Newborn Health and Child Mortality Across England

Key Points Question Is neonatal illness associated with later childhood mortality? Findings In this cohort study of 4829 children who died before age 10 years, 71.6% had evidence of neonatal illness, with increased mortality across a range of diseases and underlying conditions. Meaning These findings suggest that neonatal events may be associated with death before age 10 years and that improvements to perinatal health could have positive impacts on subsequent child health.


Introduction
The death of a child soon after birth is one of the biggest contributors to childhood mortality in the developed world, with 42% of all child deaths in England occurring within 28 days of birth. 1 Although the association of preterm birth (for babies born earlier than 37 weeks' gestation) with perinatal brain injury is well recognized, 2 the wider outcomes on childhood mortality and the roles of specific diseases across childhood are unclear.Perinatal disease is associated with sociodemographic characteristics, 3 but there is little evidence from recent data suggesting that there have been improvements in the incidence of either preterm birth 4 or brain injury. 5In addition, the COVID-19 pandemic did not appear to have a measurable impact on neonatal mortality in England. 6ta collection for deaths near the time of birth is common in many countries, but these collections are rarely able to quantify the population-level outcomes across childhood.In contrast, the National Child Mortality Database (NCMD) collects data on all children who die before their 18th birthday, with data supplemented from health and social care records.Notification of deaths in England is required by law, and additional data come from detailed review of the circumstances of death by the Child Death Overview Panels (CDOPs).Two recent NCMD reports 1,7 highlighted the outcomes of preterm birth across childhood 1 and the clear associations of neonatal illness with later childhood death, with potentially modifiable factors identified in more than one-third of cases. 7This work is an extension of these reports and aims to quantify the population outcomes of birth events and neonatal conditions in England and their contribution to childhood mortality.We included all deaths of children before their tenth birthday reported to NCMD that occurred between April 1, 2019, and March 31, 2021 (24 months), after being born at or after 22 weeks' gestation.NHS numbers were used to obtain records from BadgerNet, an electronic patient data management platform used by neonatal units in the UK after 2009; deaths were restricted to those before their tenth birthday to ensure reliable data linkage.If a record existed, the child was considered to have been admitted to a neonatal unit after birth, and the discharge summary for each episode of care was returned to NCMD.Text strings containing diagnoses or possible causes of death were combined.

Methods
Baseline data were identified from the death notification, including sex, racial and ethnic group (Asian or Asian British, Black or Black British, White, multiracial, other [ie, Arab or any other ethnic group], 8 or unknown), age at death (categorized as <1 year, 1-4 years, and 5-9 years), region where the death was reported, and the Index of Multiple Deprivation, a measure of local deprivation 9 with a score of 1 to 10, with a lower value suggesting greater deprivation.Data on race and ethnicity were included because of their known associations with perinatal disease.

Neonatal Conditions
Infants who received care in a neonatal unit after birth and those who died on the first day of life were considered to have evidence of neonatal illness after birth.In addition, evidence of 3 specific neonatal conditions was identified using data from the death notification form, CDOP review, or BadgerNet record.First, preterm infants were defined as those born before 37 weeks' gestation, plus those with a coded admission definition of preterm in the BadgerNet fields or text strings (  5 preterm birth, 12 and congenital abnormalities 13 were derived from published sources and were further modeled using observed frequencies of death in younger children.A Poisson model was used to derive the relative risk (RR) and 95% CIs compared with children without the condition.This analysis was repeated for each age category and for each category of death.In addition, for each age category, we also derived the population-attributable risk fraction (PAF) and its 95% CI from the frequencies seen and the underling population estimates, using the Stata command punaf.To triangulate the possible perinatal causes and contributing conditions of death, we then identified from the child's death certificate the incidence of perinatal pathologies across the whole cohort of children using text codes (eTable 1 and eTable 2 in Supplement 1) as evidence of a contributing disorder.

