Assessment of the reliability and reproducibility of the Langenskiold classification in Blount's disease

The Langenskiold classification is the most commonly utilized classification system for the radiological features of Blount's disease. Although there is only a single study found on the interobserver variability and none found on the intraobserver variability, it is commonly used for prognostication and guiding management decisions. The aim of this study was to determine the reliability and reproducibility of the Langenskiold classification. A retrospective review of radiographs was done of patients treated for infantile and juvenile Blount's disease at Chris Hani Baragwanath Academic Hospital from 2006 to 2016. There were 70 radiographs of acceptable quality, which were reviewed and staged on two occasions according to the Langenskiold classification by three orthopaedic consultants and three orthopaedic surgery senior residents. Pearson correlation coefficients, percentage agreements, and kappa statistics were used to evaluate both the reliability and reproducibility. Of the 70 images staged, only two (2.9%) were staged the same by all six observers, and 20 (28.6%) images differed by a single stage. The consultants had 17 (24.3%) images staged the same whereas the residents had 12 (17.1%) images staged the same. The overall kappa for all six observers showed a fair agreement of 0.24. Again, the consultants had a higher kappa-value compared to residents of 0.25 and 0.24, respectively. The reproducibility amongst all observers was fair with a kappa-value of 0.38. The consultants had a higher mean score of 0.48 compared to 0.26 for the residents. There was only a fair overall reliability and reproducibility amongst the six observers. We recommend the Langenskiold classification be used with caution when being used for prognostication and management planning as well as when interpreting any research relying on this classification. Level of evidence: Level III, diagnostic study.