Novel Arthroscopic Classification of Osteochondritis Dissecans of the Knee: A Multicenter Reliability Study

Background:Several systems have been proposed for classifying osteochondritis dissecans (OCD) of the knee during surgical evaluation. No single classification includes mutually exclusive categories that capture all of the salient features of stability, chondral fissuring, and fragment detachment. Furthermore, no study has assessed the reliability of these classification systems.Purpose:To determine the intra- and interobserver reliability of a novel, comprehensive arthroscopic classification system with mutually exclusive OCD lesion types.Study Design:Cohort study (diagnosis); Level of evidence, 3.Methods:The Research in OsteoChondritis of the Knee (ROCK) study group developed a classification system for arthroscopic evaluation of OCD of the knee that includes 6 arthroscopic categories—3 immobile types and 3 mobile types. To optimize comprehensibility and applicability, each was developed with a memorable name, a brief description, a line diagram corresponding to the archetypal arthroscopic appearance, and an arthroscopic photograph depicting this archetype. Thirty representative arthroscopic videos were evaluated by 10 orthopaedic surgeon raters, who classified each lesion. After 4 weeks, the raters again classified the OCD lesions depicted in the 30 videos in a new, randomly selected order. Reliability was assessed via the intraclass correlation coefficient (ICC).Results:The interobserver reliability of this novel arthroscopy classification was estimated by an ICC of 0.94 (95% CI, 0.91-0.97) for the first round and 0.95 (95% CI, 0.93-0.98) for the second round. According to the standards for the magnitude of the reliability coefficient of Altman, these ICCs indicate that interobserver reliability was very good. The intraobserver reliability was estimated by an ICC of 0.96 (95% CI, 0.95-0.97), which indicates that the intraobserver reliability was similarly very good.Conclusion:The ROCK OCD knee arthroscopy classification system demonstrated excellent intra- and interobserver reliability. In light of this reliability, this classification system may be used clinically and to facilitate future research, including multicenter studies for OCD.