A prospective study of nonoperative and operative management for Perthes' disease.

Two groups of children with Perthes' disease, randomized according to their source of referral, were treated by bed rest and skin traction, followed by either use of a weight-relieving caliper or a proximal femoral varus osteotomy. The outcome was similar in both groups and could be predicted more effectively by the arthrographic shape of the femoral head at presentation than by the Catterall grouping. The principal influences were the age of the child when treated and the sphericity of the femoral head. Walking speed at time of review was related to the height of the child and, at this age, was not affected by the shape of the healed femoral head.