The POOL Study was conducted to determine whether, among women without ‘risk-factors’, who use water immersion during labour, having a waterbirth is as safe for themselves and their babies as leaving the water prior to birth. The POOL Study provides compelling evidence that among women who use water immersion during labour, waterbirth, as practiced by NHS midwives, is as safe for mothers and their babies as birth out of water.
The cohort study had a non-inferior design, that is, the study tested whether the rates of adverse outcomes were higher following waterbirth than among births out of water. Data were extracted from the MIS from 26 NHS maternity services, and for babies admitted to a NNU, date were extracted from the NNRD, relating to births between January 2015 and June 2022.
After data cleaning, including removal of duplicates, 869,744 individual birth records were received over the study period, of which 87,040 (10.0%) related to women who used a pool during labour or birth, and 782,704 (90.0%) records related to women who did not use a pool. The number of births an individual site contributed to the study ranged from 6804 to 65,667. Rates of pool use varied by site between 1.6% and 22.9%, and in the proportion of women using a pool who gave birth in water, it ranged across sites from 32.7% to 82%. Overall, 15.9% of women using a pool had a risk factor recorded in their antenatal record or at the time of pool entry, although this may be an overestimate of the true rate, as it may include some women unnecessarily due to issues with data storage within the MIS.
Key study findings
Rates of pool use and waterbirth
Over the study period, the proportion of women in England going into spontaneous labour declined from 59% in 2015 to 47% in 2022.63 This decline was reflected as a reduction in the rate of pool use, and of waterbirths, over the latter part of the study period. The proportion of women using a pool during labour was highest in quarter 3, 2016, when 11.9% of women used a pool and 6.0% of woman gave birth in water; by quarter 2, 2022, the proportions were 7.7% and 4%, respectively. Reflecting closure of MLUs and redeployment of staff to centralised services,64 the beginning of the COVID-19 pandemic in quarter 2, 2021, was associated with a decline in POOL use and rates of waterbirths, but by the autumn quarter of 2021, the rates had returned to pre-pandemic levels.
Transfer to obstetric care
Among women without complicating factors at pool entry, 29% of nulliparous women (N = 11,139) and 5% of parous women (N = 1688) received additional monitoring, obstetric interventions or regional analgesia before or during birth. Reflecting clinical need, 97.1% of women (N = 12,455) receiving additional care prior to or during birth gave birth in an OU, with rates of SVBs, instrumental birth and birth by caesarean section being 27.9%, 52.2% and 19.8%, respectively. The rate of the neonatal primary outcome among babies born to women who received additional monitoring, obstetric interventions or regional analgesia, before or during birth, was 11.7% (N = 434), with 13.6% (N = 1,744) of babies being admitted to a NNU.
The rates of additional care before or during birth were similar to the rates of transfer to obstetric care found among nulliparous and parous women, without complicating factors at labour onset when commencing labour care in a MLU or at home, in the Birthplace in England study (32.9% and 6.7% respectively).
Mode of birth among women using a pool
Overall, among women who used a pool without risk factors, the rates of SVBs were high (78.0% nulliparous women, 97.6% parous women), with low rates of instrumental births (10.9% nulliparous women, 1.6% parous women) and birth by caesarean section (5.9% nulliparous women, 0.7% parous women). Rates of instrumental and operative births were similar to those found in the Birthplace in England study among women planning birth outside of an OU.
Primary analysis group
The primary analysis excluded women with risk factors recorded in the antenatal notes or at pool entry and those who received additional monitoring, obstetric care or epidural analgesia prior to birth. The primary analysis included 39,627 births in water, and 20,775 births out of water, and compared rates of adverse events among births in water with births out of water.
