The Route, Dose, and Interval of Epinephrine for Neonatal Resuscitation: A Systematic Review

The systematic review protocol was registered in the International Prospective Register of Systematic Reviews (CRD42019132219).

With this systematic review, we assessed the question, “Among neonates (of any gestation) at ≤28 days of age who have no detectable cardiac output, have asystole, or have a heart rate <60 beats per minute despite ventilation and chest compressions, does any nonstandard dose, interval, or route of epinephrine (adrenaline) improve the primary outcome (death at discharge) or any secondary outcomes compared to standard-dose IV epinephrine (0.01–0.03 mg/kg at intervals of every 3–5 minutes)?”

Eligible participants were neonates of any gestational age within 28 days of birth who had either no detected cardiac output or asystole or heart rate <60 beats per minute despite ventilation and chest compressions and had received epinephrine for resuscitation. In eligible studies, authors either compared predefined outcomes between infants receiving epinephrine by using standard or nonstandard routes, doses, or intervals of administration or they reported predefined outcomes of infants receiving nonstandard administration of epinephrine. For this review (reflecting the ILCOR 2010 CoSTR), the standard administration of epinephrine was defined as IV administration of 0.01 to 0.03 mg/kg per dose, and if return of spontaneous circulation (ROSC) did not occur in response, it was defined as repeated administration at 3- to 5-minute intervals.² 

Randomized controlled trials (RCTs) and nonrandomized studies (non-RCTs, interrupted time series, controlled before-and-after studies, and cohort studies) were eligible for inclusion. Unpublished studies (eg, conference abstracts, trial protocols), case series in which authors could not provide an estimate of incidence of outcomes, and animal studies were excluded. In cohort studies, authors may compare different interventions or report outcomes for a single arm receiving one intervention. For this review, observational studies were considered to be cohort studies eligible for inclusion if the authors used a defined strategy to ensure that the participants were either all of those who received an exposure of interest in a defined population (eg, infants born at a hospital between specified dates) or were sampled in such a way as to be representative of such a population.⁶  Otherwise, the study was considered to be an ineligible case series. All languages were eligible if there was an English abstract.

The prespecified primary outcome was death at hospital discharge, and secondary outcomes were rate of failure to achieve ROSC (defined as a sustained audible heart rate >60 beats per minute),^7,8  time until ROSC, moderate or severe hypoxic ischemic encephalopathy (only for term infants), intraventricular hemorrhage grade III or IV, other morbidities in early infancy (eg, necrotizing enterocolitis, retinopathy of prematurity, bronchopulmonary dysplasia, periventricular leukomalacia [only for preterm infants]), and neurodevelopmental outcomes. The need for repeat doses of epinephrine was evaluated as a post hoc secondary outcome.

An information specialist (C.Z.) conducted database searches on March 6, 2019, from Ovid Medline, Embase, the Cochrane Central Register of Controlled Trials, the Cochrane Database of Systematic Reviews, and Cumulative Index to Nursing and Allied Health Literature. The search strategies, adapted for each database, used a comprehensive combination of subject headings and keywords for neonates, cardiac arrest, and epinephrine, combined by using the Boolean operator “AND.” The Medline search strategy is found in the Supplemental Information. The databases were searched from inception, and no language limits were applied. The Medline search strategy was peer reviewed by another information specialist using the Peer Review of Electronic Search Strategies checklist.⁹  References from included studies and review articles were also hand searched, and ILCOR Neonatal Life Support Task Force and Working Group members were consulted. Trial registries in the United States, Europe, and Australia and/or New Zealand were searched.^10–12 

Titles and abstracts were independently screened by two authors (T.I. and H.G.L.) who then screened full-text articles selected by either reviewer. Discrepancies were resolved by consensus of all investigators. Cohen’s κ statistic for agreement was calculated. The same two authors independently extracted the data on study designs, study population, interventions, and outcomes. Study authors were contacted to obtain missing data or for clarification.

