Share this article on:

Estimating MET Values Using the Ratio of HR for Persons with Paraplegia

LEE, MIYOUNG1; ZHU, WEIMO2; HEDRICK, BRAD2; FERNHALL, BO2

Medicine & Science in Sports & Exercise: May 2010 - Volume 42 - Issue 5 - p 985-990
doi: 10.1249/MSS.0b013e3181c0652b
Applied Sciences

The current compendium of physical activity (CPA) cannot be applied to persons with disabilities due to the lack of physical activity (PA) they are regularly engaged in and inaccurate MET values when applied to persons with disabilities.

Purpose: The purposes of this study were (a) to determine whether HR ratio during PA and resting can be used to accurately predict MET values of PA in persons with paraplegia, (b) to compare individual calibration (IC) with group calibration (GC) in error reduction, and (c) to examine prediction generalizability through a cross-validation design.

Methods: Twenty-seven participants (aged 18-45 yr) with complete and incomplete paraplegia at T6 to L4 participated in this study. Oxygen uptake (V˙O2) and HR were measured simultaneously at rest and during 10 PA using indirect calorimetry and a Polar HR monitor. Predicted METs were calculated using the HR ratio for six activities by applying regression analysis by group (GC) and individuals (IC), respectively. The derived equations were then cross-validated using the four other activities, and corresponding METs were calculated. Absolute error rates (AC), paired t-test, and correlation (r) were used to determine the absolute and relative difference between observed and predicted METs.

Results: The overall correlation coefficient (r) between HR ratio and observed METs was 0.77 using group regression and 0.93 ± 0.05 using individual regression. GC (R 2 = 0.59, AC = 0.07%-65.25%) was less accurate than IC (R 2 = 0.90 ± 0.10, AC = 1.64%-10.26%). Cross-validation results also showed higher correlations for IC (r = 0.90 in IC and 0.72 in GC) between observed and predicted METs.

Conclusions: HR ratio was able to accurately predict METs of persons with paraplegia. IC estimated METs more accurately than GC.

1Oregon State University, Corvallis, OR; and 2University of Illinois at Urbana-Champaign, Champaign, IL

Submitted for publication May 2009.

Accepted for publication September 2009.

Address for correspondence: Miyoung Lee, Ph.D., Movement Studies in Disability, Department of Nutrition and Exercise Sciences, Oregon State University, 103 Women's Bldg, Corvallis, OR 97331; E-mail: miyoung.lee@oregonstate.edu.

Because physical activity energy expenditure (PAEE) is inversely correlated with obesity and other risk factors of chronic disease (5,19,25), tremendous efforts have been made to develop accurate methods for measuring PAEE. Although PAEE can be accurately measured using criterion measures (e.g., respiratory chamber, indirect calorimetry, and doubly labeled water), population-based surveillance or large-sample studies still depend on self-report measures (e.g., questionnaire or diary) because of lower cost, time commitment, and convenience. To estimate PAEE using a self-report, however, the MET values of physical activities (PA) must be known (27,40).

MET values have been used as an index in public health recommendations for both PA promotion and weight management. For example, in the 2008 Physical Activity Guidelines Advisory Committee Report (31), moderate-intensity activity was defined as 3.0-5.9 METs, and it was recommended that people participate in moderate-intensity activities for 150-300 min·wk−1 to maintain and enhance health (15). A major national effort, therefore, had been made to compile the MET values of more than 600 PA together, and the complied list is called the compendium of physical activity (CPA) (2,3).

The current CPA, however, cannot be applied to persons with disabilities. There are two reasons for this: (a) there are almost no disability-specific activities (e.g., wheelchair activities) included in the current CPA and (b) the MET values in the CPA may not be accurate when applying them to persons with disabilities because these values were determined on the basis of the general population. To be able to apply the CPA to a specific subpopulation of persons with disabilities, MET values need to be determined for the activities regularly engaged in by that subpopulation.

