Original Article | Published:

Maternal and pediatric nutrition

Evaluation of the wrist-worn ActiGraph wGT3x-BT for estimating activity energy expenditure in preschool children

European Journal of Clinical Nutrition volume 71, pages 12121217 (2017) | Download Citation



Easy-to-use and accurate methods to assess free-living activity energy expenditure (AEE) in preschool children are required. The aims of this study in healthy preschool children were to (a) evaluate the ability of the wrist-worn ActiGraph wGT3x-BT to predict free-living AEE and (b) assess wear compliance using a 7-day, 24-h protocol.


Participants were 40 Swedish children (5.5±0.2 years) in the Mobile-based intervention intended to stop obesity in preschoolers (MINISTOP) obesity prevention trial. Total energy expenditure (TEE) was assessed using the doubly labeled water method during 14 days. AEE was calculated as (TEEx0.9) minus predicted basal metabolic rate. The ActiGraph accelerometer was worn on the wrist for 7 days and outputs used were mean of the daily and awake filtered vector magnitude (mean VM total and mean VM waking).


The ActiGraph was worn for 7 (n=34, 85%), 6 (n=4, 10%), 5 (n=1, 2.5%) and 4 (n=1, 2.5%) days (a valid day was 600 awake minutes). Alone, mean VM total and mean VM waking were able to explain 14% (P=0.009) and 24% (P=0.001) of the variation in AEE, respectively. By incorporating fat and fat-free mass in the models 58% (mean VM total) and 62% (mean VM waking) in the variation of AEE was explained (P<0.001).


The wrist-worn ActiGraph wGT3x-BT in combination with body composition variables explained up to the 62% of the variation in AEE. Given the high wear compliance, the wrist-worn ActiGraph has the potential to provide useful information in studies where physical activity in preschool children is measured.


Current physical activity guidelines recommend that children aged 5 years and older acquire a minimum of 60 min of moderate-to-vigorous intensity physical activity each day.1 In many countries, children do not meet this recommendation,2 which is concerning as a combination of low physical activity and high sedentary time are related to overweight and obesity.3 As physical activity behaviors later in life can be impacted by childhood experiences,4 intervention trials aimed at increasing physical activity and decreasing sedentary behaviors are needed in preschool children. To carry out such interventions, accurate and easy-to-use methods to assess physical activity are required.

Overweight and obesity is caused by an imbalance between energy intake and expenditure, which can lead to numerous health consequences. This imbalance could be counteracted by increasing physical activity to increase activity-related energy expenditure. Thus, an important aspect to measure is energy expenditure in response to physical activity (i.e. activity energy expenditure, AEE). Doubly labeled water (DLW) is the gold standard to assess total energy expenditure (TEE) in humans under free living conditions.5 AEE can then be determined using measured or predicted basal metabolic rate (BMR)6 after correcting for dietary-induced thermogenesis. However, this procedure is time consuming and expensive, which limits its use in larger populations. Accelerometers measure bodily accelerations and is a common method for objectively measuring the duration and intensity of physical activity in a range of populations, and also has the potential to indirectly estimate AEE under free-living conditions.6, 7 However, there are numerous types of monitors available, as well as a variety of placement and processing criteria options, which all need to be validated in the desired population.

Traditionally, accelerometers have been placed on the waist; however, recently there has been increased interest in placing them on the wrist.8, 9 This placement enables the measurement of both physical activity and sleep, and it has also been suggested to increase wear compliance in older children8 and adults.10, 11 Although wrist-worn accelerometry may perform less well in capturing all body movements contributing to AEE,11 van Hees et al.10 reported that the wrist-worn accelerometer outputs contributed significantly to the explained variance of AEE in women. In preschool-aged children, a few studies have validated the use of waist and chest worn accelerometers to measure AEE or TEE using the DLW method.12, 13, 14, 15, 16, 17 The results are mixed with two studies reporting no associations,15, 18 while the other four demonstrated positive correlations.13, 14, 16, 17 However, to date, the potential of wrist-worn accelerometers to assess AEE has not been evaluated under free-living conditions in any preschool-aged population.

