Figure 1, A. Relation between increasing age and postoperative atrial fibrillation as occurring in the PSOCS trial , demonstrating the higher rate of postoperative atrial fibrillation in older patients. B. Similar age distribution curve for atrial fibrillation occurring in the general population as occurring in the Framingham population , (©American College of Surgery, 1997 and ©1994 American Medical Association. Reproduced with permission).
i graphically demonstrates this comparison, using the incidence of AF by age in the Framingham population  as representative of the general population. in addition, it is also important to note that direct evidence connecting the above-
described age-related changes in atrial substrate to the age-related increase in ostoperative AF does not exist.
Of note, several small studies have failed to demonstrate an association between age and postoperative AF [14, 21-23], However, all were studies of 400 or fewer patients and simply compared mean ages of patients with and without postoperative AF.
While a strong association between male gender and AF seems well established in the general population, this association remains somewhat more controversial in postoperative AF. For example, large single center trials have reported no association between gender and postoperative AF. Creswell et al  evaluated 3983 consecutive patients undergoing cardiac surgery; 1328 (33%) were women. In this study, rates of AF were similar between men and women. In addition, many previously published studies [1,13,19,25-29] have failed to demonstrate an association between gender and postoperative AF when multivariate analysis was performed. However these studies have relatively small numbers of patients and low percentages (10-20%) of women. It is possible that these relatively small samples may be unable to detect differences in rates of AF occurrence relative to gender.
In contrast, two large studies support the association of male gender to postoperative AF. The MCSPI trial  included 537 women out of 2417 total patients (24%). In this setting, male gender demonstrated an independent association with AF with an odds ratio of 1.41 (95% confidence intervals: 1.041.81). Similarly, Aranki et al  with 175 women of 570 total patients (31%) demonstrated a significant association of AF to male gender with an odds ratio of 1.7 (95% confidence intervals: 1.1-2.7). These findings confirm the results of several previously published smaller studies [19, 14, 26], As a result of these findings, Fuller and co-authors  have hypothesized that a hormone-related protective mechanism may account for this gender difference. Others have postulated the existence of hormonal effects on autonomic tone and/or gender differences in 3-dimensional myocardial architecture . These hypotheses remain to be proven.
Congestive heart failure. The association of a preoperative history of congestive heart failure with the development of postoperative AF is also somewhat controversial. In the MCSPI trial , 689 of the 2417 patients (28.5%) had a preoperative history of congestive heart failure defined by >NYHA Class II symptoms; of these patients, 238 (35%) developed postoperative AF. In comparison, 24% of patients without prior congestive heart failure symptoms developed postoperative AF. The odds ratio for the association of congestive heart failure with postoperative AF was 1.31 (95% confidence intervals: 1.04-1.64). This association seems plausible given that in the setting of congestive heart failure atrial stretching, thinning, and scarring have been reported; these changes can result in anisotropic conduction, slowing of conduction and dispersion of atrial repolarization. However, in the MCSPI trial, the history of congestive heart failure was determined by a retrospective review of patient records and physician interviews which may in fact have introduced bias. Interestingly, objective measurements associated with congestive heart failure, such as decreased left ventricular ejection fraction and elevated left ventricular end-diastolic pressure, did not predict postoperative AF in this study (see below).
Other large, single center studies have failed to corroborate these findings. Hashimoto et al , failed to demonstrate an association between preoperative symptoms of congestive heart failure and postoperative AF; however, in this study elevated left ventricular end-diastolic pressure (>20 mm Hg) was associated with postoperative AF. Likewise, Aranki et al  failed to demonstrate an association of previous congestive heart failure to postoperative AF. Several smaller studies have also failed to demonstrate an association [28, 14].
