Vandromme Marianne J, Griffin Russell L, McGwin Gerald, Weinberg Jordan A, Rue Loring W, Kerby Jeffrey D
Section of Trauma, Burns, and Surgical Critical Care, University of Alabama at Birmingham School of Medicine, Birmingham, Alabama 35294, USA.
Am Surg. 2011 Feb;77(2):155-61.
Most retrospective studies evaluating fresh-frozen plasma:packed red blood cell ratios in trauma patients requiring massive transfusion (MT) are limited by survival bias. As prospective resource-intensive studies are being designed to better evaluate resuscitation strategies, it is imperative that patients with a high likelihood of MT are identified early. The objective of this study was to develop a predictive model for MT in civilian trauma patients. Patients admitted to the University of Alabama at Birmingham Trauma Center from January 2005 to December 2007 were selected. Admission clinical measurements, including blood lactate 5 mMol/L or greater, heart rate greater than 105 beats/min, international normalized ratio greater than 1.5, hemoglobin 11 g/dL or less, and systolic blood pressure less than 110 mmHg, were used to create a predictive model. Sensitivity (Sens), specificity (Spec), positive predictive value (PPV), and negative predictive value (NPV) were calculated for all possible combinations of clinical measurements as well as each measure individually. A total of 6638 patients were identified, of whom 158 (2.4%) received MT. The best-fit predictive model included three or more positive clinical measures (Sens: 53%, Spec: 98%, PPV: 33%, NPV: 99%). There was increased PPV when all clinical measurements were positive (Sens: 9%, Spec: 100%, PPV: 86%, NPV: 98%). All combinations or clinical measures alone yielded lower predictive probability. Using these emergency department clinical measures, a predictive model to successfully identify civilian trauma patients at risk for MT was not able to be constructed. Given prospective identification of patients at risk for MT remains an imprecise undertaking, appropriate resources to support these efforts will need to be allocated for the completion of these studies.
大多数评估创伤患者大量输血（MT）时新鲜冰冻血浆与红细胞输注比例的回顾性研究都受到生存偏差的限制。由于正在设计前瞻性资源密集型研究以更好地评估复苏策略，因此尽早识别出有高MT可能性的患者至关重要。本研究的目的是建立一个针对平民创伤患者MT的预测模型。选取了2005年1月至2007年12月入住阿拉巴马大学伯明翰创伤中心的患者。采用入院时的临床测量指标，包括血乳酸≥5毫摩尔/升、心率＞105次/分钟、国际标准化比值＞1.5、血红蛋白≤11克/分升以及收缩压＜110毫米汞柱，来建立预测模型。针对临床测量指标的所有可能组合以及每个指标单独计算敏感性（Sens）、特异性（Spec）、阳性预测值（PPV）和阴性预测值（NPV）。共识别出6638例患者，其中158例（2.4%）接受了MT。最佳拟合预测模型包括三项或更多阳性临床指标（Sens：53%，Spec：98%，PPV：33%，NPV：99%）。当所有临床测量指标均为阳性时PPV升高（Sens：9%，Spec：100%，PPV：86%，NPV：98%）。所有组合或单独的临床指标产生的预测概率较低。利用这些急诊科临床指标，未能构建出一个成功识别有MT风险的平民创伤患者的预测模型。鉴于对有MT风险患者的前瞻性识别仍然是一项不精确的工作，需要分配适当资源来支持这些研究的完成。