Maternal primary outcome
The study found no evidence of increased rates of OASI among women who gave birth in water. After adjusting for differences in the characteristics of women who gave birth in, and those who gave birth out of, water, fewer nulliparous women who gave birth in water had a recorded OASI than in those who left the pool [730 of 15,176 women (4.8%) vs. 641 of 12,210 women (5.3%); aOR 0.97, one-sided 95% CI −∞ to 1.08]. Similarly, after adjusting for differences in the characteristics of women who gave birth in, and those who gave birth out of, water, fewer parous women who gave birth in water had a recorded OASI than among those who left the pool before birth [269 of 24,451 women (1.1%) vs. 144 of 8565 women (1.7%); aOR 0.64, one-sided 95% CI −∞ to 0.78].
The upper limit of the CIs for the difference in the proportion of OASI between the groups were less than the prespecified non-inferiority margin, for both nulliparous and parous women, of 1% (OR < 1.23) and 0.6% (OR < 1.38), respectively, indicating that births in water were associated with no increase in OASI compared to births out of water.
When pooled for all women, the rate of OASI in women giving birth in water was lower than that in those who left the pool [999 of 39,627 women (2.5%) vs. 785 of 20,775 women (3.8%); aOR 0.89, one-sided 95% CI −∞ to 0.98]. Preplanned sensitivity analysis, including propensity score analysis, supported the finding of no additional risk of OASI when births occurred in water.
Infant primary outcome
The primary infant outcome for infants was ‘adverse infant outcome or treatment’ (death, NNU admission with respiratory support or receipt of IV antibiotics within 48 hours of birth) and included births during the period from the first site opening in January 2019 to June 2022. Of the 16,223 infant records, 14,946 (92%) contributed to the analysis. The remaining were excluded due to incomplete recording of some of the outcomes at sites.
The study found no evidence of increased rates of the neonatal primary outcome among babies born in water. After adjusting for differences in the characteristics of women who gave birth in, and those who gave birth out of, water, fewer babies in the waterbirth group met the primary outcome than those born out of water [263 of 9868 infants (2.7%) vs. 224 of 5078 infants (4.4%); aOR 0.65, one-sided 95% CI −∞ to 0.79]. The upper limit of the CI for the difference in infant outcome between the groups was less than the prespecified non-inferiority margin of 1.0% (OR < 1.21), indicating that being born in water was found to be associated with no increase in adverse neonatal outcomes than being born out of water.
Some components of the composite infant outcome are more common than others. In the prospective cohort, which was underpowered to draw conclusion on the important outcome of baby deaths, three babies died following waterbirth (0.3. per 1000 births) and none died following birth out of water. Among the larger cohort of retrospective and prospective births, N = 60,402, there were 13 intrapartum or neonatal deaths: 7 following waterbirth (0.18 per 1000 births) and 6 (0.29 per 1000 births) among babies born out of water, with evidence of non-inferiority (aOR 0.22, one-sided 95% CI −∞ to 0.80); 0.8% (N = 329) and 1.5% (N = 320) received respiratory support on a NNU; and 1.8% (N = 629) and 2.9% (N = 535) received IV antibiotics within 48 hours of birth. When rates of the individual component outcomes were examined individually, all were lower among births in water than among births out of water; this included the rate of deaths, but due to small numbers, the CIs were wide.
Secondary outcomes
Third-stage management
Lower rates of active management of the third stage of labour occurred among waterbirths compared to births out of water (72.9% vs. 89.1%; aOR 0.31, one-sided 95% CI −∞ to 0.33). For women giving birth in water during the period of prospective data collection, information was captured on whether it was intended that the placenta would be delivered in or out of water and whether placental delivery occurred as intended. Among the 69% (N = 7,422) of waterbirths, where it was intended that the placenta would be delivered once the woman had left the pool, the placenta was delivered out of the pool in 97% (7200) cases, with 1.9% (N = 143) of women delivering the placenta in water. Among the 8.5% (N = 910) of waterbirths, where it was intended that the placenta would be delivered in water, the placenta was delivered in water in 59.5% (N = 541) of cases, with 39.8% (N = 362) of women delivering the placenta out of water. Overall, the placenta was delivered in water following 8.6% (N = 926) of waterbirths. Among women who gave birth in water, rates of PPH of ≥ 1000 ml were similar among women delivering the placenta in water compared to women delivering the placenta out of the water (1.9% compared to 3.3%; aOR 0.70, one-sided 95% CI −∞ to 1.18, p = 0.181).