Two authors (T.I. and L.M.) assessed the risk of bias of each included study as low, moderate, serious, or critical using the Risk of Bias in Nonrandomized Studies - of Interventions (ROBINS-I) tool for comparative cohort studies.¹³  For single-arm (uncontrolled) cohort studies, a modification of a tool proposed by Murad et al¹⁴  was used.

For studies in which authors compared either primary or secondary outcomes between two interventions, meta-analyses were performed with random-effect models because heterogeneity was anticipated. Review Manager version 5.3 (Nordic Cochrane Centre, Copenhagen, Denmark) was used for the analyses.¹⁵  The results are shown with risk ratio (RR) for binary outcomes and mean differences (MDs) for continuous variables along with their 95% confidence intervals (CIs). For studies in which authors assessed one intervention (single arm), the proportions of outcomes were calculated. Statistical significance was determined by using two-sided P values <.05. No adjustment was conducted for multiple outcomes, which is typical of systematic reviews. Heterogeneity was evaluated among included studies by examining forest plots, the I² statistic, and χ² tests for heterogeneity.¹⁶ 

The GRADEpro Guideline Development Tool (software) was used to assess the certainty of evidence as high, moderate, low, or very low for each outcome.^5,17  For nonrandomized studies assessed by using ROBINS-I, in accord with recent recommendations of GRADE methodologists, an initial assumption of high certainty of evidence was applied.¹⁸  Ratings were downgraded on the basis of the assessment of risk of bias, inconsistency, indirectness, imprecision, and other considerations where applicable.¹⁸ 

Because there were so few eligible studies during screening, articles in which outcomes for nonstandard doses, routes, or dosing intervals were reported but did not meet inclusion criteria for the systematic review were set aside to include in a table and narrative summary.

Because there were few eligible human studies, the searches were subsequently rerun to include animal studies. They were expected to contribute information regarding mechanisms, pharmacokinetics or pharmacodynamics, or toxicity and adverse effects. Animal studies were not submitted to GRADE analysis, combined with human studies, or used for estimates of magnitude of effect.

Among 593 unique articles retrieved, 98 articles were selected for full-text screening, and 4 studies were considered eligible (Supplemental Fig 2, Table 1).^7,8,19,20  Initial agreement between the two reviewers was fair (κ = 0.34), but differences were resolved by discussion. All 4 included studies were single-center, retrospective cohort studies and included only newly born infants. Three studies were from the same institution, although the study populations did not overlap.^7,8,20  In two sequential studies from one institution, authors reported outcomes of infants receiving chest compressions for resuscitation, some of whom received epinephrine initially via endotracheal and some via IV administration (referred to hereafter as two-arm studies).^7,8  Of note, the endotracheal epinephrine doses used in both studies (0.01 mg/kg per dose⁷  and 0.03–0.05 mg/kg per dose⁸ ) were lower than those currently recommended (0.05–0.1 mg/kg per dose).²  Furthermore, the proportion of infants receiving initial endotracheal epinephrine decreased from 44 of 47 infants in the first to 30 of 50 infants in the second study, indicating a practice shift. In both studies, some infants who initially received endotracheal epinephrine subsequently received ≥1 doses of IV epinephrine. In the other two studies, authors reported the outcomes of preterm infants receiving endotracheal epinephrine for resuscitation (single-arm studies).^19,20  Jankov et al¹⁹  included only infants ≤750 g birth weight. Although Perlman et al²⁰  included preterm and term infants, the subgroup of term infants (with one group receiving IV epinephrine) were excluded from this review because their outcomes were not reported by route of administration. Therefore, gestation and birth weight characteristics were lower for included infants from these two studies than for the two-arm studies (Table 2). None of the included studies were specifically designed to address the question for this systematic review, so we extracted such data that did allow any comparison of review outcomes (Table 2). Reasons for exclusion of other studies are summarized in Supplemental Table 5.

