Confounder Selection and Sensitivity Analyses in Epidemiological Studies
Author Information
Author(s): Inoue Kosuke, Sakamaki Kentaro, Komukai Sho, Ito Yuri, Goto Atsushi, Shinozaki Tomohiro
Primary Institution: Kyoto University
Hypothesis
How can we effectively select confounders and assess sensitivity to unmeasured confounding in epidemiological studies?
Conclusion
The modified disjunctive cause criterion is a useful approach for confounder selection, enhancing the accuracy of epidemiological studies.
Supporting Evidence
- Confounders must be carefully selected to avoid bias in observational studies.
- The modified disjunctive cause criterion helps in identifying relevant confounders.
- Sensitivity analyses can assess the robustness of findings against unmeasured confounding.
Takeaway
This study helps researchers figure out how to choose the right factors to consider when studying health effects, so they don't get confused by other things that might affect the results.
Methodology
The study reviews various epidemiological and statistical methods for confounder selection and sensitivity analysis, illustrated with examples from the National Health and Nutritional Examination Survey.
Potential Biases
There is a risk of collider-stratification bias and M-bias when selecting confounders.
Limitations
The study may not cover all potential confounding variables and relies on the assumption of missing completely at random for data analysis.
Participant Demographics
Participants were from the National Health and Nutritional Examination Survey, targeting the U.S. civilian non-institutionalized adult population, with oversampling of specific groups.
Statistical Information
P-Value
0.05
Confidence Interval
[0.005–0.083]
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website