Measured versus predicted resting energy expenditure in infants: A need for reappraisal

Measured versus predicted resting energy expenditure in infants: A need for reappraisal M. A. Thomson, MRCP(UK), S. Bucoio, GDID, P. Quirk, BSc(Hons), a n d R. W. S h e p h e r d , MD, FRACP From the Department of Child Health, Children's Nutrition Research Center, Royal Children's Hospital, and the Queensland Institute of Medical Research, Brisbane, Australia The reliability of c o m m o n l y used predictive equations for estimating energy expenditure in infants in both health and disease was assessed by c o m p a r i n g resting energy expenditure (REE, measured by indirect calorimetry) in relation to weight, height, and b o d y cell mass (by total b o d y potassium analysis) with predictive equations (Harris-Benedict, Food and Agriculture Organization/ World Health Organization/United Nations University [FAO/WHO/UNU], Schofield weight-only, and Schofield weight-and-height equations) in 36 healthy infants ( a g e 0.43 ± 0.27 years; 19 male) and in 9 infants with cystic fibrosis ( a g e 0.41 ± 0.30 years; 4 male). Mean ± SD REE for healthy boys was 0.205 ± 0.019 MJ k g - l d a y -1 and for healthy girls 0.217 _ 0.026 MJ kg -1 day -I. Infants with cystic fibrosis had a significantly higher REE (0.258 ± 0.034 vs 0.210 ± 0.024 MJ kg - I d a y - l ; p <0.005). C o m p a r e d with measured values, predicted REEvalues varied markedly a m o n g equations, overestimating REE in healthy infants (Harris-Benedict equation, 182% _+ 63% [SD] of measured values; FAO/WHO/UNU equation, 104% ___14%; Schofield weight-only equation, 107.5% ± 14%; and Schofield weight-and-height equation, 106% ± 11%) and underestimating REE in those with cystic fibrosis (84% to 88% for the FAO/WHO/UNU, Schofield weight-only, and Schofield weight-and-height equations) e x c e p t the Harris-Benedict equation (152%). On regression analysis both weight and b o d y cell mass were related significantly to REE (r 2 = 0.87 and r 2 = 0.61, respectively) for normal infants and (r 2 = 0.92 and r 2 = 0.94) for those with cystic fibrosis. Using a generalized linear model of variance, we saw a significant (p <0.001) variability a m o n g all REE measures. Thus we could rely on none of the predictive equations to give an a c c u r a t e estimate of REE, and h e n c e energy and fluid requirements, in individual infants. We suggest that when a c c u r a t e estimates are n e e d e d , measurement of REE in individual infants should be attempted, especially in disease states, and that the continued use of current formulas should be reexamined. (J PEDIATR 1995;126:21-7)

For many years the determination of nutritional and fluid requirements in infants has been based on equations used to predict values of basal metabolic rate derived from studies

Submitted for publication March 3, 1994; accepted July 26, 1994. Reprint requests: R.W. Shepherd, MD, FRACP, Department of Child Health, Royal Children's Hospital, Brisbane, Queensland 4029, Australia. Copyright ® 1995 by Mosby-Year Book, Inc. 0022-3476/95/$3.00 + 0 9/20/59481

using indirect calorimetry. Forster 1 measured the gaseous metabolism of infants in 1877, and a number of metabolic studies using indirect calorimetry were performed in the first half of this century. 2-4 Harris and Benedict,5 in 1919, formulated their predictive equations based on weight, height, and age to the nearest whole year. The data of Talbot et al. 2 formed the basis for the 1973 Food and Agriculture Organization/World Health Organization/United Nations University recommendations for energy requirements in childhood6, 7 but their reliability was questioned by

21

22

Thomson et al.