The Mobile-based intervention intended to stop obesity in preschoolers (MINISTOP) study is a randomized controlled trial that aimed to evaluate the effectiveness of a mobile-phone-based intervention to promote better body composition, dietary habits and physical activity in healthy preschool-aged children.19, 20 A nested validation study within this trial was conducted to: (a) evaluate the capacity of the wrist-worn ActiGraph wGT3x-BT to predict free-living AEE and (b) assess wear compliance using a 7-day, 24-h protocol in healthy preschool-aged children.

Materials and methods

Participants and study design

Forty parents and their child participating in the MINISTOP trial19, 20, 21 partook in this nested validation study.22 The recruitment and overall procedures have been published elsewhere.22 In brief, the child’s TEE was measured during 14 days using the DLW method.22 The child’s activity was measured using the ActiGraph wGT3x-BT accelerometer (ActiGraph Corporation, Pensacola, FL, USA) for the first 7 days of this period. Ethical approval was obtained from the Research and Ethics Committee (Stockholm, Sweden; 2013/1607-31/5; 2013/2250-32) and both parents provided informed consent. MINISTOP is registered as a clinical trial (https://clinicaltrials.gov/ct2/show/NCT02021786).

Body composition and reference estimate of AEE

All children consumed an accurately weighed dose of stable isotopes.21 Urine samples were collected, stored and analyzed for isotope enrichments as described previously.22 The equation published by Davies et al.23 (assumption: 27.1% of the water losses being fractionated22) was used to derive carbon dioxide production. TEE was calculated by means of the Weir equation24 assuming a food quotient of 0.85.25 The quotient between 2H dilution space (ND) and 18O dilution space (NO) was 1.039±0.008 for the 40 children in this study. Total body water was computed as the average of ND/1.041 and NO/1.007.5 Fat-free mass (FFM) was determined by dividing total body water by 0.764.26 Fat mass (FM) was then calculated as body weight minus FFM. There was no change in body weight during the 14-day study period (0.07±0.32 kg).

BMR was calculated using prediction equations based on weight.27 AEE was calculated as: (TEE × 0.9)−BMR, assuming that dietary-induced thermogenesis corresponds to 10% of TEE. Physical activity level (PAL) was calculated as: TEE/BMR.


The ActiGraph wGT3x-BT triaxial accelerometer (http://www.ActiGraphcorp.com) was used to assess bodily movements 24 h per day for seven consecutive days and was worn on the non-dominant wrist. We collected data at 50 Hz, as this sampling frequency has shown to sufficiently capture body movement28 and allow for a whole week of data collection. The low-frequency filter included in the ActiLife Software (version 6.13.0, ActiGraph Corporation) was used to process the raw data to derive filtered sum of vector magnitudes (VM) in activity counts (calculated as the square root of the sum of the square of acceleration for each of the three axes). Raw acceleration data and the filtered VM data in one second epochs were then exported to comma separated value files using the ActiLife Software and imported into SAS (version 9.3, Cary, NC, USA) for processing. As we have described previously,21 non-wear time was determined by means of the raw acceleration data using the approach adapted from van Hees et al.10 The calculated non-wear time was confirmed by a diary kept by the parents during the measurement period. Wear time was then classified into sleep or awake time using the Sadeh algorithm29, 30 using the count data. A valid day was when the awake wearing time (non-sleeping data) was 600 min31 and a minimum of 3 days of valid data were included in the analyses. We calculated the mean per minute filtered VM during all wear time (mean VM total) and the mean per minute filtered VM during only minutes classified as awake (mean VM waking), both expressed in counts per minute (c.p.m.).


