Ainsworth and associates (^2,3) published an adult compendium of physical activities that designates the intensity level of most common activities by units of metabolic equivalents (MET), using “multiples of one MET, or the ratio of the associated metabolic rate for the specific activity divided by the resting metabolic rate (RMR).” The assigned MET values for activity intensities are generalized estimates based on the literature and expert opinion (²; p. 72). The authors of the compendium define a MET as the energy expenditure of sitting quietly, equivalent to resting oxygen intake (3.5 mL·kg⁻¹·min⁻¹ or 1 kcal·kg⁻¹·h⁻¹). Compendium values are commonly used to convert subjects’ physical activity recalls into an estimation of calorie use. However, the compendium MET levels are applicable for adults, not for children (²). Currently, it is not clear to what degree or at what age the values of the Compendium can be applied to children.

Research indicates that resting energy expenditure (REE) is higher in children than in adults (^5,6,11) and somewhat higher in boys than in girls (⁵). The increased relative REE in children compared with adults is probably due to a variety of factors, including growth and puberty, and differences in body mass. In addition, energy cost may be greater for children due to the greater proportional amount of internal organs in children, their shorter legs, and smaller muscle mass. Puberty increases muscle mass, especially in boys, which could reduce the differences in EE between children and adults (^8,9,23,24). Bitar and associates (⁵) found that absolute EE (kJ·min⁻¹) was higher in pubertal than in prepubertal youth. Conversely, Roemmich and others (²⁴) found that REE relative to body mass was higher in prepubertal youth.

The energy cost of completing a task may also be greater for children than for adults, although research on the actual energy expenditure of children has been limited to very few activities. Early work by Robinson (²³) showed that the energy cost of walking for boys under 13 yr of age was approximately 5.5 mL·kg⁻¹·min⁻¹ more than for those over 13 yr of age and was related to their REE. Other studies compared the energy expenditure of running and walking in men and boys, finding the energy cost was greater for boys than for men (^8,9,22). However, small sample sizes and a general failure to include females reduce the generalizability of the studies.

Some studies have evaluated the age-related EE of exercise through childhood and adolescence to determine at what point the disparity in metabolic cost between children and adults disappears. Studies suggest that: 1) from 7 yr up to 12 yr of age, there is no difference in energy cost between boys and girls; 2) at about 12–14 yr of age girls may have attained adult responses; and 3) boys do not achieve adult responses until they near the end of adolescence (^8,9,26). However, these hypotheses have been directly tested only for walking, running, and cycling. We know very little about the age shift in energy expenditure (EE) for other physical activities that are common in youth.

Data are needed on the EE of children and adolescents to provide researchers and clinicians with a means of more accurately assessing EE during physical activity in youth and evaluating the effectiveness of interventions to increase children’s daily activity (²⁰). Such information is needed for studies of physical activity in youth with obesity, diabetes, cystic fibrosis, and asthma, as well as studies of primary prevention of obesity, diabetes, and CVD in youth.

The primary aim of the Energy Expenditure of Physical Activity in Youth (EEPAY) study was to determine the energy expenditure and age-adjusted metabolic equivalents (A-AME) of activities commonly performed by children and adolescents. Secondary aims were to determine at what age and pubertal developmental stage values approach those of adults.

METHODS

Ethical conduct.

Clearance for human subject study was obtained through a multiple assurance institutional review board (IRB). All study procedures were conducted according to ethical and legal parameters of the IRB, and participation was strictly voluntary. Subjects who were 18 yr of age provided written consent to participate, and parents of all other subjects gave written consent for their children to be in the study, whereas their children gave written assent. To assure safety, research assistants accompanied the subjects during each activity. Weight-training activities used have been shown to be safe with children as young as age 6.

Subjects and setting.

We studied 295 youth 8–18 yr of age, 53% males and 47% females, with at least 10 boys and 10 girls at each age (see Table 1). Because the earlier smaller studies of children indicated that energy expenditure differs for boys and girls, we tested a similar number of boys and girls at each age range, using the same activities for both. There were 223 (75.6%) white youth, 46 (15.6%) African Americans, and 26 (8.8%) other races. To be included in the study, the youth had to be essentially healthy and have no limitations in exercise participation. Subjects were recruited via advertisements placed in local newspapers, mass e-mails sent to departments on the UNC campus, flyers placed around the campus and other locations around Chapel Hill, as well as flyers sent home from local schools. The study was conducted primarily in the Applied Physiology Laboratory on the campus of the University of North Carolina at Chapel Hill.

