Benefit Absence of Therapeutic Moderate Hypothermia in the Treatment of Cardiac Arrest: A Systematic Review and Meta-Analysis

Background: Hyperthermia is frequent after cardiac arrest, and is associated with poor vital and neurological prognosis. In the last few years there have been published some studies that show benefits with moderate hypothermia in these patients, and other studies haven ́t shown such benefits. Aim: To collect all clinical trials evaluating the utility of moderate therapeutic hypothermia in survivors of a cardiac arrest. Method: A comprehensive search of clinical trials evaluating moderate hypothermia in patients who survive a cardiac arrest was carried out. The mortality and quality of life of the survivors were evaluated. The quality of the included studies, the publication bias and the heterogeneity of the results were evaluated. Results: there is no significant reduction in mortality (RR 0.97, 95% CI 0.93-1.01) or increase in quality of life (RR 1.07, 95% CI 0.94-1.21) of the patients undergoing moderate hypothermia versus those not treated with that. There are no different results in patients with cardiac arrest with defibrillable and non – defibrillable rhythms, with the different used cooling methods, or even with the induction of intra – cardiac arrest hypothermia. The mortality of these patients is high, and there are no significant differences in relation to the age or sex of them. Conclusion: In patients who survive a cardiac arrest, the induction of moderate hypothermia is not recommended.


Introduction
Cardiac arrest (CA) is defined as the cessation of heart beat of myocardial contraction [1]. In Europe, there are about 350,000-375,000 sudden CA a year outside the hospital, with estimated numbers of 17 -53 patients with CA treated by the Emergency Medical Systems / 1000,000 habitants / year [2]; the one year-survival of patients suffering from CA varies between countries, being the average is 10% [3,4]. Currently it is demonstrated that if a serie of successive actions, which can be performed by most people, called "Cardiopulmonary Resuscitation (CPR) maneuvres", included into the "Chain of Survival", could increase survival in 2-3 times; however, these maneuvers are only carried out in 1 of every 5 people who suffer CA outside the hospital; hence, the importance of education to the general population [5].
In the first 48 hours after CA, a period of hyperthermia is frequent. A relationship between post-CA hyperthermia and poor prognosis has been demonstrated [6]. It is reasonable to treat hyperthermia in unconscious patients who survive a CA. Several studies done in animals and humans show that mild hypothermia is neuroprotective and improves prognosis after García García Miguel Ángel et al.: Benefit Absence of Therapeutic Moderate Hypothermia in the Treatment of Cardiac Arrest: A Systematic Review and Meta-Analysis a period of global cerebral ischemia [7]. The promising findings of 2 clinical trials of moderate hypothermia (MH) in survivors of a CA caused by ventricular fibrillation (VF) [8,9] supported the recommendation of MH in the following CPR Guidelines of the European Resuscitation Council (ERC) [10], and even by analogy in CA situations of non-shockable rhythm. The inconclusive recent results of the Nielsen study [11] reduced the euphoria about the usefulness of this therapy, which became only a careful control of temperature and not of MH [12]. It is not clear if an intervention such as MH in CA is more or less useful depending on several factors: the observed cardiac rhythms in a CA situations; if the beginning of hypothermia at extra or intrahospital level, or even coinciding with cardiopulmonary resuscitations maneuvers; or the applied mechanism of hypothermia.
Hence the reason for this systematic review and meta-analysis: to collect all clinical trials conducted in recent years that evaluate the usefulness of MH to have a global view of its effectiveness, and perform subgroup analyzes to analyze whether this therapy may be more effective in some subgroups of patients.

Initial Objective
Our objective was to assess the efficacy of inducing MH in reducing mortality and/or increasing the survival of patients with good neurological prognosis. A board search in several dabatases was carried out, trying to find randomized controlled trials in which MH was compared with the absence of hypothermia. A comprehensive systematic review was carried out according to the recommendations concerning design quality of the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analysis) statement [13].

Included Population
Patients were adults who have had recovery of spontaneous circulation (ROSC) after CA. There are a lot of causes of CA. The study did not included patients in the context of severe trauma, a severe toxicological or metabolic disturb / drug overdose, severe hypothermia (<= 30ºC) or being pregnant women.

