Bayesian Method for Correcting Measurement Error in Case-Control Studies
Author Information
Author(s): Gabriela Espino-Hernandez, Paul Gustafson, Igor Burstyn
Primary Institution: University of British Columbia
Hypothesis
Can a Bayesian method effectively correct for measurement error in multiple continuous exposures in individually matched case-control studies?
Conclusion
The proposed Bayesian method can correct for measurement error in continuous exposures, showing little adjustment needed for the level of measurement error observed.
Supporting Evidence
- The Bayesian method was illustrated using data from a study on perfluorinated acids and thyroid hormone levels.
- Sensitivity analysis showed that larger measurement errors lead to more substantial adjustments.
- The method can be implemented in WinBUGS software for practical use.
Takeaway
This study created a new way to fix mistakes in measuring things in health studies, helping to understand if certain chemicals affect pregnant women’s thyroid health.
Methodology
A Bayesian approach was used to correct for measurement error in exposures, employing conditional logistic regression and random-effect models.
Potential Biases
Potential biases may arise from the assumptions of non-differential measurement error.
Limitations
The method assumes no misclassification of case/control status and may not generalize well to other settings.
Participant Demographics
Pregnant women aged 18 or older, with 96 cases and 175 matched controls.
Statistical Information
Confidence Interval
95% credible intervals were computed for odds ratios.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website