The medial approach for the treatment of children with developmental dysplasia of the hip (DDH) in whom closed reduction has failed requires minimal access with negligible blood loss. In the United Kingdom, there is a preference for these children to be treated using an anterolateral approach after the appearance of the ossific nucleus. In this study we compared these two protocols, primarily for the risk of osteonecrosis.

Data were gathered prospectively for protocols involving the medial approach (26 hips in 22 children) and the anterolateral approach (22 hips in 21 children) in children aged <  24 months at the time of surgery. Osteonecrosis of the femoral head was assessed with validated scores. The acetabular index (AI) and centre–edge angle (CEA) were also measured.

The mean age of the children at the time of surgery was 11 months (3 to 24) for the medial approach group and 18 months (12 to 24) for the anterolateral group, and the combined mean follow-up was 70 months (26 to 228). Osteonecrosis of the femoral head was evident or asphericity predicted in three of 26 hips (12%) in the medial approach group and four of 22 (18%) in the anterolateral group (p = 0.52). The mean improvement in AI was 8.8° (4° to 12°) and 7.9° (6° to 10°), respectively, at two years post-operatively (p = 0.18). There was no significant difference in CEA values of affected hips between the two groups.

Children treated using an early medial approach did not have a higher risk of developing osteonecrosis at early to mid-term follow-up than those treated using a delayed anterolateral approach. The rates of acetabular remodelling were similar for both protocols.

Cite this article: Bone Joint J 2014;96-B:406–13.

We present the results of treatment of developmental dysplasia of the hip in infancy with the Pavlik harness using a United Kingdom screening programme with ultrasound-guided supervision. Initially, 128 consecutive hips in 77 patients were reviewed over a 40-month period; 123 of these were finally included in the study. The mean age of the patients at the start of treatment was five weeks (1 to 12). All hips were examined clinically and monitored with ultrasound scanning. Failure of treatment was defined as an inability to maintain reduction with the harness. All hips diagnosed with dysplasia or subluxation but not dislocation were managed successfully in the harness. There were 43 dislocated hips, of which 39 were reducible, but six failed treatment in the harness. There were four dislocated but irreducible hips which all failed treatment in the harness. One hip appeared to be successfully treated in the harness but showed persistent radiological dysplasia at 12 and 24 months. Grade 1 avascular necrosis was identified radiologically in three patients at 12 months.

Reimers’ hip migration percentage is commonly used to document the extent of subluxation of the hip in children with spasticity. In this study, two measurers, with six months paediatric orthopaedic experience, measured the migration percentage on 44 pelvic radiographs of children with cerebral palsy, aged between two and eight years. Unknown to the measurers, each radiograph was duplicated, giving 22 non-identical radiographs (44 hips) which were measured twice at time 0 and twice six weeks later. The intra-measurer, intra-sessional absolute differences between the first and second measurements ranged from 0% to 23%, with median values of 2.5% to 3.6%. The intra-sessional median absolute differences were not statistically different between the two measurers and measuring sessions (p = 0.42, Kruskal-Wallis test). The inter-sessional absolute differences for measurements made by the same measurers ranged from 0% to 18% with a median absolute difference of 1.7% to 3.2%. Overall, only 5% of the intra-measurer measurement differences, within and between sessions, were above 13%. Repeated measurements by one measurer over time must, therefore, vary by more than 13% in order to be 95% confident of a true change. The inter-measurer error was higher with median absolute differences between the two measurers’ measurements of the same hip of 3.25% to 5% (0% to 26%) and a 95^th upper confidence interval of 21% to 23%. Averaging the four separate measurements over the two sessions reduced the inter-measurer error to a median absolute difference of 2.8%, but did not improve the 95^th upper confidence interval, which measured 22.4%.

Such inter-measurer errors may be clinically unacceptable.