Diagnostic Test Evaluation

Page 1: Presentation Introduction

  • Title: Diagnostic Test Evaluation

  • Presented by: Talha Rafiq

  • Course: HTH SCI 2G03

Page 2: Diagnostic Test Validity

  • Definition: The ability of a test to accurately identify diseased and non-disease individuals.

Page 3: Evaluating Test Validity

  • Proposed Questions for Evaluation:

    • How many people are correctly diagnosed with the disease?

    • How many people are correctly diagnosed without the disease?

    • How many people are misdiagnosed with the disease but do not have it?

    • How many people are misdiagnosed as no disease but have it?

Page 4: Aspects of Test Validity

  • Validity evaluated through:

    • Sensitivity: Proportion of people who have the disorder and test positive.

    • Specificity: Proportion of people who do not have the disorder and test negative.

    • False Positive Rate: Proportion of people who do not have the disorder but test positive.

    • False Negative Rate: Proportion of people who have the disorder but test negative.

Page 5: Fill in the Blanks Activity

  • Fill in the blanks regarding test results based on the presence or absence of disease.

Page 6: True Positive Definition

  • True Positive: Those who test “positive” and ARE cases.

Page 7: True Negative Definition

  • True Negative: Those who test “negative” and are NOT cases.

Page 8: False Positive Definition

  • False Positive: Those who test “positive” but ARE NOT cases.

Page 9: False Negative Definition

  • False Negative: Those who test “negative” but ARE cases.

Page 10: Understanding Sensitivity and Specificity

  • Sensitivity: Ability of the test to correctly identify who has the disease (denominator: all people with disease).

  • Specificity: Ability of the test to correctly identify who does NOT have the disease (denominator: all people without disease).

Page 11: Formula for Sensitivity

  • Formula: Sensitivity = True Positives / (True Positives + False Negatives)

Page 12: Formula for Specificity

  • Formula: Specificity = True Negatives / (False Positives + True Negatives)

Page 13: Test Data

  • Data Overview:

    • Disease Present (Yes/No): 10 / 40

    • Disease Absent (Yes/No): 5 / 45

Page 14: Calculation of Sensitivity and Specificity

  • Calculated Sensitivity: 67% (10 / (10 + 5))

  • Calculated Specificity: 53% (45 / (40 + 45))

Page 15: Interpretation of Sensitivity and Specificity

  • Question raised: Is 53% specificity and 67% sensitivity good or bad?

Page 16: Further Exploration

  • Further discussion on interpreting test values.

    • Source: U. Penn Law Review, 146 (173).

    • Link: https://scholarship.law.upenn.edu

Page 17: Decision Tolerance

  • Decision points regarding tolerance for:

    • False positives:

      • Tolerate or not tolerate?

    • False negatives:

      • Tolerate or not tolerate?

Page 18: Costs of Misdiagnosis

  • Costs of False Positives (low specificity):

    • Emotional distress of being told one is sick.

    • Financial costs due to re-testing or needless treatment.

  • Costs of False Negatives (low sensitivity):

    • Untreated sick individuals potentially leading to severe consequences.

Page 19: ROC Curve

  • Definition: Receiver Operating Characteristic (ROC) Curve, useful for comparing screening or diagnostic tests.

Page 20: Comparing Test Performance

  • Example of ROC with varying True Positive (TP) and False Positive (FP) rates, indicating better performance tests.

Page 21: ROC Curve Analysis

  • ROC Curve insights on test performance—less false positives and more true positives are desired.

Page 22: Visual Representation of ROC Curves

  • Graphical analysis indicating accuracy levels from worthless to excellent based on false positive rate.

Page 23: Predictive Values

  • Positive Predictive Value: Probability the disease is present when the test is positive (denominator: positive tests).

  • Negative Predictive Value: Probability the disease is not present when the test is negative (denominator: negative tests).

Page 24: Calculation of Positive Predictive Value (PPV)

  • Formula: PPV = True Positives / (True Positives + False Positives)

Page 25: Calculation of Negative Predictive Value (NPV)

  • Formula: NPV = True Negatives / (False Negatives + True Negatives)

Page 26: Example Calculations for PPV and NPV

  • PPV Calculation: 10 / (10 + 40) = 20%

  • NPV Calculation: 45 / (5 + 45) = 90%

Page 27: Interpretation of PPV and NPV

  • PPV Insight: "Understood as the actual chance of having the disease after a positive result—20% chance."

  • NPV Insight: Likely not having the disease after a negative result—90% chance.

Page 28: Down’s Syndrome Study Overview

  • Summary of a study reporting instances of Down's Syndrome among fetuses using sensitivity, specificity, PPV, and NPV methods.

Page 29: Results of Down’s Syndrome Study

  • Probability insights on test results for affected versus unaffected fetuses based on previous findings on sensitivity, specificity, PPV, and NPV.

Page 30: COVID-19 Example

  • Summary of a COVID-19 testing scenario:

    • Pre-test probability: 5.0%

    • Test sensitivity: 91.4%

    • Test specificity: 98.7%

    • Implications of results: The balance of true and false test results and resulting behavioral decisions.

Page 31: Summary of Key Concepts

  • Sensitivity: Ability to correctly identify who has the disease.

  • Specificity: Ability to correctly identify who does NOT have the disease.

  • ROC Curves: Useful for comparing the validity of diagnostic tests.