FormalPara Key Points

Despite therapeutic hypothermia in newborn infants being associated with stress and pain, there is a lack of studies investigating sedation and pain relief during this treatment.

More clinical trials are needed to determine the most effective intervention.

1 Introduction

Three to five newborns out of 1000 births are affected by peripartum asphyxia followed by subsequent moderate to severe hypoxic-ischemic encephalopathy (HIE) [1, 2]. The risk for subsequent disabilities from the asphyxia is reduced when the infant is treated with therapeutic hypothermia (TH) [3, 4]. The treatment should be initiated as soon as possible within the first 6 h after birth and then maintained for 72 h [2]. Infants undergoing TH are subjected to intensive care including painful procedures [5] as well as the TH, which is also associated with pain and stress [6, 7]. Additionally, painful experiences during the newborn period can alter future pain responses [8] and can also impair the infants’ brain development [9]. Therefore, managing infants’ pain and stress during TH is of great importance.

By reducing the body’s temperature during TH, serum clearance of morphine, fentanyl and midazolam is prolonged [10]. TH also redistributes regional blood flow, impacting both drug distribution and clearance. It has also been associated with a decreased glomerular filtration rate in animal studies and may consequently decrease renal excretion of drugs in humans [11]. The effects of sedative and analgesic treatment during TH are unclear and since TH can affect pharmacokinetics [12] caution is often recommended. In addition, infants undergoing TH might suffer from hepatic and renal injuries due to the asphyxia further impacting the way the infant will metabolize the drugs [11]. Morphine, midazolam, and dexmedetomidine are examples of drugs that might be used during TH for pain relief and/or sedation [13]. In addition to pharmacological interventions of pain management, non-pharmacological interventions could be good additional options since they have no known adverse effects and also facilitate parental involvement in infants’ care [14].

In summary, pain needs to be managed in all patients, including newborn infants, and there is strong evidence of the negative short- and long-term effects of painful procedures in this population. Pain management and sedation of infants receiving TH has not been systematically assessed for best practice. A comprehensive synthesis is needed to determine the best available evidence on pain relief and sedation in infants treated with TH.

2 Methods

2.1 Review Question

The aim of this study was to assess the benefits and harms of pharmacological and non-pharmacological interventions for the management of pain and sedation in newborn infants undergoing TH for HIE. The protocol of the review was registered in PROSPERO and submitted for publication before performing the search and data collection [15, 16]. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2 guidelines were used in preparing this article.

2.2 Study Selection

The study included randomized, quasi-randomized controlled trials (RCTs) and non-randomized studies of intervention using any type of drug or any type of non-pharmacological intervention used for the management of pain and/or sedation during TH. Crossover and cluster-randomized trials were excluded.

2.3 Primary Outcomes

The four primary outcomes were (1) analgesia and sedation assessed using validated pain scales in the neonatal population (the Echelle Douleur Inconfort Nouveau-ne [EDIN] scale, the COMFORTneo, Faces Pain Scale-revised, the Neonatal Pain, Agitation and Sedation scale [N-PASS], Pain Assessment Tool, the Astrid Lindgren and Lund Children’s Hospital’s Pain and Stress Assessment Scale for Preterm and Sick Newborn Infants [ALPS-neo], the Neonatal Facial Coding system [NFCS], and the Crying, Requires oxygen, Increased vital signs, Expression, Sleepless [CRIES] scale); (2) circulatory instability in the 12 h following the initiation of the intervention; (3) mortality to discharge; and (4) neurodevelopmental disability, defined as a composite outcome of cerebral palsy, developmental delay (Bayley Scales of Infant Development-Mental Development Index Edition II [BSID-MDIII]; Bayley Scales of Infant and Toddler Development—Edition III Cognitive Scale [BSITD-III] or Griffiths Mental Development Scale—General Cognitive Index (GCI) assessment greater than two standard deviations [SDs] below the mean); intellectual impairment (intelligence quotient [IQ] greater than two SDs below the mean); and blindness (vision less than 6/60 in both eyes) or sensorineural deafness requiring amplification [4].

2.4 Secondary Outcomes

The secondary outcomes were neonatal mortality; duration of hospital stay; days to reach full enteral feeding; analgesia assessed with neurophysiological measures such as near-infrared spectroscopy (NIRS) or galvanic skin response (GSR); focal gastrointestinal perforation; episodes of bradycardia; signs of distress, e.g. heart rate > 100 beats/min, or as reported by study authors, and each of the components of the primary outcome ‘moderate-to-severe neurodevelopmental disability’.

2.5 Search Strategy

A systematic literature search was conducted using the following databases: PubMed, Embase, CINAHLComplete, Cochrane CENTRAL, Scopus, and Web of Science. Ongoing studies were searched for in ClinicalTrials.gov (see Appendix 1 for the full search history). Studies were included regardless of language, publication date, or publication status. We checked the reference lists of the included studies. The search strategy was run on January 2021.

