Contents - Previous - Next

Estimation and validation of energy expenditure obtained by the minute-by-minute measurement of heart-rate

1. Heart-rate method

1.1. Subject calibration
1.2. Calculation of energy expenditure from fH

2. Validation of heart-rate method

2.1. Sources of error

3. Applications of the heart-rate method

3.1. Daily pattern of EE and TDEE
3.2. Pattern of relative effort
3.3. Comparison of EE pattern of individuals



* Department of Physiology, Medical College of Wisconsin and Research Service, Zablocki VA Medical Center, Milwaukee, WI, U.S.A., and Departments of Physiological Sciences and Pediatrics, Universidad del Valle, Cali, Colombia.


Validation studies have shown that the appropriate use of the minute-by-minute heart-rate method for estimating energy expenditure provides a suitable alternative when high accuracy in individual subjects is not required. It can be used under field conditions and gives excellent measurements of average energy expenditure in small groups of subjects. It is also useful for obtaining the activity pattern of individuals or groups.

Interest in the measurement of energy expenditure (EE) in free-living subjects is evident from the articles presented in this volume. There is widespread attraction to the application of the doubly-labelled water (2H218O) method because of its accuracy in individual subjects (WESTERTERP et al., 1988). However, it has two drawbacks that limit its use in many settings and in some kinds of research project. The expectations that its cost (~300 US$/subject; SCHOELLER, 1988) would be reduced have not been realized. In addition, the highly technical nature of mass-spectrometric analysis requires equipment and training which are formidable. The second drawback is that it gives an average of the energy expenditure over the period of measurement, i.e., it cannot provide the pattern of energy expenditure.

The purpose of this article is to describe an alternative to the 2H218O method which may prove useful in many research situations.

1. Heart-rate method

1.1. Subject calibration
1.2. Calculation of energy expenditure from fH

The idea of using the well-known relationship (BOOYENS and HERVEY, 1960) between heart-rate (fH) and oxygen consumption (VO2) to estimate energy expenditure in free-ranging subjects has a long and checkered history. Although early discussions about the use of fH in energy expenditure measurements mentioned telemetry to nearby observers (BRADFIELD, 1971; BERG, 1971), this technique apparently had little advantage over existing factorial methods and was not much used except for obtaining ancillary data to measurements of VO2 (SPURR et al., 1975). The first fH monitors which allowed subjects freedom from observation accumulated fH over some period of time and gave an average fH when divided by the time of accumulation (HEYWOOD and LATHAM, 1971; GANDRA and BRADFIELD, 1971).

At first, the measurement of 24-hour EE was attempted by accumulation of the fH for this period. Despite reports of the suitability of using mean fH values for 24 hours in estimating total daily energy expenditure (TDEE) (ACHESON et al., 1980; WARNOLD and LENNER, 1977), there was justifiable dissatisfaction with the use of a single averaged fH (DAUNCEY and JAMES, 1979; CHRISTENSEN et al., 1983) because in sedentary subjects mean 24-hour fH is only a few beats above resting values where the VO2:fH relationship is notoriously unreliable (BOOYENS and HERVEY, 1960). In children, when the fH accumulation occured only during times of the day when they were awake and active, with basal metabolic rate (BMR) and resting metabolic rate (RMR) supplying EE for the remainder of the day, better estimates were obtained (SPADY, 1980; SPURR, REINA and BARAC-NIETO, 1986) because the average fH while awake is on the active portion of the calibration curve where reliability is much higher (SPURR et al., 1986).

The development of small instruments which measure and store fH minute-by-minute for long periods of time has considerably improved the applicability of this technique to the measurement of TDEE. These may be worn on the waist (BAHARESTANI et al., 1979) or wrist (Vantage Performance Monitor, Polar Electro, Inc., Hartland, WI, USA) and offer no interference with the subjects' ability to work or play. This approach offers the advantages of relatively low cost and the ability to measure activity patterns of subjects away from observers.

