Interpretation of associations between the accelerometry physical activity spectrum and cardiometabolic health and locomotor skills in two cohorts of children using raw, normalized, log-transformed, or compositional data

ABSTRACT

It is discussed whether associations between accelerometer-derived physical activity intensities and outcomes should be analysed as absolute or relative data. The aim of the present study was to compare interpretation of association patterns of spectrum physical activity descriptions with outcome using raw, normalized, log-transformed, or compositional data. We used two datasets including 1) 841 schoolchildren and a cardiometabolic health outcome and 2) 1081 preschool children and a locomotor skill outcome. Accelerometry (ActiGraph GT3X+) data were described using multiple variables across the intensity spectrum. We varied the binning of variables to examine sensitivity of the compositional analyses to changes in the distribution centre. We used multivariate pattern analysis for all analyses and interpretations of data. Analyses of absolute (i.e., non-compositional) data showed weak associations for lower intensities and strongest associations with cardiometabolic health and locomotor skills for vigorous intensities. The same association patterns were partly observed for the compositional data, but association patterns were in some cases conflicting. The binning of variables had a major influence on associations for compositional data, but not for absolute data, meaning that conclusions depend on the operationalization of compositional data. These differences challenge and confuse interpretation of association patterns derived from the different approaches.

KEYWORDS:

• Multivariate pattern analysis
• compositional data analysis
• multicollinearity
• singularity
• log-transform
• log-ratio
• children
• accelerometer

Acknowledgments

We thank all children, parents and staff at the participating preschools (PRESPAS) and schools (ASK) for their excellent cooperation during the data collection. We also thank colleagues and students at the Western Norway University of Applied Sciences (formerly Sogn og Fjordane University College) for their contribution to the ASK and PRESPAS studies. We thank participants at the International Workshop: A focus on statistical methods to analyse accelerometer-measured physical activity, Granada, Spain, October 21.-22. 2019 for their valuable perspectives on best practices for analyzing physical activity data.

Availability of data and materials

The datasets used in the current study are available from the corresponding author on reasonable request.

Disclosure statement

The authors declare that they have no competing interests.

Supplementary material

Supplemental data for this article can be accessed online https://doi.org/10.1080/02640414.2020.1796462.

Additional information

Funding

The ASK study was funded by the Research Council of Norway (grant number 221047/F40) and the Gjensidige Foundation (grant number 1042294). The PRESPAS study was funded by the Sogn og Fjordane County Municipality. Jairo H. Migueles is supported by a Grant from the Spanish Ministry of Education, Culture and Sport (FPU15/02645). Additional funding was obtained from the SmarterMove project supported by the MINECO/FEDER (DEP2016‐79512‐R) and the University of Granada, Plan Propio de Investigación 2016, Excellence actions: Units of Excellence; Scientific Excellence Unit on Exercise and Health (UCEES). Junta de Andalucía, Consejería de Conocimiento, Investigación y Universidades and European Regional Development Funds (ref. SOMM17/6107/UGR). None of the funding agencies had any role in the study design, data collection, analyzing or interpreting data, or in writing the manuscripts;Ministerio de Educación, Cultura y Deporte [FPU15/02645];Norges Forskningsråd [221047/F40];MINECO/FEDER [DEP2016‐79512‐R];Gjensidige Foundation [1042294].