[Reliability of digital measurement of acetabular index in hip dysplasia to the time children start
[Reliability of digital measurement of acetabular index in hip dysplasia to the time children start walking.]
AIM: An assessment of the reproducibility of the acetabular index on the basis of anteroposterior radiographs of the pelvis was performed in order to rule out residual dysplasia of the hip in the second year of life. METHOD: 51 consecutive patients (41 girls, 10 boys) with congenital dysplasia of the hip (35 bilateral, 16 unilateral) were included. Approximately at the time children started walking (mean age: 13.9 months; range: 10 - 22 months), digitised anteroposterior radiographs of the pelvis were recorded. Acetabular indices were measured bilaterally and blinded by two observers (O1 and O2), then classified according to the normal values of the hip joint described by Tönnis et al. (normal = lower single standard deviation [s], light dysplasia = above single standard deviation [s - 2 s] and severe dysplasie = above double standard deviation [2 s]). O2 repeated a second set of observations 6 weaks after the first. 306 values of the acetabular index (O1: 102 values; O2: 204 values) were statistically evaluated. Reliability was expressed by Pearson\’s correlation coefficient (PCC) and 95 % confidence interval. The rater agreement of the classification into normal values of the hip joint was calculated with the kappa coefficient. RESULTS: The PCC for intraobserver/interobserver reliability was 0.928/0.875. The 95 % confidence interval for intraobserver reliability was +/- 3.1 degrees and for interobserver reliability - 3.1 degrees /+ 5.9 degrees . A comparison of the first series of measurements by O1 with the series of measurements by O2 presented congruence in 66 of 102 hip joints classified according to the normal values of the hip joint as described by Tönnis. In 36 hip joints the classification was deviant, in 15 the classification differed between normal and s, in 21 between s and 2 s. The kappa coefficient of 0.51 showed moderate interobserver agreement. CONCLUSION: The reproducibility of radiological surveillance in congenital hip dysplasia, approximately to the time children start walking, can be considered as adequate. Correct positioning of the patient and exact definition of radiological landmarks are essential. In spite of good statistical values for reliability, decreasing accuracy in classifying into Tönnis\’ normal values of the hip joint is seen with increasing degree of dysplasia. In marginal cases, before coming to the decision for operative treatment, a second reading, ideally by the same observer and with a chronological shift, is advisable.
Dornacher D, Cakir B, Reichel H, Nelitz M.
Orthopädie, Orthopädische Universitätsklinik Ulm am RKU.