Screening is the testing of people who have no symptoms to find disease (or a risk factor) at an early, more treatable stage. It is secondary prevention. A positive screening test is not a diagnosis; it must be followed by a diagnostic (confirmatory) test.

• Mass (population) screening: everyone in a group (for example, newborn screening).
• Selective (targeted) screening: only high-risk groups (for example, hepatitis C testing in people who inject drugs).
• Multiphasic screening: several tests at one event (a health fair).

Validity vs. reliability

A test can be reliable but not valid (a scale that is consistently 2 kg too heavy). A test that is not reliable cannot be valid.

The 2 × 2 table

Memory aids: Sensitivity reads down the disease column; predictive values read across the test row. SnNout — a highly sensitive test with a negative result rules disease out. SpPin — a highly specific test with a positive result rules disease in.

Prevalence affects predictive values, not sensitivity or specificity.

• Sensitivity and specificity are properties of the test (with a fixed cutoff) and do not change with prevalence.
• As prevalence falls, PPV falls (more false positives relative to true positives) and NPV rises.
• This is why screening is most efficient in high-risk (higher-prevalence) groups.

Choosing the cutoff (trade-off). Lowering the cutoff for a positive result raises sensitivity and lowers specificity; raising it does the opposite.

• Favor high sensitivity when missing a case is dangerous and treatment is effective (newborn phenylketonuria, HIV screening in blood donors, a serious infectious disease). The first test in a screening sequence is usually highly sensitive.
• Favor high specificity when false positives cause harm, cost, or invasive follow-up, and for rare diseases, where even a small false-positive rate swamps the true positives. The confirmatory test is usually highly specific.

Two tests together

• Sequential (series) testing: positive only if both tests are positive → net specificity rises, net sensitivity falls. Typical screening-then-confirmation pathway (for example, HIV antigen/antibody screen followed by a differentiation assay).
• Simultaneous (parallel) testing: positive if either test is positive → net sensitivity rises, net specificity falls.

Criteria for a good screening program (Wilson and Jungner, WHO 1968, summarized)

1. The condition is an important health problem.
2. It has a recognizable latent or early symptomless stage.
3. An accepted, effective treatment exists, and early treatment improves outcomes.
4. A suitable, acceptable, safe test is available.
5. Facilities for diagnosis and treatment are available.
6. The natural history of the condition is understood.
7. There is an agreed policy on whom to treat.
8. The cost is balanced against overall health spending, and case finding is a continuing process.

Screening biases (why survival data can mislead)

• Lead-time bias: earlier detection makes survival after diagnosis look longer even if death occurs at the same time.
• Length-time bias: screening preferentially finds slow-growing disease with a better prognosis.
• Overdiagnosis: finding disease that would never have caused symptoms (a concern in prostate and some thyroid and breast cancers).
• Volunteer (healthy screenee) bias: people who choose screening tend to be healthier.

Randomized trials that measure disease-specific mortality avoid these biases.

US recommendation system. The US Preventive Services Task Force (USPSTF) grades preventive services: A (recommend; high certainty of substantial net benefit), B (recommend; high certainty that the net benefit is moderate, or moderate certainty that the net benefit is moderate to substantial), C (offer selectively based on professional judgment and client preference), D (recommend against), I (insufficient evidence). Under the Affordable Care Act, non-grandfathered private health plans must cover A- and B-graded services without cost sharing; the US Supreme Court upheld this requirement in Kennedy v. Braidwood Management (2025).

Each answer below was calculated twice.

Example 1 — Full 2 × 2 analysis. A new test is given to 200 people; a gold-standard test later confirms disease status.

• Sensitivity = 45 ÷ 50 = 0.90 = 90%
• Specificity = 135 ÷ 150 = 0.90 = 90%
• PPV = 45 ÷ 60 = 0.75 = 75%
• NPV = 135 ÷ 140 = 0.964 = 96.4%
• Accuracy = (45 + 135) ÷ 200 = 180 ÷ 200 = 90%
• Prevalence = 50 ÷ 200 = 25%
• Positive likelihood ratio = 0.90 ÷ (1 − 0.90) = 0.90 ÷ 0.10 = 9

Example 2 — Same test, low prevalence. The same test (sensitivity 90%, specificity 90%) is used in 10,000 people where prevalence is 1%.

