老年股骨转子间骨折患者术后输血的相关因素△

Factors related to postoperative blood transfusion of femoral intertrochanter fractures in elderly

摘要:

[目的]分析老年转子间骨折术后输血的危险因素，建立预测模型并对其验证。[方法]回顾性分析本院 2016 年 12 月—2021 年 12 月收治的老年转子间按骨折共 358 例患者临床资料。按术后是否输血，分为输血组和非输血组，采用单因素和多因素分析得出术后输血的危险因素。建立预测模型，对模型效能进行评定。[结果] 358 例患者中，共 168 例术后输血，占 46.9%。单项因素比较显示，输血组患者较非输血组年龄更大 [(80.7±8.5) 岁 vs (76.6±8.7) 岁, P＜0.001]、BMI 更小 [(20.7±3.3) kg/ m² vs (21.5±3.7) kg/m² , P=0.015]、高血压占比更高 [(105/63) vs (93/97), P=0.010]、骨折类型更严重 [A1/A2/A3, (40/93/35) vs (80/83/ 27), P＜0.001]、手术时间更长 [(122.9±56.6) min vs (101.4±30.5) min, P＜0.001]、术中出血量更多 [(258.3±218.9) ml vs (130.5±94.9) ml, P＜0.001]、术前 HB 更低 [(93.1±14.8) g/L vs (111.6±14.2) g/L, P＜0.001]、术前 Alb 更低 [(34.7±8.2) g/L vs (36.8±4.5) g/L, P= 0.002]，差异均有统计学意义。逻辑回归分析显示：年龄 （OR=1.053, P=0.007）、手术时间 （OR=1.008, P=0.025）、术中出血量 （OR=1.007, P＜0.001） 是术后输血发生的独立危险因素；而 BMI （OR=0.900, P=0.025） 和术前血红蛋白 （OR=0.904, P＜0.001） 是术后输血的独立保护因素。基于逻辑回归结果所得的术后输血预测方程曲线下面积 AUC 为 0.894，95%CI 0.862~0.927，最佳临界值为 0.578，敏感度为 74.4%，特异度为 88.4%。[结论]高龄、手术时间长、术中出血多、低 BMI 和低血红蛋白是术后发生输血的独立危险因素。术后输血预测模型有助于临床判断输血发生风险并提前预防。

Abstract:

[Objective] To analyze the risk factors of postoperative blood transfusion of femoral intertrochanteric fracture in the elderly, and an established prediction equation was verified. [Methods] A retrospective study was done on 358 elderly patients who received surgi- cal treatment for femoral intertrochanteric fractures in our hospital from December 2016 to December 2021. The patients were divided into transfusion group and non-transfusion group according to whether postoperative blood transfusion occurred. The univariate comparisons and multi-factor binary logistic regression analysis were conducted to search the relative factors, and then a prediction model was estab- lished with its efficiency evaluated. [Results] Among the 358 patients, 168 (46.9%) received postoperative blood transfusion. In term of uni- variate comparison, the transfusion group were significantly older in age [(80.7±8.5) years vs (76.6±8.7) years, P＜0.001], lower BMI [(20.7± 3.3) kg/m² vs (21.5±3.7) kg/m² , P=0.015], higher prevalence of hypertension [(105/63) vs (93/97), P=0.010], and more severe fracture types [A1/A2/A3, (40/93/35) vs (80/83/27), P＜0.001], longer operation time [(122.9±56.6) min vs (101.4±30.5) min, P＜0.001], more intraopera- tive blood loss [(258.3±218.9) ml vs (130.5±94.9) ml, P＜0.001], lower preoperative Hb [(93.1±14.8) g/L vs (111.6±14.2) g/L, P＜0.001], and lower preoperative Alb [(34.7±8.2) g/L vs (36.8±4.5) g/L, P=0.002] than the non-transfusion group, and all abovesaid were statistically sig- nificant between the two groups. As results of logistic regression analysis, the age (OR=1.053, P=0.007), operation time (OR=1.008, P= 0.025), intraoperative blood loss (OR=1.007, P＜0.001) were the independent risk factors for postoperative transfusion, while the BMI (OR= 0.900, P=0.025) and preoperative hemoglobin (OR=0.904, P＜0.001) were the independent protective factor. Based on the logistic regres- sion results, a prediction equation was established, which proved area under curve (AUC) of 0.894, 95%CI 0.862~0.927，with cutoff of 0.578, the sensitivity of 74.4%, and the specificity of 88.4% by the ROC analysis. [Conclusion] The advanced age, longer operation time,more intraoperative bleeding, lower BMI and lower hemoglobin are independent risk factors for postoperative blood transfusion. The predic- tion model of postoperative blood transfusion might be helpful to judge the risk of transfusion and prevent it in advance.