What question are we trying to answer? Embracing causal inference.

Area of Science:

• Veterinary Epidemiology
• Preventive Medicine
• Causal Inference

Background:

• Veterinary epidemiologists aim to identify causes of disease using observational research.
• Current practices in veterinary epidemiology often involve controlling for confounding variables and using causal language in interpretations.
• Established frameworks for studying causes include hypothesis articulation, variable selection, statistical estimation, and causal interpretation.

Purpose of the Study:

• To summarize current approaches to studying causes in veterinary epidemiology.
• To highlight differences in methodology between veterinary and human population observational studies.
• To advocate for a community discussion on improving causal inference in veterinary research.

Main Methods:

• The study reviews empirical evidence of causal inference methods in veterinary epidemiology.
• It compares approaches used in veterinary and human population studies.
• It discusses the consequences of current methodologies on research validity and replicability.

Main Results:

• Veterinary observational studies frequently aim to identify causes, control for confounders, and use causal language.
• Significant differences exist in the application of causal inference frameworks compared to human studies, particularly in a priori hypothesis definition and confounding variable selection.
• Data-driven approaches for variable selection are common in veterinary studies, contrasting with the prior knowledge-based approach in human studies.

Conclusions:

• Current methods in veterinary epidemiology may increase the probability of biased results and reduce replicability.
• A critical discussion among researchers is warranted to refine approaches for causal inference in veterinary observational studies.
• Improving methodological rigor can enhance the validity and impact of veterinary epidemiological research.