On the basis of the ROBINS-I, both two-arm studies had very serious risk of bias because of the risk of confounding (Supplemental Table 6).^7,8  On the basis of the modified Murad’s tool,¹⁴  both single-arm studies had serious risk of bias (or indirectness) because the thresholds for administering epinephrine appeared to be lower than those of current ILCOR recommendations (Supplemental Table 7).^19,20 

In only one study did authors provided data used to address the primary outcome of the systematic review. This study (reporting 50 infants treated with epinephrine) revealed no difference in death at hospital discharge between initial endotracheal and IV epinephrine (RR = 1.03 [95% CI 0.62 to 1.71]; absolute risk difference [ARD] = 17 more [209 fewer, 391 more] per 1000 infants; very low certainty of evidence, downgraded for very serious risk of bias and very serious imprecision) (Table 3, Fig 1).⁷  For the secondary outcome of failure to achieve ROSC, meta-analysis of two studies (97 infants treated with epinephrine) revealed no difference between initial endotracheal and initial IV administration (RR = 0.97 [95% CI 0.38 to 2.48]; ARD = 7 fewer [135 fewer, 322 more] per 1000 infants; very low certainty of evidence, downgraded for very serious risk of bias and very serious imprecision).^7,8  For time to ROSC, one study (50 infants treated with epinephrine) revealed no difference between initial endotracheal and initial IV epinephrine (MD = 2.00 [95% CI −0.60 to 4.60] minutes; very low certainty of evidence, downgraded for very serious risk of bias and very serious imprecision).⁷  Post hoc analysis of the proportion of infants who received subsequent doses of epinephrine (two studies; 97 infants) revealed no difference after initial endotracheal versus initial IV administration (RR = 1.94 [95% CI 0.18 to 20.96]; ARD = 654 more [570 fewer, 1000 more] per 1000 infants; very low certainty of evidence, downgraded for very serious risk of bias and very serious imprecision).^7,8 

In two studies, authors provided data for infants only exposed to initial endotracheal epinephrine.^19,20  In one study, authors reported that mortality to hospital discharge among these infants was 0.38 (3 of 8 infants born at a mean gestation of 30 weeks)²⁰  In the second study, the proportion who died or had severe neurodevelopmental impairment at a median assessment age of 2 years was 0.58 (7 of 12 infants born at a mean gestation of 25 weeks).¹⁹ 

The rates of death at discharge and failure to achieve ROSC were similar for infants treated with each of these two doses (30 infants in one study with very serious risk of bias; Supplemental Fig 3).⁷ 

No studies eligible for this systematic review evaluated alternative doses of IV epinephrine, routes other than IV and endotracheal (eg, intraosseous, intramuscular [IM]) administration, or dosing intervals. We summarized results of ineligible studies assessing these points in Table 4. Data on the use of high-dose IV epinephrine for neonatal resuscitation are meager (Table 4). In adults and children, a systematic review of 15 RCTs in which authors compared high-dose versus standard-dose epinephrine for cardiac arrest (typical starting dose for children of 0.1 mg/kg vs 0.01 mg/kg) reported a slight reduction in survival to admission and time until ROSC with high-dose epinephrine; however, the quality of evidence was very low, the subgroup of pediatric studies revealed no difference, and the applicability to newborn infants is uncertain.²¹  Although authors of a few case series reported high-dose IV or endotracheal epinephrine in preterm infants, the safety and effectiveness of high doses remain unknown.^22,23  Longer latency between doses was examined in an observational cohort study of children with cardiac arrest and was associated with improved survival. However, the direction of effect had reversed after adjustment for confounders, suggesting a high risk of bias.²⁴ 

In a few studies, authors described routes for epinephrine administration other than via endotracheal or IV during neonatal resuscitation. In a cohort study of 27 neonates within 5 hours of birth (median gestation of 31 weeks [range 25–41 weeks]), the authors commented that intraosseous access was always accomplished within 2 minutes (although data were not shown) and incurred few major complications.²⁵  Nevertheless, in a few case reports, authors have described severe complications including amputation of an extremity due to extravasation and compartment syndrome attributable to intraosseous lines in infants.^26,27  In one case report, authors described severe skin lesions at the site of an IM epinephrine injection.²⁸ 

