I), the majority of the equations showed significant relationships (P, 0.03) between the individual differences in Fig. 1. Detailed comparisons of basal metabolism rate (BMR) predicted by using the equations shown in Table 1 with BMR by whole body calorimetry. A-J: solid line represents mean difference between predicted and measured BMR values. The 2 dashed lines represent upper and lower limits of agreement, calculated as mean difference 6 2 SD of differences or as 2 SE of estimate around the regression line if slope relating the between-method differences and the average BMR values was significant. Symbols represent individual differences between predicted and measured BMR values of Caucasian (s)and African-American (l) girls, respectively. Numerical values above and below the 2 dashed lines are upper and lower limits of agreement at corresponding BMR values of 900 and 2,100 kcal/day. P value is significance level for slope relating differences between predicted and measured BMR values to average BMR.

6 2412 BMR EQUATIONS VS. INDIRECT CALORIMETRY BMR (predicted BMR 2 measured BMR) and the average BMR values. As shown in Fig. 1, A-C, G, H, and J, the average differences in BMR using these equations were not constant but rather varied depending on the BMR values. For some individuals, a BMR value of 2,100 kcal/day could be underestimated by as much as 378 kcal/day or 18% (Fig. 1J, Maffeis et al.) or overestimated by as much as 503 kcal/day or 24% (Fig. 1B, Boothby et al.). At a BMR value of 900 kcal/day, these equations could underestimate an individual BMR value by 278 kcal/day or 31% (Fig. 1H, Schofield, based on weight) or overestimate it by 304 kcal/day or 34% (Fig. 1B, Boothby et al.). Among the four equations (Fig. 1D, Talbot, based on height; E, Robertson and Reid; F, FAO/WHO/UNU, based on weight; I, Schofield, based on weight and height) that yielded constant mean differences over the range of BMR values between 900 and 2,100 kcal/day, the Schofield equation yielded the smallest mean differences for both ethnic groups ( kcal/day for the Caucasians and kcal/day for the African- Americans) vs. the measured values. Mean differences using the other three equations were larger on average or more variable for individuals (Talbot, kcal/day for Caucasians and kcal/day for African-Americans; Robertson and Reid, kcal/ day for Caucasians and kcal/day for African- Americans; FAO/WHO/UNU, kcal/day for Caucasians and kcal/day for African- Americans). As shown in Fig. 1D, Talbot s BMR table based on height could underestimate or overestimate individual BMR values by as much as 287 or 387 kcal/day, respectively. Therefore, further statistical analyses were performed only on the BMR predicted, using the equation formulated by Schofield based on weight and height (23) against the BMR measured by whole body calorimetry. With the Schofield equation, the difference between predicted and measured BMR values by univariate analysis differed by ethnicity (P, 0.01) and by sexual maturation (P, 0.01). The differences were smaller among the Caucasians than among the African- Americans. The differences also were smaller at the early stages of sexual maturity in both ethnic groups. Because the African-American subjects were older, more mature, and had more body mass than the Caucasian subjects, ANCOVA was applied. This analysis indicated that the mean difference between predicted and measured BMR remained significantly affected by ethnicity (P, 0.04) after controlling for differences in age, body weight, and sexual maturation between the two ethnic groups (Table 2). DISCUSSION The majority of the equations for prediction of BMR were formulated based on BMR measurements done.40 years ago. After elimination of erroneous data due to clerical errors, repeated measurements on the same individual, duplication of data, ill subjects, and outliers, Schofield (23) produced revised equations for prediction of BMR. In his article, Schofield indicated that the revised equations worked well with Caucasians but overestimated the BMR of Indians. Other studies also reported overestimation of BMR in adults, using the available equations. For example, the Harris and Benedict equations, which were based on BMR measurements of Caucasians, have been shown to overestimate BMR in healthy adults by % (4). Because,45% of the BMR measurements used in the formulation of the Schofield (23) and the FAO/WHO/UNU (30) equations were collected from young and physically active Italian subjects, the applicability of these