Following cardiac surgery, 10% to 50% of patients develop atrial fibrillation,1,2 which has been shown to result in increased hospital costs.1,3 The causes of atrial fibrillation after surgery are not clear, although the
condition is common. Atrial hypertrophy, atrial dilatation, and patchy fibrosis, with damage to the sinoatrial node, have been shown to occur as a result of aging.4 The effect of age on the development of atrial fibrillation may be better understood by assessing the function and measurements of the atria.
Left atrial function is usually described as consisting of three elements—a reservoir, a conduit function, and a booster function. During ventricular systole, blood accumulates and is stored in the reservoir. During the conduit function, blood travels from the pulmonary veins to the left ventricle throughout passive ventricular filling in early diastole. During the booster function, left ventricular filling at end diastole is supplemented by contraction of the atrium.5,6 Irregularity in any of these functions results in dysfunction of the left atrium.
We hypothesized that patients who develop atrial fibrillation after surgery had preexisting structural remodeling, and the trauma of surgery caused atrial function to diminish, resulting in atrial fibrillation. Through the use of echocardiography, these changes can be measured noninvasively.
Patients and methods
Patients who underwent coronary artery bypass graft (CABG) surgery elec-
tively were included in the study. These patients did not undergo any other procedures at the time of their surgery. Patients with a preoperative pacemaker, cur-
rent atrial fibrillation, recent atrial fibrillation (within a week of the surgery),
or inadequate or unavailable echocardiographic studies were not included in
The anesthetic management and surgical procedures were conducted according to our customary procedures and were not controlled. Measurements were
taken throughout the surgery using transesophageal echocardiography, as well as after the surgery. Doppler echocardiography was used to assess mitral inflow measurements. The measurements included peak velocity during early filling (E), late filling from atrial contraction (A), and deceleration time. In addition, pulmonary vein diameter and pulmonary venous flow velocities were measured. Other measurements included left atrial area, left atrial length, left atrial appendage area, left atrial appendage peak velocity, left ventricular ejection fraction, and hepatic vein size and velocity.
Left atrial reservoir function was estimated by left atrial ejection fraction (LAEF), which was determined by the following formula: LAEF = (left atrial maximal area — left atrial minimal area) left atrial maximal area ¥ 100%. Dividing the E wave velocity time in-tegral (VTI) by the A wave VTI provided an estimate for left atrial conduit function. Left atrial systolic function was estimated by the atrial filling
fraction, obtained by dividing the
A wave VTI by the total VTI of mitral inflow.7 The peak velocity of the A wave was used to estimate the left atrial booster pump function.8 Hemodynamic and echocardiographic data were measured simultaneously to provide information on loading conditions. Other potential predictors of postoperative atrial fibrillation, including patient demographic and surgical data, were recorded. We used continuous electrocardiographic te-lemetry to monitor for the development of atrial fibrillation in all patients. Two blinded investigators validated all episodes of atrial fibrillation from echocardiographic printouts.
The association of potential clinical echocardiographic predictors with the occurrence of postoperative atrial fibrillation was evaluated using multivariate logistic regression. Group-by-time interaction and repeated measures analysis of variance were used to analyze hemodynamic information. The Mann-Whitney test was used to compare echocardiographic information before and after surgery.
None of the patients developed atrial fibrillation intraoperatively. Postoperative atrial fibrillation occurred in 84 (28%) of the 300 patients. As shown in Table 1, those with postoperative atrial fibrillation were older than those who did not develop atrial fibrillation. There also was a correlation between postoperative atrial fibrillation and use of antiarrhythmic agents postoperatively, postoperative mediastinal exploration, race, body surface area, and a history of atrial fibrillation. The risk of developing postoperative atrial fibrillation was higher for whites and Asians compared with African Americans and Hispanics. Compared with patients who did not develop atrial fibrillation after surgery, hospital stay was significantly longer for patients who developed atrial fibrillation after surgery (median days, 4.5 versus 6.0 respectively; P < .001).
As shown in Table 2, compared with patients who did not develop postoperative atrial fibrillation, the LAEF was lower and the left atrial appendage area and left atrial area were larger in patients who developed atrial fibrillation postoperatively.