Values are presented as means and standard deviations. Regression analyses were used to determine the amount of variation in AEE explained by the ActiGraph wGT3x-BT variables alone and in combination with weight, gender, age, FFM and FM. First, regression models were fitted using AEE (kJ/24 h) as the dependent variable (y) and the independent variables (x) were fitted starting with mean VM total (c.p.m.), followed by sex, age and weight (kg) or the combination of FFM (kg) and FM (kg). Thereafter, the same regression model was fitted with the mean VM waking (c.p.m.). We also fitted models with PAL as the dependent variable. The multiple regression models were cross-validated using the leave-one-out approach.10, 32, 33 Briefly, AEE for each child was predicted from the equation derived from the whole sample minus that child and then this was repeated for all children. Thereafter, the agreement between predicted and measured AEE was evaluated according to Bland and Altman procedure.34 It was estimated that 40 children would provide 80% power (α=0.05) to detect correlations between ActiGraph variables and AEE of 0.43 or higher (r2=0.16). A two-sided test (P<0.05) was considered statistically significant. Analyses were performed using SPSS Statistics Version 23 (IBM, Armonk, NY, USA). The required assumptions (independence, linearity, homoscedasticity and normality) for the regression models were not violated.35


Age, weight, height, body composition, ActiGraph outputs and energy expenditure are presented in Table 1. Overall, there was considerable variation in body composition, AEE and ActiGraph outputs.

Table 1: Descriptive characteristics of the participating children

Thirty-four children (85%) wore the ActiGraph for 7 days, 4 children (10%) for 6 days, 1 child (2.5%) for 5 days and 1 child (2.5%) for 4 days. Table 2 shows the results when mean VM total, sex, age and weight or FFM and FM were included as independent variables and AEE as the dependent variable. Mean VM total explained 14% of the variation of AEE (P=0.009) (model 2A). When sex, age and body weight were added as independent variables, 1% more in the variation of AEE was explained (15%, P=0.048) (model 2B). If FFM and FM were substituted into this model for weight, 58% of the variation of AEE was explained (P<0.001) (model 2C).

Table 2: Regression models for activity energy expenditurea and mean VM total obtained using the ActiGraph (n=40)

Table 3 displays the results when AEE was included as the dependent variable and mean VM waking, sex, age and weight or FFM and FM were included as independent variables. Alone, mean VM waking explained 24% (P=0.001) of AEE (model 3A). When sex, age and weight were added to the model, an additional 2% of the variation was accounted for (26%, P=0.005) (model 3B). When FFM and FM were used instead of weight in the same model, 62% of the variation of AEE was explained (P<0.001) (model 3C).

Table 3: Regression models for activity energy expenditurea and mean VM waking obtained using the ActiGraph (n=40)

When PAL was substituted for AEE as the dependent variable in the regression models similar results were obtained (data not shown). For example, when only using mean VM waking, 25% (P=0.001) of the variation in PAL was explained. When sex, age, FFM and FM were added to the models as independent variables, 57% (P<0.001) of the variation in PAL was explained.

Figure 1 shows the Bland and Altman plot using the leave-one-out approach for the predicted AEE (mean VM total (a) and mean VM waking (b)) versus measured AEE. For both plots, the mean difference was 0.2%, the limits of agreement were wide and no significant association was found between the average and difference between the two methods.

Figure 1
Figure 1

(a) A Bland and Altman plot of the difference versus the mean AEE using DLW and predicted AEE using the leave-one-out cross-validation approach estimated using the ActiGraph (mean VM total), sex, age, fat-free mass and fat mass in 40 children. The mean difference between the methods was 2.4 kJ/24 h with limits of agreement (2 s.d.) of 621.6 kJ/24 h. (b) A Bland and Altman plot of the difference versus the mean AEE using DLW and predicted AEE using the leave-one-out cross-validation approach estimated using the ActiGraph (mean VM waking), sex, age, fat-free mass and fat mass in 40 children. The mean difference between the methods was −2.4 kJ/24 h with limits of agreement (2 s.d.) of 584.8 kJ/24 h.


