运用倾向性评分方法探索真实世界苦碟子注射液治疗冠心病的疗效

Use propensity score method to analyze data about effectiveness of coronary heart disease treated by Kudiezi injection

该研究旨在评价真实世界苦碟子注射液(KDZ)治疗冠心病结局的疗效。从全国18家三甲医院信息管理系统(hospital information system,HIS)中提取数据,共获得5 597例诊断为"冠心病"的患者住院信息,其中4 040例为使用KDZ患者,1 557例为未使用KDZ的患者。基于generalized boosted models(GBM)倾向评分加权法,平衡大量混杂因素偏倚,同时使用3种Logistic回归分析对比分析平衡后结果。2组之间已知的72个在组间有差异的混杂因素,如性别、用药剂量、用药疗程等得以平衡。使用KDZ和不使用KDZ 2组人群治疗冠心病的疗效("治愈")通过3种Logistic方法回归方法检验,系数均为负值,且检验P小于0.05,从统计学分析上来说使用KDZ较之不使用KDZ有利于治疗冠心病。KDZ治疗冠心病有一定的疗效。倾向性评分方法是去除已知的混杂因素的有效方法。由于本研究是一种回顾性分析方法,且倾向性评分方法无法对未知混杂因素进行分析,故分析结果有一定局限性。

Objective： Real world clinical research to evaluate the effect on coronary heart disease （CHD） of parenterally ad- ministered Kudiezi. Method： Eighteen nationwide general hospitals＇ hospital information systems （HIS） were searched for information on in patients treated with parenterally administered Kudiezi. Evaluation used generalized boosted models （GBM）, and three types of logistic regression based on adjusted propensity score to balance confounding factors. Result： Two groups of coronary heart patients were compared, one group received parenterally administered Kudiezi and one did not. Seventy two confounding factors were identified and balanced by GBM. The three methods of logistic regression demonstrated that patients receiving parenterally administered Kudiezi had improved outcomes compared to those who did not receive the drug and this was statistically significant. Conclusion：Parenterally administered Kudiezi is a potential adjuvant therapy in the treatment of coronary heart disease. Logistic regression based on adjusted propensity score can balance confounding factors in retrospective data analysis. However it＇s limitation is that it is not applicable in pro- spective studies.