We analysed the incidence of avascular necrosis in 101 hips of 90 infants with congenital dislocation treated with the Pavlik harness and followed up for more than one year. Using ultrasonography in the flexed-abducted position the hips were classified as type A when the femoral head was displaced posteriorly, but within the socket and making contact with the posterior inner wall of the acetabulum; type B when it was in contact with the posterior margin of the acetabulum, with its centre at this level or anterior to it; and type C when it was displaced out of the socket, with its centre posterior to the posterior rim of the acetabulum.

Eighty-seven hips were reduced by the harness (86%), and seven of these developed avascular necrosis (8%). All 69 hips with type-A dislocation were reduced, and only one (1.4%) showed slight avascular necrosis. Eighteen (78.3%) of 23 hips with type-B dislocation were reduced, and six developed avascular necrosis (33.3%). In one hip, the femoral head was severely damaged. None of the nine hips with type-C dislocation was reduced in the harness.

We conclude that the Pavlik harness is indicated for type-A, but not for type-B or type-C dislocations.

We describe a new technique for examining the infant hip using ultrasound. Both hips are imaged simultaneously via an anterior approach. The examination can be done with the hip either extended or flexed and abducted. The method has three advantages: 1) since both hips are imaged simultaneously, lines can be drawn to assist in determining the relationship between the femoral head and the pelvis; 2) proximal, anteroposterior and lateral displacement of the femoral head can all be demonstrated; 3) the method is applicable to the infant in a harness or a plaster cast to demonstrate maintenance of reduction of a dislocated hip. The usual direction of dislocation of the femoral head was anterior and lateral. Proximal migration was also observed in cases with more severe dislocation. In flexion, the dislocated head of the femur often moved posterior to the acetabulum. Of 1276 hips, in 638 infants aged from three weeks to one year, 49 showed congenital dislocation. The accuracy of our anterior method of sonography in diagnosing congenital dislocation of the hip compared well with the method of Graf and with radiography.

We used ultrasonography to examine 36 children suffering from transient synovitis and 12 children with early Perthes' disease. Widening of the joint space was revealed by ultrasonography in all affected hips with either disease. In the patients with transient synovitis, capsular distension was attributed to synovial effusion, while in the patients with Perthes' disease it was produced by thickening of the synovial membrane. Neither capsular distension nor thickening of the joint cartilage was seen in the contralateral normal hip in the patients with transient synovitis, but they were common in early Perthes' disease. Ultrasonography may provide significant diagnostic clues to differentiate early Perthes' from transient synovitis.