A prospective cohort study comparing neonatal outcomes of waterbirth and land birth in an Australian tertiary maternity unit

Maternal preference for warm water immersion (WWI) and waterbirth is increasing, but adoption into obstetric guidelines and clinical practice remains limited. Concerns regarding safety and a paucity of evidence have been cited as reasons for the limited adoption and uptake.


INTRODUCTION
In 2018 the World Health Organization 1 published recommendations for intrapartum care, with an intentional focus on a positive childbirth experience as a significant outcome for women undergoing labour and birth. Several studies suggest that warm water immersion (WWI) and waterbirth result in increased maternal satisfaction with the birth experience, 2-4 likely due to an increased sense of control 3,4 and improved pain management. 5,6 The practice is supported by several studies which demonstrate reassuring clinical outcomes, concurrent with increasing maternal preference and requests for waterbirth. 7 Two recent systematic reviews 7,8 demonstrate no increased risk of harm to the mother or neonate; rather, evidence suggests afforded benefit, including a reduction in the use of regional anaesthesia and length of labour, with no significant differences in neonatal morbidity. 7,8 Despite this high-quality evidence, adoption into clinical guidelines and thus endorsed practice are conflicting, with the American College of Obstetricians and Gynaecologists (ACOG) continuing to recommend that birth occur on land. 9 However, more recently, the Royal Australian and New Zealand College of Obstetricians and Gynaecologists (RANZCOG) has updated its position statement to recognise the increasing evidence surrounding WWI. It acknowledges the changing cohort of women who may now have to access water immersion, stating 'Women requiring continuous electronic fetal monitoring (CEFM) during labour may utilise water immersion, provided that adequate telemetry equipment is available'. 10 Furthermore, qualitative studies have reported a lack of support for WWI from obstetricians, most notably where women have additional risk factors. 11,12 This was particularly prevalent within obstetric units where senior staff did not appreciate the benefits of WWI and were observed to promote a medicalised approach to labour and birth. 13 RANZCOG also acknowledges this lack of support for waterbirth among the medical community, suggesting that more research is required in the area. 10 During a new hospital and maternity unit construction (completed in 2017), our birth suite was generously furnished with birthing pools in 12 of 15 birth suite rooms. All rooms were equipped with wireless, waterproof continuous electronic fetal monitoring (CEFM), providing a range of women unprecedented access to WWI and waterbirth as a pain relief strategy.
Considering the adaptations to the RANZCOG guidelines, as well as increasing maternal preference to use WWI and waterbirth, this study aimed to investigate the neonatal and maternal outcomes after WWI and/or waterbirth compared with land birth, including for women requiring CEFM.

MATERIALS AND METHODS
We conducted a prospective cohort study of all women who had a vaginal birth within the Sunshine Coast University Hospital, Queensland, Australia, between April 2019 and April 2020. All women who were eligible to enter the birth pool (WWI ± waterbirth) were included. The study site is a tertiary, publicly funded maternity unit with approximately 3600 births per year, in a coastal region of Queensland, Australia. Eligibility criteria and subsequent exclusion criteria were guided by the local Workplace Instruction on WWI and Waterbirth 14 Appendix (S1).  All women who entered the water were cared for in accordance with our local WWI policy (S1). During established labour women in each of the three groups were provided one-to-one midwifery care, with two health professionals present at all births. At the study site, all birth-suite midwives are credentialled to perform waterbirths, and as such, there was limited variance in the knowledge and practice of the accoucheur between the groups. The WWI policy (S1), followed for all women who enter the pool, includes clearly defined inclusion/exclusion criteria, guidance for maternal and fetal observations in all stages of labour, maintenance of water temperature and cleanliness as well as indications to leave the pool. Strict infection control measures are also in place to reduce the risk of water contamination and ensure that the pool is cleaned to the appropriate standard.

Data collection
Data were collected from two linked data sources: Birth Registration

Short-Form and the Queensland (Qld) State-Wide Perinatal Data
Collection. The primary outcome measure was admission to the neonatal unit (NNU) (see Table 1 for a full list of variables).
All NNU admissions recorded on the perinatal database were manually cross-checked against the electronic medical record.
This confirmed that the data had been accurately recorded, thus enhancing internal reliability.

Data analysis
Descriptive analyses were performed to compare the maternal and fetal demographical and outcome measures for the three groups: land, water and WWI-only births. Data are presented using median and inter-quartile range (IQR) for continuous data (due to skewed distributions) and frequencies and proportions for categorical variables. χ 2 tests were performed to investigate the associations between the groups (land, water and WWI) and categorical variables (eg, NNU admission, perineal damage). χ 2 tests and Mann-Whitney U-tests were used to study the associations between NNU admission and maternal variables. Multivariate logistic regressions were used to model the odds of NNU admission, where variables were selected for inclusion based on their clinical relevance. These models contained the following variables: birth group (WWI vs waterbirth vs land birth), maternal age, maternal BMI, previous birth mode, epidural use, length of labour (first and second stages) and labour onset. All analyses were performed in R (version 4.0.2).

