Post-Asphyxial Aftercare and Management of Neonates in Low- and Middle-Income Countries: A Systematic Evidence Synthesis

Birth asphyxia, also commonly termed birth complications, presents a disproportionate disease burden, with the incidence of hypoxic-ischemic encephalopathy (HIE) estimated around 2.3–26.5 per 1,000 live births in low- and middle-income countries (LMICs) compared to 1–2 per 1,000 live births in high-income countries (HICs) [1‒3]. Despite advancements in fetal monitoring, intrapartum asphyxia and consequent HIE persist as significant causes of neonatal morbidity and mortality accounting for 23–35% of global neonatal deaths [4]. Without access to neonatal resuscitation, around 60% of intrapartum-related deaths occur before neonatal encephalopathy (NE) develops. In addition, an estimated 1.2 million stillbirths annually have been attributed to intrapartum-related hypoxic events [5]. Of the annual 1 million neonatal deaths attributed to NE globally, 99% occur in LMICs with around two-thirds occurring in the African (AFR) and South-East Asian (SEA) regions of the World Health Organization (WHO) [1, 6, 7]. The severity of NE ranges from mild to severe, with moderate cases associated with a 60% chance of major deficits, including cognitive impairments and cerebral palsy (CP), and severe cases showing over 90% incidence of death or disability despite available treatments [8‒10].

NE is difficult to predict prior to onset due to its varying clinical presentation [11]. Since it is difficult to prevent through early detection, research to date has focused on minimizing subsequent brain injury and morbidities through supportive care to ensure adequate perfusion of blood and nutrients to the brain [12]. This includes adequate oxygenation, ventilation and perfusion, careful fluid management, avoidance of hypoglycemia and hyperglycemia, and treatment of seizures [13, 14]. Following resuscitation, it is crucial to continuously monitor the neonate to correct hemodynamic and pulmonary disturbances along with timely neurological assessment as part of routine care (Fig. 1). Fluid management, or restricted fluid intake, is recommended for mitigating the consequences of NE, especially cerebral edema. However, this practice has not been well studied, and to date, recommendations in neonates are based on studies conducted in older children or adults [15]. Therefore, we sought to identify any trials recruiting in LMICs that evaluated the effect of fluid restriction in newborns.

Seizures are also common in encephalopathic newborns, with a known probability of seizure occurrence post-HIE ranging from 26 to 65% [16, 17]. High seizure burden is found to be associated with adverse outcomes in neonates with encephalopathy of any severity, especially at its peak in the secondary phase of injury [17, 18] (Fig. 1). Currently, the recommended management is continuous electroencephalographic monitoring (cEEG) which represents a significant challenge for physicians in LMICs who rely on bedside clinical assessment [19, 20].

Following initial resuscitation and stabilization, the standard of care and most commonly deployed neuroprotective therapy is therapeutic hypothermia (TH), which involves controlled cooling of the body, either through selective head (SH) or whole-body (WB) hypothermia, ideally commenced within 6 h of birth for term infants with moderate to severe encephalopathy, followed by gradual rewarming to a euthermic state [21]. However, several emerging trials and reviews have suggested that TH is ineffective in significantly reducing neonatal mortality in LMIC settings [22, 23]. Concern has also been expressed about the possible differences in the nature of encephalopathy in settings where the fetus may have been subjected to prolonged and repeated episodes of intrauterine compromise [23‒25]. Due to conflicting evidence on the safety and efficacy of TH in different contexts, there is a need to comprehensively review evidence from LMICs on TH and to also explore new neuroprotective treatment strategies. Pharmacological agents, including but not limited to erythropoietin, melatonin, allopurinol, and magnesium sulfate (MgSO₄), are being investigated as monotherapy or adjuncts alongside TH to reduce free radical production and enhance neuroprotective effects [26, 27].

Early childhood development (ECD) intervention according to the WHO encompasses physical, socioemotional, cognitive, and motor development interventions implemented from birth to 8 years of age. ECD-specific interventions may benefit high-risk infant survivors of NE prone to developmental disabilities [28, 29]. Trials and reviews, especially on neonatal ECD, have been limited; hence, we sought to review current evidence on early interventions that improve developmental outcomes post-asphyxia in infants [30].