HR may aid the accuracy of MET estimations because of the well-known linear relationship between HR and energy expenditure (EE) (1,39). Because HR increases linearly when the intensity of a PA and its corresponding EE increases, the ratio of HR (HR ratio) during a PA to the resting HR should be able to accurately estimate MET values of the PA, which is the ratio of V˙O2 during the PA to resting V˙O2. In fact, HR has long been used in PA assessment practice (27). As an example, ratios of HR responses between resting and physical activities (e.g., physiological cost index = HR walking − HR rest/walking speed and PA level = TEE/basal metabolic rate) have been used to estimate EE or to obtain the energy efficiency in persons with and without disabilities (6,12,21,26,37). These studies have shown that HR can be an accurate, inexpensive, and efficient method to estimate EE.

There is, however, a large range of individual variation in HR that may affect the MET values of PA. For instance, the range of prediction error was approximately ±15%-20% when using the physiological cost index to estimate EE (1). This error rate can be reduced by using individual calibration (IC).

When a calibration is derived on the basis of a group (e.g., a relationship between an outcome measure and a predictor), it is called group calibration (GC), which may have a large prediction error. When a prediction equation is developed for each individual, the calibration is called IC. Although IC has been shown to be more accurate in estimating PAEE than GC (33,34,38), it has several limitations, e.g., subject burden, high cost, and it is time consuming (7,38). Furthermore, it is important to use cross-validation (CV) to examine the accuracy of the prediction.

Although technically sound, the HR ratio has not been used to determine MET values, and the impact of IC on prediction error reduction and the prediction generalizability are still unknown in people with disabilities. Using a sample of persons with paraplegia, the purposes of this study were threefold: (a) to determine whether the HR ratio can accurately be used to predict the MET values of a set of selected PA, (b) to examine the impact of IC on error reduction, and (c) to examine the prediction generalizability by using CV.

Back to Top | Article Outline

METHODS

Participants.

A total of 27 participants with paraplegia between T6 and L4 (15 males and 12 females) were recruited for the study, which included 12 subjects with a complete lesion and 15 subjects with an incomplete lesion. The participants exhibited a wide range of American Spinal Injury Association Impairment Scale levels, from A to D (A = 12, B = 5, C = 4, and D = 6 participants, respectively). The duration of spinal cord injury ranged from 6 to 45 yr (16 ± 8 yr), and participants verbally provided information on the stability of their spinal cord injury. All participants signed informed consent before participation. The study was approved by the institutional review board at the University of Illinois at Urbana-Champaign.

Back to Top | Article Outline

Measures.

All participants completed measures of resting metabolic rate (RMR; i.e., 1 MET value) and MET values of selected PA. They were asked not to consume alcohol or food for at least 12 h and not to participate in vigorous PA 24 h before the testing day.

The protocol of RMR data collection has been previously described (16). Briefly, the participants came to the laboratory in the morning, shortly after their daily awaking time. Testing took place in a quiet room with dim lighting, and the temperature was maintained between 21°C and 24°C. Subjects rested quietly in a supine position for 30 min, and then data were collected for an additional 30 min. V˙O2 was measured using Canopy (Quark b2 with Canopy Kit; COSMED, Rome, Italy), which is an open-circuit spirometer. Only the last 15 min of data were used for RMR calculation.

On the basis of an extensive literature review of disability researches and expected exercise intensities, 10 representative PA for persons with paraplegia were carefully selected by a panel of experienced PA researchers, wheelchair sport coaches, and individuals who use wheelchairs. The selected activities included working on a computer, sitting and watching TV, reading, pushing the wheelchair on a tiled floor, vacuuming, mopping, moving chairs, pushing the wheelchair on a sidewalk, pushing up and down a ramp, and exercising in an arm ergometer.

Participants performed each PA for 10 min, and their oxygen consumption during the performances was measured by an indirect calorimetry (K4b2; COSMED). Testing of all 10 PA was completed within 14 d, and a counterbalanced design was used to avoid any carryover effects (20). V˙O2peak was measured using a discontinuous arm ergometry protocol (14,28,32).

