The importance of having a conceptual stage when reporting non-randomized studies

Area of Science:

• Biostatistics
• Epidemiology
• Health Research Methods

Background:

• Non-randomized studies are frequently used in scientific research.
• Statistical reporting guidelines exist but may be insufficient for causal conclusions.
• Ensuring the validity of causal inferences from observational data is critical.

Purpose of the Study:

• To highlight limitations in current statistical reporting guidelines for non-randomized studies.
• To propose improvements for drawing causal conclusions from such studies.
• To introduce the concept of a 'conceptual stage' in statistical analysis.

Main Methods:

• Review and critique existing statistical reporting guidelines.
• Illustrate the proposed 'conceptual stage' for analysis.
• Emphasize the formal description of non-randomized studies within hypothetical randomized experiments.

Main Results:

• Current guidelines may not adequately support causal inference.
• A formal 'conceptual stage' is identified as a crucial initial step.
• This stage involves embedding the non-randomized study design within a hypothetical randomized experiment framework.

Conclusions:

• Enhanced statistical reporting guidelines are needed for non-randomized studies aiming for causal conclusions.
• The proposed 'conceptual stage' offers a structured approach to improve analytical rigor.
• Implementing this conceptual stage can lead to more reliable causal inferences in research.