Given the disproportionately high burden of asphyxia and NE in LMIC settings, a comprehensive synthesis of the evidence on management of NE following asphyxial injury is lacking within this context. The purpose of this paper was thus to describe the findings regarding post-asphyxial aftercare strategies which have contributed to the upcoming Lancet Global Newborn Care Series and form part of a supplement describing an extensive synthesis of effective newborn interventions in LMICs. This evidence synthesis can be used to inform recommendations and clinical guidelines based on context-specific evidence for management of newborns post-asphyxia in LMICs. The following review questions were assessed:

• 1.

  Is fluid management effective in reducing morbidities and mortality associated with NE in LMICs?

• 2.

  What are the most effective antiseizure medications (ASMs) for neonates with encephalopathy in LMICs?

• 3.

  What is the effectiveness of TH in newborns with hypoxic encephalopathy in LMICs?

• 4.

  What are the effects of potential new pharmacological measures for neuroprotection in LMICs?

• 5.

  What are the effects of early postnatal interventions on cognitive developmental outcomes of asphyxiated infants in LMICs?

Given the scope of this topic and the existence of both recent and outdated systematic reviews of relevance, the present review combines different methodological approaches to synthesize evidence to assess the effectiveness of key interventions for the management and aftercare of neonates post-asphyxial injury in low-resource settings. Herein, we provide a high-level summary of topics reviewed under each methodological approach; further details of all review methods can be referenced here [31].

The initial scoping exercise could not identify recent, high-quality reviews for topics of fluid restriction and ASMs; thus, we performed de novo reviews. The protocol was registered on International prospective register of systematic reviews (PROSPERO# CRD42022382952).

Search strategies were developed using the Population Intervention, Comparison, Outcome (PICO) methodology and can be found in the online supplementary material (1.1, 1.2) (for all online suppl. material, see https://doi.org/10.1159/000541862). For both topics, the original search date was September 14, 2022. For the topic of ASMs, the search strategy was initially run using the terms “anti-epileptic therapy” and “anticonvulsants.” All searches were rerun on October 28, 2022, with the additional term – “anti-seizure medications.” Searches were conducted in the following electronic databases: MEDLINE, Embase, Cochrane CENTRAL, and CINAHL. Gray literature searches were also conducted in key organizational websites, as well as recent trials from the WHO, International Clinical Trials Registry Platform (ICTRP), and the International Standard Randomised Controlled Trial Number (ISRCTN) registry. References of relevant systematic reviews [15, 32], other reviews, and included trials were hand-searched to ensure all relevant studies were captured.

We included all randomized and quasi-randomized studies enrolling term or preterm newborns following intrapartum asphyxia with suspected or diagnosed NE in LMICs. Studies including newborns without encephalopathy were excluded if disaggregated results were not available. For fluid restriction, studies had to compare the effect of any fluid restriction to control or standard care on one of the primary or secondary outcomes. Primary outcomes include severe neurodevelopmental disability, neonatal mortality, and infant mortality. For ASMs, studies had to compare any ASM to any other ASM for the treatment of neonatal seizures on one of the following outcomes: seizure control, death, neurodevelopmental disability, or any adverse effects from the drug. For prophylactic ASMs, studies had to compare any ASM to a control group (e.g., placebo, no drug, and supportive care including TH). If we found limited eligible studies conducted in LMICs, as defined by the World Bank, we planned to leverage findings from high-income settings as prespecified in the protocol. Expert consultation was sought to determine the appropriateness of using evidence from HICs on a per-intervention basis. Detailed eligibility criteria per topic are shown in the online supplementary material (table 2).

For the topic of fluid restriction, we computed a risk ratio (RR) and 95% confidence interval (CI) for all dichotomous data using the Mantel-Haenszel (M-H) fixed-effects model. For the topic of ASMs, we grouped interventions by type of ASM and performed standard meta-analyses on dichotomous data to generate a summary RR and 95% CI. Effect estimates were calculated using the M-H random-effects model, as there were differences among included studies’ intervention, settings, and methodological characteristics. Further, data were sparse both in terms of event risks being low and the study size being small. Continuous data were analyzed using the inverse-variance model. Subgroup meta-analyses were conducted on all outcomes according to a priori defined sources of potential clinical and methodological heterogeneity: study design (e.g., randomized controlled trials [RCTs] and quasi-experimental studies), gestational age (e.g., preterm and term), type of ASM (e.g., first-line or second-line ASM), country context (e.g., LMIC and HIC), facility setting (e.g., primary, secondary, or tertiary), dosage of ASM administered (e.g., 20 mg/kg, 40 mg/kg or 60 mg/kg), duration of administration (e.g., <2 weeks and >/ = 2 weeks), and by route of administration (e.g., oral formulation of drug via nasogastric tube and intravenous (IV) formulation).

