Factors influencing perioperative blood transfusions in patients with gastrointestinal cancer

Background. Patients undergoing major cancer resections often receive blood transfusions (TFs). Preoperative erythropoietin (EPO) offers the rationale to reduce TFs and related morbidity. Methods. Perioperative TF information was collected prospectively in a single surgeon practice over 5 years. Results. Three hundred forty-four patients underwent a major procedure, including pancreatic (n = 130, 38%), hepatobiliary (n = 87, 25%), gastroesophageal (n = 69, 20%), and other operations (n = 58, 17%). Median estimated blood loss (EBL) was 375 mL. PRBC TFs were given in 83 cases (24%), at a median of 2 units [1-16]. TF frequency and EBL did not differ between diagnoses. Multivariate TF associations existed for Hgb (P < 0.0001, OR 0.335), EBL (P < 0.0001, OR 1.007), serum CI (P = 0.004, OR 1.25), serum Na (P = 0.02, OR 0.810), and age (P = 0.04, OR 1.033). TFs (versus no TFs) were linked to major complications (43 versus 20%, P = 0.0002), mortality (12% versus 3%, P = 0.001), and increased LOS (9 versus 7 days, P < 0.0001). A potential benefit for preoperative EPO to avoid TFs could be derived for only 31 patients (9%). Conclusions. In this low TF rate of 24% for major visceral resections, few preoperative parameters are able to identify subgroups at risk for TFs aside from blood counts. Our data would not support generalized preoperative EPO administration. (C) 2007 Elsevier Inc. All rights reserved.