On the joint use of propensity and prognostic scores in estimation of the average treatment effect on the treated: a simulation study

Propensity and prognostic score methods seek to improve the quality of causal inference in non-randomized or observational studies by replicating the conditions found in a controlled experiment, at least with respect to observed characteristics. Propensity scores model receipt of the treatment of interest; prognostic scores model the potential outcome under a single treatment condition. While the popularity of propensity score methods continues to grow, prognostic score methods and methods combining propensity and prognostic scores have thus far received little attention. To this end, we performed a simulation study that compared subclassification and full matching on a single estimated propensity or prognostic score with three approaches combining the estimated propensity and prognostic scores: full matching on a Mahalanobis distance combining the estimated propensity and prognostic scores (FULL-MAHAL); full matching on the estimated prognostic propensity score within propensity score calipers (FULL-PGPPTY); and subclassification on an estimated propensity and prognostic score grid with 5 × 5 subclasses (SUBCLASS(5*5)). We considered settings in which one, both, or neither score model was misspecified. The data generating mechanisms varied in the degree of linearity and additivity in the true treatment assignment and outcome models. FULL-MAHAL and FULL-PGPPTY exhibited strong to superior performance in root mean square error terms across all simulation settings and scenarios. Methods combining propensity and prognostic scores were no less robust to model misspecification than single-score methods even when both score models were incorrectly specified. Our findings support the joint use of propensity and prognostic scores in estimation of the average treatment effect on the treated.

[1]  C. Michael Stein,et al.  Oral erythromycin and the risk of sudden death from cardiac causes , 2004 .

[2]  Donald B. Rubin,et al.  Principal Stratification Approach to Broken Randomized Experiments , 2003 .

[3]  J. Myers,et al.  Effects of adjusting for instrumental variables on bias and precision of effect estimates. , 2011, American journal of epidemiology.

[4]  Joseph Kang,et al.  Demystifying Double Robustness: A Comparison of Alternative Strategies for Estimating a Population Mean from Incomplete Data , 2007, 0804.2958.

[5]  P. Rosenbaum Dropping out of High School in the United States: An Observational Study , 1986 .

[6]  E F Cook,et al.  Performance of tests of significance based on stratification by a multivariate confounder score or by a propensity score. , 1989, Journal of clinical epidemiology.

[7]  B. Hansen,et al.  Optimal Full Matching and Related Designs via Network Flows , 2006 .

[8]  T. DiPrete,et al.  7. Assessing Bias in the Estimation of Causal Effects: Rosenbaum Bounds on Matching Estimators and Instrumental Variables Estimation with Imperfect Instruments , 2004 .

[9]  Gary King,et al.  Matching as Nonparametric Preprocessing for Reducing Model Dependence in Parametric Causal Inference , 2007, Political Analysis.

[10]  J. Brooks-Gunn,et al.  Maternal employment and child development: a fresh look using newer methods. , 2005, Developmental psychology.

[11]  Til Stürmer,et al.  Confounder summary scores when comparing the effects of multiple drug exposures , 2010, Pharmacoepidemiology and drug safety.

[12]  W. Ray,et al.  Performance of disease risk scores, propensity scores, and traditional multivariable outcome regression in the presence of multiple confounders. , 2011, American journal of epidemiology.

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

[14]  Jeremy Arkes,et al.  Marijuana use and depression among adults: Testing for causal associations. , 2006, Addiction.

[15]  B. Hansen The prognostic analogue of the propensity score , 2008 .

[16]  Thomas Lumley,et al.  Analysis of Complex Survey Samples , 2004 .

[17]  J. Avorn,et al.  High-dimensional Propensity Score Adjustment in Studies of Treatment Effects Using Health Care Claims Data , 2009, Epidemiology.

[18]  D. Rubin,et al.  Reducing Bias in Observational Studies Using Subclassification on the Propensity Score , 1984 .

[19]  Gary King,et al.  MatchIt: Nonparametric Preprocessing for Parametric Causal Inference , 2011 .

[20]  K. Murray,et al.  Cyclic antidepressants and the risk of sudden cardiac death , 2004, Clinical pharmacology and therapeutics.