This is the first evaluation of the ability of a wrist-worn accelerometer to assess free-living AEE in preschool children using DLW. We found that the ActiGraph wGT3x-BT outputs accounted for 24% of the variation in AEE, and that wear compliance was very high with 85% of the children wearing the monitor for all 7 days and 95% wearing it for at least 6 days. Only one other study has evaluated the accuracy of a wrist-worn accelerometer to assess AEE under free-living conditions. In adults, van Hees et al.10 found that the wrist-worn GENEA accelerometer (Unilever Discover, Sharnbrook Bedfordshire, UK) explained between 21 and 24% of the variation in AEE, which compares well to our findings. Comparisons between studies are difficult, because of the use of different devices, body placement and data analytic approaches;7 however, three studies in young children have validated waist and chest-attached accelerometers and explained 6,14 2217 and 31%13 of the variation in AEE. Of these, the Tracmor accelerometer (Philips DirectLife, Amsterdam, The Netherlands) appeared to account for the greatest variation in AEE.13 One explanation for the lower values obtained in our study compared with Sijtsma et al.13 is the placement of the monitors, where they placed the Tracmor accelerometer in the small area of the lower back, whereas we placed the ActiGraph on the non-dominant wrist. Indeed, in a lab setting it was found that the waist-worn Wockets accelerometer (http://wockets.wikispaces.com/) was able to explain more of the variation in AEE than a corresponding wrist-worn monitor, which the authors hypothesized was the result of the waist-worn monitor capturing greater movement from the larger muscle groups because of its central placement.36

In 2011, the National Health and Nutrition Examination Study (NHANES) switched from waist-worn to wrist-worn monitors.9 In the 2003–2006 NHANES survey, between 40 and 70% of participants provided at least 6 days of valid data (including ~10 h of wear time). With the wrist-worn devices 70 to 80% of subjects provided at least 6 days of valid data, with a median wear time between 21 and 22 h per day in the 2011–2012 NHANES survey.9 This is a significant increase in compliance; however, these results need to be interpreted with caution as the same protocol was not applied in both surveys. Another study in 6–9-year-old children compared the wear compliance of wrist and waist-worn accelerometers and found that significantly more children wore the wrist-worn monitor.8 Although our study was not a direct comparison for wrist and waist-worn monitors, we can conclude that a 7-day, 24-h protocol for wrist-worn accelerometers was well accepted.

One important observation from this study is the amount of variation of AEE that could be explained was much higher when body composition variables were included in the regression models. This finding is in accordance with previous studies in adults7, 37, 38 and young children14 and highlights the importance of including body composition measures when predicting AEE from accelerometers. Furthermore, it is also important to note that in this study both FFM and FM were independently associated with AEE. This finding is in agreement with our previous study in 1.5-year-old children.14 Further studies are warranted to elucidate the nature of the relationships between movements, body composition and energy expenditure in young children.

Our regression models for AEE was cross-validated using the leave-one-out approach as applied previously.10, 33, 39 On average, predicted AEE and measured AEE agreed well and we did not observe any trends in the differences in the agreement between these two methods across the levels of AEE. Together, these results indicate that the wrist-worn ActiGraph recordings may provide accurate group estimates of AEE. However, these results should be interpreted with caution as we cannot exclude that the use of the leave-one-out approach has produced a stronger agreement than would have been observed for an independent population. In this context, it is relevant to note that our results are very similar to those obtained by Assah et al.33 and Corder et al.39 who used a similar analytical approach. Hence, we recommend that our regression equations should also be evaluated in another population.

A major strength of this study was that we used DLW to assess TEE. This study was limited by use of a predicted BMR to calculate AEE. Owing to the extensive protocol for the MINISTOP trial we were unable to measure sleeping metabolic rate. We cannot exclude that this fact may have contributed to the lower degree of variation explained in this study compared with Sijtsma et al.13 study who measured sleeping metabolic rate. In addition, it may be argued that predicting BMR from body weight to calculate AEE increases the risk for a spurious correlation in regression models with body weight (models 2B and 3B); however, this is unlikely as body weight only explained a small fraction of AEE. We evaluated our ActiGraph protocol for the MINISTOP trial (7 days) using the DLW method (14 days) as 10–14 days provides the best precision for this method.5 It may be argued that this difference in time periods affected our results; however, this is unlikely as day-to-day variation in TEE is small40, 41 and the results were similar when using AEE assessed during 7 days (results not shown). Finally, we only used counts which is very good for clinical use, as they are simple to work with, but further work is needed using raw data features (e.g. Euclidean Norm Minus One), which could possibly have better associations with AEE and health outcomes.

