Accuracy of electrocardiographic criteria for atrial enlargement: validation with cardiovascular magnetic resonance

Connie W Tsao¹ , Mark E Josephson¹ , Thomas H Hauser¹ , T David O'Halloran¹ , Anupam Agarwal¹^,2 , Warren J Manning¹^,3 and Susan B Yeon¹

¹Harvard-Thorndike Laboratory and the Department of Medicine, Cardiovascular Division, Beth Israel Deaconess Medical Center and Harvard Medical School, 330 Brookline Avenue, Boston, Massachusetts, USA

²Cardiovascular and Metabolic Division, GlaxoSmithKline Pharmaceuticals, 1250 Collegeville Road, Collegeville, Pennsylvania, USA

³Department of Radiology, Beth Israel Deaconess Medical Center and Harvard Medical School, 330 Brookline Avenue, Boston, Massachusetts, USA

author email corresponding author email

Journal of Cardiovascular Magnetic Resonance 2008, 10:7doi:10.1186/1532-429X-10-7


Published:	25 January 2008

Abstract

Background

Anatomic atrial enlargement is associated with significant morbidity and mortality. However, atrial enlargement may not correlate with clinical measures such as electrocardiographic (ECG) criteria. Past studies correlating ECG criteria with anatomic measures mainly used inferior M-mode or two-dimensional echocardiographic data. We sought to determine the accuracy of the ECG to predict anatomic atrial enlargement as determined by volumetric cardiovascular magnetic resonance (CMR).

Methods

ECG criteria for left (LAE) and right atrial enlargement (RAE) were compared to CMR atrial volume index measurements for 275 consecutive subjects referred for CMR (67% males, 51 ± 14 years). ECG criteria for LAE and RAE were assessed by an expert observer blinded to CMR data. Atrial volume index was computed using the biplane area-length method.

Results

The prevalence of CMR LAE and RAE was 28% and 11%, respectively, and by any ECG criteria was 82% and 5%, respectively. Though nonspecific, the presence of at least one ECG criteria for LAE was 90% sensitive for CMR LAE. The individual criteria P mitrale, P wave axis < 30°, and negative P terminal force in V1 (NPTF-V1) > 0.04s·mm were 88–99% specific although not sensitive for CMR LAE. ECG was insensitive but 96–100% specific for CMR RAE.

Conclusion

The presence of at least one ECG criteria for LAE is sensitive but not specific for anatomic LAE. Individual criteria for LAE, including P mitrale, P wave axis < 30°, or NPTF-V1 > 0.04s·mm are highly specific, though not sensitive. ECG is highly specific but insensitive for RAE. Individual ECG P wave changes do not reliably both detect and predict anatomic atrial enlargement.