The contribution of hypertensive disorders of pregnancy to late preterm and term admissions to neonatal units in the UK 2012–2020 and opportunities to avoid admission: A population‐based study using the National Neonatal Research Database

Abstract Objective To quantify maternal hypertensive disorder of pregnancy (HDP) prevalence in late preterm and term infants admitted to neonatal units (NNU) and assess opportunities to avoid admissions. Design A retrospective population‐based study using the National Neonatal Research Database. Setting England and Wales. Population Infants born ≥34 weeks’ gestation admitted to NNU between 2012 and 2020. Methods Outcomes in HDP infants are compared with non‐HDP infants using regression models. Main outcome measures Hypertensive disorder of pregnancy, primary reason for admission, clinical diagnoses and resource use. Results 16 059/136 220 (11.8%) of late preterm (34+0 to 36+6 weeks’ gestation) and 14 885/284 646 (5.2%) of term (≥37 weeks’ gestation) admitted infants were exposed to maternal HDP. The most common primary reasons for HDP infant admission were respiratory disease (28.3%), prematurity (22.7%) and hypoglycaemia (16.4%). HDP infants were more likely to be admitted with primary hypoglycaemia than were non‐HDP infants (odds ratio [OR] 2.1, 95% confidence interval [CI] 2.0–2.2, P < 0.0001). 64.5% of HDP infants received i.v. dextrose. 35.7% received mechanical or non‐invasive ventilation. 8260/30 944 (26.7%) of HDP infants received intervention for hypoglycaemia alone (i.v. dextrose) with no other major intervention (respiratory support, parenteral nutrition, central line, arterial line or blood transfusion). Conclusions The burden of maternal HDP on late preterm and term admissions to NNU is high, with hypoglycaemia and respiratory disease being the main drivers for admission. Over one in four were admitted solely for management of hypoglycaemia. Further research should determine whether maternal antihypertensive agent choice or postnatal pathways may reduce NNU admission.


| I N TRODUC TION
Infant admission to a neonatal unit (NNU) often leads to separation of mother and baby during a critical period of bonding. 1 Previous studies have estimated that one in five term admissions to NNU could be avoided. 2The most common primary reasons for term admissions to NNU are respiratory disease (~24% of admissions), infection (~18% of admissions) and hypoglycaemia (~10% of admissions). 24][5][6][7] Whereas the high perinatal risk of hypertensive disorders is well described, the burden presented on NNU at a population level in the UK, and opportunities to avoid NNU admission in this cohort are not well characterised.
In comparison with infants born to normotensive mothers, infants born to mothers with hypertensive disorders of pregnancy (HDP) have a higher risk of preterm birth, 5,8 with concomitant risks of prematurity including respiratory morbidity. 96][17] Whereas antenatal corticosteroid use to reduce respiratory morbidity in preterm infants is well established, 18 strategies to reduce infant hypoglycaemia in the context of HDP are lacking.
As previous studies have demonstrated that almost 100% of infants born at 25-33 weeks' gestation are admitted to NNU for management of prematurity, 19 infants born at ≥34 weeks' gestation were the focus of this study where there may be opportunities to avoid NNU admission.We aimed to quantify the contribution of maternal HDP on NNU at a population level and identify opportunities to avoid NNU admission among late preterm and term infants.

| M ET HODS
This was a retrospective cohort study using prospective, deidentified, routinely recorded neonatal electronic health record data from the National Neonatal Research Database (NNRD, Research Ethics Service approval [10/ H0803/151]).The study was prospectively registered (Clinical Trials registration: NCT05015049) and received ethical approval (21/ES/0061).Findings are reported in line with RECORD guidelines. 20Data items are limited to what is held in the NNRD, but outcomes have been chosen to align with the neonatal core outcome set where possible. 21Parents were not involved in the development of this study.

