CT-based five-segment classification in Freiberg's infarction: Evaluation of its intraobserver and interobserver reliability.

ABSTRACT

BACKGROUND: According to the location and frequency of osteonecrosis zone of metatarsal head, a map of osteonecrosis zone was drawn for Freiberg's infarction. The objective of the current study is to develop a new computed tomography-based Five-Segment classification system for Freiberg's infarction and testify if it has good intra- and interobserver reliability or not. METHODS: According to the location and its frequency of osteonecrosis zone of metatarsal head, a map of osteonecrosis zone was drawn. According to the distribution of osteonecrosis zones of metatarsal heads, we proposed the Five-Segment classification system. Four evaluators evaluated each radiography and computed tomography (CT) twice at 8-week intervals. To test the reproducibility of the Five-Segment classification system, the interobserver and intraobserver reliability of this classification system comparing with that of the Smillie classification by four observers using the kappa statistic. RESULTS: The 80 cases were classified into five reproducible types by using Five-Segment classification system type Ⅰ, two (2.5%) cases; type Ⅱ, ten (12.5%) cases; type Ⅲ, 42 (52.5) cases; type Ⅳ, 24 (30.0%) cases; type Ⅴ, two (2.5%) cases. The mean kappa value for interobserver reliability using Smillie classification systems was 0.562 (95% CI 0.531-0.585), whereas the mean kappa value was 0.828 (95% CI 0.801-0.852), by using Five-Segment classification; the mean kappa values for intraobserver reliability by using Smillie classification and Five-Segment classification were 0.777 (95% CI 0.762-0.792) and 0.860 (95% CI 0.843-0.895), respectively. CONCLUSIONS: The new Five-Segment classification system demonstrated perfect interobserver and intraobserver agreement between evaluators in the management of Freiberg's infarction. Prospective studies should be done to evaluate its prognostic value and utility in clinical practice. LEVEL OF EVIDENCE Level IV, retrospective.