ANSeR1 indicates The Algorithm for Neonatal Seizure Recognition Study; BiHiVE1, Biomarkers in Hypoxic-Ischemic Encephalopathy; BiHiVE2, The Investigation and Validation of Predictive Biomarkers in Hypoxic-Ischaemic Encephalopathy; HIE, hypoxic ischemic encephalopathy; and PA, perinatal asphyxia.
Eight of 55 infants (14.5%) diagnosed as having mild HIE and treated with TH were excluded.
eTable 1. Inclusion and Exclusion Criteria for the 4 Cohort Studies
eTable 2. Maternal and Infant Characteristics, Including Children Lost to Follow-up
Customize your JAMA Network experience by selecting one or more topics from the list below.
Identify all potential conflicts of interest that might be relevant to your comment.
Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.
Err on the side of full disclosure.
If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.
Not all submitted comments are published. Please see our commenting policy for details.
Finder M, Boylan GB, Twomey D, Ahearne C, Murray DM, Hallberg B. Two-Year Neurodevelopmental Outcomes After Mild Hypoxic Ischemic Encephalopathy in the Era of Therapeutic Hypothermia. JAMA Pediatr. 2020;174(1):48–55. doi:10.1001/jamapediatrics.2019.4011
Do infants with mild hypoxic ischemic encephalopathy at birth have worse neurodevelopmental outcomes compared with their unaffected peers?
Among 471 patients, this multicenter cohort study found that children with mild hypoxic ischemic encephalopathy at birth have significantly lower cognitive composite scores than healthy controls as measured with the Bayley Scales of Infant and Toddler Development, Third Edition at age 2 to 3 years.
This study’s findings suggest that clinical trials of neuroprotective therapies, including therapeutic hypothermia, are warranted for patients with mild hypoxic ischemic encephalopathy.
Therapeutic hypothermia reduces risk of death and disability in infants with moderate to severe hypoxic ischemic encephalopathy (HIE). Randomized clinical trials of therapeutic hypothermia to date have not included infants with mild HIE because of a perceived good prognosis.
To test the hypothesis that children with mild HIE have worse neurodevelopmental outcomes than their healthy peers.
Design, Setting, and Participants
Analysis of pooled data from 4 prospective cohort studies in Cork, Ireland, and Stockholm, Sweden, between January 2007 and August 2015. The dates of data analysis were September 2017 to June 2019. Follow-up was performed at age 18 to 42 months. In this multicenter cohort study, all children were born or treated at the tertiary centers of Cork University Maternity Hospital, Cork, Ireland, or Karolinska University Hospital, Stockholm, Sweden. In all, 690 children were eligible for this study.
At discharge, all children were categorized into the following 5 groups using a modified Sarnat score: healthy controls, perinatal asphyxia (PA) without HIE, mild HIE, moderate HIE, and severe HIE.
Main Outcomes and Measures
Cognitive, language, and motor development were assessed with the Bayley Scales of Infant and Toddler Development, Third Edition (BSITD-III). The BSITD-III scores are standardized to a mean (SD) of 100 (15), with lower scores indicating risk of developmental delay.
Of the 690 children eligible for this study, 2-year follow-up data were available in 471 (mean [SD] age at follow-up, 25.6 [5.7] months; 54.8% male), including 152 controls, 185 children with PA without HIE, and 134 children with HIE, of whom 14 had died. Infants with mild HIE (n = 55) had lower cognitive composite scores compared with controls, with a mean (SD) of 97.6 (11.9) vs 103.6 (14.6); the crude mean difference was −6.0 (95% CI, −9.9 to −2.1), and the adjusted mean difference was −5.2 (95% CI, −9.1 to −1.3). There was no significant difference in the mean cognitive composite scores between untreated children (n = 47) with mild HIE and surviving children with moderate HIE (n = 53) treated with therapeutic hypothermia, with a crude mean difference for mild vs moderate of −2.2 (95% CI, −8.1 to 3.7).
Conclusions and Relevance
This study’s findings suggest that, at age 2 years, the cognitive composite scores of children with a history of mild HIE may be lower than those of a contemporaneous control group and may not be significantly different from those of survivors of moderate HIE treated with therapeutic hypothermia.
