The search and selection of studies happened between July 2020 and January 2022, without language restrictions. The Medical Literature Analysis and Retrieval System Online (MEDLINE/PubMed), Scientific Electronic Library Online (SciELO), Cochrane, and Latin American and Caribbean Literature in Health Sciences (LILACS) databases were used in our search. The following descriptors and their synonyms, according to the Medical Subject Headings (MeSH), were used in the search phrase: “hypothermia, induced,” “heart arrest,” and “child.”

Item 4 of the PRISMA methodology refers to the PICO (patient/population, intervention, comparison, outcomes) strategy to guide systematic reviews, and it is presented in Chart 1 .¹¹

Aiming at the scientific value of the analyzed studies, only randomized controlled trials (RCTs) were selected. Inclusion criteria were: studies with humans that investigated the use of TH after CA in people under 18 years old (using the MEDLINE filters “infant,” “preschool child,” “child,” and “adolescent”). Other types of articles, published more than 10 years ago, with volunteers aged 18 years or older, not directly related to the subject, and duplicates of articles found in more than one database were excluded.

The articles found with the search phrase were analyzed by two independent researchers in the following stages: 1) title analysis; 2) abstract analysis; 3) full article reading when the abstract was related to the topic or when the reading was not sufficient to define its eligibility. Divergences were solved by a third independent author.

Two independent authors used the Jadad scale and the Cochrane Collaboration tool for randomized clinical trials to assess the quality of information and risk of bias.^{12 , 13} The Jadad scale assesses whether the research was adequately randomized and described as double-blind, identifying sample losses and exclusions to quantify study quality¹² . On the other hand, the Cochrane risk-of-bias tool for randomized trials assesses risks involving allocation, selection, outcome analysis, and other biases.¹³

Using our search phrase, 446 studies were identified in the databases. When applying the inclusion and exclusion criteria, 22 studies were obtained. After reading the full texts and eliminating those that were not directly related to the topic or that were secondary analyses of the studies by Moler et al.⁷ and Moler et al.,⁹ 3 studies were selected, as shown in Figure 1 .^{7 , 9 , 10}

The selected studies comprised a sample of 658 patients under 18 years old, without gender restrictions^{7 , 9 , 10} . Table 1 summarizes the general characteristics of the included publications.

The 3 selected studies assessed mortality and adverse events. Studies 1 and 2 measured survival rates and survival with favorable neurobehavioral outcomes (Vineland Adaptive Behavior Scale, Second Edition – VABS ≥ 70) 1 year after CA. In addition, the impact on neurobehavior and overall cognitive function were analyzed separately.^{7 , 9} Study 3, on the other hand, showed serum concentrations of biomarkers, unfavorable outcomes, and compared the duration of TH interventions (24 h or 72 h). Table 2 presents the main results.¹⁰

The VABS-II scale assesses the impacts of CA on neurobehavioral function by analyzing deficiencies in adaptive behavior and development, in addition to motor, communication, socialization, and daily life skills (self-care, personal hygiene, and feeding).¹⁴

In studies 1 and 2, when VABS-II could not be applied, the Pediatric Overall Performance Category (POPC) and Pediatric Cerebral Performance Category (PCPC) scores were used for sample selection^{7 , 9} . In order to be eligible, children should score 1 (good overall performance) or 2 (mild disability) on these scales. Study 3 used PCPC to identify cases of unfavorable outcomes.¹³ Both POPC and PCPC are based on the Glasgow scale¹⁵ and quantify the morbidity of patients after critical conditions. While POPC classifies functional morbidity, PCPC evaluates cognitive disorders.¹⁶

The Wechsler Abbreviated Scale of Intelligence (WASI) and the Mullen Scales of Early Learning were used to analyze global cognitive function. The WASI is applied to people aged from 6 to 89 years and investigates aspects such as verbal knowledge, visual information processing, spatial and non-verbal reasoning, and fluid and crystallized intelligence through the evaluation of vocabulary, block design, similarities, and matrix reasoning.¹⁷ The Mullen Scales are applied to children aged 68 months or older and is used to analyze data such as visual perception, expression, and receptive language, as well as fine and gross motor skills.¹⁸

This review focused only on RCTs, and studies 1 and 2 are considered references on the subj v ect.^{7 , 9} These used the same method and duration of intervention (120 h): volunteers were sedated, paralyzed, and had their central temperatures monitored at 2 sites (esophageal, rectal, or bladder) using the automatic mode of the temperature control unit. Temperature management was conducted in 2 stages: initially, central temperature was maintained at 33 ºC for 48 h using blankets, which were later reheated for 16 h or more up to 36.8 ºC.^{7 , 9}

