Background It is unclear whether the sandwich vertebra, is at higher risk of new symptomatic fractures (NSFs), and whether prophylactic augmentation might benefit patients with sandwich vertebrae.

Objective To compare fracture-free probabilities of sandwich, ordinary-adjacent, and non-adjacent vertebrae, and identify predictors of NSFs.

Methods Data were retrospectively analyzed for patients who had undergone vertebral augmentation resulting in sandwich vertebrae. NSF rates were determined and predictors were identified using Cox proportional hazard models.

Results The analysis included 1408 untreated vertebrae (147 sandwich, 307 ordinary-adjacent, 954 non-adjacent vertebrae) in 125 patients. NSFs involved 19 sandwich, 19 ordinary-adjacent, and 16 non-adjacent vertebrae. The NSF rate was significantly higher in the patients with sandwich vertebrae (27.2%) than among all patients (14.8%). At the vertebra-specific level, the NSFs rate was 12.9% for sandwich vertebrae, significantly higher than 6.2% for ordinary-adjacent and 1.7% for non-adjacent vertebrae. The corresponding fracture-free probabilities of sandwich, ordinary-adjacent, and non-adjacent vertebrae were 0.89, 0.95, and 0.99 at 1 year, and 0.85, 0.92, and 0.98 at 5 years (p<0.05). Cox modeling identified the following as predictors for occurrence of an NSF in a given vertebra: vertebra location, type of vertebrae, number of augmented vertebrae, and puncture method.

Conclusion Sandwich vertebrae are at higher risk of NSFs than ordinary-adjacent and non-adjacent vertebrae, and several NSF risk factors were identified. Since 85% of sandwich vertebrae are fracture-free for 5 years and NSF risk increases with the number of augmented vertebrae, prophylactic augmentation of every sandwich vertebra may be unnecessary.