Related Experiment Videos
Limitations of relative sensitivity in detecting differential misclassification in case-control studies
J A Schwartzbaum1, R W Setzer, L L Kupper
1Department of Preventive Medicine, Ohio State University, Columbus 43210.
Epidemiology (Cambridge, Mass.)
|May 1, 1994
Summary
Apparent relative sensitivity can indicate differential misclassification of exposure status. However, it may falsely suggest misclassification or misrepresent its degree, depending on specific epidemiological conditions.
Area of Science:
- Epidemiology
- Biostatistics
Background:
- Apparent relative sensitivity (ARS) measures observed exposure sensitivity in cases versus controls.
- An ARS differing from 1.0 typically suggests differential misclassification of exposure status.
Purpose of the Study:
- To identify conditions where ARS > 1.0 accurately reflects differential misclassification.
- To assess if ARS quantifies the degree of differential misclassification accurately.
Main Methods:
- Derived an algebraic relationship linking ARS to true sensitivities, specificities, odds ratio, external standard accuracy, and disease incidence.
- Analyzed scenarios with perfect and imperfect external standards.
Main Results:
- ARS > 1.0 correctly indicates differential misclassification if the external standard is perfect, or if the odds ratio is 1.0 with specific true sensitivity/specificity values.
- ARS > 1.0 can falsely suggest differential misclassification when true relative sensitivity is 1.0 and the odds ratio is > 1.0 with an imperfect external standard.
- ARS may misrepresent the magnitude of differential misclassification even when correctly detected.
Conclusions:
- The interpretation of ARS > 1.0 requires careful consideration of the external standard's accuracy and underlying epidemiological parameters.
- ARS is not always a reliable indicator or accurate quantifier of differential exposure misclassification.