Predicting physical activity change in men and women in two New England communities.

We describe the predictors of change in physical activity in a population-derived sample of residents of two New England communities. In 1986-1987 and again in 1990-1991, we assessed the self-reported physical activity of 1,081 survey respondents using a single question concerning the frequency of exercise-induced sweating. We measured and analyzed sociodemographic variables, cardiovascular disease risk factors, health knowledge and beliefs, social support variables, and exercise performance variables at baseline as potential predictors of physical activity change. We defined physical activity change in subjects as follows: maintainers (12.2%) persisted with > or = 3 times per week of activity, adopters (16.0%) increased activity to > or = 3 times per week, quitters (11.9%) decreased activity to < 3 times per week, and sedentary (59.9%) individuals persisted with < 3 times per week of activity. Using polychotomous multiple logistic regression analysis, we found that short-term success with exercise and weight loss, school children's recommendation of exercise, and work outside the home predicted adoption or maintenance of increased physical activity in women. For men, short-term success with exercise and weight loss and health beliefs regarding the role of exercise in preventing heart attacks and stroke were the major determinants of maintaining or adopting increased physical activity. This identification of the predictors of physical activity change provides the rationale for future sex-specific strategies aimed at increasing regular physical activity and long-term adherence to an active lifestyle.