Statin Therapy for the Prevention of Atrial Fibrillation

A Meta-analysis of Randomized Controlled Trials

Zhongsu Wang; Yong Zhang; Mei Gao; Jiangrong Wang; Qing Wang; Xiaojun Wang; Lequn Su; Yinglong Hou

Disclosures

Pharmacotherapy. 2011;31(11):1051-1062. 

In This Article

Results

Search Results

We identified nine statins from the PubChem Compound search: atorvastatin, cerivastatin, fluvastatin, lovastatin, pravastatin, mevastatin, meglutol, rosuvastatin, and simvastatin. From the total number of articles identified (1219 from MEDLINE, 2068 from EMBASE, and 1481 from Cochrane Controlled Trials Register), 20 RCTs were deemed eligible for our meta-analysis (Figure 1).8–26

Figure 1.

Schematic of the trial selection process.

Study Characteristics

The 20 eligible RCTs included 16,203 patients randomly assigned to receive statins and 16,108 patients randomly assigned to receive a placebo or active control regimen. Of the 20 trials, atorvastatin was studied in 11, pravastatin in five, rosuvastatin in three, and simvastatin in one. The characteristics of these trials are shown in Table 1.

According to the quality assessment, all the studies followed the randomization principles. In seven trials,[11,16,19–23] atrial fibrillation was not a prespecified end point. Atrial fibrillation was diagnosed through electrocardiogram and/or 24- hour Holter monitor or pacemaker in 16 studies;[8–19,24–26] no detection method was mentioned in the other four studies.[20–23] Three trials provided no clear definition of atrial fibrillation: one trial with postoperative transitory arrhythmias or transitory atrial fibrillation as the end point,[23] another with an atrial high-rate episode of 10 minutes or longer,[25] and the third with atrial fibrillation or atrial flutter.[26]

Meta-analysis

The meta-analysis indicated that statins were effective for the prevention of atrial fibrillation (OR 0.59, 95% CI, 0.45–0.76; Table 2). However, the evidence for statistical heterogeneity was substantial (Q( df =19)=65.63, p<0.00001, I2=71%; Figure 2).

Figure 2.

Primary and secondary prevention of atrial fibrillation with statins. M-H = Mantel-Haenszel; CI = confidence interval; ATAHEB = Atorvastatin Trial for Atrial High Rate Episodes in Patients with Bradycardia; ARMYDA = Atorvastatin for Reduction of Myocardial Dysrhythmia After Cardiac Surgery; ALLHAT = Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial; MIRACL = Myocardial Ischemia Reduction with Aggressive Cholesterol Lowering.

A subgroup analysis evaluating the effects of statins on primary and secondary prevention was performed. Primary prevention meant prevention in patients with no history of atrial fibrillation, whereas prevention in patients with a known history of atrial fibrillation was defined as secondary prevention. Statins were effective for both primary prevention (OR 0.67, 95% CI 0.51–0.88) and secondary prevention (OR 0.40, 95% CI 0.20–0.83) of atrial fibrillation. However, the heterogeneity was substantial (primary prevention: Q( df =13)=38.24, p=0.0003, I2=66%; secondary prevention: Q( df =5)=22.06, p=0.0005, I2=77%) in both groups, and the difference between the two subgroups was not statistically significant (Figure 2).

According to another subgroup analysis, the effect sizes for the four drugs were as follows: OR 0.43 (95% CI 0.27–0.66) for atorvastatin (11 studies, OR 1.03 (95% CI 0.77–1.37) for pravastatin (five studies), OR 0.53 (95% CI 0.26–1.05) for rosuvastatin (three studies), and OR 0.32 (95% CI 0.01–8.25) for simvastatin (one study. These data indicated a significant benefit in the atorvastatin-treated subgroup; however, no protective effect was observed in the pravastatin subgroup. No RCTs studying the effects of the other five statins on atrial fibrillation were identified. The heterogeneity for atorvastatin and rosuvastatin was substantial (atorvastatin: Q( df =10)=28.75, p=0.001, I2=65%; rosuvastatin: Q( df =2)=7.61, p=0.02, I2=74%; Figure 3).

Figure 3.

