Combining Propensity Score Matching with Additional Adjustments for Prognostic Covariates

Abstract Propensity score matching refers to a class of multivariate methods used in comparative studies to construct treated and matched control samples that have similar distributions on many covariates. This matching is the observational study analog of randomization in ideal experiments, but is far less complete as it can only balance the distribution of observed covariates, whereas randomization balances the distribution of all covariates, both observed and unobserved. An important feature of propensity score matching is that it can be easily combined with model-based regression adjustments or with matching on a subset of special prognostic covariates or combinations of prognostic covariates that have been identified as being especially predictive of the outcome variables. We extend earlier results by developing approximations for the distributions of covariates in matched samples created with linear propensity score methods for the practically important situation where matching uses both the estimated linear propensity scores and a set of special prognostic covariates. Such matching on a subset of special prognostic covariates is an observational study analog of blocking in a randomized experiment. An example combining propensity score matching with Mahalanobis metric matching and regression adjustment is presented that demonstrates the flexibility of these methods for designing an observational study that effectively reduces both bias due to many observed covariates and bias and variability due to a more limited subset of covariates. Of particular importance, the general approach, which includes propensity score matching, was distinctly superior to methods that focus only on a subset of the prognostically most important covariates, even if those covariates account for most of the variation in the outcome variables. Also of importance, analyses based on matched samples were superior to those based on the full unmatched samples, even when regression adjustment was included.

[1]  P. Rosenbaum.  A Characterization of Optimal Designs for Observational Studies , 1991 .

[2]  C. Drake.  Effects of misspecification of the propensity score on estimators of treatment effect , 1993 .

[3]  Donald B. Rubin,et al.  Multivariate matching methods that are equal percent bias reducing , 1974 .

[4]  D. Rubin,et al.  The bias due to incomplete matching. , 1983, Biometrics.

[5]  W. G. Cochran.  The effectiveness of adjustment by subclassification in removing bias in observational studies. , 1968, Biometrics.

[6]  D B Rubin,et al.  Matching using estimated propensity scores: relating theory to practice. , 1996, Biometrics.

[7]  B. Efron.  The Efficiency of Logistic Regression Compared to Normal Discriminant Analysis , 1975 .

[8]  R. G. Carpenter,et al.  Matching when covariables are normally distributed , 1977 .

[9]  D. Rubin,et al.  Constructing a Control Group Using Multivariate Matched Sampling Methods That Incorporate the Propensity Score , 1985 .

[10]  Donald B. Rubin,et al.  Characterizing the effect of matching using linear propensity score methods with normal distributions , 1992 .

[11]  Donald B. Rubin,et al.  MULTIVARIATE MATCHING METHODS THAT ARE EQUAL PERCENT BIAS REDUCING, I: SOME EXAMPLES , 1974 .

[12]  Wojtek J. Krzanowski,et al.  Discrimination and Classification Using Both Binary and Continuous Variables , 1975 .

[13]  Paul R. Rosenbaum,et al.  Comparison of Multivariate Matching Methods: Structures, Distances, and Algorithms , 1993 .

[14]  J. Daudin,et al.  Selection of variables in mixed-variable discriminant analysis , 1986 .

[15]  D. Rubin,et al.  Using Multivariate Matched Sampling and Regression Adjustment to Control Bias in Observational Studies , 1978 .

[16]  D. Rubin.  Matched Sampling for Causal Effects: The Use of Matched Sampling and Regression Adjustment to Remove Bias in Observational Studies , 1973 .

[17]  D. Rubin.  Matched Sampling for Causal Effects: Matching to Remove Bias in Observational Studies , 1973 .

[18]  D. Rubin.  Bias Reduction Using Mahalanobis-Metric Matching , 1980 .

[19]  Donald B. Rubin,et al.  Affinely Invariant Matching Methods with Ellipsoidal Distributions , 1992 .

[20]  W. G. Cochran,et al.  Controlling Bias in Observational Studies: A Review. , 1974 .

[21]  Ingram Olkin,et al.  Multivariate Correlation Models with Mixed Discrete and Continuous Variables , 1961 .

[22]  Paul R. Rosenbaum,et al.  Optimal Matching for Observational Studies , 1989 .

[23]  D. Rubin,et al.  The central role of the propensity score in observational studies for causal effects , 1983 .

[24]  W J Krzanowski,et al.  Mixtures of continuous and categorical variables in discriminant analysis. , 1980, Biometrics.

[25]  P. C. Chang,et al.  Robustness of Fisher's Linear Discriminant Function under Two-Component Mixed Normal Models , 1981 .