RESUMO
BACKGROUND: The standard evaluation of epiphyseal involvement in Perthes disease is lateral pillar classification. However, it needs to be compared with contralateral normal hip leading to limited use in bilateral disease. We, therefore, develop a ratio between epiphyseal height and metaphyseal width of affected hips to estimate lateral pillar involvement. This study aimed to assess the height-width ratio of the proximal femoral epiphysis in non-Perthes children, and to find the relationship between the height-width ratio and lateral pillar classification in Perthes disease. METHODS: A cross-sectional study was conducted between 2009 and 2015. Phase I included children aged 2 to 15 years who did not have Perthes disease. Phase II included children aged 2 to 15 years who had Perthes disease. Other abnormal proximal femoral epiphysis was excluded. Lateral pillar height and metaphyseal width were independently measured twice by 2 assessors in each phase. Intraobserver and interobserver levels of agreement, height-width ratio and cut-off points to differentiate lateral pillar types were determined. RESULTS: There were 69 children (87 hip radiographs) who had non-Perthes hips, and 18 boys with Perthes disease (20 hip radiographs). Height-width ratio in the non-Perthes group increased from 0.38 to 0.48 at 2 to 10 years of age and remained constant until maturity. Average height-width ratio in lateral pillar A/non-Perthes hip was 0.47±0.05, lateral pillar B or B/C 0.32±0.05, and lateral pillar C 0.18±0.05. Intraobserver and interobserver level of agreements of height-width ratio in Perthes disease were 0.007 (95% confidence interval, -0.030 to 0.043) and 0.006 (95% confidence interval, -0.119 to 0.107), respectively. Cut-off values to differentiate lateral pillar A and B or B/C was 0.40, and to differentiate lateral pillar B or B/C, and C was 0.25 with 90% accuracy and area under receiver operating characteristic curve of 0.9. CONCLUSIONS: Height-width ratio is useful for grading severity in unilateral and bilateral Perthes disease. It has excellent reliability and validity with exact cutoff values to estimate lateral pillar classification. LEVEL OF EVIDENCE: Level II-diagnostic study.
