Postoperative supraventricular tachyarrhythmia is a common complication of surgery. Because chemical cardioversion is often ineffective, ventricular rate control remains a principal goal of therapy. The authors hypothesized that patients with supraventricular tachyarrhythmia after major noncardiac surgery who receive intravenous beta-adrenergic blockade for ventricular rate control would experience conversion to sinus rhythm at a rate that differs from those receiving intravenous calcium channel blockade.
The rate of conversion to sinus rhythm at 2 and 12 h after treatment was examined in 64 cases of postoperative supraventricular tachyarrhythmia. After adenosine administration, patients who remained in supraventricular tachyarrhythmia were prospectively randomized to receive either intravenous diltiazem or intravenous esmolol for ventricular rate control (unblinded). Loading and infusion rates were adjusted to achieve equivalent degrees of ventricular rate control.
Patients were similar with regard to age and Apache III score. Most patients in both groups had atrial fibrillation (esmolol, 79%; diltiazem, 81%), and none experienced stable conversion with adenosine. Patients randomized to receive esmolol experienced a 59% rate of conversion to sinus rhythm within 2 h of treatment, compared with only 33% for patients randomized to receive diltiazem (intention to treat, P = 0.049; odds ratio, 2.9; 95% confidence interval, 1.046 to 7.8). After 12 h of therapy, the number of patients converting to sinus rhythm increased in both groups (esmolol, 85%; diltiazem, 62%), and the rates of conversion no longer differed significantly. Ventricular rates when supraventricular tachyarrhythmia began and after 2 and 12 h of rate control therapy were similar in the two treatment groups. The in-hospital mortality rate and length of stay in the intensive care unit were not significantly influenced by treatment group.
Among adenosine-resistant patients in the intensive care unit with atrial fibrillation after noncardiac surgery, intravenous esmolol produced a more rapid (2-h) conversion to sinus rhythm than did intravenous diltiazem.
This article is accompanied by an Editorial View. Please see: Pearson KS: Emergency informed consent. Anesthesiology 1998; 89:1047-9.
SUPRAVENTRICULAR tachyarrhythmia (SVT) occurs in 11-40% of patients after cardiac surgery, is an independent factor for postoperative stroke, and is associated with a 3- to 5-day lengthening of hospital stay. [1-3]It also occurs after major noncardiac operations. [4-6]The incidence is highest among patients with multiple medical illnesses, and the mortality rate for them is markedly increased. [5-8]Efforts to hasten conversion of SVT with anti-arrhythmic agents have been unsuccessful because of poor efficacy and undesired side effects ; thus control of the ventricular rate remains a principal goal of therapy for patients in the intensive care unit (ICU). [7,11]Digitalis, once the preferred agent, is now used less commonly because of its ineffectiveness and slow onset of action. Calcium channel blockers and [small beta, Greek]-blockers provide rate control within minutes and are used interchangeably in patients with normal cardiac contractility.
The cause of SVT in patients after operation is multifactorial but may involve neurohumoral factors related to the perioperative stress response. Serum catecholamine levels are high after surgery, [13,14]and prophylactic administration of [small beta, Greek]-blockade reduces the incidence of SVT after cardiothoracic surgery. [15,16]Catecholamines modulate voltage-gated ion channel function and may induce SVT by shortening the refractory period, accelerating conduction velocity, or increasing the rate of spontaneous diastolic depolarization. Although they slow AV nodal conduction and control ventricular rate, [small beta, Greek] blockers have indirect anti-arrhythmic effects in atrial cells stimulated by catecholamines. Calcium channel blockers have little acute electrophysiologic effect on atrial fibers but may influence electric remodeling during sustained atrial tachycardia. 
We hypothesized that postoperative patients with SVT who receive [small beta, Greek]-blockade for ventricular rate control would experience conversion to sinus rhythm at a rate that differs from those receiving calcium channel blockade. To examine this hypothesis, we prospectively compared the rate of SVT conversion in noncardiac surgical ICU patients during rate-control therapy with either intravenous esmolol or intravenous diltiazem.
With the consent of our institutional review board, 63 noncardiac surgical ICU patients with recent-onset SVT were randomized to receive intravenous diltiazem or intravenous esmolol. Although recent-onset typically refers to SVT present for as long as 24 h, our standard of practice is to administer intravenous rate control therapy within 15 min of the time when tachyarrhythmia is detected. Both agents are Federal Drug Administration-approved therapies and are used interchangeably for SVT rate control in our patients when there are no contraindications to [small beta, Greek]-blockade (see below). Because most noncardiac surgical ICU patients cannot provide informed consent when SVT commences (because of intubation, sedation, delirium, anxiety, and so forth), and recognizing the potential risks of delaying rate control therapy to obtain consent from relatives (in cases of cardiac ischemia, for example), our institutional review board waived prospective written consent. However, patients or their nearest relatives were contacted within 24 h to obtain permission to use clinical data for publication purposes.