Neonatal hypoxic ischemic encephalopathy (HIE) is one of the most common causes of neonatal death and long-term disability, occurring in 2 to 3 per 1000 newborn children.1-3 It accounts for 6% to 9% of all neonatal deaths and 21% to 23% of deaths in term infants. Globally, HIE is estimated to cause more than 1 million deaths each year.4 Outcome varies considerably depending on the grade of encephalopathy.
Therapeutic hypothermia (TH) is the only neuroprotective treatment proven to improve outcome in HIE and has been shown in randomized clinical trials (RCTs) to reduce risk of cerebral palsy and significant disability in children with moderate and severe HIE.5,6 In all relevant RCTs to date, TH was not offered to infants with mild HIE because of a perceived good prognosis.5,7-9 Recent studies have challenged this perception. Higher-than-expected rates of disability have been reported after mild HIE in cohorts recruited before the introduction of TH.10 To date, no evidence is available to guide clinicians regarding the benefit of TH in mild HIE. Despite this deficit, there has been significant therapeutic drift in some centers,11 where many infants with mild HIE are being offered TH. However, the literature provides little information on outcome in mild HIE, with almost no data available from the era of TH.12,13 The RCTs of TH that have included mild HIE did so inadvertently, when such patients were initially misclassified as having moderate HIE.14 Therefore, these infants may not be representative of the broader mild HIE population. Our aim in this study was to test the hypothesis that children with mild HIE have worse neurodevelopmental outcomes than their healthy peers at age 2 to 3 years in large, prospective cohorts of all grades of HIE (mild, moderate, and severe) in the era of TH.
In this multicenter cohort study, participants were recruited prospectively to 4 sequential cohorts between January 2007 and August 2015 who were born or treated at the following 2 tertiary centers: Cork University Maternity Hospital, Cork, Ireland, and Karolinska University Hospital, Stockholm, Sweden. The dates of analysis were September 2017 to June 2019. These cohorts had been recruited separately at each site in the 2 earlier cohorts (NeoCool and Biomarkers in Hypoxic-Ischemic Encephalopathy [BiHiVE1]) and as part of multicenter studies in 2 later cohorts (The Investigation and Validation of Predictive Biomarkers in Hypoxic-Ischaemic Encephalopathy [BiHiVE2]15 and The Algorithm for Neonatal Seizure Recognition Study [ANSeR1]16). Complete follow-up data were only available for the Cork University Maternity Hospital arm of the ANSeR1 in the time frame of the present study. Therapeutic hypothermia was the standard of care at both sites during the study period. The individual inclusion criteria for all 4 studies (NeoCool, BiHiVE1, BiHiVE2, and ANSeR1) have been previously reported15-19 and are listed in eTable 1 in the Supplement. All 4 cohort studies received approval from the Clinical Ethics Committee of the Cork Teaching Hospitals or the Stockholm Ethical Review Board. Written informed consent was obtained in BiHiVE1, BiHiVE2, and ANSeR1. Oral informed consent was obtained in NeoCool.
All data were collected using a prespecified, electronic clinical research form. Information on maternal demographics, health data, and delivery details were obtained from the maternal medical records. Neonatal data were prospectively collected after birth.
All children in BiHiVE1 and BiHiVE2 were recruited at birth. Cases were defined as having 1 or more of the following: umbilical cord pH less than 7.1, Apgar score not exceeding 6, and need for assisted ventilation and/or cardiopulmonary resuscitation. Infants meeting these criteria were defined as having perinatal asphyxia (PA) and were followed up during the neonatal course for the development of HIE. Control infants with uneventful deliveries, normal umbilical cord pH, normal Apgar scores, and normal newborn examination findings were recruited prenatally and contemporaneously. The neonatal course and results of early neurological examinations were recorded. For all infants with PA in the 4 studies, the worst grade of encephalopathy during the first 24 hours was assigned by the treating physician using a modified Sarnat score without electroencephalographic (EEG) staging for mild HIE.7,20,21 At discharge, the children were categorized into the following 5 groups: healthy controls, PA without HIE, mild HIE, moderate HIE, and severe HIE. Infants who were diagnosed as having neonatal encephalopathy of any cause other than HIE were excluded (Figure 1). All infants with moderate to severe HIE had continuous monitoring with amplitude-integrated EEG or multichannel EEG.