equations to prediction of BMR in other ethnic groups has been questioned. Indeed, the Schofield and the FAO/ WHO/UNU equations have been shown to overestimate BMR of non-european adults (5, 11, 12, 17). For children and adolescents, conflicting results were reported in four recent studies. In 1991, using the FAO/ WHO/UNU equations (30), Dietz et al. (6) reported good agreement between the predicted and the measured BMR values of 54 adolescents. However, the FAO/WHO/UNU equations were reported by Henry and Rees (12) to overestimate the BMR of children and adolescents living in the tropics. Using the Schofield equations, Spurr et al. (25) reported overestimation of BMR in mestizo boys but not in girls. The resting metabolic rates (RMR) of 33 obese and 97 nonobese Italian children between 6 and 10 yr of age were reported by Maffeis et al. (15) to be overestimated when the equations formulated by FAO/WHO/UNU (30), Robertson and Reid (22), Talbot (26) and Boothby et al. (3) for estimation of BMR were used. The overestimation was higher in the obese children than in the nonobese children. The Robertson and Reid equations (22) were based on BMR measurements done on normal people between 3 and 80 yr of age in Britain. Presumably, these subjects were primarily of European descent. No specific information on the ethnic origins of the children studied by Talbot (26) was given. The Boothby et al. (3) equation was based on BMR data collected from children attending the schools in Rochester, MN, employees of the Mayo Clinic, and patients at the clinic, but without specificity as to their ethnic origins. We assume Caucasians were the majority of the subjects in their study. Several explanations have been offered to the observed overestimation of BMR when the prediction equations were used. In a study of nine men highly trained in exercise and nine sedentary men (21), the RMR of the trained men was found to be higher than that of the untrained men. Because the BMR data used in the formulation of the Schofield and the FAO/WHO/ UNU equations consisted of a significant portion of young and physically active subjects, it is reasonable to expect that these equations will overestimate the BMR of sedentary subjects. Difference in racial abilities to produce different degrees of muscular relaxation and temperature-induced changes in thyroid gland activity that might lower BMR have been postulated to be responsible for the lower BMR in people living in the tropics (12, 16). The thrifty genotype or increased efficiency in intake and utilization of food also has been

7 BMR EQUATIONS VS. INDIRECT CALORIMETRY 2413 hypothesized to be responsible for the lower energy expenditure in Pima Indians (14, 20) and in Gambian men (17). The exaggerated overestimation of RMR in obese children might be due to the reduced diet-induced thermogenesis and RMR in obese and African-American subjects (1, 24). Based on the comparisons shown in Table 4, it is easy to misinterpret that most of the equations are appropriate for estimation of BMR, particularly in our Caucasian girls. However, the detailed comparisons shown in Fig. 1 indicated that only three equations [Robertson and Reid (22), FAO/WHO/UNU with weight (30), Schofield with weight and height (23)] yielded constant mean differences over the range of BMR values between 900 and 2,100 kcal/day. Among these three equations, the Schofield and the FAO/WHO/UNU equations yielded average BMR values that were in closest agreement with the mean values measured by whole body calorimetry. This is consistent with the most recent observation made by Kaplan et al. (13) on 102 diseased subjects between 0.2 and 10.5 yr of age. However, detailed comparisons as shown in Fig. 1 also indicated that by using these equations, individual BMR values between 900 and 2,100 kcal/day could be underestimated by 200 kcal/day (Schofield, Fig. 1H) or overestimated by 286 kcal/day (FAO/WHO/UNU, Fig. 1F). Because the Du Bois body surface area equation (7), which was validated mainly for adults, was used in the BMR prediction equations of Boothby et al. (3) and Robertson and Reid (22), the use of the Du Bois body surface area equation might not be appropriate in children and adolescents. However, we found no significant improvement in agreement between the predicted and measured BMR values when body surface areas of our subjects were calculated with the use of the equation of Haycock et al. (10). The latter body surface area equation has been validated in infants, children, and adults