Reliability of the Severin classification in the assessment of developmental dysplasia of the hip.

This study was undertaken to investigate the reliability of Severin's classification at various ages, and to determine whether the reliability is improved by careful measurement rather than subjective assessment. The radiographs taken at ages 6, 10 and 16 years of 20 randomly selected patients treated for developmental dysplasia of the hip were graded by six observers on two separate occasions using parameters measured according to Severin's criteria. In addition, four of these observers regraded the same radiographs using subjective assessment without measurements being made on two other separate occasions. Agreements between and within observers were evaluated using the weighted Cohen's kappa statistic for each age group. Intraobserver reliability was good, there being a close association between the measured and the subjectively observed ratings. This accords with the subjective nature of this classification. The interobserver reliability was found to be poor although it improved when direct measurements were made. Overall agreement between observers improves as patient age increases. It is concluded that comparisons between different observers using the Severin classification system are not reliable. However, a single investigator comparing treatment modalities in the same study allowing for individual bias in assessing deformity and subluxation would produce reliable results.