BCM CF FAO/WHO/UNU

H-B REE TBK VQO2 ~/C02

Body cell mass Cystic fibrosis Food and Agriculture Organization/ World Health Organization/United Nations University [equation] Harris-Benedict [equation] Resting energy expenditure Total body potassium Volume of oxygen consumption Volume of carbon dioxide production

Schofield, 7 who presented new equations derived from pooled measurements of resting energy expenditure from a number of studies divided into gender and limited age categories (e.g., infants were in the birth to 3-year age group) and based on height and weight or weight alone. In addition, the data collected by Talbot 8 have provided the theoretical cornerstone for prediction of infant caloric and, hence, maintenance fluid requirements in infants developed by Holliday and Segar. 9 These formulas are widely quoted today in standard pediatric texts. 10These equations have also been used by some groups to evaluate changes in energy expenditure in various disease states,ll, t2 but their validity in disease states might be questioned partly because of the alterations in body composition, particularly of the body cell mass, which represents the oxygen-exchanging, work-performing body compartment. Studies in healthy individuals who have altered, body composition (such as obesity) have highlighted their unreliability in these situations. 13' 14 Review of the derivation of these equations calls into question the validity of using them in the infant age group at all, particularly in predicting an individual infant's energy expenditure, or if a disease state with altered body composition is present. All current equations have been either extrapolated down from older age groups, or have been developed by calorimetric techniques of limited accuracy for infants. We therefore measured REE using a dedicated infant indirect calorimeter in relation to weight, height, and BCM (by total body potassium). The BCM appears to be the most appropriate frame of reference against which energy expenditure should be studied. 15

METHODS Subjects. Thirty-six healthy infants (age 0.43 + 0.27 years; 19 male) and nine infants with cystic fibrosis homozygous for the 6F508 mutation (age 0.41 _+ 0.30 years; 4 male) were studied. Sixteen of the healthy infants received more than 90% of their daily caloric intake as human milk, and neither the infants nor any of the breastfeeding mothers were taking medications, or had intercurrent infections or pyrexia. Infants were volunteered for this noninvasive study by their parents, and written informed

The Journal of Pediatrics January 1995

f O b l e I. Equations used for estimation of basal energy expenditure in infants 1. Harris-Benedict (kcal/day)* Boys: REE = 66.473 + 5.003 X Height (cm) + 13.752 X Weight (kg) - 6.755 × age (yr) Girls: REE = 655.096 + 1.850 X Height (cm) + 9.563 X Weight (kg) - 4.676 X Age (yr) 2. Schofield,weight only (MJ/day) Children less 3 years of age Boys: REE = 0.249 X Weight (kg) - 0.127 Girls: REE = 0.244 X Weight (kg) - 0.130 3. Schofield,weight and height (MJ/day) Children less than 3 years of age Boys: REE = 0.0007 X Weight (kg) + 6.349 x Height (m) - 2.584 Girls: REE = 0.068 X Weight (kg) + 4.281 X Height (m) - 1.730 4. FAO/WHO/UNU (kcal/day)* Children less than 3 years of age Boys: REE = 60.9 x Weight (kg) - 54 Girls: REE = 61.0 X Weight (kg) - 51 *1 kcal = 4,186 kJ.

consent was obtained. The study was approved by the ethics committee of the Royal Children's Hospital, Brisbane.