Follow-up was performed at age 18 to 42 months, when certified psychologists or dedicated research fellows assessed cognitive, language, and motor development using the Bayley Scales of Infant and Toddler Development, Third Edition (BSITD-III). The BSITD-III composite scores are standardized to a mean (SD) of 100 (15), with lower scores indicating risk of developmental delay. In accord with other studies,22 we regarded a mean group difference greater than 5 points as clinically important. Cognitive, language, and motor development was considered normal if the BSITD-III composite scores were within 1 SD of the mean of the control group. Children unable to participate in any BSITD-III assessment because of severe motor dysfunction or severe autistic traits were recorded as having a severe outcome, with a composite score of −3 SDs compared with the control group. When a language other than English or Swedish was spoken at home, the language composite score of that child was excluded from analysis.
A post hoc analysis was performed that was limited to those infants who were treated according to current guidelines.23 Therefore, infants diagnosed as having mild HIE treated with TH and infants having moderate to severe HIE who did not receive TH were excluded from the post hoc analysis.
Differences in demographic characteristics between the exposure groups were evaluated with the Kruskal-Wallis test for continuous variables and with the Fisher exact test for categorical and binary variables. Gestational age at birth and birth weight were centered on the study population mean. The mean BSITD-III composite scores and the mean differences with their respective 95% CIs were estimated using linear regression with robust estimates of the SEs. Comparative analyses of the patient groups were related to the control group, and all adjusted analyses were adjusted for country of birth. Stepwise linear regression of the variables HIE grade, country of birth, maternal age, maternal work status, maternal tertiary educational level, gestational age at birth, sex, and birth weight was performed; independent variables with P < .20 were entered into a multivariable regression model. For postestimation comparison of outcome in children with mild HIE vs children with moderate HIE, an F test was used. Cohen d with adjustment for unequal variances was used to estimate effect sizes comparing the difference in the means in children with mild HIE and controls. For binary outcome variables, odds ratios were estimated using unadjusted and adjusted logistic regression with and without adjustment for country of birth. The significance level was set at 2-sided P = .05. For missing covariates, logistic and predictive mean matching was used as appropriate. Chained imputation with 20 imputed data sets was used, including neurodevelopmental outcome, HIE grade, country of birth, and an interaction term containing both exposure and outcome. Differences between children lost to follow-up and children with available outcome data were evaluated using the Fisher exact test and Wilcoxon rank sum test as appropriate. Statistical analyses were performed using Stata statistical software, version 14.2 (StataCorp LLC).
In total, 741 live-born infants were recruited to the 4 cohort studies. Overall, the families of 66 children provided consent to participate in NeoCool, 117 in BiHiVE1, 501 in BiHiVE2 (30 of whom were included in ANSeR1), and 87 in ANSeR1. Fifty-one children were excluded. Six children were excluded because of prematurity (<36 weeks), and 3 children were excluded because of lack of postnatal data. Forty-two children were excluded because their final diagnosis was other than healthy control, PA without HIE, or HIE; of these, 4 had stroke without signs of PA, 9 had sepsis or meningitis, 7 had metabolic disorders, 6 had other genetic disorders, 5 had postnatal collapse, 1 had a surgical condition, and 10 had other causes of encephalopathy (Figure 1).
Of the remaining 690 who were eligible for this study, 471 children (mean [SD] age at follow-up, 25.6 [5.7] months; 54.8% [258 of 471] male) were followed up until at least age 2 years or death, with 14 children dying before age 2 years. Three children (2 with PA without HIE and 1 with mild HIE) could not be assessed because of severe autistic traits. Six children (4 with moderate HIE and 2 with severe HIE) with severe cerebral palsy were deemed too disabled to test in 1 or more scales (cognitive, language, or motor) with the BSITD-III.
