A Randomized Controlled Trial of Magnesium Sulfate, in Addition to Usual Care, for Rate Control in Atrial Fibrillation

Background 

Atrial fibrillation is the most common sustained clinically significant arrhythmia.1 The majority of patients with a rapid ventricular response rate (rapid atrial fibrillation) do not require immediate electrical cardioversion. However, if allowed to remain in sustained rapid atrial fibrillation for hours, tachycardia-induced left ventricular dysfunction may result.2, 3 Ventricular rate control is generally the primary therapeutic objective. Early conversion to sinus rhythm is of secondary importance and, in many cases, may not be desirable until anticoagulation has been effected.4, 5

A variety of pharmacologic agents are used for early rate control. Calcium-channel blockers, β-blockers, sotalol, and Vaughan Williams Class 1C agents must be used with caution in patients with poor left ventricular function—precisely the group in which pulse rate reduction is most urgent. For this group, digoxin or amiodarone has been recommended.4, 6, 7 Digoxin has recently been the mainstay of therapy, and although inexpensive, it has a slow onset of action and is significantly less effective in states of increased sympathetic tone.4, 6, 8, 9, 10 Despite the widespread use of amiodarone in this setting, the evidence for its efficacy is somewhat equivocal. It also has a high adverse-effect profile with chronic dosing.11

Magnesium sulfate has a number of biochemical and electrophysiologic properties that might make it useful in the treatment of rapid atrial fibrillation. Specifically, its role as a coenzyme for the sodium-potassium ion exchange pump (the Na-K ATPase) and its effects on potassium channels and intracellular calcium accumulation all serve to decrease automaticity.12, 13, 14 Clinically, it has been found that atrioventricular nodal conduction is prolonged in patients treated with magnesium sulfate infusions.15 Such effects might serve to decrease the ventricular response rate to atrial fibrillation. It also has a high therapeutic to toxic ratio16, 17, 18, 19, 20 and is relatively devoid of negative inotropic effects.21

Importance 

To date, there have been 7 published clinical trials that have studied the effects of magnesium sulfate therapy in patients with supraventricular tachyarrhythmias, including patients in rapid atrial fibrillation.22, 23, 24, 25, 26, 27, 28 Each of the studies concluded that magnesium sulfate reduced ventricular response rates with minimal adverse sequelae. All but 1 concluded that the rate reductions were clinically significant.27 The rate of conversion to sinus rhythm in patients treated with magnesium sulfate varied considerably in these studies. Unfortunately, the sample sizes of these studies were small (ranging from 15 to 57 patients), and differences in study design make it difficult to combine their results for analysis.

The question of whether magnesium sulfate is of particular benefit in hypomagnesemic patients with rapid atrial fibrillation has been examined in only 1 small clinical study that showed its effect to be independent of magnesium status.27 Problematic in this determination is the fact that serum magnesium concentrations may not accurately reflect total body or intracellular magnesium. Patients receiving regular diuretics or patients who are chronic alcoholics may be total body magnesium deplete but may have normal serum magnesium levels.29, 30

Goals of This Investigation 

This study aimed to answer the question of whether magnesium sulfate could, within the first 2.5 hours of presentation and when added to usual care, safely and effectively reduce the ventricular response rate in adult emergency department (ED) patients with rapid atrial fibrillation. Additionally, we wished to determine whether it increased the rate of conversion to sinus rhythm.

Theoretical Model of the Problem 

Magnesium sulfate is believed to reduce the rate of impulse transmission through the atrioventricular node. An intravenous infusion might augment the effect of coadministered antiarrhythmics and more rapidly reduce the rate of transmission of atrial impulses in atrial fibrillation. Because of its relatively wide therapeutic window, it would be expected that, at appropriate dosing, it would do this with few significant adverse effects. It might also be expected not to have a significant effect on the rate of conversion to sinus rhythm.

This study aimed to examine these questions in the clinical setting of acute presentations of atrial fibrillation in the ED. Endpoints of achieving a pulse rate of less than 100 beats/min, a relative reduction in pulse rate of 15 beats/min or more, and conversion to sinus rhythm were measured throughout the period of an infusion of magnesium sulfate. Administration of the study solution was in addition to any standard therapies thought appropriate by the treating physician.