Effect of different types of statins on the prevention of atrial fibrillation. M-H = Mantel-Haenszel; CI = confidence interval; ATAHEB = Atorvastatin Trial for Atrial High Rate Episodes in Patients with Bradycardia; ARMYDA = Atorvastatin for Reduction of Myocardial Dysrhythmia After Cardiac Surgery; ALLHAT = Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial; MIRACL = Myocardial Ischemia Reduction with Aggressive Cholesterol Lowering.

Subgroup analysis was also performed in the atorvastatin subgroup. The results showed that not only was atorvastatin effective for the prevention of atrial fibrillation (OR 0.43, 95% CI 0.27–0.66), but the benefit was more marked in the 10–40-mg subgroup (OR 0.29, 95% CI 0.19–0.45) than the intensive therapy (80 mg) subgroup (OR 0.85, 95% CI 0.62–1.16; Figure 4). Furthermore, no significant heterogeneity was detected in the subgroups (atorvastatin 10–40 mg/day: Q( df =7)=9.09, p=0.25, I2=23%; atorvastatin 80 mg/day: ( df =2)=2.01, p=0.37, I2=0%; Figure 4).

Figure 4.

Effect of different doses of atorvastatin on the prevention of atrial fibrillation. M-H = Mantel-Haenszel; CI = confidence interval; ATAHEB = Atorvastatin Trial for Atrial High Rate Episodes in Patients with Bradycardia; ARMYDA = Atorvastatin for Reduction of Myocardial Dysrhythmia After Cardiac Surgery; MIRACL = Myocardial Ischemia Reduction with Aggressive Cholesterol Lowering.

Sensitivity Analysis

A meta-analysis concerning the 16 trials[8–19,24–26] in which atrial fibrillation was diagnosed through electrocardiogram and/or 24-hour Holter monitor or pacemaker was conducted, and a similar result (OR 0.59, 95% CI 0.45–0.79) was obtained. After excluding the three studies[23,25,26] that had no clear definition of atrial fibrillation, we obtained a similar result (OR 0.56, 95% CI 0.41– 0.76). Atrial fibrillation was not a prespecified end point in seven trials;[11,16,19–23] after excluding them, however, a similar result was obtained (OR 0.56, 95% CI 0.41–0.77). These studies were also excluded when performing the subgroup analyses, and no obvious change in effect size was detected (Table 2).

When conducting the subgroup analysis evaluating the relationship between the dose and effect size in the atorvastatin subgroup, we noticed that five studies[17,18,20,22,24] involving participants undergoing cardiac surgery were included, and these patients were administered atorvastatin 20 or 40 mg/day. Since significant benefits of statins on atrial fibrillation has been observed in postoperative patients,[5,28] those five studies were excluded in the sensitivity analysis, and a significant benefit on atrial fibrillation (OR 0.13, 95% CI 0.05–0.33) could still be observed. Thus, inclusion of these studies did not influence the beneficial effect of statins in our analysis.

Publication Bias

Funnel plots and Egger's regression tests showed significant publication bias among the 20 studies (intercept = −2.104, p=0.001). Publication bias was also evaluated in the subgroups (atorvastatin, pravastatin, rosuvastatin, and simvastatin). The funnel plots were still visually asymmetrical in the atorvastatin and rosuvastatin subgroups, and Egger's regression test suggested that publication bias was significant in the atorvastatin subgroup (intercept = −2.605, p=0.030) but not in the pravastatin (intercept = −1.672, p=0.164) and rosuvastatin (intercept = −2.548, p=0.101) subgroups. However, funnels plots were virtually symmetrical, and Egger's regression test suggested no significant publication bias in both the atorvastatin 10–40- mg and 80-mg subgroups (intercept = −1.183, p=0.298; and intercept = 20.857, p=0.536, respectively; Figure 5) when we performed the third subgroup analysis.

Figure 5.

Evaluation of publication bias among all 20 randomized controlled trials (RCTs); among the pravastatin, atorvastatin, and rosuvastatin subgroups; and among the atorvastatin 10–40-mg and 80-mg subgroups. Publication bias is significant among all 20 studies and in the atorvastatin subgroup. *Publication bias was not significant in the pravastatin and rosuvastatin subgroups, and in both atorvastatin dose subgroups.

Comments

3090D553-9492-4563-8681-AD288FA52ACE
Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.

processing....