Predictive factors for postoperative deformity in thoracolumbar burst fractures. A statistical approach.

Abstract

Objectives: To assess, by means of statistical analysis, if, and to what extent, the final outcome of the surgical treatment of burst fracture depends on procedure, the fracture level and initial deformity severity. Materials and Methods: Database of 287 patients with single level thoracal and lumbar spine fractures, analysed with simple and multiple linear regression equations, where the dependent variable was the last follow-up kyphotic angle, and the predictor variables were the operation type (anterior approach - AA, posterior short segment fixation - PSSF, and posterior monosegmental fixation - PMF); the fracture level: (T11-L1, L2-L3, L4-L5); and the preoperative angle. The models were applied on either the whole lot or on the operation type subgroups. Results: The fracture level in the simple linear regressions was a significant explanatory factor only for AA and PMF subgroups, accounting for 32% and 18% respectively. The multiple linear regression models for the same subgroups explained up to 40% of the variation in LFU. Surgical treatment as predictor variable, depicted that patients operated by PSSF developed a post-surgical kyphotic angle by 8.51 degrees more severe than patients operated by AA. However, the model explained only 2% of the variation of the regressand. The simple linear regressions performed on each subgroup with preoperative angle as independent variable, explained 15% (PSSF subgroup), 17% (AA subgroup) and 34% (PMF subgroup) of the variation of the LFU. Conclusions: All valid regression models displayed modest explanatory power, suggesting that there are, still, other relevant factors than those taken into consideration. DOI: 10.3944/AOTT.2015.14.0274 This abstract belongs to the un-edited version of the article and is only for informative purposes. Published version may differ from the current version.