Impact of Measurement Error on Diagnostic Test Accuracy
Author Information
Author(s): Fosgate GT
Primary Institution: Texas A&M University
Hypothesis
How does non-differential measurement error affect diagnostic likelihood ratios and accuracy in brucellosis testing?
Conclusion
Non-differential measurement error can bias diagnostic test results towards the null value, affecting the accuracy of diagnostic tests.
Supporting Evidence
- Non-differential measurement error can lead to biased results in diagnostic tests.
- The accuracy of a diagnostic test decreases with added measurement error.
- Likelihood ratios derived from tests can be biased towards the null value under certain conditions.
Takeaway
When tests for diseases like brucellosis have measurement errors, they can give misleading results that make it seem like the test is less effective than it really is.
Methodology
The study used simulation to evaluate the effects of non-differential measurement error on diagnostic accuracy measures like likelihood ratios and area under the ROC curve.
Potential Biases
The study could not determine the exact individuals misclassified due to the lack of a gold standard.
Limitations
The true infection status of individuals was not known, leading to potential misclassification.
Participant Demographics
The study involved 391 cattle and 381 domestic water buffalo from Trinidad.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website