245. Predicting massive intraoperative blood loss in adult spinal deformity surgery

Abstract

BACKGROUND CONTEXT Due to inherent patient and surgical factors, some adult spinal deformity patients are at higher risk of larger blood loss. This is associated with increased risk of complications, coagulopathy, and higher requirements for blood component replacement intraoperatively. The medical team should be aware of expected blood loss in order to adequately plan patient monitoring and blood product requirements. PURPOSE This study aims to develop and validate a model based on patient characteristics and surgical strategies to predict which patients undergoing adult spinal deformity surgery are at greater risk of massive intra-operative blood loss. STUDY DESIGN/SETTING Retrospective analysis of a multicenter prospective database. PATIENT SAMPLE Surgical ASD patients undergoing long fusions (>5 levels). OUTCOME MEASURES Massive intraoperative blood loss (>7% of a patient's ideal body weight), ICU stay, complications. METHODS Surgical adult spinal deformity patients undergoing long fusions (> 5 levels) were analyzed. Massive blood loss was defined as an intraoperative blood loss over 7% of a patients’ ideal body weight, in accordance to the anesthesia literature. The impact of mEBL on complications and ICU stay was assessed through multivariate poisson and logistic regression, adjusting for baseline patient demographics and magnitude of surgery. Multivariate logistic regression modeling was used to build a prediction model of mEBL. Potential predictors were identified using univariate analysis. The model was built using a combination of backward elimination and bootstrap selection. Model fit was assessed using the Hosmer-Lemeshow test and the ROC curve. Split sample internal cross-validation was performed. RESULTS A total of 1,205 patients met inclusion criteria, and mEBL occurred in 7.8% (94/1205) of cases. In this cohort, mEBL was an independent risk factor for major complications (IRR 2.89, p=0.0001), and postoperative ICU stay (OR 4.68, p=0.001). Predictors of mEBL included: revision surgery (OR 1.66, p=0.05), performing a three-column osteotomy (OR 2.65, p=0.0001), number of levels fused (OR 1.19, p=0.0001), number of levels decompressed (OR 1.15, p=0.004), and number of LLIF interbodies (OR 0.69, p=0.066). The predictive model had a good discrimination (ROC curve 0.76), and good calibration (Hosmer-Lemeshow p=0.39), which held true when performing split-sample cross-validation. CONCLUSIONS We built and validated a model to predict massive intra operative blood loss in ASD surgery. In this model, baseline patient demographics and elements of the surgical strategy are used to predict which patients are at greater risk for massive blood loss. This tool would be of particular help to the anesthesia team, allowing to identify preoperatively which patients are at greater risk and plan intraoperative care and fluid management accordingly. FDA DEVICE/DRUG STATUS This abstract does not discuss or include any applicable devices or drugs.