Antiarrhythmia drugs for cardiac arrest: a systemic review and meta-analysis

Introduction Antiarrhythmia agents have been used in the treatment of cardiac arrest, and we aimed to review the relevant clinical controlled trials to assess the effects of antiarrhythmics during cardiopulmonary resuscitation. Methods We searched databases including Cochrane Central Register of Controlled Trials; MEDLINE, and EMBASE. Clinical controlled trials that addressed the effects of antiarrhythmics (including amiodarone, lidocaine, magnesium, and other new potassium-channel blockers) on the outcomes of cardiac arrest were included. Data were collected independently by two authors. The risk ratio of each outcome was collected, and meta-analysis was used for data synthesis if appropriate. Heterogeneity was assessed with the χ2 test and the I2 test. Results Ten randomized controlled trials and seven observational trials were identified. Amiodarone (relative risk (RR), 0.82; 95% confidence interval (CI), 0.54 to 1.24), lidocaine (RR, 2.26; 95% CI, 0.93to 5.52), magnesium (RR, 0.82; 95% CI, 0.54 to 1.24) and nifekalant were not shown to improve the survival to hospital discharge compared with placebo, but amiodarone, lidocaine, and nifekalant were shown to be beneficial to initial resuscitation, assessed by the rate of return of spontaneous circulation and survival to hospital admission, with amiodarone being superior to lidocaine (RR, 1.28; 95% CI, 0.57 to 2.86) and nifekalant (RR, 0.50; 95% CI, 0.19 to 1.31). Bretylium and sotalol were not shown to be beneficial. Conclusions Our review suggests that when administered during resuscitation, antiarrhythmia agents might not improve the survival to hospital discharge, but they might be beneficial to initial resuscitation. This is consistent with the AHA 2010 guidelines for resuscitation and cardiovascular emergency, but more studies with good methodologic quality and large numbers of patients are still needed to make further assessment.


Introduction
Sudden cardiac arrest (SCA) is an emergency with high incidences but poor outcomes. Summary data indicate that the annual incidence of emergency medical service (EMS)-treated out-of-hospital cardiac arrest (OHCA) is about 50 to 55 per 100,000 persons in North America, and the annual incidence of in-hospital cardiac arrest (IHCA) ranges from three to six per 1,000 admissions. Fewer than 10% of them have been survived to discharge [1][2][3][4].
For the victims presenting with ventricular fibrillation (VF)/pulseless ventricular tachycardia (VT), antiarrhythmia agents are a kind of fundamental medication recommended by the resuscitation guidelines [5,6]. Some clinical trials studied the effects of antiarrhythmics on the outcomes of cardiac arrest. The number of participants in each study might be limited, and the individual studies might be different in methodologic quality, according to the study designs and methods. Several variables should be considered when evaluating the effects of resuscitation drugs, such as the scene of collapse or the quality of cardiopulmonary resuscitation (CPR). Although some antiarrhythmics have been recommended by the guidelines [5,6], evidence does not indicate that they could increase the survival from cardiac arrest, and more data are needed for the assessments of these drugs administered during CPR.
Besides the drugs recommended for routine use during CPR currently, such as amiodarone, lidocaine, and magnesium, we also supposed that the new potassiumchannel blockers (such as ibutilide and so on), which emerged as important kinds of agents for various arrhythmias, might have the potential to be used during CPR [7][8][9]. Although reviews about antiarrhythmics used during CPR were undertaken as part of the ILCOR Consensus on CPR Science in 2010, however, new studies conducted after 2010 and also some new drugs were not included, and no meta-analyses were performed to evaluate the results quantitatively. Thus, we conducted the systematic review and meta-analysis to review the current literature, assess the relevant clinical trials quantitatively and qualitatively, and provide better evidence regarding the effects of antiarrhythmia agents on the outcomes of cardiac arrest.