Childbirth emergencies
Being in a pool during birth may have the potential to delay emergency treatments, such as in the event of shoulder dystocia, or PPH, where the woman needs to leave the pool for treatment.
Shoulder dystocia was recorded as having occurred in 0.6% of waterbirths and in 3.1% of births out of water (aOR 0.16, one-sided 95% CI −∞ to 0.18). There was no evidence that waterbirth increased the chance of shoulder dystocia or extended the head-to-body delivery interval, with a mean head-to-body delivery interval of 3.5 minutes and 3.3 minutes among births in water and out of water.
Among women who gave birth in water, compared to women who gave birth out of water, fewer had a recorded PPH of ≥ 1000 ml (2.9% vs. 3.8%), although differences in measurement techniques mean that direct comparisons should not be made. Among women who gave birth in water, compared to women who gave birth out of water, fewer underwent a MROP (1.8% vs. 2.5%) or received obstetric care following birth (6.5% vs. 9.6%).
Other maternal intrapartum secondary outcomes
Rates of women sustaining any perineal trauma were lower among waterbirths compared to that among women who gave birth out of water (62.2% vs. 68.3%; aOR 0.84, one-sided 95% CI −∞ to 0.88), and similarly, rates of the woman undergoing perineal repair were lower following waterbirths compared to following birth out of water (31.9% vs. 44.3%; aOR 0.78, one-sided 95% CI −∞ to 0.81).
Among women who gave birth in an OU or MLU, the mean duration of postnatal stay was shorter among women who had a waterbirth compared to the duration among women who gave birth out of water [mean 11.7 hours, N = 35,514, vs. 15.5 hours, N = 17,420; adjusted incidence rate ratio (AIRR) 0.85, one-sided 95% CI −∞ to 0.87]. Rates of women receiving higher levels of care, rates of blood transfusion and rates of women returning to hospital for assessment or admission in the postnatal period were similar among women giving birth in and out of water.
Infant care and feeding
A higher rate of the umbilical cord snapping before it was clamped occurred among births in water compared to births out of water (1.0% vs. 0.3%, aOR 3.89, one-sided 95% CI −∞ to 6.88). Among the 122 babies in both groups, where their cord snapped prior to clamping, compared to babies where the cord was clamped prior to cutting, there were higher rates of: NNU admission (13.1% vs. 4.8%); NNU admission with respiratory support (3.3% vs. 1.3%) and administration of antibiotics with 48 of birth (5.7% vs. 2.3%). No babies with a ‘snapped cord’ died or received therapeutic hypothermia.
There was no evidence that waterbirth was associated with babies receiving less recommended care at birth. Compared with babies born out of water, waterbirth was associated with lower proportions of babies with cord clamping within a minute of birth (5.4% vs. 8.1%, aOR 0.73, one-sided 95% CI −∞ to 0.85); no record of skin-to-skin contact between mother and her baby at birth (5.6% vs. 7.6%, aOR 0.74, one-sided 95% CI −∞ to 0.80) and babies not breastfed within the first hour of birth (17.6% vs. 27.7%, aOR 0.63, one-sided 95% CI −∞ to 0.70).
Rates of breastfeeding initiation and breastfeeding at the time of discharge from community midwifery care were both higher following births in water compared to births out of water (83.8% vs. 79.8%; 72.1% vs. 67.8%, respectively). Waterbirth was non-inferior to birthing out of water at both time points (initiation and community discharge of care) (aOR 0.79, one-sided 95% CI −∞ to 0.83; aOR 0.83, one-sided 95% CI −∞ to 0.88, respectively).
