艾滋病合并非皮疹型马尔尼菲篮状菌感染临床诊断预测模型的建立及验证  

Establishment of a model for clinical diagnosis and prediction of non-rash Talaromyces marneffei infection in AIDS patients and its validation

作　　者：胡家光[1,2] 李旭 邬佩云 覃凌[1,2] 唐贵 黄小红[1,2] 莫胜林 张鹏 蒋忠胜 HU Jia-guang;LI Xu;WU Pei-yun;QIN Ling;TANG Gui;HUANG Xiao-hong;MO Sheng-lin;ZHANG Peng;JIANG Zhong-sheng(The People's Hospital of Liuzhou,Liuzhou,Guangi 545006,China;不详)

机构地区：[1]柳州市人民医院感染性疾病科,广西柳州545006 [2]柳州市传染病免疫研究重点实验室,广西柳州545006 [3]广西艾滋病防治研究重点实验室,广西南宁530021

出　　处：《中华医院感染学杂志》2023年第5期657-662,共6页Chinese Journal of Nosocomiology

基　　金：十三五国家重大科技专项基金资助项目(2018ZX103020104);广西卫健委自筹课题基金资助项目(Z20190034,Z20200476,Z20200504);广西艾滋病防治研究重点实验室开放课题基金资助项目(gklapt201904);柳州市科技计划基金资助项目(2022SB007)。

摘　　要：目的建立艾滋病(AIDS)合并非皮疹型马尔尼菲篮状菌(TM)感染的临床诊断预测模型并进行验证。方法选择2012年1月-2018年12月在柳州市人民医院感染科住院的AIDS患者376例为研究对象,其中155例合并TM感染,221例无TM感染。通过SPSS 25.0软件将376例研究对象按7∶3比例随机分为建模组、验模组。收集所有患者临床、实验室资料,单因素分析有统计学差异的因素进入多因素Logistic模型,最后采用列线图对多因素结果进行评分,在建模组及验模组中采用接受者操作特性(ROC)曲线、校准图检测模型效能。结果多因素Logistic回归分析显示,体质量下降、外周或腹腔淋巴结肿大、天门冬氨酸氨基转移酶(AST)升高和CD4+T淋巴细胞计数<50个/μl是TM感染诊断的预测因素(P<0.05);通过列线图建立临床诊断预测模型并对其进行验证,结果显示建模组、验模组ROC曲线下面积分别为0.813和0.808,提示该模型具有较好预测能力。结论在TM感染流行地区,体质量下降、外周或腹腔淋巴结肿大、AST升高、CD4+T淋巴细胞计数<50个/μl 4个因素组成的临床诊断模型可用于早期预测CD4+T淋巴细胞计数<200个/μl的AIDS患者是否合并非皮疹型TM感染的发生。OBJECTIVE To establish and validate the model for clinical diagnosis and prediction of non-rash Talaromyces marneffei(TM)infection in acquired immune deficiency syndrome(AIDS)patients.METHODS Totally 376 AIDS patients who were hospitalized in infection department of the People′s Hospital of Liuzhou from Jan 2012 to Dec 2018 were recruited as the research subjects,155 of whom were complicated with TM infection,and 221 were not complicated with TM infection.The 376 subjects were randomly divided into the model group and the validation group in a 7∶3 ratio by using SPSS 25.0 software.The clinical data and laboratory test data were collected from all of the patients.The multivariate logistic model was established based on the factors with significant differences through univariate analysis,the result of multivariate analysis was scored by means of nomogram.The efficiency of the model was detected for the model group and the validation group by receiver operating characteristic(ROC)curves and calibration chart.RESULTS The multivariate logistic regression analysis showed that the decline of body weight,enlargement of peripheral or abdominal lymph nodes,rise of aspartate aminotransferase(AST)and CD4+T lymphocytes counts less than 50 perμl were the predictive factors for diagnosis of TM infection(P<0.05).The clinical diagnosis and prediction model was established based on the nomogram,and the result of validation for the model showed that the area under curve(AUC)of the model group was 0.813,the validation group 0.808,indicating that the model had favorable predictive capability.CONCLUSION In the areas where TM infection is prevalent,the clinical diagnosis model established based on the four factors decline of body weight,enlargement of peripheral or abdominal lymph nodes,rise of AST and CD4+T lymphocytes counts less than 50 perμl can be used for prediction of the non-rash TM infection in the AIDS patients with CD4+T lymphocytes counts less than 200 perμl.

关 键 词：获得性免疫缺陷综合征 马尔尼菲篮状菌感染 临床诊断 预测模型 列线图 

分 类 号：R512.91[医药卫生—内科学]