Related Experiment Videos
Misclassification of exposure in case-control studies: assessment by quality indices
1Department of Community Health, University of Auckland, New Zealand.
Epidemiology (Cambridge, Mass.)
|May 1, 1994
Summary
Misclassification in case-control studies is better measured by quality indices than sensitivity and specificity. These indices reveal when exposure misclassification bias affects odds ratios, offering a new correction method.
Area of Science:
- Epidemiology
- Biostatistics
Background:
- Misclassification of binary exposures in case-control studies is typically evaluated using sensitivity and specificity.
- Sensitivity and specificity alone are insufficient for quantifying misclassification's impact as they omit predictive values.
Purpose of the Study:
- To introduce quality indices as a more comprehensive measure of exposure misclassification.
- To explore the relationship between measured and actual odds ratios using these quality indices.
Main Methods:
- Defining quality indices as rescaled sensitivity, specificity, and predictive values.
- Deriving the relationship between measured and actual odds ratios based on quality indices.
Main Results:
- Quality indices provide a more adequate assessment of misclassification than traditional measures.
- Nondifferential or proportional quality indices indicate no bias in the odds ratio.
Conclusions:
- Quality indices offer a superior method for assessing exposure misclassification in epidemiological studies.
- The derived relationship facilitates a novel approach to correcting measured odds ratios for misclassification bias.