Need for additional neonatal care
Rates of babies with an Apgar score of < 7 at 1 and 5 minutes were lower among babies born in water than among babies born out of water (3.6% vs. 5.1%; aOR 0.72, one-sided 95% CI −∞ to 0.78) and (0.5% vs. 0.9%, aOR 0.54, one-sided 95% CI −∞ to 0.62). Reflecting this, the rates of resuscitation were also lower among babies born in water than among babies born out of water (4.1% vs. 6.3%; aOR 0.61, one-sided 95% CI −∞ to 0.65).
Access to pools and waterbirth
The site case studies explored the context of care in maternity units, and this was found to have a substantial influence on pool use. Considerable differences were found between OUs and midwifery units in relation to equipment and resources, staff attitudes and confidence, senior staff support and women’s awareness of water immersion.
The online discussion groups and individual interviews, with women, midwives, obstetricians, neonatologists and doulas, identified various factors influencing the use of birth pools in the UK, including resource availability (including midwives with experience of waterbirth), unit culture and guidelines and staff endorsement. Demand for waterbirth was perceived to vary across the UK and across different socioeconomic groups: health professionals suggested that greater promotion of pool use by maternity unit staff could increase awareness of this option among women. Proactive support for waterbirth in obstetric-led units was seen as lacking.
Women with risk factors using a pool
When restricted to risk factors present at the time of pool entry, as recorded by the midwife following birth among records completed after sites opened to the study, 11.4% (N = 2725) records were flagged as the mother having at least one risk factor at pool entry. The most common risk factors reported were: induction of labour 2.3% (N = 583); GBS carrier 1.9%, 475; gestational diabetes/diabetes 0.8% (N = 188); vaginal birth after caesarean (VBAC) 0.7% (N = 83); and BMI > 35 kg/m2, 0.6% (N = 152).
There was some indication that, compared to women who did not use a pool in labour, women who used a pool without risk factors and women with identified risk factors who used a pool in labour, there were differences in the proportions of woman of (1) White ethnicity (70.1%, 81.0%, 83.3%, respectively); (2) women who were fluent in English (79.9%, 85.0% and 90.4%, respectively) and (3) women living in least deprived areas (15.2%, 20.7% and 20.5%, respectively).
Interpretation of results
It is important that the results of the study are interpreted as reflecting outcomes observed in the NHS clinical practice rather than evidence of a causal relationship between birth in water and lower risk of OASI, NNU admission or treatment for potential early onset neonatal infection.
The most plausible explanation of the differences found in outcomes between groups is that some women who gave birth out of water did so when those women or midwives had unrecorded concerns, or when there were other reasons to recommend birth was out of water.
When a woman is in a pool and abnormalities occur during labour, providing there is time for her to do so safely, the woman should be advised to leave the pool for care and to give birth out of water.65 Indications for advising a woman to leave the pool, for example, concerns in the fetal HR or slow advancement of the fetal head, are also associated with the primary maternal and neonatal outcomes. A higher proportion of births with a midwife concern were reported in the out-of-water group (N = 1862, 9.0%) compared to the waterbirth group (N = 273, 0.7%). This reflects that there is usually sufficient time between a concern becoming evident for a woman to leave the pool and give birth out of water and that most women are willing, and able, to follow such advice. Some midwife concerns, or considerations, which led to women giving birth out of water may not have been documented in the MIS following birth, concealing unidentified differences between women who gave birth in and out of water.
Although the study was designed to reduce bias, its observational design means that the potential remains for unknown or incompletely measured confounders to exist between women giving birth in and out of the water, which have not been controlled for; for example, a woman’s individual determination to give birth in water, or the experience of the midwife caring for her. While this is a limitation of observational studies, it reflects ‘real-life’ choices of women and NHS practice, and their associated outcomes, and should be interpreted as such.
For some outcomes, particularly involving blood loss measurement, direct comparisons between groups should be interpreted with caution. Estimation of blood loss while a woman remains in the water is mostly based on visualisation which is known to be inaccurate,66 as more accurate blood loss measurement techniques suitable for pool use are yet to be developed.
POOL Study findings in context
The POOL Study provides new knowledge about outcomes for women and their babies following pool use in labour. The findings of the POOL Study add to previous large observational and small randomised studies, conducted in the UK,18 USA17 and Australia,16,67 in finding waterbirth to be associated with no additional morbidity for mothers or their babies.