Study Design 

This was a randomized, prospective, double-blinded, placebo-controlled study. We compared the effects of magnesium sulfate and placebo on acute rate control in patients presenting to the ED with atrial fibrillation and with ventricular response rates greater than 120 beats/min. Study solutions were given in addition to any therapy the treating physician would normally consider appropriate, including the use of standard rate-reducing agents.

Setting 

The study was conducted in 2 major academic, tertiary referral EDs of a city of 1 million inhabitants with a combined adult ED census of 80,000 per year. Patients were enrolled from August 1999 to June 2002. The study was approved by the ethics committees of both participating hospitals, and all patients consented to be involved.

Selection of Participants 

Patients older than 18 years and presenting to the ED with atrial fibrillation and a ventricular response rate greater than 120 beats/min were eligible for inclusion. Exclusion criteria are listed in Figure 1. Randomization was achieved by block randomization of 50 consecutive study numbers as either solution “A” or “B” by drawing lots blindly from an equal number of marked tokens. The pharmacy department of each hospital prepared all study solutions and determined which of “A” or “B” was to be the study and which the placebo solutions. Patients were then allocated consecutive study numbers as they were enrolled. Enrollment was convenience based, at the discretion of the ED treating physician.

Interventions 

Enrolled patients were randomized to receive either 40 mEq (5 g, 20 mmol) of magnesium sulfate in 100 mL of 5% dextrose solution, with 20 mEq (2.5 g, 10 mmol) given intravenously over a 20-minute period, followed by the remaining 20 mEq (2.5 g, 10 mmol) intravenously over the next 2 hours, or an equivalent volume of 5% dextrose solution at the same rates of infusion. Patients, staff, treating physicians, and investigators were blinded to the study solution contents. Statistical analysis was also performed blinded to which arm of the study received magnesium sulfate.

Methods of Measurement 

Enrolled patients received continuous electronic vital sign monitoring, with half-hourly automatic recording of pulse rate, oxygen saturation, and noninvasive blood pressure. Before infusion of the study solution, an ECG was performed, and blood was drawn for serum creatinine and electrolyte concentrations, including magnesium and calcium.

The treating physicians and nurses recorded data on a preformatted data collection sheet and in the case records. Where data were incompletely recorded on the study data sheet, the principal investigator at each site reviewed the case records to extract results. Predetermined validity of data extracted from the case record included pulse rate data if recorded on the hospital nursing chart, conversion to sinus rhythm if documented by an ECG with a time stamp indicating when it was performed or documentation by the treating physician, chronic alcoholism if documented in the current medical history by the admitting physician, and current use of diuretics if listed by the admitting physician as “current medication.”

Data collected included duration of symptoms (<24 hours, >24 hours, or unknown); history of heart disease or alcohol abuse; current use of diuretics, digoxin, and antiarrhythmics; drug therapy administered during the trial period; pulse rates at baseline and half hourly for 150 minutes after commencement of the infusion; whether there was conversion to sinus rhythm and the time it occurred; and adverse effects and their nature. Creatinine clearance was calculated using the formula: creatinine clearance=(140−age)×weight (kg)×0.85 (if female)/serum creatinine (mmol/L)×815. Hypomagnesemia was defined as a measured serum concentration of less than 1.4 mEq/L (0.70 mmol/L).

Outcome Measures 

Outcome measures during the infusion were a pulse rate of less than 100 beats/min; mean changes in pulse rate at 30, 60, 90, 120, and 150 minutes; and conversion to sinus rhythm. A 20% absolute greater incidence (25% versus 5%) of pulse rate reduction to less than 100 beats/min and a difference in the mean pulse rate of 15 beats/min at the measured points were predetermined to be clinically significant. Using these criteria with a significance level of 0.05, 45 and 39 patients (assuming an SD of approximately 25 beats/min), respectively, would be needed in each group to achieve a power of 80%. A predetermination of clinical significance for differences in the rates of conversion to sinus rhythm and of adverse effects was not made.

