Estimating treatment effect via simple cross design synthesis.

Area of Science:

• Biostatistics
• Epidemiology
• Health Services Research

Background:

• Randomized controlled trials (RCTs) are the gold standard but have limited external validity due to strict eligibility criteria.
• Observational studies offer broad generalizability but are prone to selection bias.
• A gap exists in leveraging the strengths of both study designs for robust medical evidence.

Purpose of the Study:

• To introduce and evaluate a novel Cross-Design Synthesis (CDS) approach for combining RCT and observational data.
• To develop a simple effect size estimator utilizing CDS.
• To compare the performance of the CDS estimator against traditional single-source estimators.

Main Methods:

• Proposed a simple effect size estimator based on the Cross-Design Synthesis (CDS) framework.
• Evaluated the estimator's properties within a causal estimation framework.
• Compared the CDS estimator with traditional estimators using simulated or real-world data.

Main Results:

• The simple CDS estimator demonstrated unbiasedness under specific conditions where observational data selection error is constant across RCT-eligible and non-eligible populations.
• Under plausible data assumptions, the CDS estimator exhibited reduced bias and improved coverage compared to single-source estimators.
• The CDS approach effectively integrates internal validity from RCTs and external validity from observational studies.

Conclusions:

• Cross-Design Synthesis (CDS) offers a valuable method for synthesizing evidence from both RCTs and observational studies.
• The proposed simple CDS estimator provides a potentially more accurate and generalizable measure of effect size.
• This approach can enhance the applicability of research findings to broader patient populations in medical decision-making.