The Effect of Therapeutic Hypothermia after Cardiac Arrest on the Neurological Outcome and Survival—A Systematic Review of RCTs Published between 2016 and 2020

Therapeutic hypothermia is a treatment used for patients who have suffered cardiorespiratory arrest and remain conscious after the recovery of spontaneous circulation. However, its effectiveness is controversial. The objective of this systematic review is to summarize the scientific evidence available about the effect of therapeutic hypothermia on neurological status and survival in this type of patients. Methodology: A primary search in CINAHL, CUIDEN, Pubmed, Web of Science, and Scopus databases was carried out. Randomized clinical trials (RCT) published from 2016 to 2020 were selected. Results: 17 studies were selected for inclusion and most relevant data were extracted. Methodological quality was assessed by the RoB tool. Conclusions: Although therapeutic hypothermia is a safe technique with few adverse and manageable effects, it has not shown to improve survival rate and neurological status of adult nor pediatric patients. It is possible that its positive effect on neuroprotection could be achieved only by preventing hyperthermia although further investigation is needed.


Introduction
Cardiorespiratory arrest (CA) is one of the main causes of death in developed and developing countries. CA is defined as the cardiac failure caused by the cessation of mechanic activity, which is confirmed by poor circulation. The cardiopulmonary resuscitation (CPR) procedure aims to restore circulation by ensuring ventilation and gas exchange through CPR and defibrillation, among other emergency techniques [1].
Worldwide out-of-hospital cardiorespiratory arrest (OHCA) rate is 83.7 cases per 100,000 per year. In adults, the incidence increases to 95.9 cases per 100,000 annually. These numbers vary between countries and are higher in North America (98.1) or Australia (112.9) in comparison with Asia (52.5). Annually, 300,000 OHCA take place in North America and 275,000 approximately in Europe. Most of them occur at home (66%), followed by workplaces and public spaces (20%). The average age of patients that suffer a CA is 71 years, being more frequent among men than women [2,3].

Materials and Methods
A review of TTM studies was carried out to address our scientific question according to Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA). The protocol is registered in PROSPERO coded as CRD42020207405 and available at https: //www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42020207405 (accessed on 6 November 2021).
Randomized clinical trials (RCTs) relating to TTM and temperature control interventions in patients who suffer a CA and its influence on survival and neurological status were included.
CINAHL, CUIDEN, Scopus, Cochrane, Pubmed, and Web of Science databases and Dialnet journal were consulted. The search was limited to articles written in English or Spanish, and only articles published in the last 5 years were included.
Our electronic search identified 606 potentially relevant studies. Once duplicates were removed (N = 260), 211 studies were selected as potentially relevant and reviewed by title and abstract. Seventeen studies were finally included according to the inclusion criteria for full reading and data extraction [17] (Figure 1).
In order to assess the methodological quality of the selected studies, we assessed the risk of bias in all included studies in duplicate by two independent reviewers (B.R.G. and J.C.S.G.) using the Cochrane risk of bias assessment tool. The following items were assessed: randomization and sequence generation, allocation concealment, blinding and performance, outcome assessment, completeness of outcome data, and selective outcome reporting. Unblinded studies were not penalized in the risk of bias assessment due to the nature of the treatments that makes blinding non-feasible [18].
The information was collected and divided into two groups: adult and pediatric patients. Studies were examined in depth, and the most relevant data were extracted: title, authorship, year of publication, sample size, main objective, treatment in intervention, and control groups and outcomes.  In order to assess the methodological quality of the selected studies, we assessed the risk of bias in all included studies in duplicate by two independent reviewers (B.R.G. and J.C.S.G.) using the Cochrane risk of bias assessment tool. The following items were assessed: randomization and sequence generation, allocation concealment, blinding and performance, outcome assessment, completeness of outcome data, and selective outcome reporting. Unblinded studies were not penalized in the risk of bias assessment due to the nature of the treatments that makes blinding non-feasible [18].
The information was collected and divided into two groups: adult and pediatric patients. Studies were examined in depth, and the most relevant data were extracted: title, authorship, year of publication, sample size, main objective, treatment in intervention, and control groups and outcomes.

