Inter-observer agreement of ultrasonographic measurement of alpha and beta angles and the final type classification based on the Graf method.

Abstract

The aim of our study was to investigate the agreement of the assessment of hip ultrasonograms by different observers. In the period from June 3rd to December 9th 2002, four different (by experience and field) groups of observers rated all first time sonograms obtained in our hospital. The results in terms of angle and type classification were compared. 158 ultrasonographic images were evaluated. The inter-observer agreement for the classification "normal" (type I) versus "abnormal" (types IIa+ to IV) varied from 83% to 90% with kappa coefficients which indicated moderate (kappa 0.55) to substantial (kappa 0.71) inter-observer agreement. For one pair of observers, a better agreement could be demonstrated for the assessment of immature hips than for mature ones. The deviation for the a-angle was 0 to 16 degrees with a standard deviation of 3.15 degrees (95% CI 2.95, 3.37), and for the b-angle 0 to 26 degrees with a standard deviation of 6.1 degrees (95% CI 5.7, 6.5). The intra-class correlation coefficient was estimated to be 0.72 and 0.34 for the alpha and beta angles respectively. If the hip was immature there was no increase in the discrepancy in assessment between observers. The least agreement existed between the less experienced and the most experienced. It has not been possible to make a statement on the discrepancy with regard to initial signs of instability or decentralization of the hip joints because of the small number of hips of this type. Although the spread in measured a- and b-angles is large, the inter-observer agreement for the classification showed good results. No disagreement occurred in the diagnosis of normal vs. dysplastic hips, so no severe cases have been missed. The experience and training of the investigators seemed to play an important role with regard to variability and agreement. The agreement in the assessment of immature hips was better than that of mature hips. Therefore, ultrasound examination of infant hips would appear to be a trustworthy screening method.