We updated existing reviews captured during the initial scoping exercise if the search was conducted prior to 2020 by conducting a systematic literature search to identify new trials per our review methodology [31].

The original search syntax reported by the Cochrane review on allopurinol [33] was recreated and run in the same databases (online suppl. material 1.3). A custom date range from April 2012, which was the original review’s last search date, to September 2023 was applied. Additionally, we snowballed reference lists of included trials and searched gray literature sources (e.g., OpenGrey, ProQuest, and Google Scholar) by adapting the original search syntax.

The original review’s [33] eligibility criteria were applied to our search selection, and our inclusion criteria were limited to trials conducted in LMICs, as defined by the current World Bank classification (see online supplementary material [table 2] for detailed eligibility criteria).

We used the standard methods of the Cochrane Neonatal Review Group as specified in the original review [33] and extracted data from included trials. Trials which were multicenter were included if data were disaggregated by country income classification. We summarized extracted data in tabular form, calculating pooled effects using the M-H fixed-effects model.

Recent systematic reviews on TH [34], erythropoietin [35], MgSO₄ [36], melatonin [37], and ECD interventions [38] were identified by our Technical Advisory Group (TAG) of experts in newborn asphyxia; thus, for these topics, we have extracted trial data as-is from studies that were conducted in LMICs without an updated search [31]. A pragmatic decision was taken to include additional trials on WB cooling conducted in LMICs that were known to our TAG members but had not been captured by the previous reviews since they were published recently. This reanalysis is a secondary analysis of five existing systematic reviews and meta-analyses.

All studies included in the original reviews conducted in LMICs were part of the reanalysis. We searched lists of ongoing trials if they were provided by the review to identify any completed trials with reported results (see online supplementary material 1.4–1.8 for search dates and list of electronic databases per review).

The source reviews were inclusive of studies conducted in all settings. While we adhered to the criteria specified in the original reviews, the present study is focused on low-resource settings; thus, we only included trials conducted in LMICs. Detailed eligibility criteria for each review adapted for use in this study are shown in the online supplementary material (table 2).

We extracted all key outcomes from studies meeting eligibility criteria and pooled data using the same statistical method described in the original reviews to produce effect estimates specific to LMIC context. For the topic of TH, we disaggregated results by method of cooling (e.g., WB and SH cooling) and ran subgroup analyses by study site (e.g., multicenter studies and single-center studies) and by type of cooling device (e.g., servo-controlled device and low-cost or manual cooling device). For neuroprotective therapies, we disaggregated results by the presence or absence of TH (e.g., sole therapy and adjunct to TH). In the case that missing data on study characteristics or outcome measures limits our ability to use the study in analyses, we planned to contact the corresponding author for more details. Where study protocols or trial registration details were available, we compared these to the full publications to determine the likelihood of reporting bias. Funnel plots were constructed to explore publication bias if a sufficient number of studies (n > 10) are included within an outcome. A visual inspection of the plot, Egger’s test, Begg’s test, and Harbord’s test were used to assess small-study effects.

The range of interventions reviewed and associated methodology for evidence synthesis is shown in Figure 1 and Table 1 (PRISMA diagrams per topic can be found in the online suppl. Fig. 3.1–3.9). After removing duplicates for de novo reviews, a total of 5,246 records were screened by two reviewers, and 48 records were retrieved at full text. One pilot trial on fluid restriction [39], eight trials on prophylactic ASMs [40‒47], and four trials assessing ASMs for treatment of neonatal seizures [48‒51] in term infants with encephalopathy met the inclusion criteria for our review.

For the topic of TH, 29 studies were included originally by the review [34]. Ten of these studies were conducted in high-income settings and were thus excluded. Additionally, one study was excluded given that some infants’ initiation of treatment was delayed past the 6-hour therapeutic window [52]. Lastly, our TAG members identified four new studies [53‒56] for inclusion. Our reanalysis therefore included 22 trials on TH conducted in LMICs. For the topic of allopurinol, the original review [33] included three trials, one of which was conducted in Egypt [57]. Our updated systematic search found two new trials [58, 59] conducted in Pakistan for a total of three LMIC trials included in this review. From recent systematic reviews on pharmacological interventions for neuroprotection [35‒37], we identified 16 trials on MgSO₄, three unique trials on erythropoietin [60‒62], and three trials on melatonin [63‒65] conducted in LMICs. Lastly, for ECD interventions, we identified eight trials conducted in LMICs in the original review [38]. Narrowing the scope to asphyxiated infants as the population, we identified one trial of relevance to report [66]. List of excluded studies across all topics can be found in online supplementary material (table 4).