Results
In total, 17 randomized trials reporting on 5813 adults and 712 children were included; Tables 1 and 2 show the characteristics of the RCTs carried out in adult and pediatric  patients, respectively. Tables 3 and 4 summarize the main results in adult and  The time after which the survival rate was analyzed was very heterogeneous. It was determined after hospital discharge in three studies Bernard et al. [18], Look et al. [19] and Scales et al. [20] after one month of follow-up in Meert et al. [21], after three months of follow-up in Lascarrou et al. [22], Lopez-de-Sa et al. [23] and Nordberg et al. [24], after six months of follow-up in Cronberg et al. [25], Kirkegaard et al. [26], Lilja et al. [27], Pang et al. [28] and Fink et al. [29], and after one year after suffering CA in Maynard et al. [30], Moler et al. [31] Moler et al. [32], Silverstein et al. [33]. The quality of included studies was overall moderate, with some studies demonstrating a high risk of bias ( Figure 2). One study had a high risk of bias for randomization, and 10 (10/17, 58.82%) had some concerns. Two studies had a high risk of bias for randomization, and six had some concerns. One study had a high risk of bias for allocation concealment. Outcomes assessment (i.e., attrition) was judged to have a high risk of bias in two studies and was inadequate in three studies but good in 12 studies (12/17 70.58%). Thirteen studies had a low risk of bias for detection (i.e., selective reporting), and 10 had some concerns in outcomes reporting (i.e., incomplete data).   To analyze survival rate after hospital discharge and neurological status in survivors of CA after undergoing intravascular and surface cooling devices for TTM.
Patients were randomized to intravascular and surface cooling devices for TTM or to non-therapeutic hypothermia treatment.
Intravascular TTM showed more accurate temperature control, and non-significant higher rates of survival were found in comparison with surface cooling. To evaluate 30-day survival, TTM efficacy and serum levels of S100B protein as a prognostic biomarker.
Blood samples of 20 CA patients randomly selected from TTM of 32-34 • C and standard care groups were taken at admission and after 12 and 36 h after recovery to measure S100B levels.
TTM showed no improvement on survival and S100B serum levels.

Nordberg et al. (2019) [24] 677
To determine whether prehospital trans-nasal evaporative intra-arrest cooling improves survival with good neurologic outcomes compared with cooling initiated after hospital arrival.
Patients were randomly assigned to receive trans-nasal evaporative intra-arrest cooling or standard care. Both groups received TTM at 32 • C to 34 • C for 24 h.

Therapeutic Hypothermia in Adults
Cardiorespiratory arrest data in adults was homogeneous. CA patients were aged between 61 and 68 years. Most studies reported a higher proportion of men (74.8-85.7%) [19,[24][25][26][27] Nonetheless, in other studies, that proportion dropped to 65% [20,22]. Two studies highlighted that 60% of cases of cardiorespiratory arrest are witnessed by someone [19,20]. In contrast, in two studies, this percentage rose to 90% [22,27]. However, only 46% of CA are attended by witnesses or non-medical personnel before emergency services arrival [20]. Regarding the etiology of CA, 85-96% present a probable cardiac cause [18,19]. However, the study by Lascarrout et al. that analyzed non-shockable rhythms CA reported asphyxia as the main cause of CA in 56% of cases [22].
A total of 41% to 47% of patients showed shockable rhythm when a defibrillator was connected [19,20,24]. Cronberg et al. reported that approximately 93-94% of patients who suffered a CA and survived at least six months had showed an initial shockable rhythm [25]. In addition, Lascarrout's study showed that only 17.7% of patients who suffered a CA with an initial non-shockable rhythm were alive at 90 days [22]. Furthermore, nonshockable rhythms are associated with worse neurologic status [36]. Results from another study stated that those patients that were defibrillated by a witness of the CA increased by 67% the chance of ending with a good neurological state at discharge [23]. In this context, early defibrillation is essential since it is the most effective treatment against ventricular fibrillation and increases survival rates of CA patients [37]. A rang of 54.4-57% of the