Primary Data Analysis 

All analyses were performed on the basis of intention to treat. The effect of treatment on the binary outcomes “any pulse rate less than 100,” conversion to sinus rhythm, and any adverse effects was analyzed using χ2 tests, and the results were reported as relative risks (RRs) and 95% confidence intervals (CIs). Log-binomial regression was used to adjust the analyses for confounders. Log-binomial regression is a generalized linear model similar to logistic regression, except that the log of the probability is analyzed rather than the logit. Log-binomial regression was used because we wished to report results in terms of RRs rather than odds ratios.

The effect of treatment on pulse rate measured at various points was analyzed using mixed model analysis of variance (with a random intercept and slope for each subject), with contrasts to test preplanned hypotheses of interest. Confounders were added to the model as necessary. The effect of treatment on the time taken to achieve a pulse rate of less than 100 beats/min was analyzed using a log-rank, with life-table estimates of the probability of achievement. Cox's proportional hazards regression was used to adjust the previous analysis for confounding factors.

The statistics package used for analyses was the SAS version 8.2 (SAS Institute, Inc., Cary, NC).

Characteristics of Study Subjects 

One hundred ninety-nine patients were enrolled in the study. Seventeen patients were withdrawn from the study treatment, but their data were included in analyses (Table 1). There were 14 protocol violations, the majority the result of too-rapid infusion of the test solution. Three of the 14 were randomized but converted to sinus rhythm before the test solution was administered, and in 3 further cases no pulse rate or reversion status data were available. Subjects were enrolled at the discretion of the physician treating them in the ED, and the total number and characteristics of patients who were eligible for inclusion in the study but not included are unknown.

Table 1. Study withdrawals.

	Reason	No.	MgSO4/Placebo
	Creatinine clearance <30 mL/min	3	1/2
	Unknown	2	1/1
	Hypermagnesemia	1	1
	Patient or treating physician request
	Hypotension	4	3/1
	Other, minor adverse effect	5	2/3
	Other	2	1/1
	Total	17	9/8

MgSO4, Magnesium sulfate.

Baseline characteristics were similar between the randomized groups, except for duration of symptoms, whether digoxin was given, presence or likelihood of hypomagnesemia, and baseline pulse rate (Table 2). Digoxin was the most commonly coadministered antiarrhythmic drug selected by treating physicians (157 patients [78.9%]). A β-blocker was administered to 20 (10%) patients and verapamil to 6 (3%) patients. For 24 patients, the first dose of antiarrhythmic was administered more than 20 minutes after the study solution (10 having received magnesium sulfate, 14 placebo). For 11 patients, no antiarrhythmic, other than the study solution, was recorded as being administered (7 having received magnesium sulfate and 4 placebo). For 9 patients (2 having received magnesium sulfate and 7 placebo), multiple doses of antiarrhythmic were administered during the study.

Table 2. Comparison of characteristics between treatment groups.∗

	Characteristic	MgSO4	Placebo
	No. randomized	n=102	n=97
	Sex, No. (%)
	Unknown	3 (2.9)	1 (1.0)
	Male	46 (45.1)	45 (46.4)
	Female	53 (52.0)	51 (52.6)
	Age, y, median (IQR)	72.0 (60.0–80.0)	73.0 (61.0–81.0)
	History of alcohol abuse, No. (%)	8 (8.2)	10 (10.8)
	History of heart disease, No. (%)	46 (46.9)	47 (50.5)
	Currently receiving diuretic, No. (%)	13 (13.4)	25 (26.9)
	Currently receiving digoxin, No. (%)	14 (14.4)	12 (12.9)
	Currently receiving other antiarrhythmics, No. (%)	28 (27.5)	29 (29.9)
	Duration of symptoms, No. (%)
	Unknown	13 (12.7)	15 (15.5)
	<24 h	70 (68.6)	54 (55.7)
	≥24 h	19 (18.6)	28 (28.9)
	Digoxin given, No. (%)	76 (77.6)	81 (86.2)
	Verapamil given, No. (%)	2 (2.0)	4 (4.3)
	β-Blocker given, No. (%)	11 (10.8)	9 (9.3)
	Mean serum magnesium, mmol/L (SD)	0.85 (0.10)	0.88 (0.14)
	Magnesium <0.7 mmol/L, No. (%)	5 (5.7)	5 (6.0)
	Low magnesium or alcohol abuse or diuretic, No. (%)	22 (22.2)	34 (36.2)
	Baseline pulse rate, beats/min (SD)	141.6 (17.7)	143.2 (16.9)
	Protocol violation, No. (%)	7 (6.9)	7 (7.2)
	Withdrawn, No. (%)	9 (8.8)	8 (8.2)

IQR, Interquartile range.