Across all topics, 61 RCTs were included (Table 1). All trials were facility-based, except one which was conducted in rural communities [66]. Eight trials were multicenter [23, 43, 44, 48, 51, 66‒68]. Low and lower middle-income countries included Bangladesh, Egypt, India, Pakistan, Sri Lanka, Uganda, and Zambia. Only for the topic of ASMs, trials conducted in HICs were included [43‒46, 48, 51]. The included studies in this review targeted 5,046 term infants with asphyxia and/or encephalopathy. The trial on fluid restriction assessed restriction of fluids by 40–80 mL/kg compared to a normal fluid range of 60–120 mL/kg in the first 4 days of life [39]. For prophylactic ASMs, five trials assessing phenobarbital (20–40 mg/kg dose) were included [40‒42, 46, 69]. Trials conducted in HICs on thiopental [45] and topiramate [43, 44] were leveraged since no LMIC studies were available for these drugs. For ASMs as a first-line agent in treatment of neonatal seizures, two trials [49, 50] conducted in LMICs compared levetiracetam (20 and 60 mg/kg dose) to phenobarbital (40 mg/kg dose). Additionally, a multicenter trial [48] conducted in HICs assessing a dose of 40 mg/kg levetiracetam compared to phenobarbital on a subgroup of neonates with HIE and electrical seizure activity was also included. For second-line ASMs, one trial [51] in the UK compared lidocaine to midazolam and clonazepam in neonates with seizures persisting after phenobarbital. Two trials [58, 59] on allopurinol assessed oral formulations delivered through the nasogastric tube, while one study [57] evaluated IV formulation. For TH, 15 studies [23, 53‒55, 70‒80] implemented WB cooling and seven studies [56, 67, 81‒85] implemented SH cooling. All trials were conducted in tertiary neonatal units with intensive care and facilities for ventilatory support except one [71] which was conducted in a special care baby unit. Three unique trials [60‒62] on erythropoietin and two multi-arm trials [47, 75] assessed different treatment regimes. Majority of the 16 trials on MgSO₄ assessed 250 mg/kg/dose IV MgSO₄ administered within 6 h of birth with maintenance for 3 days. Only two studies [68, 86] were conducted on infants who also received TH. Melatonin was evaluated in three trials as a monotherapy [65], adjunct to TH [64], or combined with MgSO₄ [87]. Lastly, for ECD interventions, one included multicenter trial [66] assessed interventions delivered by parents who received training in biweekly home visits initiated during the neonatal period and continuing until 36 months. Outcomes were review specific; however, each topic assessed neonatal mortality and developmental outcomes in infancy. Detailed characteristics of all included studies by topic are available in online supplementary material (table 5).

Of the 61 RCTs across all intervention domains evaluated using the Cochrane Risk of Bias (RoB) 2.0 tool [88], eight (13%) were rated as having low RoB, nine (15%) were rated as having moderate RoB and 44 (72%) were rated as having a high RoB (Fig. 2). Trials rated as “high risk” primarily had concerns around performance bias due to post-randomization exclusions, sample attrition (>10%), missing information, and lack of blinding of outcome assessors (see online supplementary material [Fig. 6.1, 6.2] for detailed quality ratings per included study.

We identified one pilot RCT comparing a strategy of fluid restriction (40–80 mL/kg) to normal fluids (60–120 mL/kg) in the first 4 days of life in 80 infants with moderate to severe encephalopathy also undergoing cooling with ice packs [39]. The study was conducted in a tertiary care academic institute in urban southern India and found no impact on mortality or morbidities of the newborn (online suppl. material, Table 7.1).

Prophylactic ASMs. Pooled effect of three studies [40, 41, 47] with 184 term neonates found a significant 49% (95% CI: 18–68%) reduction in incidence of clinical (probable) seizures in groups that received prophylactic phenobarbital compared to controls (Table 2). Subgroup analysis by dosage (20 mg/kg vs. 40 mg/kg) did not show significant differences. A significant pooled effect was reported for seizures diagnosed only clinically by two studies [40, 69] as opposed to a nonsignificant effect reported by one study [47] where clinical correlates were confirmed with aEEG. Incidence of mortality and abnormal neurological outcome at discharge were not significantly different between groups. One study conducted in a HIC [45] assessing prophylactic thiopental found no significant results. Lastly, no significant reduction in risk of seizures (measured using cEEG), abnormal MRI, and various neurodevelopmental assessments at different time points were reported by two studies [43, 44] assessing prophylactic topiramate as an adjunct to TH in HICs.