Resting energy expenditure Predicted REE. Details of the equations used in prediction of the REE are set out in Table I. Measured REE. The REE was measured with a portable indirect calorimeter designed primarily for use with spontaneously breathing children and infants. The system is based on the open-circuit principle16 and uses a small canopy to collect expired respiratory gases. Gas flow is measured with a mass flow meter (Bronkhorst F-103D-HB; Anri Instruments, Melbourne, Australia), which has an accuracy of --+1% over the flow range of 0 to 100 L/min. Carbon dioxide is measured by a single-beam infrared analyzer (ADC, type RF; Analytical Development Co., Hoddeston, England), which has a response time of <4 seconds, a resolution of 0.001%, and a nonlinearity error of 0.05% for carbon dioxide concentrations up to 10%. Oxygen is measured by a differential paramagnetic analyzer (Datex OM11; Instrumentarium OY, Helsinki, Finland), which has a response time of 130 msec, resolution of 0.01%, and a nonlinearity error of <1% for oxygen concentrations up to 100%. The associated software was designed to enable calculations of the volumes of oxygen consumption and carbon dioxide production to be displayed continuously, to define steady state concentrations, and to enable later editing for noise from episodes of agitation and so forth. Validation of the system was based on the combustion of methanol to simulate infant values of Vo2, Vco2, and respiratory quotient.17 Thirty studies of 15 minutes were performed, with a flame and chimney substituted for the canopy used in patient studies. The amount of methanol con-

The Journal o f Pediatrics Volume 126, Number 1

T h o m s o n et al.

23

T a b l e II. Characteristics of infants

Healthy infants

Age (yr) Weight (kg) Height (cm) %Predicted weight/height %Predicted weight/age TBK* (gin) %Expected TBK*/age %Expected TBK*/weight

Total (n = 36)

Male (n = 19)

Female (n = 17)

CF infants (total) (n = 9)

0.43 _+ 0.27 6.75 + 2.07 64.8 + 7.6 97.5 _+ 10.6 102.0 _+ 10.9 10.1 +_ 3.4 98.8 _+ 17.2 98.9 _+ 16

0.43 +_ 0.29 7.24 + 2.40 65.5 _+ 8.2 99.5 -+ 8.9 107.6 _+ 10.3 10.9 -+ 3.9 99.4 _+ 18 98.1 _+ 14

0.43 _+ 0.26 6.20 _+ 1.44 63.9 _+ 6.7 95.9 _+ 7.8 95.9 _+ 11.2 9.1 _+ 2.6 97.9 _+ 16.4 99.2 -+ 17

0.41 + 0.30 5.75 _+ 2.25 61.5 + 9.1 96 + 9.5 91 _+ 11.3 9.1 -+ 3.8 89.3 _+ 10.4 94.1 _+ 7.8

Values expressed as mean ± SD. *TBK, total body potassium from 4°K analysis.

sumed was controlled by adding an additional flame inside the chimney; this allowed varying values of ~'o2 and ~'co2 to be simulated within a range expected of the study population (approximately 20 to 250 ml/min). This validation study showed that the system has an accuracy of 0.59% _+ 2.21% (mean + SD) for Vo2, 1.053% _+ 1.45% (mean _+ SD) forVeoz, and 1.01% + 1.93% (mean _+ SD) for respiratory quotient. Measurement protocol, Weight and height of each naked infant were obtained by a single observer before each measurement. The z scores for weight and height, and the weight-for-height and weight-for-age percentages were calculated with the Anthro software program, is and standard definitions of underweight and overweight were adopte d 19 with the use of Australian reference standards. 2° Two measurements of REE were obtained in the mid morning, 2 to 3 hours after a feeding, and separated by a 30-minute rest period, and the average was taken. The infants were quiet and awake during the 3-minute stabilization period, followed by a 20-minute period of sampling, which was performed while the infant was enclosed in a humidity- and temperature-controlled environment within a canopy system. Measurement started only after a steady state in Vo2 and ~rco2 was obtained as defined by variance of <5% for a 30-second period. Thereafter, ~ro2 and ~'co2 were calculated every 20 seconds, and any episodes of crying, agitation, coughing, and so forth could be edited out later. Body composition analysis. The BCM, which is the actively metabolising, the work performing, and in children the actively growing body compartment, was measured by TBK analysis.21-23Potassium is the major intracellular cation and is maintained within strict concentration limits within each cell; 98% of body potassium is located within the BCM 21 and contains a small, fixed proportion of the natural isotope 4°K. The TBK was measured with a bodycounter facility adapted for use in infants, as previously described.22, 23 Calculation of the percentage of expected TBK

