This Article Full Text Full Text (PDF) Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal My Folders Download to citation manager Google Scholar Articles by Yiannakopoulos, C. K. Articles by Hartofilakidis, G. PubMed PubMed Citation Articles by Yiannakopoulos, C. K. Articles by Hartofilakidis, G.

Inter- and intra-observer variability of the Crowe and Hartofilakidis classification systems for congenital hip disease in adults

¹ Nikis 2, 14561, Athens, Greece.
² North Manchester General Hospital, Crumpsall, Manchester M8 5RB, UK.
³ Coxa Hospital for Joint Replacement, Biokatu 6b, 33250 Tampere, Finland.
⁴ The Centre for Hip Surgery Wrightington Hospital, Hall Lane, Appley Bridge, Wigan WN6 9EP, UK.
⁵ University of Athens Medical School, Fotiou Patriarchou 21, Athens 11471, Greece.

Correspondence should be sent to Dr C. K. Yiannakopoulos; e-mail: cky{at}ath.forthnet.gr

Our study evaluated the reliability of the Crowe and Hartofilakidis classification systems for developmental dysplasia of the hip in adults. The anteroposterior radiographs of the pelvis of 145 patients with 209 osteoarthritic hips were examined twice by three experienced hip surgeons from three European countries and the abnormal hips were rated using both classifications. The inter- and intra-observer agreement was calculated.

Interobserver reliability was evaluated using weighted and unweighted kappa coefficients and for the Crowe classification, among the three pairs there was a minimum kappa coefficient with linear weighting of 0.90 for observers A and C and a maximum kappa coefficient of 0.92 for observers B and C. For the Hartofilakidis classification, the minimum kappa value was 0.85 for observers A and B, and the maximum value was 0.93 for observers B and C. With regard to intra-observer reliability, the kappa coefficients with linear weighting between the two evaluations of the same observer ranged between 0.86 and 0.95 for the Crowe classification and between 0.80 and 0.93 for the Hartofilakidis classification.

The reliability of both systems was substantial to almost perfect both for serial measurements by individual readers and between different readers, although the information offered was dissimilar.