In the general population, African Americans experience atrial fibrillation (AF) less frequently than European Americans. This difference could also exist in the incidence of this arrhythmia after cardiac surgery, but this possibility has been insufficiently examined. To test the association of such an ethnic difference, we compared the incidence of postoperative AF in a consecutive series of 2,312 African Americans and 6,054 European Americans who underwent isolated coronary artery bypass grafting from July 2000 to June 2007. Raw differences between the cohorts in the incidence of new AF were adjusted to take into account the baseline differences. Postoperatively, new-onset AF developed in 504 (22%) of 2,312 African-American patients and in 1,838 (30%) of 6,054 European-American patients (p <0.01). After adjustment with logistic regression analysis for numerous baseline differences, African Americans remained less likely to develop AF (odds ratio 0.63, 95% confidence interval 0.55 to 0.72; p <0.001). Risk was also adjusted using propensity matching. In that analysis, 457 (22%) of 2,059 African-American patients had postoperative AF, as did 597 (29%) of 2,059 matched European-American patients (p <0.01). In conclusion, AF was significantly less common among African-American patients than among European-American patients after coronary artery bypass grafting.
In population-based studies, African-American patients have been found to have a lower prevalence of atrial fibrillation (AF) than European-American patients. Similarly, new AF in African-American patients after coronary artery bypass grafting (CABG) might be less frequent than in European-American patients. In an investigation of the predictors of incident AF after cardiac surgery, Magee et al found African-American ethnicity to be a “protective factor.” This association between race and the incidence of AF after cardiac surgery had not been examined in a direct comparison before the recent report by Lahiri et al. We undertook the present study to confirm in a larger study population the association demonstrated by Lahiri et al between ethnicity and postoperative AF.
The clinical data from all patients who undergo cardiac surgery in our institution are prospectively entered into a computerized registry. The 10,556 who underwent isolated CABG from January 2000 to July 2007 were reviewed for inclusion in this comparison. The assigned ethnic group was self-declared by the patient at hospital registration. We excluded 1,734 patients who described themselves as belonging to an ethnic group other than African American or European American and 456 patients with a history of chronic or paroxysmal AF or atrial flutter. Thus, the final study population consisted of 2,312 African-American and 6,054 European-American patients without previous AF or atrial flutter.
The baseline demographics, procedural details, and perioperative outcomes are collected and entered into the database during the patients’ hospitalization for the surgery by a data coordinating center as a part of routine clinical practice. The variables are defined according to the Society of Thoracic Surgeons National Cardiac Surgery Database Guidelines and Definitions (available at: www.STS.org/STS national database).
The MedStar Health Research institutional review board approved the present study.
The rhythm of all patients was monitored by telemetry throughout their postoperative course. New-onset AF was diagnosed when continuously present for ≥20 minutes or cumulatively for ≥1 hour within a 24-hour period. Also included was AF that required postoperative drug therapy or cardioversion.
In the absence of contraindications, the use of low-dose β-blocker agents has been routine clinical practice in this institution since 2003. Before 2003, they were used at the discretion of the operating surgeon. No other prophylactic strategies were systematically used.
The categorical variables are reported as the frequency and percentage and differences were tested using 2-sided chi-square analysis. Continuous variables are reported as the mean and standard deviation for normally distributed variables and as the median with interquartile range for nonparametric data. Differences were tested for significance using 2-sided Student’s t tests for normally distributed data and the Wilcoxon 2-sided test for nonparametric data. The Cochran-Armitage trend test was used to compare ordinal data. To reduce the likelihood of type 1 errors, a p value of <0.01, rather than <0.05, was required for significance.
To adjust for the effect of racial differences on incident AF for differences in baseline and procedural characteristics, a multivariate model using stepwise logistic regression analysis was developed. The tested covariates were selected on the basis of their univariate association with the outcome (p <0.15). A Hosmer goodness-of-fit statistic was then computed to examine the calibration of the regression model.
To examine further the effect of intergroup variability on the association between ethnicity and new-onset AF, we used a propensity score analysis. Each patient received a propensity score according to the risk of postoperative AF conferred by each of 33 variables listed in Tables 1 and 2 . The “risk” (propensity) of postoperative AF for each variable (as determined by logistical regression analysis) was then used to compute the propensity for each patient. We were able to pair 2,059 (89%) of 2,312 African Americans with an equal number of European Americans with similar propensity scores. Multivariate logistic regression analysis was performed to confirm that the matched pairs were indeed comparable with regard to potentially confounding variables. Comparability having been established, the incidences of postoperative AF for African-American and European-American patients were compared.
|Characteristic||African Americans (n = 2,312)||European Americans (n = 6,054)||p Value|
|Age (years)||63 ± 10||65 ± 10||<0.01|
|Women||1,041 (45%)||1,428 (24%)||<0.01|
|Body mass index (kg/m 2 )||30 ± 6||29 ± 5||<0.01|
|Diabetes mellitus||1,199 (52%)||1,936 (32%)||<0.01|
|Hypertension||1,976 (85%)||4,341 (72%)||<0.01|
|Renal insufficiency||201 (9%)||140 (2%)||<0.01|
|Hypercholesterolemia||1,601 (69%)||4,489 (74%)||<0.01|
|Current smoker||496 (21%)||1,016 (17%)||<0.01|
|Chronic lung disease||97 (4%)||386 (6%)||<0.01|
|Previous myocardial infarction||1,043 (45%)||2,506 (41%)||<0.01|
|Previous peripheral vascular disease||411 (18%)||848 (14%)||<0.01|
|Previous cerebral vascular accident||229 (10%)||345 (6%)||<0.01|
|Previous carotid artery disease||19 (1%)||130 (2%)||<0.01|
|Previous aortic disease||22 (1%)||112 (2%)||<0.01|
|Myocardial infarction <24 hours||69 (3%)||171 (3%)||0.70|
|Left ventricular ejection fraction <0.35||483 (21%)||1,115 (18%)||0.01|
|Left main stenosis||526 (23%)||1,274 (21%)||0.09|
|Chronic heart failure||355 (15%)||476 (8%)||<0.01|
|Preoperative shock||49 (2%)||138 (2%)||0.66|
|Ventricular arrhythmia||65 (3%)||187 (3%)||0.51|
|Preoperative intra-aortic balloon pump||89 (4)||282 (5)||0.11|
|Preoperative angina pectoris||1,364 (59)||3,526 (58)||0.53|
|β-Blocking agents||1,191 (52%)||2,833 (47%)||<0.01|
|Calcium channel blockers||474 (21%)||99 (15%)||<0.01|
|Angiotensin-converting enzyme inhibitors/angiotensin receptor blocking agents||992 (43%)||2,066 (34%)||<0.01|
|Lipid-lowering agents||1,081 (47%)||2,867 (47%)||0.62|
|Variable||African Americans (n = 2,312)||European Americans (n = 6,054)||p Value|
|Elective coronary artery bypass grafting||1,479 (64%)||3,857 (64%)||0.82|
|Reoperation||59 (3%)||245 (4%)||<0.01|
|Off pump||1,011 (44%)||3,482 (58%)||<0.01|
|Saphenous vein graft||2,093 (91%)||5,167 (85%)||<0.01|
|Left internal mammary||2,170 (94%)||5,677 (94%)||0.88|
|Both internal mammary arteries||208 (9%)||967 (16%)||<0.01|