Perineal trauma
In the mid-1990s, the rate of OASI among women experiencing a SVB in the UK and participating in a randomised trial of perineal care was 1.3%.68 Low rates of OASI were also reported in out-of-hospital births in the USA between 2012 and 2018 among a cohort of 17,530 women who had a waterbirth propensity matched to 17,530 land births.17 After matching, the cohort included 26.5% nulliparous women, with rates of OASI of 0.75% and 0.84%, respectively, among births in and out of water.
Over recent years, rates of detected OASI following vaginal birth in the UK have increased.28 The increasing rates have been attributed to improved detection, women being at greater risk due to the increasing age and BMI of women and potentially other undetermined factors.30
The POOL Study was designed to be sufficiently powered to be able to conduct an analysis of OASI rates by parity subgroups, and it found no evidence to indicate that birth in water is associated with increased rates of OASI in either group. The rates of OASI in the POOL Study cohort were 4.6% and 5.0%, respectively, among nulliparous women giving birth in and out of water and were 1.1% and 1.5%, respectively, for parous women giving birth in and out of water.
Rates of OASI found in the POOL Study were higher than those reported in a hospital-based cohort study of 1007 waterbirths at a hospital in England over the 10-year period from 2007 to 2017, which compared outcomes with 36,924 similar ‘low-risk’ women experiencing a SVB over the same period.69 The rates of OASI in the period 2007–17 were 4.3% and 4.0%, respectively, among nulliparous women giving birth in and out of the water, and these were 0.9% and 0.9%, respectively, for parous women giving birth in and out of the water.
A UK-based cluster randomised trial explored the impact of a care bundle on rates of OASI between 2016 and 2018. The care bundle included antenatal information for women on OASI, recommending an episiotomy when clinically indicated, perineal protection during all vaginal births and systematic perineal examination following birth. Overall OASI rates among SVBs were 2.6% before and 2.2% after the implementation of the care bundle (aOR 0.75, 95% CI 0.60 to 0.93); rates among nulliparous women were 5.2% before and 4.9% after implementation of the care bundle (aOR 0.81, 95% CI 0.65 to 1.00); and among parous women, the rates were 1.7% before and 1.5% after the implementation of the care bundle (aOR 0.78, 95% CI 0.61 to 1.01).70 Subsequently, uptake of the care bundle was encouraged across all UK maternity units. Where perineal protection cannot be implemented, such as during waterbirth, midwives are required by the care bundle to document their non-compliance and are encouraged to document the reason for their non-compliance for this aspect of the bundle.
In the UK, two midwives are usually expected to be present at waterbirths but not during SVBs occurring out of water. Evidence from a randomised trial in Sweden, including women experiencing their first vaginal birth (N = 3776), found that the presence of a second midwife during SVBs reduced the rate of OASI from 5.7% to 3.9% (aOR 0.69, 95% CI 0.49 to 0.97).71 A second midwife as standard at SVBs out of water is not currently a component of the RCOG OASI care bundle, but such an addition may have the potential to reduce OASI rates if implemented in the UK.
Rates of OASI found in the POOL Study were similar to those in other studies from the UK, where rates have been reported by parity. This indicates that around 1 in 20 women experiencing a SVB of their first baby will sustain identified damage to their anal sphincter regardless of whether birth is in or out of water. Although the OASI care bundle has shown some effectiveness, further research is required to reduce such trauma.
Neonatal primary outcome
The composite neonatal primary outcome was selected to reflect adverse neonatal outcomes of concern to neonatal and paediatric practitioners, as reported in case series and in previous neonatal studies.26 After adjusting for differences in the characteristics of women who gave birth in, and those who gave birth out of, water, fewer infants born in water met the primary outcome (2.7% vs. 4.4%, aOR 0.65, one-sided 95% CI −∞ to 0.79). This finding will form important evidence on which professionals can base informed opinions on neonatal outcomes associated with waterbirth.
