Correlation of examiner judgement and radiological digital pictures in infants with upper cervical spine dysfunction: a cross-sectional study
TL;DR
Although lateral atlantodental distance measurements showed excellent inter-rater reliability, the SEM was relatively large compared with the small absolute values of the measurements, which limits the ability to detect subtle side-to-side asymmetries.
Abstract
This study aimed to (1) assess the inter-rater reliability of lateral atlantodental distance measurements in digital open-mouth radiographs of infants with upper cervical spine dysfunction, and (2) investigate correlations between lateral atlantodental distance values and clinical findings. 72 digital radiographs of the odontoid region from infants aged 3–12 months were analyzed (Aug 2021–Dec 2023). Lateral atlantodental distance was measured bilaterally by three independent raters. Inter-rater reliability and measurement precision was assessed using intraclass correlation coefficients (ICC), standard error of measurement (SEM), and minimal detectable change (MDC). Side differences were evaluated using paired t-tests. Mean lateral atlantodental distance values were further examined for their association with the clinically assessed side of cervical rotational malposition using logistic regression and receiver operating characteristic (ROC) analysis. Inter-rater reliability was excellent for right (ICC=0.91 [95% CI: 0.86 – 0.94]) and left (ICC=0.91 [95% CI: 0.86 – 0.95]). The SEM was 0.62 mm for both sides, the MDC was 1.72 mm (right) and 1.73 mm (left). In infants with left-sided cervical rotational malposition, lateral atlantodental distance was significantly greater on the left side, whereas in infants with right-sided cervical rotational malposition, lateral atlantodental distance was significantly greater on the right side. Right-sided lateral atlantodental distance was a significant predictor of clinically assessed right-sided cervical rotational malposition (OR 1.80, 95% CI 1.14–2.84, P=0.01). ROC analysis demonstrated moderate discriminatory performance (AUC = 0.74), meaning that the model is reasonably capable of distinguishing patients with right-sided cervical rotational malposition from those without. Although lateral atlantodental distance measurements showed excellent inter-rater reliability, the SEM was relatively large compared with the small absolute values of the measurements (in millimeters), which limits the ability to detect subtle side-to-side asymmetries. Lateral atlantodental distance should be interpreted as a reproducible radiographic finding associated with cervical rotational malposition rather than as an independent diagnostic marker or pathognomonic sign in infants. ClinicalTrials.gov ID NCT04981782 (28.07.2021).