Neonatal hip ultrasound: tangent to lateral roof angle correlates better with stability than alpha angle according to Graf or percentage femoral head coverage

• Published in: Journal of pediatric orthopaedics. B Vol. 29; no. 3; pp. 219 - 227
• Main Authors: Wilkinson, A. Graham, Wilkinson, Sally, Elton, Robert A., Davis, Joanna
• Format: Journal Article
• Language: English
• Published: United States Wolters Kluwer Health, Inc. All rights reserved 01-05-2020
• Subjects: Female; Femur Head - diagnostic imaging; Hip Dislocation, Congenital - classification; Hip Dislocation, Congenital - diagnostic imaging; Humans; Infant, Newborn; Joint Instability - classification; Joint Instability - diagnostic imaging; Male; Neonatal Screening - classification; Neonatal Screening - standards; Retrospective Studies; Ultrasonography - classification; Ultrasonography - standards
• ISSN: 1060-152X; 1473-5865
• DOI: 10.1097/BPB.0000000000000709

The aim of this study was to determine the relationship of tangent to the lateral roof angle (TLRA), Graf’s alpha angle and percentage femoral head cover (PHC) and to observed stability, and to establish intraobserver and interobserver errors for measurement of alpha angle and TLRA. In total, 2235 babies referred to the hip review clinic over a period of 8  years. Intraobserver and interobserver errors were calculated from readings of 383 hip images by an experienced paediatric radiologist and a trainee. Taking TLRA <70° as abnormal resulted in sensitivity for unstable right hips of 88% compared with 77% for alpha angle <60° (P  =  0.002) and 81% for PHC <60% (P  =  0.028) and specificity of TLRA 89%, alpha angle 90% (P  =  0.07) and PHC 83% (P  <  0.001). Corresponding figures for left hips are sensitivity of TLRA 99%, alpha angle 91% (P  <  0.001) and PHC 96% (P  =  0.013), and specificity TLRA 86%, alpha angle 83% (P  =  0.001) and PHC 67% (P  <  0.001). Mean intraobserver errors for alpha angle were 1.85° and 1.81° for consultant and trainee compared to 2.54 and 2.55 for TLRA. Mean interobserver errors were 2.22 for alpha angle and 3.42 for TLRA. TLRA, a new parameter, correlated better with observed stability with significant improvement in sensitivity in both hips and specificity in left hips compared with Graf’s alpha angle, and significantly improved sensitivity and specificity in both hips compared with percentage femoral head cover.