Determinants of physical activity maintenance: a systematic review and meta-analyses

Abstract

The aim of this study was to identify the psychosocial and socio-demographic determinants of physical activity maintenance (PAM) among adults by examining baseline differences between individuals who did and did not maintain physical activity participation over time (Part-I) and by examining how well combinations of psychosocial constructs and socio-demographic characteristics predict PAM (Part-II). Longitudinal and experimental studies (sample mean age 18–64 years) published between 1980 and 2010 were included. Independent data extraction was performed by two reviewers. All pooled effect sizes were calculated with the inverse-variance method under the random-effects model assumption. A total of 31 and 22 studies met all inclusion criteria and were included in Part-I and Part-II meta-analysis, respectively. Maintainers had higher [standard mean difference (95%CI)] self-efficacy (0.62 [0.49, 0.76]) and intention (0.65 [0.52, 0.79]) compared with relapsers. Health status directly affected PAM and moderated the relationship between BMI, education as well as perceived negative consequences and PAM. The overall model adjusted R ² [95%CI] for the prediction of PAM was 0.20 [0.14, 0.27]. The present findings suggest that beliefs about capabilities and motivation and goals are among the strongest variables associated with PAM. However, other variables rarely investigated so far (e.g., post-intentional constructs/self-regulatory processes) may also influence PAM.

Keywords:

• physical activity
• maintenance
• determinant
• systematic review
• meta-analysis

Acknowledgements

S.A. is supported by the Training Program in Obesity of the Merck Frosst-Canadian Institute of Health Research (CIHR) Research Chair on Obesity.

Notes

1. Although ‘six months’ is the period of time generally used in the physical activity domain to consider individuals as maintainers, it can be argued that a definition of PAM based solely on a specific period of time is arbitrary. Therefore, we included all studies meeting the specified inclusion criteria, irrespective of the length of the follow-up period.

2. Eight studies using the TTM also assessed the experiential and behavioural processes of change. These latter variables were considered as behaviour change techniques. Thus, they were not considered in the present review.

3. One study using the HAPA assessed maintenance self-efficacy. According to the HAPA [action/task] self-efficacy and maintenance self-efficacy are conceptually two different constructs. Most studies that assessed self-efficacy did it according to Marcus, Selby, Niaura, and Rossi's (1992) self-efficacy items, which is in accordance with the operationalisation of action/task self-efficacy. Thus, maintenance self-efficacy was not included in the pooled analysis for self-efficacy.