National policy for increasing leisure physical activity in the United States is impeded by a poor understanding of interventions that can be implemented by community and clinical medicine. To clarify the literature in this area, we conducted a quantitative, meta-analysis of 127 studies that examined the efficacy of interventions for increasing physical activity among similar to 131,000 subjects in community, worksite, school, home, and health care settings; 445 effects were expressed as a Pearson correlation coefficient (r) and examined as they varied according to moderating variables important for community and clinical intervention. The mean effect was moderately large, r = 0.34, approximating three-fourths of a standard deviation or an increase in binomial success rate from 50% to 67%. The estimated population effect weighted by sample size was larger, r = 0.75, approximating 2 standard deviations or increased success to 88%. Contrasts between levels of independent moderating variables indicated that effects weighted by sample size were larger when the interventions: 1) employed the principles of behavior modification, 2) used a mediated delivery, 3) targeted groups, 4) of combined ages, 5) sampled apparently healthy people, or 6) measured active leisure, of 7) low intensity, 8) by observation. Independently of sample size, effects were larger when interventions 1) used behavior modification, 2) employed a pre- or quasi-experimental design, or 3) were of short duration, regardless of features of the people, setting, or physical activity. Our results show that physical activity can be increased by intervention. The optimal ways for selecting intervention components, settings, and population segments to maintain increases in physical activity and the relative contributions by community and clinical medicine toward successful physical activity intervention require experimental confirmation, warranting accelerated attention in clinical trials.
