Development, test reliability and validation of a classification for revision hip arthroplasty

The objective of the study was to develop a valid and reliable classificati on system for failed hip arthroplasties. The study uses research principals derived from multi-attribute utility theory and consensus group techniques . The development of the severity measure was done in two phases. Phase I o f the study included: (a) questionnaire development, (b) submission of the questionnaire to the respondents. (c) data synthesis of the responses and i tem reduction, and id) classification development and inter-observer reliab ility testing. Phase II included: (a) resubmission of the instrument to the respondents for suggestions/feedback, (b) instrument revision by the co-in vestigators based on the respondents' second feedback, and (c) inter-observ er reliability testing and intraoperative validity testing of the instrumen t. The questionnaires sought to capture expert opinion as to what clinical determinants obtained preoperatively (during patient interview, physical ex am and review of plain radiographs - AP pelvis and hip lateral) that would in their clinical experience reveal intraoperative severity. There was an 8 0% (16/20) response rate from the outside experts invited to participate in the study. Based on item reduction and test-retest analysis, a five-grade radiographic classification for the acetabulum as well as the femur was dev eloped. This system was then reviewed by 13 of the initial outside experts (16, 80%) who participated iu the first round. Inter-rater reliability test ing of the final format of the classification revealed a weighted kappa sta tistic value of 0.88 between the two-blinded raters tinter-rater reliabilit y) and 0.87 between the blinded raters and the reference standard (intraope rative validity). We conclude that the study developed a reliable and valid radiographic classification system for failed hip arthroplasty. (C) 2001 O rthapaedic Research Society. Published by Elsevier Science Ltd. All rights reserved.