In several studies, authors addressed the time to obtain access suitable for administering epinephrine (Table 4). In a cohort of neonates who had a 1-minute Apgar score of 0 (median gestation of 36 weeks [range 27–41 weeks]), among 23 for whom the time was documented, umbilical venous catheter (UVC) placement was accomplished at a median time of 9 (95% CI 7 to 14) minutes after birth, probably longer than ideal for asystolic neonates, whereas endotracheal tubes were placed much earlier at a median time of 3.8 (95% CI 2.0 to 7.5) minutes after birth.²⁹  Simulation studies in which authors used neonatal manikins suggest shorter times to intraosseous than UVC placement, but variation in study design, participants, and models makes it difficult to know whether these results would correspond to clinically meaningful time differences in practice.^31,32  Furthermore, selection of participants in some of these studies may have been weighted to those who had greater previous familiarity with intraosseous insertion than UVC placement.

Several animal studies provided relevant information, although there were variations in animal models and interventions (Supplemental Table 8). This systematic review was based on the presumption that epinephrine is efficacious, but this is based on expert opinion because there are no adequate human newborn infant trials.²  Animal studies are used to address this question. In near-term lambs undergoing perinatal transition with asphyxia-induced cardiac arrest, administration of IV epinephrine (0.01–0.015 mg/kg) was critical for increasing carotid arterial pressure and carotid blood flow and thereby achieving ROSC.³³  However, authors of other animal models after the perinatal transition (porcine asphyxial cardiac arrest) found no benefit of epinephrine.^34,35  Wagner et al³⁵  reported that 50% of animals achieving ROSC did so without epinephrine, whereas Linner et al³⁴  revealed that 80% of saline controls achieved ROSC. Both reported that epinephrine did not improve the rate of ROSC.^34,35  Furthermore, early epinephrine (0.01 mg/kg administered before closed-chest cardiac massage) did not reduce the time to achieve ROSC, improve the rate of ROSC, or improve postresuscitation survival compared with saline controls.³⁴ 

A neonatal lamb study (asphyxial cardiac arrest) revealed that epinephrine doses administered into the right atrium at 0.05 or 0.1 mg/kg resulted in a greater increase in heart rate and systemic vascular resistance than lower doses (0.001 and 0.01 mg/kg).³⁶  However, the higher dose (0.1 mg/kg) blunted the rise in cardiac output and stroke volume, probably because of a dose-dependent rise in peripheral vascular resistance.³⁶  Vali et al³⁷  examined asphyxiated term lambs in cardiac arrest with transitioning fetal circulation and fluid-filled lungs to compare IV versus endotracheal epinephrine. In this study, epinephrine administered via the right atrium or UVC (0.03 mg/kg every 3 minutes; maximum 4 doses) achieved higher and faster peak plasma concentrations than epinephrine administered via endotracheal (0.1 mg/kg) and resulted in a shorter time to ROSC with fewer doses.³⁷ 

The human infant data on which to base guidelines for epinephrine treatment during neonatal resuscitation are sparse. In this systematic review, we included 4 retrospective cohort studies in which authors reported outcomes for 117 eligible neonates. Only two studies reporting 97 neonates allowed comparison of endotracheal and IV epinephrine administration, and only one of these allowed analysis of the primary outcome for this systematic review: death at hospital discharge.^7,8  The combined results revealed no difference in the primary outcome or the other critical and important outcomes that were reported: failure to achieve ROSC, time to ROSC, or the proportion of infants receiving additional epinephrine doses (post hoc analysis). The very low certainty of the evidence indicates that the findings of “no difference” should be interpreted cautiously; the wide CIs mean that the data could still be consistent with large, clinically meaningful differences between IV and endotracheal epinephrine for neonatal resuscitation.

