Adequate physical activity (PA) is a critical component of chronic disease prevention and a healthy lifestyle. Unfortunately, studies suggest that US children do not meet the recommended 60 minutes of PA per day (11, 68). However, recent advances in measurement techniques are enabling researchers to gather more detailed objective PA data, allowing for an improved understanding of children’s PA accumulation and patterns. This information will enable researchers and policy makers to better design and evaluate interventions aimed at increasing PA, ultimately reducing the prevalence of chronic disease. These ongoing advances in objective PA monitoring devices call for studies to test and refine the methods by which PA data are processed and interpreted. Specifically, although these novel PA devices and methods (e.g., accelerometers and activity intensity classification methodologies) are being calibrated and validated using laboratory protocols, their accuracy in estimating children’s free-living PA has not been well-established. Additionally, given the well-established sporadic nature of children’s activity, it is critical to measure activity during very short time intervals (i.e., 1-2 second bouts), requiring devices that can record and store acceleration data at a relatively high-resolution (e.g. 30-100 Hz). Importantly, though many intervention studies have been conducted with the goal of increasing daily PA, none have used high frequency acceleration data to examine the accumulation of PA in a free-living setting, nor to evaluate the effectiveness of these PA interventions. However, the need to do so is widely recognized among the PA monitoring community (7, 21, 34). Therefore, the following dissertation describes a series of experiments with the overall aim of improving PA measurement tools and behaviors in children.

In the first study (Chapter 2), we attempt to establish cutpoints to distinguish between sedentary, light, moderate and vigorous activity using a novel wrist-mounted accelerometry device. We also examine the effects of various bout lengths (periods of consecutive seconds of activity above the moderate threshold) on the estimated MVPA accumulation. Moderately accurate cutpoints resulted (~70-75% accuracy). We also found very high estimates of daily MVPA (>300 minutes). Because of the high estimates of daily MVPA as well as the relative difficulty in distinguishing between light and moderate activity by the confusion matrix, we began to further investigate the effects of the specific processing methodologies we used.

This led us to the second study (Chapter 3), whereby we attempted to investigate the ability of three different processing methodologies to accurately detect MVPA. In this study, we applied three different processing methodologies (band pass filtered:​ BPEN, unfiltered:​ ENMO, and low pass filtered:​ LPENMO) to three separate independent samples of children:​ a calibration sample, a direct observation (classroom/recess) sample, and a multi-day, free-living sample. Results from this study suggested that BPEN is likely overestimating MVPA. ENMO and LPENMO both appeared to accurately detect MVPA compared to direct observation data (~85%). Because of these relatively good accuracies, and because low pass filtering is considered a best practice in signal processing, we elected to move forward with the low pass filtering methodology.

Once we had established a methodology that we felt accurately detected MVPA, we were able to process and analyze data from the IPLAY (Intervention of PhysicaL Activity in Youth) study. IPLAY is a large-scale, school-based intervention aimed at increasing activity through either curriculum intervention (SPARK), environmental intervention (renovated playgrounds), or the combination of the two (see Chapter 4 for a more detailed description of the intervention). Results revealed no differences in lunch recess, school day or full day MVPA between the groups. In addition, relatively high estimates of daily MVPA resulted (~140 minutes), as well as a lack of effect of BMI z-score on MVPA accumulation.

The combination of these studies adds a significant contribution to the literature around PA in children. Specifically, the investigation into processing methodologies demonstrates how critical this step is in being able to interpret acceleration data. It also provides a framework for other investigators to process acceleration data, with the goal of producing comparable results. The evaluation of the IPLAY study suggests the need for additional opportunities for children to be active during the day. The high estimates of daily MVPA suggest the need to further investigate how/when activity is being accumulated. Finally, an investigation into whether the PA guidelines ought to be re-established given novel methodologies for quantifying PA is warranted.