The relatively small sample may limit generalizability; however, the participating children were representative of the whole study in regards to age, BMI and time spent in moderate-to-vigorous physical activity as well as parental age, BMI and education.21 Parental BMI was similar to the general Swedish population,42 the children’s TEE was similar to previously collected data in children from western countries43 and their body size was comparable to Swedish reference data.44 Participating parents had a higher education level with 73% of mothers and 65% of fathers having a university education, in comparison with the general Swedish population of 52% and 39%, respectively.45 We cannot exclude that this fact may have contributed to the high compliance, and further studies in families with lower educational level need to be performed.

We explained 24% of the variation in AEE by accelerometer outputs. To the best of our knowledge, there is no consensus regarding the amount of the variation in AEE a monitor should be required to explain to be useful for specific practical applications. However, 24% represents a moderate amount and two previous studies10, 33 in adults concluded that accelerometers have the potential to assess AEE based on similar findings. It is also relevant to note that a recent review of 28 studies in adults demonstrated a large heterogeneity across the studies in the explained variance of AEE where accelerometer outputs alone accounted for 4–80% (median 26%), and when adding other predictors such as body composition, this figure increased to 12.5–86% (median 41%).7 We were able to explain 58% (mean VM total) and 62% (mean VM waking) of the variation in AEE when using the ActiGraph wGT3x-BT outputs and body composition variables. This is a considerable amount and part of the remaining variation may be because of experimental errors as well as variations in efficiency in energy utilization during movements, dietary-induced thermogenesis and energy expenditure in response to growth. The latter only corresponds to ~1% in this age group;46, 47 however, variations between individual children likely occur. Thus, the wrist-worn ActiGraph explained a significant amount of the variation in AEE, demonstrating its potential in physical activity studies in preschool-aged children. However, future studies should investigate how such predictions of AEE can be improved; for instance, through technological advancements in monitors and data processing as well as the use of multiple sensors at different placement sites to see which are better at predicting the amount of explained variance.


In conclusion, the wrist-worn ActiGraph wGT3x-BT in combination with body composition variables explained up to 62% of the variation in AEE. Although confirmation is required, given the high wear compliance, the wrist-worn ActiGraph wGT3x-BT accelerometer has the potential to provide useful information in studies where physical activity in preschool children is measured.


  1. 1.

    World Health Organization. Global Recommendations on Physical Activity for Health. Global Strategy on Diet, Physical Activity and Health. World Health Organization: Geneva, Switzerland, 2010.

  2. 2.

    , , , , , et al. Global Matrix 2.0: report card grades on the physical activity of children and youth comparing 38 countries. J Phys Act Health 2016; 13 (Suppl 2), S343–S366.

  3. 3.

    , , . Addressing childhood obesity through increased physical activity. Nat Rev Endocrinol 2010; 6: 543–549.

  4. 4.

    World Health Organization Report of the Commission on Ending Childhood Obesity. Global Strategy on Diet, Physical Activity and Health. World Health Organization: Geneva, Switzerland, 2016.

  5. 5.

    International Atomic Energy Agency Assessment of Body Composition and Total Energy Expenditure in Humans Using Stable Isotope Techniques, IAEA Human Health Series No. 3. International Atomic Energy Agency: Vienna, Austria, 2009.

  6. 6.

    , , . Assessing physical activity using wearable monitors: measures of physical activity. Med Sci Sports Exerc 2012; 44 (Suppl 1), S5–S12.

  7. 7.

    , . Prediction of activity related energy expenditure using accelerometer derived physical activity under free-living conditions—a systematic review. Int J Obes 2016; 40: 1187–1197.

  8. 8.

    , , , , , . Wear compliance and activity in children wearing wrist and hip-mounted accelerometers. Med Sci Sports Exerc 2016; 48: 245–253.

  9. 9.

    , . Comment on 'estimating activity and sedentary behavior from an accelerometer on the hip and wrist'. Med Sci Sports Exerc 2013; 45: 962–963.

  10. 10.

    , , , , , et al. Estimation of daily energy expenditure in pregnant and non-pregnant women using a wrist-worn tri-axial accelerometer. PLoS One 2011; 6: e22922.

