What are the specificity, sensitivity, and accuracy of an electrocardiogram (ECG) in diagnosing cardiovascular disease?

ECG Diagnostic Performance for Cardiovascular Disease

The ECG has variable sensitivity and specificity that depends heavily on what you're trying to diagnose: it excels at detecting arrhythmias (consistently high sensitivity and specificity) but performs poorly for structural heart disease, with sensitivity around 68% and specificity around 77% for detecting coronary artery disease. 1

Performance Varies Dramatically by Clinical Question

The critical concept is that the ECG is not a single test—it's composed of multiple waveforms, each with distinct diagnostic characteristics 1:

For Arrhythmias and Conduction Disorders

• Sensitivity and specificity are consistently high 1
• The ECG is the only practical noninvasive method for recording and analyzing cardiac rhythm abnormalities 1
• This is where the ECG performs best because it directly measures electrical activity rather than inferring structural changes 1

For Coronary Artery Disease Detection

• Mean sensitivity: 68% (range 23-100%) 1
• Mean specificity: 77% (range 17-100%) 1
• The enormous ranges reflect differences in patient populations, disease severity, effort level during testing, and anti-ischemic medications 1
• Using the standard 0.1 mV ST-segment depression cutpoint provides this baseline performance 1

For Acute Myocardial Infarction (STEMI)

• Prehospital/ED ECG sensitivity: 68-76% for acute cardiac ischemia 1
• Specificity: 88-97% for acute MI 1
• Admission ECG sensitivity improves to 76% with specificity of 88% 1
• Computer-interpreted ECG for acute MI shows sensitivity 53-71% and specificity 98-99% when optimized 2, 3
• Adding posterior and right-sided leads (16-lead or 24-lead ECG) increases sensitivity to 37-49% but slightly reduces specificity to 93-95% 4

For Structural Heart Disease (Hypertrophy, Infarction)

• ST-segment and T-wave changes are the most sensitive abnormalities but least specific 1
• Diagnoses are made by inference from electrical activity, making them inherently subject to error 1
• The same ECG pattern can result from different structural and pathophysiologic states, explaining the frequent low specificity 1

Critical Pitfalls That Reduce Accuracy

Computer Interpretation Errors

• Computer-interpreted ECGs have significant error rates: 9.9% of all ECGs and 15.9% of abnormal ECGs show major disagreements with cardiologist interpretation 5
• 86.4% of computer errors involve arrhythmias, conduction disorders, and pacemakers 5
• Computer diagnosis of acute MI is inaccurate 40.7% of the time; high-degree AV blocks are missed 75% of the time 5
• All computer-interpreted ECGs must be verified by an experienced physician—computer interpretation alone is not recognized as properly interpreted 1, 6

Technical Factors That Degrade Performance

• Electrode misplacement significantly alters interpretation and creates false diagnoses 6
• Inadequate high-frequency response (minimum 150 Hz for adults, 250 Hz for children required) causes systematic underestimation of signal amplitude and smoothing of Q waves 6
• Noise in ECG signals compromises the tradeoff between sensitivity and specificity 6
• Baseline wander and electrical interference must be minimized 6

Clinical Context Errors

• The same ECG abnormality has vastly different implications depending on clinical presentation 1, 6
• For example, moderate T-wave inversion predicts 21% annual mortality with known heart disease versus only 3% without heart disease 1
• Interpreting ECG findings in isolation without clinical correlation leads to serious iatrogenic disease 1
• An abnormal T-wave is often incorrectly equated with ischemia when its specificity for any single cause is actually low 1

Optimizing ECG Diagnostic Accuracy

Serial ECGs and Biomarkers

• Diagnostic accuracy improves substantially by repeating the ECG on ED arrival and using serial cardiac markers 1
• Combined use of serial biomarkers and ECGs in low-risk patients safely identifies those who can be discharged 1

Enhanced Lead Systems

• Using inverted leads (19-lead or 24-lead approach) increases sensitivity from 33% to 45-49% for acute MI 4
• This requires no additional electrodes—just examining both positive and negative aspects of the conventional 12-lead 4
• Adding posterior leads (V7-V9) and right-sided leads (V4R, V5R) increases sensitivity to 37% with specificity 95% 4

Proper Interpretation Standards

• Interpretation must follow standardized guidelines for optimal electrocardiography 1
• Physicians require interpretation of a large number of ECGs with experienced faculty review to maintain competence 1
• Continuing medical education is essential for physicians who read ECGs infrequently 6

When ECG Performance Is Inadequate

If the ECG is nondiagnostic with negative biomarkers in suspected ACS, consider advanced imaging 1:

• CT angiography: 95% sensitivity, 90% specificity 1
• Myocardial perfusion imaging: 89% sensitivity, 77% specificity 1
• Cardiac MRI: 85% sensitivity, 84% specificity 1
• Rest echocardiography: 93% sensitivity, 66% specificity 1