Underlying Conditions
NCMD-notified deaths were linked to NHS HES admitted patient care data, using the NHS number, to identify exiting comorbidities.We compared the proportion of deaths of children with previous neonatal illness with the proportion of those without for the presence of preceding chronic condition.Three a priori categories were investigated: behavioral or developmental disorders, chronic neurological disease (ie, cerebral palsy, epilepsy, neurological congenital anomalies, and other neurological disease), and chronic respiratory disease (ie, asthma and chronic lower respiratory disease, respiratory congenital anomalies, and other respiratory disease).Relevant ICD-10 codes are shown in eTable 3 in Supplement 1.The relative odds of having 1 of the chronic conditions for each neonatal category compared with those without evidence for such a condition were derived using a random-effects logistic regression model, with age at death as the grouping variable, to allow for the different likelihoods of diagnosis of age-dependent conditions (eg, cerebral palsy) alongside the different age profiles of the exposure and those who were unexposed.

Primary Category of Death
The final category of death was identified from data received from the CDOPs in England, which perform a detailed review of the circumstances of death in all cases.The review process is statutory and can take months to complete; thus, the primary category of death was available for only a subset of 2484 (51.4%) children (eTable 4 in Supplement 1).Because of the small numbers, the 12 children who died of suicide or deliberate self-inflicted harm, deliberately inflicted injury, abuse, or neglect were excluded from this analysis and are not presented.A Poisson model was used to derive the RR comparing children with vs those without neonatal illness.
Comparisons were made using the χ 2 or likelihood ratio test (as appropriate), and P < .05 was considered evidence of statistical significance.Data flow for the different analyses is summarized in the eFigure in Supplement 1.

Results
A years.Neonatal illness was associated with sex, multiple births, and race and ethnicity for infant deaths and with multiple births for deaths occurring between ages 1 and 4 years (Table 1).The characteristics of deaths occurring between age 5 and 9 years were not patterned by neonatal illness, but the numbers and, hence, the precision were lower.There was weak evidence for differences in the proportion of deaths associated with neonatal illness by region of England for deaths before age 1 year and between ages 1 and 4 years.A total of 13.9% of all child deaths (672 deaths) before age 10 years occurred before admission to neonatal units; however, even for deaths between 1 and 9 years, 373 children (33.9%) had been admitted to a neonatal unit after birth.

Discussion
The findings of this cohort study reveal significant associations of neonatal illness with childhood mortality up to age 10 years.Nearly 14% of all child deaths before age 10 years occur before admission to neonatal units, and most deaths (82.7%) before 1 year are associated with evidence of neonatal illness.However, even for deaths between 1 and 9 years, 33.9% of children had been admitted to a neonatal unit after birth.Although the absolute risk of death is low, children who are admitted to neonatal units continue to have higher risk of death than others throughout the next 10 years, with 62.5% having an identifiable perinatal condition listed as a cause of or contributor to their deaths (the most common being prematurity and congenital abnormalities), comparable to the estimated PAF of 66.4% derived from the regression analysis.Children with evidence of neonatal illness had increased risks of death from a range of causes, including infections and SUDIC.
Although this work was only able to quantify mortality, similar trends and outcomes are seen with disability. 2 Preterm birth is common, 14 and, if the association seen here is causal, it may contribute to 1 in 3 deaths before age 10 years.We know that preterm birth is associated with sociodemographic characteristics, 3 suggesting that there may be a component of potential preventability.However, in the UK, we have seen little change in the prevalence of preterm birth over the last decade. 4Although we have therapies to optimize outcomes after preterm birth, [15][16][17] clinical use is inconsistent. 18Congenital abnormalities also represent a major factor in these deaths, and although such abnormalities are heterogeneous, many may also be avoidable, 19 and population-level interventions can empower affected families and reduce unexpected affected births. 20 addition to preterm birth itself, brain injury around birth, both in term and preterm infants, appears to be associated with childhood mortality: 12.2% of children in this study had evidence of brain injury (7.8% with HIE and 4.5% with ICH) around birth.For children who had been admitted to a neonatal unit and died after age 1 year, most had evidence of chronic neurological disease.Although there have been ambitions to reduce the prevalence of chronic neurological disease, 21 few changes have been seen over recent years. 5No novel neuroprotection therapies have been implemented in the last decade, and given the broad impact seen here and the financial 22 and personal outcomes, 23 such new therapies are urgently needed.
Interpretation of any regional variation is difficult within this work but warrants further study.
Patterns appear to be different between age groups, but region-specific demographics (eg, preterm birth rates) are needed to guide interpretation further.