  11. 11.

    , , , . Evolution of accelerometer methods for physical activity research. Br J Sports Med 2014; 48: 1019–1023.

  12. 12.

    , , , , , et al. Ability of the actiwatch accelerometer to predict free-living energy expenditure in young children. Obesity Res 2004; 12: 1859–1865.

  13. 13.

    , , , , , et al. Validation of the TracmorD triaxial accelerometer to assess physical activity in preschool children. Obesity 2013; 21: 1877–1883.

  14. 14.

    , , . Evaluation of Actiheart and a 7 d activity diary for estimating free-living total and activity energy expenditure using criterion methods in 1.5- and 3-year-old children. Br J Nutr 2014; 111: 1830–1840.

  15. 15.

    , , , , , et al. Relation between physical activity and energy expenditure in a representative sample of young children. Am J Clin Nutr 2004; 80: 591–596.

  16. 16.

    , , , , , . Prediction of energy expenditure and physical activity in preschoolers. Med Sci Sports Exerc 2014; 46: 1216–1226.

  17. 17.

    , , , , , et al. Validity of hip-mounted uniaxial accelerometry with heart-rate monitoring vs. triaxial accelerometry in the assessment of free-living energy expenditure in young children: the IDEFICS Validation Study. J Appl Physiol 2012; 113: 1530–1536.

  18. 18.

    , , , , , et al. Validation of Actigraph accelerometer estimates of total energy expenditure in young children. Int J Pediatr Obes 2006; 1: 161–167.

  19. 19.

    , , , , , et al. A web- and mobile phone-based intervention to prevent obesity in 4-year-olds (MINISTOP): a population-based randomized controlled trial. BMC Public Health 2015; 15: 95.

  20. 20.

    , , , , , et al. Mobile-based intervention intended to stop obesity in pre-school children: the MINISTOP randomized controlled trial. Am J Clin Nutr 2017; 105: 1327–1335.

  21. 21.

    , , , , , et al. Physical activity intensity, sedentary behavior, body composition and physical fitness in 4-year-old children: results from the MINISTOP trial. Int J Obes 2016; 40: 1126–1133.

  22. 22.

    , , , , , et al. A mobile phone based method to assess energy and food intake in young children: a validation study against the doubly labelled water method and 24h dietary recalls. Nutrients 2016; 8(1). doi:10.3390/nu8010050.

  23. 23.

    , , , , . Total energy expenditure and energy intake in the pre-school child: a comparison. Br J Nutr 1994; 72: 13–20.

  24. 24.

    . New methods for calculating metabolic rate with special reference to protein metabolism. J Physiol 1949; 109: 1–9.

  25. 25.

    , , . Use of food quotients to predict respiratory quotients for the doubly-labelled water method of measuring energy expenditure. Hum Nutr Clin Nutr 1986; 40: 381–391.

  26. 26.

    , , , , , et al. Pediatric reference data for lean tissue properties: density and hydration from age 5 to 20 y. Am J Clin Nutr 2010; 91: 610–618.

  27. 27.

    The Nordic Council of Ministers Nordic Nutrient Recommendations 2012, Integrating Nutrition and Physical Activity, 5th edn. The Nordic Council of Ministers: Copenhagen, Denmark, 2012.

  28. 28.

    , , , , . A triaxial accelerometer and portable data processing unit for the assessment of daily physical activity. IEEE Trans Biomed Eng 1997; 44: 136–147.

  29. 29.

    , , , , . Actigraphic home-monitoring sleep-disturbed and control infants and young children: a new method for pediatric assessment of sleep–wake patterns. Pediatrics 1991; 87: 494–499.

  30. 30.

    , , . Activity-based sleep–wake identification: an empirical test of methodological issues. Sleep 1994; 17: 201–207.

  31. 31.

    , , , , , et al. Physical activity intensity, sedentary time, and body composition in preschoolers. Am J Clin Nutr 2013; 97: 1020–1028.

  32. 32.

    , , . Statistical Methods in Medical Research. Blackwell Science: Oxford, UK, 2002.