of various body shapes, sizes, and ethnicity. It is interesting to note that the RMR of 33 obese and 97 nonobese Italian children measured by Maffeis et al. (15) by indirect calorimetry were consistently lower than those estimated using the equations formulated by FAO/WHO/UNU (30), Robertson and Reid (22), Talbot (26), and Boothby et al. (3). Although RMR by definition is higher than BMR, the RMR values predicted using the equation formulated by Maffeis et al. (15) also were found to be lower than the measured BMR values of our volunteers (Table 4). Therefore, it is reasonable to suspect that there might be a systematic error in the RMR measurements reported by Maffeis et al. (15). In the formulation of the prediction equations, body weight has been considered to be the major determinant of BMR. Addition of height to the equations has been shown to contribute insignificantly to the accuracy and precision of the predicted BMR values. However, as shown in Fig. 1 (C vs. D, F vs. G, H vs. I), the use or inclusion of height in the prediction equations significantly changed the comparisons between the predicted and the measured BMR values in our subjects. The different responses to the inclusion of height in the prediction of BMR between the FAO/WHO/UNU and the Schofield equations could be due to the elimination of 220 outliers in the FAO/WHO/UNU database before the formulation of the Schofield equation. Therefore, height should not be completely disregarded in the BMR prediction equations. It is well documented that African-Americans grow faster and have more lean body mass than Caucasians from 2 yr of age (8). Lean body mass, estimated by total body electrical conductivity (28), of our African- American girls was significantly higher than that of the Caucasian girls. However, after controlling for differences in age, sexual maturation, and lean body mass, the overestimation of BMR using the Schofield equation with weight and height remained significantly higher (P, 0.03) among the African-American girls than among the Caucasian girls. In conclusion, we demonstrated in this study that the magnitude of the differences between predicted and the measured BMR values is significantly associated with ethnicity. The prediction equations were found to overestimate BMR more in African-American girls than in Caucasian girls. Although several of the prediction equations yielded average BMR values similar to the mean values measured by whole body calorimetry, detailed comparisons as shown in Fig. 1 indicated that significant underestimation (22%) and overestimation (31%) can occur on an individual basis. Therefore, we conclude that although some prediction equations might be appropriate for estimation of mean BMR on a population basis, they are not appropriate for estimating BMR of individual female children and adolescents. Because the magnitude of the differences between predicted and measured BMR values is significantly associated with ethnicity after controlling for differences in age, weight, sexual maturation, and lean body mass, we recommend that ethnicity should be included in future refinement of these prediction equations. The authors are indebted to the volunteers; to the staff of the Metabolic Research Unit for meeting the needs of the subjects during the study; to Dr. J. Hoyles at the Pediatric Department of Kelsey- Seybold West Clinic, Dr. M. desvignes-kendrick at the City of Houston Health and Human Services Department, X. Earlie of the Aldine Independent School District, S. Wooten at the Teague Middle School, Dr. B. Shargey and C.C. Collins at the High School for Health Professions, Mt. Carmel High School, and K. Wallace for subject recruitment; to Dr. J. Moon, M. Puyau, and F. A. Vohra for the calorimetric measurements; and to L. Loddeke for editorial assistance. This work is funded in part with federal funds from the US Department of Agriculture (USDA), Agricultural Research Service, under Cooperative Agreement 58 7MNI The contents of this publication do not necessarily reflect the views or policies of the USDA, nor does mention of trade names, commercial products, or organization imply endorsement by the US Government. Address for reprint requests: W. W. Wong, USDA/ARS Children s Nutrition Research Center, 1100 Bates St., Houston, TX Received 23 February 1996; accepted in final form 25 July REFERENCES 1. Albu, J., M. Shur, M. Curi, L. Murphy, S. Heymsfield, and F. X. Pi-Sunyer. Resting metabolic rate in African-American women (Abstract). FASEB J. 10: A727, 1996.