The 2010 ILCOR recommendations were informed by indirect evidence from animal or pediatric studies of uncertain applicability to newborns.²  The human studies provided evidence against using doses >0.03 mg/kg IV. A pediatric case series suggested some improvement in ROSC if a higher dose was used after two lower doses had failed to achieve ROSC,²²  authors of another study that included children found no associated benefit,³⁸  and the only pediatric RCT suggested harm.³⁹  Of note, the “high” epinephrine doses of these pediatric studies (0.1–0.2 mg/kg per dose) were 3 to 6 times higher than the maximum dose currently recommended for newborns (0.03 mg/kg per dose). Taken together, the evidence at that time was assessed as suggesting that endotracheal might be less effective than IV epinephrine, with the possibility that lower response rates could be mitigated by using a higher endotracheal dose.²  The evidence from the current systematic review is insufficient to confirm or refute these recommendations, and the available studies highlight the difficulties of studying, in vulnerable patients, rare events (eg, 0.05%–0.06% of all births) that are difficult to predict.^7,8  Even if multihospital databases were used to increase the sample size, the high risk of bias seems inescapable in retrospective studies of routine practice for rare events. The reasons for choice of route of epinephrine administration were not predefined in either of the two studies that allowed comparison of routes. The choice was likely to have been influenced by factors such as the extent of forewarning and the characteristics of the resuscitation team available at the time.^7,8  Because the use of epinephrine in newborn resuscitation is rare and unpredictable and the decision to use it must be made rapidly, it is difficult to design and perform adequate, ethical randomized trials in human infants, especially if previous parental informed consent is required. Prospective, multicenter cluster-randomized trials could offer a way forward.

We justified the decision to include single-arm cohort studies that did not allow any comparison between our predefined “intervention” and “control” by the extreme paucity of any evidence from comparative studies, but these studies did not alter the direction or strength of evidence. They provided data on preterm infants, but no robust conclusion about the efficacy or safety of endotracheal epinephrine in preterm infants can be drawn.^19,20  In the cohort studied by Jankov et al,¹⁹  as many as 6% of all infants ≤750 g received epinephrine, raising the possibility that not all the infants were asphyxiated. A proportion of those receiving chest compressions and epinephrine might have responded to better lung ventilation alone, leading to overestimation of survival and quality of survival after epinephrine.

The lack of studies in which authors evaluated high doses, administration routes other than endotracheal or IV, and nonstandard intervals of administration of epinephrine precluded any conclusions about efficacy and risks. The narrative summary of human studies that did not meet the inclusion criteria of this systematic review yielded little more. The applicability of very low certainty evidence from a recent systematic review of adults and children in cardiac arrest to newborns is uncertain.²¹  There are major differences in the conditions that necessitate chest compressions and epinephrine in adults (predominance of asystole due to cardiac causes), in children (mixed causes), and in neonates (mostly intrapartum hypoxia or asphyxia related, against the background of the perinatal cardiorespiratory transition). Therefore, the extrapolation of study findings from adults and children to neonates requires caution. In the study by Halling et al,⁷  most of the infants (11 of 15 [73%]) who achieved ROSC after IV epinephrine required a cumulative dose of 0.03 mg/kg or more (median of ≥2 doses). This raises the possibility that the initial IV dose of 0.01 mg/kg per dose was too low for most infants. However, considering all the evidence, the balance of benefits and harms of higher doses epinephrine for neonates remains unresolved.

The best route of administration for epinephrine is also uncertain. Although in their study, Halling et al⁷  reported similar median time for first epinephrine administration via endotracheal and IV (median 5.1 and 5.4 minutes after birth, respectively), further research is needed to determine if this is achievable in other institutions. It is suggested in other evidence that there may be potential for endotracheal administration to be achieved faster, but the impact on efficacy is unknown.^29,30  Although simulation studies (also of uncertain external validity with respect to resuscitation of newborn infants) indicated faster time for insertion of intraosseous than umbilical venous lines,^31,32  caution is needed because severe complications of intraosseous lines in neonates have been reported.^26,27  There is a need for future research to investigate any advantages and the safety of intraosseous lines in neonatal resuscitation.

