Duration of Resuscitation at Birth, Mortality, and Neurodevelopment: A Systematic Review

This systematic review was undertaken as part of the continuous evidence evaluation process for the production of the ILCOR CoSTR. This was a Neonatal Life Support Task Force–led review. An information specialist from Apex Information (Vancouver, British Columbia, Canada) assisted with the literature search. The protocol for this review was registered on the International Prospective Register of Systematic Reviews (PROSPERO) on January 17, 2020 (CRD42020157370). The protocol included newborns immediately after birth presenting with at least 10 minutes of asystole, bradycardia, or pulseless electrical activity for which CPR is indicated. Older newborns and infants receiving CPR in the ICU or other settings were not the focus of this review.

Outcomes were selected in accordance with priority outcomes identified through consensus of the ILCOR Neonatal Life Support Task Force.⁶  The primary outcome was survival to the latest follow-up reported by the authors of the study. Secondary outcomes included NDI and the composite outcome of survival without moderate or severe NDI. Neurologic outcomes were as defined by authors of individual studies, provided that age-appropriate assessment tools were used. A post hoc outcome was survival at hospital discharge.

Ovid Medline, Embase, all Evidence-Based Medicine Reviews (including the Cochrane Central Register of Controlled Trials), Cumulative Index to Nursing and Allied Health Literature (CINAHL) Plus with Full Text (EBSCOhost), and Scientific Electronic Library Online (SciELO) were searched. The search strategy is included in the Supplemental Information and in Supplemental Tables 7 through 11. There were no restrictions on year of publication or gestational age. The initial search was conducted October 17, 2019, and an updated search of the Medline, Embase, the Cochrane Central Register of Controlled Trials, CINAHL, and SciELO was performed on February 29, 2020. In addition, reference lists of included studies were hand screened for potential studies. One study was identified from public comments for the draft CoSTR posted on the ILCOR Web site.

We included randomized controlled trials (RCTs) and ancillary analyses of RCTs, non-RCTs, interrupted time series, and controlled before-and-after studies. Cross-sectional or cohort studies were eligible for inclusion if it was clear how the study population was generated (ie, population level studies or inclusive cohort of all deliveries within a given period).⁷  Studies in all languages were eligible if there was an English abstract. Unpublished studies (eg, conference abstracts, trial protocols) and animal studies were excluded.

Covidence systematic review software (Veritas Health Innovation, Melbourne, Australia) was used for study selection in 2 steps. Two authors (E.E.F. and R.G.) independently screened titles and abstracts for all studies identified in the search. Selected studies were subject to full-text review and assessment. Differences were resolved through discussion and consensus. Two investigators (E.E.F. and R.G.) independently extracted the following data from all included studies: study author, location, year, study design, population, and outcome measures. Study authors were contacted to request missing data.

Two authors (E.E.F. and R.G.) assessed the risk of bias for all included studies. Authors of most included studies reported on single-arm cohorts without a comparison group. Therefore, the risk of bias was assessed for domains adapted from the Risk of Bias in Non-Randomized Studies - of Interventions tool.⁸  These included subject selection, confounding, exposure measurement, outcome measurements, missing data, and selective reporting.

When resuscitation efforts are discontinued before ROSC, newborns rarely survive. Thus, we did not generate risk ratios or other measures of comparison between infants with ongoing CPR beyond 10 minutes and those for whom resuscitation was discontinued. Because the data permitted the calculation of absolute risks at defined time intervals, we aggregated data from all included studies by simple, unweighted summation to describe the impact of ongoing duration of CPR on the probability of the primary and secondary outcomes at different time periods. Because of the heterogeneity of included studies, we did not calculate confidence intervals or attempt to calculate weighted summary estimates for these outcomes. Because not all researchers specified the neurodevelopmental assessment tools used, we performed a sensitivity analysis of neurodevelopmental outcomes excluding studies without a specified tool.