| Data sources
The NNRD has complete coverage of infants admitted to National Health Service (NHS) NNU in England and Wales since 2012. 19High completeness and accuracy (>95%) of neonatal data held in the NNRD has been confirmed by formal comparison with data from a multi-centre, randomised placebo-controlled trial. 19Population-level Office of National Statistics (ONS) England and Wales annual livebirth data by gestation were accessed for denominator data. 22

| Study population
Infants born between 1 January 2012 and 31 December 2020 at ≥34 weeks' gestation admitted to an NHS NNU in England or Wales were eligible for inclusion in this study.Infants with a confirmed congenital abnormality, those with implausible birthweight Z-scores (≤−4 or ≥4), those with unspecified location of death (NNU versus home or hospice), and those with missing final discharge destination were excluded.We defined late preterm infants as those born between 34 +0 and 36 +6 weeks' gestation and term infants as those born at ≥37 weeks' gestation.

| Exposure
To identify infants of mothers with HDP, maternal HDP diagnoses were extracted from antenatal, delivery and neonatal NNRD data items (Table S1).A single, mutually exclusive diagnosis was assigned (Table S2), with aggregation of diagnoses into three groups with shared pathophysiology for analysis: (i) 'gestational hypertension' -gestational hypertension or HDP not otherwise specified (maternal antihypertensive agent exposure in labour), (ii) 'pre-eclampsia' -pre-eclampsia or superimposed pre-eclampsia (pre-eclampsia in women with chronic hypertension) and (iii) chronic hypertension.

| Outcomes
The primary outcome was primary reason for admission (single reason permitted on electronic health record system) extracted from the infant's first NNU admission.Secondary outcomes were survival, clinical diagnoses (ascertained from diagnosis codes across total NNU stay) and resource use (ascertained from daily data across total NNU stay).Daily level of care was determined according to Health Resource Group (HRG) definitions 23 based on category of care received. 24ajor NNU interventions were defined as any form of respiratory support, parenteral nutrition, central line, arterial line or blood transfusion.Primary management of hypoglycaemia alone was defined as receipt of i.v.dextrose and no other major intervention on infants cared for on special care (SC) only.We used number of days of i.v.dextrose as a proxy for hypoglycaemia severity.As this was highly skewed (median 2 days, range 1-121 days), we categorised this to a binary outcome variable of <3 days and ≥3 days.Birthweight Z-scores were calculated with UK90 reference data using the LMS2z function of the SITAR package. 25Derived variable extraction procedures are listed in Table S3.

| Statistical methods
Population proportions are reported with 95% confidence intervals (CIs) calculated using the exact binomial test.Differences in characteristics, outcomes and resource use between HDP and non-HDP infants are described, with use of unadjusted and adjusted linear and logistic regression models for continuous and binary outcomes, respectively, and chi-square tests for formal testing of differences.
To assess opportunities to avoid admissions in HDP infants, the intersection of maternal HDP status with currently recognised risk factors for infant hypoglycaemia as specified in the 2017 British Association of Perinatal Medicine (BAPM) Infant Hypoglycaemia guidelines 26 (maternal diabetes, birthweight <2nd centile, late preterm birth) is reported.In addition, unadjusted and adjusted logistic regression models were used to test the association between maternal HDP, recognised infant hypoglycaemia risk factors and severity of hypoglycaemia (≥3 days of i.v.dextrose administration).
For all models, highly skewed variables violating model assumptions were log-transformed.Missing data are reported in descriptive tables.All data processing and analysis was performed in R version 4.1.3(10 March 2022). 27

| Characterisation of HDP infants
Infant antenatal and delivery characteristics are shown in Tables 1 and S4.HDP infants were born at earlier gestational ages than non-HDP infants, with pre-eclampsia infants being most likely to be born preterm (birth on average 12.9 days earlier than non-HDP infants [95% CI 12.6-13.2days]).Compared with non-HDP infants, HDP infants also had higher odds of being small for gestational age (birthweight <10th centile), with the highest risk in the pre-eclampsia subgroup (odds ratio [OR] 2.7, 95% CI 2.6-2.8).HDP infants were more likely to be born to women of black ethnic backgrounds in multivariable models adjusted for maternal age, parity and index of multiple deprivation compared with non HDP infants (aOR HDP infant black versus white ethnic background: 1.8, 95% CI 1.7-1.9,P < 0.0001).
A total of 71.8% (95% CI 7.1-7.2) of HDP infants were born following obstetric intervention (pre-labour caesarean section or induction of labour) compared with 41.1% (95% CI 41.0-41.2) of non-HDP infants, with the highest proportion in the pre-eclampsia subgroup (80.3%, 95% CI 79.6-81.1).HDP infants also had a high rate of emergency caesarean section (which includes expedited but non-urgent caesarean sections) overall, with the highest rate in the pre-eclampsia subgroup (Table 1).