Limitations
This study has limitations that should be mentioned.Data used in this work were derived from statutory data reported to NCMD and were linked to routine patient electronic record data used by all neonatal units in England.Although previous work 24,25 has shown good validation and coverage of both data sources, it is likely that some linkage failed (ie, we were unable to identify a BadgerNet record for a child death).In addition, the population estimate for neonatal illness is based on neonatal intensive care unit admissions, which does not include those who died before their transfer, although they represent a small proportion of the estimated population at risk, and so any overestimate of risk is likely small.Our case definitions for preterm birth, HIE, and congenital abnormalities may also have missed children who were not identified in our data sources (eg, missing gestational age data).For the causes of death analysis, which was designed to support the findings of the main analysis and was derived from the death certification data, we were limited to text searches, and misspellings or misclassifications may have introduced some bias.However, any missing data bias may be conservative, because it likely relates to missing diagnoses or evidence of neonatal care (rather than false-positive cases); any bias from nonascertainment of the exposure is, therefore, likely to underestimate the point estimates presented here.We also limited the work to infants born at or after 22 weeks' gestation, as recognition that registration below this age is likely inconsistent, 26 and further work including stillbirths in any measures of population impact would be useful.For the analysis of underlying comorbidities, we were able to link HES data to 90.6% of cases, and CDOP reviews had been performed for only 51.4% of the cohort, so interpretation of these sections should bear this in mind.In addition, our estimates of at-risk populations are derived used differing methods, and so absolute measures of risk should be interpreted cautiously; for one analysis (of category of death), we were limited to a subset of deaths reviewed by the CDOP panel, and so absolute risks were not derivable.

Conclusions
Although the outcomes of neonatal events are most obvious shortly after birth, they can continue to reverberate across the first decade of childhood.In this cohort study, most children who died before age 10 years had some evidence of a neonatal disorder, and they died of a range of causes, including infections and SUDIC.Improvements to perinatal health, preterm births, and reductions to the number of children with brain injury around birth, all areas with an evidence base for improvement, would likely have impacts on child health across the next decade.

Population
This cohort study was reviewed by the Chair of the Central Bristol National Health Service (NHS) Research Ethics Committee, who confirmed that NHS ethical approval, including obtaining individual consent, was not required because this work is classified as Usual Practice in Public Health and Health Protection.This work follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for observational studies.

JAMA Network Open | Pediatrics
details, suspected cause of death, and cause of death [1a, 1b, or 2]) if death occurred in the first day of life, or from the CDOP review cause of death (immaturity or prematurity).Children without evidence of premature birth were assumed to not be preterm.Second, infants were identified as having likely hypoxic-ischemic encephalopathy (HIE) from text codes (eTable 1 in Supplement 1) or from the CDOP review final category of cause of death (ie, perinatal asphyxia or HIE).Third, congenital anomalies were identified as per the dedicated field within the BadgerNet ie, principal diagnosis at discharge, neurological diagnosis, cause of death [1a, 1b, or 2], principal reason JAMA Network Open.2023;6(10):e2338055. doi:10.1001/jamanetworkopen.2023.38055(Reprinted) October 17, 2023 2/12 Downloaded From: https://jamanetwork.com/ on 11/10/2023 for admission, and cause of death) (eTable 1 in Supplement 1) in the discharge record, or NCMD codes (ie, notification