for age, height, and weight was performed by comparing with pooled normal data as described previously.2224 Statistical methods, Analysis of variance was performed to determine whether there were any differences between measured REE and that predicted by the four different equations for each individual. This was performed for all healthy subjects (and within the following subsets: male; female; those from birth to 6 months of age; those from 6 to 12 months of age; those with ideal body weight; and those with less than ideal body weight) and subjects with CF. Multiple regression analysis was performed to determine the influence of patient age, weight, height, and TBK on measured REE. Analysis of covariance was performed in the healthy group to determine whether there were any differences between (1) breast-fed and formula-fed infants and (2) boys and girls in REE after adjustment for the covariates age, weight, length, and TBK. All analyses were performed with the SAS statistical package (SAS Institute Inc., Cary, N.C.). RESULTS The age, weight, height, weight/height percentage, BCM represented by the TBK in grams, and the TBK expected for age and weight are represented in Table II. The CF group had a lower-than-expected BCM for age but also a lower BCM per unit of weight than normal infants, indicating altered body composition. The measured REE expressed in terms of weight and TBK are represented in Figs. I and 2 for the healthy group (also divided into subsets of male, female, birth to 6 months of age, and 6 to 12 months of age) and for the CF group. The infants with CF had a significantlyhigher mean REE than the healthy group expressed both per unit of weight (0.258 _+ 0.036 vs 0.21 _+ 0.024 MJ/kg per 24 hours) and per unit of TBK (0.163 _+ 0.027 vs 0.142 + 0.027 M J/gin K per 24 hours). Infants' mean REEs for weight, for male versus female infants, were 0.217 _+ 0.027 versus 0.205 _+ 0.019 MJ/kg per 24 hours and for TBK were 0.15 + 0.03 versus

24

Thomson et al.

The Journal of Pediatrics January 1995

Fig. 1. Measured REE (expressed in terms of weight and TBK). It was significantly greater in infants with CF than in healthy infants (p <0.005). gK, Grams of potassium.

Fig. 2. Measured REE (expressed in terms of weight and TBK), analyzed for age, source of energy intake, and state of nutrition. No statistically significant differences emerged, gK, Grams of potassium.

0.135 + 0.021 M J / g m K per 24 hours; these differences were not statistically significant. Similarly, no significant differences were noted between mean R E E for weight when the groups less than versus more than 6 months of age, less

than versus more than the 90th percentile weight-forlength, or bottle versus breast fed were compared. Using a generalized linear model of variance, we saw a significant (p <0.001 ) variability in the variance of the five

The Journal of Pediatrics Volume 126, Number 1

Thomson et al.

25

Fig, 3. Predicted/measuredREEf•rthef•urequati•nsexamined.•nmu•tip•e-rangetestana1ysis•theH-Bequati•nwas significantly worse than the other equations (p 0.05), but all four overestimated REE in healthy infants and (except H-B) underestimated REE in infants with CF. When a generalized linear model of variance was used, all five equations were at significant variance (p <0.001) for all subjects, as well as for subsets in Fig. 2. (Age, breast vs bottle, weight/height percentile data not shown.) different values for REE (measured and four predicted), for all subjects combined as well as for the subsets analyzed (i.e., male/female; birth to 6 months/6 to 12 months of age; and those with ideal body weight/those below ideal body weight). Subsequent use of the studentized Newman-Keuls multiple range test showed a significant difference between the mean predicted energy expenditure from the HarrisBenedict equation and the other three predicted REEs, as well as the measured REE (p <0.05) in both the healthy and the CF groups, and in all subsets of healthy subjects (Fig. 3). There was a consistent overestimation of REE, by up to 16% in one subset (Schofield weight-only equation in the 6to 12-month age group) in healthy infants, and a consistent underestimation in infants with CF (excepting the H-B equation). The 95% confidence limits of up to 12% for predicted/measured REE with the Schofield weight-only, Schofield weight-and-height, and F A O / W H O / U N U equations (and up to 48% if one uses the H-B equation) were unacceptably large and, if added to the aforementioned overestimation, could result in a variation of up to 23% of measured REE. For example, the measured REE for a 7 kg male infant in this study was 1.43 M J/day; by F A O / W H O / U N U equation the REE would be 1.56 M J/day, but the 95% confidence limits would be 1.45 to 1.67 M J/day. In the healthy infants <90% ideal body weight, the 95% confidence limits were wider: 1.37 to 1.75 M J/day. The infants with CF who have altered body composition had wider prediction confidence limits (for a 7 kg infant with CF, REE