Therapeutic Hypothermia in Adults
Cardiorespiratory arrest data in adults was homogeneous. CA patients were aged between 61 and 68 years. Most studies reported a higher proportion of men (74.8-85.7%) [19,[24][25][26][27] Nonetheless, in other studies, that proportion dropped to 65% [20,22]. Two studies highlighted that 60% of cases of cardiorespiratory arrest are witnessed by someone [19,20]. In contrast, in two studies, this percentage rose to 90% [22,27]. However, only 46% of CA are attended by witnesses or non-medical personnel before emergency services arrival [20]. Regarding the etiology of CA, 85-96% present a probable cardiac cause [18,19]. However, the study by Lascarrout et al. that analyzed non-shockable rhythms CA reported asphyxia as the main cause of CA in 56% of cases [22].
A total of 41% to 47% of patients showed shockable rhythm when a defibrillator was connected [19,20,24]. Cronberg et al. reported that approximately 93-94% of patients who suffered a CA and survived at least six months had showed an initial shockable rhythm [25]. In addition, Lascarrout's study showed that only 17.7% of patients who suffered a CA with an initial non-shockable rhythm were alive at 90 days [22]. Furthermore, non-shockable rhythms are associated with worse neurologic status [36]. Results from another study stated that those patients that were defibrillated by a witness of the CA increased by 67% the chance of ending with a good neurological state at discharge [23]. In this context, early defibrillation is essential since it is the most effective treatment against ventricular fibrillation and increases survival rates of CA patients [37]. A rang of 54.4-57% of the reported CA cases occurred at home and 38.7% in a public place [26,27]. Studies indicated that angiography procedure appears to be the most frequent post-CA intervention (56-82% of cases). TTM can be applied simultaneously with angiography, so the application of one technique should not be delayed by the other [26,27].

Targeted Temperature Management
Results from the study by Lascarrout et al. reported better neurologic status 90 days after resuscitation in CA patients treated with TH at 33 • C compared to the control group using the CPC Scale [22]. In accordance with this evidence, patients with initial nonshockable rhythms may also benefit from TH. However, in this study, patients in the control group were not excluded if they experienced fever (temperature > 38 • C). This fact could be responsible for harmful effects, thereby overestimating the beneficial effects of TTM. Hyperthermia prevention could have been enough to achieve these benefits.
In contrast, Lilja et al. compared the effectiveness of TH at 33 • C and 36 • C, concluding that no benefits on neurologic status were found according to the CPC scale [27]. Moreover, these results were in accordance with another study that compared TH at 32 • C, 33 • C, and 34 • C, and no differences in neurologic status were found [27]. It is worth mentioning that the sample size in that study was a great deal smaller, whereby results should be analyzed with caution.
Cronberg et al. concluded that there were no significant differences in any of the scales that measured cognitive function, quality of life, and neurological status in patients who had suffered a CA and had been treated with TH at 33 • C or 36 • C [25].
Therefore, the evidence appears to indicate that the prevention of hyperthermia and TTM at 36 • C are more effective than TH application. Nonetheless, it is possible that specific subgroups of patients that have suffered a CA may benefit from TH.

Adverse Events
As the Lopez-de-Sa et al. study indicated, the group assigned to TH at 32 • C showed a lower risk of respiratory infections, however, they suffered arrhythmias more frequently, specifically bradycardias [35]. Results from another RCT came to similar conclusions when patients treated with TH at 32 and 34 • C were compared. Patients assigned to the 32 • C TH group showed lower prevalence of seizures but more cases of bradycardia [38]. Nevertheless, one study concluded that bradycardia during TH intervention was associated with better outcomes at hospital discharge [39].
On the other hand, another frequent adverse effect associated with TH reported by Mérei et al. is the likelihood of developing thrombocytosis [35].
These adverse events are in line with the current evidence [16]. Additionally, it has been shown that these events have no serious health consequences, so TH is considered a safe therapy, and it can be implemented together with other interventions, such as extracorporeal life support (ECLS), as Pang et al. reported [28].
Bernard et al. concluded that patients that received TH during their CA showed higher rates of pulmonary edema (10%) than those who received standard treatment (4.5%) [19]. Thus, ice-cold saline for the induction of mild hypothermia is discouraged during CA. Another study pointed that saline infusion leads to a rise central venous pressure (CVP), compromising coronary perfusion [40].
On the contrary, Scales et al. reported no adverse effects associated to TH application five minutes after CA. Furthermore, pulmonary edema was lower in the group that received out-of-hospital TH [20]. One explanation could be that only 72% received intravenous outof-hospital TH and that the volume infused was smaller than in the Bernard et al. study.
None of the studies included in this review showed significant differences in survival rate.

Prognostic Factors
Mérei et al. observed higher rates of mortality in patients with lower Glasgow Coma scale (GCS) score at the time of admission to intensive care [35]. They concluded that neurologic status at discharge was a long-term indicator of neurologic status and survival [37]. This could mean that the chances of improving neurological status of CA patients is very low.

Quality of Life in CA Patients
Cronberg et al. reported in their study that 49.4% of CA survivors were active workers before the event, and only 30% continued working after the event [25]. Therefore, it can be concluded that surviving a CA will likely affect the daily life of these patients due to CA sequelae.

