中国中西医结合杂志2025，Vol.45Issue(3)：283-289,7.DOI:10.7661/j.cjim.20250209.069
倾向性评分和疾病风险评分匹配法控制中医治疗艾滋病真实世界研究混杂因素
Propensity Score Matching and Disease Risk Score Methods to Control Confounding Factors in Real-World Studies of Chinese Medicine Treatment for AIDS
摘要
Abstract
Objective To compare the effect of propensity score(PS)and disease risk score(DRS)matching methods on controlling confounding factors in the real world study(RWS)of Chinese medicine for AIDS clinical efficacy evaluation,and to provide a reference for similar research.Methods In the database of acquired immune deficiency syndrome(AIDS)conventional treatment,AIDS patients who were transferred to second-line antiretroviral therapy(ART)from May 2010 to May 2016 were selected.They were assigned into Chinese medicine(CM)group and non-CM group,and the two groups were matched 1∶1 by PS and DRS method respectively,the matched patients were followed up for 10 years.Cox regression model was used to analyze the factors affecting the death of AIDS patients and the effect of two methods to control confounding factors was compared.Results Totally 1 220 AIDS patients were included in each group after PS and DRS matching,and the distribution of PS and DRS was consistent in the two groups.After PS matching,the death density was 1.47/100 person-years in the CM group and 2.05/100 person-years in the non-CM group(P＜0.05).After DRS matching,the death density was 1.46/100 person-years in the CM group and 2.10/100 person-years in the non-CM group(P＜0.05).After PS matching,the Akaike information criterion(AIC)and Bayesian information criterions(BIC)values of the Cox regression model were 2 374.819 and 2431.682,respectively.The results of the multivariate analysis showed that the risk of death for patients who received CM treatment was 0.73 times that of those who did not(95％CI:0.54-0.98).Additionally,gender,age(51～60 years),duration of ART(3～6 years),and CD4+cell count(＞500 cells/L)were identified as influencing factors for the death of AIDS patients(P＜0.05).After DRS matching,the AIC and BIC values of the Cox regression model were 2 384.621 and 2 441.792,respectively.The results of the multivariate analysis showed that the risk of death for patients who received CM treatment was 0.76 times that of those who did not(95％CI:0.55-1.04).Gender,age(51～60 years),marital status,educational level,duration of HIV positivity(＞8 years),duration of ART(3～6 years),CD4+cell count(＞350 cells/L),and whether platelet count was abnormal were identified as influencing factors for the death of AIDS patients(P＜0.05).Conclusions Under the same matching model,the balance of covariates between groups after PS matching is better than that after DRS matching.There is little difference in the evaluation results of clinical efficacy of CM between the two methods,and both can provide reliable statistical support for the efficacy evaluation of CM in the treatment of AIDS.关键词
倾向性评分/疾病风险评分/混杂因素/真实世界研究/艾滋病/中医Key words
propensity score/disease risk score/confounding factor/real world study/acquired immune deficiency syndrome/Chinese medicine引用本文复制引用
康秋佳,金艳涛,孙若寒,马秀霞,袁君,许前磊,郭会军,马彦民..倾向性评分和疾病风险评分匹配法控制中医治疗艾滋病真实世界研究混杂因素[J].中国中西医结合杂志,2025,45(3):283-289,7.基金项目
河南省中医药研究专项(No.2022JDZX080,No.2023ZXZX1160) （No.2022JDZX080,No.2023ZXZX1160）
国家自然科学基金资助项目(No.81803953) （No.81803953）
河南中医药大学"双一流"创建工程中医学学科项目(No.HSRP-DFCTCM-2023-4-01) （No.HSRP-DFCTCM-2023-4-01）