Instruments.

Energy expenditure was measured with open-circuit, indirect calorimetry, using the Cosmed K4b² (Cosmed, Rome, Italy), a portable metabolic system that measures breath-by-breath ventilation (V̇_E), fraction of expired oxygen (F_EO₂), and carbon dioxide (F_ECO₂). Subjects breathe through a rubber face mask (Hans Rudolph Inc., Kansas City, MO) that directs air into the ventilation turbine, and into the portable unit housing the O₂ and CO₂ gas analyzers. This lightweight system (total mass ∼1.5 kg) attaches by harness to the subject’s torso, allowing the subject freedom of movement during the measurement periods. The system estimates EE by measuring breath-by-breath oxygen uptake (mL O₂·min⁻¹) and converts this to kilocalories per minute (or kcal·kg⁻¹·min⁻¹) using the Haldane correction for inspired min ventilation and standard equations (³⁰; pp. 455–463). The K4b² has been shown to accurately measure V̇O₂ and estimate EE over a wide range of metabolic rates in adults (¹⁷). To validate this system in children, we conducted a pilot test with 14 children 8–14 yr of age, comparing V̇O₂ obtained with the K4b² system with results using our standard laboratory metabolic system (Parvo Medics Truemax 2400, Salt Lake City, UT) during rest, walking (4.0 km·h⁻¹), and running (8.0 km·h⁻¹). The K4b² system showed a small positive bias of 5.1 mL·min⁻¹ at rest, 32.7 mL·min⁻¹ while walking, and 43.9 mL·min⁻¹ when running. However, the differences in V̇O₂ were less than 6% at each time point (P > 0.05) (Baggett, McMurray, Harrell, Bangdiwala, unpublished data).

Anthropometrics.

Height was measured to the nearest 0.1 cm with a stadiometer (Perspective Enterprises, Kalamazoo, MI), with the subjects shoeless. Body mass was measured to the nearest 0.1 kg using a calibrated balance beam metric scale (Detecto Scales, Inc., Brooklyn, NY), with the subjects clothed but shoeless. Body mass index (BMI) was calculated: body mass (kg) divided by height squared (m²).

Puberty.

Each subject’s stage of puberty was estimated by self-report using the Pubertal Development Scale (PDS) based on the five stages characterized by Marshall and Tanner (^15,16). The PDS has two five-item subscales, one for each gender, which consist of specific developmental characteristics. The scale is reliable, with Cronbach’s alphas ranging from 0.68 to 0.83 (median of 0.77) (²¹). Validity has been assessed in several ways. Moderate to high correlations have been found between the PDS and physician ratings of sexual maturity based on the Tanner scale (r = 0.61–0.67), between the PDS and adolescent self-reports based on Tanner scale pictures (r = 0.72–0.80), and between the PDS and age at peak height velocity and interviewer assessments of maturity (²¹).

Procedures.

The activities were determined by examining a variety of relevant studies (^7,19,25). Once potential subjects were identified, subjects were scheduled to come to the laboratory on three separate occasions to perform three sets of activities. Consent and assent were obtained in the first visit to the laboratory. During visit 1, height and body mass were measured and the youth were fitted with the K4b², and taken to a private room where they became accustomed to the instruments and filled out the questionnaires. The children then completed one of the three sets of activities. Each set began with sedentary activities, followed by activities of moderate intensity, and ended with high intensity activities. The order of the sets for each child was determined in computer-generated random order. See Table 2 for the activities in each set. Standard procedures were followed for each activity.

To measure REE, subjects were placed in a quiet darkened room in the supine position and allowed to rest for 5 min before values were recorded. Energy expenditure was then measured for 15 min with the subject quiet, but awake. The first 5 min as well as the last minute of measurement were eliminated and the REE was obtained from the average of 9 min. To measure EE during the activities, all activities were performed continuously for 10 min, unless the subjects could not continue, as was sometimes the case with rope skipping. To ensure that steady state was attained and maintained, data from the first 2 min and last minute of each activity were not used in the analyses. Thus, EE was averaged for 7 min of the activities. Steady state was verified for each subject by one of the investigators (RGM). (See the Appendix for activity descriptions.)

