Mediators of change in physical activity following an intervention in primary care: PACE

Abstract

Background. The current study evaluated the construct validity of a physical activity intervention in primary care by testing whether the intervention changed hypothesized mediators and whether changes in the mediators were associated with behavior change. Methods. Matched physician offices were nonrandomly assigned to intervention or control. Apparently healthy, sedentary, adult patients (N = 255) were recruited from physician offices. The intervention was brief, behaviorally based counseling by physicians, plus a telephone follow-up 2 weeks later. Assessments of physical activity and mediators were collected at baseline and at 4- to 6-week follow- up. Hypothesized mediators were processes of change, self-efficacy, and social support for exercise. Results. Patients who were counseled improved significantly more than those in the control group on behavioral and cognitive processes of change. Other changes in mediators were nonsignificant. Behavioral processes of change and self-efficacy made significant contributions to the multiple regression model explaining self- report and objective measures of physical activity. Conclusions. The intervention affected some of the targeted mediators of physical activity change. Two of three mediator variables were associated with changes in physical activity regardless of experimental condition and other variables. The construct validity of the intervention was partially supported.