Study eligibility
We indentified studies according to the following criteria: the studies were randomized controlled designs or prospective/retrospective cohort designs; the studies recruited adult (older than 18 years) cardiac arrest patients (OHCA and IHCA); all arrest rhythms were included; antiarrhythmic agents were administered during advanced cardiac life support (ACLS), including amiodarone, lidocaine, magnesium, other new potassium-channel blockers, such as nifecalant, ibutilide, dofetilide, and others, such as bretylium.
The studies had reported the following outcomes: ROSC; short-term survival: survival to hospital intensive care unit (ICU) admission for OHCA patients/survival to 24 hours for IHCA patients; survival to hospital discharge; and neurologic outcomes at discharge. Neurologic outcomes at hospital discharge were measured as cerebral performance category (CPC): good recovery (defined as a CPC score of 1 or 2) and unfavorable recovery (defined as a CPC score of 3, 4, or 5).

Data sources
We searched the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE (Ovid), and EMBASE. We also screened the reference lists of relevant trials and reviews. We searched the following databases for unpublished or ongoing studies: http://www.controlled-trials. com and http://clinicaltrials.gov.

Study selection and data extraction
Two authors (Yu Huang and Qing He) independently screened all the titles and abstracts for eligibility. If we doubted whether a title or abstract should be included or excluded, then we read the full text to make a decision. The full texts were also read independently. Disagreements were solved by discussion with other authors, if any (Min Yang and Lei Zhan).
Two authors (Yu Huang and Qing He) extracted and collected data independently. The following data were abstracted: publication information mainly including first author's last name and publication year; the settings of the study; the study design; characteristics of types of the included patients and the rhythms of arrest; information about sample collecting; the regimens of drug administration; and outcomes reported. Disagreements were resolved by discussion.

Quality assessment of included studies
We assessed the methodologic quality of eligible trials by using the Risk of Bias tool recommended by the Cochrane Collaboration [10]. We assessed each trial across the following quality domains: (a) random-sequence generation; (b) allocation concealment; (c) blinding: according to special properties of CPR, we considered blinding adequate if the professional rescuers, the physicians in the hospital or the intensive care unit and the outcome assessors were blinded, regardless of the blinding of patients; (d) incomplete outcome data or loss to follow-up; and (e) selective reporting and any other potential threats to validity.
The assessment was performed by two authors (Yu Huang and Qing He) independently.

Synthesis of results
The ROSC and survival outcomes were measured and derived as risk ratios (RRs) with 95% confidence intervals (CIs). Neurologic outcomes, grouped into the two categories of good recovery and unfavorable recovery, were also measured as RR with 95% CI to adapt them for meta-analysis. We synthesized the data by metaanalysis if available. Meta-analysis was performed by using Review Manager (RevMan 5.1). Heterogeneity was tested by using the χ 2 test, and P ≤ 0.10 was considered significant. The I 2 statistic also was used (I 2 % ≤ 25% for low, 25% < I 2 % < 50% for moderate, and I 2 % ≥ 50% for high). According to the variability of practice of CPR, including the different comorbidities of the included patients, the different treatment strategies and so on, we expected that a random-effects model would be suitable for this meta-analysis.

Subgroup analysis
We planned to perform subgroup analysis of OHCA and IHCA patients when applicable.

Sensitivity analysis
We planned to perform sensitivity analysis by excluding trials with high risk of bias when applicable.

