Sensitivity & Specificity Calculator

Compute sensitivity, specificity, PPV, NPV and accuracy with 95% CIs from a 2×2 table, plus likelihood ratios and the Youden index. For clinical research.

What each index of the 2×2 table measures

• Sensitivity = TP / (TP + FN): of the people who truly have the condition, the proportion the test detects.
• Specificity = TN / (TN + FP): of the people without the condition, the proportion the test rules out.
• Positive predictive value (PPV) = TP / (TP + FP): if the test is positive, the probability of having the condition.
• Negative predictive value (NPV) = TN / (TN + FN): if it is negative, the probability of not having it.
• Likelihood ratios: how much a positive (LR+) or negative (LR−) result multiplies the odds of having the condition. They do not depend on prevalence.
• Youden index, J = sensitivity + specificity − 1 (Youden, 1950): it sums both up in one number and is the usual criterion for choosing the cutoff on an ROC curve.

Why PPV changes with prevalence

Sensitivity and specificity are properties of the test; predictive values also depend on how many people have the condition where you use it. Take a screening test with a sensitivity of 0.85 and a specificity of 0.90.

• In a specialist clinic with 30% prevalence (1,000 people: 300 with the condition), you get 255 true positives and 70 false positives. PPV = 255 / 325 = 78.5%; NPV = 630 / 675 = 93.3%.
• In the general population with 5% prevalence (10,000 people: 500 with the condition), you get 425 true positives and 950 false positives. PPV = 425 / 1,375 = 30.9%; NPV = 8,550 / 8,625 = 99.1%.

Same test: in one setting almost 8 in 10 positives are real, in the other 3. That is why the calculator also reports your sample prevalence: if your sample does not reflect the prevalence of the population you want to generalize to (for example, because you recruited cases and controls separately), do not report PPV and NPV as if they held for that population.

Wilson confidence intervals

All five percentages come with a 95% confidence interval by Wilson’s (1927) score method. The textbook interval (p ± 1.96·√(p(1 − p)/n)) fails exactly where diagnostic accuracy uses it most: with proportions close to 100% and small groups it can exceed 100% or have zero width. Wilson’s stays within 0-100% and has better coverage with small n.

How to report it

Always give sensitivity and specificity with their intervals, the cutoff and the reference standard, plus the full 2×2 table (in the text or in supplementary material). The reporting guideline for these studies is the STARD 2015 checklist, with its 30 items; you can check your paper against STARD before you submit. If you still need to choose the cutoff, the ROC curve calculator gives the AUC and the Youden cutoff, and the full explanation is in sensitivity, specificity and ROC curves.

How to cite this tool

If you use it in your thesis or in a paper, cite it so that readers of your Methods section know how the calculation was done.

Ciudad-Fernández, V. (2026). Sensitivity and specificity calculator [Web application]. Ciudad Science. Retrieved October 6, 2026, from https://analisisdedatospsicologia.com/en/tools/sensitivity-specificity-calculator

For your Methods section: Sensitivity, specificity and predictive values were computed with Wilson (1927) 95% confidence intervals using the Ciudad Science sensitivity and specificity calculator (Ciudad-Fernández, 2026).

@misc{ciudadfernandez2026sensibilidad,
  author       = {Ciudad-Fern{\'a}ndez, V{\'i}ctor},
  title        = {{Sensitivity and specificity calculator}},
  year         = {2026},
  publisher    = {Ciudad Science},
  howpublished = {Web application},
  url          = {https://analisisdedatospsicologia.com/en/tools/sensitivity-specificity-calculator},
  urldate      = {2026-10-06},
  note         = {Accessed 2026-10-06},
}

You may also be interested in

• All statistical consulting services
• Free statistical calculators
• Statistics blog for psychology researchers
• About Víctor Ciudad, PhD in Psychology

Home · Contact for a quote