Prespecified subgroup analyses included therapeutic hypothermia postresuscitative care, gestational age ≥36 weeks versus <36 weeks, birth weight ≥2500 versus <2500 g, and infants enrolled in population-level cohort studies. We considered population-level cohort studies as those in which authors reported the outcome of all births, including unsuccessful resuscitations, within the defined population during the specified study period. In a post hoc subgroup analysis, we identified the primary and secondary outcomes for infants with first detectable heart rate or heart rate ≥100 beats per minute reported at or beyond 20 minutes after birth. Subgroups were defined using study-level characteristics, study-specific subgroups, or infant-level characteristics as reported in individual studies.

Using our search, we identified 761 studies (Fig 1). After duplicates were removed, 706 original references were screened and 92 full texts were reviewed for eligibility. Authors of 2 studies presented outcome data from the same cohort^9,10 ; only the report from the latest follow-up was included. Sixteen eligible studies enrolling 579 newborns were included in this review.^10–25 

All studies were conducted in well-resourced settings, representing Australia, Canada, Japan, the United Kingdom, and the United States (Table 1). There was important clinical and methodologic heterogeneity among studies. Enrollment years ranged from 1982 to 2017 and the number of enrolled infants ranged from 3 to 177. Authors of 8 studies only included late preterm and/or term infants, whereas infants across the spectrum of gestation were assessed in the remaining 8 studies. When reported, resuscitation and postresuscitation interventions varied across studies.

Natarajan et al¹⁰  reported on outcomes of a subset of infants enrolled in a trial of therapeutic hypothermia. The remaining studies were single-arm cohort studies that met specified enrollment criteria (Table 1). Six were population-based studies. Patients in the remaining 10 studies were included if they survived to neonatal unit admission or transfer or received an indicated treatment, such as therapeutic hypothermia. In 13 studies, the authors reported neurodevelopmental outcomes of survivors. There was important heterogeneity between, and in some cases within, studies regarding the timing and tools used for these assessments.

Overall, all studies were judged to have a high risk of bias (Table 2). Ten studies were considered to have high risk of bias for subject selection because enrolled patients had to survive resuscitation to a specific entry point to be included. This risk was unclear in the 6 population-level studies because these used variable or unspecified definitions of stillbirths for infants who did not respond to resuscitation. We considered all studies at high risk of bias caused by confounding because no authors provided specific information regarding the decision to discontinue resuscitation. There was potential differential use of cointerventions during resuscitation or postresuscitation, including withdrawal of intensive care, which may have been influenced by knowledge about the duration of resuscitation for enrolled infants.

Studies were at high risk of bias for exposure and outcome measurement. It is unclear how the duration of resuscitation and heart rate were measured. In most studies, authors reported duration of CPR in discrete time intervals instead of in minutes. In addition, infants were included on the basis of the Apgar score, which is subjective, prone to interobserver variability, and does not account for interventions in place.⁴  It was unclear whether physiologic monitors were used for heart rate assessment, but many studies were conducted before 2010 when electronic physiologic monitoring was first recommended for the delivery room setting.²⁶  There was unclear risk of bias for outcome assessment. Although the primary outcome of survival is less susceptible to measurement bias, it is unclear whether assessors of secondary outcome measures were aware of the duration of resuscitation or hospital course. Except for the report by Natarajan et al,¹⁰  the remaining studies were not prospectively registered, leading to unclear risk of bias from missing data or selective reporting. Finally, there is risk of publication bias because these studies may have preferentially represented centers with better outcomes among infants who receive ongoing CPR after birth.

For the primary outcome of survival until last follow-up, in all 16 studies, authors reported survival outcomes of 579 newborns (Table 3). The last documented follow-up ranged from hospital discharge through 12 years of age. Survival rates within studies ranged from 2% to 100%. Among all newborns included, 237 of 579 (40.9%) survived to last follow-up. The vast majority of deaths occurred before discharge. Across 13 studies in which authors reported specifically on survival at hospital discharge, 176 of 432 (40.7%) patients were alive at discharge.

There were 13 studies enrolling 277 infants that reported neurodevelopmental outcomes among survivors (Table 4); 86 (31.0%) infants survived and 79 were assessed for neurodevelopmental outcomes. Of these, 30 of 79 (38%) did not have moderate or severe NDI. When considering the composite outcome among all enrolled infants, 10.8% survived without moderate or severe NDI. In sensitivity analysis, excluding 4 studies in which authors did not specify the assessment tools used, 71 of 216 (33%) infants survived and 29 of 216 (13%) survived without moderate or severe NDI (Supplemental Table 6).

