Questions: Screening, Positive Predictive Value, and Disease Prevalence

Question 1

A test for Disease X has 95% sensitivity and 95% specificity. Applied to a population where prevalence is 1%, approximately what fraction of positive test results will be true positives (the positive predictive value)?

• A. 95%, because the test is 95% accurate in both directions
• B. About 50%, because random chance in a near-zero-prevalence population gives roughly even odds
• C. About 16%, because the large pool of disease-free people at 1% prevalence generates far more false positives (495) than true positives (95) among 10,000 screened people
• D. About 5%, because only the false positive rate (5%) matters in a low-prevalence population

Answer: (C)

At 1% prevalence in 10,000 people: 100 have disease → 95 true positives (95% sensitivity). 9,900 are disease-free → 495 false positives (5% of 9,900). PPV = 95 / (95 + 495) ≈ 16%. The test's 95% accuracy is real, but in a low-prevalence population the sheer number of disease-free people means even a 5% false positive rate generates far more false alarms than true detections. This is the core counterintuitive result — the same test with 68% PPV at 10% prevalence has only 16% PPV at 1% prevalence.

Question 2

A public health department screens the general adult population for a rare autoimmune condition (prevalence 0.1%) using a test with 99% sensitivity and 99% specificity. What is the primary concern with this program?

• A. The sensitivity is too low — the test will miss most cases in such a rare disease
• B. The overwhelming majority of positive results will be false positives, subjecting many healthy people to unnecessary follow-up procedures, anxiety, and potential iatrogenic harm
• C. The specificity is insufficient for a general population screening program
• D. Rare diseases cannot be detected through screening regardless of test performance

Answer: (B)

At 0.1% prevalence in 100,000 people: 100 have disease → 99 true positives. 99,900 disease-free → 999 false positives (1% of 99,900). PPV ≈ 99/1,098 ≈ 9%. Even a 99%/99% test produces roughly 10 false positives for every true positive in this setting. Each false positive may receive biopsy, imaging, or specialist referral — procedures that carry their own risks. The harm-to-benefit ratio turns unfavorable when the false positive cascade affects many more people than the disease itself.

Question 3

A test with 95% sensitivity and 95% specificity correctly identifies disease in 95% of people who test positive.

• T. True
• F. False

Answer: False

This is the most common misconception about diagnostic test performance. Sensitivity (95%) is the probability of a positive result given disease is present — it tells you nothing about what a positive result means in a given population. The probability of disease given a positive result (PPV) depends critically on prevalence. At 1% prevalence, the PPV is approximately 16% despite 95%/95% test accuracy. Sensitivity and specificity are intrinsic test properties; PPV is a population-dependent property. Conflating them is a dangerous clinical error.

Question 4

Lead time bias can make a screening program appear to improve survival even when detected patients die at the same calendar time they would have died without screening.

• T. True
• F. False

Answer: True

Lead time bias occurs when earlier detection via screening advances the diagnosis date but does not alter the date of death. The patient's measured survival (from diagnosis to death) is artificially extended because the clock started earlier, even though their life was not actually prolonged. This creates the illusion of improved survival statistics from screening. It is one of the principal mechanisms by which screening programs can show apparent benefit — longer survival from diagnosis — while failing to reduce mortality in randomized trials where the comparison is death rates per 100,000, not survival time from diagnosis.

Question 5

A test has 90% sensitivity and 90% specificity. Explain why the positive predictive value will be much lower when screening a population where 1% have the disease versus a population where 20% have the disease.

Model answer: At 1% prevalence (1,000 people): 10 have disease → 9 true positives; 990 disease-free → 99 false positives (10% of 990). PPV = 9/108 ≈ 8%. At 20% prevalence (1,000 people): 200 have disease → 180 true positives; 800 disease-free → 80 false positives. PPV = 180/260 ≈ 69%. The difference is determined by the ratio of true positives to false positives. When disease is rare, the large pool of healthy people generates many more false positives than there are true cases, swamping the signal. When disease is common, true positives dominate. Sensitivity and specificity are unchanged — only prevalence differs.

This numerical exercise is the core reasoning tool for evaluating screening programs. The key insight is that false positives grow with the size of the disease-free population, which is vast when prevalence is low. PPV is essentially asking: 'Given a positive result, what are the odds the disease is actually present?' Bayes' theorem formalizes this, but the 2×2 table arithmetic makes the mechanism transparent and is the approach recommended for clinical decision-making.