Antiarrhythmia agents versus placebo
We identified eight eligible studies that compared the effects of antiarrhythmia agents versus placebo for cardiac arrest. Five of the included studies were randomized designed trials, and the other three were retrospective observational studies [11][12][13][14][15][16][17][18]. The details of study characteristics are shown in Table 1.
One randomized controlled trial studied the use of amiodarone during CPR, and 504 patients were enrolled. Kudenchuk and colleagues performed a study conducted in urban and suburban emergency medical service (EMS) systems in the United States [11]; the CPR protocol was according to American Heart Association (AHA) guidelines published in 1982. OHCA patients (504) were enrolled, and 300 mg amiodarone or placebo was administered randomly if no resuscitation was gained after three shocks. The rescuers responsible for drug administration, advanced life support (ALS), or treatment in the emergency department (ED), and the investigators responsible for data collection and analyses, were all blinded adequately. The recipients of amiodarone were more likely to be resuscitated and survive to admission (RR, 1.27; 95% CI, 1.02 to 1.59); however, the proportion of patients who survived to discharge did not differ significantly in the two groups (RR, 1.02; 95% CI, 0.65 to 1.59). Thus although ROSC was not reported, amiodarone had significantly improved initial resuscitation compared with placebo, according to survival to hospital admission.
Another retrospective study conducted by the Helsinki EMS systems studied the use of undiluted amiodarone in the management of VF/pulseless VT resistant to three shocks [12]. No significant differences were shown in ROSC, survival to hospital admission, and discharge, but data demonstrated that the patients who received amiodarone had a more complicated prehospital course compared with those who did not receive amiodarone.
Four trials studied the effects of magnesium for cardiac arrest [13][14][15][16]; all were double-blinded randomized controlled trials. Three trials enrolled 370 OHCA patients [13][14][15]. In the case of Hassan et al. [13], the study was conducted in the county of Leicestershire, UK, and CPR was performed according to ERC guidelines published in 1992. Patients who had either VF resistant to three shocks or a second episode of VF during resuscitation were included. Magnesium sulfate (2 g) or placebo was given intravenously, and a repeated dose of 2 g magnesium sulfate might be given after six ineffective shocks. In the case of Allegra et al. [14], a multicenter prehospital study clinical trial conducted in New Jersey, USA. was found. Patients with VF refractory to three shocks were included, and magnesium sulfate (2 g) or placebo was given intravenously. In the case of Fatovich et al. [15], the study was undertaken at the ED of a large university   hospital in Perth. OHCA patients were enrolled if they were still undergoing arrest when they arrived at the ED, and the patient management followed standard ALS guidelines (5 mg MgSO 4 or 10 ml saline was given randomly at arrival at the ED). These three individual studies showed that no significant differences were found between magnesium and placebo groups in either ROSC or survival to discharge. One trial was conducted by the Duke Hospital code team, and enrolled 67 IHCA patients from intensive care or general wards [16]. The primary diagnosis included circulatory/neurologic/respiratory disorders. The resuscitation protocol was according to AHA guidelines, MgSO 4 (2-g bolus followed by an infusion of 8 g over a 24-hour period) or placebo given in a double-blinded manner. The results showed that in IHCA patients, no significant differences occurred between the magnesium and placebo groups in all outcomes.
Two retrospective studies compared lidocaine with placebo for OHCA patients with shock-resistant VF/VT [17,18]. In total, 406 patients were enrolled, and either individual study showed that lidocaine was superior to placebo in initial resuscitation but not survival to discharge.
We assessed the quality of the evidence by using the Cochrane Risk of Bias tool (see details in Table 2). In the five randomized trials, random sequence was generated by computer in the case of Allegra et al. [14], but no detailed information about the allocation concealment was found in the text. Sealed envelopes and central allocation were used for allocation in the case of Fatovich et al. [15], but no detailed information on randomized sequence generation was reported. No detailed information about random-sequence generation and allocation concealment could be found in the articles about other studies [11,13,16]. However, they are all randomized prospective designs, according to the texts, so unclear selection bias was considered [11,[13][14][15][16]. Double-blinding was performed in all trials [11,[13][14][15][16], and in the case of Kudenchuk et al. [11], the rescuers in the ED and the investigators for data collection and analyses were also blinded. Because considering the blinding of outcome assessors would not influence the analysis of ROSC or survival rate, a low risk of performing bias was considered.
In the three retrospective studies, high risk of allocation and performing bias was considered [12,17,18]. Moreover, low risk of attribution and other bias was considered in all studies.
No trials reported data of neurologic outcomes assessed with CPC, so no meta-analysis was performed. In addition, only two trials reported data of neurologic outcomes measured by Glasgow Coma Scale score in OHCA and IHCA patients [13,16], and both trials showed that magnesium had not improved neurologic outcomes at hospital discharge compared with placebo.

