Association between physical activity levels in mid-life with physical activity in old age: a 20-year tracking study in a prospective cohort

OBJECTIVES This study aims to examine the tracking and predictability of physical activity in old age from overall physical activity and participation in sport, recreational activity and walking in mid-life. DESIGN Prospective population-based cohort study. SETTING British Regional Heart Study participants recruited from primary care centres in the UK in 1978-1980. PARTICIPANTS AND OUTCOME MEASURES Men (n=3413) self-reported their physical activity at baseline, 12, 16 and 20-year follow-ups and were categorised as inactive or active and having high or low participation in sport, walking and recreational activities. Tracking was assessed using kappa statistics and random effects models. Logistic regression estimated the odds of being active at 20-year follow-up according to physical activity participation in mid-life. RESULTS Among 3413 men (mean age at baseline 48.6±5.4 years) with complete data, tracking of overall physical activity was moderate (kappa: 0.23-0.26). Tracking was higher for sports participation (kappa: 0.35-0.38) compared with recreational activity (kappa: 0.16-0.24) and walking (kappa: 0.11-0.15). Intraclass correlation coefficients demonstrated similar levels of stability and only marginally weakened after controlling for covariates. Compared with inactive men, being active at baseline was associated with greater odds of being active at 20-year follow-up (OR 2.7, 95% CI 2.4 to 3.2) after adjusting for sociodemographic, health and lifestyle variables. Playing sport in mid-life was more strongly associated with being active at 20-year follow-up than other domains, particularly when sport participation began earlier in life. CONCLUSION Being physically active in mid-life increases the odds of being active in old age. Promoting physical activity in later life might be best achieved by promoting sport participation earlier in the life course.