Background. Data on time spent in physical activity, sedentary behavior and sleep during a day is compositional in nature, i.e. they add up to a constant value, typically 100 %time. Compositional data have fundamentally different properties from unconstrained data, and require other processing and analysis procedures, referred to as compositional data analysis (CoDA). Most studies of physical activities, however, still apply analysis procedures adapted to standard data, which can lead to misleading results. The present study describes the CoDA approach when comparing time spent sedentary and in physical activity between age groups and genders, and investigates the extent to which results obtained by CoDA differ from those obtained using standard analysis procedures.
Methods. Proportions of time spent sedentary (sitting/lying), standing, and in physical activity (walking/running/stair climbing/cycling) during work and leisure was determined for 1-4 days in each of 692 blue-collar workers using accelerometry. Data were analyzed for differences between genders and age groups using MANOVA, following either a standard approach or a CoDA approach based on isometric log-ratio transformed data.
Results. When comparing genders at work, the effect size obtained using standard analysis (F=15.89, P<0.001) was 16% smaller than that obtained with CoDA (F=18.45, P<0.001), although both approaches suggested a statistically significant difference. When comparing age groups, CoDA resulted in a 61% larger, and significant, effect size (F=4.21, P=0.02) than that obtained with the standard approach (F=2.62, P=0.07). During leisure, results with standard (age; F=2.44, P=0.09; gender; F=18.36, P<0.001) and CoDA (age; F=2.37, P=0.09; gender; F=18.19, P<0.001) analysis were similar.
Conclusion. Results differ between CoDA and standard analysis when investigating age and gender-based group differences in time spent sedentary and in physical activity at work. We recommend future studies to use CoDA, which can correctly account for the compositional nature of data describing physical activity and sedentary behavior.