Methods

The inclusion criterion for the study was patients undergoing isolated CABG from June 1, 2001 through December 31, 2009 at hospitals in Australia participating in the Australasian Society of Cardiac and Thoracic Surgeons (ASCTS) Cardiac Surgery Database. Patients having concomitant valve surgery or other concurrent cardiac surgical procedures were excluded from this study. Moreover, only patients with documented preoperative sinus rhythm without a history of AF were included. All 6 Victorian public hospitals that perform adult cardiac surgery—Royal Melbourne Hospital, Alfred Hospital, Monash Medical Centre, Geelong Hospital, Austin Hospital, and St. Vincent’s Hospital, Melbourne—were involved in the prospective data collection during the entire period. In addition, 14 cardiac surgical units from South Australia, New South Wales, and Queensland entered the database project in the last 30 months of the study period and contributed 41.4% of the total patient numbers.

The ASCTS database contained detailed information on patient demographics, preoperative risk factors, operative details, postoperative hospital course, and morbidity and mortality outcomes. These data were collected prospectively using a standardized dataset and definitions. Data collection and audit methods have been previously described. In the State of Victoria, the collection and reporting of public hospital cardiac surgery data are compulsory and mandated by the state government; hence, it is all-inclusive. Data validation has been a major focus since the establishment of the ASCTS database. Data are subjected to local validation and an external data quality audit program, which is performed on site to evaluate the completeness (defined as <1% missing data for any variable) and accuracy (97.4%) of the data held in the combined database. Audit outcomes are used to assist in further development of appropriate standards. The ethics committee of each participating hospital had previously approved the use of de-identified patient data contained within the database for research and waived the need for individual patient consent.

POAF was defined as evidence of new AF that required treatment by electrocardiography or continuous monitoring during the postoperative period. Although treatment of AF may vary slightly between hospitals, it is general practice in the participating institutions to restore sinus rhythm in most patients within 24 hours after the onset of POAF using electrolyte replacement, antiarrhythmic drugs (AADs), or by electrical cardioversion. Patients in AF who are discharged home are maintained on warfarin (in the absence of any contraindication) and usually referred for cardioversion after 3 to 6 weeks. Patients discharged home on AADs are followed up in clinics in 6 weeks. In the absence of evidence of AF recurrence, their AADs are stopped. The decision to stop warfarin in this instance was left to the discretion of the treating physician.

For this study patients were separated into 2 groups based on the development of POAF (POAF group) or not (no-POAF group). Preoperative characteristics, early outcomes, and long-term survival were compared between the 2 groups. Late mortality was defined as death from any cause that occurred at any time after hospital discharge.

Fourteen early postoperative outcomes were analyzed: (1) 30-day mortality, defined as death within 30 days of operation; (2) permanent stroke, defined as a new central neurologic deficit persisting for >72 hours; (3) transient stroke, defined as a transient neurologic deficit (transient ischemic attack or reversible ischemic neurologic deficit); (4) postoperative acute myocardial infarction, defined as ≥2 of the following: enzyme level elevation, new cardiac wall motion abnormalities, or new Q waves on serial electrocardiograms; (5) new renal failure, defined as ≥2 of the following: serum creatinine increased to >200 μmol/L, ≥2 times increase in creatinine compared to the preoperative value, or new requirement for dialysis or hemofiltration; (6) prolonged ventilation (>24 hours); (7) multisystem failure; defined as concurrent failure of ≥2 of the cardiac, respiratory, or renal systems for ≥48 hours; (8) septicemia, defined as positive blood cultures supported by ≥2 of the following indexes of clinical infection: postoperative fever, increased granulocyte cell counts, increased C-reactive protein, and increased erythrocyte sedimentation rate; (9) gastrointestinal complications, defined as postoperative occurrence of any gastrointestinal complication; (10) deep sternal infection involving muscle and bone as demonstrated by surgical exploration and 1 of the following: positive cultures or treatment with antibiotics; (11) pneumonia diagnosed by 1 of the following: positive cultures of sputum, blood, pleural fluid, empyema fluid, transtracheal fluid, or transthoracic fluid consistent with the diagnosis and clinical findings of pneumonia; (12) red blood cell transfusion postoperatively; (13) return to the operating theater for any cause; and (14) return to the operating theater for bleeding.

To assess the impact of POAF on each outcome, logistic regression analysis was used to adjust for 20 preoperative patient variables, with the outcome as the dependent variable. Long-term survival status was obtained from the Australian National Death Index. The closing date was March 18, 2010. A Kaplan–Meier estimate of survival was obtained. Differences in long-term survival were assessed by log-rank test. The role of POAF in long-term survival was assessed by constructing a Cox proportional hazards model using octogenarian status and other preoperative patient characteristics as variables. Continuous variables are presented as mean ± 1 SD. Mann–Whitney U test was used to compare 2 groups of continuous variables. Fisher’s exact test or chi-square test was used to compare groups of categoric variables. All calculated p values were 2-sided, and a p value <0.05 was considered statistically significant. Statistical analysis was performed using SPSS 17.0 for Windows 0 (SPSS, Inc., Munich, Germany).