What is the difference between sensitivity and specificity in diagnostic testing?

Difference Between Sensitivity and Specificity

Sensitivity measures how well a test identifies people who actually have the disease (true positives), while specificity measures how well a test identifies people who do not have the disease (true negatives). 1

Core Definitions

Sensitivity answers the question: "If the patient has the disease, what is the probability the test will be positive?" 1

• Calculated as: true positives ÷ all patients with disease 1
• Represents the percentage of diseased individuals who will have abnormal test results 1

Specificity answers the question: "If the patient does not have the disease, what is the probability the test will be negative?" 1

• Calculated as: true negatives ÷ all patients without disease 1
• Represents the percentage of non-diseased individuals who will have normal test results 1

Key Conceptual Distinction

These metrics describe the test itself, not the patient. 2 Sensitivity and specificity are derived from populations where the true disease status is already known—they tell you how the test performs, not whether an individual patient has the disease. 2

The Critical Difference from Predictive Values

• Sensitivity and specificity remain constant regardless of disease prevalence 1
• In contrast, positive and negative predictive values (which tell you about the patient's actual disease probability) change dramatically with prevalence 2, 3
• This is why sensitivity and specificity are robust test characteristics, while predictive values are context-dependent 1

Mathematical Framework

Using a 2×2 table where disease status is known:

	Disease Present	Disease Absent
Test Positive	True Positives (TP)	False Positives (FP)
Test Negative	False Negatives (FN)	True Negatives (TN)

• Sensitivity = TP / (TP + FN) 2
• Specificity = TN / (TN + FP) 2

Clinical Factors That Affect These Metrics

Sensitivity is influenced by: 1

• Disease severity (more severe disease → higher sensitivity)
• Patient effort level during testing
• Anti-ischemic medications
• If the study population has more severe disease, sensitivity will be artificially elevated 1

Specificity is affected by: 1

• Medications
• Baseline ECG abnormalities
• Left ventricular hypertrophy
• Using healthy controls instead of disease mimics artificially inflates specificity 1

The Trade-off Relationship

Sensitivity and specificity change in opposite directions. 4 When you adjust a test's threshold to increase sensitivity (catch more disease), you inevitably decrease specificity (more false positives), and vice versa. 4 This inverse relationship is fundamental to diagnostic testing and is visualized through ROC curves. 5

Common Pitfalls to Avoid

1. Never interpret sensitivity and specificity in isolation 6—they must be considered alongside predictive values and disease prevalence for meaningful clinical application 6

2. Beware of workup bias 2—when positive test results determine who gets the reference standard (e.g., angiography), this artificially inflates apparent test performance 2

3. Selection bias distorts estimates 1—using inappropriate control groups (healthy volunteers instead of patients with similar symptoms) makes specificity appear better than it truly is 1

4. The reference standard must be accurate 1—when the "gold standard" itself is imperfect, sensitivity and specificity estimates become untrustworthy 1

5. Don't apply predictive values as if they're independent of prevalence 3—this is a frequent error that leads to misinterpretation of test results 3

Practical Clinical Application

For clinical decision-making, you need both the test characteristics (sensitivity/specificity) AND the clinical context (pretest probability) to determine what a positive or negative result means for your specific patient. 2 The same test with 90% sensitivity and 90% specificity will have vastly different implications in a high-prevalence versus low-prevalence population. 2

In screening contexts, predictive values are more clinically useful than sensitivity and specificity 7, though sensitivity and specificity can guide individual screening decisions when they are extremely high (e.g., ≥95% sensitivity and ≥98% specificity). 1, 7