measured 1.81 [1.63 to 1.99] M J/day, and the predicted value was 1.56 M J / d a y [1.36 to 1.76]). In the healthy group, by linear regression, subject weight and TBK were significantly related to REE (r 2 = 0.87, p <0.0001 and r 2 __ 0.61, p <0.0001, respectively); however, subject age and sex were not (p >0.4 and p >0.7, respectively). In the CF group, similar relationships were found with REE for weight (r 2 = 0.92, p <0.0001) and for TBK (r 2 __ 0.94, p <0.0001). When the relation of TBK to REE was compared between the healthy group and the CF group by analysis of covariance, the difference was not significant (p = 0.08). DISCUSSION Estimation of nutritional and fluid requirements based on determination of energy expenditure by the use of predictive formulas is, on the basis of this study, prone to inaccuracy in individual infants, consistently overestimating energy and fluid requirements in healthy infants, and sometimes significantly underestimating them in disease states. Schofield7 emphasised the wide coefficient of variation when these formulae are used to predict REE in healthy subjects, as opposed to group estimates (n > 100), for which their accuracy was not disputed. The 95% confidence limits of up to 12% for predicted versus measured REE (and up to 48% if one uses the H-B equation) are unacceptably large and, if added to the aforementioned overestimation, could result in a variation of up to 23% of measured REE. Although by multiple range test the calculated REEs (oth-

26

Thomson et al.

er than by H-B) were not statistically different from the measured REE, these variations nevertheless illustrate the magnitude of error that may occur when one employs equations to predict values for specific individuals. This point was underlined by Schofield7 when he pointed out that in the group <3 years of age a coefficient of variation for prediction of REE for an individual was 19.4%, for a group of ten 6.3%, for one hundred 2.5%, and for one thousand 1.6%. This demonstrates the danger of using pro rata formulas on which to base estimates of energy expenditure for individual cases, especially in this age group. In our relatively small population group, no differences between male and female infants or between those being breast fed or formula fed were seen. However, in this study we have highlighted not only the inaccuracy of predictive equations in individuals but also the dubious reliability of these equations in the infant population, whether healthy or sick. The H-B equation 5 is grossly inaccurate in this age group and, despite its continued use, is obviously inappropriate for infants: the intercept (655.096 kcal, or 2.74 M J) is too great for a female infant. This is also obvious from the fact that the H-B predicted mean value for girls is 1.69 that for males. Nevertheless, when the newer equations are used, there remains a consistent overestimation of REE in healthy infants. The F A O / W H O / U N U equation particularly overestimates REE in males and in infants 6 to 12 months of age. Recent work from the Dunn Nutrition Unit on total energy expenditure in infants and young children suggests that a decline in energy intake in recent years may be one reason for the apparent gradual decrease in TEE through the years 25,26 and this may be a contributing factor in the overestimation of REE in our study, given that the predictive equations are largely based on infant studies from the first half of this century. In contrast, however, in the group with CF the predictive equations markedly underestimated the actual REE, indicating that CF infants have a higher energy requirement than realized. 27 Schofield7 noted that many of the studies on which these predictive equations are based were carried out with a single measurement only, which may also bias the data. 28 We took the average of two measurements under strictly controlled conditions and were able to edit out "noise," a practical problem when studying infants. These data also highlight the fact that body composition is important in the context of energy expenditure. Differences in body composition occur in obese but healthy children, in whom these equations have a large predictive error.13, 14 Previous studies of infants with CF had mean REE values for weight which were significantly different from those predicted by these equations 29, 3o but body composition was not measured. Because all the equations cited are based on standard anthropometric variables, they fail to take into account possible alterations in body composition.

