Preoperative electrocardiographic risk assessment of atrial fibrillation after coronary artery bypass grafting.

INTRODUCTION: This study evaluated the role of surface ECG in assessment of risk of new-onset atrial fibrillation (AF) after coronary artery bypass grafting surgery (CABG). METHODS AND
RESULTS: One hundred fifty-one patients (126 men and 25 women; age 65 +/- 10 years) without a history of AF undergoing primary elective and isolated CABG were studied. Standard 12-lead ECGs and P wave signal-averaged ECG (PSAE) were recorded 24 hours before CABG using a MAC VU ECG recorder. In addition to routine ECG measurements, two P wave (P wave complexity ratio [pCR]; P wave morphology dispersion [PMD]) and six T wave morphology descriptors (total cosine R to T [TCRT]; T wave morphology dispersion of ascending and descending part of the T wave [aTMD and dTMD], and others), and three PSAE indices (filtered P wave duration [PD]; root mean square voltage of terminal 20 msec of averaged P wave [RMS20]; and integral of P wave [Pi]) were investigated. During a mean hospital stay of 7.3 +/- 6.2 days after CABG, 40 (26%) patients developed AF (AF group) and 111 remained AF-free (no AF group). AF patients were older (69 +/- 9 years vs 64 +/- 10 years, P = 0.005). PD (135 +/- 9 msec vs 133 +/- 12 msec, P = NS) and RMS20 (4.5 +/- 1.7 microV vs 4.0 +/- 1.6 microV, P = NS) in AF were similar to that in no AF, whereas Pi was significantly increased in AF (757 +/- 230 microVmsec vs 659 +/- 206 microVmsec, P = 0.007). Both pCR (32 +/- 11 vs 27 +/- 10) and PMD (31.5 +/- 14.0 vs 26.4 +/- 12.3) were significantly greater in AF (P = 0.012 and 0.048, respectively). TCRT (0.028 +/- 0.596 vs 0.310 +/- 0.542, P = 0.009) and dTMD (0.63 +/- 0.03 vs 0.64 +/- 0.02, P = 0.004) were significantly reduced in AF compared with no AF. Measurements of aTMD and three other T wave descriptors were similar in AF and no AF. Significant variables by univariate analysis, including advanced age (P = 0.014), impaired left ventricular function (P = 0.02), greater Pi (P = 0.012), and lower TCRT (P = 0.007) or dTMD, were entered into multiple logistic regression models. Increased Pi (P = 0.038), reduced TCRT (P = 0.040), and lower dTMD (P = 0.014) predicted AF after CABG independently. In patients <70 years, a linear combination of increased pCR and lower TCRT separated AF and no AF with a sensitivity of 74% and specificity of 62% (P = 0.005).
CONCLUSION: ECG assessment identifies patients vulnerable to AF after CABG. Combination of ECG parameters assessed preoperatively may play an important role in predicting new-onset AF after CABG.