Ignoring regression to the mean leads to misleading interpretation about muscle strength responsiveness in obese elderly women