15.3 Evaluation of Screening Tests: Sensitivity, Specificity & Predictive Value

Evaluation of Screening Tests: Sensitivity, Specificity & Predictive Value

1. Screening Concept vs. Diagnostic Test

Screening is defined as the search for unrecognized disease or defect by applying rapid tests, examinations, or other procedures in apparently healthy target populations.

Wilson and Jungner Criteria for Screening (WHO)

1. The condition should be an important health problem.
2. There should be an accepted treatment for patients with recognized disease.
3. Facilities for diagnosis and treatment should be available.
4. There should be a recognizable latent or early symptomatic stage.
5. There should be a suitable test or examination.
6. The test should be acceptable to the population.
7. The natural history of the condition should be adequately understood.
8. There should be an agreed policy on whom to treat as patients.
9. The cost of case-finding should be economically balanced in relation to overall medical care expenditure.
10. Case-finding should be a continuing process and not a 'once and for all' project.

Key Differences Between Screening and Diagnostic Tests

---

2. The 2x2 Evaluation Matrix

To evaluate any screening test against a gold standard diagnostic benchmark, a 2x2 contingency matrix is constructed:

---

3. Core Validity Metrics: Sensitivity & Specificity

A. Sensitivity (True Positive Rate)

The probability that a diseased individual tests positive.

$\text{Sensitivity} = \frac{a}{a+c} \times 100$

• Clinical Rule (SNOUT): A Sensitive test, when Negative, rules OUT the disease. (High sensitivity means very few false negatives $c$).
• Indication: Mandatory when screening for serious, highly transmissible, or treatable conditions (e.g., Blood donor screening for HIV/Hepatitis B; Pap smear screening).

B. Specificity (True Negative Rate)

The probability that a non-diseased individual tests negative.

$\text{Specificity} = \frac{d}{b+d} \times 100$

• Clinical Rule (SPIN): A Specific test, when Positive, rules IN the disease. (High specificity means very few false positives $b$).
• Indication: Mandatory when false-positive results cause severe physical, psychological, or financial harm (e.g., Confirmatory Western Blot; initiating cancer chemotherapy).

C. Error Rates

• False Positive Rate (FPR) = $1 - \text{Specificity} = \frac{b}{b+d}$
• False Negative Rate (FNR) = $1 - \text{Sensitivity} = \frac{c}{a+c}$

---

4. Predictive Values & Influence of Disease Prevalence

A. Positive Predictive Value (PPV)

The probability that a person who tests positive actually has the disease.

$\text{PPV} = \frac{a}{a+b} \times 100$

B. Negative Predictive Value (NPV)

The probability that a person who tests negative is truly free of the disease.

$\text{NPV} = \frac{d}{c+d} \times 100$

C. Impact of Disease Prevalence on Predictive Values

UPSC CMS High-Yield Concept: Sensitivity and Specificity are intrinsic properties of a screening test and DO NOT change with disease prevalence. However, PPV and NPV are heavily dependent on disease prevalence:

• As Prevalence Rises: PPV increases dramatically; NPV decreases slightly.
• As Prevalence Falls: PPV drops sharply (resulting in high proportions of false positives among test-positives); NPV increases toward 100%.

+---------------------------------------------------------------------------------------------------+
|                     EFFECT OF PREVALENCE ON PREDICTIVE VALUE (Fixed Sens = 90%, Spec = 90%)       |
+-----------------------+-----------------------+-----------------------+---------------------------+
| Disease Prevalence    |  True Positives (a)   |  False Positives (b)  | Positive Predictive Value |
+-----------------------+-----------------------+-----------------------+---------------------------+
| High Prevalence (20%) | 180 per 1,000         | 80 per 1,000          | PPV = 180/260 = 69.2%     |
| Low Prevalence (1%)   | 9 per 1,000           | 99 per 1,000          | PPV = 9/108 = 8.3%        |
+-----------------------+-----------------------+-----------------------+---------------------------+

---

5. Multiple Testing Strategies: Parallel vs. Serial

A. Parallel Testing (Simultaneous)

Two or more screening tests are administered concurrently. A positive result on ANY single test classifies the individual as test-positive.

• Effect: Maximizes overall Sensitivity and NPV; reduces overall Specificity.
• Clinical Application: Emergency room evaluation (e.g., combining ECG + Troponin T for acute myocardial infarction).

B. Serial / Sequential Testing

Tests are performed in sequence; the second test is administered ONLY if the first test is positive. An individual is classified as positive only if ALL tests are positive.

• Effect: Maximizes overall Specificity and PPV; reduces overall Sensitivity.
• Clinical Application: Two-tier screening (e.g., initial ELISA screening followed by confirmatory Western Blot or PCR).

---

6. Advanced Screening Metrics & Evaluation Biases

Receiver Operating Characteristic (ROC) Curve

An ROC curve plots Sensitivity (y-axis) against 1 - Specificity (x-axis) across continuous diagnostic cut-off thresholds.

• The Area Under the Curve (AUC) evaluates overall diagnostic accuracy (AUC = 1.0 represents a perfect test; AUC = 0.5 represents a useless test equivalent to chance).

Screening Evaluation Biases

1. Lead Time Bias: The apparent inflation in survival time caused by detecting disease earlier in its natural history (at screening) without actually altering the ultimate age at death.
2. Length Time Bias: The over-representation of slowly progressing, indolent cases (having longer preclinical phases) among screen-detected cases compared to rapidly fatal cases.