ECG Sensitivity and Specificity
The ECG has variable sensitivity and specificity depending on the clinical condition being diagnosed, with high accuracy for arrhythmias (>90%) but more limited sensitivity for acute myocardial infarction (approximately 43-56% for initial ECG, improving to 67-79% with serial recordings), though specificity remains high (>90%) when proper criteria are applied.
For Acute Myocardial Infarction
Initial ECG Performance
The initial 12-lead ECG has limited sensitivity but high specificity for detecting acute MI 1. The ECG by itself is often insufficient to diagnose acute myocardial ischemia or infarction, since ST deviation may be observed in other conditions such as acute pericarditis, left ventricular hypertrophy, left bundle branch block, Brugada syndrome, stress cardiomyopathy, and early repolarization patterns 1.
Key performance metrics:
- Sensitivity: 43-56% for STEMI criteria on initial presentation 2
- Specificity: 97-98% for STEMI criteria 2, 3
- Positive Predictive Value: 96-97% 2
Improving Sensitivity with Serial ECGs
Serial ECG monitoring significantly improves diagnostic accuracy 4:
Repeat ECG at 3-4 hours:
- Sensitivity: 39% for AMI
- Specificity: 88% for AMI 4
Serial 12-lead ECG monitoring:
Enhanced Lead Configurations
Using additional leads improves sensitivity 3:
- 12-lead ECG: 28% sensitivity (postadmission), 33% (admission)
- 16-lead ECG (adding V4R, V5R, V8, V9): 33% sensitivity (postadmission)
- 24-lead ECG: 37% sensitivity (postadmission), 49% (admission)
- All maintain specificity >93% 3
Critical pitfall: The ECG is often equivocal in early hours, and even in proven infarction it may never show classical features of ST-segment elevation and new Q waves 5. Posterior and right ventricular infarctions are particularly prone to being missed without additional leads (V7-V9 for circumflex occlusion, V3R-V4R for RV infarction) 1.
Modified STEMI Criteria
Research suggests that adding ST-depression criteria boosts sensitivity to 79-84% but reduces specificity to 79-90% 2. A proposed STEMI13 approach (using 3 additional contiguous lead pairs) achieves 67% sensitivity while maintaining 97% specificity 2.
For Atrial Fibrillation
The ECG demonstrates excellent diagnostic accuracy for atrial fibrillation, as it is the only practical noninvasive method for recording and analyzing cardiac rhythm abnormalities 6.
Standard 12-Lead ECG
ECG documentation is required to establish the diagnosis of AF 7. A standard 12-lead ECG recording or single-lead ECG tracing of ≥30 seconds showing heart rhythm with no discernible repeating P waves and irregular RR intervals (when atrioventricular conduction is not impaired) is diagnostic of clinical AF 7.
Single-Lead ECG Devices
Lead-I ECG devices show high accuracy for AF detection 8:
- Sensitivity: 93.9% (95% CI 86.2-97.4%)
- Specificity: 96.5% (95% CI 90.4-98.8%)
Patient-Operated ECG Devices
Handheld, patient-operated devices demonstrate excellent performance 9:
- Sensitivity for AF: 99%
- Specificity for AF: 96%
- Sensitivity for normal ECG: 91%
- Sensitivity for T-wave abnormalities: 90%
Important consideration: These devices are particularly valuable for detecting paroxysmal AF, which may be missed on single time-point standard ECGs 9.
Clinical Application Algorithm
For Suspected Acute MI:
- Obtain initial 12-lead ECG within 10 minutes of presentation 1
- If initial ECG non-diagnostic but high clinical suspicion:
- Do not wait for biomarker results to initiate reperfusion therapy if ECG shows STEMI criteria 5
- Always compare with prior ECG tracings when available 1
For Suspected AF:
- Any ECG recording ≥30 seconds showing irregular RR intervals without discernible P waves is diagnostic 7
- For intermittent symptoms: Consider patient-activated single-lead ECG devices (sensitivity 93.9%, specificity 96.5%) 8
- Continuous monitoring may be needed for paroxysmal AF 9
Critical caveat: The sensitivity and specificity of the ECG vary considerably depending on the clinical question asked. While sensitivity and specificity for arrhythmias are consistently high, they are more variable for structural abnormalities like myocardial infarction, where diagnoses are made by inference from electrical activity 10, 6.