Amiodarone versus lidocaine
Two studies compared the effects of amiodarone and lidocaine during CPR (Table 1) [19,20]. Dorian and colleagues [19] performed a randomized study conducted under the auspices of the Toronto EMS system, and 247 patients were enrolled. OHCA patients with VF/pulseless VT refractory to defibrillation shocks were included, and the CPR was performed based on the 2000 AHA guidelines. Amiodarone (5 mg/kg) or lidocaine (1.5 mg/kg) was administered in a randomized, double-blinded manner. If VF persisted after a further shock, a second dose was given (1.5 mg/kg lidocaine; 2.5 mg/kg amiodarone), and resuscitation was continued. The rate of survival to admission was significantly higher in the amiodarone group compared with the lidocaine group (RR, 1.90; 95% CI, 1.16 to 3.11), however, no significant difference was found in the survival to discharge between the two groups (RR, 1.67; 95% CI, 0.57 to 4.88). Rea and colleagues [20] retrospectively reviewed the data of IHCA (pulseless VT/VF) from three large academic medical centers. Cardiac arrest was caused by circulatory or respiratory diseases. All CPR treatments and drug doses were according to 2000 AHA guidelines. The results showed that no difference was found in the proportion of patients alive 24 hours after arrest and at discharge between the two groups, the survival to ICU admission was reported. However, amiodarone was administered 8 minutes later on average, compared with lidocaine.
In Dorian et al. [19], adequate allocation concealment and binding was considered, according to the article, but no information about randomized-sequence generation was reported, so unclear risk of selection bias was considered. For the retrospective study [20], high risk of selection, performing, and detection bias was considered. Details are shown in Table 2.
Both of the two studies reported survival to discharge and survival to admission. The pooled results showed that no significant differences in survival to discharge  Figure 5).

Nifekalant versus amiodarone/lidocaine
In our review, we attempted to study the use of new potassium-channel blockers during CPR, and only four trials that studied the effects of nifekalant were identified (Table 1) [21][22][23][24]. Amino and colleagues [21] compared nifekalant versus amiodarone for OHCA patients (shock-resistant VF) who were transferred to the EMS of Tokai University [21]. Thirty patients were enrolled, and the study drugs were given randomly; amiodarone was shown to be superior to nifekalant in all outcomes ( Figure 6). Shiga and colleagues [22] conducted a prospective observational study to compare the effects of nifekalant versus lidocaine for IHCA (shock-resistant VF/VT) [22]. Tahara and colleagues [23] performed a retrospective, historical controlled study that enrolled OHCA patients with shock-resistant VF. Both studies showed that nifekalant significantly improved ROSC compared with lidocaine but not the live discharge. Additionally, Igarashi and colleagues [24] conducted a controlled study with only 22 patients to compare nifekalant and lidocaine for OHCA patients with VF. Nifekalant (0.2 to 0.4 ml/kg) or lidocaine (1 to 2 mg/kg) was given respectively after arrival at the ED and confirmation of VF. All patients died before discharge, but nifekalant was shown to improve the return of sinus rhythms compared with lidocaine.
In the assessments of methodologic qualities (Table 2), unclear risk of selection bias was considered for the randomized trial, but high risk of selection and performing bias was considered for the other two observational studies [21][22][23]. In the case of Igarashi et al., because no detailed information about study design was found in the text, a high risk of bias was considered for this study [24]. The overall pooled results showed that nifekalant significantly improved initial resuscitation compared with lidocaine ( Figure 7), but no significant difference was found in the live discharge between the two groups (χ 2 = 0.59; P = 0.44; I 2 = 0; RR, 1.40; 95% CI, 0.88 to 2.25).