8 2414 BMR EQUATIONS VS. INDIRECT CALORIMETRY 2. Bland, J. M., and D. G. Altman. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1: , Boothby, W. M., J. Berkson, and H. L. Dunn. Studies of the energy of metabolism of normal individuals: a standard for basal metabolism, with a nomogram for clinical application. Am. J. Physiol. 116: , Daly, J. M., S. B. Heymsfield, C. A. Head, L. P. Harvey, D. W. Nixon, H. Katzeff, and G. D. Grossman. Human energy requirements: overestimation by widely used prediction equation. Am. J. Clin. Nutr. 42: , De Boer, J. O., A. J. H. van Es, L. E. Voorrips, F. Blokstra, and J. E. Vogt. Energy metabolism and requirements in different ethnic groups. Eur. J. Clin. Nutr. 42: , Dietz, W. H., L. G. Bandini, and D. A. Schoeller. Estimates of metabolic rate in obese and nonobese adolescents. Pediatrics 118: , Du Bois, D., and E. F. Du Bois. A formula to estimate the approximate surface area if height and weight be known. Arch. Intern. Med. 17: , Garn, S. M., and D. C. Clark. Nutrition, growth, development, and maturation: findings from the Ten-State Nutrition Survey of Pediatrics 56: , Harris, J. A., and F. G. Benedict. A Biometric Study of Basal Metabolism in Man. Washington, DC: Carnegie Institute of Washington, 1919, p (Publ. no. 279) 10. Haycock, G. B., G. J. Schwartz, and D. H. Wisotsky. Geometric method for measuring body surface area: a height-weight formula validated in infants, children, and adults. Pediatrics 93: 62 66, Hayter, J. E., and C. J. K. Henry. A re-examination of basal metabolic rate predictive equations: the importance of geographic origin of subjects in sample selection. Eur. J. Clin. Nutr. 48: , Henry, C. J. K., and D. G. Rees. New predictive equations for the estimation of basal metabolic rate in tropical peoples. Eur. J. Clin. Nutr. 45: , Kaplan, A. S., B. S. Zemel, K. M. Neiswender, and V. A. Stallings. Resting energy expenditure in clinical pediatrics: measured versus prediction equations. Pediatrics 127: , Knowler, W. C., D. J. Pettitt, P. H. Bennett, and R. C. Williams. Diabetes mellitus in the Pima Indians: genetic and evolutionary considerations. Am. J. Phys. Anthropol. 62: , Maffeis, C., Y. Schutz, R. Micciolo, L. Zoccante, and L. Pinelli. Resting metabolic rate in six- to ten-year-old obese and nonobese children. Pediatrics 122: , Mason, E. D., and M. Jacob. Variations in basal metabolic rate responses to changes between tropical and temperate climates. Hum. Biol. 44: , Minghelli, G., Y. Schutz, A. Charbonnier, R. Whitehead, and E. Jequier. Twenty-four-hour energy expenditure and basal metabolic rate measured in a whole-body indirect calorimeter in Gambian men. Am. J. Clin. Nutr. 51: , Moon, J. K., F. A. Vohra, O. S. V. Jimenez, M. R. Puyau, and N. F. Butte. Closed-loop control of carbon dioxide concentration and pressure improves response of room respiration calorimeters. J. Nutr. 125: , Must, A., G. E. Dallal, and W. H. Dietz. Reference data for obesity: 85th and 95th percentiles of body mass index (wt/ht 2 ): a correction. Am. J. Clin. Nutr. 54: 773, Neel, J. V. Diabetes mellitus: a thrifty genotype rendered detrimental by progress? Am. J. Hum. Genet. 14: , Poehlman, E. T., C. L. Melby, and S. F. Badylak. Resting metabolic rate and postprandial thermogenesis in highly trained and untrained males. Am. J. Clin. Nutr. 47: , Robertson, J. D., and D. D. Reid. Standards for the basal metabolism of normal people in Britain. Lancet 1: , Schofield, W. N. Predicting basal metabolic rate, new standards and review of previous work. Hum. Nutr. Clin. Nutr. 39C: 5 41, Schutz, Y., T. Bessard, and E. Jequier. Diet-induced thermogenesis measured over a whole day in obese and nonobese women. Am. J. Clin. Nutr. 40: , Spurr, G. B., J. C. Reina, and R. G. Hoffmann. Basal metabolic rate of Colombian children 2 16 yr of age: ethnicity and nutritional status. Am. J. Clin. Nutr. 56: , Talbot, F. B. Basal metabolism standards for children. Am. J. Dis. Child. 55: , Tanner, J. M., and R. H. Whitehouse. Variations of growth and development at puberty. In: Atlas of Children s Growth, Normal Variation and Growth Disorders. New York: Academic, 1982, p Van Loan, M. D. Assessment of fat-free mass in teen-agers: use of TOBEC methodology. Am. J. Clin. Nutr. 52: , Weir,J.B.D.New methods for calculating metabolic rate with special reference to protein metabolism. J. Physiol. Lond. 109: 1 9, World Health Organization. Estimates of energy and protein requirements of adults and children. In: Energy and Protein Requirements. Geneva: World Health Organization, 1985, p