In the first few months of life, a notch may be seen on radiographs of the superolateral margin of the infant hip. It may be associated with a steeply inclined acetabular roof and may be an indicator of persistent neonatal instability or displacement of the femoral head.

The epidemiological data and intra-operative findings from 260 consecutive arthroscopically-diagnosed acetabular labral tears seen over a ten-year period were analysed. Radiographs of 128 of these patients were assessed for dysplasia using established radiological parameters. Patients with acetabular dysplasia were then compared against those without in order to identify any differences in gender, age, the side of the tear, the pattern of the tear, the number of quadrants involved, the quadrant preference and the prevalence of intra-articular comorbidity. Dysplasia was found in 46% (59 of 128) of the hips. No significant differences existed between the dysplastic and non-dysplastic subgroups as regards gender distribution, the side of the lesion, the number of quadrants involved or the distribution of tears among the quadrants. However, tears associated with dysplasia were diagnosed in an older age group, had a different pattern and were associated with a much higher prevalence of osteoarthritis.

We studied the reproducibility of ultrasonographic screening examination of the hip when read by diagnostic radiographers. In order to determine interobserver variability, 200 ultrasonograms were classified according to Graf’s method by five observers (four radiographers and one radiologist). The kappa values for interobserver variability indicated moderate agreement (kappa 0.47) for the exact Graf classification and substantial agreement (kappa 0.65) for the classification of normal (type I) versus abnormal (type IIa-IV). Agreement was significantly different for normal, immature and abnormal hips. Comparison of the findings in our interobserver study with existing information based on other examinations and treatment revealed that only a small number of infants with mildly dysplastic hips would have been typed as normal by some observers as a result of observer variability. In conclusion, the interobserver agreement on the ultrasound assessment of the hip was good enough for screening purposes. Observer variability did not result in any severe cases being missed.

Ultrasonography of the hip was performed sequentially by two different examiners in 75 infants. The ultrasound strips were reviewed twice by three paediatric orthopaedic surgeons and classified by the Graf method. The intraobserver and interobserver agreement between the interpretations was analysed using simple and weighted kappa coefficients calculated for agreement on the Graf classification and for grouping as normal (types 1A to 2A), and abnormal requiring treatment (types 2B to 4).

When examining the same ultrasound strip, intraobserver agreement for the Graf classification was substantial (mean kappa 0.61), but interobserver agreement was only moderate (kappa 0.50).

For the grouping into normal and abnormal, the mean kappa value for intraobserver agreement was 0.67 and for interobserver agreement 0.57. Because of the significant differences in agreement between normal and abnormal hips, we analysed a subgroup of those with at least one abnormal interpretation. Intraobserver agreement within this subgroup showed moderate reliability (kappa 0.41), but interobserver agreement was only fair (kappa 0.28).

Interpretations of two different strips performed sequentially showed significantly lower agreement with an intraobserver kappa value of 0.29 and an interobserver value of 0.28. In the subgroup with at least one abnormal reading, the intraobserver kappa was 0.09 and the interobserver 0.1.

Our findings suggest that both the technique of performing ultrasonography and the interpretation of the image may influence the result.

Ten children who had clinically stable hips at birth were radiographed at one month because they had factors predisposing to hip dislocation. In all cases one or both hips gave rise to a suspicion of dysplasia, though clinical abnormalities were slow to appear. Four of these hips subsequently dislocated. We believe that infants with late presentation of acetabular dysplasia and clinical abnormality belong to a different aetiological group from those with neonatal instability due to ligamentous laxity. The significance of this differentiation is that some infants presenting late have only recently dislocated, and the diagnosis has not necessarily been "missed" at neonatal examination.

Failure of normal acetabular development is inevitable in congenital dislocation of the hip when it is unrecognised until late infancy or early childhood. Numerous stabilising procedures have been described, one of the first being the shelf operation or acetabuloplasty. This procedure fell into disrepute because the shelf was often located too high or was too small to prevent upward migration of the femoral head. The technique described emphasises the importance of resecting the thickened capsule and exposing the lateral margin of the acetabulum to allow a substantial portion of the outer cortex of the ilium to be turned down both laterally and anteriorly over the femoral head. Twenty cases of acetabuloplasty have been followed up for an average period of nine years. The results have been assessed from the point of view of function, relief of pain, and degree of support of the femoral head as demonstrated radiologically. The shelf operation is a simple and safe procedure, and provided it gives adequate support to the subluxated femoral head it will ensure a stable, mobile joint capable of standing up to reasonable use for many years. If deterioration should occur in later life, total hip replacement may be undertaken more easily than if the joint had been allowed to undergo progressive subluxation.

Six boys were examined during the second year of life, each with symptoms in one hip. The affected femoral capital epiphysis was smaller or more irregular (or both) than that of the unaffected hip but was neither subluxed nor dislocated. The subsequent radiographs revealed changes similar to those in severe Perthes' disease. Nevertheless, we give reasons for believing that this disorder in boys under two years of age should be distinguished from Perthes' disease.

Children who present late with hip dislocation may require femoral osteotomy after reduction, to correct valgus and anteversion deformity of the femoral neck. After these procedures proximal femoral growth is unpredictable. We have studied proximal femoral growth in 40 children who had been treated by femoral osteotomy.

Preoperatively, the mean femoral neck-shaft angle was 5° greater on the affected side than on the contralateral side. Postoperatively, it was 28° less. There was progressive recorrection; after five years the angle was not significantly different from that on the contralateral side. In our series 70% of the capital epiphyses became abnormally shaped, taking the appearance of a ‘jockey’s cap’. All the growth plates became angulated but this corrected with time.

Correction of the neck-shaft angle probably results from the more normal mechanical environment provided by reduction. The abnormal radiographic appearance of the epiphysis and growth plate is probably due to the rotation produced by the osteotomy.