Results from subgroup analyses are shown in Table 5. Subgroup analyses based on birth weight were not conducted because this information was not reported in all studies. Among 131 infants enrolled in population-based studies with neurodevelopmental follow-up, 9 (7%) survived without moderate or severe NDI. Of 105 newborns who survived to NICU admission, received therapeutic hypothermia, and underwent neurodevelopmental assessment, 21 (20%) survived without moderate or severe impairment.

Because of the high risk of bias and inconsistency, the evidence in this systematic review is considered to have low certainty. Given the heterogeneity of included studies and the lack of weighting, the summed outcome estimates should be interpreted cautiously.

The duration of resuscitation that optimizes either survival or neurodevelopmental outcomes after prolonged CPR after birth has not been identified. In this systematic review of 16 studies including 579 newborns with ongoing need for CPR at 10 minutes after birth, 237 (40.9%) patients survived to last follow-up. Among 277 newborns included in 13 studies in which authors reported neurodevelopmental outcomes, 49 (17.6%) survived with moderate or severe NDI, and 30 (10.8%) survived without moderate or severe impairment.

The 2015 ILCOR Neonatal Life Support Task Force reviewed the available evidence on this topic and suggested it may be reasonable to discontinue resuscitation after 10 minutes of resuscitation if the heart rate remains undetectable.¹  Eight studies reviewed in the present analysis, including the single largest population-level cohort study,¹²  were published in 2015 or later. In the present review, the number of survivors without moderate or severe NDI at or beyond 10 minutes of resuscitation suggests that cessation of resuscitation before 10 minutes may preclude survival of some infants who may survive without moderate or severe neurodisability.

Factors that may influence a provider’s decision to discontinue CPR include whether optimal resuscitative efforts were performed throughout the 10-minute period and whether the infant was pulseless or profoundly bradycardic throughout the 10 minutes or just at 10 minutes. No studies included in this review were used to report specific details about the timing or efficacy of resuscitative interventions or information about the presenting cardiac rhythm. In addition, because the onset of bradycardia or pulseless state is frequently before birth, the total duration of compromise may or may not be known but could influence the outcome of resuscitation. Finally, advances in postresuscitative care may improve the probability of survival without neurologic impairment after prolonged CPR after birth.

Equally important is whether all recommended resuscitation interventions have been provided. The time taken to perform all recommended steps, including the administration of ≥1 doses of epinephrine, varies but may be as long as 20 minutes.^23,27–29  The variation in the interval from birth to completion of these steps may depend on the availability, composition, and procedural skill of the resuscitation team. As a result, recommending a single time interval after birth to cessation of intensive resuscitation for all newborns might mean that the full repertoire of resuscitation interventions has not yet been provided before CPR is discontinued.

Extremely limited data are available regarding outcomes after specific durations of CPR beyond 10 minutes after birth. The duration of CPR after which continued efforts no longer improve the probability of intact survival has not been defined. In a retrospective multicenter analysis of registry data including 659 newborns who survived delivery room CPR, lower gestational age and prolonged duration of CPR were independently associated with decreased odds of survival to hospital discharge; however, 25% of survivors received ≥10 minutes of CPR.³⁰  In the present review, we identified outcomes for 39 infants with ROSC or increase in heart rate at or beyond 20 minutes of life. Within this small cohort, 15 (38%) survived to last follow-up and 6 of 15 (40%) survivors did not have moderate or severe NDI. Although these limited data may be influenced by selective reporting, the potential for survival without moderate or severe NDI after 20 minutes of intensive resuscitation is noteworthy.

Another concern is the potential burden of adverse outcomes on infants and their families after hospital discharge. The discharge survival rates were similar to survival rates at last follow-up, suggesting that most deaths occurred in the delivery room or during the initial hospitalization. For those infants who ultimately die during their initial hospitalization, achieving even short-term survival may provide the family the time and opportunity to participate in decision-making and care of their infant. Moreover, among those infants who do survive, survival without moderate or severe NDI is possible. In this systematic review, 38% of survivors assessed in follow-up did not have moderate or severe NDI.