Patients were excluded from the study if they were hemodynamically unstable and required DC cardioversion (systolic blood pressure, <80 mmHg), had hemodynamic contraindications to intravenous [small beta, Greek]-blockade (ejection fraction <or= to 30%; cardiac index <2.0 1/min/m2), depended on cardiac inotropes, or had documented bronchospasmic disease. Patients with preoperative chronic SVT and patients with “do not resuscitate” orders who were experiencing SVT as a preterminal event were not enrolled. Patients with a heart rate > 100 beats/min and not in sinus rhythm first received an adenosine challenge (6 mg, followed by 12 mg if no response) to rule out either reentrant AV nodal rhythms or ventricular tachycardias. Patients who experienced transient slowing of the ventricular rate without conversion to sinus rhythm (atrial fibrillation, atrial flutter, or other atrial tachyarrhythmias) were randomized to receive esmolol or diltiazem (open label) for ventricular rate control. Patients who converted to sinus rhythm with adenosine administration (suspected nodal reentrant rhythms) were randomized only if the rhythm recurred.
Loading and infusion rates were adjusted to achieve similar degrees of ventricular rate control with standard dosing regimens used in our surgical ICU. Patients randomized to received diltiazem were given a loading infusion of 20 mg over 2 min, followed immediately by a 10-mg/h maintenance infusion. After 15 min, patients with heart rates exceeding 110 beats/min received an additional loading infusion of 25 mg and a 5-mg/h increment in their maintenance infusion rate. After 30 min, patients receiving a maintenance infusion < 15 mg/h with a heart rate > 100 beats/min received an additional 5-mg/h increment in their infusion rate. Patients randomized to receive esmolol were given a 12.5-mg intravenous bolus, followed by additional 25- to 50-mg boluses every 3 to 5 min until the heart rate decreased <110 beats/min or a total loading dose of 250 mg was attained. The maintenance infusion was 50 [micro sign]g [middle dot] kg-1[middle dot] min-1for patients receiving < a 30-mg load, and 100 [micro sign]g [middle dot] kg-1[middle dot] min-1for patients receiving more than 30 mg. After 15 min, patients with heart rates > 110 beats/min received one to four boluses of 25 mg, followed by a 50-[micro sign]g [middle dot] kg (-1)[middle dot] min-1increment in their maintenance infusion. This was repeated after 30 min for patients with a heart rate > 100 beats/min. Beyond 30 min, esmolol and diltiazem infusion rates were adjusted by the physician staff to maintain heart rates between 80 and 100 beats/min. If at any time a patient experienced symptomatic hypotension or the systolic blood pressure was <80 mmHg, the infusion rate was decreased by 50%, and a fluid bolus of 500 ml or a phenylephrine infusion, or both, was administered. Temporary discontinuation of the agent or DC cardioversion were implemented at the discretion of the physician staff. Similarly, drug infusions were titrated down or discontinued for heart rates <60 beats/min or symptomatic bradycardia (e.g., heart rates <75 beats/min with systolic blood pressure <80 mmHg, cardiac ischemia, mental status changes).
Twelve-lead electrocardiograms and rhythm strips were obtained before esmolol or diltiazem were given (after adenosine), after 2 h (our primary end point), and after 12 h (our secondary end point). These tracings were subsequently reviewed by a cardiologist (G.F.T.) blinded to patient treatment. At the time of SVT onset, potassium and magnesium levels were checked. Patients received immediate potassium chloride repletion to levels >or= to 4 mEq/l. Because of the longer period required to obtain magnesium chemistries from the laboratory, intravenous magnesium sulfate (2-4 g) was administered for levels <2 mEq/l after the 2-h primary end point. Digoxin (0.25 mg every 6 h x 4) was also administered after the initial 2 h at the discretion of the ICU staff for long-term rate control. Beyond 12 h, patients continued to receive their rate control therapy but were no longer monitored for study purposes.
Continuous variables are expressed as means +/− SD. Categorical variables (Table 2) were expressed as actual numbers (numerator/denominator) and also as percentages. Conversion frequencies in the two groups were compared using a two-tailed Fisher's exact test. The influence of patient characteristics (Table 1) on conversion frequency was analyzed using logistic regression analysis. Significance was considered at the P < 0.05 level.
