Chemistry, toxicology & urinalysis

Management

Sensitivity and specificity



Last staff update: 9 April 2025 (update in progress)

Copyright: 2022-2025, PathologyOutlines.com, Inc.

PubMed Search: Sensitivity and specificity

Rongrong Huang, Ph.D.
Page views in 2025 to date: 69
Cite this page: Huang R. Sensitivity and specificity. PathologyOutlines.com website. https://www.pathologyoutlines.com/topic/chemistrysensitivityspecificity.html. Accessed April 29th, 2025.
Definition / general
  • Diagnostic (or clinical) sensitivity and specificity are measures of diagnostic accuracy of a test, which are essential indicators to assess clinical test performance; analytical sensitivity and specificity are out of the scope of this chapter
  • Understanding test sensitivity and specificity and how they apply to various patient populations with different disease prevalence is important for result interpretation and test selection
Essential features
  • Sensitivity and specificity represent a summary of the diagnostic accuracy of a test; they are indicators of clinical test performance and independent of disease status or disease prevalence
  • Receiver operating characteristic (ROC) curve and area under the curve (AUC) score are useful tools for assessing sensitivity and specificity, selecting optimal cutoff points and comparing diagnostic accuracy among different methods
  • Predictive value is influenced by both diagnostic accuracy and disease prevalence; it is useful information to aid in result interpretation and clinical decision making
  • 2 x 2 truth table is a useful tool for calculation of sensitivity, specificity, false positive rate, false negative rate, positive predictive value (PPV), negative predictive value (NPV) and likelihood ratio
Terminology
  • Diagnostic accuracy (clinical accuracy): the ability of a diagnostic test to discriminate between diseased and nondiseased subjects or between 2 or more clinical states
  • Sensitivity (diagnostic / clinical): the ability of a test to give a positive result for subjects who have the disease or condition for which they are being tested
  • Specificity (diagnostic / clinical): the ability of a test to give a negative result for subjects who do not have the disease or condition for which they are being tested
  • True positive (TP): positive test result in a subject in whom the disease or condition of interest is present
  • True positive rate: equivalent to sensitivity
  • True negative (TN): negative test result in a subject in whom the disease or condition of interest is absent
  • True negative rate: equivalent to specificity
  • False positive (FP): positive test result for a subject in whom the disease or condition of interest is absent
  • False positive rate: equivalent to 1 - specificity
  • False negative (FN): negative test result in a subject in whom the disease or condition of interest is present
  • False negative rate: equivalent to 1 - sensitivity
  • Receiver operating characteristic (ROC) curve: a graphical description of test performance representing the relationship between the true positive rate (sensitivity) and the false positive rate (1 - specificity)
  • Likelihood ratio (LR): calculated by dividing the probability of test result in persons with disease by the probability of same result in persons with no disease
  • Prevalence: the probability of a particular clinical state in a specified population or subpopulation at a given point in time, also known as pretest probability
  • Positive predictive value (PPV): the probability that a positive test indicates presence of disease
  • Negative predictive value (NPV): the probability that a negative test indicates absence of disease
  • Reference: CLSI: CLSI EP24 [Accessed 17 February 2025]
Diagrams / tables

Contributed by Rongrong Huang, Ph.D.
Example of ROC curve

Example of ROC curve



Truth table format
Result Disease Nondisease Total
Positive TP FP TP+FP
Negative FN TN FN+TN
Total TP+FN FP+TN TP+FP+FN+TN

PPV = TP/(TP+FP); NPV = TN/(TN+FN); sensitivity = TP/(TP + FN); specificity = TN/(TN+FP)

Example truth table
Result Disease Nondisease Total
Positive 190 30 220
Negative 10 170 180
Total 200 200 400