Overall, 152 of 471 children (32.3%) were categorized as healthy controls, 185 (39.3%) as having PA without HIE, and 134 (28.5%) as having HIE at birth (55 with mild HIE, 56 with moderate HIE, and 23 with severe HIE). Baseline characteristics are listed in Table 1. Country of birth varied significantly between patient groups, with more who were classified as controls, PA without HIE, and mild HIE born at Cork University Maternity Hospital and a greater proportion with moderate HIE and severe HIE born at Karolinska University Hospital. There was a significant difference across the groups in the median age at BSITD-III follow-up, ranging from 23.1 (interquartile range, 20.9-25.7) months (controls) to 26.6 (interquartile range, 24.9-29.7) months (severe HIE).
Outcome data were available for 471 children. Of these, 449 completed the BSITD-III cognitive scale, 414 completed the motor scale, and 381 completed the language scale (the language scales of 29 children were excluded because alternative languages were spoken at home). Six children were assigned a severe outcome in 1 or more of the 3 scales because of severe motor dysfunction and 3 children because of severe autistic traits.
The BSITD-III composite scores for all groups are listed in Table 2. The crude mean (SD) cognitive composite scores were 103.6 (14.6) for the control group, 102.6 (15.7) for PA without HIE, 97.6 (11.9) for mild HIE, 98.4 (18.1) for moderate HIE, and 88.3 (19.0) for severe HIE.
Comparisons of composite scores between the controls and the mild HIE group are summarized in Table 3. There was a significant difference in the mean cognitive composite score in children with mild HIE (−6.0; 95% CI, −9.9 to −2.1), with a Cohen d effect size of −0.43 (95% CI, −0.12 to −0.74). When adjusted for country of birth, the mean difference compared with the control group was −4.9 (95% CI, −8.7 to −1.2) in the mild HIE group. There was no significant difference in the means in any of the BSITD-III scales between the children with mild HIE and the children with moderate HIE, with a crude mean difference in the cognitive composite score of −2.2 (95% CI, −8.1 to 3.7).
A stepwise regression analysis was performed that included country of birth, maternal age, maternal work status, maternal tertiary educational level, sex, and birth weight. Maternal age, maternal tertiary educational level, and gestational age at birth were excluded because of no significant association with the outcome. After multivariable regression analysis that included 392 children with complete data, the difference in the mean cognitive composite score of children with mild HIE compared with controls was −5.2 (95% CI, −9.1 to −1.3). The proportion of variance explained (R2) in the full model was 0.14. Multiple imputation was used for missing covariates in the multivariable regression analysis. One value of birth weight and 65 values of maternal work status were imputed. The difference in the mean cognitive composite score between mild HIE vs controls remained significant at −4.5 (95% CI, −8.3 to −0.7).
In total, 219 of 690 patients (31.7%) were lost to follow-up. We found no significant difference in umbilical cord pH, Apgar score, and maternal work status between children lost to follow-up and children with available outcome data in the control and mild HIE groups. There was a significant difference in maternal tertiary educational level within the control group (eTable 2 in the Supplement).
Survival with a normal cognitive composite score was seen in 92.8% (141 of 152 of controls, 88.1% (163 of 185) of children with PA without HIE, 85.5% (47 of 55) of children with mild HIE, 76.8% (43 of 56) of children with moderate HIE, and 39.1% (9 of 23) of children with severe HIE. Rates of intact survival (composite scores within 1 SD below the mean of the control group) were comparable in controls (76.3% [116 of 152]), PA without HIE (74.6% [138 of 185]), and mild HIE (74.5% [41 of 55]). Intact survival was seen in 67.9% (38 of 56) of children with moderate HIE and in 21.7% (5 of 23) of children with severe HIE.
There was no significant difference in composite scores in any BSITD-III scale between the 8 children with mild HIE treated with TH compared with the 47 nontreated children with mild HIE. Their crude mean differences in scores were 7.2 (95% CI, −3.1 to 17.4) for the cognitive composite score, 6.8 (95% CI, −5.2 to 18.8) for the language composite score, and 13.5 (95% CI, −1.3 to 28.3) for the motor composite score.