1.1. Subject calibration

The importance of individual calibration of a subject to obtain his/her VO2:fH relationship has long been recognized (BRADFIELD, HUNTZICKER and FRUEHAN, 1970) and is a cardinal feature of most studies using fH to obtain EE. Some suggest that the VO2 and fH should be measured while performing activities which will be carried out during the independent collection of fH data (TREMBLAY and BOUCHARD, 1987) and it may be that in individuals who must sustain relatively high levels of activity for long periods of time, measurements of the energy cost of the activity would improve accuracy. However, validation studies with whole-body calorimetry and doubly-labelled water have shown excellent agreement when the calibration procedure was carried out with bicycle ergometry alone (SPURR et al., 1988) or a combination of bicycle ergometry and step-testing (LIVINGSTONE et al., in press) and in-place jogging, even when activity in the calorimeter included the arm exercise of rowing (CEESEY et al., 1989). Our own studies in Colombian children have been done with calibration of subjects using a treadmill protocol (see below). The procedure involves obtaining simultaneous VO2 and fH measurements while at rest lying down, sitting and standing quietly to obtain RMR calculated as the average of all resting values, and while performing exercise using one of the mentioned modalities. A typical calibration curve for a young male subject while walking on a treadmill and during rest is shown in Figure 1. The FHFLEX can be calculated as the mean of the highest resting and the lowest exercise fH or obtained by visual estimation from the plotted data. If a particular fH is below FHFLEX then RMR is used for the EE, if above the least squares regression line of the exercise curve is used to calculate VO2.

Figure 1. Calibration to demonstrate how the VO2:fH relationship of a young subject is used to estimate VO2 from fH at values of fH above FHFLEX. Below FHFLEX, VO2 at rest (RMR) is used. From SPURR, in press.

1.2. Calculation of energy expenditure from fH

With the fH monitors currently in use, the minute-by-minute fH values accumulated during the period of measurement are transferred into a desk top computer and stored on floppy disk for subsequent calculations. Figure 2 is a plot of the individual data points for the subject in Figure 1 during 14 hours of fH monitoring from 6 a.m. to 8 p.m. The upper curve is of fH and shows the FHFLEX value for the 14-hour period as a horizontal line during this period. When fH is below FHFLEX, the RMR value is used and so the plot of VO2 and EE show this value as a minimum.

Figure 2. Plot of minute-by-minute heart-rate during 14 hours for the subject in Figure 1 showing the conversion of fH to VO2 and to the rate of energy expenditure. Below FHFLEX, the resting metabolic rate is used and so appears as a minimum value. From SPURR, in press.

The calculations applied to each heart rate to obtain EE in kJ/min are:

EE = (m * fH + b) * 20.48 at fH > FHFLEX


EE = RMR at fH (FHFLEX (1)

where m = the slope and b = the intercept of the activity calibration curve (Figure 1) and 20.48 kJ/L is the caloric equivalent of O2 at an assumed respiratory quotient (RQ) of 0.88. RMR is the mean of the various measurements of resting metabolism.

The energy expenditure while on the fH monitor (EEM) is then:

EEM = (EE (2)

In field studies there may be a short period (extra time; ET) after the end of fH recording and before bed that is not accounted for by monitor time (MT) and sleep time (ST). This can be assigned RMR values also if activity levels are known to be low such that the total daily energy expenditure (TDEE) is:

TDEE = EEM + (RMR * ET) + (BMR ST) (3)

where BMR and RMR are in kJ/min. The use of BMR for sleep EE has been shown to overestimate sleep EE by about 5% during the actual hours of sleep, but when applied to the TDEE the error is negligible (GOLDBERG et al., 1988). Also, if one is sure that the Schofield equations (SCHOFIELD, 1985) for estimating BMR apply to the subjects under study, they may be used to estimate sleep EE (SPURR et al., 1988). BMR should be measured if there is doubt.

The maintenance energy expenditure (MEE) is defined as the minimum EE and can be written as:

MEE = (ST * BMR) + [(1440 - ST) * RMR] (4)

and the EE in activity (EAC) is:

EAC = TDEE - MEE (5)

Once minute-by-minute values are obtained, the data can be handled in a number of ways. Plots of individual points may be useful (Figure 2) or the data can be averaged by various time intervals to establish patterns such as those seen in Figure 3. Obviously, the amount of detail desired in establishing the pattern will dictate the time interval of the averaging (Figures 2 and 3).

Figure 3. Examples of various time intervals for averaging the energy expenditures plotted in Figure 2. From SPURR, in press.

2. Validation of heart-rate method

2.1. Sources of error

There have been several validation studies of the minute-by-minute heart-rate method. Two have compared it against whole-body indirect calorimetry (SPURR et al., 1988; CEESAY et al., 1989) and one against the 2H218O method in free-living subjects (LIVINGSTONE et al., in press). The first study (SPURR et al., 1988) employed the bicycle ergometer for calibration, and for the exercise protocols followed while in the calorimeter. The second (CEESAY et al., 1989) used the bicycle ergometer, stepping and in-place jogging for calibration and these activities plus rowing for the exercise protocols while in the calorimeter. The regression and correlation analyses of these three studies are shown in Figure 4. Despite a wide variety of activities and ranges of TDEE, the data all fall on essentially the same regression line.

