Can propensity score matching replace randomized controlled trials?

Area of Science:

• Clinical Research Methodology
• Biostatistics
• Epidemiology

Background:

• Randomized controlled trials (RCTs) are the gold standard but face limitations like ethical concerns and poor external validity due to strict inclusion criteria.
• Existing observational data sources, such as registries and electronic health records, offer valuable information for clinical research.
• Propensity score matching (PSM) is a statistical method that uses observational data to create comparable groups for treatment effect estimation.

Purpose of the Study:

• To review the applications, advantages, and considerations of integrating propensity score matching (PSM) with randomized controlled trials (RCTs).
• To highlight how PSM can refine randomization, enhance external validity, and account for protocol non-compliance in clinical research.
• To advocate for the synergistic use of PSM and RCTs to improve the generalizability of clinical trial outcomes.

Main Methods:

• Literature review of propensity score matching (PSM) applications in conjunction with randomized controlled trials (RCTs).
• Analysis of how PSM utilizes observational data to match participants based on propensity scores, considering covariates like age, gender, and comorbidities.
• Examination of case studies where PSM was incorporated into RCT analysis to address baseline characteristic imbalances.

Main Results:

• PSM circumvents ethical issues associated with withholding treatment in RCTs by using retrospective observational data.
• PSM enhances external validity by allowing the inclusion of diverse populations often excluded from RCTs (e.g., elderly, pregnant women, children).
• Integration of PSM in an RCT example (mannitol in acute cerebral hemorrhage) demonstrated a fairer comparison by matching baseline characteristics.

Conclusions:

• The synergistic integration of PSM with RCTs offers superior research outcomes compared to either method alone.
• PSM refines RCTs by improving randomization, increasing external validity, and managing non-compliance, leading to more generalizable results.
• Future research should prioritize the integration of PSM in RCTs to enhance the applicability of findings to broader patient populations in clinical practice.