HR during both RMR and PA performance were measured using an HR monitor (Polar, Kempele, Finland). Because fat-free mass and percent body fat may affect on RMR, body composition was measured by dual-energy x-ray absorptiometry (QDR 4500 Elite; Hologic, Bedford, MA).

Back to Top | Article Outline

MET value and HR ratio computations.

MET values of PA collected were calculated using equation 1. Of the 10-min data during the performance of PA, 6 min was used after eliminating the first 3 min and the last minute of V˙O2.

Using the corresponding HR data, the HR ratio was computed with equation 2.

Back to Top | Article Outline

Statistics analysis.

Descriptive statistics (e.g., mean, SD, and skewness) were first computed for all the variables. For GC, IC, and CV, a series of linear regression analyses were then conducted to predict MET values using HR ratio, respectively.

Using all participants' data of the six activities (i.e., working on computer, sitting and watching TV, vacuuming, moving chairs, pushing the wheelchair, and exercising in an arm ergometer), MET values were predicted with equation 3.

where a is intercept and b is slope. To examine the accuracy of the prediction, SEE and absolute error percentage were computed, and the latter was determined using equation 4.

where METpredicted values were derived from equation 3 and the METobserved value was measured in the laboratory. The prediction accuracy was also examined by paired t-test of the observed and predicted MET values. Type I error (α) was set at 0.05 with an adjustment using Bonferroni techniques when needed. Finally, Pearson product-moment correlation coefficients (r) were calculated to examine the relative relationship between observed and predicted METs. All the IC computations were the same as those in the GC analyses above except that predictions using equation 3 were now derived for each participant individually.

To examine whether generated equations can be applied to other PA's MET values, a CV study was conducted. Instead of using a new sample of participants, we evaluated if the regression equation could predict the MET for another set of PA. Therefore, derived prediction equations from both GC and IC were applied to the rest of four activities (i.e., reading, pushing the wheelchair on a tiled floor, mopping, and pushing up and down a ramp). Similarly, all accuracy and relationship indices used in GC and IC analyses were computed. Statistical packages (e.g., SPSS 17 SPSS Inc., Chicago, IL) and Microsoft Excel (Bellevue, WA) were used for the computation. Research design of the study is summarized in Figure 1.

FIGURE 1

FIGURE 1

Back to Top | Article Outline

RESULTS

The data were normally distributed (e.g., Kolmogorov-Smirnov and Shapiro-Wilk, P > 0.05). The participants' demographic information is summarized in Table 1. There was no statistical difference in the demographic variables, observed MET, and HR ratio by the completeness of lesion groups, with P > 0.05.

TABLE 1

TABLE 1

HR ratios and observed MET values are summarized in Table 2. Four PA of the 10 activities were low-intensity activities, ranging from 1.2 METs (sitting and watching TV) to 2.7 METs (pushing the wheelchair on a tiled floor). All other activities were moderate-intensity activities, ranging from 3 to 6 METs. The overall correlation coefficient (r) between the HR ratio and the observed MET values was 0.77. The individual relationships between HR ratio and the observed MET values ranged from 0.78 to 0.99, with a mean ± SD of 0.93 ± 0.05.

TABLE 2

TABLE 2

The regression analysis results of GC and IC of the six original PA are summarized in Table 2. For GC-based analysis, the derived relationship between HR ratio and observed MET values was as follows: METpredicted = −0.99 + 2.49(HR ratio), with R 2 = 0.59 (SEE = 1.10). The correlation between observed and predicted METs was moderately high with r = 0.78 in GC. The absolute error percentage of GC was rather high with a mean ± SD of 28.83% ± 28.21%. There was a large variation in prediction error among the PA used. Vacuuming showed the lowest error at 0.07%, and sitting and watching TV showed the highest error of 65.25% (Table 2 and Fig. 2).