Of the 63 patients enrolled, one patient entered the study on two occasions separated by 4 days and was randomly assigned to receive diltiazem on both occasions. Therefore, we studied a total of 64 cases of SVT, with 34 patients randomized to receive esmolol and 30 to receive diltiazem. Although adenosine administration sometimes induced transient (1 or 2 min) conversion, SVT invariably recurred and no patients were excluded from enrollment based on adenosine-induced cardioversion. Because of enrollment errors or patient intolerance (see Morbidity section), 55 patients with nonsinus tachyarrhythmias continued to receive rate control therapy until the primary 2-h end point (31 esmolol, 28 diltiazem). Table 1provides a breakdown of the patient characteristics. Patients in the two treatment groups were similar with regard to age, sex, and Apache III score. Patients experienced SVT after various surgical procedures, most within 3 days of surgery (82%). Although a significant number of SVT cases occurred after thoracic surgery (27%), major abdominal procedures (Whipple, hepatic surgeries, and so on) accounted for the largest number of cases studied.
Most patients in both groups had atrial fibrillation, whereas a few had either atrial flutter or paroxysmal supraventricular tachyarrhythmias that could not be diagnosed specifically from the electrocardiograph tracings. The most common comorbidities among patients with postoperative SVT were coronary artery disease, hypertension, and chronic obstructive pulmonary disease. Three patients had SVT in a setting of recent or ongoing myocardial ischemia. Both treatment groups contained similar numbers of patients taking [small beta, Greek]-blockers or calcium channel blockers when SVT developed (Table 1). A few patients were taking digoxin (two esmolol, two diltiazem) or intravenous vasoactive agents (one esmolol, three diltiazem) when SVT began (Table 1).
Patients from both groups had similar ventricular rates at SVT onset and throughout drug administration (Figure 1). In both groups, the mean ventricular rate was reduced to 100 beats/min after 2 h of therapy, and to approximately 85 beats/min by 12 h. Although rate control with the two agents was similar, 59% of the patients enrolled and randomized to receive esmolol converted to sinus rhythm within 2 h, compared with only 33% of the patients randomized to receive diltiazem (intention to treat:Table 2; n = 64, P = 0.049). Among the patients with nonsinus supraventricular tachyarrhythmias who received rate control therapy until the 2-h primary end point, the rate of conversion to sinus rhythm was 68% for those receiving esmolol compared with 37% for those receiving diltiazem (n = 55, P = 0.031;Figure 2). Differences in rates of conversion among only patients with atrial fibrillation who received rate control therapy for 2 h were not significant (esmolol, 59%; diltiazem, 27%; P = 0.067;Table 2). Between 2 and 12 h, a number of patients converted to sinus rhythm in both treatment groups. By 12 h, 85% of patients who continued to receive esmolol had converted to sinus rhythm, compared with 62% of those receiving diltiazem, and the between-group difference was no longer significant (P = 0.116;Figure 2). Approximately 40% of the patients in both groups received magnesium between 2 and 12 h (Table 2), potentially contributing to enhanced rate control (Figure 1) and an increased rate of conversion (Figure 2) by the 12-h time point. Of the 11 patients who converted to sinus rhythm between 2 and 12 h, three received magnesium (one of five esmolol cases, two of six diltiazem cases). The rate of SVT recurrence between 2 and 12 h for those patients converting to sinus rhythm by 2 h was low in both groups (Table 2).
Logistic regression analysis indicated that the patient variables listed in Table 1(age, Apache score, and so on) did not significantly modify the effect of esmolol or diltiazem on the rate of 2-h conversion. The number of SVT cases occurring more than 72 h after surgery was small (22% for diltiazem compared with 14% for esmolol), and the postoperative day did not modify the effect of treatment on the overall rate of conversion. Nonetheless, because a few patients developed SVT more than 72 h after surgery, there was insufficient statistical power to determine whether late-developing SVT was more or less resistant to conversion. In both treatment groups, preexisting [small beta, Greek]-blockade or calcium channel blockade did not produce significant differences in the rate of conversion at 2 h (Table 2).