  33. 33.

    , , , , , et al. Predicting physical activity energy expenditure using accelerometry in adults from sub-Sahara Africa. Obesity 2009; 17: 1588–1595.

  34. 34.

    , . Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1: 307–310.

  35. 35.

    , , . Applied Regression Analysis and Other Multivariable Methods, 4th edn. Thomson: Belmont, CA, USA, 2008.

  36. 36.

    , , , , , . Estimating activity and sedentary behavior from an accelerometer on the hip or wrist. Med Sci Sports Exerc 2013; 45: 964–975.

  37. 37.

    , , , . Estimation of free-living energy expenditure using a novel activity monitor designed to minimize obtrusiveness. Obesity 2010; 18: 1845–1851.

  38. 38.

    , , , , , . Influence of body composition on physical activity validation studies using doubly labeled water. J Appl Physiol 2004; 96: 1357–1364.

  39. 39.

    , , , , , et al. Physical activity energy expenditure of adolescents in India. Obesity 2010; 18: 2212–2219.

  40. 40.

    Coward WA, Cole TJ. The double labeled water method for the measurement of energy expenditure in humans: risks and benefits. In: Whitehead RG, Prentice A (eds). New Techniques in Nutritional Research, 9th edn. Academic Press: San Diego, CA, USA, 1991; pp 139–176.

  41. 41.

    . Double Labelled Water. Theory and Practice, 1st edn. Chapman & Hall: London, UK, 1997.

  42. 42.

    Public Health Agency of SwedenOverweight and Obesity National Statistics, 2014. Available at: (last accessed September 2016).

  43. 43.

    , , , , , et al. Validity of 24-h recalls in (pre-)school aged children: comparison of proxy-reported energy intakes with measured energy expenditure. Clin Nutr 2014; 33: 79–84.

  44. 44.

    , , , . Swedish population-based longitudinal reference values from birth to 18 years of age for height, weight and head circumference. Acta Paediatr 2002; 91: 739–754.

  45. 45.

    Statistics Sweden. Educational Attainment of the Population, 2015. Available at: (last accessed September 2016).

  46. 46.

    , , , . Body composition of reference children from birth to age 10 years. Am J Clin Nutr 1982; 35 (Suppl), 1169–1175.

  47. 47.

    , . Energy costs of fat and protein deposition in the human infant. Am J Clin Nutr 1988; 48: 951–995.

  48. 48.

    , . Extended international (IOTF) body mass index cut-offs for thinness, overweight and obesity. Pediatr Obes 2012; 7: 284–294.

Download references


CDN was supported by the SNF Swedish Nutrition Foundation; ML by the Swedish Research Council (project no. 2012–2883), the Swedish Research Council for Health, Working Life and Welfare (2012-0906), Bo and Vera Axson Johnsons Foundation and Karolinska Institutet; and PH by Henning and Johan Throne-Holst Foundation.

Author information


  1. Department of Biosciences and Nutrition, NOVUM, Karolinska Institutet, Huddinge, Sweden

    • C Delisle Nyström
    • , P Henriksson
    • , F B Ortega
    •  & M Löf
  2. Marshfield Clinic Research Institute, Marshfield, WI, USA

    • J Pomeroy
  3. PROFITH PROmoting FITness and Health through physical activity Research Group, Department of Physical Education and Sport, Faculty of Sport Sciences, University of Granada, Granada, Spain

    • P Henriksson
    • , F B Ortega
    •  & J H Migueles
  4. Department of Clinical and Experimental Medicine, Linköping University, Linköping, Sweden

    • E Forsum
  5. School of Exercise and Nutrition Sciences, Deakin University

    • R Maddison


  1. Search for C Delisle Nyström in:

  2. Search for J Pomeroy in:

  3. Search for P Henriksson in:

  4. Search for E Forsum in:

  5. Search for F B Ortega in:

  6. Search for R Maddison in:

  7. Search for J H Migueles in:

  8. Search for M Löf in:

Competing interests

The authors declare no conflict of interest.

Corresponding author

Correspondence to C Delisle Nyström.

About this article

Publication history