| Primary reasons for admission in HDP infants
Primary recorded reasons for admission are illustrated in Figure 1C and listed in Tables S5 and S6.The most common primary reason for admission in HDP infants overall was respiratory disease (27.8%, 95% CI 27.3-28.3),followed by prematurity (22.3%, 95% CI 21.8-22.8)and hypoglycaemia (16.1%, 95% CI 15.7-16.5).The majority of infants born from 34 +0 to 35 +6 weeks' gestation were admitted with prematurity.From 36 +0 to 41 +6 weeks' gestation, respiratory disease and hypoglycaemia were the most common reasons for admission.Infants in the gestational hypertension and pre-eclampsia subgroups had the highest proportion of hypoglycaemia and prematurity admissions, respectively (Table S6).
Odds of admission with primary hypoglycaemia were twice as high in HDP as in non-HDP infants across gestational age categories (overall OR 2.1, 95% CI 2.0-2.2,P < 0.0001), with the highest odds in the gestational hypertension subgroup (OR 2.5, 95% CI 2.4-2.6,P < 0.0001).A lower proportion of HDP compared with non-HDP infants were admitted with respiratory disease and infection.

| Opportunities to avoid admission in HDP infants
A Venn diagram of maternal HDP status and recognised neonatal hypoglycaemia risk factors 26 in infants who received primary management for hypoglycaemia alone in the whole study population (n = 89 381) is shown in Figure 2. Of 8260 HDP infants in this group, the majority had HDP alone (n = 3030/8260 [36.7%]) or were born late preterm with HDP (2707/8260 [32.8%]) and 2523/8717 (30.5%) had multi-factorial recognised risks for hypoglycaemia including maternal diabetes and/or birthweight <2nd centile.In sensitivity analysis using a birthweight centile threshold <10 and including antenatal steroid exposure where recorded, 2330/8260 (28.2%) of HDP infants had HDP alone, 702/8260 (8.5%) were additionally born late preterm and 996/8260 (12.1%) were additionally late preterm and had antenatal steroid exposure.

| Main findings
Over one in ten late preterm and one in 20 term infants without congenital abnormalities admitted to NNU in England and Wales between 2012 and 2020 were associated with a recorded maternal HDP.Respiratory disease and hypoglycaemia were the primary pathologies driving HDP infant admission when considering both primary admission reason and interventions received.Overall, HDP infants were more than twice as likely than non-HDP infants to be admitted with a primary reason of hypoglycaemia and have a hypoglycaemia diagnosis during admission.Abbreviations: HD, high dependency care; IC, intensive care; NG, naso-gastric; NJ, naso-jejunal; OG, oro-gastric; SC, special care.
We estimate approximately one in ten late preterm and term HDP infants admitted to NNU receive primary management of hypoglycaemia alone without additional recognised risk factors 26 for infant hypoglycaemia.These infants were more likely to receive a short duration of i.v.dextrose (≤2 days).Infants born to mothers with gestational hypertension were the most likely to be admitted with a primary reason of hypoglycaemia, have a hypoglycaemia diagnosis, receive primary management of hypoglycaemia alone and have no additional risk factors.These findings suggest that interventions to reduce the risk and severity of infant hypoglycaemia in HDP could reduce neonatal unit admissions of late preterm and term infants of HDP mothers by at least 10%, especially in mothers with gestational hypertension.These infants may also be suitable for care in a transitional care setting without separation of mother and baby.
Compared with non-HDP infants, infants in the preeclampsia subgroup were the most likely to be born preterm, be delivered by emergency caesarean section, be of low birthweight centile and have a diagnosis of growth restriction.The finding of higher survival among HDP than non-HDP infants is expected given the comparison group were sick infants requiring admission to NNU, in addition to the fact that delivery in HDP is often indicated for primarily maternal as opposed to fetal reasons.

