Predictors of physical activity change among adults using observational designs

Regular physical activity (PA) is foundational to human health, yet most people are inactive. A sound understanding of the determinants of PA may be instructive for building interventions and/or identifying critical target groups to promote PA. Most research on PA correlates has been biased by cross-sectional or passive prospective designs that fail to examine within-person analysis of PA change.

The purpose of this review was to collect and appraise the available literature on the predictors of PA change conceived broadly in terms of increases/decreases from baseline assessment as well as specifically in terms of adoption and maintenance.

Eligible studies were from English, peer-reviewed published articles that examined predictors of natural change of PA over 3 months + using observational (non-experimental) data in adult samples. Searches were performed from June 2012 to January 2014 in eight databases.

Sixty-seven independent data-sets, from 12 countries, primarily of medium quality/risk of bias, were identified with 26 correlates spanning demographic, behavioral, intra-individual, inter-individual, and environmental categories. Only intention and the onset of motherhood could reliably predict overall PA change. Among datasets configured to predict PA adoption, affective judgments and behavioral processes of change were the only reliable predictors, although both only have a small number of available studies. There were no reliable predictors of maintenance when compared to PA relapse.

The results underscore the importance of individual-level motivation and behavioral regulation in PA change, but also denote critical social variables. These findings, however, are constrained by PA measurement bias and limited studies that employed time-varying covariation between predictor variables and PA.