Ethics
This study was granted ethical approval through The Prince Charles Hospital Human Research Ethics Committee (HREC/17/ QPCH/480). All data were deidentified before export and analysis. Individual consent was not required as approval was granted through the Public Health Act (RD007551) to access routinely collected data. The study was conducted in accordance with the ethical principles from the National Health and Medical Research Council guidelines. 15

RESULTS
A total of 1665 women met inclusion criteria. There were 1265 land births, 243 waterbirths and 157 WWI-only land births. Baseline demographical and maternal delivery characteristics were similar between groups, as presented in Table 2.
Our primary outcome measure was admission to the NNU.
Univariate analyses found that babies born in water had significantly lower rates of NNU admission (11.23, 10.19 and 5.35% (land birth, WWI and waterbirth, respectively); P = 0.022). Reasons for admission to the NNU across all groups were similar and are included in the Appendix (S2). No babies required newborn intensive care unit admission for water inhalation or near-drowning.
There were no intrapartum deaths in any groups. A sub-analysis, where instrumental births and those who had regional analgesia were removed, showed no significant association between NNU admission and birth group (P = 0.506) for those who had a spontaneous vaginal birth.
There was a statistically significant association between land birth and admission to the NNU for a suspected infectious condition (6.17, 5.1 and 2.06% for land birth, WWI and waterbirth, respectively; P = 0.035); 5.45% of neonates born on land required antibiotics compared with 2.55% in the WWI group and 2.47% in the waterbirth group, which had borderline significance (P = 0.053). No maternal pyrexia was found in the WWI or waterbirth group (see Table 3).
Multivariate logistic regressions of neonatal admission (yes/ no) adjusted for maternal age, maternal BMI, previous birth mode, epidural use, total length of labour and onset labour (in addition to birth group) were performed. After accounting for these variables, there was no significant difference in the rates of NNU admission between groups (P = 0.167). See Table 4 for a full summary.  The risk of neonatal sepsis is often cited by those who oppose WWI/waterbirth. 16 However, case reports of neonates infected with Legionella pneumophila and Pseudomonas aeruginosa 17 appear to have resulted when a deviation from established waterbirth protocols had occurred or from the use of unclean water.
Overwhelmingly, recent literature suggests no increase in infective sequalae from WWI/waterbirth. 18,19 Most women (95.5%) who were febrile intrapartum had an epidural, likely explaining the increased rates of pyrexia in the land birth group. The association between maternal pyrexia and regional anaesthesia is well established, with one retrospective study of 261,457 women finding those with an epidural five times more likely to experience pyrexia and neonates 1.6 times more likely to receive antibiotics. 20 A sub-group analysis performed in our study, which removed women with epidurals from the cohort, subsequently showed no differences in NNU admission between the groups. This is a rare event, and studies are inevitably underpowered to prove a statistically significant association with waterbirth.
According to Schafer, 22 based on pooled data of 10,000 waterbirths, the incidence of cord avulsion appears increased at 3.10 per 1000. However, this risk can be mitigated by clear waterbirth protocols, midwifery education and discussion with women before pool entry. 22,23 Demand for WWI and waterbirth is increasing, and where facilities and policies are supportive of this practice (as within this study site), access and usage increase (24% in this series). There are some consumer perceptions that obstetricians are keen to 'over-medicalise' both labour and birth. 12 The 'medicalisation' of birth has diminished women's satisfaction with their experience, 23 and as obstetricians, we must tread a careful line between over-and under-intervention. suggesting that waterbirth can be achieved safely but falls short of endorsing the practice, 10 and ACOG recommends that birth occur on land. 9 Furthermore, there is also no national registry or entity collecting Australian waterbirth data, where potential research and quality improvement could be optimised. 25

Strengths and limitations
This is a large sample size relative to other studies, with 400 women using WWI during their labour and/or birth; 24% of our cohort used WWI for labour and/or birth, which is significantly higher than the state average of 14.8%. 24 The study also included 157 women utilising WWI and waterbirth while using CEFM, the first Australian data reporting this practice. The groups had similar baseline characteristics, and the multivariable analysis allowed for the control of confounding factors. Our data were prospectively collected and input in the research database during birth, thereby increasing the reliability of our findings.
Due to the nature of the intervention, women self-select waterbirth/WWI, creating selection bias. While a randomised con- This study reinforces the safety of WWI and waterbirth, including, for the first time, a selected group of women requiring CEFM.
Demand for access to non-pharmacological analgesia continues to increase, and maternity healthcare providers must be flexible and proactive in meeting these demands. We recommend increasing access to water immersion, including for those women who require CEFM. This must coincide with the development of evidence-based guidelines, audit of practice and provision of First-and second-stage length (min).