• With disease = 10,000 × 0.01 = 100 → TP = 100 × 0.90 = 90; FN = 10
• Without disease = 9,900 → TN = 9,900 × 0.90 = 8,910; FP = 990
• PPV = 90 ÷ (90 + 990) = 90 ÷ 1,080 = 8.3%
• NPV = 8,910 ÷ (8,910 + 10) = 8,910 ÷ 8,920 = 99.9%

Interpretation: at 25% prevalence, 3 of 4 positives are true; at 1% prevalence, fewer than 1 in 10 are. Sensitivity and specificity did not change.

Example 3 — Two tests (assume the tests are independent). Test A: sensitivity 90%, specificity 80%. Test B: sensitivity 80%, specificity 95%.

• Series: net sensitivity = 0.90 × 0.80 = 72%; net specificity = 1 − (0.20 × 0.05) = 1 − 0.01 = 99%
• Parallel: net sensitivity = 1 − (0.10 × 0.20) = 1 − 0.02 = 98%; net specificity = 0.80 × 0.95 = 76%

Running a community screening event

1. Confirm the condition meets screening criteria and the target group follows current USPSTF or specialty recommendations.
2. Arrange referral pathways before screening: every positive needs confirmatory testing and access to treatment.
3. Train staff to a standard technique (improves reliability); calibrate equipment.
4. Obtain informed consent; explain that a positive result is not a diagnosis.
5. Record results, notify clients, and track follow-up completion.
6. Evaluate: yield (new cases found), PPV, follow-up rate, and cost per case detected.

• Select evidence-based screening for the right population and interval.
• Perform tests with a standardized technique (correct blood pressure cuff size, calibrated devices) to improve reliability and validity.
• Explain results honestly, including the chance of false positives and false negatives.
• Ensure follow-up: screening without diagnosis and treatment offers no benefit and may cause harm.
• Reduce barriers (language, transportation, cost, trust) to reach underserved groups.
• Interpret program data (PPV, yield) and recommend changes.

• Informed consent for voluntary screening; explain benefits, harms, and the need for confirmation. Newborn screening is run by each state; consent and opt-out rules and the list of conditions vary by state.
• Confidentiality: protect results under HIPAA; positive results for reportable conditions must be reported to the health department.
• Beneficence and nonmaleficence: do not offer screening without a path to diagnosis and treatment; weigh harms of overdiagnosis and false positives (anxiety, invasive procedures).
• Justice: target screening to groups with higher prevalence and poorer access, without stigmatizing them.
• Genetic Information Nondiscrimination Act (GINA) limits use of genetic test results by health insurers and employers.

Case 1 — Explaining a positive result. A client whose screening mammogram is positive asks whether she has cancer.

Correct action: explain that the screen is not diagnostic, most positive screens turn out not to be cancer, and help schedule diagnostic imaging promptly.

Why: PPV of screening tests in average-risk groups is modest; follow-up completion is the nurse's priority.

Case 2 — Low-prevalence population. A school plans to screen all students for a rare condition with a 95%-specific test.

Correct action: advise that most positives will be false and recommend targeting higher-risk students or using a more specific confirmatory test.

Why: low prevalence lowers PPV.

Case 3 — Poor agreement between examiners. Two nurses reading the same tuberculin skin tests record different induration sizes.

Correct action: retrain both on measuring induration (not redness) with a ruler and recheck agreement.

Why: this is a problem of inter-rater reliability.

• Believing sensitivity and specificity change with prevalence — only predictive values do.
• Reversing sensitivity (true positives among the diseased) and PPV (true positives among those who test positive).
• Calling screening diagnostic or primary prevention.
• Choosing high sensitivity for a rare disease with harmful follow-up — specificity matters more there.
• Confusing reliability (consistency) with validity (correctness).
• Judging a screening program by survival after diagnosis, which is inflated by lead-time and length-time bias.
• Screening without a referral and treatment plan.

• Sensitivity = TP ÷ (TP + FN); specificity = TN ÷ (TN + FP).
• PPV = TP ÷ (TP + FP); NPV = TN ÷ (TN + FN); accuracy = (TP + TN) ÷ total.
• Validity = accuracy; reliability = reproducibility (test–retest, inter-rater, intra-rater).
• Prevalence changes PPV and NPV, not sensitivity or specificity; lower prevalence → lower PPV, higher NPV.
• SnNout (sensitive test, negative result rules out); SpPin (specific test, positive result rules in).
• High sensitivity for serious, treatable disease and initial screens; high specificity for confirmation and rare disease.
• Series testing raises specificity; parallel testing raises sensitivity.
• Lead-time, length-time, and overdiagnosis bias distort survival statistics.
• USPSTF grades A–D and I; A and B services are generally covered without cost sharing under the ACA.