Using Propensity Score Matching in Clinical Investigations: A Discussion and Illustration

Authors

  • Carrie Hosman From the Department of Statistics, University of Michigan, Ann Arbor MI, USA
  • Hitinder S. Gurm From the Department of Statistics, University of Michigan, Ann Arbor MI, USA

DOI:

https://doi.org/10.6000/1929-6029.2015.04.02.7

Keywords:

Propensity Score Matching, Observational Data, Clinical Investigations.

Abstract

Propensity score matching is a useful tool to analyze observational data in clinical investigations, but it is often executed in an overly simplistic manner, failing to use the data in the best possible way. This review discusses current best practices in propensity score matching, outlining the method’s essential steps, including appropriate post-matching balance assessments and sensitivity analyses. These steps are summarized as eight key traits of a propensity matched study. Further, this review illustrates these traits through a case study examining the impact of access site in percutaneous coronary intervention (PCI) procedures on bleeding complications. Through propensity score matching, we find that bleeding occurs significantly less often with radial access procedures, though many other outcomes show no significant difference by access site, a finding that mirrors the results of randomized controlled trials. Lack of attention to methodological principles can result in results that are not biologically plausible.

Author Biographies

Carrie Hosman, From the Department of Statistics, University of Michigan, Ann Arbor MI, USA

Statistics

Hitinder S. Gurm, From the Department of Statistics, University of Michigan, Ann Arbor MI, USA

Internal Medicine

References

Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika 1983; 70: 41-55. http://dx.doi.org/10.1093/biomet/70.1.41 DOI: https://doi.org/10.1093/biomet/70.1.41

Luo Z, Gardiner JC, Bradley CJ. Applying propensity score methods in medical research: pitfalls and prospects. Medical Care Research and Review 2010; 67: 528-554. http://dx.doi.org/10.1177/1077558710361486 DOI: https://doi.org/10.1177/1077558710361486

Brookhart MA, Schneeweiss S, Rothman KJ, Glynn RJ, Avorn J, Stürmer T. Variable selection for propensity score models. American journal of epidemiology 2006; 163: 1149-1156. http://dx.doi.org/10.1093/aje/kwj149 DOI: https://doi.org/10.1093/aje/kwj149

Rubin DB, Thomas N. Matching using estimated propensity scores: relating theory to practice. Biometrics 1996: 249-264. http://dx.doi.org/10.2307/2533160 DOI: https://doi.org/10.2307/2533160

Rosenbaum PR, SpringerLink (Online service). Design of observational studies. New York: Springer, 2010: 1 online resource (xviii, 384 p.

Austin PC, Grootendorst P, Anderson GM. A comparison of the ability of different propensity score models to balance measured variables between treated and untreated subjects: a Monte Carlo study. Statistics in medicine 2007; 26: 734-753. http://dx.doi.org/10.1002/sim.2580 DOI: https://doi.org/10.1002/sim.2580

Griswold ME, Localio AR, Mulrow C. Propensity score adjustment with multilevel data: setting your sites on decreasing selection bias. Annals of internal medicine 2010; 152: 393-395. http://dx.doi.org/10.7326/0003-4819-152-6-201003160-00010 DOI: https://doi.org/10.7326/0003-4819-152-6-201003160-00010

Arpino B, Mealli F. The specification of the propensity score in multilevel observational studies. Computational Statistics & Data Analysis 2011; 55: 1770-1780. http://dx.doi.org/10.1016/j.csda.2010.11.008 DOI: https://doi.org/10.1016/j.csda.2010.11.008

Rubin DB. The design versus the analysis of observational studies for causal effects: parallels with the design of randomized trials. Statistics in medicine 2007; 26: 20-36. http://dx.doi.org/10.1002/sim.2739 DOI: https://doi.org/10.1002/sim.2739

Hansen BB, Klopfer SO. Optimal full matching and related designs via network flows. Journal of Computational and Graphical Statistics 2006; 15. DOI: https://doi.org/10.1198/106186006X137047

Gu XS, Rosenbaum PR. Comparison of multivariate matching methods: Structures, distances, and algorithms. Journal of Computational and Graphical Statistics 1993; 2: 405-420. http://dx.doi.org/10.2307/1390693 DOI: https://doi.org/10.1080/10618600.1993.10474623

Ming K, Rosenbaum PR. Substantial gains in bias reduction from matching with a variable number of controls. Biometrics 2000; 56: 118-124. http://dx.doi.org/10.1111/j.0006-341X.2000.00118.x DOI: https://doi.org/10.1111/j.0006-341X.2000.00118.x

Imai K, King G, Stuart EA. Misunderstandings between experimentalists and observationalists about causal inference. Journal of the royal statistical society: series A (statistics in society) 2008; 171: 481-502. http://dx.doi.org/10.1111/j.1467-985X.2007.00527.x DOI: https://doi.org/10.1111/j.1467-985X.2007.00527.x

Austin PC. The relative ability of different propensity score methods to balance measured covariates between treated and untreated subjects in observational studies. Medical Decision Making 2009. http://dx.doi.org/10.1177/0272989X09341755 DOI: https://doi.org/10.1177/0272989X09341755

Hansen BB, Bowers J. Covariate balance in simple, stratified and clustered comparative studies. Statistical Science 2008: 219-236. http://dx.doi.org/10.1214/08-STS254 DOI: https://doi.org/10.1214/08-STS254

Rosenbaum PR. Observational studies: Springer, 2002. http://dx.doi.org/10.1007/978-1-4757-3692-2 DOI: https://doi.org/10.1007/978-1-4757-3692-2

Hosman CA, Hansen BB, Holland PW. The sensitivity of linear regression coefficients’ confidence limits to the omission of a confounder. The Annals of Applied Statistics 2010; 4: 849-870. http://dx.doi.org/10.1214/09-AOAS315 DOI: https://doi.org/10.1214/09-AOAS315

Lin D, Psaty BM, Kronmal R. Assessing the sensitivity of regression results to unmeasured confounders in observational studies. Biometrics 1998: 948-963. http://dx.doi.org/10.2307/2533848 DOI: https://doi.org/10.2307/2533848

Gurm HS, Smith DE, Collins JS et al. The relative safety and efficacy of abciximab and eptifibatide in patients undergoing primary percutaneous coronary intervention: insights from a large regional registry of contemporary percutaneous coronary intervention. Journal of the American College of Cardiology 2008; 51: 529-35. http://dx.doi.org/10.1016/j.jacc.2007.09.053 DOI: https://doi.org/10.1016/j.jacc.2007.09.053

Moscucci M, Rogers EK, Montoye C et al. Association of a continuous quality improvement initiative with practice and outcome variations of contemporary percutaneous coronary interventions. Circulation 2006; 113: 814-22. http://dx.doi.org/10.1161/CIRCULATIONAHA.105.541995 DOI: https://doi.org/10.1161/CIRCULATIONAHA.105.541995

Kline-Rogers E, Share D, Bondie D et al. Development of a multicenter interventional cardiology database: the Blue Cross Blue Shield of Michigan Cardiovascular Consortium (BMC2) experience. Journal of interventional cardiology 2002; 15: 387-92. http://dx.doi.org/10.1111/j.1540-8183.2002.tb01072.x DOI: https://doi.org/10.1111/j.1540-8183.2002.tb01072.x

Downloads

Published

2015-05-21

How to Cite

Hosman, C., & Gurm, H. S. (2015). Using Propensity Score Matching in Clinical Investigations: A Discussion and Illustration. International Journal of Statistics in Medical Research, 4(2), 208–216. https://doi.org/10.6000/1929-6029.2015.04.02.7

Issue

Section

General Articles