Positive Predictive Value: Why a Good Test Can Still Produce Mostly False Alarms

One Test, Two Waiting Rooms

Imagine testing 10,000 people. For illustration only, hold sensitivity and specificity fixed at 90% in both settings. These are hypothetical inputs, not performance estimates for any real test.

In the first setting, 900 of 1,800 positive results are true positives. In the second, only 90 of 1,080 are. A positive result did not become less technically accurate by arithmetic alone; the pool of people without the condition became much larger, creating more opportunities for false positives.

What the Calculation Does—and Does Not—Assume

PPV equals true positives divided by all positive results. If sensitivity, specificity, and condition prevalence are known for the same intended-use setting, the equivalent calculation is:

PPV = (sensitivity × prevalence) ÷ [(sensitivity × prevalence) + ((1 − specificity) × (1 − prevalence))]

The table changes prevalence while deliberately freezing sensitivity and specificity. Real patient groups can differ in disease severity, competing diagnoses, prior testing, specimen quality, and how a test is performed. Sensitivity and specificity may change too. Recomputing PPV from a new prevalence is therefore a scenario analysis, not external validation.

A Five-Question Review Before Deployment

1. Intended use: Is the test for screening, triage, diagnosis, or another role? At what point in the care pathway is it applied?
2. Population: Were participants recruited from the same setting and eligibility rules as the proposed users, or were obvious cases compared with healthy controls?
3. Reference standard: How was the target condition established, and did people with negative index tests receive adequate verification?
4. Full 2×2 table: Are true positives, false positives, false negatives, and true negatives reported alongside uncertainty intervals?
5. Consequences: What action follows a positive result, and are the downstream false-positive burden and missed cases acceptable for that action?

A high PPV alone does not prove clinical utility. That requires a decision context and evidence about what happens when the result changes care.

Where Aqrab Fits

When reviewing a diagnostic accuracy paper or clinical-AI claim, use Aqrab to pressure-test the intended population, reference standard, 2×2 counts, and leap from test performance to clinical usefulness. The critique should expose which assumptions still need local validation; it cannot make a transported PPV true by assertion.

Source Anchors

The FDA diagnostic-test reporting guidance defines sensitivity, specificity, PPV, and the 2×2 table. The STARD 2015 statement and its explanation and elaboration guide reporting and applicability assessment. These are reporting and methodological sources, not validation of the hypothetical test above.