Table 3.

Summary table of literature exploring physical activity correlates and predictors among BCS

Demographics
Income	• (+) Cross-sectional. Higher income associated with increased PA [33]
Age	• (+) Longitudinal. Younger age associated with lower PA post diagnosis [46]
	• (/) Cross-sectional. Age not associated with meeting the guidelines [47]
	• (/) Intervention study. Age not associated with exercise adherence [48]
Education	• (/) Intervention study. Education did not predict exercise adherence [48]
Marital status	• (/) Intervention study. Marital status did not predict exercise adherence [48]
Health status
Co-morbidities	• (-) Cross-sectional. Higher co-morbidity associated with lower PA [47]
Weight	• (+) Longitudinal. Normal weight pre-diagnosis associated with less PA post-diagnosis [46]
	• (-) Cross-sectional. Higher BMI associated with reduced likelihood of exercising [47]
	• Cross-sectional. Lower sense of exercise self-efficacy among women who were overweight [49]
HRQL	• (+) Cross-sectional .Poorer HRQL was related to relapsing from active exercising to not exercising [50]
	• Longitudinal. HRQL (mental scale) significant predictor of rate of change of PA [51]
Fatigue	• (-) Longitudinal. Fatigue associated with lower PA at baseline but not associated with rate of change in PA [51].
Time since diagnosis	• (/) Intervention study. Time since diagnosis did not predict exercise adherence [48]
Stage of cancer	• (/) Intervention study. Stage of cancer did not predict exercise adherence [48]
Social cognitive
Self-efficacy	• (+)Cross-sectional. Self-efficacy association with positive exercise changes [49]
	• Cross-sectional. Self-efficacy correlated with current PA levels independent of pre-treatment PA levels [31].
	• (+) Cross-sectional. Task self-efficacy highly predictive for both PA and exercise in the overall sample and in the subgroup of younger women. Barrier self-efficacy followed the same trend [47]
	• (+) Intervention study. Baseline self-efficacy significant predictor of mean minutes of weekly exercise and of meeting weekly goals. [48].
Social support	• (+) Cross-sectional. Having an exercise partner or role model associated with increased PA [33]
	• (+) Longitudinal. Family support predicts change in PA behaviour [51]
	• (+) Cross-sectional. Perceived social support related to increases in PA after diagnosis, even up to five years later [52]
	• (/) Longitudinal. Social support of friend (not exercise specific) not a predictor of PA at baseline [51]
Intention	• (+) Cross-sectional. Intention significantly predicted PA behaviour [53]
	• Cross-sectional. Intention explained 35% of the variance in exercise adherence [54]
Personality	• (+) Cross-sectional. Neurotic breast cancer survivors more like to relapse [43]
	• (+) Intervention study. Extraversion related to increased exercise [44]
	• (+) Cross-sectional. Optimism related to reports of increased exercise frequency in the past 6 months, although the amount of variance accounted for was small [45]
Perceived control	• (/) Cross-sectional. General locus of control unrelated to improvements in survivors PA [55]
Outcome expectation	• (+) Cross-sectional. Outcome Expectations significant predictor of PA and exercise in [47]
	• (+) Mediation analysis. Positive beliefs about PA and cancer recurrence are related to increased PA levels [56]
Decisional balance	• (/) Cross-sectional. Decisional balance did not predict exercise adherence [57]
Physical activity behaviour
Pre-diagnosis PA level	• (-) Longitudinal. Women reporting more PA pre diagnosis had lower levels of PA post diagnosis [46]
	• (+) Cross-sectional. Prior exercise was a significant positive predictor of overall PA [47]
	• Cross-sectional. Direct association with Pre-treatment PA level and current PA level [31]
Baseline PA level	• (+) Intervention study. Baseline PA a significant predictor of mean minutes of weekly exercise [48]