Clinical Chemistry I - Lab Test Evaluation

Clinical Chemistry I - 0202304

• Course instructor: Mohammad Qabajah
• Email: [email protected]

Week 05&06 - Clinical Evaluation of Laboratory Tests

• Objectives:
  • State formulas for calculating clinical sensitivity, specificity, and predictive value of a lab test.
  • Explain how prevalence affects a lab test's predictive value.
  • Construct a receiver operating characteristic (ROC) plot using diagnostic test data.
  • Interpret difference plots used in method comparison.

Agreement of Test Results: Accurate Diagnosis

• Key concepts:
  • Sensitivity and Specificity
  • Predictive Values
  • Prevalence
  • Likelihood Ratio (LR)
  • Odds Ratio (OR)

Review: Gaussian Probability Distribution

• Distribution graph shown, with key percentiles marked:
  • 99.7%
  • 95%
  • 68%
• Values are shown on the horizontal axis (e.g., -3σ, -2σ, -1σ, μ, +1σ, +2σ, +3σ).

Sensitivity and Specificity

• True Positive Rate:
  • Sensitivity—measures the proportion of true positives correctly identified.
  • Predictive models or algorithms are used to correctly identify true positive cases.
• True Negative Rate:
  • Specificity — measures the proportion of true negatives correctly identified.
  • Predictive models or algorithms correctly identify true negative cases.

Negative and Positive

• Graph depicting true negatives (disease absent) and true positives (disease present).
• Shows false negatives and false positives as areas on the graph.

Source of the False

• Graph similar to the previous one, but with a focus on false negatives and false positives.

Reference Method (Gold/Reference Standard)

• "Best current practice" for confirming a disorder.
• Screening:
  • High sensitivity (high false positives, low false negatives)
  • Finds all possible positives for further testing
  • Ensures no one is missed
• Diagnostic:
  • High specificity (low false positives, high false negatives)
  • Excludes individuals without the disease;
  • Ensures no one with the disease is missed.
  • Diagnostic testing should be done on all positive screened results.

Sensitivity and Specificity (Table)

• Table outlines True Positives (TP), False Negatives (FN), False Positives (FP), and True Negatives (TN).
• Shows the relationship to how the results are categorized and used for analysis.

Sensitivity and Specificity (Formulas)

• Sensitivity: TP / (TP + FN)

• Specificity: TN / (FP + TN)

• % Sensitivity: (TP / (TP + FN)) * 100

• % Specificity: (TN / (FP + TN)) * 100

Predictive Values

• Positive Predictive Value (PV+): TP / (TP + FP)
• Negative Predictive Value (PV−): TN / (TN + FN)

Prevalence

• Frequency of a condition in a given population.
• Example: 1 in 10 (0.1) in a population of 1,000,000

Example: Sensitivity, Specificity, PV+, and PV-

• Example data provided, showing how to calculate Sensitivity, Specificity, positive predictive value , and negative predictive value.
• Prevalence of the disease affects the positive and negative predictive values.

Likelihood Ratio (LR)

• Likelihood that a symptom is related to the target disease, compared to the likelihood of the symptom in the absence of the disease.
• Positive Likelihood Ratio (+LR): Sensitivity/(1-Specificity)
• Negative Likelihood Ratio (-LR): (1-Sensitivity)/Specificity

Likelihood Ratio (LR) - Table

• Table showing Likelihood Ratio values and their relationship to change in probability of disease (e.g., LR of 0.1 decreases probability by 45%).
• Good LR tests are +LR > 10, or - LR < 0.1
• Relationship between sensitivity, specificity, and likelihood ratios: a nomogram is provided using this info to show the zones of good tests (strong positive/negative findings).

Odds Ratio (OR)

• Probability of a disease's presence divided by probability of absence.
• The calculation of Pretest odds, Posttest odds, and Posttest probability
• Example of a blood pressure test and how to analyze the likelihood results.

LR Nomogram

• The graph shows different likelihood ratios and how they correlate to sensitivity vs specificity
• How these measures can be used to create a quick analysis on the LR test value.