For the key questions of this systematic review, animal studies reveal mixed results, which might be because of the variation in models. These include term or near-term lambs or piglets either during the perinatal transition or at a few days of age. The studies reflecting transitional physiology, that is, before closure of fetal shunts and before establishment of sustained lung aeration,^34,38  might provide more valid evidence for newly born infants than studies in which authors use older animals. These transitional model studies revealed benefit of IV epinephrine treatment compared to no epinephrine or endotracheal epinephrine.^33,37 

Strengths of our study include rigorous literature searching and systematic review by using a preregistered protocol, assessment of certainty of evidence by using GRADE,¹⁸  and input from experts from the ILCOR Neonatal Life Support Task Force. Limitations include the absence of RCTs and the small number of eligible studies and infants. Because all eligible studies included only newly born infants, the findings may not be applicable later in the neonatal period. Furthermore, no eligible studies examined alternative doses of IV epinephrine, nonstandard intervals of repeated epinephrine, and routes other than IV and endotracheal.

We found evidence that administration of epinephrine by endotracheal versus IV routes was associated with similar rates of death at discharge, failure to achieve ROSC, time to ROSC, and the proportion of patients receiving additional doses of epinephrine, but the evidence is of very low certainty. However, indirect evidence from animal studies supports IV administration over endotracheal administration at the 0.01 to 0.03 mg/kg dose that has been recommended by ILCOR. Although epinephrine can be administered via intraosseous instead of IV routes, further human infant data are needed.

Several of the authors of this study were ILCOR Neonatal Life Support Task Force members (T.I., L.M., G.M.S., H.-S.K., Y.R., and H.G.L.). The following additional task force members provided input to the review protocol, interpretation of the results, and on the article as experts in neonatal resuscitation: Dr Myra Wyckoff (The University of Texas Southwestern Medical Center, Dallas, TX); Dr Jonathan Wyllie (James Cook University Hospital, Middlesbrough, Cleveland, United Kingdom); Dr Khalid Aziz (Royal Alexandra Hospital, Edmonton, Alberta, Canada); Dr Maria Fernanda de Almeida (Federal University of Sao Paulo, Sao Paulo, Brazil); Dr Ruth Guinsburg (Federal University of Sao Paulo, Sao Paulo, Brazil); Dr Shigeharu Hosono (Jichi Medical University Saitama Medical Center, Saitama, Japan); Dr Vishal Kapadia (The University of Texas Southwestern Medical Center, Dallas, TX); Dr Jeffrey Perlman (Weill Medical College, Weill Cornell Medicine, New York, NY); Dr Charles Roehr (University of Oxford and John Radcliffe Hospital, Oxford, United Kingdom); Dr Edgardo Szyld (University of Oklahoma, Oklahoma City, OK); Dr Daniele Trevisanuto (University of Padua, Padua, Veneto, Italy); and Dr Sithembiso Velaphi (Chris Hani Baragwanath Academic Hospital, Johannesburg, South Africa).

We thank the authors of two included studies (Dr Cecilie Halling, The University of Texas Southwestern Medical Center, Dallas, TX; Dr Robert P. Jankov, Children’s Hospital of Eastern Ontario and the Ottawa Hospital, Ontario, Canada) for providing additional data for this systematic review. We also thank Dr Monica Kleinman (Boston Children’s Hospital, Boston, MA) for overseeing and monitoring the progress of our review as an ILCOR domain lead.

ARD

absolute risk difference

CI

confidence interval

CoSTR

consensus on science with treatment recommendations

GRADE

Grading of Recommendations Assessment, Development and Evaluation

ILCOR

International Liaison Committee on Resuscitation

IM

intramuscular

IV

intravenous(ly)

MD

mean difference

RCT

randomized controlled trial

ROBINS-I

Risk of Bias in Nonrandomized Studies - of Interventions

ROSC

return of spontaneous circulation

RR

risk ratio

UVC

umbilical venous catheter