Hypertension. While in the general population the association of systemic hypertension with the development of AF is clearly documented, its association with the development of postoperative AF remains controversial. In data from Leitch et al , the largest series to date, postoperative AF developed in 18.2% of patients with a history of hypertension as compared to 16.2% of patients without hypertension (p< 0.05). Almassi et al , in a population of 3855 predominantly male veterans, noted a history of hypertension in 702 of the 1143 patients who developed postoperative AF (61.4%); this compared to a history of hypertension in 1571 of the 2711 patients (58%) who did not develop postoperative AF (p=0.046). Resting systolic blood pressure > 120 mm Hg was an independent predictor of postoperative AF in this population. Aranki et al , in a single center study, also noted borderline significance to the association of hypertension and postoperative AF: 122/165 (65%) of patients with postoperative AF had previously diagnosed hypertension as compared to 214/381 (56%) of patients without postoperative AF.
Offering conflicting data, the MCSPI trial , with 2417 patients from 24 centers, noted a similar incidence of AF in patients with and without a history of hypertension (29% vs. 26%, respectively; p=0.18). Likewise Hashimoto et al  demonstrated no association between preoperative hypertension and postoperative AF. Multiple smaller studies have also failed to demonstrate an association [17-19, 22, 26, 28, 30],
Data published from our institution  also failed to demonstrate an association between hypertension and postoperative AF; interestingly, however, the presence of left ventricular hypertrophy on the surface ECG was associated with the development of postoperative AF in this same study. Since the ECG changes associated with hypertrophy occur most often in the setting of longstanding hypertension, these findings seem to suggest that changes in the myocardial substrate resulting from hypertension play an important and possibly necessary role in the development of postoperative AF.
Chronic lung disease. The association of preoperative lung disease with the development of postoperative AF is also somewhat controversial with conflicting data present in previously published reports. Almassi et al  noted that 219/1141 (19.57%) patients in a Veterans Administration trial who developed postoperative AF had chronic obstructive pulmonary disease (defined as FEV1 <1.5 L); this compared to chronic obstructive pulmonary disease in 367/2714 (13.57%) of patients without postoperative AF. Leitch et al , in a series of 5807 consecutive patients, showed a significant association of postoperative AF with "chronic airflow limitation" as defined by "usual clinical criteria". In this series, 25.5% of patients with "chronic airflow limitation" developed postoperative AF as compared to 16.8% without; this difference did not achieve statistical significance. In addition, the MCSPI trial  failed to demonstrate and association of with "chronic lung disease" as defined in medical records and physician interviews; further breakdown of the diagnostic category into emphysema, bronchitis or asthma also failed to elucidate an association. Likewise, while Aranki et al  demonstrated a trend toward the development of postoperative AF in patients with "chronic lung disease"; this difference failed to reach statistical significance.
Most previously published series have failed to demonstrate a significant association between the development of postoperative AF and the presence of diabetes mellitus [10, 12, 19, 24], history of previous myocardial infarction [3, 8, 10, 19] or prior cardiac surgery [8, 10, 12].
3. Adrenergic Stimulation, Preoperative Beta-Blocker Usage and Beta-Blocker Withdrawal
The sympathetic nervous system has long been hypothesized to play a significant role in the development of AF especially in the postoperative setting . In the experimental and clinical electrophysiology laboratory, adrenergic stimulation has been shown to shorten the refractory period of myocardial tissue . In addition, adrenergic stimulation may increase the frequency of ectopic beats. These changes can have a critical impact on the occurrence of AF , As such, the implication that adrenergic stimulation can provide an adequate trigger for the development of AF has intuitive appeal. It is from this hypothesis that the postoperative use of beta-blocker therapy as a preventative measure for AF had arisen. While this therapy has been extensively studied [1,34-40], the role of adrenergic stimulation, preoperative beta-blocker usage and beta-blocker withdrawal also deserve study. As suggested by Frost et al , many postoperative beta-blocker prevention trials fail to account for these effects in control populations.
