Pragmatic considerations for negative control outcome studies to guide non-randomized comparative analyses: A...

Area of Science:

• Pharmacoepidemiology
• Biostatistics
• Health Services Research

Background:

• Non-randomized studies are increasingly used for drug effectiveness and safety evaluations.
• Unmeasured or mismeasured confounders pose a significant threat to the validity of these studies.
• Negative control outcomes (NCOs) offer a method to assess and mitigate confounding bias.

Purpose of the Study:

• To describe the application of NCO methods for bias assessment in non-randomized drug comparisons.
• To illustrate NCO implementation using real-world data from osteoporosis and lipid-lowering therapy studies.
• To provide recommendations for future NCO study design and interpretation.

Main Methods:

• Narrative review of NCO methodology.
• Application of NCOs in claims-based data sources for two drug classes.
• Focus on identifying confounding mechanisms and selecting appropriate NCOs.

Main Results:

• NCO studies can reveal potential uncontrolled confounding in comparative drug studies.
• Results from NCO analyses can inform decisions on proceeding with comparative effectiveness or safety research.
• Illustrative examples demonstrate how confounding can manifest in observational data.

Conclusions:

• NCO methods enhance the design, analysis, and interpretation of real-world data.
• Proposing NCO studies prior to comparative research strengthens the credibility of real-world evidence.
• Application of NCOs is crucial for reliable regulatory decision-making using non-randomized evidence.