Interobserver agreement and clinical disparity between the Graf method and femoral head coverage measurement in developmental dysplasia of the hip screening: A prospective observational study of 198 newborns.

ABSTRACT: Ultrasonography is the ideal tool for assessing hip morphology in infants younger than 6-month-old. This study assessed the interobserver agreement and clinical disparities of the 2 most widely used ultrasound (US) methods, the Graf method, and femoral head coverage (FHC) measurement.A prospective observational study (STROBE compliant) of 2024 newborns was conducted between January 2017 and December 2018. Hip US was conducted on all newborns with abnormal Barlow and Ortolani maneuvers as well as on 50 randomly selected normal newborns. The physical examination and US were performed by a senior pediatric orthopedic surgeon with musculoskeletal sonography certification. Three observers with different levels of experience interpreted the images by using the Graf method and FHC. We analyzed the intraclass correlation coefficient, Cohen kappa, and the disparity between the clinical findings of the 2 methods.A total of 198 newborns (9.8%) presented with clinical instability, including 193 subluxatable hips in 168 patients (84.8%) and 45 dislocatable/dislocated hips in 30 patients (15.2%). The mean age at US examination was 11.69 days (range: 0-18 days). The intraclass correlation coefficient was .71 (95% CI: 0.55-0.83) for FHC, 0.63 (95% CI: 0.38-0.78) for the alpha angle, and 0.47 (95% CI: 0.16-0.69) for beta angle. The Cohen kappa coefficients of Graf type were 0.19 (95% CI: 0.03-0.35), 0.39 (95% CI: 0.20-0.58), and 0.17 (95% CI: 0.02-0.32) between observers 1 and 2, observers 1 and 3, and observers 2 and 3, respectively. Based on the Graf method, 14% of the stable hips had abnormal USs; by contrast, 19.2% of the subluxatable hips and 17.8% of the dislocatable/dislocated hips had normal Graf morphologies. In USs interpreted using FHC, 16% of stable hips demonstrated abnormal coverage, whereas 13.5% of subluxatable hips and 4.4% of dislocatable/dislocated hips had normal FHC.Incidence of clinically detectable hip instability was 9.8% among newborns in our series. Both alpha angle and FHC ratio revealed substantial interobserver agreement while beta angle achieved moderate agreement. FHC ratio possesses higher sensitivity and similar specificity compared with the Graf method when screening unstable hips.Level II, development of diagnostic criteria on basis of consecutive patients.