Intervention
The evaluated intervention was the application of MH to survivors of CA. All types of hypothermia were assessed: external -applied on skin and mucous membranes-versus internal -infusion of cold fluid or cooling of blood in extracorporeal circuit-; extrahospital vs intrahospital, even intra-CA; etc.

Search of Clinical Trials
Several electronic databases were consulted: i. Pubmed, with a broad strategy and search syntax:  Register of Controlled Trials) and OVID, with a similar strategy -(cardiopulmonary resuscitation OR heart arrest) AND (hypothermia induced)-. These databases were finally revised on May 6th, 2.018. All this information was supplemented with bibliographic references in several found systematic reviews and meta-analysis. We searched for the results of completed studies referred in ClinicalTrials.gov.
A priori there was no limits on language or date of publication.

Obtaining Results
Basic information (population, intervention, outcomes) and methodological quality (risk of bias) were obtained from the finally accepted studies in a peer review process; in aspects where there was no coincidence, an agreement was reached with the help of a third evaluator. Hypothermia was achieved with external and/or internal methods, and with adequate doses of sedation, analgesia and muscle relaxation.

Mathematical Analysis
Efficacy of MH was measured with reduction of global mortality and increased survival of patients with good neurological prognosis. This good neurological prognosis was measured with 2 scales: Pittsburgh Cerebral Performance Cathegory (CPC) of 1 (good cerebral performance: conscious, alert, able to work, might have a mild neurological or psychological déficit) or 2 (moderate cerebral disability: conscious, sufficient cerebral function for independent activities in daily life, able to work in sheltered environment) [14]; and modified Rankin scale of 0 (no symptoms), 1 (no significant disability) or 2 (slight disability) [15]. These dichotomous variables were evaluated with a relative risk (RR) (probability of patients with the event in the moderate hypothermia group / probability of patients with the event in the control group). There would be a positive effect if the RR of mortality is <1, and if the RR of patients with good neurological prognosis is > 1.
Since the heterogeneity between studies was probable, the random effects model was assumed and calculations were made using the Maentel-Haenszel model.
The quality of included studies was assessed with 2 methods: the 5 items included in the Bias Assessment Tool of the Rev Man programme: generation of randomisation sequence, concealment of randomisation sequence, patient and doctor blinding, blind assesment of outcomes and incomplete follow-up [16]; and Jadad scale, that evaluates randomisation sequence, blinding and incomplete follow-up [17].
The assessment of the publication bias was made with 2 methods: graphs (funnel-plot) and numerical, with the calculation of the number of unpublished studies with the methods of Rosenthal and Glesser-Olkin.

Employed Software
Calculations were made using the Cochrane Collaboration Review Manager 5.3 program, the J Primo EXCEL spreadsheet posted on the CASPe website (www.redcaspe.org ) and with STATA / IC v.14.2.

Results
The bibliographic search described above was carried out (figure 1). 38 clinical trials and 19 meta-analyzes were obtained. The abstract of a clinical trial of the year 2000 [18] was obtained; it was randomized, but with insufficient methodological data, so it was not included in the meta-analysis. Finally, 16 clinical trials were evaluated [8,9,11,19,32]. They are described in the Table of Characteristic of Included Studies (Appendix). The overall quality of the studies is shown in Fig. 2. Blinding of the patients and clinicians in this situation is impossible, so they can not be double blind; as an internal validity finding, in most studies blinding in the evaluation of the functional prognosis of the patient is blind. Randomization is usually well done, with the exception of one of the initial pivotal work [19] that could clearly magnify its beneficial effect. The individual evaluation of each work is shown in Figure A1     The overall effectiveness of MH in terms of mortality variation is shown in Figure 3. Few studies show a significant reduction in mortality [8,25]. The overall estimate shows no benefit with this intervention, with a RR of 0.97 and the estimation by confidence interval that exceeds the unit, and estimates of mortality in each branch of 67.7 and 68.9%. The funnel plot (fig 4) of these trials shows that there seems to be no publication bias; Neither the Rosenthal nor the Glesser-Olkin methods yield results that guide the presence of unpublished studies with non-significant results.
The separate assessment of the data of patients with CA by defibrillable and non-defibrillable rhythms shows analogous data. Figure 5 shows no benefit with MH in defibrillable rhythms, with risks of mortality in each branch of 50.6 and 50.4%. Figure 6 shows a similar result, with slightly better results, although without statistical significance non-significant (p = 0.13, RR 0.98, with 95% CI from 0.95 to 1.01), and with significantly higher mortality risks (87.3 and 89.1%).

