Doctor of Philosophy
School of Education
Janssen, Xanne C.J, The objective measurement of physical activity and sedentary behaviour in preschool-aged children, Doctor of Philosophy thesis, School of Education, University of Wollongong, 2013. https://ro.uow.edu.au/theses/3988
Physical activity and sedentary behaviour have been associated with several health outcomes among adults and school-aged children. In addition, the evidence base linking physical activity and sedentary behaviour to health outcomes in preschool-aged children is increasing. Yet, a major issue in this field of research is the accuracy of methods used to measure physical activity and sedentary behaviour in 4- to 6-year-old children.
Currently, hip-mounted accelerometers such as the ActiGraph and Actical are the most common devices used to objectively measure free-living physical activity and sedentary behaviour in preschool-aged children. Several calibration studies have been conducted to develop equations and cut-points to predict energy expenditure and classify physical activity intensity or sedentary behaviour in young children. However, the resulting equations and cut-points differ substantially among calibration studies. As a result, researchers are challenged with the decision of which equation and/or cut-point to use to estimate energy expenditure or time spent physically active. The choice of an accurate equation or cut-point is of great importance as the use of different cut-points results in significant differences in the reported outcomes such as time spent in different physical activity intensities. Another problem with hip-mounted accelerometers is their inability to differentiate between sitting and standing. This is of importance when examining sedentary behaviour which is defined as any activity with an energy expenditure value of less than or equal to 1.5 times resting metabolic rate while in a sitting or reclining position. More recently a thigh-mounted accelerometer (the activPALTM) has been developed to measure posture and postural changes. However, the validity of the activPALTM for examining sedentary behaviour in preschool-aged children remains unclear.
To overcome aforementioned issues, methodological studies comparing the accuracy of current accelerometer equations and cut-points against appropriate criterion measures in preschool-aged children are needed. The outcomes of such studies might improve the consistency of physical activity and sedentary behaviour measurement in this age group. Therefore, the primary aim of this thesis was to examine the validity of several accelerometer cut-points and equations developed to classify physical activity intensity and sedentary behaviour, and to estimate energy expenditure in preschool-aged children. Additionally, the validity of the activPALTM for assessing sedentary behaviour, physical activity and energy expenditure was investigated. The aims were addressed by simultaneously examining current ActiGraph and Actical energy expenditure equations and cut-points against energy expenditure measured by room calorimetry and physical activity intensity classified by direct observation. In addition, the validity of the activPALTM for classifying sedentary behaviour and moderate- to vigorous-intensity physical activity, and estimating energy expenditure in preschool-aged children was examined against these criterion methods. The studies in this thesis will make a unique contribution to the evidence related to the objective measurement of physical activity and sedentary behaviour in preschool-aged children.
The first study described in this thesis explored the feasibility of the use of a room calorimeter in preschool-aged children. In total, 40 children, aged 4- to 6-years, were recruited to the study and asked to follow a graded activity protocol while in a room calorimeter. In addition, six children participated in two additional resting protocols to examine the effect of a light standardised breakfast followed by a 90-minute fast on resting energy expenditure measures. The results of this study supported the use of room calorimetery in preschool-aged children. In addition, the effect of a light standardised breakfast followed by a 90-minute fast on resting energy expenditure appeared to be small. Nevertheless, measured resting energy expenditue values were slightly higher than predicted resting energy expenditure using the Schofield equation. The biological importance of this difference is unclear. Therefore, the use of predicted resting energy expenditure was recommended for future studies where an overnight fast is not feasible.
The studies reported in Chapter 4 and Chapter 5 examined the predictive validity of ActiGraph and Actical energy expenditure equations, respectively. In addition, the classification accuracy of current accelerometer cut-points was tested. For these studies, 40 children completed a 150-minute activity protocol involving age-appropriate sedentary, light- and moderate- to vigorous-intensity physical activities. Energy expenditure measured by room calorimetry and physical activity intensity classified by direct observation were used as criterion measures. The results of these studies showed all Actical equations and the ActiGraph equation developed by Pate et al. performed reasonably well when predicting energy expenditure during moderate- to vigorousintensity physical activity. However, none of the equations estimated energy expenditure accurately over the complete range of intensities. For the Actical, the classification accuracy was highest when using cut-points of ≤ 6 counts per 15 seconds, between 7-286 counts per 15 seconds, and ≥ 287 counts per 15 seconds for sedentary behaviour, light-intensity physical activity and moderate- to vigorous-intensity physical activity, respectively. For the ActiGraph, using cut-points ≤ 25 counts per 15 seconds, between 26-419 counts per 15 seconds, and ≥ 420 counts per 15 seconds resulted in the highest accuracy for classifying sedentary behaviour, light-intensity physical activity and moderate- to vigorous-intensity physical activity, respectively, in preschool-aged children.
In Chapter 6, the validity of the activPALTM for assessing postural allocation was examined using direct observation as the criterion measure. The activPALTM showed good (ROC-AUC = 0.84) classification accuracy for sitting/lying (i.e. sedentary behaviour), whereas classification accuracy for standing and walking was found to be fair (ROC-AUC = 0.76 and 0.73, respectively). Nevertheless, time spent sitting/lying and standing was overestimated (mean difference = 5.9% and 14.8%, respectively) and time spent walking was underestimated (mean difference = 10.0%). In addition, the total number of breaks in sedentary behaviour was overestimated (mean difference = 35). One of the factors influencing the accuracy of the activPALTM appeared to be the classification of activities defined as ‗other‘, such as kneeling down on one knee. It was recommended future studies should examine the importance of classifying these postures before conclusions around the validity of the activPALTM in preschool-aged children can be made.
The study in Chapter 7 examined the predictive validity of the energy expenditure equation embedded in the activPALTM software. Additionally, an activPALTM cut-point for the classification of moderate- to vigorous-intensity physical activity was developed and validated. The activPALTM energy expenditure equation overestimated energy expenditure during sedentary behaviour whereas energy expenditure during light- and moderate- to vigorous-intensity physical activities was underestimated. Further development of this prediction equation is needed before it can be used with confidence in studies examining energy expenditure among preschool-aged children. For the classification of moderate- to vigorous-intensity physical activity, an activPALTM cutpoint of ≥ 1418 counts per 15 seconds was established with excellent classification accuracy. In addition, validation of this cut-point in a separate sample resulted in good classification accuracy for moderate- to vigorous-intensity physical activity (ROC-AUC = 0.88). However, studies validating this cut-point under free-living conditions are needed to confirm its accuracy for assessing habitual moderate- to vigorous-intensity physical activity in preschool-aged children.
To conclude, this thesis aimed to provide evidence to better understand the accuracy of methods used to objectively measure physical activity, sedentary behaviour and estimate energy expenditure in preschool-aged children. If implemented consistently the recommendations from the reported studies could improve measurement accuracy and comparability between studies in this age group. The decision as to which methodology to use depends on the research question under investigation. Using the accelerometer physical activity cut-points identified as most accurate in this thesis might be the best available approach when examining physical activity intensities and/or adherence to current guidelines. However, if sedentary behaviour is the focus of the study, using the activPALTM accelerometer might be an appropriate alternative to using physical activity intensity cut-points. Finally, approaches more refined than single linear regression equations might be needed to accurately predict free-living energy expenditure from accelerometer data in preschool-aged children.