∗ Percentages may be calculated from less than 102 or 97 due to missing data.

Main Results 

Patients who received magnesium sulfate were more likely to achieve a pulse rate of less than 100 beats/min during the study period (63 [65%] of 97 versus 32 [34%] of 93; RR 1.89; 95% CI 1.38 to 2.59; P<.0001) (Figure 2, Figure 3). Adjusting for baseline differences between the 2 groups for duration of symptoms and baseline pulse rate made little difference to the result (adjusted RR 1.74; 95% CI 1.25 to 2.41; P=.001).

View Large Image Download to PowerPoint

Figure 2. Cumulative total percentage of patients achieving a pulse rate of less than 100 beats/min from onset of study solution.

View Large Image Download to PowerPoint

Figure 3. Flow diagram of patients achieving a pulse rate less than 100 beats/min at any time during the study. PR, Pulse rate.

Subjects in the magnesium sulfate group were more likely than those in the placebo group to convert to sinus rhythm (25 [27%] of 94 versus 11 [12%] of 91; RR 2.20; 95% CI 1.15 to 4.21; P=.01) (Figure 4). Adjustment for baseline differences in duration of symptoms, hypomagnesemia, and baseline pulse rate made little difference to the result (adjusted RR 2.05; 95% CI 1.05 to 4.03; P=.04).

View Large Image Download to PowerPoint

Figure 4. Flow diagram of patients converting to sinus rhythm.

Subjects in the magnesium sulfate group were more likely than those in the placebo group to have an adverse event (14 [15%] of 95 versus 5 [5%] of 92; RR 2.71; 95% CI 1.02 to 7.23; P=.04). The majority of adverse events were considered minor, with flushing the most frequently reported (Table 3).

Table 3. Reported adverse effects.

	Adverse Effect	Nature	MgSO4	Placebo
	Minor	Flushing sensation	6
		Local pain/irritation	3
		Nausea		2
		Abdominal discomfort		2
		Headache		1
		Dizziness	1
		Hypotension (blood pressure >100 mm Hg systolic and not withdrawn)	1
		Bradycardia (transient, no intervention required)	1
	Major	Hypotension (blood pressure <100 mm Hg systolic, withdrawn from study, or other intervention)	4	1
		Bradycardia (intervention required)	1
	Total	23 Adverse effects in 21 patients

There was a statistically significant reduction in mean pulse rate in the magnesium sulfate group relative to placebo at points 30 through 150 minutes on the unadjusted analysis and the analysis adjusting for duration of symptoms, digoxin given, and hypomagnesemia (Figure 5). However, our predetermined criteria for clinical significance of a mean pulse rate of greater than or equal to 15 beats/min difference between the groups was not achieved, although it did lie within the CIs at points 120 and 150 minutes (Table 4).

View Large Image Download to PowerPoint

Figure 5. Mean pulse rate (with standard error bars) over time from commencement of the study.

Table 4. Mean pulse rate differences (beats/min) between treatment groups, placebo versus magnesium sulfate, adjusted repeated-measures analysis.

	Time Point	Mean Pulse Rate Difference, Beats/Min (95% CI)
	Change at 30 min	−5.94 (−11.66 to −0.23)
	Change at 60 min	−6.06 (−12.08 to −0.05)
	Change at 90 min	−7.79 (−14.54 to −1.05)
	Change at 120 min	−11.61 (−19.01 to −4.14)
	Change at 150 min	−10.58 (−18.87 to −2.30)

In Retrospect 

Reliance on the clinical staff treating the patient to record data proved less reliable than hoped. Having a dedicated study data recorder might have resulted in more complete results. To simplify interpretation, the effects of magnesium sulfate in conjunction with other specific antiarrhythmics would be better examined separately.

In summary, we have demonstrated that an infusion of magnesium sulfate marginally improves early ventricular response rate control and increases rates of conversion to sinus rhythm in patients in rapid atrial fibrillation when added to usual treatment. A small risk of significant hypotension and bradycardia accompanies these effects.