FIGURE 2

FIGURE 2

With IC, the prediction errors were significantly reduced. The mean ± SD of R 2 from IC was 0.90 ± 0.10, and constants (a) and slope (b) of individual equations were −3.28 ± 1.63 and 3.67 ± 1.46, respectively. The average correlation between observed and predicted METs with IC was very high with r = 0.93. Absolute error percentage of IC (6.05% ± 2.94%) was much lower than those of GC. Similarly, vacuuming showed the lowest error (1.64%), and sitting and watching TV showed the highest one (10.26%; Table 2 and Fig. 2).

Results of CV using reading, pushing wheelchair on a tiled floor, mopping, and pushing up and down a ramp are also summarized in Table 2. The correlation coefficients (r) between observed and predicted MET values in the CV were 0.72 and 0.90 after applying the equation from GC and IC, respectively. In addition, there was no statistical difference between MET values using a paired t-test with Bonferroni corrections (α = 0.05/4 = 0.012). Absolute error rates were 25.79% ± 27.90% from the GC equation and 8.38% ± 6.11% from the IC equations.

Back to Top | Article Outline

DISCUSSION

Overall, the HR ratio of persons with paraplegia showed a good relationship with the observed MET values (r ≥ 0.72) in this study. This result is similar to the findings of Wareham et al. (39), in which EE between HR and the ratio of total EE/resting EE (n = 167 and 30-40 yr old) were r = 0.77 in males and r = 0.71 in females, respectively. There was also no statistical difference between observed and predicted MET values on the basis of either the IC or the GC in this study. As expected, IC reduced the prediction error significantly by approximately 23% (Table 2). The IC error rate in the current study was lower than previously reported for individuals with spinal cord injury (SCI) (17). For example, Hayes et al. (17) estimated EE (kcal·min−1) using HR in 13 persons with SCI including tetraplegia and paraplegia. The EE in that study was estimated by IC using an arm ergometer exercise test in a laboratory setting. Derived prediction equations were applied to estimate the EE of PA in free-living setting (e.g., desk work, doing dishes, transferring, wheeling on tile, doing laundry, and resting), and the EE predicted by HR showed a 25% higher overall than the observed EE.

Compared with the current study, the higher error rate observed by Hayes et al. (17) could be due to several reasons. First, it may have been carryover effects among the activities previously performed because participants only rested for 3 min between activities. In contrast, we asked the participants in our study to rest at least 20-30 min between activities until the HR came back to baseline. In addition, we applied a counterbalanced design to avoid fatigue and to reduce carryover effects from the previous activities. Second, the study participants in Hayes et al. (17) performed each activity for 5 min and used all 5 min for the data analysis, but our study participants performed each activity for 10 min, and the first 3 min and the last 1 min were not used to reduce the error from the slow HR responses to the activity intensity in the beginning and at the end of each activity. Finally, the higher error of EE in the study of Hayes et al. may have been due to a wide range of SCI (C5 to S12) with a small sample size (n = 13). In contrast, the participants in our study had a limited range of SCI at the paraplegic level and included a larger number of participants (n = 27).

In our study, lower-intensity activities showed higher error rates than higher-intensity activities using both GC and IC. This was similar to the result of Hayes et al. (17), in which the lower-intensity PA showed a lower relationship (r = 0.53) between observed and predicted EE in contrast to a higher correlation (r = 0.94), when a wide range of PA intensities was selected. Interestingly, other studies showing larger variability at lower-intensity activities (9,22-24) applied a variety of statistical approaches to correct the estimations (e.g., multiple prediction equations or adjusting the resting or lower intensity of activities to derive prediction equations). However, most studies that used statistical adjustments estimated free-living EE for 24 h rather than the EE of an individual as we did in this study.

