The studies presented in this thesis highlight important issues relating to the measurement of physical activity (PA) and sedentary behaviour (SB) in older adults, especially relating to the transfer of these methods to free-living environments.
Study One examined cut-points for ActiGraph GT3X+ (AG) for estimating time spent in SB in older adults in free-living environments. ActivPAL3™ (AP) was used as the reference standard. Movement patterns were measured in 41 older adults over seven consecutive days with both AG and AP. The results suggested that different cut-points must be used in older adults. In the case of older adults when using 1-minute epochs, for example, 25 counts/minute must be considered when using the AG vertical axis only, or 200 counts/minute when using the three axes of the accelerometer.
Study Two assessed the reliability and validity of a single question about sitting time on weekend and weekdays and a 24-hour recall (MARCA) for the last two days (n=41) that provides contextual information about use of time. The single question showed poor validity, but fair agreement for classifying individuals into tertiles. The MARCA showed good validity and moderate agreement for categorising participants into tertiles of time spent in SB. It is important to highlight that both instruments showed more accurate estimates for groups than for individuals.
Study Three assessed MET values of common daily activities in middle age and older adults in free-living environments and compared these with MET values listed in the Compendium of Physical Activities (CPA). Sixty participants completed a semi-structured protocol of sitting, lying, self-paced walking and four self-selected activities in their residences. 20 different activities were measured among the participants, observing a wide range of movement patterns. Measured METs were significantly different from values in the CPA for several activities (sitting, walking, sweeping, trimming, and laundry). It was concluded that using values from the CPA in older adults may result in overestimation of total daily energy expenditure, particularly when considering that most of the day is spent in sedentary activities.
Study Four assessed the validity of four different AG equations for predicting energy expenditure (EE) in older adults; and assessed agreement between physical activity levels obtained from these equations and from four sets of published AG equations, with those from indirect calorimetry. The equations (Freedson, Crouter (refined equation), and Santos-Lozano (vertical axis [VT] and vectormagnitude [VM])) were compared with measured EE. 40 participants completed the study and 16 different activities were assessed. It was concluded that when selecting an equation to estimate total daily EE from accelerometer data, equations with more accurate estimates of sedentary behaviour might be preferred. However, when the main interest is to estimate total time spent in different physical activity levels, a method with higher agreement for categorising activities may be preferred.
The work contained in this thesis will help researchers to 1) implement more accurate tools in the surveillance of PA and SB in larger studies, 2) better understand the strengths and limitations of current predictive methods when applying them in free-living environments, and 3) improve interpretation and translation of energy expenditure measures in older adults.