Others
One randomized trial enrolled 91 out-of-hospital patients with VF refractory to four countershocks and compared the effect of bretylium with lidocaine [25]. Bretylium was not shown to have better effect than lidocaine on survival to hospital admission (RR, 1.01; 95% CI, 0.48 to 2.15) and survival to hospital discharge (RR, 0.45; 95% CI, 0.09 to 2.18). Another randomized double-blinded trial compared the effect of bretylium with placebo [26]. Bretylium was not shown to improve In a randomized double-blinded trial of sotalol versus lignocaine in out-of-hospital VF patients [27], 129 patients with VF refractory to four shocks were enrolled, and the results showed that sotalol was not superior to lignocaine for improving survival to hospital admission (RR, 0.60; 95% CI, 0.33 to 1.10), survival to hospital discharge (RR, 0.33; 95% CI, 0.07 to 1.52), and neurologic outcome (RR, 0.13; 95% CI, 0.01 to 2.32).

Subgroup analysis
In the comparison of magnesium with placebo and amiodarone with lidocaine, both OHCA and patients were included, so we performed subgroup analysis of OHCA and IHCA patients.
In the comparison of amiodarone with lidocaine, one double-blind, randomized controlled trial with 347 patients showed that amiodarone improved the survival to ICU admission (RR, 1.90; 95% CI, 1.16 to 3.11) in OHCA patients compared with lidocaine [19]. However, in IHCA patients, the other retrospective study that enrolled 118 patients, showed no significant difference in survival to 24 hours between the two treatments (RR, 0.91; 95% CI, 0.69 to 1.18) [20]. No significant difference was found between the effects of amiodarone and lidocaine on survival to hospital discharge (RR, 0.99; 95% CI, 0.54 to 1.82) in both OHCA and IHCA patients [19,20].

Sensitivity analysis
In the comparisons of amiodarone with placebo and amiodarone with lidocaine, only one trial could be identified if the other trials with high risk of bias were excluded [11,19]. In the comparison of amiodarone with placebo, amiodarone was shown to improve ROSC rate   and survival to admission (RR, 1.27; 95% CI, 1.02 to 1.59) compared with placebo; however, the survival to discharge did not differ significantly in the two groups (RR, 1.02; 95% CI, 0.65 to 1.59) [11]. These results were consistent with those synthesized before exclusion of the retrospective designed study, which was considered to be with high risk of bias.
In the comparison of amiodarone with lidocaine, when the retrospective designed study with high risk of bias was excluded, only one randomized blinded trial was included, and no significant difference was found in the survival to discharge between the two groups (RR, 1.67; 95% CI, 0.57 to 4.88), but the survival to admission was significantly better in the amiodarone group compared with the lidocaine group (RR, 1.90; 95% CI, 1.16 to 3.11) [19]. This was not consistent with the totally pooled results. This result might come from the design of two studies. In the retrospective designed study, no randomization and blinding were performed, so high risk of selection bias and detection bias were considered. However, statistical heterogeneity (I 2 = 88%) and clinical heterogeneity (the retrospective designed study enrolled IHCA patients) were found.

