The 2x2 Diagnostic Test Table
The 2x2 table organizes every patient into one of four cells based on test result (positive or negative) and true disease status (present or absent). From these cells you derive the core metrics of diagnostic testing: sensitivity and specificity (fixed properties of the test), predictive values (which depend on prevalence), and likelihood ratios (which work across populations). Mastering the layout and knowing which cells feed which formula is the foundation of clinical test interpretation.
Test-Positive Row (TP and FP)
A positive test in a patient who truly has disease is a True Positive (TP). A positive test in a patient who does NOT have disease is a False Positive (FP). This row is the numerator/denominator source for Positive Predictive Value: PPV = TP / (TP + FP), the probability that disease is actually present given a positive result. False positives drive real harm — unnecessary procedures, sedation and perforation risks, and overdiagnosis — which is why we don't screen everyone for rare disease.
Test-Negative Row (FN and TN)
A negative test in a patient who truly has disease is a False Negative (FN) — a missed diagnosis. A negative test in a patient without disease is a True Negative (TN). This row feeds Negative Predictive Value: NPV = TN / (TN + FN), the probability of no disease given a negative result. False negatives are the danger of low-sensitivity screening tests; missing disease is exactly what a good screening test must avoid.
Column Totals: Disease Present vs Disease Absent
Reading down the columns, not across the rows, gives the test's intrinsic properties. The 'disease present' column (all diseased patients) is used for Sensitivity = TP / (TP + FN), the proportion of diseased correctly identified. The 'disease absent' column (all non-diseased) is used for Specificity = TN / (TN + FP), the proportion of non-diseased correctly identified. Sensitivity and specificity are properties of the TEST and stay fixed regardless of population.
Likelihood Ratios from the Table
Likelihood ratios combine the vertical (test-property) metrics and express how much a result shifts disease probability. LR+ = Sensitivity / (1 − Specificity); LR− = (1 − Sensitivity) / Specificity. Interpretation: LR+ >10 is strong evidence for disease, 5–10 moderate, 2–5 weak, and 1 useless; LR− <0.1 is strong evidence against disease and 0.1–0.2 moderate. Because they derive from sensitivity and specificity, LRs are independent of prevalence and can be applied across different populations using: post-test odds = pre-test odds × LR.
Prevalence and Predictive Values
Predictive values, unlike sensitivity and specificity, depend on prevalence. When prevalence is LOW, even a highly specific test produces many false positives, so PPV falls while NPV rises (a negative test is reassuring). When prevalence is HIGH, PPV rises and NPV falls. Classic example: a test with 90% sensitivity and 85% specificity in a 1% prevalence population yields a PPV of only about 5% — most positives are false positives. This is the statistical reason we avoid screening low-prevalence populations.
High-yield
- SnNout: a highly SeNsitive test, when Negative, rules OUT disease.
- SpPin: a highly SPecific test, when Positive, rules IN disease.
- Sensitivity = TP/(TP+FN); Specificity = TN/(TN+FP) — read down the columns.
- PPV = TP/(TP+FP); NPV = TN/(TN+FN) — read across the rows.
- Sensitivity and specificity are fixed properties of the test; predictive values vary with prevalence.
- Low prevalence → low PPV, high NPV even for an excellent test.
- LR+ = Sens/(1−Spec); LR− = (1−Sens)/Spec; LRs are prevalence-independent.
- LR+ >10 strong for disease; LR− <0.1 strong against disease; LR = 1 is useless.
- Screening tests prioritize high sensitivity (don't miss disease); confirmatory tests prioritize high specificity (don't misdiagnose).
- Post-test odds = pre-test odds × likelihood ratio.
Pitfalls
- Confusing sensitivity/specificity (test properties) with PPV/NPV (population-dependent) — a frequent trap.
- Assuming a positive test means disease is present when prevalence is low; most positives may be false positives.
- Reading the table across the rows for sensitivity/specificity — those come from the columns.
- Forgetting that likelihood ratios, unlike PPV, stay constant across populations.
- Interpreting 'finding disease' as 'preventing death' — high sensitivity for polyps does not mean every detected lesion is lethal.
- Believing a good test justifies universal screening; low prevalence generates more false positives than true positives.
Clinical pearls
- A negative result on a highly sensitive test is powerful for ruling disease out (SnNout).
- A positive result on a highly specific test is powerful for ruling disease in (SpPin).
- The same test performs very differently in a high- vs low-prevalence population.
- When counseling on screening harms, false positives and overdiagnosis are the concrete downsides of a positive test in a healthy person.
- Use likelihood ratios at the bedside — they let you update probability without recalculating for each population.
Frequently asked
Which cells of the 2x2 table do I use to calculate sensitivity?
Sensitivity uses the 'disease present' column: Sensitivity = TP / (TP + FN), the proportion of all diseased patients correctly identified as positive.
Why does PPV drop in low-prevalence populations even with an excellent test?
With few diseased people, the large non-diseased group generates many false positives. Since PPV = TP/(TP+FP), those false positives swamp the true positives — for example, a 90% sensitive, 85% specific test at 1% prevalence gives a PPV of only about 5%.
How do sensitivity/specificity differ from predictive values?
Sensitivity and specificity are fixed properties of the test itself. Predictive values (PPV and NPV) depend on the prevalence of disease in the population being tested and therefore vary by setting.
What do SnNout and SpPin mean?
SnNout: a highly SeNsitive test, when Negative, rules OUT disease. SpPin: a highly SPecific test, when Positive, rules IN disease. Sensitivity is prioritized for screening; specificity for confirmatory testing.
Why are likelihood ratios more useful across populations than PPV?
Likelihood ratios are derived from sensitivity and specificity, which are prevalence-independent. So an LR can be applied to any population by combining it with that patient's pre-test odds: post-test odds = pre-test odds × LR.
How do I interpret an LR+ of 12 versus an LR− of 0.08?
An LR+ greater than 10 is strong evidence for disease, so 12 substantially raises the probability of disease. An LR− below 0.1 is strong evidence against disease, so 0.08 substantially lowers it.
What is the difference between a false positive and a false negative in the table?
A false positive is a positive test in a person without disease (test-positive row, disease-absent column). A false negative is a negative test in a person who actually has disease (test-negative row, disease-present column) — a missed diagnosis.
Turn this into reasoning you can use on exam day — practice The 2x2 Diagnostic Test Table on branching cases where your decisions shape the patient.