Reprodutibilidade no diagnóstico de fraturas vertebrais osteoporótica x neoplásica seguindo instrumento específico para diferenciação entre tais fraturas

Abstract

Abdala EHC. Reproducibility in the diagnosis of osteoporotic x neoplastic vertebral fractures following a specific instrument for differentiation between such fractures. [Dissertation] Ribeirão Preto: University of São Paulo, Faculty of Medicine of Ribeirão Preto; 2019. Non-traumatic spinal fracture in the elderly has as its main etiologies metastasis and osteoporosis. Differentiating these diagnoses in the acute phase can be difficult even to experienced physicians. Recently, a score named META based on magnetic resonance imaging (MRI) was created to differentiate benign vertebral fracture (FVB) from malignant vertebral fracture (FVM) with excellent interobserver correlation and accuracy between the authors. However, the score was evaluated by independent physicians presenting a different result from the META creators. The present study aims to evaluate if this score has a good consistency between the examiners and if it can be used as a tool for differentiation FVB from FVM. A retrospective observational study was performed with 63 MRI patients with BVF or MVF. Two spine surgeons and two medical students evaluated the images using the META score. The intraclass correlation coefficient (ICC) and Kappa (k) were used to evaluate the inter and intraobserver relationship in the diagnostic relationship between the evaluators and also for the correlation of the individual score criteria. The area under the Receiver Operating Characteristic Curve (ROC) was calculated to establish the accuracy. Interobserver evaluation was excellent among both spine surgeons (ICC = 0.822) and medical students (ICC = 0.835) and Kappa was considered substantial in both groups (0.64 and 0.61). Both surgeons and students respectively intraobserver evaluation were excellent [ICC = 0.935 (0.908-0.955)], [ICC = 0.939 (0.914-0.957)], as well as the relationship with Kappa between surgeons (0.81) and students (0.84). The Kappa evaluation of each criterion ranged from 0.74 to 0.18 in all groups. Accuracy improved for students by comparing their evaluation by their experience and the META score (0.57 to 0.76). Nevertheless, for surgeons there was no difference (0.76 to 0.75). The ROC curve was 0.79 for surgeons and 0.71 for students (p <0.0001). Good accuracy was observed using the score and excellent interobserver correspondence. We evaluated the score as a good tool for differentiating between fractures for students. For surgeons, however, despite the good accuracy, there was no difference regarding their subjective assessment.