A clinical calculator for predicting intraoperative blood loss and transfusion risk in spine tumor patients

BACKGROUND CONTEXT: Surgery for vertebral column tumors is commonly associated with intraoperative blood loss (IOBL) exceeding 2 liters and the need for transfusion of allogeneic blood products. Transfusion of allogeneic blood, while necessary, is not benign, and has been associated with increased rates of wound complication, venous thromboembolism, delirium, and death. PURPOSE: To develop a prediction tool capable of predicting IOBL and risk of requiring allogeneic transfusion in patients undergoing surgery for vertebral column tumors. STUDY DESIGN/SETTING: Retrospective, single-center study. PATIENT SAMPLE: Consecutive series of 274 patients undergoing 350 unique operations for primary or metastatic spinal column tumors over a 46-month period at a comprehensive cancer center OUTCOME MEASURES: IOBL (in mL), use of intraoperative blood products, and intraoperative blood products transfused. METHODS: We identified IOBL and transfusions, along with demographic data, preoperative laboratory data, and surgical procedures performed. Independent predictors of IOBL and transfusion risk were identified using multivariable regression. RESULTS: Mean age at surgery was 57.0±13.6 years, 53.1% were male, and 67.1% were treated for metastatic lesions. Independent predictors of IOBL included en bloc resection (p<.001), surgical invasiveness (β=25.43 per point; p<0.001), and preoperative albumin (β=−244.86 per g/dL; p=0.011). Predictors of transfusion risk included preoperative hematocrit (odds ratio [OR]=0.88 per %; 95% confidence interval [CI, 0.84, 0.93]; p<0.001), preoperative MCHgb (OR=0.88 per pg; 95% CI [0.78, 1.00]; p=0.048), preoperative red cell distribution width (OR=1.32 per %; 95% CI [1.13, 1.55]; p<0.001), en bloc resection (OR=3.17; 95%CI [1.33, 7.54]; p=0.009), and surgical invasiveness (OR=1.08 per point; [1.06; 1.11]; p<0.001). The transfusion model showed a good fit of the data with an optimism-corrected area under the curve of 0.819. A freely available, web-based calculator was developed for the transfusion risk model (https://jhuspine3.shinyapps.io/TRUST/). CONCLUSIONS: Here we present the first clinical calculator for intraoperative blood loss and transfusion risk in patients being treated for primary or metastatic vertebral column tumors. Surgical invasiveness and preoperative microcytic anemia most strongly predict transfusion risk. The resultant calculators may prove clinically useful for surgeons counseling patients about their individual risk of requiring allogeneic transfusion.