The only surgical variable that correlated with developing postoperative atrial fibrillation was return to the operating room for mediastinal exploration, and this was only slightly so. In the final model, this variable was not shown to portend postoperative atrial fibrillation. Aortic cross-clamp time, cardiopulmonary bypass time, number of grafts, inotropic drugs, magnesium sulfate and beta blocker use post-CABG, return to cardiopulmonary bypass, use of intra-aortic balloon pump, intraoperative defibrillation, and postoperative myocardial infarction were not associated with postoperative atrial fibrillation.
There was no difference between patients with and without postoperative atrial fibrillation in the subjective assessment of myocardial protection. For patients who developed postoperative atrial fibrillation, mean systolic blood pressure was the only hemodynamic factor that was considerably greater preoperatively (130 ± 26 mm Hg versus 113 ± 20 mm Hg; P = .082), although there were no echocardiographic changes when this occurred.
As shown in Table 3, in addition to age, race, and body surface area, atrial filling fraction (lowest 1st quartile) and mitral inflow duration (highest 4th quartile) following surgery were independent predictors of postoperative atrial fibrillation, based on multivariate analysis.
The results of our study showed that patients who developed postoperative atrial fibrillation were more likely to be older, to have increased body surface area, and to be white or Asian. As a result of a reduction in myocardial fibers and increased fibrosis, aging may result in atrial remodeling.9,10 In the Framingham study, it was recently reported that the excess risk of atrial fibrillation associated with obesity appears to be mediated by left atrial dilatation,11 which was also shown in our study. In addition, we found that Hispanics and possibly African Americans were not as likely as whites to develop postoperative atrial fibrillation. Two previous studies of nonsurgical patients similarly showed that African Americans were not as likely as whites to develop atrial fibrillation.12,13 The reasons for these associations with race are still not clear.
In addition to these clinical factors, our study also showed that those
who developed atrial fibrillation postoperatively had signs of structural remodeling preoperatively, as shown by a larger left atrium and a tendency toward having a reduced LAEF and a larger left atrial appendage measurement. Compared with patients who did not develop postoperative atrial fibrillation, those who did had evidence of left ventricular relaxation impairment, increased atrial conduit function, and decreased atrial systolic function.
In our study, there was a decrease in the atrial filling fraction and an increase in the VTI E/A fraction among those who developed atrial fibrillation after surgery, which are indices of atrial systolic function and left atrial conduit function (analogous to stroke volume), respectively. The increase in left atrial stroke volume may be secondary to an increase in autonomic nervous system activity.14 The decrease in atrial filling fraction and increase in left atrial conduit function may be an indication of increased left atrial afterload, possibly as a result of the increased left ventricular stiffness, as shown by abnormal ventricular relaxation in our study. Furthermore, an increase in left atrial volume15 and dilatation of the left atrium and pulmonary veins may result from abnormal relaxation of the left ventricle, which might result in higher atrial pressures during atrial diastole. This hypothesis is supported by our findings that patients with postoperative atrial fibrillation had larger pulmonary veins, left atrial appendages, and left atria. There is also evidence of an increase in right atrial pressure, as reflected by an increase in hepatic vein velocity, in these patients.
Are these changes in atrial function related to atrial stunning or ischemia after open-heart surgery, or both? Because the atrial functional abnormalities occur predominantly after cardiopulmonary bypass, the evidence indirectly suggests that surgery or cardiopulmonary bypass, or both, may have been an inciting factor. How-
ever, a more sensitive technique may
be necessary to show the adequacy
of atrial protection during cardiopulmonary bypass because the adequacy of myocardial protection as assessed
by the surgeons subjectively was similar between patients with and with-
out postoperative atrial fibrillation in our study.
The findings from our study provide evidence that patients who develop atrial fibrillation after bypass surgery have some of the same functional and structural alterations in the atria as those found in elderly patients with long-term atrial fibrillation. Furthermore, the development of atrial fibrillation after cardiac surgery is probably preceded by atrial dysfunction.
Because prophylactic antiarrhythmic therapy may be prohibitive in cost and may have unwanted side effects, carefully choosing patients for therapeu-
tic intervention before cardiac surgery based on the prediction of clinical risk should be explored as an alternative approach. Some of the risks identified in our study include a decrease in atrial systolic function and diastolic filling as shown on Doppler echocardiogram, white race, body surface area above 2.12 m2, and age older than 70 years. Additionally, future studies should evaluate methods to improve left ventricular diastolic filling and decrease left atrial size. Studies focusing on how atrial function is affected by loading conditions may also provide insight into how atrial dysfunction prior to surgery influences the occurrence of atrial fibrillation after cardiac surgery.