2.6 Data Extraction and Risk-of-Bias Assessment

Two independent researchers screened the titles and abstracts followed by full-text screening using an online tool for the preparation of systematic reviews [17]. Disagreements were solved by a third researcher or through discussion within the group. Two researchers independently performed data extraction and assessed the included studies for risk of bias. For the assessment of risk of bias in the RCTs and observational studies, the Cochrane ‘Risk of Bias’ tool [18] and ‘Risk Of Bias In Non-randomized Studies of Interventions’ (ROBINS-I) tool [19] were used, respectively. Any disagreements were resolved through discussion between the researchers.

2.7 Measures of Treatment Effect

The plan was to summarize data in a meta-analysis if they were sufficiently homogeneous, both clinically and statistically. For dichotomous data, we planned to present results using risk ratios (RRs), odds ratios (ORs) and risk differences (RDs) with 95% confidence intervals (CIs). For continuous data, we planned to use the mean difference (MD) when outcomes were measured in the same way between trials. We planned to report outcome data in tables if meta-analysis is deemed not appropriate, for example because of clinical or statistical heterogeneity.

2.8 Dealing with Missing Data, Assessment of Heterogeneity and Subgroup Analysis

Planned methods are reported in the review protocol but are not reproduced here as meta-analysis was deemed not appropriate [15, 16].

3 Results

3.1 Search Results

In total, 5145 studies were screened for title and abstract, of which 5102 were deemed irrelevant (Fig. 1). The remaining 43 studies were read in full text and 30 were excluded, leaving 10 studies to be included (one RCT and nine observational studies) (Table 1) and three ongoing studies (two RCTs and one observational study) (Table 2). Meta-analysis was deemed not feasible due to the heterogeneity among the included studies.

Fig. 1
figure 1

PRISMA flow chart

Table 1 Overview of the included studies
Table 2 Characteristics of the included ongoing studies

3.2 Non-Pharmacological Interventions

Only 1 of the 13 studies assessed a non-pharmacological intervention, i.e. maternal holding (NCT038377172019).

3.3 Pharmacological Interventions

Most of the included studies examined the use of phenobarbital or other antiepileptic drugs with primary outcomes related to seizure activity. Only one study [20] and one of the ongoing studies (NCT03177980) used a pain/sedation scale as an outcome. Sample sizes in the included studies varied between 19 and 2621; the studies were conducted in the US (n = 5), Canada (n = 1), Egypt (n = 1), The Netherlands (n = 1), Ukraine (n = 1) and the UK (n = 1). Table 3 presents the outcomes of the included studies. The RCT by Shady et al. [21] compared pentoxifylline with placebo in a group of 20 infants, while the study by Surkov and colleagues [22] (described as a randomized trial in 2019; correction in 2021 reporting that it was an observational study) compared dexmedetomidine with the control group, in which morphine, sodium oxybutyrate, and diazepam were also used. Liow et al. [23] compared the use of pre-emptive morphine with no morphine. O’Mara et al. [20] also use dexmedetomidine in their study, although there was no control group and 17/19 infants also received fentanyl as co-intervention. Favié et al. [24] included infants who were divided into three groups according to the treatment received: phenobarbital alone, phenobarbital with midazolam, or midazolam alone. Meyn et al. [25] used prophylactic phenobarbital and compared it with TH itself. Similarly, Sant’Anna et al. [26] compared phenobarbital prior to TH with TH itself. Rao et al. [27] assessed the first-line treatment of neonatal seizures comparing phenobarbital used first or levetiracetam used first. Berube et al. [28] published an analysis of 2621 infants divided into four groups: unexposed, exposed to opioids, exposed to benzodiazepines, or exposed to both opioids and benzodiazepines.

Table 3 Outcomes of the included studies

3.4 Primary Outcomes

Among the four primary outcomes of this review, five studies reported on circulatory instability and mortality to discharge, two studies reported on neurodevelopmental disability, and one study reported on pain scale. Primary and secondary outcome data are reported in Table 3.

Regarding circulatory instability, one study showed a decrease in the need for dobutamine [22], three studies found no difference in the use of vasopressors [20, 26, 27], while Berube et al. [28] showed that more infants receiving opioids alone or in combination with benzodiazepines required inotropic support than infants unexposed to those drugs.

Mortality to discharge was shown to be reduced in infants receiving treatment in three of the studies [21, 25, 28]. In the study by Rao et al. [27], infants who received phenobarbital first had a significantly higher mortality than infants who received levetiracetam first. In the study by Favié et al. [24], mortality (unclear whether the timeframe was until discharge) was reported as 28.3%, 39.7%, and 2.0% for infants receiving phenobarbital, phenobarbital+midazolam and midazolam, respectively. In the study by O’Mara [20], one patient died but it is unclear if the patient was in the dexmedetomidine and fentanyl group or in the dexmedetomidine-alone group.

