Purpose: To develop a regression equation using preoperative variables to predict the likelihood of transfusion intra- or post- operatively for patients undergoing elective spinal fusion surgery.

Methods: This study was a retrospective chart review of patients (n=774; 51% female; mean age=56 years; mean BMI=29) who underwent elective spinal fusion surgery between February 2001 and May 2005 at the QEII Health Sciences Centre in Halifax, Nova Scotia. The variables recorded were incidence of allogeneic and autologous transfusions (intra- and post-operatively), body mass index, age, gender, number of spinal levels fused, preoperative hemoglobin and hematocrit, ASA grade and pedicle screw and bone graft use.

Results: Spinal fusion of 4 or more levels (n=24) resulted in a transfusion rate of 83%, therefore, the regression equation was calculated using only spinal fusions of 1 to 3 levels (n=750; transfusion rate=14%). The number of spinal levels fused and preoperative hemoglobin levels were found to be significantly related to the likelihood of receiving a transfusion. A predictive logistic regression equation was derived (Prediction = 3.615 + 1.018 x Levels fused–0.052 x Preop Hgb) where a probability cutoff of 0.160 gave a sensitivity of 63.5% and a specificity of 81.2%.

Conclusions: The high sensitivity and specificity of the regression equation suggest that the results are clinically useful for determining whether alternatives to allogeneic transfusions should be investigated preoperatively for each patient undergoing elective spinal fusion surgery. The results could also have economic implications because implementing the use of a cell salvage device would be less costly than allogeneic units if the patient required a transfusion.