| Strengths and limitations
The strengths of this study include the population-level coverage of NNU admissions enabling the national burden of HDP on NNU to be assessed and known high-quality neonatal data within the NNRD. 19In addition, availability of maternal comorbidities and pregnancy complications in the NNRD enabled characterisation of HDP subtypes.Although there was variation in HDP subtype coding between 2012 and 2016, the characteristics of HDP infants described in this study are in keeping with anticipated phenotypes, which is reassuring and highlights the potential utility of maternal variables in the NNRD for studies at the intersection of obstetrics and neonatology.
The richness of the NNRD data extract enabled detailed characterisation of the study cohort and adjustment for relevant confounders in statistical models, such as demonstrating that HDP infants were more likely to receive primary management of hypoglycaemia alone after adjustment for maternal diabetes status, birthweight centile, gestational age at delivery and fetal growth restriction diagnosis.The variable 'primary reason for neonatal admission' reflected the admitting clinician's perception of the single most clinically significant issue, enabling us to highlight the burden of hypoglycaemia specifically in HDP infants.To mitigate the potential limitation of a single primary reason for admission, particularly prematurity, which can encompass multiple pathologies including respiratory disease and hypoglycaemia, we were able to use available daily resource data to confirm that i.v.dextrose and respiratory support were the most common major interventions late preterm and term HDP infants receive.The depth of information available also allowed us to examine days of i.v.dextrose as a proxy for hypoglycaemia severity, enabling further assessment of avoidability of hypoglycaemia admissions.
We were unable to examine effects of specific maternal antihypertensive agents, as these data are not captured routinely in the NNRD.Therefore third-trimester in utero beta-blocker exposure, which is another currently recognised risk factor for infant hypoglycaemia, 26 could not be ascertained.As a result, we were unable to determine how many infants were exposed to labetalol and would have been additionally identified as high risk.
Maternal variables including HDP also may have been mis-or under-ascertained, particularly because maternal diagnosis codes for missingness cannot be distinguished from absence of a condition.Efforts to provide robust national maternity data are ongoing. 28Should HDP infant live birth denominator data become available, linkage to the NNRD would enable interrogation of risk factors for NNU admission and hypoglycaemia in HDP.Formal validation of maternal variables in the NNRD with comparison with clinical codes and technical definitions, e.g.hypertension and proteinuria, would also be valuable.

| Interpretation in the light of other studies
This is the first study, to our knowledge, of the populationlevel contribution of HDP to late preterm and term infant NNU admissions in the UK and analysis of their outcomes, resource use and opportunities to avoid admission.International studies are limited.A population study of 6180 HDP infants born at 34 +0 to 36 +6 weeks' gestation from the Netherlands also reported a higher risk of hypoglycaemia in infants born to women with pre-eclampsia. 29A singlecentre UK study of 474 infants ≥35 weeks' gestation reported maternal HDP were associated with infant hypoglycaemia admissions. 30

| Clinical relevance and future research
This study has highlighted that hypoglycaemia is a primary driver of late preterm and term HDP infant admissions in the UK.We estimate that over one in four admitted HDP infants receive management for hypoglycaemia alone, and one in ten additionally have no other recognised risk factors for hypoglycaemia, with gestational hypertension being the most common HDP subtype in these infants.It is likely that a high proportion of HDP infants in this study were exposed to a beta-blocker in utero, given that current NICE Hypertension in pregnancy guidelines (CG107) recommend labetalol, a beta-blocker, as first-line treatment for pregnancy hypertension, 31 primarily as labetalol has a licence for this use in pregnancy.
Maternal antihypertensive agent choice is a potentially modifiable risk factor for infant hypoglycaemia in HDP.Although epidemiological evidence strongly suggests betablockers are associated with neonatal hypoglycaemia, 12 a recent single-centre study also reported nifedipine was associated with length of stay in infant hypoglycaemia. 30larification of the neonatal effects of antenatal and peridelivery exposure to antihypertensive agents is required.Should risk of neonatal hypoglycaemia vary substantially across antihypertensive agents, appropriate antenatal and peri-delivery antihypertensive agent selection and postnatal risk stratification pathways may avoid at least 10% of current HDP NNU admissions.

