We searched studies in PubMed, Embase, and Cochrane Library from inception through 25 Jul 2021, to identify potentially relevant studies. Language restriction was limited in English and Chinese. We also reviewed reference lists of relative articles. Details of the search strategy are provided in Supplementary Additional File S2.

After screening titles, we evaluated abstracts for relevance and identified them as included, excluded, or requiring further assessment. Studies were considered for inclusion if they focused on CA survivors during TTM and compared different NMBA strategies, including but not limited to prophylactic NMBA (continuous or scheduled), bolus if demanded or managed without NMBA (defined as the use of placebo, saline, or no use; patients are allowed to receive emergent NMBA use to control shivering episodes). We excluded studies enrolling children, pregnant women, or patients with pre-existing dementia or brain injury. Articles published in editorials, narrative reviews without data on predefined outcomes available were also excluded.

Two reviewers (L-JL and H-BH) independently extracted data from the included studies on the first author, year of publication, country, sample size, study design, disease severity, NMBA and TTM regimens, methodological quality, and all outcomes of interest. L-JL and H-BH also evaluated the quality of included studies using the risk of bias tool recommended by the Cochrane Collaboration in randomized clinical trials (RCTs) (Higgins et al., 2011) and the Newcastle-Ottawa scale for assessing the risk of bias in observational studies (Stang, 2010). Discrepancies were identified and resolved through discussion.

We aimed to explore the effectiveness and safety of NMBA strategies during TTM, including 1) with or without NMBA strategy; and 2) NMBA administration methods (i.e., continuous vs. intermittent). The primary outcomes were mortality at the longest follow-up available and the neurological outcome. Secondary outcomes included duration of MV, ICU or hospital stay, lactate clearance, time to targeted temperature, and NMBA associated complications (i.e., pneumonia).

The results from all relevant studies were combined to estimate the pooled risk ratio (RR) and associated 95% confidence intervals (CIs) for dichotomous outcomes. As to the continuous outcomes, mean differences (MD) and 95% CI were estimated as the effect results. We assessed heterogeneity using the Mantel-Haenszel χ two test and the I ² statistic (Higgins et al., 2003). An I ² < 50% was considered to indicate insignificant heterogeneity, and a fixed-effect model was used, whereas a random-effect model was used in cases of significant heterogeneity (I ² > 50%). Before data analysis, we estimated mean from median and standard deviations (SD) from IQR using the methods described in previous studies (Wan et al., 2014). We conducted subgroup analyses basing NMBA strategies. Sensitivity analyses were performed by excluding trials that potentially biased the results of primary outcomes. We also conducted sensitivity analyses for the primary outcomes by pooling only RCTs or studies focusing on targeted temperature of 32–34°C to investigate the potential affecting factors among the included studies. Publication bias was evaluated by visually inspecting funnel plots. All analyses were performed using Review Manager version 5.3.

The literature search yielded 881 records through database searching, and 12 studies with 11,317 patients who fulfilled inclusion criteria were eligible for final analysis (Jurado and Gulbis, 2011; Snider et al., 2012; Salciccioli et al., 2013; Curtis et al., 2014; Lascarrou et al., 2014; Lee et al., 2017; Stöckl et al., 2017; Lee et al., 2018; May et al., 2018; Moskowitz et al., 2020; Hifumi et al., 2021; Takiguchi et al., 2021). Additionally, in replying to the letter comment on their study (Salciccioli et al., 2013), Salciccioli et al. provided some related data Salciccioli and Donnino, (2014), which were also included in our meta-analysis. The overview of the study selection process is presented in Figure 1.

The main characteristics of the 12 included studies [3 RCTs (Stöckl et al., 2017; Moskowitz et al., 2020; Lee et al., 2018) and nine observational studies (Lascarrou et al., 2014; Takiguchi et al., 2021; Hifumi et al., 2021; Salciccioli et al., 2013; May et al., 2018; Lee et al., 2017; Jurado and Gulbis, 2011; Curtis et al., 2014; Snider et al., 2012)] are shown in Table 1. The NMBA regimens described in the included studies were presented in Supplementary Additional File S3. Of these studies, Six and five were single-center (Jurado and Gulbis, 2011; Snider et al., 2012; Curtis et al., 2014; Lascarrou et al., 2014; Lee et al., 2017; Stöckl et al., 2017) and multi-center studies (Salciccioli et al., 2013; Lee et al., 2018; May et al., 2018; Moskowitz et al., 2020; Hifumi et al., 2021), respectively, and one used data from an international cardiac arrest registry (May et al., 2018). Nine studies (Snider et al., 2012; Curtis et al., 2014; Lascarrou et al., 2014; Lee et al., 2017; Stöckl et al., 2017; Lee et al., 2018; May et al., 2018; Moskowitz et al., 2020; Hifumi et al., 2021) compared patients receiving prophylactic NMBA with the absence of NMBA regimen, and five (Jurado and Gulbis, 2011; Salciccioli et al., 2013; Lee et al., 2018; May et al., 2018; Takiguchi et al., 2021) evaluated the effects of continuous NMBA with bolus NMBA if demanded. The duration of NMBA used ranged from 24 to 37 h among studies. Most studies assessed neurological outcome based on Cerebral Performance Category score (CPC), with good outcome defined as CPC of one or two and poor outcome as CPC of 3–5. Sedation and anesthetic schemes varied across the included studies and were summarized in the Supplementary Additional File S4. Overall, the quality of the included studies was low to medium (Supplementary Additional File S5).