Figure 6. Comparison of mortalithy of patients with non-shockable rhythms.
Overall results in terms of improvement in neurological prognosis also show no significant differences. Figure 7 shows a minimal non-significant trend towards improvement of the RR (1.07, with 95% CI between 0.94 and 1.21) with 27.5 and 26.6% of patients with good prognosis in each branch. In the funnel plot (fig 8) a striking asymmetry of studies is observed, which may be related to publication bias -not publication of some works without absence of neurological prognosis benefit-. However, as in the previous section on mortality, the Rosenthal and Glesser-Olkin methods provide non-significant results. Figures A2 and A3 (Appendix) show analogous results to that described for the analysis of subgroups against mortality: absence of benefit in both defibrillable and non-defibrillable rhythms, and percentages of patients with good neurological results of 50.4 and 49.3% for the assumption of defibrillable rhythms, and 12.3 and 10.9% in rhythms that cannot be defibrillated.   Assessing mortality, the cumulative meta-analysis of the 16 studies shows a striking curve ( fig 9): initial benefit, with an estimate clearly below the unit in the first studies / years, with a trend towards no benefit from the year 2010, which is maintained from 2013. The type of cooling is an element that can influence the final outcome of the patient. Figure 10 shows the subgroup analysis of the studies according to the type of used cooling method. The internal methods (mostly obtained by infusion of cold saline, at 4ºC) and the mixed, combination of internal and external methods, do not provide mortality benefit, while the aggregated estimation of the works with external cooling seems to show certain benefit (RR 0.85, with CI 95% from 0.76 to 0.95); however, this estimate is made on a small group of patients (only 10.87% of the total number of patients included in the meta-analysis). In recent years there is a greater use of intranasal cooling, although mostly in combination with other methods: the only found studies in which this method is used alone are Castren [24] (without significant difference in mortality) and Belohlavek [31] (without significant differences in neurological prognosis). No significant differences were found either in mortality -RR 1.01 (0.97 -1.04) -nor in neurological prognosis -RR 1.01 (0.75 -1.35) -in the studies that perform intra-CA cooling [30,33,34]. The evaluation of the methodological quality of the studies provides surprising data. We can use the Jadad scale, and describe the low quality studies with a score <3 points, and high quality with> = 3 points. Subgroup analysis ( Figure 11) shows no benefit in high quality studies (RR = 1) and a trend to statistical significance in studies with lower quality (RR = 0.8). The result is analogous for the assessment of neurological prognosis (figure A4). If the quality is managed as a continuous numerical variable, and a meta-regression is made, the result does not become statistically significant, but a greater slope of a line between a higher quality and an absence of effect (RR next to 1). In other words, at a lower methodological quality of the works, a greater reduction in mortality is observed ( Figure A5). There are also no notable differences if the studies are published in Core Clinical Journals (journals of greater impact and quality) than in other journals (figure A6). The influence that the sex or age of the patients can have is evaluated in figures A7 and A8 of the Appendix. Meta-regression (in which are assessed the average / median age of the patients included in each study, their effect and sample size, in relation to the RR of mortality) does not yield statistically significant data. A higher percentage of men and older populations are relationated with a greater mortality (RR> 1), with a slope of the line slightly higher for a higher age than for a higher percentage of men, although in both cases this slope seems minimally positive.
Finally, the assessment of the degree of efficiency of the technique as a function of the mortality rate in each study is assessed with the L'Abbé chart (Figure A9 Appendix). We can see most of the effect estimates on the bisector of the graph, which means that the RR of death is 1; in some studies with a risk of death of 0.4-0.7 the RR seems to be <1 (there are several circles plotted below that bisector); but the number and area of trials with RR of 1 is much higher, so the overall effect must be close to unity.