The other RCT analyzed in this review used a similar intervention, but with different temperature management: it was kept at 33 ± 1 ºC for 24 h or 72 h, then patients were reheated by 0.5 ºC every 4 h up to 37 ºC.¹⁰

Table 3 shows the Jadad scale, used for evaluating study quality, and the Cochrane Collaboration tool, employed in the analysis of risk of bias.^{12 , 13} All studies scored a 3 (moderate quality) on the Jadad scale, as they were not described as double-blind. As for the Cochrane tool, the aspect with the highest risk of bias in the 3 studies was the blinding of participants, but other variables still presented uncertain risks.

In articles 1 and 2, mortality was evaluated after 28 days, and in study 3, after 6 months. None of the studies found a statistically significant difference between groups, thus TH in the 120-, 24-, and 72-h regimens had no significant effects on mortality in people under 18 years old after CA.^{7 , 9 , 10} It is worth mentioning that differences in confidence intervals for this variable can weaken the comparison between findings.

Hypothermia did not significantly alter the occurrence of adverse events, and its duration (24 or 72 h) did not influence this outcome. Seizures were a frequent adverse effect reported in all 3 studies. The main adverse effects are mentioned in Table 2 and these findings are compatible with those of a meta-analysis that showed that TH may be related to changes in coagulation, neurological function, and serum ion concentrations.¹

Articles 1 and 2 concluded that there was no significant difference between groups in terms of survival one year after CA,^7,9,19-25 so it is not possible to state that TH was able to increase 12-month survival.

Secondary analyses of this evidence point to some risk factors such as: children under 1 year old, serum lactate concentrations ≥ 10 mmol / L, the use of more than 4 doses of adrenaline, and chest compressions performed for more than 30 minutes.^21,34 Patients who had CA during the weekends or caused by a condition that threatens life or sudden unexpected death also had lower survival rates as a consequence of prolonged exposure to the absence of blood flow.²¹

Other risk factors for survival are preexisting health conditions and the requirement for renal replacement therapy. Renal impairment before or after CA decreases survival rates by intensifying electrolyte and blood pH changes caused by the arrest.^{24 , 25}

Protective factors are: administration of low doses of adrenaline, heart rate before CA with ventricular tachycardia and ventricular fibrillation, cardiac etiology, CA occurring during the week and the day, and in-hospital CA. Patients who had in-hospital CA were twice as likely to survive when compared to those who had out-of-hospital CA.^21,22

Articles 1 and 2 defined survival with favorable neurobehavior as scoring at least 70 points on the VASB II scale in the first year after CA. Both studies reported that there was no statistically significant difference between groups.^{7 , 9 , 19 – 22 ,24-26}

Secondary analyses suggest some risk factors: children under 1 year old or 5–12 years old, CA caused by a life-threatening condition or by sudden unexpected infant death , low blood pH and platelet count, serum lactate concentrations ≥ 10 mmol / L, and obesity.^{21 , 22}

Factors such as CA , the administration of a lower dose of adrenaline, and underweight children were listed as protective. In-hospital CA was considered a protective factor, and patients with this condition were 3 times more likely to survive with favorable neurobehavior when compared to those who had out-of-hospital arrests.^21,22 Obesity influences this variable because it compromises the precision of the concentration of drugs administered for resuscitation and the correct energy for defibrillation, which require the patient’s weight for their calculations. In addition, the quality of cardiopulmonary resuscitation is affected in obese patients.²¹

It is thus not possible to state that TH was effective in guaranteeing a good neurobehavioral outcome in 1-year CA survivors, as it is not the only factor that influences survival with favorable neurobehavior.

An adverse outcome was established by article 3 as a PCPC score of 4, 5, or 6, or an increase > 1. Eleven children in the 24-h TH group (65%) and 10 in the 72-h TH group (59%) had an adverse outcome, with no difference between groups 6 months later. It is noteworthy that S100b levels were increased in participants who underwent 24-h therapy before this change could have a direct relationship with the intervention. Since these levels are linked to brain damage , it cannot be said that the interventions were the sole responsible for the observed results.¹⁰

While studies 1 and 2 prioritized the use of VABS-II, study 3 used exclusively PCPC to analyze the impact of TH on neurobehavioral outcome, impairing the analysis of this variable.