| Conclusion
This study demonstrates the high prevalence of maternal HDP in NNU admissions: one in ten late preterm and one in 20 term infant admissions to NNUs are associated with a maternal HDP.Respiratory disease and hypoglycaemia are the major drivers of admission in HDP infants.Examination of resource use data demonstrated that one in four HDP infants are being managed primarily for hypoglycaemia on SC.These admissions were assessed as being the most likely to be avoidable, particularly those infants with no additional risk factors for hypoglycaemia (one in ten HDP admissions overall).Our findings suggest that understanding the risk of hypoglycaemia across different maternal antihypertensive agents, appropriate postnatal pathways and consideration of transitional neonatal care, particularly in mothers with gestational hypertension, may substantially reduce the burden of maternal HDP on NNU and separation of mothers and their infants by reducing late preterm and term infant admissions for hypoglycaemia.

C ON F L IC T OF I N T E R E S T S TAT E M E N T
CB is funded by the UK NIHR through an Advanced Fellowship Award, has received support from Chiesi Pharmaceuticals to attend educational conferences, and been investigator on research grants from the National Institute of Health Research, CB is deputy chair for the NIHR Prioritisation committee for Hospital-based care.

F
I G U R E 1 (A,B) Percentage of live-born babies admitted to Neonatal Units in England and Wales 2012-2020 with a maternal record of hypertensive disorder of pregnancy by gestational age at delivery (A) and by year of birth (B).(C) Bar plot of proportion of infant admissions to a neonatal unit with primary diagnosis of prematurity, respiratory disease, hypoglycaemia, infection and growth restriction by gestational age (late preterm [34 +0 to 36 +6 ] by gestational week, early term [37 +0 to 38 +6 ], term [39 +0 to 41 +6 ] and post-term [≥42 +0 ]) stratified by exposure to maternal HDP.FGR, fetal growth restriction; HDP, hypertensive disorder of pregnancy; SGA, small for gestational age.T A B L E 1 Baseline, antenatal and delivery characteristics of infants admitted to a neonatal unit at ≥34 weeks' gestation in England and Wales 2012-2020 comparing those with and without a record of maternal hypertensive disorder of pregnancy (HDP) stratified by HDP type.

T A B L E 2
Neonatal outcomes and resource use of infants admitted to a neonatal unit at ≥34 weeks' gestation in England and Wales 2012-2020 comparing those with and without a record of maternal hypertensive disorder of pregnancy (HDP) stratified by HDP type.

F I G U R E 2
Venn diagram of recognised hypoglycaemia risk factors as defined in British Association of Perinatal Medicine 2017 Guidelines on Hypoglycaemia in term infant risk factors: maternal diabetes (DM, diabetes mellitus [Type 1 or 2]; GDM, gestational diabetes mellitus), birthweight [BW] <2nd centile and late preterm birth [34 +0 to 36 +6 weeks], in addition to hypertensive disorder of pregnancy [HDP] in late preterm and term infants admitted to NNU in England and Wales 2012-2020 who received primary management of hypoglycaemia alone (special care [SC] level care only, i.v.dextrose and no other major intervention).Total n = 89 381.
FCR, LCC and CB conceived the study.JL accessed the NNRD database and performed data extraction.JL and JF performed data cleaning.FCR, JF and JL carried out data analysis.FCR drafted the first version of the paper, which was edited and approved by all the other authors.AC K NO W L E D GE M E N T SNone.F U N DI NG I N FOR M AT IONThis work was supported by an Isaac Schapera Trust grant and the Medical Research Council (MR/V006835/1, external peer review).Professor Lucy Chappell is supported by an NIHR Senior Investigator Award.Dr Cheryl Battersby is funded by the UK NIHR through an Advanced Fellowship Award.The funders have played no role in conducting the research or writing the paper.For the purpose of open access, the author has applied a 'Creative Commons Attribution (CC-BY)' licence to any Author Accepted Manuscript (AAM) version arising.