DiscussionThe results of this meta-analysis show that therapeuticexercise, in general, improves habitual gait speed incommunity-dwelling elderly people, although the relationshipbetween the 2 factors was not large. A positiveeffect of therapeutic exercise is consistent with Keysorand Jette’s19 review, which reported that exerciseimproved strength, flexibility, or endurance in 14 of 21studies of elderly people. As the fail-safe number was 109studies, we feel confident about this finding because it isunlikely that there are 109 additional unpublished studieswith nonsignificant findings. An overall change ingait speed between 0.01 to 0.02 m/s, however, is notclinically meaningful in the opinion of the authors ofthis review. Furthermore, it is not possible to determine,at this time, whether this finding was due to the variety ofinterventions, different gait speed measures used acrossstudies, or some other factor that is not apparent fromthe review.