Discussion
In our review, we included all adult nontraumatic patients. The included studies were all conducted in urban or suburban settings, and the medical treatment was performed by European/American/Asian EMS systems or EDs, respectively. The CPR treatment during ALS was according to AHA or ERC guidelines published in 2000 or before. The majority of participants' rhythms were presented as VF/VT.
In the comparison of magnesium versus placebo, four studies with good methodologic quality were included [13][14][15][16], and no significant differences could be found in ROSC, survival to admission, and survival to hospital discharge between the two treatments. In subgroups of OHCA and IHCA patients, no significant differences were found in all outcomes. Although two trials included few patients with other rhythms, we did not perform subgroup analysis for these rhythms because of such small sample size [15,16]. Thus, our results demonstrate that magnesium has no significant benefits when used for CPR, for either OHCA or IHCA.
In the assessments of the effects of amiodarone, no improvement of survival to admission was found in ROSC, survival to admission, and survival to hospital discharge when treated with amiodarone according to the pooled results. But when excluding a retrospective observational study, one randomized trial that had enrolled 504 patients and was considered to have good quality showed that amiodarone had improved the survival to admission but not the survival to discharge, compared with placebo [11]. All high statistical (I 2 = 86%), methodologic, and clinical heterogeneity was concerned between the two studies included in the synthesis. The heterogeneity derived from the retrospective study with high risk of bias [12], and in this study, the prehospital condition was more severe in the amiodarone group [20]. So the pooled results of survival to admission must be interpreted with caution. According to the high-quality study [11], amiodarone is still likely to be beneficial to initial resuscitation, although the effect size is rather small (RR, 1.27; 95% CI, 1.02 to 1.59).
We also identified two studies that had compared amiodarone with lidocaine. The pooled results showed that no significant differences could be found in the survival to hospital discharge and survival to admission/ 24 hours between the two study groups. However, one good-quality study indicated that amiodarone was superior to lidocaine in initial resuscitation (evaluated by the survival to admission) [19]. This was inconsistent with the results reported from the other retrospective designed study and also the pooled result [20]. According to sensitivity analysis, the pooled results did not seem to be robust. However, the retrospective observational study included IHCA patients, whereas the RCT included OHCA patients, and it was considered to be with high risk of bias. Moreover, in the observational study, amiodarone was administered later than lidocaine, and it was not administered with the recommended doses to all the patients. So it is suggested that the reduced demonstrated effects of amiodarone on survival to 24 hours in this study is due to such clinical heterogeneity. Significant statistical heterogeneity (I 2 = 88%) was also considered between these two studies. Thus, in the comparison of amiodarone with lidocaine, the pooled result of the survival to admission still must be interpreted with caution. We generally suggest the use of amiodarone instead of lidocaine, according to the findings of the randomized trial.
Two studies with high risk of bias compared lidocaine with placebo [17,18], and showed that no significant differences were found in those alive at discharge between the two groups. Although it is indicated that lidocaine has the potential to improve initial resuscitation compared with placebo, the quality of evidence is low, and we suggest that the replacement of lidocaine with amiodarone in clinical use seems to be appropriate. Moreover, we suggest that the new potassium-channel blockers might have the potential to be used during CPR. However, after our literature search, no more trials were identified, except for four trials that focused on the effects of nifekalant, and they were all conducted in Japan [21][22][23][24]. The pooled results of three low-quality trials indicated that nifekalant was superior to lidocaine in initial resuscitation but not in the survival outcomes [22][23][24]. By contrast, a randomized trial with small sample size showed that nifekalant was inferior to amiodarone in either initial resuscitation or survival outcomes [21]. Thus, no present evidence can support the use of such potassium-channel blocker instead of amiodarone. Although nifekalant may be superior to lidocaine in initial resuscitation, considering the low quality of evidence, it is still hard to suggest the use of nifekalant.
Some studies focused on the use of bretylium and sotalol, but the very limited data showed no benefits of the two drugs.
Above all, all antiarrhythmic drugs were not shown to be significantly beneficial for cardiac arrest. However, because many variables might affect the practice of resuscitation, such as the cause of arrest, the response interval of the EMS system, the chest-compression fraction, the quality of CPR, and postresuscitation treatment. Thus, regarding these issues, the retrospective design of a study might bring high risk of overall bias through the study, including allocation, performing, and detection bias. All of these might have the potential to affect the demonstration of treatment effect. Therefore, according to the high-quality studies [11,19], it is suggested that amiodarone still seems to be superior to placebo (RR, 1.27; 95% CI, 1.02 to 1.59) and lidocaine (RR, 1.90; 95% CI, 1.16 to 3.11) for improving initial resuscitation assessed by the survival to admission, although the effect sizes are not big.
