Evaluating and managing type 2 odontoid fractures: an interrater reliability study assessing agreement among spine...

Consensus agreement exists regarding the management of Anderson and D'Alonzo type 1 and type 3, however management of type 2 odontoid fractures remains largely controversial. Though displaced type 2 odontoid fractures are generally considered operative, in the appropriate patient, the parameters that define "displacement" and their relation to fracture stability and outcomes, are poorly defined in the literature. Sagittal fracture displacement, sagittal fracture angulation, fracture comminution, and presence of local cervical deformity impact surgical decision-making, but the effect each characteristic has on clinical decisions has yet to be defined.

Our goal in this study is to 2-fold: (1) define agreement among spine surgeons, as it relates to 5 type 2 odontoid parameters: presence of local cervical deformity, presence of fracture comminution, sagittal displacement >5 mm, sagittal angulation >11°, and clinical management, (2) assess the impact each variable has on the likelihood of surgical indication.

Radiographic analysis study of spine surgeons assessing odontoid fracture morphology and clinical management, using retrospectively collected imaging data.

Patients 65 years or older, treated within the military health system, with type 2 odontoid fractures and CT imaging.

Our outcome measures of interest in this study are (1) agreement among spine surgeons, as it relates to radiographic measurements, and (2) the relative impact, measured by odds ratio, that each radiographic parameter has on clinical decision making, METHODS: We queried the Military Health System (MHS) for all type 2 odontoid fractures, between 2016 and 2023, which resulted in 441 patients, of which 37 had viewable CT scans. Six spine surgeons, 3 orthopedic-trained spine surgeons and 3 neurosurgery-trained spine surgeons, reviewed 37 sagittal CT scans of acute type 2 odontoid fractures and recorded "0' or "1" for the presence of local cervical deformity, presence of fracture comminution, sagittal displacement >5 mm, sagittal angulation >11° and surgical (1) or nonsurgical management (0). We performed an interrater reliability analysis using Fleiss' kappa coefficient to assess agreement among raters and binary multivariate regression analysis to quantify the effect of each variable on eventual clinical management.

Among all spine surgeons, there was substantial agreement with sagittal angulation measurements (k=0.69, p<.01), moderate agreement with sagittal displacement measurements (k=0.55, p<.01 and comminution (k=0.40, p<.01), and fair agreement regarding surgical decision-making (k=0.262, p<.000001). Subspecialty subgroup analysis demonstrated slight agreement with operative management in, both, orthopedic-trained spine surgeons (k=0.10, p>0.05) and neurosurgery-trained spine surgeons (k=0.02, p<.05). Binary univariate regression analysis identified each variable as significantly associated with surgical management. Binary multivariate regression analysis indicated sagittal displacement (OR=21.3, [9.19-54.0, 95% CI]), fracture comminution (OR=6.29, [1.84-23.1, 95% CI]), and local cervical deformity (OR=11.0, [3.87-35.2, 95% CI]), as independently associated with surgical management. ROC and AUC analysis identified sagittal displacement as an excellent predictor (AUC=0.96 [0.903-1.00]) relative to surgical management, while the combinations of deformity and comminution (AUC=0.78, [61.99-94.57]), deformity and angulation (AUC=0.79, [64.94-93.50]), and angulation and comminution (AUC=0.75, [56.61-93.96]), represent fair predictors of surgical management.

Our findings quantify the magnitude of surgical agreement of type 2 odontoid fractures among spine surgeons. Our analysis identified displacement >5mm, angulation >11°, and presence of local cervical deformity as independent variables associated with a higher likelihood of surgical management in patients with type 2 odontoid fractures. Though there is general agreement upon radiographic assessment, there lacks a corresponding agreement in clinical management. This indicates the need for prospective studies that identify predictive preoperative characteristics that correlate with optimal postoperative patient-related outcomes.