中国全科医学 ›› 2021, Vol. 24 ›› Issue (2): 190-195.DOI: 10.12114/j.issn.1007-9572.2020.00.426

• 专题研究 • 上一篇    下一篇

获得性免疫缺陷综合征合并肺孢子菌肺炎患者预后模型的建立及评价

陈涛,蒋忠胜*,李敏基,莫胜林,张鹏,胡家光,覃锦玉,蒙达礼   

  1. 545006广西壮族自治区柳州市人民医院感染病科
    *通信作者:蒋忠胜,主任医师;E-mail:jiangzs1111@126.com
  • 出版日期:2021-01-15 发布日期:2021-01-15
  • 基金资助:
    “十三五”国家科技重大专项课题(2018ZX10302104-001-008);广西卫生厅自筹课题(Z2016799)

Establishment and Value Evaluation of Prognostic Model of Acquired Immunodeficiency Syndrome Patients Complicated with Pneumocystis Pneumonia 

CHEN Tao,JIANG Zhongsheng*,LI Minji,MO Shenglin,ZHANG Peng,HU Jiaguang,QIN Jinyu,MENG Dali   

  1. Department of Infectious Diseases,Liuzhou People's Hospital,Liuzhou 545006,China
    *Corresponding author: JIANG Zhongsheng,Chief physician;E-mail: jiangzs1111@126.com
  • Published:2021-01-15 Online:2021-01-15

摘要: 背景 肺孢子菌肺炎(PCP)是由肺孢子菌(PJ)感染引起的呼吸系统机会性感染,是获得性免疫缺陷综合征(AIDS)的指征性疾病。AIDS合并PCP患者多以亚急性发病为主,部分患者病情进展迅速、内科治疗效果差、短期内病死率高,因此对该类患者近期预后需进行准确预测,以便指导临床救治。目的 分析影响AIDS合并PCP患者预后危险因素,并建立预后预测模型。方法 选取2009年1月—2017年8月柳州市人民医院收治的AIDS合并PCP患者300例,通过SPSS 19.0统计学软件随机分组(建模组241例和验模组59例)。回顾性分析建模组患者的临床资料,包括基本资料、合并症、实验室检查等。采用单因素及多因素Logistic回归分析筛选出独立危险因素,建立预后模型,用验模组患者资料评价预后模型的预测能力。结果 多因素Logistic回归分析结果显示,清蛋白(ALB)〔OR=0.759,95%CI(0.595,0.967)〕、乳酸脱氢酶(LDH)〔OR=1.009,95%CI(1.003,1.015)〕、CD4+T淋巴细胞计数〔OR=0.878,95%CI(0.790,0.975)〕、肺泡-动脉血氧分压差〔P(A-a)O2〕〔OR=1.164,95%CI(1.073,1.262)〕是AIDS合并PCP患者的预后独立因素(P<0.05)。预测模型为P=1/(1+e-y),Y=-0.278-0.276×ALB-0.131×CD4+T淋巴细胞计数+0.009×LDH+0.152×P(A-a)O2,其中P为患者的恶化概率,Y为预测指数。Hosmer-Lemeshow检验结果显示,模型拟合度好(χ2=3.974,df=8,P=0.859)。受试者工作特征(ROC)曲线下面积(AUC)为0.986〔95%CI(0.970,1.000),P<0.001〕。验模组验证结果显示:所构建模型预测的特异度、灵敏度、阳性预测值、阴性预测值和总正确率分别为97.50%、89.47%、94.44%、95.12%和94.91%。结论 ALB、LDH、CD4+T淋巴细胞计数、P(A-a)O2作为独立的危险因素可用于预后判断模型的构建。本研究中所构建的预后模型能够较准确的预测AIDS合并PCP患者的近期预后。

关键词: 获得性免疫缺陷综合征, 肺孢子菌肺炎, 预测模型, 预后, Logistic回归分析, 特异度, 灵敏度

Abstract: Background Pneumocystis pneumonia (PCP),a respiratory opportunistic infection caused by pneumocystis,is an indicative disease of acquired immune deficiency syndrome (AIDS).AIDS with PCP occurs subacutely in most patients,and progresses rapidly in some patients,which has poor response to medical treatment,with short-term high mortality.Therefore,the recent prognosis of such patients should be accurately predicted in order to guide clinical treatment.Objective To identify the risk factors associated with the prognosis of AIDS patients with PCP,to develop a prognostic model for such patients.Methods From January 2009 to August 2017,300 cases of AIDS with PCP from Liuzhou People's Hospital were enrolled,and were randomized into modelling group (n=241) and validation group (n=59) with a grouping scheme developed using SPSS 19.0.The clinical data of the modelling group were analyzed retrospectively,including essential data,complication,laboratory examination,and so on.Univariate and multivariate Logistic regression analyses were used to screen independent prognostic risk factors to construct the prognostic model.The data of validation group were used to evaluate the predictive ability of the prognostic model.Results Multivariate Logistic regression analysis showed that the independent prognostic factors of AIDS with PCP were albumin (ALB)〔OR=0.759,95%CI(0.595,0.967)〕,lactate dehydrogenase(LDH)〔OR=1.009,95%CI(1.003,1.015)〕,CD4+T lymphocyte count 〔OR=0.878,95%CI(0.790,0.975)〕,alveolar-arterial gradient 〔P(A-a)O2〕〔OR=1.164,95%CI(1.073,1.262)〕(P<0.05).The algorithm for the prediction model was: P=1/(1+e-y),Y=-0.278-0.276×ALB-0.131×CD4++0.009×LDH+0.152×P(A-a)O2,(P refers to the deteriorative probability,and Y is the prognostic index).Hosmer-Lemeshow test results showed that the model fitting degree was good(χ2=3.974,df=8,P=0.859).The AUC for the model's predictive value was 0.986〔95%CI(0.970,1.000),P<0.001〕.In predicting the prognosis of the validation group,the model showed 97.50% specificity,89.47% sensitivity,94.44% positive predictive value,95.12% negative predictive value and 94.91% total accuracy.Conclusion As independent risk factors for AIDS with PCP,ALB,LDH,CD4+T lymphocyte count and P(A-a)O2 can be used to construct the prognostic model.The prognostic model in this study may accurately predict the short-term prognosis of such patients.

Key words: Acquired immune deficiency syndrome, Pneumocystis pneumonia, Prediction model, Prognosis, Logistic regression analysis, Specificity, Sensitivity