Stability and prediction of physical activity in 5-, 10-, and 28-year follow-up studies among industrial employees.

The aim of the study was to examine the stability of voluntary and household physical activity (PA) and to compare it with that of the use of the most common stimulants. The prospective cohort study comprised of follow-ups at 5, 10, and 28 years at baseline in 1973 in four plants of an industrial corporation in Finland. A systematic, non-proportional sample (n=902, age range 18-64 years) stratified for age, gender, and occupational status was drawn from the employees. Scores of PA were based on a questionnaire and interviews. Logistic regression models with proportional odds assumptions were counted. The 5-year stability (Spearman's rho) of PA time was 0.44 (PA intensity 0.44), the respective 10-year coefficient was 0.26 (0.32), and that in the 28-year follow-up was 0.18 (0.20). The stability of PA decreased rapidly from 1973 to 1983 and more slowly thereafter. Changes along the follow-up reflect a polarization of the distributions of PA within the sample. Age and an initially low level of activity were the strongest predictors of inactivity. The stability coefficient of smoking and alcohol consumption was twice as high as that of PA. Stimulant use was a greater factor in the individual's lifestyle than PA.