Literature DB >> 34809559

Oversampling and replacement strategies in propensity score matching: a critical review focused on small sample size in clinical settings.

Daniele Bottigliengo1, Ileana Baldi1, Corrado Lanera1, Giulia Lorenzoni1, Jonida Bejko2, Tomaso Bottio2, Vincenzo Tarzia2, Massimiliano Carrozzini2, Gino Gerosa2, Paola Berchialla3, Dario Gregori4.   

Abstract

BACKGROUND: Propensity score matching is a statistical method that is often used to make inferences on the treatment effects in observational studies. In recent years, there has been widespread use of the technique in the cardiothoracic surgery literature to evaluate to potential benefits of new surgical therapies or procedures. However, the small sample size and the strong dependence of the treatment assignment on the baseline covariates that often characterize these studies make such an evaluation challenging from a statistical point of view. In such settings, the use of propensity score matching in combination with oversampling and replacement may provide a solution to these issues by increasing the initial sample size of the study and thus improving the statistical power that is needed to detect the effect of interest. In this study, we review the use of propensity score matching in combination with oversampling and replacement in small sample size settings.
METHODS: We performed a series of Monte Carlo simulations to evaluate how the sample size, the proportion of treated, and the assignment mechanism affect the performances of the proposed approaches. We assessed the performances with overall balance, relative bias, root mean squared error and nominal coverage. Moreover, we illustrate the methods using a real case study from the cardiac surgery literature.
RESULTS: Matching without replacement produced estimates with lower bias and better nominal coverage than matching with replacement when 1:1 matching was considered. In contrast to that, matching with replacement showed better balance, relative bias, and root mean squared error than matching without replacement for increasing levels of oversampling. The best nominal coverage was obtained by using the estimator that accounts for uncertainty in the matching procedure on sets of units obtained after matching with replacement.
CONCLUSIONS: The use of replacement provides the most reliable treatment effect estimates and that no more than 1 or 2 units from the control group should be matched to each treated observation. Moreover, the variance estimator that accounts for the uncertainty in the matching procedure should be used to estimate the treatment effect.
© 2021. The Author(s).

Entities:  

Keywords:  Monte Carlo simulations; Oversampling; Propensity score matching; Replacement; Small samples

Mesh:

Year:  2021        PMID: 34809559      PMCID: PMC8609749          DOI: 10.1186/s12874-021-01454-z

Source DB:  PubMed          Journal:  BMC Med Res Methodol        ISSN: 1471-2288            Impact factor:   4.615


  50 in total

1.  Long-term safety and efficacy of stenting versus coronary artery bypass grafting for unprotected left main coronary artery disease: 5-year results from the MAIN-COMPARE (Revascularization for Unprotected Left Main Coronary Artery Stenosis: Comparison of Percutaneous Coronary Angioplasty Versus Surgical Revascularization) registry.

Authors:  Duk-Woo Park; Ki Bae Seung; Young-Hak Kim; Jong-Young Lee; Won-Jang Kim; Soo-Jin Kang; Seung-Whan Lee; Cheol Whan Lee; Seong-Wook Park; Sung-Cheol Yun; Hyeon-Cheol Gwon; Myung-Ho Jeong; Yang-Soo Jang; Hyo-Soo Kim; Pum Joon Kim; In-Whan Seong; Hun Sik Park; Taehoon Ahn; In-Ho Chae; Seung-Jea Tahk; Wook-Sung Chung; Seung-Jung Park
Journal:  J Am Coll Cardiol       Date:  2010-05-06       Impact factor: 24.094

Review 2.  Effects and non-effects of paired identical observations in comparing proportions with binary matched-pairs data.

Authors:  Alan Agresti; Yongyi Min
Journal:  Stat Med       Date:  2004-01-15       Impact factor: 2.373

Review 3.  Model-based estimation of relative risks and other epidemiologic measures in studies of common outcomes and in case-control studies.

Authors:  Sander Greenland
Journal:  Am J Epidemiol       Date:  2004-08-15       Impact factor: 4.897

4.  Interval estimation for treatment effects using propensity score matching.

Authors:  Jennifer Hill; Jerome P Reiter
Journal:  Stat Med       Date:  2006-07-15       Impact factor: 2.373

5.  Evaluating uses of data mining techniques in propensity score estimation: a simulation study.

Authors:  Soko Setoguchi; Sebastian Schneeweiss; M Alan Brookhart; Robert J Glynn; E Francis Cook
Journal:  Pharmacoepidemiol Drug Saf       Date:  2008-06       Impact factor: 2.890

6.  Absolute risk reductions, relative risks, relative risk reductions, and numbers needed to treat can be obtained from a logistic regression model.

Authors:  Peter C Austin
Journal:  J Clin Epidemiol       Date:  2009-02-20       Impact factor: 6.437

7.  Metrics for covariate balance in cohort studies of causal effects.

Authors:  Jessica M Franklin; Jeremy A Rassen; Diana Ackermann; Dorothee B Bartels; Sebastian Schneeweiss
Journal:  Stat Med       Date:  2013-12-09       Impact factor: 2.373

8.  Uncertainty in the design stage of two-stage Bayesian propensity score analysis.

Authors:  Shirley X Liao; Corwin M Zigler
Journal:  Stat Med       Date:  2020-05-24       Impact factor: 2.497

9.  Balance diagnostics for comparing the distribution of baseline covariates between treatment groups in propensity-score matched samples.

Authors:  Peter C Austin
Journal:  Stat Med       Date:  2009-11-10       Impact factor: 2.373

10.  A comparison of different methods to handle missing data in the context of propensity score analysis.

Authors:  Jungyeon Choi; Olaf M Dekkers; Saskia le Cessie
Journal:  Eur J Epidemiol       Date:  2018-10-19       Impact factor: 8.082

View more

北京卡尤迪生物科技股份有限公司 © 2022-2023.