Targeted Hypothermia in Out-of-Hospital CA Patients
The devices most frequently used in the studies analyzing TH effectiveness in the out-of-hospital environment are cold saline infusion and intranasal evaporative catheter.
Bernard et al. reported that patients who received cold saline infusion during CA took more time to return to spontaneous circulation than those who underwent standardized treatment. In addition, mortality rates were higher in CA patients treated with TH during CA compared to those who received normal treatment. However, no significant survival differences were found between groups [19]. One possibility is that saline infusion TH treatment could be delaying other procedures aimed at CA reversion. These findings are in line with the study by Maynard et al., where they concluded that patients that received out-of-hospital TH showed similar outcomes in terms of neurological status to patients that were treated with TH after hospital arrival [30]. Moreover, patients who received out-of-hospital TH were more likely to also receive hospital TH according to Scales et al. Nonetheless, this fact did not lead to higher survival rates or better neurologic status [20]. In contrast, the study carried out by Garrett et al. showed that 36.5% of the CA patients that received TH recovered spontaneous circulation compared to 26.9% of those who did not receive it. This, however, did not lead to a rise in survival rate [41].
Nordberg et al. concluded that using an intranasal device during CA could reduce body temperature to the target temperature faster than standard care. However, survival rate and neurological outcomes were similar between groups in this case, too [24].
Considering the results of these studies, it seems more reasonable to initiate TH using intranasal devices rather than cold saline infusion. Moreover, these findings also indicate that TH may possibly not have the beneficial effect on patients that Holzer et al. and Bernard et al. reported in their studies back in 2002 [42,43].

Maintenance Phase
Kirkegaard et al. reported no differences in survival rates between patients treated with TH during 24 or 48 h. However, patients that showed optimal neurologic status at six months after suffering a CA was 5% higher in the 24-h TH group. Although no significant differences were found, it is not possible to completely dismiss possible beneficial effects. In addition, that study showed higher prevalence of adverse events such as hypotension in the 48-h TH group [26]. Another observational study concluded that increasing TH time led to higher rates of hemorrhage, pneumonia, and arrhythmia [44]. Therefore, a 24-h TH intervention seems more effective.

Internal and External Cooling Devices
The study by Look et al. showed that patients treated with internal devices reached higher survival rates than patients that did not receive TH. They also found that those devices allowed for a more accurate temperature control. Nonetheless, survival rates were not statistically different and the insufficient sample size ought to be taken into consideration [34].
Another observational study came to a similar conclusions, reporting no significant differences in survival rates in terms of neurological status between patients that received invasive TH in comparison to surface cooling with blankets or ice [45]. Another study also pointed out that invasive device use increases the risk of infection and bleeding [46]. Therefore, internal device use entails a higher risk of adverse events despite offering a more accurate temperature control.

Risk of Bias
In order to carry RCTs on TTM, a large enough sample size is required since TH beneficial effect rates may not be higher that 5-10%. In this context, this kind of RCTs are usually multicenter, but in some cases, the choice of certain TH parameters are left to the individual centers, which can lead to important differences between hospitals. In the cases of Kirkegaard et al. and Lascarrout et al., different cooling devices were used in the TH intervention, which could lead to biased results [22,26]. Moreover, studies like those by Mérei et al. or Scales et al. have an insufficient sample size [20,35]. On the other hand, it is not possible to interview all CA survival patients to assess neurologic status at discharge. In the study by Maynard et al., for example, the proportion of patients interviewed was only 73% [37]. Bernart et al. had to end the study when targeted temperature protocol at the hospitals participating in the trial changed from 33 • C to 36 • C. That change was imposed after TTM study publication [47]. Subsequently, lower rates of compliance of TH at 33-36 • C were reported, as were higher rates of fever and worse clinical prognosis. We assessed the risk of bias using the Cochrane risk of bias assessment tool (Higgins et al., 2011), which demonstrated moderate risk of bias in the majority of included studies [18].
As a result, Bray et al. recommended that hospitals that decide to practice TH at 36 • C should adequately sedate patients and consider the use of muscle relaxants in order to prevent fever onset [48].
The impossibility of blinding the TH interventions is another determinant of bias. Finally, the most commonly used scales to measure neurologic status in this kind of patients are CPC and mRS. Rittenberger et al. concluded that those instruments have not been validated, and more precise tools are needed to assess CA survivors [49].

