Sample size of 25 per group was determined by a power analysis conducted in advance of recruitment and was based on measures of physical function, assuming that improvements in physical function would be smaller and thus more difficult
to detect than changes in perceived well-being.