[21]  N. Christakis,et al.  The health impact of health care on families: a matched cohort study of hospice use by decedents and mortality outcomes in surviving, widowed spouses. , 2003, Social science & medicine.

[22]  G. Imbens Nonparametric Estimation of Average Treatment Effects Under Exogeneity: A Review , 2004 .

[23]  W. Ray,et al.  Risk of peptic ulcer hospitalizations in users of NSAIDs with gastroprotective cotherapy versus coxibs. , 2007, Gastroenterology.

[24]  D. Rubin,et al.  Combining Propensity Score Matching with Additional Adjustments for Prognostic Covariates , 2000 .

[25]  O S Miettinen,et al.  Stratification by a multivariate confounder score. , 1976, American journal of epidemiology.

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

[27]  C Michael Stein,et al.  Non-steroidal anti-inflammatory drugs and risk of serious coronary heart disease: an observational cohort study , 2002, The Lancet.

[28]  W A Ray,et al.  Antipsychotics and the risk of sudden cardiac death. , 2001, Archives of general psychiatry.

[29]  B. Hansen Propensity Score Matching to Extract Latent Experiments from Nonexperimental Data: A Case Study , 2011 .

[30]  S. Shoor,et al.  Risk of acute myocardial infarction and sudden cardiac death in patients treated with cyclo-oxygenase 2 selective and non-selective non-steroidal anti-inflammatory drugs: nested case-control study , 2005, The Lancet.

[31]  J. Avorn,et al.  Variable selection for propensity score models. , 2006, American journal of epidemiology.

[32]  M. Pike,et al.  Some insights into Miettinen's multivariate confounder score approach to case-control study analysis. , 1979, Journal of epidemiology and community health.

[33]  Vincent Mor,et al.  Principles for modeling propensity scores in medical research: a systematic literature review , 2004, Pharmacoepidemiology and drug safety.

[34]  B. Hansen Full Matching in an Observational Study of Coaching for the SAT , 2004 .

[35]  Elizabeth A Stuart,et al.  Improving propensity score weighting using machine learning , 2010, Statistics in medicine.

[36]  Elizabeth A Stuart,et al.  Matching methods for causal inference: A review and a look forward. , 2010, Statistical science : a review journal of the Institute of Mathematical Statistics.

[37]  J. Lunceford,et al.  Stratification and weighting via the propensity score in estimation of causal treatment effects: a comparative study , 2004, Statistics in medicine.

[38]  S. Schneeweiss,et al.  Evaluating uses of data mining techniques in propensity score estimation: a simulation study , 2008, Pharmacoepidemiology and drug safety.

[39]  J. Carpenter,et al.  Variance estimation for stratified propensity score estimators , 2012, Statistics in medicine.

[40]  S. Morgan,et al.  Matching Estimators of Causal Effects , 2006 .

[41]  C Michael Stein,et al.  COX-2 selective non-steroidal anti-inflammatory drugs and risk of serious coronary heart disease , 2002, The Lancet.

[42]  Kosuke Imai,et al.  Causal Inference With General Treatment Regimes , 2004 .

[43]  J. Pearl Invited commentary: understanding bias amplification. , 2011, American journal of epidemiology.

[44]  Sebastian Schneeweiss,et al.  Role of disease risk scores in comparative effectiveness research with emerging therapies , 2012, Pharmacoepidemiology and drug safety.

[45]  Xiao-Hua Zhou,et al.  The use of propensity scores in pharmacoepidemiologic research , 2000, Pharmacoepidemiology and drug safety.

[46]  M Alan Brookhart,et al.  Analytic strategies to adjust confounding using exposure propensity scores and disease risk scores: nonsteroidal antiinflammatory drugs and short-term mortality in the elderly. , 2005, American journal of epidemiology.

[47]  Ingeborg Waernbaum,et al.  Model misspecification and robustness in causal inference: comparing matching with doubly robust estimation , 2012, Statistics in medicine.

[48]  W. Ray,et al.  Use of disease risk scores in pharmacoepidemiologic studies , 2009, Statistical methods in medical research.