Finally, interpretation of these results should be influenced by the local and regional context. All studies included in this review occurred in high-resource settings. In low-resource settings, where resuscitation may include only face mask ventilation,³¹  advanced resuscitation procedures and prolonging resuscitation may not be an option. In addition, the quality of available antenatal and neonatal care as a whole and access to neuroprotective strategies in particular are likely to influence outcomes after delivery room CPR.³² 

Given these considerations, it is unlikely that 1 uniform time interval or duration of CPR after birth will be appropriate for all newborns. Relevant contextual factors include gestational age, the presence of congenital anomalies, the timing of perinatal insult, the perceived adequacy of resuscitative interventions performed, the family’s stated preferences and values, and the availability of postresuscitative resources such as neonatal intensive care and neuroprotective strategies. In addition, cultural and religious differences, including different perceptions on the value of extending life, the quality of life, and the acceptance of comfort care as an option, may influence this decision.^33–35  ILCOR treatment recommendations based on this evidence are available online,³⁶  and authors of previous work have proposed a practical approach for discontinuing resuscitation.³⁷ 

This systematic review includes a comprehensive assessment of the best available evidence; however, the certainty of evidence is considered very low. The most significant weakness is the potential for selection bias. Only 6 studies included in this review were population-level cohort studies. In the remaining studies, authors reported outcomes of infants who survived resuscitation, were admitted or transferred to the NICU, and in some cases met criteria for therapeutic hypothermia treatment. Because infants who died in the delivery room were not included in those studies, mortality estimates likely underestimate the actual probability of death when all resuscitations in a defined birth cohort are considered.

In addition, the details, efficacy, and timing of resuscitation procedures were not reported in most studies. Furthermore, the impact of prolonged resuscitation in preterm infants is uncertain because of limited data. We reported results of only 144 infants <36 weeks’ gestation, of whom 41 survived and only 8 were assessed for NDI; results should be cautiously applied to preterm infants. Last, in most available studies, authors characterized the infant’s response to resuscitation with the 10 minute Apgar score, which is subjective and does not provide information about ongoing assessments or response to resuscitation beyond 10 minutes.⁴  To address these gaps, additional cohort studies in which authors report outcomes of all infants who present without signs of life and receive intensive resuscitation are needed. Ideally, these would include consistent definitions of the at-risk population, standardized definitions of stillbirths, objective physiologic assessments during CPR, detailed information about the occurrence and timing of resuscitation and postresuscitative interventions, and accurate and consistent assessments of outcomes among survivors.

Infants with ongoing need for CPR at 10 minutes after birth are at high risk for mortality and neurodisability, but survival without NDI is possible. Exposure to 1 specified duration of CPR is unlikely to uniformly predict survival or survival without NDI for all newborns.

Besides the authors Maria Fernanda de Almeida, Ruth Guinsburg, Yacov Rabi, Gary Weiner, Myra Wyckoff, and Jonathan Wyllie, members of the ILCOR Neonatal Life Support Task Force include the following: Khalid Aziz, Jorge Fabres, Joe Fawke, Shigeharu Hosono, Tetsuya Isayama, Vishal Kapadia, Han-Suk Kim, Helen G. Liley, Lindsay Mildenhall, Chris McKinlay, Jeffrey Perlman, Charles Christoph Roehr, Georg Schmölzer, Edgardo Szyld, Daniele Trevisanuto, and Sithembiso Velaphi. We acknowledge the following authors of studies included in this review for sharing additional data as needed: Osuke Iwata, Dr Jun Shibasaki, Thomas Sproat, and Marya Strand.

CINAHL

Cumulative Index to Nursing and Allied Health

CoSTR

consensus on science with treatment recommendations

CPR

cardiopulmonary resuscitation

ILCOR

International Liaison Committee on Resuscitation

NDI

neurodevelopmental impairment

RCT

randomized controlled trial

ROSC

return of spontaneous circulation

SciELO

Scientific Electronic Library Online