After excluding the children with severe autistic traits, the crude mean difference in the cognitive composite score compared with the control group was −5.3 (95% CI, −9.0 to −1.6) for the children with mild HIE. When adjusted for country of birth, the mean difference was −4.2 (95% CI, −7.8 to −0.7).
A post hoc analysis was performed to adjust for therapeutic drift and to adjust for the possible association of TH with outcome in mild HIE. This analysis was limited to infants treated according to current guidelines. Eight of 55 infants (14.5%) diagnosed as having mild HIE and treated with TH were excluded, as were 5 of 79 infants (6.3%) diagnosed as having moderate to severe HIE not treated with TH. The crude mean difference in the cognitive composite score in children with mild HIE compared with the control group was −7.0 (95% CI, −11.0 to −3.0), with an effect size of 0.51 (95% CI, 0.17-0.84). The distribution of cognitive composite scores in both groups is shown in Figure 2. Again, there was no significant difference in the means in any of the BSITD-III scales between the infants with mild HIE not treated with TH and the infants with moderate HIE treated with TH. Their crude mean differences in scores were −2.2 (95% CI, −8.1 to 3.7) for the cognitive composite score, 0.2 (95% CI, −7.5 to 8.0) for the language composite score, and −2.5 (95% CI, −9.3 to 4.3) for the motor composite score.
We have observed that children with mild HIE have significantly lower cognitive composite scores measured with the BSITD-III compared with a healthy control group. As our group has previously observed in a pre-TH cohort,10 untreated children with mild HIE have higher rates of intact survival, and herein we find no difference in performance compared with surviving children with moderate HIE treated with TH.
Handley-Derry et al9 compared children with PA without HIE vs healthy controls and observed (similar to our results) no significant difference between these 2 patient groups on the BSITD-III. Similarly, Robertson and Finer24 found no increased risk for outcomes, but they did not investigate minor disabilities. In another study,8 Robertson et al found lower IQs as measured with the Wechsler Intelligence Scales at age 8 years in children with mild HIE at birth, with an IQ of 106 compared with 112 in healthy peers; however, this difference was not statistically significant. The conclusion since then has been that children with mild HIE have normal developmental outcomes. Similarly, our study indicates that children with mild HIE have developmental scores within the normal range, but our findings support the increasing amount of evidence that children with mild HIE are not following the same developmental trajectory as their peers.10,11,14,25-29 Our group has earlier observed that more than 70% of children with moderate HIE have learning difficulties at late adolescence,30 suggesting that subtle disabilities at an early age might become more significant over time. In this study, we have only assessed these children at age 2 to 3 years. Longer follow-up is needed to investigate how these subtle disabilities will influence the school-age performance of these children.
In the pre-TH era, HIE grading was based on the worst grade of encephalopathy recorded during the first days of life. Since the introduction of TH, HIE grading has been needed within the first 6 hours of life to decide if treatment should be commenced. Pressure to decide eligibility within this therapeutic window has led to a therapeutic drift to lower grades of HIE11 and other patient groups31 being treated with TH. Gagne-Loranger et al13 have reported that 15% of infants with mild HIE referred to a tertiary center were treated with TH. On a recent survey in the United Kingdom, 75% of centers had at times treated infants with mild HIE with TH.32 The grade of encephalopathy is often difficult to assess in the first postnatal hours. As the grade of encephalopathy evolves over time, children with an initial diagnosis of mild HIE might progress to develop moderate HIE.11 However, none of the nontreated children with mild HIE worsened after 24 hours in our study.
At present, clinicians are operating in an evidence vacuum, with indications (but no clear evidence) of the benefit of TH in this patient group. A large, adequately powered trial with an appropriate control group is needed to assess the consequences, if any, of TH in mild HIE.
