Abstract:

Objective　To construct a clinical blood transfusion prediction model for patients with dangerous placenta previa, to verify its prediction efficacy, and to guide the decision of clinical blood transfusion. Methods　A retrospective analysis was performed on 200 patients with dangerous placenta previa who were hospitalized in Northwest Women's and Children's Hospital from January 2021 to December 2023, including 90 patients in the blood transfusion group, who were (33.55±3.95) years old and (34.45±2.12) weeks pregnant, and 110 patients in the non-transfusion group, who were (32.50±4.21) years old and (35.27±1.85) weeks pregnant. The clinical data and pregnancy outcomes were compared between the two groups. The logistic regression algorithm was used to screen out the independent risk factors related to blood transfusion. A nomogram prediction model was constructed by R language, and the performance of the model was evaluated. t, χ², and rank sum tests were applied. Results　There were no statistical differences in age, body mass index, pregnant times, abortion history, hypertension, diabetes, thyroid function, and newborn's body weight between the two groups (all P>0.05). The proportions of the patients with pregnant times ≥3, prenatal anemia, and placenta implantation, ultrasound score, postpartum bleeding volume, and hospital stay in the transfusion group were higher than those in the non-transfusion group [5.6% (5/90) vs. 2.8% (3/110), 60.0% (54/90) vs. 30.0% (33/110), 90.0% (81/90) vs. 32.7% (36/110), 9 (8, 11) vs. 6 (5, 7), 1 800 (1 200, 2 500) ml vs. 600 (500, 800) ml, 8 (6, 12) d vs. 6 (5, 8.3) d]; the gestational weeks in the transfusion group was lower than that in the non-transfusion group [(34.45±2.12) weeks vs. (35.27±1.85) weeks]; there were statistical differences (χ²=23.660, 18.120, and 66.880; Z=8.080, 10.590, and 3.730; t=-2.820; all P<0.05). The multivariate logistic regression analysis showed that postpartum hemorrhage, prenatal anemia, placenta implantation, and placenta ultrasound score were independent risk factors for blood transfusion in the patients (all P<0.05). A nomogram prediction model was constructed; the area under the curve was 0.96 (95%CI 0.939-0.986), indicating that the model had good prediction performance. Conclusions　Postpartum hemorrhage, prenatal anemia, placenta implantation, and placenta ultrasound score are independent risk factors leading to blood transfusion in patients with dangerous placenta previa. The constructed nomogram prediction model can predict the patients' blood transfusion needs, and has high clinical decision-making value.

Key words:

Postpartum hemorrhage, Dangerous placenta previa, Placental implantation, Blood transfusion, Prediction model

摘要：

目的　构建凶险性前置胎盘（PPP）患者临床输血预测模型并检验其预测效能，为临床输血提供决策。方法　回顾性分析，选取2021年1月至2023年12月在西北妇女儿童医院住院分娩的PPP患者200例作为研究对象，根据输血情况分为输血组（90例）和未输血组（110例）。输血组年龄（33.55±3.95）岁，孕周（34.45±2.12）周；未输血组年龄（32.50±4.21）岁，孕周（35.27±1.85）周。对比两组患者临床资料及妊娠结局，采用logistic回归法筛选出输血相关独立危险因素，通过R语言构建列线图预测模型，并对模型进行性能评价。统计学方法采用t检验、χ²检验、秩和检验。结果　两组患者年龄、体质量指数、孕次、流产史、高血压、糖尿病、甲状腺功能异常及新生儿体质量比较，差异均无统计学意义（均P>0.05）。输血组剖宫产次数≥3次、产前贫血、胎盘植入占比及胎盘超声评分、产后出血量均高于非输血组［5.6%（5/90）比2.8%（3/110）、60.0%（54/90）比30.0%（33/110）、90.0%（81/90）比32.7%（36/110）、9（8，11）分比6（5，7）分、1 800（1 200，2 500）ml比600（500，800）ml］，住院天数长于非输血组［8（6，12）d比6（5，8.3）d］，孕周低于未输血组［（34.45±2.12）周比（35.27±1.85）周］，差异均有统计学意义（χ²=23.660、18.120、66.880，Z=8.080、10.590、3.730，t=-2.820；均P<0.05）。两组患者1 min阿氏评分、5 min阿氏评分、10 min阿氏评分比较，差异均有统计学意义（Z=4.210、3.730、2.960，均P<0.05）。多因素logistic回归分析显示，产后出血、产前贫血、胎盘植入、胎盘超声评分均是导致PPP患者输血的独立危险因素（均P<0.05）。构建列线图预测模型，曲线下面积为0.963（95%CI 0.939～0.986），表明模型具有较好的预测性能。结论　产后出血、产前贫血、胎盘植入、胎盘超声评分是导致PPP患者输血的独立危险因素，构建列线图预测模型能较好预测患者输血需求，具有较高的临床决策价值。

关键词:

产后出血, 凶险性前置胎盘, 胎盘植入, 输血, 预测模型