Diagnostic Screening and Epidemiology

Introduction to Two by Two Table

• A two by two table is a versatile tool used in epidemiology for:
  • Understanding associations between two discrete variables (exposure and outcome).
  • Assessing the validity and reliability of diagnostic screening tools.

Key Terminology

• Exposure Variable: Refers to any risk factor that may cause or contribute to an outcome (e.g., smoking, contaminated food).
• Outcome Variable: The result of the exposure, typically a health-related event or condition (e.g., food poisoning, HIV, lung cancer).

Variable Types

• Categorical Variables:
  • Nominal (e.g., has disease or not).
  • Ordinal (e.g., severity of illness).
• Continuous Variables: Data that can fall along a number line (e.g., blood pressure, cholesterol levels).

Structure of Two by Two Table

1. The table consists of four cells:

   	Disease Present	Disease Absent
   Test Positive	True Positive	False Positive
   Test Negative	False Negative	True Negative

2. Each cell contains counts of individuals fitting those outcomes.

Example of Two by Two Table Counts

• Total individuals: 100
  • True positive (exposed and have outcome): 39
  • False negative (exposed and don't have outcome): 18
  • False positive (not exposed but have outcome): 19
  • True negative (not exposed and don't have outcome): 24

Sensitivity and Specificity

• Sensitivity: The test's ability to identify true positives.
  ext{Sensitivity} = \frac{True Positives}{True Positives + False Negatives}

• Specificity: The test's ability to identify true negatives.
  ext{Specificity} = \frac{True Negatives}{True Negatives + False Positives}

Reliability vs. Validity

• Validity: The ability of a test to accurately measure what it is supposed to measure (getting true results).
• Reliability: The ability of a test to produce consistent results over repeated applications.

Bull's Eye Metaphor

• Illustrates the concepts of reliability and validity:
  • Ideal Test: Dots all clustered center (valid & reliable).
  • Valid but Not Reliable: Dots are scattered but center is captured.
  • Reliable but Not Valid: Dots are clustered but not centered around the bull's eye.
  • Neither Valid nor Reliable: Dots are scattered and not clustered.

Diagnostic Cut Points

• Cut Points: Thresholds used to determine if a diagnosis is made based on continuous measures.
• Bimodal vs. Unimodal Distributions:
  • Bimodal: Distinct peaks corresponding to sick and well populations (easier to identify cut points).
  • Unimodal: Single peak complicates cut point determination.

Predictive Value of a Test

• Positive Predictive Value (PPV): The likelihood that a positive test indicates true disease.
  ext{PPV} = \frac{True Positives}{True Positives + False Positives}

• Negative Predictive Value (NPV): The likelihood that a negative test indicates no disease.
  ext{NPV} = \frac{True Negatives}{True Negatives + False Negatives}

Impact of Test Outcomes

• False Positives: Emotional and financial distress due to false indication of disease.
• False Negatives: Risk of untreated disease can have serious health consequences.

Reliability Assessment

• Factors influencing reliability:
  • Intrasubject Variation: Variability in an individual over time.
  • Intraobserver Variation: Variability by the same observer using the test.
  • Interobserver Variation: Variability between different observers assessing the same result.

Calculating Percent Agreement

• Method to determine the level of agreement between two observers:
  • Sum true agreements and divide by total observations.
    ext{Percent Agreement} = \frac{(Agreements)}{(Total Observations)}
• Example results in 80% agreement indicating a moderate level of consensus in diagnostics.

Recap of Key Concepts

• Understanding the use of two by two tables is crucial for evaluating diagnostic tests.
• Recognizing the balance between sensitivity and specificity is essential for test utility.
• Validity and reliability are foundation stones in the assessment of any diagnostic tool.