Abstract
(1) Background: Propensity score methods gained popularity in non-interventional clinical studies. As it may often occur in observational datasets, some values in baseline covariates are missing for some patients. The present study aims to compare the performances of popular statistical methods to deal with missing data in propensity score analysis. (2) Methods: Methods that account for missing data during the estimation process and methods based on the imputation of missing values, such as multiple imputations, were considered. The methods were applied on the dataset of an ongoing prospective registry for the treatment of unprotected left main coronary artery disease. The performances were assessed in terms of the overall balance of baseline covariates. (3) Results: Methods that explicitly deal with missing data were superior to classical complete case analysis. The best balance was observed when propensity scores were estimated with a method that accounts for missing data using a stochastic approximation of the expectation-maximization algorithm. (4) Conclusions: If missing at random mechanism is plausible, methods that use missing data to estimate propensity score or impute them should be preferred. Sensitivity analyses are encouraged to evaluate the implications methods used to handle missing data and estimate propensity score.
Highlights
Non-interventional studies are increasingly being used to infer causal relationships between new treatments and health outcomes in real clinical settings [1]
When the mechanism that generated missing data is Missing At Random (MAR), i.e., the missingness depends only on observed variables, or Missing Not At Random (MNAR), i.e., missing data are generated given information not observed in the dataset, there may be the risk of obtaining a biased estimate of the treatment effect [34]
The most notable differences between angiography-guidance and imagingintravascular ultrasound (IVUS)/optical coherence tomography (OCT) were observed in terms of the proportion of subjects with NSTEMI, the EuroSCORE II, the proportion of patients with an FE between 30–50%, which was lower in the IVUS/OCT group, the Syntax score, higher for patients who underwent angiography-guidance
Summary
Non-interventional (or non-randomized) studies are increasingly being used to infer causal relationships between new treatments and health outcomes in real clinical settings [1]. In non-randomized clinical studies, the allocation of subjects into treatment groups often depends on their characteristics collected at baseline. The groups of individuals systematically differ from each other in terms of baseline covariates, leading to the so-called confounding bias. In such situations, a naïve comparison of the outcomes between the treatment and control groups would provide a biased treatment effect estimate. Propensity Score (PS) methods have been proposed to reduce confounding bias in non-interventional studies and to provide consistent treatment effect estimates [2]. PS is defined as the individual probability of being treated given baseline characteristics
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callMore From: International Journal of Environmental Research and Public Health
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