A nonlinear relationship between EE and HR during low-intensity PA (e.g., <2.5 METs, see Fig. 2) may cause substantial inaccuracy when predicting the EE of these lower-intensity PA. To solve the problem of the nonlinearity, the HR flex method has been used (22), where the "flex" point of active EE excludes the activity at rest from the prediction equation. After applying the HR flex, the absolute error rate may be reduced up to 2%-3%, but determining the "flex" point is subjective, and it may therefore affect the error rate (36,39). In our study, instead of using the HR flex method, we used a wide range of PA intensities, which were carefully selected, and it did show a linear relationship (Fig. 2) between HR and EE. Thus, our prediction equations included both low- and moderate-intensity activities (in contrast to the HR flex method), which is closer to real-life situations for most individuals who take a part in a variety of low- and moderate-intensity activities.

To examine the accuracy of the derived prediction equation, we included a CV design, which was rarely done in previous studies. After applying the prediction equation derived from the six initial activities to the four other activities for the CV, the higher-intensity activities again produced a lower error rate than the lower-intensity activities (e.g., 2.55% of activity 9). The error rate of the predicted METs using the IC equation was considerably lower than when using GC (8.38% vs 25.79%, respectively). Therefore, it is important to include an IC component when determining MET values for this population, although using IC is cumbersome. In fact, IC using HR has been used to predict EE in population studies (4,8,39), showing the feasibility of using HR in larger sample studies with persons without disabilities.

SCI can be a critical cause of altering a person's functional and physiological conditions, and it can lead to consequences of chronic diseases (e.g., cardiovascular disease or obesity) (13,30). In addition, musculoskeletal declines and sympathetic nervous system dysfunction cause lower V˙O2peak and HRpeak and require higher V˙O2 and HR during PA and exercise, especially for individuals with higher levels of paraplegia (e.g., upper thoracic cord injury) compared with able-bodied individuals (18,35). On the other hand, participating in regular PA and exercise provided health benefits, such as increasing V˙O2peak and muscular fitness (10,29). Consequently, promoting a physically active lifestyle is important for persons with SCI, and being able to estimate EE of various activities may aid in promoting PA in this population.

There are a few limitations in this study. First, the participant's trunk was not stabilized while performing each activity because we tried to simulate an individual's daily living activities and to make the test setting as natural as possible. The lack of trunk stabilization may have affected the increase in EE if participants had a higher level of injury and lack of trunk stability (11). Another limitation is the exclusion of tetraplegia and the higher level of paraplegia (above T5) due to the different HR response compared with individuals with a lower level of SCI (18). Finally, more MET values of PA, which represent the activities that persons with SCI often undertake, need to be examined to develop a full version of a CPA supplement for persons with SCI.

Back to Top | Article Outline

CONCLUSIONS

Using the HR ratio produced accurate estimates of MET values when IC was used. After determining the individual relationship between HR and EE both at rest and during a set of calibration activities, the derived relationship can be extended to determine the MET values of additional PA. A large variability in prediction error was observed among the activities used. Future studies should examine the factors causing this difference. The HR ratio method can be used to help improve PAEE assessment in persons with SCI.

This study was supported by Carl V. Gisolfi Memorial Research Fund of the American College of Sports Medicine foundation and Schneider Research Award, College of Applied Health Sciences, University of Illinois at Urbana-Champaign. The results of the present study do not constitute endorsement by the American College of Sports Medicine.

This study was not supported by private companies, manufacturers, or outside organizations providing technical or equipment. Also, this study has not been submitted to any other journal.

