Constructing a graphical model of the relationship between physical activity and cognitive function based on a systematic review of prospective evidence
Elucidating the factors that contribute to healthy ageing is an important research goal. Physical activity (PA) has been associated with benefits for cognitive function (CF). However, most of this evidence comes from longitudinal cohort studies which, in the absence of experimental design, have limited scope to make causal inferences regarding observed relationships. This review aimed to utilise recent methodological developments allowing researchers to formulate and answer stronger causal questions using observational data, by following a best-practice method for synthesising evidence to produce a graphical causal model known as a directed acyclic graph (DAG). Following a search of 3 databases (EMBASE, MEDLINE and PsycINFO), 21 observational studies on the PA-CF relationship were reviewed and their methodological quality, characteristics, and key findings were summarised. The outcomes of interest were the covariates and modelling practices employed in each study. The reported covariates were synthesised against a set of criteria to determine their role in the DAG as confounders or mediators of the PA-CF relationship. Every included study had some areas of methodological weakness. The resulting DAG included a wide range of biopsychosocial covariates spanning the entire life-course and indicated potential intermediate pathways between PA and CF via structural brain health. Strengths, limitations and implications of this review for modelling decisions are discussed, prior to the model being taken forward to inform an empirical analysis using data from the UK Biobank cohort, separate from this review.