Multi-test algorithms

Multi-Test Algorithms - More Tests, More Answers?

• Serial Testing (AND logic): Positive only if all tests are positive.

  • Maximizes Specificity (Sp) and PPV.
  • Used for confirmation, ruling in disease.

• Parallel Testing (OR logic): Positive if any test is positive.

  • Maximizes Sensitivity (Sn) and NPV.
  • Used for screening, ruling out disease.

⭐ In serial testing, a highly sensitive (Sn) test is often used first, followed by a highly specific (Sp) test to confirm positives, maximizing efficiency and accuracy (e.g., ELISA → Western Blot for HIV).

Serial Testing - One After Another

• Involves applying tests sequentially. A second, more specific test is often used to confirm the results of an initial, sensitive screening test.

• To be considered positive, the individual must test positive on all tests.

• Goal: ↑ Specificity and ↑ Positive Predictive Value (PPV).

• Effect on Diagnostic Metrics:

  • Net Sensitivity: ↓ (lower than the least sensitive test). $Sn_{net} = Sn_1 \times Sn_2$
  • Net Specificity: ↑ (higher than the most specific test). $Sp_{net} = 1 - [(1-Sp_1) \times (1-Sp_2)]$
  • PPV: ↑
  • NPV: ↓

⭐ Exam Favorite: This approach is the classic "screening followed by confirmation" strategy. A common example is using a sensitive ELISA for HIV screening, followed by a highly specific Western blot for confirmation.

📌 Mnemonic: Serial testing boosts Specificity.

Parallel Testing - All At Once

• Multiple, independent tests are performed simultaneously. A positive result on any test is considered a positive outcome.
• Primarily used to ↑ Net Sensitivity. Excellent for ruling out diseases (SNOUT).
• The trade-off is a ↓ Net Specificity, leading to more false positives.
• Net Sensitivity: $Sn_{net} = 1 - [(1 - Sn_A) \times (1 - Sn_B)]$
• Net Specificity: $Sp_{net} = Sp_A \times Sp_B$
• 📌 Mnemonic: Think of an "OR" gate. The result is positive if Test A OR Test B is positive. This increases the "OR-portunity" to detect a disease.

⭐ Exam Favorite: Parallel testing is the strategy of choice in clinical scenarios where a false negative is highly undesirable and must be avoided, such as in screening for pulmonary embolism or acute MI.

• Serial testing (e.g., ELISA then Western blot for HIV) significantly ↑ specificity and ↑ PPV.
• The trade-off for serial testing is a ↓ net sensitivity; you miss more true cases.
• Use serial testing to confirm a diagnosis, as it effectively rules in disease by reducing false positives.
• Parallel testing (multiple tests at once) dramatically ↑ sensitivity.
• This is ideal for screening or emergencies, minimizing false negatives.
• The cost of parallel testing is a ↓ net specificity.