不同混杂结构条件下各倾向性评分方法的模拟比较研究

被引:10
作者
孙婷
秦国友
武振宇
赵耐青
机构
[1] 复旦大学公共卫生学院生物统计学教研室和公共卫生安全教育部重点实验室
关键词
倾向性评分; 混杂因素; Monte Carlo模拟; 偏倚;
D O I
暂无
中图分类号
R195.1 [卫生统计学];
学科分类号
摘要
目的通过构建不同混杂结构的处理因素模型和结局模型、不同相关性的协变量,比较多种倾向性评分方法在结局模型为线性回归模型的情况下估计处理效应的优劣。方法采用Monte Carlo模拟方法,通过构建四种由简单到复杂的不同结构的混杂模型,生成相应的数据集,再分别应用倾向性评分匹配、回归调整、加权以及分层的方法估计处理效应并进行比较。评价指标包括点估计、标准误、相对偏倚、均方误差。结果在结局模型为线性回归模型情况下,倾向性评分回归调整法估计的相对偏倚最小,稳定性也最好。匹配法卡钳值取0.02较卡钳值取倾向性评分标准差的0.2倍估计的相对偏倚更小。当处理因素模型中含有非线性效应时,用逆概率加权法估计的偏倚较大,并且加权法估计的标准误也最大。倾向性评分分层法在各种情况下估计的相对偏倚都较大。结论倾向性评分回归调整法能够较好地估计处理效应,并且在各种情况下估计都较为稳健。建议当协变量与处理因素和结局变量的关系无法确定时,这四种方法中可以考虑优先使用回归调整法。
引用
收藏
页码:415 / 420
页数:6
相关论文
共 12 条
[1]   On variance estimate for covariate adjustment by propensity score analysis [J].
Zou, Baiming ;
Zou, Fei ;
Shuster, Jonathan J. ;
Tighe, Patrick J. ;
Koch, Gary G. ;
Zhou, Haibo .
STATISTICS IN MEDICINE, 2016, 35 (20) :3537-3548
[2]  
Estimating causal effects for multivalued treatments: a comparison of approaches[J] . Ariel Linden,S. Derya Uysal,Andrew Ryan,John L. Adams.Statist. Med. . 2016 (4)
[3]   On the use and misuse of scalar scores of confounders in design and analysis of observational studies [J].
Pfeiffer, R. M. ;
Riedl, R. .
STATISTICS IN MEDICINE, 2015, 34 (18) :2618-2635
[4]   Bias associated with using the estimated propensity score as a regression covariate [J].
Hade, Erinn M. ;
Lu, Bo .
STATISTICS IN MEDICINE, 2014, 33 (01) :74-87
[5]   The performance of different propensity score methods for estimating marginal hazard ratios [J].
Austin, Peter C. .
STATISTICS IN MEDICINE, 2013, 32 (16) :2837-2849
[6]   An Introduction to Propensity Score Methods for Reducing the Effects of Confounding in Observational Studies [J].
Austin, Peter C. .
MULTIVARIATE BEHAVIORAL RESEARCH, 2011, 46 (03) :399-424
[7]   A Systematic Review of Propensity Score Methods in the Social Sciences [J].
Thoemmes, Felix J. ;
Kim, Eun Sook .
MULTIVARIATE BEHAVIORAL RESEARCH, 2011, 46 (01) :90-118
[8]   The performance of different propensity-score methods for estimating differences in proportions (risk differences or absolute risk reductions) in observational studies [J].
Austin, Peter C. .
STATISTICS IN MEDICINE, 2010, 29 (20) :2137-2148
[9]   The performance of different propensity-score methods for estimating relative risks [J].
Austin, Peter C. .
JOURNAL OF CLINICAL EPIDEMIOLOGY, 2008, 61 (06) :537-545
[10]   Effect of a US National Institutes of Health programme of clinical trials on public health and costs [J].
Johnston, SC ;
Rootenberg, JD ;
Katrak, S ;
Smith, WS ;
Elkins, JS .
LANCET, 2006, 367 (9519) :1319-1327