A. J. Acuña, L. T. Samuel, A. K. Emara, A. F. Kamath, Department of Orthopaedic Surgery, Cleveland Clinic Foundation, Cleveland, OH, USA.
D. Grits, Cleveland Clinic Lerner College of Medicine, Cleveland, OH, USA.
Clin Orthop Relat Res. 2021 Mar 1;479(3):589-600. doi: 10.1097/CORR.0000000000001513.
Given the morbidity, mortality, and financial burden associated with venous thromboembolism (VTE) after TKA, orthopaedic providers continually seek to identify risk factors associated with this devastating complication. The association between perioperative transfusion status and VTE risk has not been thoroughly explored, with previous studies evaluating this relationship being limited in both generalizability and power.
QUESTIONS/PURPOSES: Therefore, we sought to determine whether perioperative transfusions were associated with an increased risk of (1) pulmonary embolism (PE) or (2) deep vein thrombosis (DVT) after primary TKA in a large, multi-institutional sample.
The American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP) database was implemented for our analysis. The definitions of complications, such as DVT and PE, and risk adjustment validation is monitored by the central ACS NSQIP office to ensure participating hospitals are adhering to the same guidelines to log patients. Additionally, both preoperative and intraoperative/72 hour postoperative transfusion status is included for all patients. Therefore, ACS NSQIP was determined to be the most appropriate database for our analysis. All patients who underwent primary TKA between 2011 and 2018 were identified using Current Procedural Terminology code 27447. Primary TKAs designated as "non-elective" were excluded, thereby providing a cohort composed solely of patients undergoing unilateral primary elective TKA for further analysis. The final analysis included 333,463 patients undergoing TKA (mean age 67 ± 9 years, 62% female). Preoperative transfusions were received by < 0.01% (48 of 333,463) of the patients, while 4% (14,590 of 333,463) received a transfusion within the interim between the start of surgery up to 72 hours postoperatively. All missing values were imputed through multiple imputation by chained equation to avoid variable availability-based selection and the subsequent listwise deletion-associated bias in the estimate of parameters. A multivariable logistic regression analysis was conducted using variables identified in a univariate model to calculate adjusted odds ratios and 95% confidence intervals for risk factors associated with symptomatic DVT and/or PE. For variables that maintained significance in the multivariable model, an additional model without confounders was used to generate fully adjusted ORs and 95% CIs. A propensity score matched comparison between recipients versus nonrecipients (1:1) of transfusion (preoperative and intraoperative/72 hours postoperative) was then conducted to evaluate the independent association between DVT/PE development and patients' transfusion status. Significance was determined at a p value < 0.05.
Adjusted multivariable regression analysis accounting for patient age, sex, race, BMI, American Society of Anesthesiologists (ASA) class and baseline comorbidities demonstrated the absence of an association between preoperative (OR 1.75 [95% CI 0.24 to 12.7]; p = 0.58) or intraoperative/72 hours postoperative (OR 1.12 [95% CI 0.93 to 1.35]; p = 0.23) transfusions and higher odds of developing PE. Similar findings were demonstrated after propensity score matching. Although multivariable regression demonstrated the absence of an association between preoperative transfusion and the odds of developing DVT within the 30-day postoperative period (OR 1.85 [95% CI 0.43 to 8.05]; p = 0.41), intraoperative/postoperative transfusion was associated with higher odds of DVT development (OR 3.68 [95% CI 1.14 to 1.53]; p < 0.001) relative to transfusion naïve patients. However, this significance was lost after propensity score matching.
After controlling for various potential confounding variables such as ASA Class, age, anesthesia type, and BMI, the receipt of an intra- or postoperative transfusion was found to be associated with an increased risk of DVT. Our findings should encourage orthopaedic providers to strictly adhere to blood management protocols, further tighten transfusion eligibility, and adjust surgical approach and implant type to reduce the incidence of transfusion among patients with other DVT risk factors. Additionally, our findings should encourage a multidisciplinary approach to VTE prophylaxis and prevention, as well as to blood transfusion guideline adherence, among all providers of the care team.
Level III, therapeutic study.
鉴于全膝关节置换术后静脉血栓栓塞症（VTE）的发病率、死亡率和经济负担，骨科医生一直在寻找与这种破坏性并发症相关的风险因素。围手术期输血状态与 VTE 风险之间的关系尚未得到充分探讨，之前评估这种关系的研究在普遍性和效力方面都受到限制。
问题/目的：因此，我们旨在确定在大型多机构样本中，全膝关节置换术后（1）肺栓塞（PE）或（2）深静脉血栓形成（DVT）是否与围手术期输血有关。
我们使用美国外科医师学会国家手术质量改进计划（ACS NSQIP）数据库进行分析。并发症（如 DVT 和 PE）的定义以及风险调整验证由 ACS NSQIP 中央办公室进行监控，以确保参与医院遵守相同的指南来记录患者信息。此外，所有患者的术前和术中/72 小时术后输血状态都包括在内。因此，ACS NSQIP 被确定为我们分析的最合适数据库。使用当前程序术语代码 27447 确定 2011 年至 2018 年间接受初次全膝关节置换术的所有患者。排除“非择期”的初次全膝关节置换术，从而提供一个仅由接受单侧初次择期全膝关节置换术的患者组成的队列进行进一步分析。最终分析包括 333463 例接受全膝关节置换术的患者（平均年龄 67±9 岁，62%为女性）。<0.01%（48/333463）的患者接受了术前输血，而 4%（14590/333463）在手术开始至术后 72 小时之间接受了输血。所有缺失值均通过链式方程的多次插补进行插补，以避免基于变量可用性的选择和参数估计的列表删除相关偏差。使用单变量模型中确定的变量进行多变量逻辑回归分析，计算与症状性 DVT 和/或 PE 相关的风险因素的调整后优势比和 95%置信区间。对于在多变量模型中保持显著的变量，使用没有混杂因素的额外模型生成完全调整后的 OR 和 95%CI。然后对输血（术前和术中/72 小时术后）接受者与非接受者（1:1）进行倾向评分匹配比较，以评估 DVT/PE 发展与患者输血状态之间的独立关联。以 p 值<0.05 为显著性水平。
多变量回归分析调整了患者年龄、性别、种族、BMI、美国麻醉医师协会（ASA）分级和基线合并症，结果表明术前（OR 1.75[95%CI 0.24 至 12.7]；p=0.58）或术中/72 小时术后（OR 1.12[95%CI 0.93 至 1.35]；p=0.23）输血与发生 PE 的几率无关。在倾向评分匹配后也得到了相似的发现。尽管多变量回归显示术前输血与术后 30 天内发生 DVT 的几率无关（OR 1.85[95%CI 0.43 至 8.05]；p=0.41），但术中/术后输血与 DVT 发展的几率较高相关（OR 3.68[95%CI 1.14 至 1.53]；p<0.001）与未输血的患者相比。然而，这种显著性在倾向评分匹配后消失了。
在控制了 ASA 分级、年龄、麻醉类型和 BMI 等各种潜在混杂变量后，发现术中或术后输血与 DVT 风险增加有关。我们的发现应该鼓励骨科医生严格遵守血液管理方案，进一步收紧输血适应证，并调整手术方法和植入物类型，以减少具有其他 DVT 风险因素的患者的输血发生率。此外，我们的发现应该鼓励所有治疗团队的提供者采取多学科方法来预防和预防 VTE，以及遵守输血指南。
III 级，治疗性研究。
