中国全科医学 ›› 2018, Vol. 21 ›› Issue (35): 4354-4357.DOI: 10.12114/j.issn.1007-9572.2018.00.066

• 专题研究 • 上一篇    下一篇

播散性马尔尼菲蓝状菌病并发消化道出血预测模型的构建

陈涛1,蒋忠胜1,兰慧慧2*,李敏基1,胡家光1,覃川1,莫胜林1   

  1. 1.545006广西壮族自治区柳州市人民医院感染病科 2.545005广西壮族自治区柳州市,广西壮族自治区龙潭医院呼吸内科
    *通信作者:兰慧慧,住院医师;E-mail:lhh_2017@126.com
  • 出版日期:2018-12-15 发布日期:2018-12-15
  • 基金资助:
    基金项目:广西卫生厅自筹课题(Z2016799)

Establishment of a Predictive Model for Disseminated Talaromycosis Marneffei Complicated by Gastrointestinal Hemorrhage 

CHEN Tao1,JIANG Zhongsheng1,LAN Huihui2*,LI Minji1,HU Jiaguang1,QIN Chuan1,MO Shenglin1   

  1. 1.Department of Infectious Diseases,Liuzhou General Hospital,Liuzhou 545006,China
    2.Department of Respiratory Medicine,Longtan Hospital of Guangxi Zhuang Autonomous Region,Liuzhou 545005,China
    *Corresponding author:LAN Huihui,Resident physician;E-mail:lhh_2017@126.com
  • Published:2018-12-15 Online:2018-12-15

摘要: 目的 分析播散性马尔尼菲蓝状菌病(DTSM)并发消化道出血的独立危险因素,并建立预测模型。方法 选取2009年1月—2017年11月柳州市人民医院收治的DTSM初治患者63例,将其分为无消化道出血组(34例)和消化道出血组(29例)。回顾性分析DTSM患者的临床资料,包括基本资料、临床资料、合并症、实验室检查、影像学资料等。采用单因素Logistic回归分析及多因素Logistic回归分析DTSM并发消化道出血的独立危险因素,并建立预测模型,采用受试者工作特征(ROC)曲线分析模型的预测价值。结果 多因素Logistic回归分析显示,血小板计数(PLT)〔β=-0.037,OR=0.946,95%CI(0.937,0.979)〕、凝血酶原时间(PT)〔β=0.376,OR=1.457,95%CI(1.082,1.961)〕是DTSM并发消化道出血的独立危险因素(P<0.05)。预测模型为P=1/(1+e-y),Y=-3.612+0.376×PT-0.037×PLT,其中P为患者的出血概率,Y为预测指数。Hosmer-Lemeshow检验结果显示,模型拟合度好(χ2=4.757,df=8,P=0.783)。ROC曲线下面积(AUC)为0.918〔95%CI(0.815,0.985),P<0.001〕,当P值的最佳临界值为0.539 3,即P≥0.539 3预测并发消化道出血,<0.539 3预测不并发消化道出血时,模型预测的灵敏度为86.21%,特异度为94.12%,阳性预测值为92.59%,阴性预测值为88.89%,总正确率为90.48%。结论 PLT、PT可用于预测模型的构建。DTSM并发消化道出血预测模型能够较准确地预测DTSM患者是否并发消化道出血。

关键词: 播散性马尔尼菲蓝状菌病, 胃肠出血, 血小板, 凝血酶原时间, 危险因素

Abstract: Objective To analyze the independent risk factors for disseminated talaromycosis marneffei(DTSM) complicated by gastrointestinal hemorrhage and to establish a predictive model.Methods Between January 2009 and November 2017,63 patients with DTSM initial treatment who were treated at Liuzhou General Hospital were selected and divided into two groups:a group with gastrointestinal hemorrhage(29 cases) and a group without gastrointestinal hemorrhage(34 cases).A retrospective analysis of the clinical data of the DTSM patients included basic data,clinical data,comorbidities,laboratory tests,and imaging data.Univariate and multivariate Logistic regression analyses were used to screen independent risk factors for DTSM complicated by gastrointestinal hemorrhage.A prediction model was established,and its predictive value was analyzed via the receiver operating characteristic(ROC) curve method.Results Multivariate Logistic regression analysis showed that platelet count(PLT)〔β=-0.037,OR=0.946,95%CI(0.937,0.979)〕 and prothrombin time(PT)〔β=0.376,OR=1.457,95%CI(1.082,1.961)〕 were independent risk factors for DTSM with gastrointestinal hemorrhage(P<0.05).The prediction model was P=1/(1+e-y),and Y=-3.612+0.376×PT-0.037×PLT,where P denotes the probability of hemorrhage,Y denotes the predictive index.The Hosmer–Lemeshow test results showed an ideal goodness of fit of the model(χ2=4.757,df=8,P=0.783).The area under the ROC curve(AUC) was 0.918〔95%CI(0.815,0.985),P<0.001〕.The P value cut-off was 0.539 3,thus indicating that P≥0.539 3 predicted complication of gastrointestinal hemorrhage,and P<0.539 3 predicted its absence.The sensitivity,specificity,positive predictive value,negative predictive value,and total accuracy of model prediction were 86.21%,94.12%,92.59%,88.89%,and 90.48%,respectively.Conclusion PLT and PT can be used to build a model for accurate prediction of the incidence of DTSM with gastrointestinal hemorrhage.

Key words: Disseminated talaromycosis marneffei, Gastrointestinal hemorrhage, Blood platelets, Prothrombin time, Risk factors