基于两水平logistic随机效应模型的乳腺癌术后并发症影响因素分析  

Analysis of influencing factors of postoperative complications of breast cancer based on two-level logistic random effects model

作　　者：刘梦洋 田园 安宁 张丽媛 魏丽 刘美娜[1] Liu Mengyang;Tian Yuan;An Ning;Zhang Liyuan;Wei Li;Liu Meina(Department of Biostatistics,Harbin Medical University,Harbin 150081,China)

机构地区：[1]哈尔滨医科大学公共卫生学院卫生统计教研室,黑龙江哈尔滨150081

出　　处：《中国医院统计》2021年第3期200-205,共6页Chinese Journal of Hospital Statistics

基　　金：国家自然科学基金(81573255)。

摘　　要：目的利用两水平logistic随机效应模型,基于医院水平和患者水平分析乳腺癌术后并发症的影响因素,为医院乳腺癌治疗质量的提高、改善患者预后提供依据。方法制定乳腺癌患者病案信息调查表,收集乳腺癌患者病例信息;利用t检验或χ^(2)检验对乳腺癌术后并发症的相关因素进行单因素分析;以医院作为水平2单位,患者作为水平1单位,患者是否发生术后并发症为结局变量,并发症的影响因素为解释变量,纳入单因素分析结果有统计学意义的因素,利用两水平logistic随机效应模型进行多因素分析。结果收集患者3224例,乳腺癌术后并发症发生率为22.52%。两水平logistic随机效应模型结果显示,医院水平随机效应在零模型中有统计学意义(P<0.05);医院水平的解释变量中,医院类型(OR=0.130,95%CI:0.027~0.638)具有统计学意义;患者水平的解释变量中,年龄(OR=0.983,95%CI:0.972~0.994)、居住地(OR=0.671,95%CI:0.534~0.843)、术前细胞学检查(OR=1.973,95%CI:1.397~2.787)以及腋窝淋巴结转移(OR=1.435,95%CI:1.168~1.762)均有统计学意义(P<0.05)。结论年龄越小、居住在城市的女性是乳腺癌术后发生并发症的高危人群;乳腺癌患者的术后治疗及护理过程中,应对有淋巴结转移、接受过术前细胞学检查的乳腺癌患者加以关注,降低术后并发症的发生几率。Objective To analyze the influencing factors of postoperative complications of breast cancer based on hospital level and patient level with the use of the two-level logistic random effect model,so as to provide a basis for improving the quality of breast cancer treatment in the hospital and improving the prognosis of patients.Methods The information questionnaire was formulated of breast cancer patients′medical records,and the information of breast cancer patients′cases was collected;univariate analysis was performed on the related factors of postoperative complications of breast cancer by t-test or chi-square test.The hospital as the two-level unit,the patient as the one-level unit,whether patients with postoperative complications as outcome variables,and the influence factors of complications as the explained variable,factors with statistical significance in univariate analysis included,two-level logistic random effect model was used for multivariate influencing factor analysis.Results Totally 3224 patients were collected;the incidence of postoperative complications of breast cancer was 22.52%;the results of two-level Logistic random effect model showed that the hospital level random effect had statistical significance in the zero model(P<0.05).Among the explanatory variables at hospital level,hospital type(OR=0.130,95%CI:0.027-0.638)was statistically significant.Among the explanatory variables for patient level,age(OR=0.983,95%CI:0.972-0.994),residence(OR=0.671,95%CI:0.534-0.843),preoperative cytological e xamination(OR=1.973,95%CI:1.397-2.787)and axillary lymph node metastasis(OR=1.435,95%CI:1.168-1.762)were statistically significant(P<0.05).Conclusion The younger the age,and women living in the city are the high-risk group of complications after breast cancer;in the process of postoperative treatment and nursing of breast cancer patients,attention should be paid to breast cancer patients with lymph node metastasis and those who have received preoperative cytological examination,reduce the probability of

关 键 词：乳腺癌 影响因素 随机效应模型 并发症 

分 类 号：R195.1[医药卫生—卫生统计学]