ASMs for Treatment of Seizures. When comparing levetiracetam to phenobarbital, three studies [48‒50] recruiting 169 asphyxiated neonates with seizures found no significant differences in seizure control after the primary drug was administered (RR = 0.60, 95% CI: 0.17–2.16) (Table 2). Heterogeneity was calculated to be 93% and tests for subgroup differences by dosage of levetiracetam reached significance (p < 0.05). Subgroup analysis by setting (LMIC vs. HIC) did not reveal significant differences (p = 0.37). Two studies [49, 50] conducted in India used bedside clinical assessment for seizure diagnosis and found no difference in seizure control between groups that received levetiracetam or phenobarbital to treat seizures. A multicenter trial [48] conducted in HICs reported a significant difference (p = 0.007) in electrographic seizure cessation rate at 24 h (RR = 0.39, 95% CI: 0.20–0.77) with phenobarbital demonstrating better seizure control. One of the LMIC trials [50] found the average time taken for seizure control to be significantly longer in group that received levetiracetam (MD = 2.33, 95% CI: 0.26–4.40). The study also reported nonsignificant higher risk for electrical seizures after clinical control in the group assigned to levetiracetam (RR = 2.17, 95% CI: 0.73–6.42). Pooled analysis of two studies [49, 50] found risk of abnormal neurological outcome at discharge to be similar between groups (RR = 0.91, 95% CI: 0.50–1.63). Incidence of mortality and abnormal neurologic outcome assessed at 3 and 6 months was lower in the phenobarbital group, though only one study contributed to these outcomes [50]. No significant differences were found for secondary outcomes of thrombocytopenia, deranged renal or liver function [49, 50]. We could not perform subgroup analyses by route of ASM administration (all IV access) and duration of ASM treatment (all short-term).

One study comparing lidocaine to midazolam as second-line treatment in 11 infants found no significant difference in efficacy of seizure control (RR = 4.00, 95% CI: 0.26–61.76) [51]. Clonazepam was also used as an add-on drug with seizure control in 0/3 infants [51].

SH Cooling. Seven studies [56, 67, 81‒83, 85, 89] recruiting 522 neonates evaluated SH cooling versus a control (Table 3). Two studies were pooled and revealed a reduced risk of death or neurological disability at ≥18 months by 45% (95% CI: 24–60%), mortality at ≥18 months by 39% (95% CI: 3–62%), and neurological disability at ≥18 months by 59% (95% CI: 27–77%). Both tests for subgroup differences by multicenter versus single-center trials, and by low-cost cooling device/materials versus servo-controlled cooling devices, revealed no impact on any outcome.

WB Cooling. Fifteen studies [23, 53‒55, 70‒80] recruiting 1,586 neonates compared WB cooling with a control. Pooled analysis of four studies [23, 55, 77, 80] found a reduced risk of death or neurological disability at ≥18 months by 17% (95% CI: 2–29%), and neurological disability at ≥18 months by 46% (95% CI: 27–61%). All TH trials are rated as having a “high” risk of bias due to lack of blinding (given the nature of the intervention). To test robustness of results, we conducted a sensitivity analysis restricted to six trials with a relatively lower risk of bias and found a higher risk of neonatal mortality before discharge in group that received TH (RR: 1.24 [95% CI: 0.95–1.62]) (Table 3).

The test for subgroup differences by multicenter versus single-center trials, and by low-cost manually controlled cooling device/materials versus servo-controlled cooling devices, revealed significant differences (p < 0.05) between groups for the composite of death or neurological disability at ≥18 months, infant mortality at ≥18 months, and neonatal mortality at discharge. All three outcomes were found to have significantly increased risk by one multicenter trial using a servo-controlled cooling device [23] compared with multiple single-center trials using low-cost manually controlled cooling device/materials reporting a pooled reduced risk

Publication bias was assessed for the outcome of neonatal mortality before discharge as it had sufficient number of included studies (n = 12 trials). Visual assessment of funnel plot suggests asymmetry (Fig. 3). Between-study heterogeneity was substantial (I² = 65%). Results of the Egger’s (p = 0.096), Begg’s (p = 0.89), and Harbord’s tests (p = 0.091) did not reach statistical significance.