Back to Top | Article Outline

REFERENCES

1. Achten J, Jeukendrup AE. Heart rate monitoring: applications and limitations. Sports Med. 2003;33:517-38.
2. Ainsworth BE, Haskell WL, Leon AS, et al. Compendium of physical activities: classification of energy costs of human physical activities. Med Sci Sports Exerc. 1993;25(1):71-80.
3. Ainsworth BE, Haskell WL, Whitt MC, et al. Compendium of physical activities: an update of activity codes and MET intensities. Med Sci Sports Exerc. 2000;32(9 suppl):S498-516.
4. Assah FK, Brage S, Ekelund U, Wareham NJ. The association of intensity and overall level of physical activity energy expenditure with a marker of insulin resistance. Diabetologia. 2008;51:1399-407.
5. Bazzarre TL, Murdoch SD, Wu SM, Hopkins RG. Associations of cardiovascular disease risk factors with measures of energy expenditure and caloric intake in a farm population. J Am Coll Nutr. 1992;11:42-9.
6. Bowen TR, Lennon N, Castagno P, Miller F, Richards J. Variability of energy-consumption measures in children with cerebral palsy. J Pediatr Orthop. 1998;18:738-42.
7. Brage S, Brage N, Franks PW, et al. Branched equation modeling of simultaneous accelerometry and heart rate monitoring improves estimate of directly measured physical activity energy expenditure. J Appl Physiol. 2004;96:343-51.
8. Brage S, Ekelund U, Brage N, et al. Hierarchy of individual calibration levels for heart rate and accelerometry to measure physical activity. J Appl Physiol. 2007;103:682-92.
9. Christensen C, Frey H, Foenstelien E, Aadland E, Refsum H. A critical evaluation of energy expenditure estimates based on individual O2 consumption/heart rate curves and average daily heart rate. Am J Clin Nutr. 1983;37:468-72.
10. Davis GM, Shephard RJ. Strength training for wheelchair users. Br J Sports Med. 1990;24:25-30.
11. Durstine JL, Moore GE. ACSM's Exercise Management for Persons with Chronic Diseases and Disabilities. Champaign (IL): Human Kinetics; 2003. p. 247-53.
12. Ekelund U, Sjostrom M, Yngve A, Nilsson A. Total daily energy expenditure and pattern of physical activity measured by minute-by-minute heart rate monitoring in 14-15 year old Swedish adolescents. Eur J Clin Nutr. 2000;54:195-202.
13. Fernhall B, Heffernan K, Jae SY, Hedrick B. Health implications of physical activity in individuals with spinal cord injury: a literature review. J Health Hum Serv Adm. 2008;30:468-502.
14. Haisma JA, van der Woude LH, Stam HJ, Bergen MP, Sluis TA, Bussmann JB. Physical capacity in wheelchair-dependent persons with a spinal cord injury: a critical review of the literature. Spinal Cord. 2006;44:642-52.
15. Haskell WL, Lee IM, Pate RR, et al. Physical activity and public health: updated recommendation for adults from the American College of Sports Medicine and the American Heart Association. Circulation. 2007;116:1081-93.
16. Haugen HA, Melanson EL, Tran ZV, Kearney JT, Hill JO. Variability of measured resting metabolic rate. Am J Clin Nutr. 2003;78:1141-5.
17. Hayes AM, Myers JN, Ho M, Lee MY, Perkash I, Kiratli J. Heart rate as a predictor of energy expenditure in people with spinal cord injury. J Rehabil Res Dev. 2005;42:617-24.
18. Jacobs PL, Nash MS. Exercise recommendations for individuals with spinal cord injury. Sports Med. 2004;34:727-51.
19. Johnstone AM, Murison SD, Duncan JS, Rance KA, Speakman JR. Factors influencing variation in basal metabolic rate include fat-free mass, fat mass, age, and circulating thyroxine but not sex, circulating leptin, or triiodothyronine. Am J Clin Nutr. 2005;82:941-8.
20. Kirk RE. Experimental Design: Procedures for the Behavioral Sciences. Pacific Grove (CA): Brooks/Cole Publishing Co.; 1995. p. 319-63.