In the postoperative period, one study has demonstrated an association of increased norepinephrine levels with the occurrence of AF. Kalman et al  prospectively studied 131 consecutive patients undergoing coronary bypass surgery of which 65 (50%) developed AF. Serum norepinephrine levels were determined immediately after surgery and every 4 hours for 48 hours following the surgery in sample drawn from the right atrial cavity via a central venous catheter. Patients who developed AF had significantly higher serum norepinephrine levels immediately following surgery (5.78±2.83 nmol/L vs. 3.57±1.31 nmol/L, pO.OOOl). In addition, the elevation in norepinephrine levels remained significant for every 4-hour sampling interval in the immediate postoperative period. While this type of sampling represented only generalized sympathetic activity, it clearly documented differences in those patients who developed AF. Not surprisingly, serum norepinephrine levels were significantly higher in those patients on beta-blockers preoperatively. Sun et al , studying the effects of hypothermia on the sympathetic response during cardiopulmonary bypass, demonstrated increased norepinephrine, epinephrine and neuropeptide Y levels in patients cooled to 28° C as compared to those cooled to 34° C. Those patients with greater elevations in adrenergic hormones were also more likely to develop postoperative AF.
Several previously published reports have investigated the association of preoperative beta-blocker use with postoperative AF. Leitch et al , in the largest study to date, showed that 17.9% of patients on preoperative beta-blockers developed AF as compared to 14.9% not on the medication (p<0.05). In contradistinction, the MCSPI trial , in which 42% of the 2417 patients included in the trial were on preoperative beta-blockers, showed no difference in the rates of postoperative AF between the 2 groups. Several smaller studies also demonstrate varying results [14,24], Important to note, however, is that in none of these studies were the preoperative indications for beta-blocker usage or the postoperative continuation of the drug commented upon. As a result, many of these trials may in fact suggest an association of beta-blocker withdrawal with postoperative AF.
The activation of sympathetic nerves and enhancement of the response to adrenergic stimulation found in the setting of beta-blocker withdrawal can provide the trigger necessary for atrial arrhythmias , In the course of studying the preventative value of postoperative beta-blocker therapy, two studies have specifically investigated the clinical impact of beta-blocker withdrawal in cardiac surgery. Abel et al  evaluated 91 patients treated with preoperative propranolol who were randomized to postoperative placebo (withdrawal) or continued propranolol treatment. In the postoperative period with 3 days of continuous telemetry monitoring and symptomatic evaluation thereafter, AF developed in 22/50 (44%) of patients withdrawn from propranolol as compared to 9/41 (22%) of those continued on therapy. In a similarly designed study, Silverman et al  demonstrated postoperative AF in 3/50 (6%) of patients (p=0.006) on continued beta-blocker therapy as compared to 14/50 (28%) in the withdrawal group. In addition, White et al  and Salazar et al  both found two- to five-fold increases in the incidence of postoperative AF when preoperative beta-blocker therapy was discontinued postoperatively. These studies form the basis for ihc proof of the role of postoperative beta-blocker withdrawal.
Diagnostic testing is typically performed for a whole host of reasons prior to cardiac surgery. The diagnostic testing to evaluate the potential for postoperative atrial arrhythmias can: 1) can dircctly or indirectly evaluate atrial electrical properties by using the surface ECG or signal-averaged P wave, 2) indirectly evaluate and quantitale structural changes in the atrial substrate, such as left atrial size in echocardiography, or 3) indirectly measure extrinsic influences on the atrial substrate, such as heart rate variability measurements for autonomic tone. In general, preoperative diagnostic testing has been disappointing, likely related to the significant impact the actual surgical event must have upon the occurrence of postoperative AF.
4.1. P wave analysis
The presence of intra-atrial conduction defects on atrial endocardial mapping represents an abnormal electrical substrate. These conduction disturbances have been associated with the occurrence of spontaneous atrial fibrillation  and are usually associated with prolongation of the P wave on the surface electrocardiogram . This finding led Buxton and Josephson  to first describe the association of preoperative P wave prolongation on the surface ECG and the development of postoperative AF. In their original paper, postoperative AF developed in 24/64 (38%) patients with P wave duration >110 ms as compared to developing in 5/35 (14%) patients with P wave duration <110 ms (p<0.05). As such, the technique had a relatively low specificity as a predictive test. In addition, a recent multivariate evaluation (15) has suggested that independent of age and body weight, P wave duration may have little real value. Figure 2 demonstrates a routine preoperative ECG with markedly prolonged P wave duration in a patient who developed AF postoperatively.
| o 30 to « :io iw :so 210 n» i~o ">00 ]$eii, ¡-j
Was this article helpful?