Figure 4. Regression and correlation of three validation studies by (A) SPURR et al. (1988), (B) CEESAY et al. (1989), (C) LIVINGSTONE et al. (in press), and of the combination of all three (D). From SPURR, in press.

Correlation and regression analyses are not always the most suitable for comparing a new with a standard method and so the statistical approach described by BLAND and ALTMAN (1986) was applied to the 56 data values from the three studies and is presented in Figure 5. The left panel shows the mean of the differences between the heart-rate method and the mean of it and the values obtained from the reference method together with the 95% confidence intervals for individuals (±2 SD) and groups (±2 SE). The right panel of Figure 5 is a regression analysis of the differences over the range of measurement and shows a lack of any dependence on the absolute TDEE.

Figure 5. Analysis of differences between the heart-rate method and each of the three validation studies shown in Figure 4 with the 95% confidence limits for individuals (±2 SD) and groups (±2 SE) according to the method described by BLAND and ALTMAN (1986). Right panel shows that there was no significant effect on the differences over the entire range of energy expenditures. From SPURR, in press.

The data in Figure 5 make it clear that the method is not suitable for application to a single measurement of TDEE in individuals. The widest discrepancies were noted in the comparison with doubly-labelled water (-22% to +52%; LIVINGSTONE et al., in press). However, the group averages in the three studies shown in Figure 4 were not significantly different from the reference methods. This was also true in even small groups, i.e. N = 4 to 6 as shown in Figure 6 (SPURR et al., 1988). The subjects were grouped by sex, according to four different exercise protocols followed in the calorimeter and when all were combined.

In none of the statistical comparisons were there significant differences between measurements in the calorimeter and those made by the heart-rate method (Figure 6). The average values for TDEE obtained by the heart-rate method varied from -2.7 to +6.8% of those measured by the calorimeter. Others have reached similar conclusions about the minute-by-minute heart-rate method (KALKWARF et al., 1989) at different average levels of TDEE. Furthermore, even the estimates of TDEE in individuals may be greatly improved by repeated measurements over several days. This type of validation has not yet been tried.

Figure 6. Means ±SE of daily energy expenditure measured by whole-body indirect calorimetry and the heart-rate method. The four exercise protocols followed by the subjects while in the calorimeter were no exercise (I), and two (II), four (III) and six (IV) 30-minute bouts of exercise at varying levels of intensity (SPURR et al., 1988). There were no statistically significant differences between the two methods of measurement in any of the groupings.

2.1. Sources of error

The heart-rate method depends primarily on the accuracy of the measurements of BMR, RMR and the VO2:fH relationship in the calibration procedure as well as on the estimation FHFLEX. The possible contributions of these to the measurement of TDEE have been discussed in detail in the three validation studies presented in Figure 4 and will not be repeated here. The measurement of TDEE by the heart-rate method depends on the three compartment systems of BMR, RMR and the VO2:fH calibration (Figure 2 and Equation 4). The first two compartments are measured by indirect calorimetry and only the last is subject to variability in the VO2:fH relationship.

The errors expected from emotional or other transitory increases in fH, not due to increased VO2, are probably not large since it is unlikely that they will be sustained for long periods of time. Averaging will usually diminish the contribution of such errors in the analysis of activity patterns (Figure 3). Improvements in the application of the method will no doubt be forthcoming. One possibility is the simultaneous measurement of heart-rate and motion, using the latter to distinguish high heart-rates due to causes other than increased activity. One such approach has been described by TAYLOR et al. (1982).

It is possible to imagine misuses of the heart-rate method; for example, calibration of the subject at sea level and fH measurement at altitude would be inappropriate. Common sense should be used. When applied appropriately, minute-by-minute heart-rate recording will give excellent estimates of TDEE in even small groups.

3. Applications of the heart-rate method

3.1. Daily pattern of EE and TDEE
3.2. Pattern of relative effort
3.3. Comparison of EE pattern of individuals

We have carried out a number of studies on the energy expenditure of school-aged, nutritionally normal and marginally malnourished children in our laboratory in Cali, Colombia, using minute-by-minute heart-rate recording (SPURR and REINA, 1988a, b, 1989 and in press). Some of these studies are discussed by Torun in this volume.

3.1. Daily pattern of EE and TDEE

The minute-by-minute heart-rate method was utilized to follow the EE while children were awake (SPURR and REINA, 1988a) from about 6 a.m. to 8-9 p.m. Since the schoolday in Cali operates on two 5-hour shifts (7 a.m. to 12 p.m. and 1 to 6 p.m.), the data were analyzed by 'school-time' and 'free-time' without regard to whether they were obtained in the morning or afternoon. In Figure 7 they are presented in 30-minute averages, during 5 hours in each category, together with the two and three-way analysis of variance (ANOVA; with repeated measures where appropriate).