Discussion
The result of our work is conclusive. There is no benefit in survival or favorable neurological prognosis of the application of MH measures. And this finding is independent of the heart rhythm present at the CA, the cooling method used, the age or sex of the patients treated, or the mortality rate of the patients included in the studies. It should be noted that heterogeneity in terms of the methods used for cooling, the temperature sought and achieved with cooling, and the time of hypothermia (see Table of Characteristic of Included Studies, Appendix); these factors can contribute to an absence of significant results. Perhaps, one specific mode of moderate hypothermia can achieve benefits, but with this study we can´t probe this assumption and it seems very unlikely. In recent years, the tendency is starting hypothermia at the beginning of CA. The current global result is the lack of benefit of this intervention. Without being conclusive, it is striking that the degree of benefit shown by the intervention is inversely proportional to the quality of the study.
The cumulative meta-analysis curve is surprising. In other published cumulative meta-analyzes that evaluate effective therapeutic interventions [33,34], the initial estimates are close to the abscense of effect, and with the accumulation of data from successive studies the effect is outlined as significant (RR or OR appart from 1). In our graph it seems to be the opposite, with initial significative effect and final estimation without effect (RR or OR = 1). The revision of the cardiopulmonary resuscitations recommendations is carried out every 5 years, the last one in 2015. Perhaps the change in the words and in the intention of the hypothermia in the CA (we remember, not as MH, but as "careful handling of the temperature ", more in the sense of avoiding hyperthermia) could have been ahead of time.
Up to now, 19 meta-analyses have been published with the aim of evaluating MH in CA (citations A to S in Appendix). Of them, 10 (#A -J) are favorable to the application of this therapy, most of them prior to 2013, although the most representative one, which is the last revision of this topic by the Cochrane Collaboration, has been made in 2016. 3 of these works (#C, D and E) give favorable results, but cast doubts (limited data, low level of evidence). The remaining 9 (#K -S), with unfavorable result to MH, are all dated 2013 or later. The results of our work are concordant with these last meta-analyses. As it has happened a lot of times in the history of Medicine, an intervention recommended at one time was not recommended a few years later. It gives the impression that the moment of MH in CA has passed.
The most recent studies, aimed at the use of intranasal cooling devices, or with the initiation of intra-CA cooling, also show no reduction in mortality or improvement in neurological prognosis. Soon the results of several works developed in recent years will be published [35,36]; with the absence of previously demonstrated benefit, it is unlikely that these data reverse the trend of no benefit. Perhaps the future of MH in the management of CA survivors is that reflected in the latest CPR Guidelines (2015) [11] as "temperature control" only in the sense of avoiding hyperthermia.

Conclusion
In patients who survive a CA, the induction of MH doesn't get improvement in survival or neurological prognosis, therefore it isn´t recommended in these patients.

Acknowledgements
Our thanks to Isabel María Álvarez, Technical Assistant of Scientific Information of GSK (Glaxo Smith Kline) company, by her help in getting several articles; to Lucía García, a future doctor who will be a better physician than her parents; to Alberto García, for his help with the editing of the images; and, in general, to all the staff of the Intensive Care Unit of the Hospital de Sagunto, for their encouragement to carry out this work.

Competing Interest
The authors declare no conflict of interests.

Funding
There is no source of funding of this study.

Bioethics Note
This study is a systematic review and meta-analysis of clinical trials. According to the declaration of Helsinki, it is not necessary for its development to be approved by the Bioethics Committee of the Hospital de Sagunto.

Note
All web links have been visited for the last time on August 20, 2018.

Key Messages
What is already known about this subject? 1. Hyperthermia in survivors of PC seems to worsen the prognosis. 2. It is advisable the careful handling of the temperature of the patients that survive a PC What does this study add? 1. Moderate hypothermia does not seem to provide a benefit in reducing mortality or improving neurological prognosis. 2. There is no benefit in any cardiac arrest situation: rhythms defibrillable or not, different hypothermia devices, etc How might this impact on clinical practice? 1. The induction of moderate hypothermia in survivors of CA should not be mandatory. 2. It seems advisable to avoid post-cardiac arrest hyperthermia.