Inter- and intra-rater reliability of predictive models for the non-operative healing potential of stable juvenile osteochondritis dissecans (JOCD) knee lesions

Abstract

Objectives: While juvenile osteochondritis dissecans (JOCD) lesions have greater healing potential than equivalent lesions in adults, only 50% of JOCD lesions demonstrate radiographic healing after 6 months of non-operative treatment. Krause et al. (2013) and Wall et al. (2008) have described models to predict a patient’s probability of healing with non-operative treatment based on patient and lesion characteristics. The Wall et al. nomogram incorporates normalized lesion length, normalized lesion width, and pain type, while the Krause et al. model includes normalized lesion width, patient age, and the size of concurrent cyst-like lesions. The purpose of this study was to determine inter- and intra-observer reliability of the predictive nomograms. Methods: We retrospectively evaluated a consecutive series of children with open physes, who underwent non-operative treatment for stable JOCD lesions. Thirty-four patients (40 knees) were included in the study. At two time points at least one week apart, two medical students, two orthopaedic surgeons, and a radiologist made measurements of the JOCD lesions as described by Wall and Krause. These measurements, along with patient ages and pain type, were used to generate a point value for each lesion based on both the Krause and Wall models. Intra-class correlations (ICC) were calculated using the point value data obtained from the nomograms to determine inter- and intra-rater reliability. Results: Based on the Landis & Koch (1977) threshold, we found near perfect intra-rater correlation for all raters in terms of individual OCD measurements, as well as total point score as calculated by both Krause and Wall methods (Table 1). Additionally, amongst the five raters there was near perfect inter-rater reliabilities in total scores, as well as substantial to near perfect reliability in measuring the individual components of the scores (Table 2). Conclusion: There is high inter- and intra-rater reliability for both the Krause and Wall point systems for predicting healing of JOCD lesions. These models appear to be important tools for helping doctors and patients reliably predict the healing potential of non-operative management of JOCD lesions.