Accelerometer-determined physical activity and its relationship with healthmethodology and application

Resumen

Accelerometers are the method of choice for the measurement of physical behaviours (i.e., physical activity [PA], sedentary behaviour [SB], and sleep) in current research. The rapid technological advances and the access to the accelerometers’ raw data addresses a series of challenges upon the need for trans-parent, comparable, and reproducible accelerometer data processing methods. The widely studied associations of physical behaviours with health outcomes should be expanded to understand the behaviours interplay in their relation-ship with health. Children with overweight or obesity might find in physical behaviours an effective tool to improve their cardiometabolic and brain health. Two main objectives are addressed in this Thesis: (i) to advance the current knowledge on accelerometer data collection and processing methods to study physical behaviours in children and adults with accelerometers; and (ii) to explore the associations of accelerometer-determined physical behaviours with cardiometabolic and brain health in children with overweight or obesity, as well as the effects of the ActiveBrains exercise randomized controlled trial The design of the studies included in this Thesis are a systematic review, a software description article, seven cross-sectional studies, a consensus statement article, and a randomized controlled trial. This Thesis encompasses data mainly from the ActiveBrains project, and complementary from the MINISTOP project and a pilot study on accelerometry. ActiGraph GT3X+ accelerometers attached to the right hip and wrists are used to quantify physical behaviours. Gold-standard measures of energy expenditure (i.e., doubly labelled-water), brain grey matter volume (i.e., magnetic resonance imaging), cardiometabolic health (i.e., blood biomarkers), and body composition (i.e., dual-energy x-ray absorptiometry) are included. Analytical approaches used include linear and quadratic regressions, analysis of variance (ANOVA), compositional data analysis, multivariate pattern analysis, and mediation models. In regards to the objective 1, this Thesis: (i) provides accelerometer data col-lection and processing recommendations based on existing literature; (ii) de-scribes an open-source software to process raw accelerometer data to quantify physical behaviours in which the PhD candidate is co-developer; (iii) finds that open-source acceleration metrics present a higher performance than proprietary activity counts to estimate energy expenditure; (iv) observes that open-source acceleration metrics are more comparable between them than with activity counts and provides cut-points to quantify PA intensity from dominant wrist-worn accelerometer data; (v) demonstrates large discrepancies in the time spent in SB and PA intensities when quantified from different cut-points, suggesting that it is not currently possible to know the prevalence of a population meeting the PA guidelines based on accelerometer data; (vi) proposes step-based metrics (including steps/day and various cadence-based intensity indicators) as a good proxy to some indicators of overall PA (i.e., counts per day, light-moderate-vigorous PA, moderate-to-vigorous PA) in children with overweight or obesity; (vii) provides a comprehensive description and international consensus on the analytical approaches most-frequently used in the field, and practical recommendations on what analytical approaches are the best-suited to a given research question. Relative to the objective 2, the current Thesis: (viii) observes that the association of PA and SB with grey matter volume in the hippocampus in children with overweight or obesity might be moderated by weight status (reallocating 20 min/day from SB to moderate-to-vigorous PA was associated with 100 mm3 more GMV in the right hippocampus in children with obesity type I); (ix) finds that sleep behaviours are associated with grey matter volume in several cortical and subcortical brain regions independently of SB and PA, and this seemed to be relevant for academic achievement in children with overweight or obesity; (x) remarks that a more stable and less fragmented activity-rest pattern (and earlier occurrence of PA) is associated with better academic achievement, executive function, and intelligence in children with overweight or obesity; (xi) demonstrates that a 20-week exercise program improves cardiometabolic health in children with overweight or obesity, while no effect is observed on mental health. The findings from this International Doctoral Thesis provide valuable recommendations on best-practice accelerometer data collection and processing techniques to measure physical behaviours, as well as consensus recommendations on analytical approaches for the field of PA epidemiology. Moreover, this Doctoral Thesis highlights the value of open-source data processing algorithms and the important role of PA, SB, sleep, and the activity-rest pattern in relation with brain health out-comes in children with overweight or obesity. Finally, this Doctoral has demonstrated that meaningful and positive changes in cardiometabolic health in children with overweight or obesity can be obtained with a 20-week exercise program, which should inform future health programs.