Atrial fibrillation after non-cardiac surgery : P-wave characteristics and Holter monitoring in risk assessment

Background: We investigated the role of 12-lead ECG P-wave duration and dispersion and of Holter monitoringas predictors of post-thoracic surgeryatrialfibrillation.Methods:Onehundredandfiveconsecutivepatients(88males—17females;age60 9),undergoingthoracicsurgery at National Cancer Institute between 2001 and 2003, were enrolled and both standard ECG and Holter monitoring were obtained from each patient. P-wave study was made on a magnified ECG paper copy. Holter monitoring was performed 1—3 days before surgery; patients were dividedintothree classesaccording tonumber and complexityofprematuresupraventricularcomplexes (0: 30/h or couplets; 2: run of supraventricular tachycardia or atrial fibrillation). Results: Atrial fibrillation was detected in 12 patients (11%) within 96 h from surgery. In univariable logistic model, P-wave duration was not associated with postoperative atrial fibrillation while P-wave dispersion and Holter monitoring demonstrated a statistically significant association with the occurrence of atrial fibrillation (OR of 30 vs 20 ms = 2.06; CI: 1.17—3.64; p = 0.012, OR of class 1—2 vs class 0 = 8.16; CI: 2.04—35.59; p = 0.003, respectively). In the multivariable model, both P-wave dispersion and Holter were shown to be significantly associated with the end-point. Holter monitoring enhanced the predictive ability of P-wave dispersion (area under the ROC curve increased from 0.64 to 0.80). Conclusions: P-wave dispersion, but not duration, was associatedwithatrialfibrillationafterthoracicsurgery.PreoperativeHoltermonitoringaddsfurtherinformationand couldbeused toenhance the P-wave predictive power. # 2007 European Association for Cardio-Thoracic Surgery. Published by Elsevier B.V. All rights reserved.