Sensitivity, Specificity, PPV and NPV

Overview Of Diagnostic Testing

Medical research continuously seeks to develop new diagnostic tests that are cheaper, easier, or less invasive than standard ones.
A new diagnostic test must be systematically compared against an established gold standard test.
The gold standard test provides a definitive and absolute diagnosis of a particular clinical condition.
Researchers apply both the new diagnostic test and the gold standard test to a specific group of individuals.
Participants will include some individuals who truly have the disease and some who do not.
The accuracy of the new test is subsequently determined by calculating specific statistical parameters from the collected data.

The results of a diagnostic accuracy study are arranged in a 2x2 contingency table.
The columns strictly represent the true disease status as determined by the gold standard test.
The rows represent the outcome results of the new diagnostic test being evaluated.

New Diagnostic Test	Disease Present (Gold Standard)	Disease Absent (Gold Standard)	Total
Test Positive	True Positive (a)	False Positive (b)	a + b
Test Negative	False Negative (c)	True Negative (d)	c + d
Total	a + c	b + d	a + b + c + d

True Positive (a): The new test correctly identifies patients who have the disease.
False Positive (b): The new test incorrectly identifies healthy individuals as having the disease.
False Negative (c): The new test incorrectly rejects the diagnosis in patients who actually have the disease.
True Negative (d): The new test correctly rejects the diagnosis in healthy individuals.

Sensitivity is a fundamental index used to measure the performance of a diagnostic test.
It evaluates how well a specific test identifies patients who truly have the disease.
It is defined as the proportion of individuals with the disease who are correctly identified by the test as positive.
It represents the true positive rate of the diagnostic test.

Sensitivity relies entirely on the left column of the contingency table.
It is calculated by dividing the number of true positives by the total number of diseased individuals.

S e n s i t i v i t y = \frac{T r u e P o s i t i v e}{T r u e P o s i t i v e + F a l s e N e g a t i v e} = \frac{a}{a + c}

A highly sensitive test yields very few false negative results.
An ideal 100% sensitive test will be positive in all patients with the disease.
Because it rarely misses the disease, a highly sensitive test is excellent for ruling out a condition.
If a highly sensitive test yields a negative result, the clinician can confidently rule out the disease.
This clinical utility is easily remembered by the mnemonic SnOUT (Sensitivity rules OUT).
However, high sensitivity often comes at the cost of a higher false positive rate.

Specificity is the complementary index used to assess diagnostic test performance.
It evaluates how accurately a diagnostic test identifies healthy individuals.
It is defined as the proportion of individuals without the disease who are correctly identified by the test as negative.
It represents the true negative rate of the diagnostic test.

Specificity relies entirely on the right column of the contingency table.
It is calculated by dividing the number of true negatives by the total number of healthy individuals.

S p e c i f i c i t y = \frac{T r u e N e g a t i v e}{F a l s e P o s i t i v e + T r u e N e g a t i v e} = \frac{d}{b + d}

A highly specific test yields extremely few false positive results.
Because it is highly selective, it rarely mistakenly identifies a healthy person as diseased.
A highly specific test is exceptionally useful for ruling in a disease.
If a highly specific test yields a positive result, the clinician can confidently rule in the disease.
This clinical utility is easily remembered by the mnemonic SpIN (Specificity rules IN).
However, high specificity often increases the risk of yielding false negative results.
Researchers map the constant trade-off between sensitivity and specificity using a Receiver Operating Characteristic (ROC) curve.

Sensitivity and specificity are inherent characteristics of the test itself.
They evaluate diagnostic accuracy from a broad population level.
However, clinicians and patients primarily want to know the probability of having the disease given a specific test result.
Predictive values answer this question and provide accuracy at the individual patient level.

PPV measures the likelihood that an individual actually has the disease if they test positive.
It is the proportion of people with a positive test who truly harbor the disease.
It relies entirely on the top row of the contingency table.

P P V = \frac{T r u e P o s i t i v e}{T r u e P o s i t i v e + F a l s e P o s i t i v e} = \frac{a}{a + b}

NPV measures the likelihood that an individual is truly healthy if they test negative.
It is the proportion of people with a negative test who are completely free of the disease.
It relies entirely on the bottom row of the contingency table.

N P V = \frac{T r u e N e g a t i v e}{F a l s e N e g a t i v e + T r u e N e g a t i v e} = \frac{d}{c + d}

Prevalence defines the percentage of the overall population that truly has the condition.
Unlike sensitivity and specificity, the PPV and NPV are heavily influenced by the disease prevalence within the tested population.
As the prevalence of a disease decreases, the NPV automatically increases.
This occurs because false negatives become extremely rare when the disease itself is very uncommon.
Conversely, as the prevalence decreases, the PPV drops significantly.
This happens because a rare disease will generate a proportionately higher number of false positives for every single true positive.

Metric	Clinical Question Answered	Formula Derivation	Key Clinical Utility
Sensitivity	How well does the test identify diseased individuals?	$T P / (T P + F N)$	High sensitivity rules OUT disease (SnOUT).
Specificity	How well does the test identify healthy individuals?	$T N / (F P + T N)$	High specificity rules IN disease (SpIN).
PPV	If the test is positive, what is the chance I am sick?	$T P / (T P + F P)$	Predicts true sickness; decreases in rare diseases.
NPV	If the test is negative, what is the chance I am healthy?	$T N / (F N + T N)$	Predicts true health; increases in rare diseases.