Figure 7. Thirty-minute averages of energy expenditure in 6-8, 1012 and 14-16-year-old normal and marginally undernourished Colombian boys and girls during 5 hours of 'school-time' and 'free-time'. Where appropriate, two-way ANOVA is with repeated measures. From SPURR and REINA, 1988a.

There are statistically significant differences in nutritional group (NG) in the youngest children and during the time course of the 5-hour period in school only in the girls. In the repeated measures ANOVA over time (hours), the only significant change occurred during the fall in EE in 14-16-year-old girls during school, and all age groups of girls had significantly lower values than boys.

These data are presented primarily to show the results of a study using minute-by-minute heart-rate measurement. Another kind of application showing the failure of malnourished boys to keep up with their well-nourished counterparts is presented in the article by Torun in this volume (his Figure 6 on p. 348).

3.2. Pattern of relative effort

The maximum oxygen consumption (VO2 max) is a measure of the physical condition of an individual. When the submaximal VO2 is expressed as a percentage of VO2 max (% VO2 max) it is a measure of relative effort (SPURR and REINA, in press). When the data shown in Figure 7 are obtained in individuals in whom VO2 max is also measured, one can express the pattern of relative effort exerted by the subjects (SPURR and REINA, in press).

3.3. Comparison of EE pattern of individuals

We are currently using this method in a study of adult women in Cali, Colombia. Figure 8 presents the average 30-minute energy expenditures of two female subjects during 24 hours. Both women worked in a preschool nursery. One was the director of the nursery and the other her secretary. The secretary spent the major portion of the day at her desk typing or carrying out other office tasks. During the morning, the director also worked in her office. After lunch she spent the afternoon working directly with the children and other teachers in supervised play. In the late afternoon, she bicycled to the store for some shopping, returning to her home about 7 p.m. The heart-rate recorders were detached from both women at 7:30 p.m. The similarities in the levels of their activities during the morning and dissimilarities in the afternoon are evident from the figure. The BMR was used as the measure of EE during sleep. With such data, it is a simple matter of summation to obtain TDEE.

Figure 8. Comparison of the pattern of energy expenditures of two women during the same 24-hour period (see text).


The work described from our laboratory was supported by the United Nations University, the Nestle Nutrition Research Grant Programme, and the Fundacion para la Educacion Superior, Cali, Colombia.


ACHESON, K.J., CAMPBELL, I.T., EDHOLM, O.G., MILLER, D.S., STOCK, M.J.: The measurement of daily energy expenditure: an evaluation of some techniques. Am. J. Clin. Nutr., 33, 1155-1164 (1980).

BAHARESTANI, H., TOMKINS, W.J., WEBSTER, J.G., MAZESS, R.B.: Heart rate recorder. Med. Biol. Eng. Comput., 79, 719-723 (1979).

BERG, K.: Heart-rate telemetry for evaluation of the energy expenditure of children with cerebral palsy. Am. J. Clin. Nutr., 24, 1438-1445 (1971).

BLAND, J.M., ALTMAN, D.G.: Statistical methods for assessing agreement between two methods of clinical measurement. Lancet, i, 307-310 (1986).

BOOYENS, J., HERVEY, C.R.: The pulse rate as a means of measuring metabolic rate in man. Can. J. Biochem. Physiol., 38, 1301-1309 (1960).

BRADFIELD, R.B.: A technique for determination of usual daily energy expenditure in the field. Am. J. Clin. Nutr., 24, 1148-1154 (1971).

BRADFIELD, R.B., HUNTZICKER, P.B., FRUEHAN, G.J.: Errors of group regressions for prediction of individual energy expenditure. Am. J. Clin. Nutr., 23, 1015-1016 (1970).

CEESAY, S.M., PRENTICE, A.M., DAY K.C., MURGATROYD, P.R., GOLDBERG G.R., SCOTT, W. SPURR, G.B.: The use of heart rate monitoring in the estimation of energy expenditure: a validation study using indirect whole-body calorimetry. Br. J. Nutr., 61, 175-186 (1989).

CHRISTENSEN, C.G., FREY, H.M.:, FOENSTELIEN, E., AADLAND, E., REFSUM, H.E.: A critical evaluation of energy expenditure estimates based on individual O2 consumption/heart rate curves and average daily heart rate. Am. J. Clin. Nutr., 37, 468-472 (1983).