A neurobehavioral evaluation was performed in studies 1 and 2, based on the difference between VABS-II right after CA and 1 year after the intervention.^{7 , 9} Patients with good results had a difference in VABS ≥ 15.^{9 , 23} Those who died and those who obtained the lowest possible score in VABS-II were classified as having the worst results.^{9 , 22}

The change between baseline scores and scores after 12 months and the difference of less than 15 points were similar between groups in both studies.^{9 , 19 , 22 , 27 – 30} Drowning was related to less negative impacts, while respiratory etiologies aggravated the condition.²⁹

Advanced age and high initial VABS-II were related to worse results. According to a secondary analysis, this may be due to the inability of young children to perform the activities measured by the test, causing their initial score to be reduced and the decline after 12 months to be smaller, making the deficit less evident when compared to older children.²⁹

The evidence indicates that TH was not beneficial in fighting brain damage from hypoxemia. Intrinsic factors, however, can change VABS-II results, but they do not reflect cognitive deficits themselves.

Global cognitive function was rated in studies 1 and 2 using the Early Learning Composite score from the Mullen Scales of Early Learning, the IQ scored at the WASI, and the combined Mullen and Wechsler tests . The WASI was used in participants aged 6 years or older and the Mullen scales, in children under 6 years old. The results did not differ between groups.^{7 , 9}

Secondary analyses investigated the impairment of specific skills due to ischemic injury to identify individual demands in the rehabilitation and education of children and teenagers after CA. Those most affected were: executive function, fine motor skills, visuomotor skills, and visual memory. On the other hand, the IQ did not change significantly.^{27 – 30}

Younger children do not have these areas fully developed, so deficiencies may be less noticeable, whereas in older children these changes are easily identified and closely linked to the sites that are most sensitive to hypoxic injury.²⁷ One of the evaluations by VABS-II aimed to discover which cognitive deficits the child already had before the arrest, reporting the effectiveness of the intervention more accurately.²⁷

Studies have not indicated a protective effect of TH on global cognitive function, and the analysis of the impact of this intervention was compromised by the influence of age and preexisting conditions.

In studies 1 and 2, the incidence of infection, the need for blood products, and the occurrence of severe arrhythmias 7 days after randomization were analyzed as safety measures. The incidence rates of these parameters were similar in both groups.^{7 , 10 , 19 , 20 , 22}

It can thus be inferred that TH in the 120-h regimen can be considered a safe intervention, since it did not expose patients to a higher risk of complications when compared to the TN group.

Study 3 evaluated the serum concentrations of biomarkers such as neuron specific enolase (NSE), S100b, and myelin basic protein (MBP). Three mL of blood were collected by a blinded technician for treatment and results, twice a day (every 12 h) for 4 days, and once on the seventh day after spontaneous return of circulation.¹⁰

These biomarkers assist in the prediction of therapeutic characteristics and neurological prognosis, also helping in the choice of treatment.^{10 , 31} Neuronal injuries and worse results were concluded to be related to higher serum concentrations of NSE and S100b.¹⁰

The concentrations of all biomarkers were higher in the 24-h group, except for MBP on day 7, and baseline concentrations were similar in both arms. It is worth mentioning that S100b was increased in the 24-h group in the 12-h to 24-h period, and it is not possible to state that this difference was due to the intervention.¹⁰

Researchers advise that these measurements should be validated for clinical use in pediatric CA, as they complement clinical results and help guide care. Moreover, these concentrations should be inclusion criteria in neuroprotective therapy trials or be used to optimize their duration.¹⁰

Although considered efficient in the analysis of neurological injuries, the measurement of these biomarkers has been little addressed in studies on the subject. In this context, more RCTs should be performed associating these biomarkers with clinical aspects to validate them as instruments for assessing TH on post-CA neurological outcomes.

Limitations of this study involve the analysis of studies with different methodologies and tools for assessing neurobehavioral state and the impossibility of blinding researchers.

It is worth mentioning that the benefits of TH cannot be totally denied by the observed results, as some statistical data showed confidence intervals that contemplated 1. Therefore, the intervention may or not be a protective factor, requiring further studies with larger samples and longer follow-up periods to reach more definitive results.

In addition, studies had some specific limitations: article 1 was interrupted because of the difficulty in obtaining an adequate sample size, associated with minor benefits. Furthermore, the delay in reaching the desired temperature delayed the return of circulation, which may have caused adverse effects. Study 2 postponed the evaluations of patients with hypothermia, which may cause a false idea of greater survival rates after the intervention. In article 3, a small sample size was used, and the PCPC was not blindly applied.^7,9,10