Radiographers' Recognition of Developmental Dysplasia of the Hip on Paediatric Pelvic Radiographs and Perceived Image-Quality Factors: An Exploratory Observer-Performance Study.

Abstract

Introduction: Developmental dysplasia of the hip (DDH) encompasses a spectrum of acetabular and hip-joint abnormalities that may progress from dysplasia and instability to subluxation or frank dislocation. Radiographers are central to the acquisition of paediatric pelvic radiographs, but their ability to recognise radiographic features suggestive of DDH is not well established. This exploratory study evaluated abnormality recognition rather than formal diagnostic reporting and described radiographers' confidence and perceived image-quality requirements.

Methods: A prospective, cross-sectional observer-performance study was conducted at the European Congress of Radiology 2023. Thirty radiographers independently reviewed 25 anonymised paediatric anteroposterior pelvic radiographs selected from the DDH teaching dataset. The observer set comprised 13 normal and 12 DDH-positive radiographs: four left-sided, four right-sided and four bilateral cases. Reference labels were provided by three specialists in the source dataset. For each case, participants selected bilateral DDH, right-sided DDH, left-sided DDH, or no DDH; rated confidence from 1 to 10; and graded subjective image quality from perfect to inadequate. For the primary binary analysis, any response indicating DDH in at least one hip was classified as DDH-positive. A unilateral response to a bilateral case was therefore counted as positive because it would prompt referral or further assessment. Observer characteristics and technical factors were described but were not tested as independent predictors of performance.

Results: Mean reported sensitivity for DDH detection was 56.2% (SD 21.6%; range 0%-92.3%), and mean reported specificity was 40.2% (SD 28.4%; range 0%-100%). Confidence was generally moderate, with image-level mean ratings predominantly between 4 and 6. Participants most frequently rated appropriate collimation, absence of pelvic rotation, lower-limb positioning, radiographic centring and removal of artefacts as important technical considerations. These descriptive findings do not establish statistically significant predictors of observer performance.

Conclusion: Radiographers demonstrated substantial variability and overall limited recognition of DDH in this small, convenience-sampled observer study. The findings support further development and prospective evaluation of targeted education and clearly defined preliminary image-evaluation pathways, but they do not establish that radiographer flagging, training, or artificial intelligence support improves diagnostic timeliness or patient outcomes.