Cervical Spine Fracture Prediction by Simple Plain X‐Ray in Ankylosing Spondylitis Patients after Low‐Energy Trauma

OBJECTIVE: Timely diagnosis is essential in the management of cervical spine fracture (CSF) in ankylosing spondylitis (AS) patients. However, the value of simple plain X‐ray in the early management of ASCSF has not been well‐studied. This study aimed to explore the prediction ability of simple plain X‐ray for CSF in AS patients who suffer from low‐energy trauma (LET). METHODS: From January 2010 to December 2020, AS patients who experienced LET were retrospectively reviewed. Clinical data including gender, age, body mass index, time interval between AS diagnosis and trauma, smoking or not, and a presence of continuous bony bridge between anterior margin of C1 and C2 body or not were collected. Morphological features including atlanto‐occipital gap, Pavlov ratio of C2–7, Angle A–D, Borden's index, and Harrison's value were measured by the lateral cervical X‐ray. All data was compared between patients who had CSF and those who did not. Binary logistic regression analysis and receiver operator characteristic (ROC) curves were applied to discriminate and assess the predictive parameters. RESULTS: A total of 129 AS patients were divided into Fracture group (41 cases) and Non‐fracture group (88 cases) based on whether CSF existed. Twelve parameters showed significant differences between two groups (p < 0.05). According to the binary logistic regression model, four of the 12 parameters showed a further correlation with the occurrence of CSF, namely, mean Pavlov ratio (p < 0.001, OR = 0.067, 95% CI: 0.023 to 0.194), Angle D (p = 0.031, OR = 1.057, 95% CI: 1.005 to 1.112), Borden's index (p = 0.042, OR = 1.131, 95% CI: 0.994 to 1.287), the time interval between the AS diagnosis and the trauma (p < 0.020, OR = 0.935, 95% CI: 0.883 to 0.990). The ROC curve further revealed the mean Pavlov ratio had the largest AUC (0.793) with the cut‐off of 0.72. While the optimal cut‐off value was 45.65° for Angle D (sensitivity = 61.0%, specificity = 78.4%), 9.79 for Borden's index (sensitivity = 87.8%, specificity = 37.5%), 15.50 years for the time interval between AS diagnosis and trauma (sensitivity = 70.7%, specificity = 56.8%). CONCLUSIONS: The time interval between the AS diagnosis and the trauma, mean Pavlov ratio, Angle D, and Borden's index showed predictive ability for the occurrence of CSF in AS patients who encounter LET. Surgeons should consider measuring these parameters in the management of AS patient.