We followed stringent quality control protocols for the activities and for mechanical and environmental controls of relative humidity, mask fitting, sampling lines, etc. The metabolic system was calibrated before each use and was tested regularly for reliability using standard gases and ventilatory volumes. Research assistants (RA) were fully trained in the use of all equipment and were observed by the investigators until all techniques were mastered. The RA practiced instructing the children in the proper and standardized performance of each activity and the investigators observed data collection continuously for the first 15 subjects, and then at frequent intervals.

Data analysis.

Our first goal was to determine age or pubertal differences in resting EE. There were significant differences in V̇O₂ by age (P < 0 0.0001); thus, we performed Newman–Keuls multiple comparison tests. Using an alpha of 0.05, we determined that in boys V̇O₂ was not significantly different for those between the ages of 8–12 yr (group 1), nor for those aged 13–15 yr (group 2), nor those in the 16–18-yr-old (group 3). We performed the same analyses with girls, and the age groups were slightly different for girls: group 1 included girls aged 8–11 yr, group 2 was those aged 12–14 yr, and group 3 was aged 15–18 yr. This is consistent with studies showing that girls enter and complete pubertal development earlier than boys (^14,15). To determine an aged-adjusted metabolic equivalent (A-AME) we divided the measured V̇O₂ of the observed activity for each subject by that subject’s resting V̇O₂. We examined the resulting values by age and pubertal stage. Because the purpose of this study was to examine energy cost of activities in the context of the compendium of physical activities, we chose to examine the data with respect to oxygen uptake (mL·kg⁻¹·min⁻¹) or EE (kcal·kg⁻¹·h⁻¹), rather with than other scaling methods, such as allometric scaling.

RESULTS

Subject characteristics by gender and age group are shown in Table 3. As expected, height, body mass, and BMI increased as the youth aged. Pubertal status was widely distributed, with 21.0% at Tanner stage 1, 14.2% at stage 2, 20.0% at stage 3, 30.5% at stage 4, and 14.2% at stage 5.

Rest results.

Resting oxygen uptake (mL·kg⁻¹·min⁻¹) did not differ between girls and boys (P = 0.13); however, it declined in both as the ages of the groups increased. The actual values representative of a metabolic equivalent at rest are shown by age group in Table 4 and by Tanner stage in Table 5 using units of V̇O₂ (mL·kg⁻¹·min⁻¹) and EE (kcal·kg⁻¹·h⁻¹). The differences by age group and by Tanner stage were statistically significant (P < 0.0001). These data demonstrate that EE at rest, estimated as 1.0 in adults, is higher in children and young adolescents than in adults. The highest REE were found in the younger children and in those at an earlier stage of physical development (lower developmental stage).

Activity Results.

Tables 6–8 present the mean V̇O₂ and the conversion to age-adjusted metabolic equivalents (A-AME) by age group for each of the activities examined, arranged by sedentary (Table 6), low to moderate (Table 7), and moderate to high intensity levels of activity (Table 8). The compendium of physical activities MET levels for each of the activities are also shown in these tables, as well as the values resulting when the actual V̇O₂ measured during each activity is divided by 3.5, the value used for 1 MET in the compendium. For all activities, the values resulting when using 3.5 as the denominator were significantly different from the A-AME values, which used the measured V̇O₂ at rest (P < 0.0001), even after Bonferroni adjustments for the 17 comparisons.

For 14 of the activities, there were no significant differences in boys and girls in A-AME when adjusting for age group, based upon a Type I error of 0.001 (adjusted for multiple comparisons). However, there were some small differences in V̇O₂ by gender in leg press, rope skipping, shoveling, and stretching in age groups 2 and 3 (P < 0.002). Because these differences by gender were minor (<0.5 A-AME) and found in only four activities, we are reporting all results with boys and girls combined. This is consistent with the approach used by the compendium of physical activities (^2,3).