The Journal of Pediatrics January 1995

Body cell mass is an important indicator of nutritional status, comprising the actively metabolizing, the work-performing, and, in children, the growing-body compartment. On multiple regression analysis of the data in our study, both weight and BCM were predictors of energy expenditure in both healthy infants and those with CF, but age was not. In healthy infants, body composition was comparatively uniform (coefficient of variation of TBK/weight was only 19%, compared with 33% for weight). We suggest that it is important to compare energy expenditure and intake data on a unit-weight and, if possible, on a BCM basis. The BCM should be the appropriate term of reference in an examination of metabolic processes, especially in states of abnormal body composition in which the relation of BCM to weight is altered. In addition to predictions for nutritional requirement, formulas for prediction of fluid requirements are based on REE prediction. The work of Holliday and Segar, 9 which forms the basis of many formulas for the calculation of energy requirements in infancy and childhood quoted in modern textbooks, l° is founded on the dubious assumption of the energy expenditure of what was termed "an average hospital patient." Their assumption that 100 to 115 ml of water is lost as insensible loss by the skin, lungs, and kidneys for each 100 calories expended led to formulas that constitute the cornerstone of our present-day fluid management in infants. However, their assumptions were based on REE data from as far back as 1914, the validity of which is highly questionable (measurements on ill children tested in the outpatient department of the Massachusetts General Hospital, Boston, were included3°-32). The inapplicability of such formulas to the energy and fluid requirements of the present-day infant cannot be too strongly emphasized, especially in disease states in which fluid and energy requirements are known to differ markedly from those in well infants, and are also prone to wide and rapid fluctuations. We advocate the wider use of actual measurement of energy expenditure in the infant population, especially in disease states, and we suggest caution in relying solely on equations for estimating energy expenditure and fluid requirements in individual infants. REFERENCES

1. Forster E, Amtl Bet d50. Versammlung deutsch Naturfarscher u Aezte in Miinchen, Munich, 1877. 2. Talbot F, Wilson E, Worcester J. Basal metabolism of girls: physiological background and application of standards. Am J Dis Child 1937;53:274-7. 3. Boothby W, Berkson J, Halbert D. Studies of the energy of metabolism of normal individuals: a standard for basal metabolism with nomogram for clinical application. Am J Physiol 1936;116:468-83. 4. Robertson J, Reid D. Standard for the basal metabolism of normal people in Britain. Lancet 1952;1:940-3. 5. Harris J, Benedict F. A biometric study of basal metabolism

The Journal of Pediatrics Volume 126, Number 1

6.

7.

8. 9. 10.

11.

12.

13.

14.

15.

16.

17.

18.