Allopurinol. Three studies [57‒59] with 172 participants assessing allopurinol versus control found no significant impact on neonatal mortality or severe neurodevelopmental disability. Pooled effect of two studies showed a statistically significant 50% (95% CI: 13–71%) reduction in risk of CP in surviving infants (Table 4).

Magnesium Sulfate. MgSO₄ exhibited a 17% (95% CI: 1–31%) reduced risk of neonatal mortality in pooled analysis of 13 trials recruiting 989 neonates [63, 68, 86, 90‒99]. Subgroup analysis based on TH as co-treatment (TH vs. No TH) did not reveal significant differences. Furthermore, MgSO₄ was found to lower the risk of poor feeding by 49% (95% CI: 32–62%) and abnormal CT scan of the brain by 38% (95% CI: 7–58%) in the absence of TH. No significant differences were found in the incidence of abnormal EEG between treatment and control groups (Table 4).

Erythropoietin. Erythropoietin, assessed in three trials [47, 60, 61] significantly reduced the risk of death or neuro-disability at 18 months of age by 44% (95% CI: 25–58%) and lessened the occurrence of CP by 53% (95% CI: 20–72%). It did not, however, have a significant impact on death in the neonatal period and at follow-up between 3 and 19 months of age (Table 4).

Melatonin. The effect of combining melatonin with hypothermia is inconclusive as only one study [64] contributed to the outcome of mortality (Table 4).

One multicenter trial [66] with 164 participants residing in rural areas in India, Pakistan, and Zambia found significantly improved cognitive development outcome at 36 months in infants with history of asphyxia who received early developmental intervention compared to controls (online suppl. material, Table 7.2). Forest plots can be referenced in online supplementary material (Fig. 7.1–7.45).

This review has summarized the most up-to-date evidence on post-asphyxial aftercare and management strategies for NE, forming part of a supplement describing an extensive synthesis of effective newborn interventions in LMICs. There is a lack of evidence to draw strong conclusions on the effectiveness of fluid restriction for management of NE. Several studies reported on prophylactic phenobarbital administration for prevention of mortality and morbidities and although a reduction in risk of clinical seizures was found, there was no impact on mortality and limited studies addressing other morbidities such as developmental outcomes. No difference was found between levetiracetam and phenobarbital in clinical seizure control reported by two LMIC trials, however one HIC trial indicated better control of electrical seizure activity in group that received phenobarbital. The evidence on TH included a number of studies with small samples sizes and several studies from the same center with potential overlapping recruitment. Notably, the composite outcome of death or neurological disability was only reported by four trials [23, 55, 77, 80]. While the outcome of neonatal mortality was reported by 12 studies, nearly half of the sample (six trials [54, 72, 73, 76, 78, 80]) were conducted in the same center, limiting generalizability to other centers and health care systems. Restricting the analysis to trials with relatively lower risk of bias (although lack of blinding was noted in all studies), did not find an impact on mortality (online suppl. Fig. 7.33). One large multicenter trial in SEA showed an increase in mortality across all trial sites. Evidence on pharmacological therapies for neuroprotection suggests that only erythropoietin reduced the risk of death or neuro-disability in infancy. No therapies were found to have an impact on mortality except for MgSO₄ which showed a reduction in the risk of neonatal mortality, but long-term outcomes were lacking. Lastly, one trial on early postnatal interventions found improvements in cognitive development at 36 months of age.

A Cochrane review from 2015 on fluid restriction found no studies to include and thus could not determine the effect of this strategy on mortality or morbidity in term neonates following birth asphyxia [15]. We identified one pilot RCT [39] conducted in South India testing a strategy of fluid restriction in neonates with encephalopathy undergoing cooling. The trial found that fluid restriction did not reduce the composite outcome of death or neurodevelopmental disability and had trends toward adverse outcomes. Given the small sample size, the trial was underpowered to detect differences in outcomes which makes conclusions drawn from study results uncertain. To the best of our knowledge, this is the first trial comparing a strategy of fluid restriction to normal fluid intake in newborns and suggests the feasibility of trials in other centers with well-defined rescue and withdrawal criteria to optimize fluid management strategies for encephalopathic neonates.