For sedentary and moderate activities, the mean A-AME values we report (see Tables 6 and 7) were slightly lower than the values presented in the compendium of physical activities, by 0.1–1.5 units, except for vacuuming and walking at 2.5 mph, where our values were higher by 0.1–0.5 units. For the higher intensity activities shown in Table 8, the A-AME were more variable. For rope skipping and running, our values were lower for the youngest age group, but similar for the other two age groups. The values we report for climbing stairs and walking at 3.5 mph are higher than the compendium values.

DISCUSSION

There are three main conclusions from this study. First, REE is greater in children than adults, so using the value of 3.5 mL·kg⁻¹·min⁻¹ to represent 1 MET for children, as used in the compendium of physical activities, can generally lead to an underestimate of energy expenditure in children. However, by the age of about 15 for girls and 16 for boys, the lower ages in our third group, the adult compendium values appear to be acceptable, keeping in mind that the 3.5 mL·kg⁻¹·min⁻¹ is an approximate average value (²). Our results are consistent with those of Roemmich and others (²⁴), who studied 31 girls and 29 boys, half prepubertal and half pubertal. They also found basal metabolic rate (BMR) was higher in children than adults until about 16 yr of age, and that using the compendium of physical activities MET value of 3.5 mL·kg⁻¹·min⁻¹ (1 kcal·kg⁻¹·h⁻¹) to assign EE to activities for youth resulted in an underestimation of energy expenditure, compared with doubly labeled water.

Our results are congruent with other studies that have shown that REE is higher in children than in adults. Early work of Boothby et al. (⁶) provided a standard for BMR in boys, showing a steady drop from age 6 to 18 yr. Several studies indicate the REE is much higher in children than in adults and somewhat lower in girls than in boys (⁵). Also, Goran et al. (¹¹) measured REE in young children and in their parents, and found that REE was somewhat higher in boys than in girls (1093 vs 1019 kcal·d⁻¹) and much higher in men than in women (2116 vs 1625 kcal·d⁻¹, respectively). However, in spite of these known differences in EE by gender in adults, the compendium of physical activities uses the same values for men and women, most likely for simplicity and ease of use.

Our second conclusion is that energy expenditure, both at rest and during activities, varies by pubertal stage, with results approximating adult values by Tanner stage 5 (see Table 5). Regarding puberty, our results agree with Bitar and associates (⁵), who found that absolute resting EE was higher in prepubertal than in postpubertal subjects, and are also similar to those of Roemmich and others (²⁴), who found that resting EE relative to body mass was higher in prepubertal youth. The decision to analyze EE by pubertal level or by age must be made by each investigator and will probably depend on the age range, gender, and ethnic background of the subjects. For example, if the subjects are in the age range during which puberty shows the greatest variation, that is, about 9–12 yr for girls and 10–14 yr for boys (^13,14), assessing pubertal level may be important. Also, if the sample contains both African Americans and white subjects, it might be wise to assess pubertal level, because African American youth, especially girls, mature earlier than white youth (¹⁸).

Our third conclusion is that the ratio of EE during activity to EE at rest is generally slightly lower for children compared with adults for most activities. Although slightly lower, this ratio of activity/rest is a better fit than the results produced using the standard definition of a MET (3.5 mL·kg⁻¹·min⁻¹ or 1 kcal·kg⁻¹·h⁻¹). Thus, using our estimates of resting EE and multiplying by compendium values for the activity is a closer estimate to the metabolic cost of the activity in children than using the standard definition of a MET. For example, if you are using a physical activity recall questionnaire that indicates a 9-yr-old who weighs 35 kg has accumulated 40 MET·h over the course of a day, with our age-adjusted resting metabolic equivalent (Table 4), this child’s total daily EE would be 35 kg × 1.71 kcal·kg⁻¹·min⁻¹ × 40 MET·h = 2394 kcal·d⁻¹. Using the adult value of 1 kcal·kg⁻¹·h⁻¹ from the compendium of physical activities to estimate total daily EE for this child would result in only 1400 kcal·d⁻¹, an underestimation of 42% (35 kg × 1 kcal·kg⁻¹·h⁻¹ × 40 MET·h = 1400 kcal·d⁻¹).