DAUNCEY, M.J., JAMES, W.P.T.: Assessment of the heart-rate method for determining energy expenditure in man using a whole body calorimeter. Br. J. Nutr., 42,1 -13 (1979)

GANDRA, Y.R., BRADFIELD, R.B.: Energy expenditure and oxygen handling efficiency of anemic school children. Am. J. Clin. Nutr., 24, 1451-1456 (1971).

GOLDBERG, G.R., PRENTICE, A.M., DAVIES, H.L., MURGATROYD, P.R.: Overnight and basal metabolic rates in men and women. Eur. J. Clin. Nutr., 42, 137-144 (1988).

HEYWOOD, P.F., LATHAM, M.C.: Use of the SAMI heart-rate integrator in malnourished children. Am. J. Clin. Nutr., 24, 1446-1450 (1971).

KALKWARF, H.J., HAAS, J.D., BELKO, A.Z., ROACH, R.C., ROE, D.A.: Accuracy of heart rate monitoring and activity diaries for estimating energy expenditure. Am. J. Clin. Nutr., 49, 37-43 (1989).

LIVINGSTONE M.B.E., PRENTICE, A.M., COWARD, W.A., CEESAY, S.M., STRAIN, J.J., McKENNA, P.G., NEVIN, G.B., BARKER M.E., HICKEY, R.J.: Simultaneous measurement of free-living energy expenditure by the doubly-labelled water (2H218O) method and heart-rate monitoring. Am. J. Clin. Nutr., in press.

SCHOELLER, D.A.: Measurement of energy expenditure in free-living humans by using doubly labelled water. J. Nutr., 118, 1278-1289 (1988).

SCHOFIELD, W.N.: Predicting basal metabolic rate, new standards and review of previous work. Hum. Nutr.: Clin. Nutr., 39C (Suppl.), 5-41 (1985).

SPADY, D.W.: Total daily energy expenditure of healthy, free-ranging school children. Am. J. Clin. Nutr., 33, 766-775 (1980).

SPURR, G.B., BARAC-NIETO, M., MAKSUD, M.G.: Energy expenditure cutting sugar cane. J. Appl. Physiol., 39, 990-996 (1975).

SPURR, G.B., REINA, J.C., BARAC-NIETO, M.: Marginal malnutrition in schoolaged Colombian boys: metabolic rate, and estimated daily energy expenditure. Am. J. Clin. Nutr., 44, 113-126 (1986).

SPURR, G.B., REINA, J.C.: Patterns of daily energy expenditure in normal and marginally undernourished school-aged Colombian children. Eur. J. Clin. Nutr., 42, 819-834 (1988a).

SPURR, G.B., REINA J.C.: Influence of dietary intervention on artificially increased activity in marginally undernourished Colombian boys. Eur. J. Clin. Nutr., 42, 835-846 (1988b).

SPURR, G.B., PRENTICE, A.M., MURGATROYD P.R., GOLDBERG, G.R., REINA, J.C., CHRISTMAN, N.T.: Energy expenditure from minute-by-minute heart-rate recording: comparison with indirect calorimetry. Am. J. Clin. Nutr., 48, 552-559, (1988).

SPURR, G.B., REINA, J.C.: Energy expenditure/basal metabolic rate ratios in normal and marginally undernourished Colombian children 6-16 years of age. Eur. J. Clin. Nutr., 43, 515-527 (1989).

SPURR, G.B.: Physical activity and energy expenditure in undernutrition. Prog. Food Nutr. Sci., in press.

SPURR, G.B., REINA, J.C.: Daily pattern of %VO2 max and heart rates in normal and undernourished children. Med. Sci. Sports Exerc., in press.

TAYLOR, C.B., KRAEMER, H.C., BRAGG, D.A., LAUGHTON E.M., RULE, B., SAVIN, W.M., DEBUSK, R.F.: A new system for long-term recording and processing of heart rate and physical activity in out-patients. Comput. Biomed. Res., 15, 7-17 (1982).

TREMBLAY, A., BOUCHARD, C.: Assessment of energy expenditure and physical activity pattern in population studies. In: Nutritional Anthropology, pp. 101-116, F.E. JOHNSTON (Ed.). Alan R. Liss, New York, 1987.

WARNOLD, I., LENNER, R.A.: Evaluating of the heart rate method to determine the daily energy expenditure in disease. A study in juvenile diabetics. Am. J. Clin. Nutr., 30, 304-315 (1977).

WESTERTERP K.R., BROWNS, F., SARIS, W.H.M., TEN MOOR, F.: Comparison of doubly labelled water with respirometry at low- and high-activity levels. J. Appl. Physiol., 65, 53-56 (1988).

Contents - Previous - Next