Table 4 )
childhood deaths were necrotizing enterocolitis (277 deaths[5.7%])andintracranialhemorrhage(ICH) (215 deaths[4.5%]).Preterm birth also caused or contributed to 21.4% of all deaths of children with HIE (81 deaths) and 26.8% of deaths of children with congenital abnormalities (557 deaths).Overall, 62.5% of all deaths (3016 deaths) had 1 of the key 7 perinatal conditions, and 547 (11.3%) had either HIE or ICH listed as a cause of or contributor to death. .This profile was also seen for children born preterm, those with HIE, and those with congenital abnormalities.The increased risk of neurological disease was seen in all subgroup comparisons, with the exception of preterm birth, which was not associated with epilepsy (OR, 0.96; For all children born in England, the risk of death before age 10 years was low (4829 deaths per 19 296 082 children [0.025%]), although for children who had neonatal illness after birth, the risk was 1.6% (3082 deaths per 190 478 children) in the first year of life (Table2).For infants with evidence of neonatal illness, the RR of death was much higher than that for those without neonatal illness (RR, 13.82 [95% CI, 13.00-14.71]).When stratifying the risk of death over the 3 age categories, infants with evidence of neonatal illness had increased risks throughout the first decade, although 5-9 years, 14.1% [95% CI, 9.0%-18.8%]).A similar profile was seen in children identified with specific neonatal conditions.When reviewing the recorded causes of death on the death certificate, 33.4% of all deaths (1614 of 4829 deaths) had preterm birth listed as a cause or contributing factor; 7.8% (376 deaths) had HIE, and 1233 (25.5%) had a congenital abnormality (Table 3).The next highest contributors for JAMA Network Open | Pediatrics Newborn Health and Child Mortality Across England JAMA Network Open.2023;6(10):e2338055. doi:10.1001/jamanetworkopen.2023.38055(Reprinted) October 17, 2023 4/12 Downloaded From: https://jamanetwork.com/ on 11/10/2023

Table 1 .
Proportion of Deaths Among Children Younger Than 10 Years in England With Evidence of Any Neonatal Illness, by Demographics and Age at Death, April 2019-March 202195% CI, 0.70-1.32).The increased risk of respiratory disease across all groups was associated with increases in respiratory congenital abnormalities and perinatally acquired lung disease (especially bronchopulmonary dysplasia), but was not associated with higher risks of asthma or other chronic lower respiratory illnesses.
cAbbreviation: NA, not available.aRefersto death on the first day after birth or admission to neonatal care.bP values were derived using χ 2 for differences in frequency between those with evidence of neonatal illness and those without for that characteristic.cSmall numbers (<6) are suppressed to prevent identifiability.d Other ethnicity includes Arab or any other racial or ethnic group.JAMA Network Open | Pediatrics Newborn Health and Child Mortality Across England JAMA Network Open.2023;6(10):e2338055. doi:10.1001/jamanetworkopen.2023.38055(Reprinted) October 17, 2023 5/12 Downloaded From: https://jamanetwork.com/ on 11/10/2023

Table 2 .
Risk and PAF of Death Before Age 10 Years in England, by Any Neonatal Illness or Specific Neonatal Condition and by Age at Death, April 2019-March 2021 Abbreviations: PAF, population-attributable risk fraction; RR, relative risk.aRefers to death on the first day after birth or admission to neonatal care.

Table 3 .
Recorded Causes of and Contributors to Childhood Death Before Age 10 Years, by Any Neonatal Illness or Specific Neonatal Condition and Age at Death, April 2019-March 2021

Table 5 .
Risk of Death Before Age 10 Years in England With or Without Neonatal Illness, by Cause of Death Identified at Child Death Overview Panel Review, April 2019-March 2021 a Refers to death on the first day after birth or admission to neonatal care.JAMA Network Open | PediatricsJAMA Network Open.2023;6(10):e2338055. doi:10.1001/jamanetworkopen.2023.38055(Reprinted) October 17, 2023 8/12 Downloaded From: https://jamanetwork.com/ on 11/10/2023 Text Used for Identification of Neonatal Conditions eTable 2. Text Used for Identification of Additional Neonatal Conditions eTable 3. ICD-10 Codes for Specific Chronic Conditions eTable 4. Name and Description for Each Category of Death on the Child Death Review Analysis Form in Hierarchical Order eFigure.Data Flow for the Different Analyses