What is the difference between Sensitivity and Positive Predictive Value (PPV)?

Sensitivity measures the probability that a diseased individual tests positive, which is an intrinsic property of the diagnostic assay. PPV measures the probability that an individual who tested positive truly has the disease, which depends directly on disease prevalence in the tested population.

Why are Likelihood Ratios (LR+ and LR-) superior to Predictive Values?

Unlike PPV and NPV, Likelihood Ratios do not fluctuate with disease prevalence across different clinical settings. They quantify exactly how many times more (or less) likely a test result is in diseased versus non-diseased patients, enabling direct Bayesian calculation of post-test probability using Fagan's nomogram.

Diagnostic Test Accuracy Calculator (Sensitivity, Specificity, PPV, NPV)

Evidence-based clinical epidemiology suite. Evaluate screening and diagnostic tests: Sensitivity, Specificity, Predictive Values (PPV/NPV), Likelihood Ratios (LR+/LR-), and Wilson score 95% confidence intervals.

Index Test Result	Gold Standard (+) Condition Present	Gold Standard (−) Condition Absent
Test Positive (+)	True Positive (TP)	False Positive (FP)
Test Negative (−)	False Negative (FN)	True Negative (TN)

Clinical Methodology: Diagnostic Accuracy Assessment

In clinical medicine and evidence-based epidemiology, diagnostic test evaluation requires quantifying how effectively an index test distinguishes patients with a target condition from those without it, benchmarked against an accepted reference standard (gold standard).

1. Core Operational Metrics

Consider a sample of size $N = TP + FP + FN + TN$. The primary performance parameters are defined as:

$$\text{Sensitivity (Se)} = \frac{TP}{TP + FN} = P(\text{Test } + \mid \text{Disease } +)$$

$$\text{Specificity (Sp)} = \frac{TN}{TN + FP} = P(\text{Test } - \mid \text{Disease } -)$$

Confidence Intervals: Exact asymptotic confidence intervals for proportions are computed via the Wilson Score Interval:

$$\tilde{p} = \frac{x + \frac{z^2}{2}}{n + z^2}, \quad CI_{1-\alpha} = \tilde{p} \pm \frac{z}{n + z^2}\sqrt{\frac{x(n-x)}{n} + \frac{z^2}{4}}$$

where $z = \Phi^{-1}(1 - \alpha/2) \approx 1.95996$ for the standard 95% level.

2. Predictive Values & Bayesian Prevalence Dependency

While Sensitivity and Specificity are intrinsic characteristics of the testing instrument, the Positive Predictive Value (PPV) and Negative Predictive Value (NPV) are strongly dependent on the pre-test disease prevalence ($\pi$) in the clinical population:

$$\text{PPV} = \frac{Se \cdot \pi}{Se \cdot \pi + (1 - Sp) \cdot (1 - \pi)}$$

$$\text{NPV} = \frac{Sp \cdot (1 - \pi)}{Sp \cdot (1 - \pi) + (1 - Se) \cdot \pi}$$

When screening in low-prevalence populations ($\pi < 1\%$), even a test with 99% specificity will yield substantial false positives, drastically reducing PPV.

3. Likelihood Ratios & Fagan's Nomogram

Likelihood Ratios ($LR^+$ and $LR^-$) overcome the prevalence-dependency limitation. They describe the relative odds of a given test result occurring in a patient with disease compared to one without:

$$LR^+ = \frac{\text{Sensitivity}}{1 - \text{Specificity}} = \frac{TP / (TP + FN)}{FP / (FP + TN)}$$

$$LR^- = \frac{1 - \text{Sensitivity}}{\text{Specificity}} = \frac{FN / (TP + FN)}{TN / (FP + TN)}$$

Likelihood Ratio Range	Clinical Shift in Likelihood (Sackett & Simel Criteria)	Diagnostic Utility
$LR^+ > 10$ or $LR^- < 0.1$	Generates large, conclusive changes in probability (> 45% shift)	Decisive diagnostic confirmation / exclusion
$LR^+ = 5 - 10$ or $LR^- = 0.1 - 0.2$	Generates moderate shifts in disease probability (30–45%)	Clinically useful in most diagnostic pathways
$LR^+ = 2 - 5$ or $LR^- = 0.2 - 0.5$	Generates small, often non-decisive shifts (15–30%)	Requires confirmatory testing
$LR^+ = 1 - 2$ or $LR^- = 0.5 - 1.0$	Negligible change in probability (< 15%)	Virtually uninformative test

4. Zero-Cell Continuity Correction

When any contingency cell equals zero ($TP=0$, $FP=0$, $FN=0$, or $TN=0$), likelihood ratios and diagnostic odds ratios incur mathematical division by zero. In compliance with Haldane (1955) and Anscombe (1956), LabStats applies a standard $+0.5$ empirical continuity correction to all four cells, guaranteeing bounded, robust estimators.

5. References & Clinical Guidelines

Bossuyt, P. M., et al. (2015). STARD 2015: An Updated List of Essential Items for Reporting Diagnostic Accuracy Studies. BMJ, 351, h5527.
Sackett, D. L., Haynes, R. B., & Tugwell, P. (1991). Clinical Epidemiology: A Basic Science for Clinical Medicine. Little, Brown and Company.
Simel, D. L., Samsa, G. P., & Matchar, D. B. (1991). Likelihood ratios with confidence: sample size estimation for diagnostic test studies. Journal of Clinical Epidemiology, 44(8), 763–770.
Wilson, E. B. (1927). Probable inference, the law of succession, and statistical inference. Journal of the American Statistical Association, 22(158), 209–212.