Comparing dichotomous screening tests when individuals negative on both tests are not verified

Area of Science:

• Medical Diagnostics
• Biostatistics
• Epidemiology

Background:

• Comparing dichotomous screening tests involves verification of positive results.
• Direct estimation of true positive and false positive rates for each test is often not feasible.

Purpose of the Study:

• To derive unbiased estimates of relative true positive and false positive rates for screening tests.
• To quantify the trade-off between increased true positives and false positives when comparing two screening tests.

Main Methods:

• Utilizing a sampled population where both tests are performed and positive results are verified.
• Deriving a ratio representing extra false positives detected per extra true positive detected.
• Calculating a 95% confidence interval for this ratio.

Main Results:

• Unbiased estimates of relative test performance can be obtained.
• A prevalence-dependent ratio quantifies the performance trade-off.
• Methods are provided to estimate this ratio for different target population prevalences under specific conditions.

Conclusions:

• The derived ratio offers a method to compare screening test performance when direct estimation is not possible.
• Understanding this trade-off is crucial for selecting appropriate screening tests.
• Estimating performance in different populations requires knowledge of test characteristics, relative prevalence, or specific assumptions.