18. Sensitivity, Specificity, PPV, and the Biases of Screening
After this lesson you can
- Place screening on the natural history timeline and distinguish it from diagnosis
- Build the screening table and compute sensitivity, specificity and both predictive values with the right denominators
- Predict how prevalence changes the predictive values while sensitivity and specificity stay fixed
- Explain the cutoff trade-off, the receiver operating characteristic idea, and serial versus parallel testing
- Recognize lead-time bias, length bias, overdiagnosis and volunteer bias
- Judge a screening program by cause-specific mortality and the nine program conditions
Video coming soon
Scheduled for release on November 15, 2026.
A lesson on screening and test evaluation using the fictional Oakhaven screening program and a test with 90 percent sensitivity and 90 percent specificity. It builds the screening table, shows how the positive predictive value falls from 90 to 50 to 8.3 percent as prevalence drops, and explains cutoffs, serial and parallel testing, and the biases that flatter screening programs.
Questions and answers
Questions a learner would ask about this lesson. Think through your own answer before you open each one.
- How is screening different from diagnosis?
Show answer
Screening tests apparently well people, usually many at once, to sort them into those likely and unlikely to have the condition. Diagnosis is applied to a person with symptoms or a positive screen, and its job is to confirm or rule out the condition. A positive screen means look closer; it is not a diagnosis.
- How is the screening table laid out, and which way do I read it?
Show answer
Test result goes in the rows with positive on top, and true status goes in the columns with disease on the left, giving the cells TP, FP, FN and TN. Sensitivity, TP / (TP + FN), and specificity, TN / (TN + FP), are read down the columns and describe the test. The positive predictive value, TP / (TP + FP), and the negative predictive value, TN / (TN + FN), are read across the rows and describe the people who got a result.
- Why does the same test give such different positive predictive values in different places?
Show answer
True positives come from the sick group and false positives come from the well group, so when prevalence falls, the well group grows and false positives multiply. With sensitivity and specificity both 90 percent, the positive predictive value was 90 percent at 50 percent prevalence, 50 percent at 10 percent prevalence, and 8.3 percent at 1 percent prevalence. The test did not change; the population did.
- What is lead-time bias?
Show answer
Lead-time bias happens when screening moves the date of diagnosis earlier without moving the date of death. In the lesson, two residents both die at year 5; one diagnosed by symptoms at year 3 survives 2 years after diagnosis, and one diagnosed by screening at year 1 survives 4 years. Survival doubled, but life did not lengthen, and the 2-year gap is the lead time.
- What outcome should be used to judge whether a screening program works?
Show answer
The right outcome is cause-specific mortality in populations offered and not offered the program, ideally randomized. Survival after diagnosis, cases found and test accuracy can all be inflated by lead time, length bias and overdiagnosis. In the two-town comparison, five-year survival rose from 50 to 80 percent while deaths stayed at 60 in each town.
Practice quiz
- Question 1A screening table has TP 90, FP 90, FN 10 and TN 810. What is the sensitivity?
- Question 2With the same cells, what is the positive predictive value?
- Question 3The same test is used in a population with lower prevalence. What happens to sensitivity and positive predictive value?
- Question 4In a low-risk group of 10,000 with 100 true cases, a test with 90 percent sensitivity and 90 percent specificity gives TP 90 and FP 990. What is the positive predictive value?
- Question 5Two residents die at year 5. One is diagnosed by symptoms at year 3, the other by screening at year 1. The longer survival after diagnosis is:
- Question 6Which outcome is right for judging whether a screening program saves lives?
- Question 7Lowering the cutoff on a test that gives a number usually does what?
- Question 8Two tests, each with 90 percent sensitivity and 90 percent specificity, are used in serial, so a person is positive only if both tests are positive. What are the combined sensitivity and specificity?
- Question 9How does overdiagnosis differ from a false positive?
- Question 10An accurate test exists for a condition, but no effective early treatment is available. What does the lesson conclude about running a screening program?
Saved in your browser only: no account, no server.
Principles of Epidemiology is education in epidemiologic reasoning, not clinical, screening, or policy guidance. Every example is set in the fictional Oakhaven Health District, and every number is invented for teaching.