The history of the MET is not clear. The concept of a MET likely originated from the early work of Dill (¹⁰), who defined the ratio of work metabolism/rest metabolism, but the origin of the specific value of 3.5 mL·kg·⁻¹·min⁻¹ as 1 MET is vague. Nevertheless, this value is commonly accepted for adults and is reported in most texts (^3,12). However, the authors of the compendium of physical activities state that the most accurate way to determine EE is to assess EE at rest and use that REE to compute energy cost of activity (²; p. 72).

This study provides a way to adapt the adult compendium of physical activities MET intensity values of common activities using the age/pubertal-adjusted REE, to determine EE for children and adolescents. Although not ideal, this methodology improves the precision in assessing energy expenditure in children and youth. The method also provides a more solid foundation for researchers and clinicians alike, to better assess physical activity levels, develop and evaluate appropriate interventions for children, and predict health outcomes related to levels of physical activity and energy expenditure. Physical activities are often evaluated from subjects’ diaries or recalls of previous days (²⁹), weeks (²⁷), or months of activity (¹). The activities reported are then evaluated for frequency, duration, and intensity to determine the caloric usage and, ultimately, daily caloric need. Using the age- or pubertal-adjusted metabolic equivalent we present, investigators could compute the energy cost for common activities of children and adolescents utilizing the adult compendium of physical activities MET levels (^2,3).

A limitation of this study is that we presented data on a relatively small number of activities (N = 18, including rest). However, the activities we measured are commonly performed in childhood and adolescence but not generally reported in the literature. Each activity was carefully measured for its energy cost using strict protocols. One reason for the differences between the A-AME and the compendium values may be the actual values we chose for comparison. The compendium of physical activities includes over 600 activities, with many choices for some activities, and it was often difficult to make an exact match between activities. For example, the compendium lists stair climbing in differing codes, varying from walking downstairs (#17070, 3.0 METs), walking upstairs (#17130, 8.0 METs), to an array of walking upstairs while carrying a variety of differing weights (codes #17025–17030, 5–12 METs). We averaged the compendium METs for walking downstairs and upstairs (5.5 METs) for our calculations. The compendium lists rope skipping with two codes, varying by qualifiers of “slow” (#15551, 8 METs) or “moderate” (#15552, 10 METs). The children in this study skipped at what we considered closer to slow than moderate speeds, thus we used the lesser MET value in our calculations.

In addition, while not all the compendium activities were directly measured, many of the activities were assigned MET levels based on indirect calorimetry similar to the methods used in our study. Thus, although not an absolute match between age-adjusted metabolic equivalents and adult MET increments, it appears logical to use the MET levels from the compendium of physical activities and the age- (pubertal-) adjusted REE obtained in the present study to compute the EE of children and adolescents.

We recognize that a relatively small number of subjects were measured for each age and gender group (Table 1). If we had studied more subjects in each group we may or may not have detected gender differences. However, for the projected uses of the findings of this study, it was not deemed necessary to provide that level of detail, especially because the compendium of physical activities is applied similarly to men and women, even though it is known that EE differs by gender in adults (⁴). The results of our study cannot be generalized to children less than 8 yr of age.

Clinicians play an important role in health promotion and disease prevention, and promoting a physically active lifestyle is an essential element in primary prevention of many diseases. Physical activity must be assessed and can be prescribed in primary, secondary, and tertiary prevention. Exercise prescription will vary in type, frequency, duration, and intensity, but to appropriately target needs in any of these areas, energy expenditure must be considered (²⁸). Nutritional prescriptive decision-making parallels similar decision-making issues—precise energy expenditure in children is integral in diagnosing and monitoring eating disorders (including under and over-nourishment and related behaviors) (²⁰).

In conclusion, resting EE is higher in children than adults. Although not a perfect match, results from the activities measured suggest that the compendium of physical activities can be used to estimate EE with children and adolescents aged 8–18, if the user adjusts for the higher resting EE of children. The adjustment might be made either for age or for pubertal stage, depending on the focus of the study. Although this estimate may produce a value that is somewhat different from the actual EE, it is closer to the actual value than the presently available norms. After age 16, or once puberty is nearly completed, the compendium of physical activities MET increments may be used without adjustment. Further research is needed to determine the EE of children younger than 8 yr of age, both at rest and during a range of activities.