|
Activity Number:
|
227
|
|
Type:
|
Invited
|
|
Date/Time:
|
Tuesday, August 8, 2006 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
| Abstract - #305278 |
|
Title:
|
Evaluation of Diagnostic Tests in Studies with Verification Bias
|
|
Author(s):
|
Marina Kondratovich*+
|
|
Companies:
|
U.S. Food and Drug Administration
|
|
Address:
|
HFZ-550, 1350 Piccard Drive, Rockville, MD, 20850,
|
|
Keywords:
|
diagnostic accuracy ; sensitivity and specificity ; verification bias ; multiple imputation ; types of missingness
|
|
Abstract:
|
It is well-known that in studies with verification bias, ratios of sensitivities and false positive rates are unbiased estimates. We will discuss whether the information about these ratios is enough to make a conclusion about effectiveness of the test under investigation. We also will discuss multiple imputation techniques for obtaining unbiased estimates of the tests' diagnostic accuracies and different mechanisms of missing disease status, missing completely at random, and missing at random. These types of missingness will be demonstrated by examples from the studies of comparison of the prostate-specific antigen assay and digital rectal exam for detection of prostate cancer and others. If different types of missingness is not addressed properly, the adjustment for the verification bias can produce biased estimates of the diagnostic accuracies of the tests.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2006 program |