Our study has some limitations. The major limitation is the number of children lost to follow-up and with missing data (eTable 2 in the Supplement), although the difference in the mean cognitive composite score remained significant after multiple imputation of missing covariates in the multivariable model. In addition, mothers of controls lost to follow-up had a lower grade of tertiary education. However, the difference in the mean cognitive composite score in mild HIE vs controls remained significant after post hoc adjustment for maternal tertiary educational level. While the different maternal work status in the control group did not change our findings, it is an important factor to consider in future research. Motivation in returning for follow-up may differ between cases and controls, with socioeconomic factors having greater consequences on retention in a control group.
Another limitation of our multicenter study is the use of different individuals to administer the BSITD-III. However, all assessors were trained in the administration of the BSITD-III and completed 5 observed deliveries before any actual study assessment. In addition, data were not available on individual assessor performance for comparison. Instead, country of birth was used as a proxy for investigator performance. Adjustment for country of birth did not significantly alter our estimates.
To our knowledge, few prospective studies of outcome in HIE have included children with mild HIE, with most focusing on moderate or severe infants only. In our study, children with signs of PA were recruited prospectively without knowledge of subsequent development of HIE to assess the full spectrum of PA at birth. This multicenter study is the largest cohort to date of children with HIE in which outcome has been assessed in all grades of HIE, providing the most robust outcome data available since the introduction of TH as the standard of care. Our findings suggest that these children should be monitored more closely and could be considered for future neuroprotective trials.
We have observed that infants who develop mild HIE at birth have impaired cognitive outcomes as measured with the BSITD-III compared with their healthy peers. However, there was no difference in cognitive outcomes in children with mild HIE compared with surviving infants with moderate HIE who have been treated with TH. Clinicians have little guidance for the management of infants with mild HIE, and neuroprotective trials are needed.
Accepted for Publication: June 20, 2019.
Published Online: November 11, 2019. doi:10.1001/jamapediatrics.2019.4011
Correction: This article was corrected on January 21, 2020, to add a new reference 12, renumber original reference 12 as new reference 14, renumber the remaining references, and correct all affected reference citations.
Corresponding Author: Deirdre M. Murray, MD, PhD, Department of Paediatrics and Child Health, Cork University Hospital, Cork T12 DC4A, Ireland (email@example.com).
Author Contributions: Dr Finder had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Finder, Boylan, Murray, Hallberg.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Finder, Boylan, Twomey, Hallberg.
Critical revision of the manuscript for important intellectual content: Finder, Boylan, Ahearne, Murray, Hallberg.
Statistical analysis: Finder, Murray, Hallberg.
Obtained funding: Boylan, Murray, Hallberg.
Administrative, technical, or material support: Finder, Boylan, Twomey, Murray, Hallberg.
Supervision: Boylan, Murray, Hallberg.
Conflict of Interest Disclosures: None reported.
Funding/Support: Biomarkers in Hypoxic-Ischemic Encephalopathy (BiHiVE1) and The Investigation and Validation of Predictive Biomarkers in Hypoxic-ischaemic Encephalopathy (BiHiVE2) were funded by the Health Research Board Ireland (HRB-CSA/2012/40) and by a Science Foundation Ireland Research Centre Award (INFANT-12/RC/2272). The Algorithm for Neonatal Seizure Recognition Study (ANSeR1) was supported by a Wellcome Trust Strategic Translational Award (098983).
Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Meeting Presentation: This paper was presented at the 7th Congress of the European Academy of Paediatric Societies; November 2, 2018; Paris, France.
Additional Contributions: The following individuals collected follow-up data and contributed to quality control: research nurses Michaela Melakari, MSc, and Anna von Vultée, MSc, psychologist Peter Lagerroos, MSc, physiotherapist Mimmi Eriksson Westblad, MSc, pediatrician Katarina Grossmann, MD, trial monitor Jackie O’Leary, MSc, and project manager Jean Conway, MSc. Matteo Bottai, ScD, and the Biostatistics Core Facility at Karolinska Institutet provided biostatistical support. Mr Lagerroos and the Biostatistics Core Facility were compensated through the research funding for their contribution. We acknowledge the nursing staff at Cork University Maternity Hospital and Karolinska University Hospital, without whom this study would not have been possible. We thank the children and their parents for participating in the included studies.
Create a personal account or sign in to: