ObjectiveTo compare the refracture risk between sandwich vertebrae and ordinary adjacent vertebrae, and to explore the risk factors related to refracture.MethodsRetrospective analysis was performed on the data of patients who received percutaneous vertebral augmentation (PVA) and formed sandwich vertebrae between April 2015 and October 2019. Of them, 115 patients were enrolled in the study. There were 27 males and 88 females with an average age of 73.9 years (range, 53-89 years). Univariate analysis was performed to analyzed the patients’ general data, vertebral augmentation related indexes, and sandwich vertebrae related indexes. Survival analysis was performed for all untreated vertebrae at T4-L5 of the included patients at the vertebra-specific level, and risk curves of refracture probability of untreated vertebrae between sandwich vertebrae and ordinary adjacent vertebrae were compared. Cox’s proportional hazards regression model was used to analyze risk factors for refracture.ResultsThe 115 patients were followed up 12.6-65.9 months (mean, 36.2 months). Thirty-seven refractures involving 51 vertebral bodies occurred in 31 patients. The refracture rate of 27.0% (31/115) in patients with sandwich vertebrae was significantly higher than that of 15.2% (187/1228) in all patients who received PVA during the same period (χ2=10.638, P=0.001). Univariate analysis results showed that there was a significant difference in the number of augmented vertebrae between patients with and without refractures (Z=0.870, P=0.004). However, there was no significant difference in gender, age, body mass index, whether had clear causes of fracture, whether had dual energy X-ray absorptiometry testing, whether the sandwich vertebra generated through the same PVA, puncture method, method of PVA, number of PVA procedures, number of vertebrae with old fracture, whether complicated with spinal deformity, bone cement distribution, and kyphosis angle of sandwich vertebral area (P>0.05). Among the 1 293 untreated vertebrae, there were 136 sandwich vertebrae and 286 ordinary adjacent vertebrae. The refracture rate of sandwich vertebrae was 11.3% which was higher than that of ordinary adjacent vertebrae (6.3%)(χ2=4.668, P=0.031). The 1- and 5-year fracture-free probabilities were 0.90 and 0.87 for the sandwich vertebrae, and 0.95 and 0.93 for the ordinary adjacent vertebrae, respectively. There was a significant difference between the two risk curves of refracture (χ2=4.823, P=0.028). Cox’s proportional hazards regression model analysis results showed that the sandwich vertebrae, thoracolumbar location, the number of the augmented vertebrae, and the unilateral puncture were significant risk factors for refracture (P<0.05).ConclusionThe sandwich vertebrae has a higher risk of refracture when compared with the ordinary adjacent vertebrae, and its 1- and 5-year fracture-free probabilities are lower than those of the ordinary adjacent vertebrae. However, the 5-year fracture-free probability of sandwich vertebrae is still 0.87, so prophylactic enhancement is not recommended for all sandwich vertebrae. In addition, the sandwich vertebrae, thoracolumbar location, the number of the augmented vertebrae, and the unilateral puncture were important risk factors for refracture.