Research Article: Associations of device-measured physical activity across adolescence with metabolic traits: Prospective cohort study

Date Published: September 11, 2018

Publisher: Public Library of Science

Author(s): Joshua A. Bell, Mark Hamer, Rebecca C. Richmond, Nicholas J. Timpson, David Carslake, George Davey Smith, Sanjay Basu

Abstract: BackgroundMultiple occasions of device-measured physical activity have not been previously examined in relation to metabolic traits. We described associations of total activity, moderate-to-vigorous physical activity (MVPA), and sedentary time from three accelerometry measures taken across adolescence with detailed traits related to systemic metabolism.Methods and findingsThere were 1,826 male and female participants recruited at birth in 1991–1992 via mothers into the Avon Longitudinal Study of Parents and Children offspring cohort who attended clinics in 2003–2005, 2005–2006, and 2006–2008 who were included in ≥1 analysis. Waist-worn uniaxial accelerometers measured total activity (counts/min), MVPA (min/d), and sedentary time (min/d) over ≥3 d at mean age 12y, 14y, and 15y. Current activity (at age 15y), mean activity across occasions, interaction by previous activity, and change in activity were examined in relation to systolic and diastolic blood pressure, insulin, C-reactive protein, and 230 traits from targeted metabolomics (nuclear magnetic resonance spectroscopy), including lipoprotein cholesterol and triglycerides, amino and fatty acids, glycoprotein acetyls, and others, at age 15y. Mean current total activity was 477.5 counts/min (SD = 164.0) while mean MVPA and sedentary time durations were 23.6 min/d (SD = 17.9) and 522.1 min/d (SD = 66.0), respectively. Mean body mass index at age 15y was 21.4 kg/m2 (SD = 3.5). Correlations between first and last activity measurement occasions were low (e.g., r = 0.40 for counts/min). Current activity was most strongly associated with cholesterol and triglycerides in high-density lipoprotein (HDL) and very low-density lipoprotein (VLDL) particles (e.g., −0.002 mmol/l or −0.18 SD units; 95% CI −0.24–−0.11 for triglycerides in chylomicrons and extremely large very low-density lipoprotein [XL VLDL]) and with glycoprotein acetyls (−0.02 mmol/l or −0.16 SD units; 95% CI −0.22–−0.10), among others. Associations were similar for mean activity across 3 occasions. Attenuations were modest with adjustment for fat mass index based on dual-energy X-ray absorptiometry (DXA). In mutually adjusted models, higher MVPA and sedentary time were oppositely associated with cholesterol and triglycerides in VLDL and HDL particles (MVPA more strongly with glycoprotein acetyls and sedentary time more strongly with amino acids). Associations appeared less consistent for sedentary time than for MVPA based on longer-term measures and were weak for change in all activity types from age 12y–15y. Evidence was also weak for interaction between activity types at age 15y and previous activity measures in relation to most traits (minimum P = 0.003; median P = 0.26 for counts/min) with interaction coefficients mostly positive. Study limitations include modest sample sizes and relatively short durations of accelerometry measurement on each occasion (3–7 d) and of time lengths between first and last accelerometry occasions (<4 years), which can obscure patterns from chance variation and limit description of activity trajectories. Activity was also recorded using uniaxial accelerometers which predated more sensitive triaxial devices.ConclusionsOur results support associations of physical activity with metabolic traits that are small in magnitude and more robust for higher MVPA than lower sedentary time. Activity fluctuates over time, but associations of current activity with most metabolic traits do not differ by previous activity. This suggests that the metabolic effects of physical activity, if causal, depend on most recent engagement.

Partial Text: Cardiometabolic diseases contribute the most to non-violent death rates worldwide [1]. Those chiefly responsible for this contribution are type 2 diabetes, coronary heart disease (CHD), and stroke [1], all of which have origins in dysfunctional metabolism [2]. Decades of observational studies have established dose–response associations of higher duration and intensity of physical activity with reduced risk of cardiometabolic diseases and their precursors [3–5]. Physical activity is now commonly measured using accelerometry devices which track intensity and duration of body movement—an advance over self-report methods, which are prone to measurement error from imprecise and biased recall and can thereby mask the true magnitude of associations [6,7]. Accelerometer data are often collected over several days to estimate mean activity level. Such measures, however, still only describe a single measurement occasion in the lifespan of a cohort and may only provide a snapshot of longer-term activity.

A prospective analysis plan written in March 2017 before analyses were conducted is included as S1 Protocol. Analyses were not changed following peer review. This study is reported according to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines (checklist included as S1 STROBE Checklist).

This study aimed to better describe associations of physical activity and sedentary time with metabolic traits by integrating repeat accelerometry data taken several years apart with data from targeted metabolomics. At age 15y, total activity was associated with a diverse set of metabolic traits. Associations were strongest between higher-intensity activity and lower cholesterol in VLDL lipoprotein particles, higher cholesterol in HDL particles, lower triglyceride in both HDL and VLDL particles, and lower inflammatory glycoprotein acetyls, although association magnitude was generally small. Associations appeared more robust for higher MVPA than for lower sedentary time and were weak in relation to change in all activity types over several years. Activity fluctuates over time, but associations of current activity with most metabolic traits did not differ by previous activity. This suggests that the metabolic effects of physical activity, if causal, depend on most recent engagement.



Leave a Reply

Your email address will not be published.