The uninvolved knee was not screened for radiographic arthritic changes but, if patients had an average knee pain greater than 4 out of 10 on a verbal analog scale or planned to have surgery on their uninvolved knee within a year, they were
excluded from the study.