3.5 Risk-of-Bias Within Studies

Details of our risk-of-bias assessments for randomized and non-randomized studies are presented in Tables 4 and 5, respectively. All of the included non-randomized studies were assessed with serious risk of bias, except one with critical risk of bias [22], due to critical bias in the classification of interventions. This study by Surkov et al., when published in 2019, was described as a randomized trial; however, in 2021 a correction was published to clarify that it was an observational study [29]. The study by O’Mara et al. [20] was assessed to have serious risk of bias due to the use of drugs that depend on the clinical stage of the infant and severity of HIE, and the relatively long period of data collection. Similarly assessed were the studies by Favié et al. [24] and Liow et al. [23], where the need for sedation and antiepileptic drugs might have been affected by the severity of HIE. The study by Sant’Anna et al. [26] was assessed with serious risk of bias because the initiation of treatment was started before the TH, that might have influenced the outcome. The study by Meyn et al. [25] had a very long data collection period and for that was assessed with serious risk of bias as the different experience of the team might have affected the outcome. For the studies by Sarkar et al. [30] and Rao et al. [27], the serious risk of bias comes from a long period of data collection followed by two different cooling methods used during the study period. The study by Berube et al. [28] was assessed to have serious risk of bias because data were collected during a long study period complicated by different practices in the participating centers.

Table 4 Risk of bias for the included randomized controlled trial
Table 5 Risk of bias for the included non-randomized studies

4 Discussion

Even though TH is an evidence-based treatment that has been available since 2006 [31], consensus and guidelines regarding appropriate analgesia and/or sedation are lacking. It is reasonable to assume the already injured asphyxiated brain would benefit from minimizing pain and stress, which are known to have negative consequences during TH, and avoiding the potential negative adverse effects from pharmacological treatments. A balanced treatment is desirable and ethically required [32, 33].

This systematic review included 10 studies on pharmacological interventions highlighting the extreme paucity of studies assessing pain and sedation management during neonatal TH. The clinical heterogenity of the included studies, e.g. different settings, comparisons, and outcomes, does not allow to draw any conclusion on the safety and efficacy of these interventions. Most of the studies examined the use of antiepileptic drugs on primary outcomes related to seizure activity; none of the included RCTs had pain relief or sedation as the primary outcome. There is a possibility that drugs primarily used for managing seizures such as phenobarbital affects infants in a way that could be interpreted as pain relieving. However, the area of use for phenobarbital is mainly antiepileptic. Certainty of the evidence is very low for all outcomes due to the imprecision of the estimates, inconsistency, and study limitations. One study [22] was listed as an RCT but after corresponding with the study author, it was clarified that it was an observational study and the manuscript was republished [29].

Previous studies have shown that TH seems to be associated with pain and stress [6, 7]; of note, this stress could even negate the beneficial effect of TH [13]. During TH in adults post cardiac arrest, sedation and analgesia are considered as essential in order to avoid shivering and achieve toleration of the treatment [32, 33]. When using opioids and other drugs during TH, there is a potential risk for adverse effects due to the effect of TH on the clearance and distribution of drugs [10, 13], which complicates dose settings. An additional challenge is posed by the lack of validated tools to assess pain during TH. Lago et al. [34] reported that there are knowledge gaps regarding both measuring and managing pain and stress during TH. We could not find other systematic reviews addressing the topic of our study. The protocol of a Cochrane review has been published, however it will include only randomized trials and a narrower list of drugs [35].

4.1 Strengths and Limitations

The strengths of this systematic review include the broad and complete search strategy, the publication of a protocol a priori, and the validated methodology to assess the included studies, e.g. Cochrane’s ‘Risk of Bias’ 2.0 tool and the ROBINS-I. We adhered to the protocol to minimize intellectual bias in conducting and reporting the findings. Screening for inclusions and risk-of-bias assessment were independently performed by two authors. Potential limitations include our broad approach, i.e. that, for example, we included all studies regardless of the type of drugs, which may have contributed to the large clinical heterogeneity of the included studies. Furthermore, our choice of the definition of outcomes could be discussed. We chose analgesia and sedation; circulatory instability; mortality to discharge; and neurodevelopmental disability as the primary outcomes while most of the included studies reported on seizures, which may be an outcome equally relevant to patients. Finally, we could not pool the included studies in any meta-analyses due to substantial heterogeneity.

4.2 Implications for Future Research

Future trials should optimize the use of tools to assess pain and report clinically relevant outcome for the infants and their families, and identify an appropriate comparator when the use of a placebo is considered unethical. Studies enrolling infants with moderate and severe HIE are needed to evaluate any differential effects of the intervention on both pain and agitation management and long-term outcomes. Observational studies might provide valuable information related to potential harms of the pharmacological interventions.

5 Conclusion

We found limited evidence to establish the benefits and harms of the interventions for the management of pain and sedation in newborn infants undergoing TH. Given the very low certainty of evidence—due to imprecision of the estimates, inconsistency and limitations in study design (all eight observational studies with overall serious risk of bias)—for all outcomes, clinical trials are required to determine the most effective interventions for the management of pain and sedation in this population.