PPV = 86%; NPV = 94%; sensitivity = 95%; specificity = 85%
Assessing sensitivity and specificity
  • Sensitivity and specificity are indicators of a test's ability to distinguish between disease / condition and absence of disease / condition at a chosen cutoff
  • They are calculated based on the observed results and the status determined by a diagnostic gold standard (i.e., truth)
  • 2 x 2 truth table as illustrated in the first table in Diagrams / tables is commonly used to guide the calculation; a calculation example is given in the second table (see Terminology for definitions) (McPherson: Henry's Clinical Diagnosis and Management by Laboratory Methods, 23rd Edition, 2016)
  • Altering a cutoff changes a test's sensitivity and specificity and there is a trade off between sensitivity and specificity (or FP and FN rates)
  • Clinical utility of a test (e.g., screening versus confirmatory) is a critical consideration for assay / method selection; for example, high sensitivity is preferred for screening tests to minimize false negative results, while high specificity is needed for confirmatory tests to reduce false positive results
ROC curve and AUC score
  • Inverse relationship between sensitivity and specificity across a range of cutoffs can be plotted in an ROC curve as illustrated in Diagrams / tables; the diagnostic accuracy of a test is indicated by the area under the curve (AUC), usually ranging between 0.5 and 1.0
  • AUC score of 0.5 is indicative of equivalence of random guess, while an AUC score of 1.0 is indicative of a perfect test (no overlapping results between disease and nondisease populations); in general, an AUC score ≥ 0.8 is considered satisfactory
  • Optimal cutoff is identified by the coordinate that maximizes discriminatory power, which is the point closest to the upper left corner (true positive rate of 1.0 and false positive rate of 0); in the example illustrated in Diagrams / tables, point C is the optimal cutoff, with point A having high sensitivity but low specificity and point B having high specificity but low sensitivity
  • ROC curve and AUC score can be used to compare the diagnostic accuracy of 2 methods; if one test has an ROC curve with a greater AUC than a comparison test, it has better sensitivity and specificity at all cutoffs and therefore, it has an overall higher diagnostic accuracy
  • For test developers or clinical laboratories, diagnostic accuracy can be evaluated according to the Clinical and Laboratory Standards Institute (CLSI) guideline EP24A2 (CLSI: CLSI EP24 [Accessed 17 February 2025])
Likelihood ratios
  • Likelihood ratio (LR) is another assessment of test performance that combines sensitivity and specificity into a single number; like sensitivity and specificity, LR is independent of disease status in the testing patient or disease prevalence in the testing population
  • Positive LR is calculated as sensitivity / (1 - specificity); negative LR is calculated as (1 - sensitivity) / specificity
Predictive value and prevalence
  • Clinically, it is useful to know how the test result changes the probability of disease (predictive value), given certain test performance and disease prevalence
  • Predictive value of a test is influenced by both diagnostic accuracy (sensitivity and specificity) and disease prevalence (pretest probability)
  • Predictive value of a positive result is highly dependent on the prevalence of the disease being tested and increases with increasing prevalence and improved accuracy
  • For a disease with low prevalence, even a test with high diagnostic accuracy will yield a low positive predictive value because most positive test results will be false positives
  • Testing patients with high clinical suspicion increases the positive predictive value of a test
  • In theory, predictive value (posttest probability) can be calculated as (sensitivity x pretest probability) / ([sensitivity x pretest probability] + [{1 - specificity} x {1 - pretest probability}]) (McPherson: Henry's Clinical Diagnosis and Management by Laboratory Methods, 23rd Edition, 2016)
Practice question #1
Which of the following best describes test specificity?

  1. Probability that someone with a negative test result does not have the disease
  2. Probability that someone with a positive test result has the disease
  3. Proportion of people without the disease who have a negative test result
  4. Proportion of people without the disease who have a positive test result
  5. Proportion of people with the disease who have a positive test result
Practice answer #1
C. Proportion of people without the disease who have a negative test result. Specificity is calculated as TN/(TN+FP), where TN and FP are the true negative and false positive numbers, respectively. Answer E is incorrect because it describes test sensitivity. Answer D is incorrect because it describes false positive rate (1 - specificity). Answer B is incorrect because it describes test positive predictive value (PPV). Answer A is incorrect because it describes test negative predictive value (NPV).

Comment Here

Reference: Sensitivity and specificity
Practice question #2

Result Disease Nondisease Total
Positive 170 30 200
Negative 10 190 200
Total 180 210 400

The laboratory is validating an assay that detects anti-dsDNA in patients with systemic lupus erythematosus (SLE). The validation study results are shown in the table above. Which of the following represents the sensitivity and specificity of the test?

  1. Sensitivity 85%; specificity 95%
  2. Sensitivity 90%; specificity 85%
  3. Sensitivity 91%; specificity 94%
  4. Sensitivity 94%; specificity 90%
  5. Sensitivity 95%; specificity 90%
Practice answer #2
D. Sensitivity 94%; specificity 90%. Sensitivity is calculated as TP/(TP+FN) where TP and FN are the true positive and false negative numbers, respectively. In this case, sensitivity = 170/(170+10) = 170/180 = 94%. Specificity is calculated as TN/(TN+FP) where TN and FP are the true negative and false positive numbers, respectively. In this case, specificity = 190/(190+30) = 190/210 = 90%. Answers A, B, C and E are incorrect because the values do not match the calculated sensitivity and specificity based on the numbers provided in the table.

Comment Here

Reference: Sensitivity and specificity
Back to top
Image 01 Image 02