Morbidity and mortality data (Table 3) are provided for all 63 patients enrolled. The length of ICU stay and the rate of in-hospital mortality did not differ significantly for the two treatment groups (Table 3). The mortality rates were not different among patients who converted by 2 h compared with those who did not (33% vs. with 32%) or for patients converting by 12 h compared with those who did not (28% vs. 40%; P = 0.4). There were no deaths directly attributable to either SVT or complications from rate-control therapy. Patients experiencing SVT as a preterminal event (designated “do not resuscitate”) were not randomized into the study, did not receive antiarrhythmic therapy, and are not included in our morbidity statistics. Significant numbers of patients in both groups (30-40%) required either intravenous fluid boluses or intravenous vasopressor therapy (phenylephrine) to maintain a minimum systolic blood pressure of 80 mmHg during the esmolol or diltiazem intravenous loading period. In most cases, the need for such therapy was transient; nonetheless, hypotension required discontinuation of rate-control therapy in three patients (Table 3) before the 2-h primary end point. One of these patients converted to sinus rhythm before discontinuation of therapy. A fourth patient did not experience a reduction in heart rate with esmolol loading and was crossed over to diltiazem therapy before the 2-h primary end point. Five additional patients were enrolled in error. One patient with moderate left ventricular dysfunction (cardiac index, 1.9 1 [middle dot] min-1[middle dot] m-2off inotropes) was inadvertently enrolled in the trial. The patient was randomized to receive esmolol but was discontinued shortly after the loading period when the error was recognized. There were no hemodynamic sequelae.
Our results (Figure 1, Table 2) indicate that [small beta, Greek]-blockade and calcium channel blockade are equally effective for controlling ventricular rate during postsurgical SVT. However, the rate of conversion to sinus rhythm during the first 2 h was significantly increased when esmolol was used for ventricular rate control rather than diltiazem (Table 2). Beyond 2 h, the conversion rates in both treatment groups increased, and the rates of conversion after 12 h did not differ for the two groups. None of the patients experienced stable conversion with adenosine, suggesting that few (if any) patients had rhythms involving AV-nodal reentry. Because few patients had atrial flutter or other SVTs (Table 1), our findings are most applicable to patients with atrial fibrillation.
Although these data suggest that intravenous [small beta, Greek]-blockade accelerates conversion of postoperative SVT, an alternate interpretation is that calcium channel blockade prolongs SVT. The use of a placebo group (no rate control therapy) to exclude this possibility was deemed unethical in ICU patients at risk for end-organ damage from hypoperfusion. The 12-h rate of conversion among patients receiving diltiazem (62%) was similar to spontaneous rates reported for nonsurgical patients with recent-onset atrial fibrillation (50-60%). [9,22]Thus it seems unlikely that calcium channel blockade prolonged SVT in this trial.
Several antiarrhythmic agents may prevent SVT after cardiac surgery when administered prophylactically, including amiodarone, [23,24]procainamide, and sotalol. [26,27]However, for patients already experiencing SVT, these agents have little benefit for accelerating conversion to sinus rhythm. Among nonsurgical patients with recent-onset atrial fibrillation, the 24-h rate of conversion on intravenous amiodarone (68%) was similar to placebo (60%). Although ibutilide is highly effective for converting recent-onset atrial fibrillation, this therapy has been associated with an 8% incidence of polymorphic ventricular tachycardia in nonsurgical patients, and more generalized use has been questioned. 
Among the standard AV nodal blocking agents, both [small beta, Greek]-blockers and magnesium seem to facilitate conversion to sinus rhythm while controlling ventricular rate. In a trial of 31 patients with recent-onset SVT, 50% of patients receiving esmolol converted to sinus rhythm within 1 h, compared with only 12% of patients receiving verapamil (P <0.03). However, this trial did not examine SVT in postsurgical ICU patients, and details regarding severity of illness are not provided. A trial of 57 nonsurgical patients with recent-onset SVT found 58% of magnesium-treated patients converted to sinus rhythm within 4 h, compared with only 23% of verapamil-treated patients (P < 0.01). By 24 h, the rate of conversion was similar (50-60%) in the two treatment groups. Another study of 42 ICU patients with SVT found a high-dose infusion of intravenous magnesium provided a higher rate of conversion to sinus rhythm than intravenous amiodarone (60% vs. 44% at 4 h; 72% vs. 50% at 12 h; 78% vs. 50% at 24 h). In our study, magnesium was not administered until after the 2-h primary end point and therefore could not have influenced the 2-h rate of conversion. However, 42% of patients receiving esmolol and 38% of patients receiving diltiazem received 2-4 g intravenous magnesium between 2 and 12 h, and 3 of the 11 patients who converted to sinus rhythm during this time interval received magnesium (Table 2: one esmolol case, two diltiazem cases). Although the increase in the rate of conversion in both groups from 2 to 12 h may be associated with part with magnesium administration, it is doubtful that magnesium therapy disproportionately biased the 12-h conversion rate in either treatment group. Magnesium influences many cardiac ion channels and second messenger systems and may mimic [small beta, Greek]-blockade by blunting catecholamine release from peripheral and adrenal sources. [33,34]
