Abstract:
Accurate assessment of physical activity (PA) is important to study the associations
between PA and health outcomes, to evaluate the effectiveness of interventions and to
derive public health recommendations. Despite limitations, accelerometry-based
methods generate the best available measures for epidemiological research involving
large number of children and adults. In this chapter, we review the most important
methodological issues pertaining to the use of accelerometers to assess the overall
volume of PA. We stress the importance of recording and keeping the raw data
whenever possible. We review the validation studies using accelerometry to
determine energy expenditure and calibration studies that attempt to derive thresholds
(“cut-offs”) for differentiating between activity intensity categories. Conceptual and
measurement issues due to the use of different thresholds are reviewed, as well as the
temporal resolution issues such as sampling rate and epoch length. Different wear
time detection algorithms and inclusion criteria are reviewed as well as options in
data reduction (deriving meaningful variables from accelerometer data). We present
an R package automatizing most of the steps in accelerometer data analysis. The
chapter concludes with some insights into the future of accelerometry given the
wearable revolution, and logistical considerations in using accelerometers in large
field studies.