Diagnostic Accuracy in Medical Testing

Core Definitions

• The American Heart Association defines sensitivity as the percentage of individuals with disease who will have abnormal test results, which answers the question: "If the patient has the disease, what is the probability the test will be positive?" 1, 2
• The American Heart Association defines specificity as the percentage of individuals without disease who will have normal test results, which answers the question: "If the patient does not have the disease, what is the probability the test will be negative?" 1, 2

Mathematical Framework

• The sensitivity calculation is defined as the number of true positives divided by all patients with disease, according to the American College of Medical Genetics 3, 4
• The specificity calculation is defined as the number of true negatives divided by all patients without disease, according to the American College of Medical Genetics 3, 4
• For FISH probes, a sensitivity of at least 95% and specificity of at least 98% are recommended for clinical testing by the American College of Medical Genetics 3, 4

Clinical Context and Interpretation

• Sensitivity is influenced by disease severity, effort level, and anti-ischemic drugs, as stated by the American Heart Association 2
• Specificity may be affected by drugs, baseline ECG patterns, and left ventricular hypertrophy, according to the American Heart Association 2
• If the population is skewed toward individuals with greater disease severity, the test will have higher sensitivity, as noted by the American Heart Association 2

Critical Distinction from Predictive Values

• Sensitivity and specificity are characteristics of the test itself and remain constant regardless of disease prevalence, while predictive values change with prevalence, as stated in The Lancet Neurology 5

Common Pitfalls

• Selection bias in case and control groups can artificially inflate or deflate sensitivity and specificity measures, according to The Lancet Neurology 5
• Using healthy controls instead of disease mimics can inflate specificity, as warned by The Lancet Neurology 5
• The reference standard must be accurate, and when the reference test is imperfect, estimates of sensitivity and specificity become untrustworthy, as noted in Clinical Microbiology Reviews 6