Hypothermia in Pediatric Patients
Mean age of pediatric patients that suffered CA was between 1 and 3.2 years old [30][31][32][33]. Most of the studies showed a higher proportion of males (59-66%) [21,31,32]. However, the study by Fink et al. included 41% of males [29]. The main cause of CA was respiratory (72-88%), with this etiology being more frequent in the pediatric population than in adults. A cardiac etiology was observed in 10% of cases [29,32]. From the studied CA events, 11-38% were witnessed, and in 65% of cases, resuscitation was initiated by non-medical professionals [21,32]. Non-shockable initial rhythms were the most frequent (72-88% of cases). Shockable rhythms constituted between 3-7%, and in 6-15%, bradycardia was detected [29,31,32]. The study by Moler et al. (2017), which analyzed intra-hospital CA, found that 57% of patients presented bradycardia, 21% pulseless electrical activity (PEA), 8% asystole, and 10% ventricular fibrillation (V-fib) [32]. One hypothesis is that intrahospital CA are treated immediately, whereas in out-of-hospital Cas, the connection to a defibrillator is somewhat delayed. It should be noted that, according to the results of Moler et al., the main cause of mortality after recovery of spontaneous circulation in out-ofhospital CA is brain death or withdrawal of life support due to poor neurological prognosis in 81% of cases. That percentage, however, drops to 36% in the intra-hospital environment. Furthermore, survival rate with good neurological status was 16% in out-of-hospital CA patients and 38% hospital patients. This demonstrates how the prognosis of a CA is intimately related to the time between onset of the arrest and initiation of resuscitation measures [31,32].
According to the study carried out by Pittl et al., factors associated with survival rate in pediatric patients were cardiac etiology, ventricular fibrillation initial rhythm, shorter resuscitation, lower epinephrine dose, and weekday and daytime events [46].

Targeted Temperature Management (TTM)
In the studies by Moler et al. comparing out-of-hospital and intra-hospital TTM at 33 and 36.8 • C in CA patients, no differences were found in one-year survival rates nor neurological status [31,32]. Meert et al. came to the same conclusion when TTM at 33 • C for 48 h was applied to the intervention group in comparison to the control group, which received normothermia [21].

Adverse Events
Meert et al. reported a lower proportion of patients that required transfusion of blood products among those treated with TTM although they suffered a higher prevalence of hypoglycemias [21]. Nonetheless, these results are probably unreliable, as the sample size was not statistically valid. Furthermore, other studies with an adequate sample size showed no differences in terms of adverse events or greater use of blood products [31,32].

Maintenance Phase Duration
The results of the study by Fink et al., which compared TTM for 24 and 72 h, showed a decrease of brain damage biomarkers in the 72-h treatment group. Nevertheless, no differences in survival rates and levels of brain damage were found after six months of follow-up [29]. These findings, notwithstanding, should be taken with care since there were already differences in these biomarkers before the TTM for 24 h intervention, so its variability cannot be attributed to the treatment. In addition, this study has sampling limitations, too, so its conclusions cannot be extrapolated.

Risk of Bias
It has been observed that studies in pediatric populations have much smaller sample sizes than studies in adults, which makes it complicated to generalize their conclusions and increases the risk of bias. This fact also limits the ability to detect the beneficial effects of TTM if rates are low ( Figure 2). This review has certain limitations. On the one hand, the lack of RCTs with a sufficient sample size makes it difficult to reach reliable conclusions on the efficacy of TRM. In addition, it is a systematic review of RCTs published between 2016 and 2020, which may limit the results presented. Nevertheless, it has been observed through the results from previous years in the different databases that the largest number of publications occurred in the inclusion period, except for 2015.

Conclusions
The evidence on TTM suggest that it is a safe procedure with few and manageable adverse effects in the hospital environment. Furthermore, it can be used in conjunction with post-resuscitation procedures, such as angiography and extracorporeal life support. Regarding TTM use in adult population after CA, no differences were found in comparison with normothermia. However, it is possible that TTM could have beneficial effects on certain subgroups of patients. Studies in pediatric patients did not report significant differences between TTM and normothermia effectiveness even if applied for more than 24 h. However, some studies point that TTM could have a beneficial effect on survival rates and neurological status in both adults and children. Evidence shows that neuroprotection can likely be achieved solely by hyperthermia prevention although more RCTs and metaanalysis of RCT with greater sample sizes are needed to determine if differences exist. Regarding out-of-hospital TTM, the results are controversial since pulmonary edema is associated with intravenous cold saline infusion for hypothermia induction. Intranasal device use appears to be a safer method. Moreover, any type of out-of-hospital TTM improves the chances of hospital TTM intervention at admission. However, its effect on survival and neurological status has not been demonstrated. Finally, more accurate scales to assess neurologic status are needed in order to reduce variability and provide more precise results.