Hans-Christoph Erben1, Florian Hess2, JoEllen Welter1,2, Nicole Graf3, Marc P Steurer4, Thomas A Neff5, Ralph Zettl2, Alexander Dullenkopf6. 1. Institute for Anesthesia and Intensive Care Medicine, Spital Thurgau, 8501, Frauenfeld, Switzerland. 2. Department of Orthopedic Surgery and Traumatology, Spital Thurgau, Frauenfeld, Switzerland. 3. Graf Biostatistics, Winterthur, Switzerland. 4. Department of Anesthesia and Perioperative Care, University of California, San Francisco, USA. 5. Department of Anesthesia and Intensive Care Medicine, Spital Thurgau, Münsterlingen, Switzerland. 6. Institute for Anesthesia and Intensive Care Medicine, Spital Thurgau, 8501, Frauenfeld, Switzerland. alexander.dullenkopf@stgag.ch.
Abstract
INTRODUCTION: Accurate identification of patients at risk of blood transfusion can reduce complications and improve institutional resource allocation. Probabilistic models are used to detect risk factors and formulate patient blood management strategies. Whether these predictors vary among institutions is unclear. We aimed to identify risk factors among our patients who underwent total hip (THA) or knee (TKA) arthroplasty, and combine these predictors to improve our model. MATERIALS AND METHODS: We retrospectively assessed risk factors among 531 adults who underwent elective THA or TKA from January 2016 to November 2018. Using relevant surgical and patient characteristics gathered from electronic medical records, we conducted univariable and multivariable analyses. For our logistic regression model, we measured the impact of independent variables (age, gender, operation type (THA or TKA) and preoperative hemoglobin concentration) on the need for a transfusion. RESULTS: Of the 531 patients, 321 had THA (uncemented) and 210 had TKA. For the selected period, our transfusion rate of 8.1% (10.6% THA and 4.3% TKA) was low. Univariable analyses showed that lower BMI (p < 0.001) was associated with receiving a transfusion. Important factors identified through logistic regression analyses were age (estimated effect of an interquartile range increase in age: OR 3.89 [CI 95% 1.96-7.69]), TKA (OR - 0.77 [CI 95% - 1.57-0.02]), and preoperative hemoglobin levels (estimated effect of interquartile range increase in hemoglobin: OR 0.47 [CI 95% 0.31-0.71]). Contrary to findings from previous reports, gender was not associated with transfusion. CONCLUSIONS: Previously published predictors such as advanced age, low preoperative hemoglobin, and procedure type (THA) were also identified in our analysis. However, gender was not a predictor, and BMI showed the potential to influence risk. We conclude that, when feasible, the determination of site-specific transfusion rates and combined risk factors can assist practitioners to customize care according to the needs of their patient population. LEVEL OF EVIDENCE: Level 3, retrospective cohort study.
INTRODUCTION: Accurate identification of patients at risk of blood transfusion can reduce complications and improve institutional resource allocation. Probabilistic models are used to detect risk factors and formulate patient blood management strategies. Whether these predictors vary among institutions is unclear. We aimed to identify risk factors among our patients who underwent total hip (THA) or knee (TKA) arthroplasty, and combine these predictors to improve our model. MATERIALS AND METHODS: We retrospectively assessed risk factors among 531 adults who underwent elective THA or TKA from January 2016 to November 2018. Using relevant surgical and patient characteristics gathered from electronic medical records, we conducted univariable and multivariable analyses. For our logistic regression model, we measured the impact of independent variables (age, gender, operation type (THA or TKA) and preoperative hemoglobin concentration) on the need for a transfusion. RESULTS: Of the 531 patients, 321 had THA (uncemented) and 210 had TKA. For the selected period, our transfusion rate of 8.1% (10.6% THA and 4.3% TKA) was low. Univariable analyses showed that lower BMI (p < 0.001) was associated with receiving a transfusion. Important factors identified through logistic regression analyses were age (estimated effect of an interquartile range increase in age: OR 3.89 [CI 95% 1.96-7.69]), TKA (OR - 0.77 [CI 95% - 1.57-0.02]), and preoperative hemoglobin levels (estimated effect of interquartile range increase in hemoglobin: OR 0.47 [CI 95% 0.31-0.71]). Contrary to findings from previous reports, gender was not associated with transfusion. CONCLUSIONS: Previously published predictors such as advanced age, low preoperative hemoglobin, and procedure type (THA) were also identified in our analysis. However, gender was not a predictor, and BMI showed the potential to influence risk. We conclude that, when feasible, the determination of site-specific transfusion rates and combined risk factors can assist practitioners to customize care according to the needs of their patient population. LEVEL OF EVIDENCE: Level 3, retrospective cohort study.
Authors: Patrick Meybohm; Toby Richards; James Isbister; Axel Hofmann; Aryeh Shander; Lawrence Tim Goodnough; Manuel Muñoz; Hans Gombotz; Christian Friedrich Weber; Suma Choorapoikayil; Donat R Spahn; Kai Zacharowski Journal: Transfus Med Rev Date: 2016-05-28
Authors: A Shander; H Van Aken; M J Colomina; H Gombotz; A Hofmann; R Krauspe; S Lasocki; T Richards; R Slappendel; D R Spahn Journal: Br J Anaesth Date: 2012-05-24 Impact factor: 9.166