21. Lazzer S, Boirie Y, Bitar A, et al. Assessment of energy expenditure associated with physical activities in free-living obese and nonobese adolescents. Am J Clin Nutr. 2003;78:471-9.
22. Leonard WR. Measuring human energy expenditure: what have we learned from the flex-heart rate method? Am J Hum Biol. 2003;15:479-89.
23. Leonard WR, Katzmarzyk PT, Stephen MA, Ross AG. Comparison of the heart rate-monitoring and factorial methods: assessment of energy expenditure in highland and coastal Ecuadoreans. Am J Clin Nutr. 1995;61:1146-52.
24. Li R, Deurenberg P, Hautvast JG. A critical evaluation of heart rate monitoring to assess energy expenditure in individuals. Am J Clin Nutr. 1993;58:602-7.
25. Luke A, Durazo-Arvizu R, Cao G, Adeyemo A, Tayo B, Cooper R. Positive association between resting energy expenditure and weight gain in a lean adult population. Am J Clin Nutr. 2006;83:1076-81.
26. Makino K, Wada F, Hachisuka K, Yoshimoto N, Ohmine S. Speed and physiological cost index of hemiplegic patients pedalling a wheelchair with both legs. J Rehabil Med. 2005;37:83-6.
27. Montoye HJ, Kemper CG, Saris WHM, Washburn RA. Measuring Physical Activity and Energy Expenditure. Champaign (IL): Human Kinetics; 1996. p. 97-115.
28. Mossberg K, Willman C, Topor MA, Crook H, Patak S. Comparison of asynchronous versus synchronous arm crank ergometry. Spinal Cord. 1999;37:569-74.
29. Muraki S, Ehara Y, Yamasaki M. Cardiovascular responses at the onset of passive leg cycle exercise in paraplegics with spinal cord injury. Eur J Appl Physiol. 2000;81:271-4.
30. Myers J, Lee M, Kiratli J. Cardiovascular disease in spinal cord injury: an overview of prevalence, risk, evaluation, and management. Am J Phys Med Rehabil. 2007;86:142-52.
31. Physical Activity Guidelines Advisory Committee. Physical Activity Guidelines Advisory Committee Report, 2008. To the secretary of health and human services. Part A: executive summary. Nutr Rev. 2009;67:114-20.
32. Rasche W, Janssen TW, Van Oers CA, Hollander AP, Van der Woude LH. Responses of subjects with spinal cord injuries to maximal wheelchair exercise: comparison of discontinuous and continuous protocols. Eur J Appl Physiol Occup Physiol. 1993;66:328-31.
33. Rennie KL, Hennings SJ, Mitchell JO, Wareham NJ. Estimating energy expenditure by heart-rate monitoring without individual calibration. Med Sci Sports Exerc. 2001;33(6):939-45.
34. Rennie KL, Rowsell T, Jebb SA, Holburn D, Wareham NJ. A combined heart rate and movement sensor: proof of concept and preliminary testing study. Eur J Clin Nutr. 2000;54:409-14.
35. Schmid A, Huonker M, Barturen JM, et al. Catecholamines, heart rate, and oxygen uptake during exercise in persons with spinal cord injury. J Appl Physiol. 1998;85:635-41.
36. Spurr GB, Prentice AM, Murgatroyd PR, Goldberg GR, Reina JC, Christman NT. Energy expenditure from minute-by-minute heart-rate recording: comparison with indirect calorimetry. Am J Clin Nutr. 1988;48:552-9.
37. Swain DP, Leutholtz BC, King ME, Haas LA, Branch JD. Relationship between % heart rate reserve and % V˙O2 reserve in treadmill exercise. Med Sci Sports Exerc. 1998;30(2):318-21.
38. Treuth MS, Adolph AL, Butte NF. Energy expenditure in children predicted from heart rate and activity calibrated against respiration calorimetry. Am J Physiol. 1998;275:E12-8.
39. Wareham NJ, Hennings SJ, Prentice AM, Day NE. Feasibility of heart-rate monitoring to estimate total level and pattern of energy expenditure in a population-based epidemiological study: the Ely Young Cohort Feasibility Study 1994-5. Br J Nutr. 1997;78:889-900.
40. Welk GJ. Physical Activity Assessment for Health-Related Research. Champaign (IL): Human Kinetics; 2002. p. 107-23.
Keywords:

MEASUREMENT; ENERGY EXPENDITURE; DISABILITY; PHYSICAL ACTIVITY; ACCURACY

©2010The American College of Sports Medicine