For the lack of data, we only performed subgroup analysis according to OHCA and IHCA in the comparison of magnesium with placebo and amiodarone with lidocaine. Similar results were found in both subgroups in the comparison of magnesium with placebo for all outcomes. Although in the comparison of amiodarone with lidocaine, the results were not consistent in the two subgroups in the analysis of initial resuscitation, this inconsistence might derive from the nonrandomization of the observational study.
Limitations in our review include that little evidence can be found after searching the literature. In the subgroups of amiodarone versus placebo, amiodarone versus lidocaine, and nifekalant versus amiodarone, only one or two controlled trials could be found. In total, 433 patients were included in the pooled results of the subgroup of magnesium versus placebo. Therefore, the results derived from present literature were rather underpowered to appraise certain ideas according to the sample sizes.
The second limitation is related to the methodologic quality of individual evidence. Many of the included studies were observational cohort studies, and especially, retrospective design. We suggest that adequate randomization and blinding of the rescuers is important to reduce the allocation, performing, and detection bias in clinical practice of CPR; thus, high risk of bias might be considered for these individual observational studies.
The third limitation is related to the heterogeneity, especially clinical heterogeneity, which is an inherent problem in data synthesis. The emergency settings in the rescuing from cardiac arrest bring particular difficulties of access to the expected homogeneity in clinical studies. The reviewed studies included patients with varied baseline characteristics (such as the settings of collapse, basic diagnosis, bystander CPR, CPR protocols performed by professional rescuers, and different regimens of antiarrhythmics and postresuscitation care, such as hypothermia). Although no significant difference was found in baseline characters between the study groups, all these variables lead to clinical heterogeneity. Significant statistical heterogeneity was found according to the χ2 test and I 2 statistic in subgroups. These encouraged us to use a random-effect model. So according to these issues, the pooled results of the data from current evidence still need to be interpreted with caution. We assessed the data according to subgroups of OHCA and IHCA, but conversely, the sample size is rather weakened.
Our review provides some implications. To our knowledge, lack of evidence focused on medications during CPR can be found. No antiarrhythmia agents have been demonstrated to improve the survival of cardiac-arrest patients. According to studies of high quality, only amiodarone seems to be beneficial in the improvement of initial resuscitation, although the strength of existing evidence is not robust enough, according to the sample size. Our review supports the summarization of the effects of antiarrhythmics and the recommendation of medication strategies in the AHA 2010 guidelines [5,6], and we suggest the use of amiodarone in patients with refractory VF/pulseless VT. A previous review provided a similar implication. Compared with the review, we performed a more up-to-date literature search; we also used an optimized tool to assess the quality of evidence, and we synthesized data by meta-analysis.
According to the review, a major issue of research in resuscitation drugs is the lack of clinical data. It is important to develop randomized prospective studies with large sample sizes to assess the effects of antiarrhythmia agents during CPR. According to our review, nonrandomization and nonblinding design seem to bring a high risk of selection bias and performing bias, which might affect the study results. Larabee et al. [28] suggested that developing prospective regional/national systems for analysis of cardiac arrest outcomes would be help to capture an appropriate sample size to examine the medications and intended outcomes.
We suggest that it is important for the articles to report details about the randomization or blinding methods (such as allocation concealment) in resuscitation studies, because in the research into resuscitation, the details of randomization and blinding are important in the assessment of the quality of evidence. Additionally, no new drugs seem to be valuable in resuscitation currently. Studies have not applied new antiarrhythmics to resuscitation, such as other potassium-channel blockers, but it is possible that these agents (such as ibutilide, dofetilide) may have the potential to be used during CPR, according to the pharmacologic mechanisms. Further studies focusing on these agents might also be expected.

Conclusions
In summary, no current drugs have been shown to improve the survival in cardiac-arrest patients. Only amiodarone may improve initial resuscitation, and no new drugs can be recommended currently. A major issue in resuscitation research is that relevant studies focused on resuscitation drugs are very limited, especially high-quality studies.

Key messages
An up-to-date systemic review and meta-analysis have found that according to present literature, only limited data about the use of antiarrhythmic drugs for cardiac arrest can be found, and the quality of evidence is rather low. No antiarrhythmic drugs have been shown to improve the survival of cardiac-arrest patients, and only amiodarone is suggested to be used for improving initial resuscitation by our review. More clinical studies focused on resuscitation drugs are expected and warranted, especially studies with good quality and focused on new types of antiarrhythmic drugs.