Accounting for Comorbidity in Etiologic Research.

Area of Science:

• Epidemiology
• Biostatistics
• Causal Inference

Background:

• Comorbidity is common in clinical research, yet its role in exposure-outcome relationships is often poorly handled.
• Failure to account for comorbidity's causal role can lead to biased effect estimates in etiologic studies.

Purpose of the Study:

• To explore causal structures involving comorbidity.
• To provide guidance on handling comorbidity in etiologic research for robust causal inference.

Main Methods:

• Utilized Directed Acyclic Graphs (DAGs) to model six comorbidity causal scenarios.
• Conducted simulations (5,000 iterations) assessing bias under different effect measures.
• Evaluated bias by comparing adjusted and unadjusted effect estimates.

Main Results:

• The impact of adjusting for comorbidity depends on its causal role: it mitigates bias as a confounder but introduces bias as a mediator or collider.
• For comorbidity as an outcome consequence, adjustment decisions are context-dependent.

Conclusions:

• Explicit causal assumptions are vital for appropriate analytical strategies in etiologic research.
• Guidance is provided for handling comorbidity measures to align study design with research objectives.