B3443 - Physical activity and mental health among adolescents and young adults A novel approach using multivariate pattern analysis - 17/01/2020

B number: 
B3443
Principal applicant name: 
Matteo Christian Sattler | NIHR Bristol Biomedical Research Centre - Nutrition Theme, Bristol Dental School, University of Bristol (United Kingdom)
Co-applicants: 
Nicola Wiles, PhD, Sam Leary, PhD, Andy Ness, PhD, Eivind Aadland, PhD, Olav Martin Kvalheim, PhD, Mireille van Poppel, PhD, Rodrigo Antunes Lima, PhD, Lars Bo Andersen, PhD
Title of project: 
Physical activity and mental health among adolescents and young adults: A novel approach using multivariate pattern analysis
Proposal summary: 

The benefits of physical activity (PA) for mental health are well established. However, little longitudinal evidence is available in adolescents regarding the association between PA and anxiety and depressive symptoms. Especially problematic is our lack of knowledge about the intensity of PA that is needed to prevent or treat clinically relevant symptoms in this critical period of life. This lack of knowledge can be strongly attributed to the way we currently analyze the intensity information from accelerometry. Data are ‘simplified’ and collapsed into only a few intensities (e.g. light, moderate-to-vigorous), which are then included in statistical models. This causes substantial loss of information and challenges the detection of the relative importance of specific intensities. However, using a larger number of intensities is not possible since traditional models (i.e. multiple linear regression) cannot handle their closed structure and multicollinearity. A novel approach is multivariate pattern analysis (MPA) (specifically: partial least squares regression [PLSR]) which was introduced to PA research by Aadland et al. in 2018. MPA overcomes these shortcomings and can be used to describe the PA spectrum with many intensities (e.g. > 20 variables) while determining most important ones. This study uses MPA to investigate the association between the entire PA intensity spectrum at age 14 years and future anxiety and depressive symptoms throughout adolescence and young adulthood.

Ref: Aadland E, Kvalheim OM, Anderssen SA, Resaland GK, Andersen LB. The multivariate physical activity signature associated with metabolic health in children. Int J Behav Nutr Phys Act. 2018;15(1):77.

Impact of research: 
Accelerometers capture detailed but complex PA information. In practice, researchers only use a fraction of the available information (e.g. time spent in moderate-to-vigorous physical activity [MVPA]) to establish associations with health outcomes. MPA allows us to use the entire intensity spectrum of PA which results in a more accurate description of PA behaviors. MPA will help us to use the intensity information from accelerometry more appropriately and to derive more reliable conclusions about how different PA intensities relate to mental health. Regarding anxiety and depressive symptoms, there is a lack of evidence for the most relevant intensities. Thus, this will be the first study worldwide applying MPA in a longitudinal setting to identify PA intensities most strongly associated with future anxiety and depressive symptoms in adolescents and young adults. Both PA guidelines for the general population and clinical practice would strongly benefit from this knowledge.
Date proposal received: 
Wednesday, 15 January, 2020
Date proposal approved: 
Friday, 17 January, 2020
Keywords: 
Epidemiology, Behaviour - e.g. antisocial behaviour, risk behaviour, etc., Mental health, Statistical methods, Physical activity, Epidemiology, Mental health, Statistical analysis