in man [publication No. 279]. Washington, D.C.: Carnegie Institute of Washington, 1919. World Health Organisation. Energy and protein requirements: report of a joint F A O / W H O / U N U expert consultation [WHO Technical Report Series No. 724]. Geneva: World Health Organization, 1985. Schofield C. Predicting basal metabolic rate, new standards and review of previous work. Hum Nutr Clin Nutr 1985; 39c(suppl 1):5-42. Talbot FB. Basal metabolism in children. In: Brennemann's practice of paediatrics. Hagerstown, Maryland: Prior, 1949. Holliday MA, Segar WE. The maintenance need for water in parenteral fluid therapy. Paediatrics 1957;19:823-32. Robson AM. Parenteral fluid therapy. In: Behrman RE, Vaughan VC, eds. Nelson's textbook of paediatrics. 13th ed. 191-4. Tilden S, Watkins S, Tong T, Jeevanandam M. Measured energy expenditure in pediatric intensive care patients. Am J Dis Child 1989;143:490-2. Vaisman N, Pencharz P, Corey M, Canny G, Hahn E. Energy expenditure of patients with cystic fibrosis. J PEDIATR 1987; 111:496-500. Maffeis C, Schutz Y, Micciolo R, Zoccante L, Pinelli L. Resting metabolic rate in six- to ten-year-old obese and nonobese children. J PEDIATR 1993;122:556-62. Dietz WH, Bandini LG, Schoeller DA. Estimates of metabolic rate in obese and nonobese adolescents. J PEDIATR 1991; 18:146-9. Kinney JM, Lister J, Moore FD. Relationship of energy expenditure to total exchangeable potassium. Ann NY Acad Sci 1963;110:711-22. Branson R. The measurement of energy expenditure: instrumentation, practical considerations and clinical applications. Respir Care 1990;35:640-59. Marks KH, Coen P, Kerrigan JR, Francalancia NA, Nardis EE, Snider MT. The accuracy and precision of an open circuit system to measure oxygen consumption and carbon dioxide production in neonates. Pediatr Res 1986;21:58-65. Anthro software program; version 1.01. Geneva: Division of Nutrition, Center for Disease Control, World Health Organization.

Thomson et al.

27

19. Dixon N J, Massey FJ. Introduction to statistical analysis. 2nd ed. New York: McGraw-Hill, 1967:61. 20. National Health and Medical Research Council. Heights and masses of Australian infants and children. Sydney: Australian Government Publishing, 1973. 21. Coward WA, Parkinson SA, Murgatroyd PR. Body composition measurements for nutrition research. Nutrition Research Reviews 1988;1:115-24. 22. Shepherd RW, Holt TL, Greer R, Cleghorn G J, Thomas BJ. Total body potassium in cystic fibrosis. J Pediatr Gastroenterol Nutr 1989;9:200-5. 23. Shepherd RW, Oxborough D, Holt TL, Thomas BJ, Thong YH. Longitudinal study of the body composition of weight gain in exclusively breast-fed and intake-measured whey-based formula-fed infants to age 3 months. J Pediatr Gastroenterol Nutr 1988;7:732-9. 24. Pierson RN, Lind DH, Phillips RA. Total body potassium in health: effects of age, sex, height and fat. Am J Physiol 1974;226:206-12. 25. Prentice AM, Lucas A, Vasquez-Velasquez L, Davies PS, Whitehead RG. Are current dietary guidelines for young children a prescription for overfeeding? Lancet 1988;2:1066-9. 26. Davies PS, Ewing G, Lucas A. Energy expenditure in early infancy. Br J Nutr 1989;62:621-9. 27. Shepherd RW, Holt TL, Vasques-Velasquez L, Coward WA, Prentice A, Lucas A. Increased energy expenditure in young children with cystic fibrosis. Lancet 1988;1:1300-3. 28. Robertson JD. The determination of basal metabolism from outpatients. BMJ 1944;1:617-27. 29. O'Rawe A, McIntosh I, Dodge JA, et al. Increased energy expenditure in cystic fibrosis is associated with specific mutations. Clin Sci (Colch) 1992;82:71-6. 30. Fried MD, Durie PR, Tsui LC, Corey M, Levison H, Pencharz PB. The cystic fibrosis gene and resting energy expenditure. J PEDIATR 1991; 119:913-6. 31. Talbot FB. Basal metabolism standards for children. Am J Dis Child 1938;55:455-9. 32. Benedict FG, Talbot FB. The gaseous metabolism of infants with special reference to its relation to pulse rate and muscular activity. Washington, D.C.: Carnegie Institute, 1914:1-168. (Publication No. 201.)