When comparing the prophylactic NMBA and without NMBA regimen, we found prophylactic NMBA strategy benefited more in CA survivors who received TTM in the outcomes of time to achieve target temperature and length of hospital stay. The duration of MV, serum lactate clearance after 24 h, and pneumonia incidence were similar between groups. Few studies compared continuous and intermittent NMBA regimens and showed continuous NMBA regimens had significantly longer ICU stay and shorter length of MV than intermittent NMBA regimens. (Table 3).

Our study found that NMBA is widely used in clinical practice, but there are differences in the strategies used and their associated clinical outcomes. The prophylactic NMBA strategy was mostly applied among the included studies, which is in line with a previous systematic review. That article included 68 IUCs in which NMBA were routinely used to prevent shivering in 54 ICUs while treat shivering in eight ICUs (Chamorro et al., 2010).

The 2010 AHA guidelines for cardiopulmonary resuscitation and Emergency Cardiovascular Care stated that the duration of NMBA use should be minimized, and the NMBA depth should be monitored (Peberdy et al., 2010). However, these conclusions were inferences from expert opinion and other studies but not supported by clear evidence. The statement is prompted by concerns that NMDA might mask epileptic activity and limit neurological assessment. Since then, neither the AHA nor the European Resuscitation Council recommended routine use of NMBA during TTM in their 2015 guidelines (Callaway et al., 2015). In the latest clinical practice guidelines for continuous NMBA in critically ill adult patients, the routine use of NMBA is not recommended for patients receiving TTM after CA (insufficient evidence) (Murray et al., 2016). Meanwhile, it is recommended that NMBA can be used to treat significant shivering during TTM, a weak recommendation based on a post-hoc analysis of only one prospective observational study (111 patients in total) (Salciccioli et al., 2013).

In our study, we added 11 newly published studies with a total sample size of 11,317 patients (Jurado and Gulbis, 2011; Snider et al., 2012; Salciccioli et al., 2013; Curtis et al., 2014; Lascarrou et al., 2014; Lee et al., 2017; Stöckl et al., 2017; Lee et al., 2018; May et al., 2018; Moskowitz et al., 2020; Hifumi et al., 2021; Takiguchi et al., 2021). Although high-quality RCTs are still lacking, our sample size allowed for better statistical power and different sensitivities and subgroup analyses. The results of subgroup analyses basing on various NMBA strategies also confirm our findings’ robustness. In addition, our results showed that using NMBA is safe, i.e., NMBA does not increase the length of stay, duration of MV, nosocomial infections, or muscle weakness in CA patients receiving TTM. Thus, our study partially fills a gap in the previous guidelines and provides additional evidence for clinical NMBA application.

We found the prophylactic NMBA strategy significantly improved mortality and neurological outcome in CA survivors undergoing TTM. Several explanations might contribute to our findings. First, NMBA can effectively control shivering, which interferes with achieving target temperatures by generating heat and increases metabolic activity, oxygen consumption, and cerebral metabolic stress (De Witte and Sessler, 2002; Oddo et al., 2010). Several included studies reported reductions in shivering episodes during NMBA therapy (Stöckl et al., 2017; Moskowitz et al., 2020). Moskowitz et al. found approximately 40% of patients in the usual care group develop shivering and required NMBA rescue administration, while no shivering episodes were observed in the NMBA group Moskowitz et al. (2020). In another RCT, patients were randomized to receive either a continuous NMBA or an on-demand rocuronium bromide (Stöckl et al., 2017). The authors found that 94% of patients in the on-demand NMBA group had detectable shivering episodes compared to 25% receiving continuous rocuronium (p < 0.01) (Stöckl et al., 2017). The authors noted that shivering occurred throughout the TTM period, rather than just at a specific stage during the TTM course. In addition, shivering may also be invisible, manifesting as ECG artifacts, EMG activity, or delayed achievement of the target temperature (Seder et al., 2011). Thus, the prophylactic NMBA strategy may control invisible shivering, which attenuates the neuroprotective effects of TTM. Meanwhile, we should note one important potential bias in on-demand NMBA strategy, that is, shivering is a natural thermoregulatory response of the body to lowering the core temperature, but require the relatively intact brain function (Nair and Lundbye, 2013; Hovdenes et al., 2016). Thus, patients with more severe brain injury who did not present shivering would not gain NMBA when administered “on-demand” but would have worse outcomes due to more severe brain injury, not due to lack of NMBA.

