The impact of theory on the effectiveness of worksite physical activity interventions: a meta-analysis and meta-regression

Abstract

Background. Despite the potential importance of worksite physical activity interventions, reviews suggest there is currently a lack of clarity regarding their effectiveness. Aim. This meta-analysis assessed the effectiveness of worksite interventions designed to promote physical activity and investigate whether (1) interventions explicitly designed based on theory are more effective, and (2) inclusion of specific behaviour change techniques (BCTs) improves effectiveness. Methods. Worksite interventions with a primary aim of increasing physical activity were systematically reviewed. Designs were experimental or quasi-experimental and outcome measures were objective or validated self-report. Interventions were coded based on the extent to which theory/predictors were used to select/develop intervention techniques. A standardised theory-linked taxonomy of 26 BCTs was also used to code interventions. Effects of explicit use of theory, individual techniques and number of BCTs used were assessed using meta-analysis and meta-regression. Results. Twenty-six studies reporting 27 evaluations were included in the meta-analysis and a random effects model produced an overall effect size (d) of 0.21 (95% CI 0.17–0.26). Subgroup analysis indicated that interventions using theory more explicitly were more effective, producing an effect size of 0.34 (95% CI 0.23–0.45; I ²=0%). No significant differences in effect sizes were found between studies that had used individual BCTs and those that had not, and studies that used more techniques were not more effective. Conclusion. Overall, worksite physical activity interventions were effective, but only produced small sized effects on physical activity. Theory-based interventions were more effective.

Keywords:

• physical activity
• worksite
• intervention
• theory

Notes

1. The correction factor, J, yields an unbiased estimate of the pooled standard deviation using the following formula: J=1–3/4df–1. Following the correction, the same equation (described above) is used to compute Hedge's g as is to compute Cohen's d (Boresntein et al., 2009).

2. The classifications for Cohen's d effect sizes also apply to Hedge's g (Cohen, 1988, 1992; Kampenes et al., 2007).

3. Interventions that used self-reported duration and energy expenditure outcome measures also achieved average effect sizes of 0.21 and 0.23, respectively.