Two patients in each treatment group were taking digoxin before developing SVT, and only 20% of the patients in both treatment groups received digoxin between 2 and 12 h. Although effective for chronic rate control, digoxin does not effect conversion to sinus rhythm and has not been useful to prevent postoperative atrial arrhythmias when administered prophylactically. Digoxin therapy is therefore unlikely to have influenced the results of this trial. The number of patients receiving [small beta, Greek]-blockade or calcium channel blockade before SVT was nearly identical in the two groups (Table 1), suggesting that preexisting drug therapy did not bias the conversion rates in the treatment groups disproportionately. However, our sample size did not allow us to determine whether preexisting [small beta, Greek]-blockade or calcium channel blockade significantly altered the rate of conversion with either agent (Table 2). Of note, 34% of our patients were already receiving [small beta, Greek]-blockade at the time of onset of SVT, suggesting either that increased postoperative sympathetic tone overwhelmed existing [small beta, Greek]-blockade to induce the arrhythmias, or that additional perioperative factors such as pericardial inflammation or pulmonary hydrostatic changes [5,35]were important arrhythmogenic mechanisms. The latter notion is supported by cardiac surgery studies showing that postoperative [small beta, Greek]-blocker prophylaxis reduces the frequency of SVT but does not eliminate the problem. [15,16]
Although thoracic (pulmonary, aortic) operations are the noncardiac surgeries most often linked to postoperative SVT, [5,7]the largest group in Table 1are major abdominal and vascular operations; importantly, these nonthoracic operations represent a majority of the admissions to our surgical ICU. Patients having noncardiac surgery share certain characteristics with cardiac surgery patients with regard to postoperative SVT. The comorbidities noted most frequently in our study (Table 2) were coronary artery disease, hypertension, and pulmonary disease. A few patients had left ventricular hypertrophy or experienced myocardial ischemia after operation. Goldman studied 916 patients prospectively to determine clinical correlates of postoperative SVT after noncardiac surgery. In addition to advanced age and major vascular surgery, significant correlates included myocardial ischemia, left ventricular hypertrophy, and other noncardiac illnesses such as chronic obstructive pulmonary disease. Similar preoperative risk factors predispose to SVT after cardiac surgery. [1-3]Furthermore, the types of SVT that occur after cardiac and noncardiac surgery are similar. In our trial, atrial fibrillation was most common (80%), with a few experiencing atrial flutter (4%) or other supraventricular tachycardis (16%, Table 2). In a study of SVT after thoracic surgery for lung cancer, atrial fibrillation was also preponderant (87%). Most SVTs that require medical intervention after cardiac surgery are atrial fibrillation. 
Although SVT after cardiac and noncardiac surgery share similarities, there are also clear differences. Although SVT is more common after cardiac surgery than noncardiac surgery, the mortality rate is <5%and does not exceed that of patients who remain in sinus rhythm. In contrast, we found a high (30-40%) mortality rate among patients with SVT having noncardiac surgery (Table 3). Goldman followed 35 patients who developed SVT after noncardiac surgery and found an even higher (49%) mortality rate. Our decision not to enroll patients who developed SVT as a preterminal event may explain our slightly lower mortality rate. A recent study of 298 patients having major noncardiac surgery found a 26% mortality rate among patients with SVT, compared with only a 1.7% mortality rate among those patients who remained free of supraventricular arrhythmias. After pulmonary resection, mortality rates in patients with postoperative SVT range from 17% to 33%. [5,8]Although none of these trials attribute death directly to the arrhythmia, [7,37]the far higher mortality rate associated with SVT after noncardiac surgery is compelling. Surgical ICU patients with postoperative SVT may suffer end-organ damage from compromised cardiac output (atrioventricular dyssynchrony, impaired diastolic filling). Although intuition suggests that ICU patients may benefit from accelerated conversion to sinus rhythm after operation, a much larger trial would be necessary to determine whether early conversion influences outcome.
The higher mortality rate of patients with postoperative SVT after noncardiac surgery may also suggest important differences in the etiologic factors precipitating SVT after cardiac and noncardiac operations. In noncardiac surgery patients, unopposed [small beta, Greek]-adrenergic stimulation has been implicated as a cause of postoperative cardiovascular complications and death. Although the cause of postoperative SVT is undoubtedly multifactorial, serum catecholamine levels are elevated after major noncardiac surgery, [13,14]the arrhythmogenic potential of catecholamines in the human atrium is well established, [18,39]and accumulating evidence suggests that agents with antiadrenergic activity ([small beta, Greek]-blockers, magnesium) accelerate conversion of SVT. [29-31]Thus the role of unopposed [small beta, Greek]-adrenergic stimulation in the genesis of postoperative SVT deserves further study.