ASMs are used prophylactically in newborns post-hypoxia to prevent seizures and avoid further brain injury [100]. Current evidence shows that prophylactic phenobarbital had no impact on mortality and abnormal neurological outcomes at discharge but may reduce the incidence of clinical seizures in newborns following asphyxia. However, since seizures were diagnosed clinically within the included studies, the effect of ASMs in reducing the true burden of electrical seizure activity is undetermined. 50–80% of seizures in term infants with NE are estimated to be electrographic only with no evident clinical manifestations [84]. Furthermore, treatment with ASMs like phenobarbital may cause “uncoupling” wherein electroclinical seizures become electrographic only [101, 102]. The Lancet Every Newborn Action Plan (ENAP) series (2014) concluded that prophylactic antiseizure therapy in term infants with birth asphyxia cannot be recommended for routine clinical practice other than in the treatment of prolonged or recurrent clinical seizures [103]. Two newer studies [43, 44] conducted in HICs on prophylactic topiramate as a potential neuroprotective co-treatment with TH using cEEG monitoring also showed no significant effect on electrical seizure activity, mortality, or neurodevelopmental disability. Thus, current evidence does not support prophylactic ASM use in routine clinical practice to prevent seizures in neonates immediately following birth asphyxia.

We identified three trials assessing the effectiveness of levetiracetam or phenobarbital as a first-line treatment for term newborns with intrapartum-related encephalopathy and seizures. The efficacy of phenobarbitone in controlling seizure with the first dose was not significantly different from levetiracetam in LMICs. Since most seizures in this population are electrographic only [104, 105] onset of seizures with no clinical manifestation could also have been missed by these studies. We leveraged results from the NEOLEV2 trial, a recent multicenter trial conducted in tertiary care hospitals in HICs, which reports that phenobarbital is more efficacious in controlling electrographic neonatal seizures in a subgroup of infants with HIE [48]. Remote cEEG monitoring in the first 24 h and up to 6 days of life may have contributed to earlier detection and treatment of seizures which may still not be feasible in most LMIC settings [106, 107]. Phenobarbital is the recommended first-line ASM for the treatment of neonatal seizures and is recommended by WHO (2011) and International League Against Epilepsy (ILAE) (2023) guidelines followed by phenytoin and benzodiazepines as second-line drugs [19, 108]. Phenobarbital is preferred due to its low-cost, ubiquity, and well-understood safety profile from decades of use [19]. However, only around 50% of neonatal seizures respond to phenobarbital and so there is also a need to identify ASMs to treat seizures in infants not responding to phenobarbital as a first-line agent [19]. We identified one RCT comparing the efficacy of lidocaine to midazolam in controlling seizures when newborn does not respond to phenobarbital, but the sample size was small (n = 11).

Evidence, mostly from high-income settings, suggests that WB cooling could reduce neonatal mortality by 25% and major neurodevelopmental disabilities by 23% [22, 34, 103]. Our review of TH solely using evidence generated in LMICs disaggregated trials by method of cooling (WB and SH). WB cooling is simpler to implement compared to SH cooling and is the preferred method due to controlled rewarming [109, 110]. WB cooling may reduce the risk (at 18 months or later) of the composite outcome of death or neurological disability, neurological disability alone, but not mortality alone when compared with control. Subgroup analyses revealed significant differences in key outcomes between one multinational trial using a servo-controlled device and several small, pilot trials using manually controlled cooling methods. We constructed a funnel plot (Fig. 3) to explore small-study effects for the outcome of neonatal mortality, however, the statistical tests were of unclear utility due to high heterogeneity between trials. Long-term studies based on national databases even in HICs such as Korea showed poor growth and developmental outcomes in infants following NE [111, 112]. Currently, TH is being implemented in some LMICs with poor reporting systems, despite recommendations by the International Liaison Committee on Resuscitation to only consider using cooling in facilities with clearly defined protocols, capabilities for multidisciplinary intensive care, and availability of adequate monitoring resources [113‒115]. Almost all included studies in this review were conducted in tertiary care centers in cities with the possibility of referral bias [116]. Some studies included outborn patients (online suppl. material; Table 8.1). Without dedicated retrieval teams, robust referral systems and transport facilities to tertiary neonatal units, this therapy cannot be implemented outside referral tertiary level centers with high-fidelity intensive care units in LMICs. Even in these institutions, there are concerns, such as delayed initiation of cooling past the therapeutic window. The case fatality rate from NE was estimated to be 27.7% in tertiary care hospitals in LMICs and reported to be much higher in a community setting (49%) [5]. There is also a significant burden in secondary care neonatal units of district hospitals [117]. Given the lack of evidence for feasible and effective therapies in settings with the highest burden of asphyxia, further effort on prevention of asphyxia insult and NE is needed.

