Using propensity score-based weighting in the evaluation of health management programme effectiveness

被引:49
作者
Linden, Ariel [1 ]
Adams, John L. [2 ]
机构
[1] Linden Consulting Grp, Hillsboro, OR 97124 USA
[2] RAND Corp, Santa Monica, CA USA
关键词
health management programmes; inverse probability of treatment weights; propensity score; MARGINAL STRUCTURAL MODELS; CAUSAL INFERENCE; REGRESSION; ESTIMATOR;
D O I
10.1111/j.1365-2753.2009.01219.x
中图分类号
R19 [保健组织与事业(卫生事业管理)];
学科分类号
摘要
When the randomized controlled trial is unfeasible, programme evaluators attempt to emulate the randomization process in observational studies by creating a control group that is essentially equivalent to the treatment group on known characteristics and trust that the remaining unknown characteristics are inconsequential and will not bias the results. In recent years, adjustment procedures based on the propensity score, such as matching and subclassification, have become increasingly popular. A new technique that has particular appeal for evaluating health management programmes uses the propensity score to create a weight based on the subject's inverse probability of receiving treatment. This weighting mechanism removes imbalances of pre-intervention characteristics between treated and non-treated individuals, and is then used within a regression framework to provide unbiased estimates of treatment effects. This paper presents a non-technical introduction of this technique by illustrating its implementation with data from a recent study estimating the impact of a motivational interviewing-based health coaching on patient activation measure scores in a chronically ill group of individuals. Because of its relative simplicity and tremendous utility, propensity-score weighting adjustment should be considered as an alternative procedure for use with observational data to evaluate health management programme effectiveness.
引用
收藏
页码:175 / 179
页数:5
相关论文
共 31 条
[21]  
Robins J. M., 2000, Proceedings of the American Statistical Association Section on Bayesian Statistical Science, V1999, P6
[22]  
Robins J M., 1998, 1997 Proceedings of the Section on Bayesian Statistical Science, P1
[23]   Marginal structural models and causal inference in epidemiology [J].
Robins, JM ;
Hernán, MA ;
Brumback, B .
EPIDEMIOLOGY, 2000, 11 (05) :550-560
[24]   MODEL-BASED DIRECT ADJUSTMENT [J].
ROSENBAUM, PR .
JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION, 1987, 82 (398) :387-394
[25]   THE CENTRAL ROLE OF THE PROPENSITY SCORE IN OBSERVATIONAL STUDIES FOR CAUSAL EFFECTS [J].
ROSENBAUM, PR ;
RUBIN, DB .
BIOMETRIKA, 1983, 70 (01) :41-55
[26]   REDUCING BIAS IN OBSERVATIONAL STUDIES USING SUBCLASSIFICATION ON THE PROPENSITY SCORE [J].
ROSENBAUM, PR ;
RUBIN, DB .
JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION, 1984, 79 (387) :516-524
[27]   Matching using estimated propensity scores: Relating theory to practice [J].
Rubin, DB ;
Thomas, N .
BIOMETRICS, 1996, 52 (01) :249-264
[29]  
Scharfstein DO, 1999, J AM STAT ASSOC, V94, P1096, DOI 10.2307/2669923
[30]   US valuation of the EQ-5D health states - Development and testing of the D1 valuation model [J].
Shaw, JW ;
Johnson, JA ;
Coons, SJ .
MEDICAL CARE, 2005, 43 (03) :203-220