A novel predictive model of perioperative blood transfusion requirement in tuberculous spondylitis patients undergoing posterior decompression and instrumentation

Study design: Retrospective study. Purpose: The aim of this study was to develop a predictive model for determining perioperative blood transfusion in tuberculous spondylitis patients undergoing posterior decompression and instrumentation. Background: Tuberculous spondylitis is a common infection found in the spine. This condition may result in the need of surgical treatment, especially when there is a delay in diagnosis with inadequate antituberculosis drug treatment. The procedure results in high amount of bleeding on many occasions, leading to the higher rate of intraoperative transfusion. We develop a predictive model that can be used to determine blood transfusion requirement in spinal tuberculosis surgery. Methods: We reviewed the medical records of 83 tuberculous spondylitis patients who underwent posterior decompression and instrumentation. The clinical characteristics of the patients were analyzed using bivariate and multivariate regression test. The impact and strength of these variables was assessed to predict the probability of intraoperative red blood cell transfusion presence based on unstandardized beta, standard error, receiver operating characteristic, and confluence of sensitivity and specificity curve analyses. Furthermore, validation of this newly proposed predictive scoring system was performed using a set of 45 patients. Results: The factors that significantly affect the need of blood transfusion during posterior spondylitis tuberculosis surgery were BMI (p = 0.005), pre-operative Hb (p < 0.001), number of affected segments (p = 0.042), and duration of surgery (p = 0.003). Our predictive model showed good sensitivity and specificity values based on a large area under the curve (0.913) and strong Pearson’s r testing (correlation coefficient of 0.752). Validation set also resulted a large area under the curve (0.905) and strong correlation coefficient of 0.713. Conclusion: BMI, pre-operative Hb, number of affected segments, and duration of surgery became the significant factors which correlated to the presence of red blood cell transfusion in patients undergoing posterior spondylitis tuberculosis surgery. This predictive scoring system can be used to further adjust blood matching and inventory, determine intraoperative blood management, and ensure the safety of surgery in a comprehensive manner.