Second, our findings suggest the safety of NMBA regimens. The previous controversy over the use of NMBA was that NMBA might be associated with the risk of early-onset pneumonia and critical illness polyneuropathy (Price et al., 2012; Lascarrou et al., 2014). It also increases the duration of MV and hospital stay. However, our findings did not reveal these results. With the development of technologies such as MV weaning, percutaneous tracheotomy, and the management of ventilator-associated pneumonia and cardiopulmonary resuscitation, most ICUs have clear protocols for managing MV during TTM and the prevention and control of nosocomial pneumonia (Callaway et al., 2015). This reduces the finding of positive clinical outcomes of adverse events in the NMBA and control groups. At the same time, the included studies showed that NMBA did not increase muscle weakness during their stay in ICU (Stöckl et al., 2017; Lee et al., 2018). This favorable result can also be partly explained by the short duration of NMBA use in all these studies (approximately 24–37 h) (Jurado and Gulbis, 2011; Snider et al., 2012; Salciccioli et al., 2013; Curtis et al., 2014; Lee et al., 2017; Lee et al., 2018; Moskowitz et al., 2020; Hifumi et al., 2021; Takiguchi et al., 2021). Similarly, a recently published meta-analysis of short-term NMBA application for ARDS treatment failed to find a correlation between NMBA and acquired muscle weakness (Tarazan et al., 2020).

However, we did not find a significant improvement in lactate levels after a prophylactic NMBA strategy. Previous theories believed that improved tissue perfusion and reduced metabolic demand were possible mechanisms for decreasing lactate levels following NMBA treatment (Salciccioli et al., 2013). Some authors explain that the duration of NMBA administration in the study was inconsistent across subjects, while the serum lactate levels were obtained at regular intervals at the specified times (Lascarrou et al., 2014). On the other hand, some patients in the control group also received a temporary bolus of NMBA for shivering episodes, which reduced lactate accumulation (Lee et al., 2018; Moskowitz et al., 2020). This may have weakened the perfusion and metabolic improvement effect in the NMDA group. We also found no significant reduction in the induction time of TTM, which might be due to the advances in cooling techniques and CPR management. As shown in the most included studies, the induction time was approximately 0.5–3 h, which might reduce shivering and other adverse events during that period (Curtis et al., 2014; Lee et al., 2017; Stöckl et al., 2017; Hifumi et al., 2021). Moreover, the initial lactate levels for the enrolling patients were not so high (1.4–3.6 mmol/L), which could partially explain the lack of differences in lactate clearance between groups (Salciccioli et al., 2013; Lascarrou et al., 2014; Lee et al., 2018; Moskowitz et al., 2020).

Our study has several limitations. First, most of the included studies were retrospective, which greatly affected the causality of our study findings. Second, some included studies also recruited patients with IHCA (Lascarrou et al., 2014; Lee et al., 2017; May et al., 2018; Moskowitz et al., 2020; Takiguchi et al., 2021), who might not benefit from TTM and even had a worse prognosis (Chan et al., 2016). Therefore, the value of NMBA for patients with IHCA still needs to be further explored. Third, there was considerable heterogeneity in the TTM regimens among the included studies in terms of cooling modalities, sedation drugs, timing, and methods of shivering monitoring. For example, apart from NMBA, other strategies to prevent or control shivering involve sedative or opioid administration, often used instead if NMBA is avoided or eliminated (May et al., 2018). Deep sedation can delay extubation, ICU transfer, lead to an increased incidence of delirium or infection, confound neurological assessment, perhaps even inappropriate withdrawal of life support (Samaniego et al., 2011; Barr et al., 2013; Sandroni et al., 2014). However, all the included studies had not provided the potential impact of assessing sedation or opioid changes during NMBA used in TTM. Fourth, although we used subgroup and sensitivity analyses to explore possible confounding factors, our results may have been influenced by unmeasured confounding factors; and the sample sizes for some of the subgroup analyses were small. Meanwhile, a sensitivity analysis that included only three small RCTs did not benefit from a preventive NMB strategy over a without NMBA strategy. Fifth, the included studies spanned an extensive range of periods, during which CPR and CA guidelines have been updated several times. Sixth, some secondary outcomes need to be treated with caution. For example, most retrospective studies may not have recognized mild or moderate weakness during routine clinical care. Thus, more studies focusing on this are required in the future. Finally, the included CA patients had different underlying diseases, demographic characteristics and used different disease severity scoring criteria. However, due to the number of studies, we could not perform subgroup analyses to clarify this point further.