We reviewed the medical records of 115 patients with 130 hips with developmental dysplasia with complete dislocation in the absence of a neuromuscular disorder, spontaneous reduction with a Pavlik harness, and a minimum of 14 years’ follow-up. The mean age at the time of harness application was 4.8 months (1 to 12) and the mean time spent in the harness was 6.1 months (3 to 12). A total of 108 hips (83.1%) were treated with the harness alone and supplementary surgery for residual acetabular dysplasia, as defined by an acetabular index > 30°, was performed in 22 hips (16.9%).

An overall satisfactory outcome (Severin grade I or II) was achieved in 119 hips (91.5%) at a mean follow-up of 16 years (14 to 32) with a follow-up rate of 75%. Avascular necrosis of the femoral head was noted in 16 hips (12.3%), seven of which (44%) underwent supplementary surgery and nine (56%) of which were classified as satisfactory. The acetabular index was the most reliable predictor of residual acetabular dysplasia.

The exact measurement of femoral head cover is essential for an assessment of reduction of congenital dislocation of the hip. We have compared standard anteroposterior radiographs with computerised tomograms and thereby classified the shape of the acetabular roof into four types. We found that the CE angle of Wiberg is a more reliable measure of head cover when the lateral point of bony condensation of the roof is chosen as the reference point rather than the edge of the bone, where these two points do not overlap. We conclude that head cover can be more accurately determined in younger children with dysplastic hips by our 'refined' CE angle, than by the original method of Wiberg.

Hip arthrography was performed in 19 patients in the initial stage of Perthes' disease. Sphericity and subluxation were measured and it was found that subluxation was independent of the femoral head deformity. We therefore tried to identify the cause of early subluxation: in seven patients a swollen ligamentum teres was thought to be responsible, and was associated with medial pooling of the contrast medium. A swollen ligamentum teres was seen in another seven cases; the other five arthrograms were normal. These findings were further clarified by enhanced CT scans, which confirmed that ligament swelling may be an important cause of early subluxation.