There is also a need to identify other neuroprotective treatment strategies that may be effective. While several pharmacological measures are studied, many are currently not ready for widespread adoption due to reasons of limited evidence, cost, or access. In our secondary analysis of pharmacological agents for neuroprotection, we found direct evidence from LMICs on, MgSO₄, erythropoietin, allopurinol and melatonin. While magnesium therapy was found to improve short-term outcomes, multicenter trials similar to the Mag Cool study evaluating neurodevelopmental disability with follow-up in infancy and beyond are needed in LMICs. Pooled effects from small trials for erythropoietin suggest it may improve neurodevelopmental outcomes as a sole therapy. However, this needs to be further evaluated in clinical trials. The effect of allopurinol on the composite outcome of neonatal mortality and severe neurodevelopmental disability was inconclusive. More research is required to substantiate these preliminary findings and study different formulations. Recent trials conducted in centers where IV formulation was unavailable used oral formulation of allopurinol administered via nasogastric tube. Lastly, the survival benefit of combining melatonin with hypothermia was uncertain due to limited data. Osmotic agents like mannitol used to manage intracranial pressure have also been investigated in HICs to no effect [118]. Future studies should include more representative preclinical models to test neuroprotective properties, rigorous dose-finding studies, and well-powered clinical trials for promising therapies.

Of infant survivors of birth asphyxia, 15–20% die in the postnatal period, and an additional 25% will sustain childhood disabilities [119, 120]. Given the limited access to institutionalized developmental follow-up and support for children with disabilities in LMICs, targeting infants at risk for neurodevelopmental disabilities to receive early interventions in community settings might help reduce the burden of NE [30, 121]. The included trial is the first multinational study of ECD interventions targeted at infants with a history of birth asphyxia and found positive effects on cognitive development at 3 years of age but none on motor development [66].

This review was primarily limited by our approach to reanalyze data from studies conducted in LMICs from existing systematic reviews where possible. While this method allowed us to review multiple interventions and thus evidence on post-asphyxia management in its totality, we were limited to the inclusion of studies captured by the source review and their methodology. We attempted to address this limitation by consulting with our TAG to identify additional trials for inclusion. Regardless, we found limited evidence for effective post-asphyxial aftercare strategies in low-resource settings with the highest burden of disease, thus there is a need to redirect research efforts toward preventive measures and early recognition of asphyxia and NE. This review was also limited by its focus on neonatal population at term due to prespecified criteria in source reviews. As a result, we were unable to review evidence on optimal therapies for preterm infants exposed to hypoxic events who have a distinct clinical profile, often requiring more intensive care due to their underdeveloped lungs and greater vulnerability [122].

Overall, there is a lack of good quality evidence to determine which interventions are most effective in low-resource settings for management and aftercare of newborns following asphyxial insult. The burden of NE remains high despite supportive care and TH indicating a much-needed focus on reducing the incidence of hypoxic events via improved access to antenatal and intrapartum care. Preventive strategies like perinatal quality audits, skilled birth attendance and early recognition of birth asphyxia in the community with prompt action, are likely to reduce the burden of NE in LMICs and must be underscored. Accessible facilities providing comprehensive supportive care are crucial for seizure control, close monitoring of the newborn, as well as prevention and management of infections. While TH may be useful for some infants in select tertiary referral centers, results are not generalizable with evidence of increased mortality in LMICs. Given damage to the brain from newborn hypoxia is permanent, larger trials with long-term follow-up are needed to evaluate the clinical significance of other neuroprotective therapies including pharmacological measures such as MgSO₄. Early intervention in community settings can be an approach used to improve developmental outcomes for infants with a history of birth asphyxia, however more studies evaluating effective interventions specific to this high-risk group are warranted.

We gratefully thank the Technical Advisory Group (TAG) of experts in newborn asphyxia for their contributions to the conceptualization of the study. We would like to thank Dr. Li Jiang for her support in translation of documents, data extraction, and analysis of TH trials.

An ethics statement is not applicable because this study is based exclusively on published studies.

The authors have no conflicts of interest.

The present evidence synthesis received support by the Gates Foundation (Grant #: INV-042789). The funders had no role in the design, analysis, or content of this article.

L.H. and Z.A.B. conceptualized the study. L.H., O.M., and S.R. designed the study, search strategy, and carried out the search and study selection. D.S., O.M., and S.R. completed data extraction, critical appraisal of included trials, and data analysis. O.M. led writing of the manuscript with contributions from L.H., S.R., and D.S. L.H., T.V., and Z.A.B. supervised the whole process and critically reviewed and revised the manuscript for important intellectual content.

Data published are available in public domain. Further inquiries can be directed to the corresponding author.