The use of composite primary outcomes for neonatal studies has been criticised on several grounds,72 including uncertain relevance of composite outcomes that combine events of different importance and severity and composites combining outcomes where effects of the intervention might be in different directions for different components. Composite outcomes can also be driven by individual components that are more common than other components. The infant composite outcome for the POOL Study combined events with very different levels severity and importance, including death and receipt of antibiotics which occurred at very different rates in the low-risk population of interest. It is therefore important to examine the analyses by individual components of the composite outcome. It is reassuring to see that, among the larger cohort of retrospective and prospective births, the POOL Study findings are consistent across all components, with no excess of stillbirth and neonatal death before discharge, admission to a NNU for respiratory support or receipt of antibiotics in the first 48 hours, among babies born in water.
Shoulder dystocia
The first-line management of shoulder dystocia in water is to stand the woman up or encourage her to leave the water. By contrast, first-line recommended management of shoulder dystocia when the woman is out of water is to use the McRoberts position.73 Among cases of shoulder dystocia occurring in water, 25.4% (N = 45) were resolved by change of the woman’s position alone, including exiting the pool, compared to 3.2% (N = 17) of cases occurring out of water. McRoberts position and suprapubic pressure were used less frequently among cases of shoulder dystocia in water compared to cases occurring out of water (78/177, 44.1% vs. 452/526 85.9% for McRoberts position) (22/177, 12.4% vs. 187/526 35.6% for suprapubic pressure). There was similar use of internal manoeuvres (44/177, 24.9% vs. 146/526, 27.8%).
The finding that 25.4% of cases of shoulder dystocia occurring in water were resolved by a change of maternal position alone suggests that a diagnosis of shoulder dystocia, on occasions, may be made at a lower threshold in births in water, compared to births out of water. In the out of water situation, a diagnosis of shoulder dystocia is made only after axial traction on the fetal head has failed to facilitate birth of the fetal shoulders.73 During births in water, midwives are discouraged from applying axial traction due to concern that this may trigger fetal water inhalation,65 and it appears shoulder dystocia may be diagnosed following failure of the fetal shoulders to be delivered through maternal effort alone.
With NICE recommending that women with a history of shoulder dystocia should be advised to plan birth in an OU,31 it is important that shoulder dystocia is not overdiagnosed, as this can limit the birth choices of women in future pregnancies. Midwives should be aware that during waterbirths, a change of maternal position with birth, facilitated by axial traction, does meet the current criteria for the diagnosis of shoulder dystocia.73 The evidence base on which discouragement of the use of axial traction in water warrants further investigation as does the threshold at which a diagnosis of shoulder dystocia is made during births in water.
Management of the third stage of labour
For women giving birth in water during the period of prospective data collection, information was captured on whether it was intended that the placenta would be delivered in or out of water and whether placental delivery occurred as intended. Among the 69% (N = 7422) of waterbirths, where it was intended that the placenta would be delivered once the woman had left the pool, the placenta was delivered out of the pool in 97% (7200) of cases.
Among the 8.5% (N = 910) of waterbirths, where it was intended that the placenta would be delivered in water, the placenta was delivered in water in 59.5% (541) of cases. Rates of PPH of ≥ 1000 ml among women delivering the placenta in water following waterbirth, compared to women delivering the placenta out of the water following waterbirth, were not increased (18/926, 1.9% compared to 322/9834, 3.3%; aOR 0.70, one-sided 95% CI −∞ to 1.08).
Neonatal secondary outcomes
Apart from cord snapping, where birth in water was associated with an increased rate, the POOL Study found no evidence that birth in water was associated with an increased rate of any of the other secondary neonatal outcomes.
Of particular concern, due for the potential for long-term hypoxic brain injury, are babies who have an Apgar score of < 7 at 5 minutes of age or require therapeutic hypothermia. Therapeutic hypothermia was recorded as having been received in 0.07% of babies born in water and 0.26% of babies born out of water (aOR 0.33, one-sided 95% CI −∞ to 0.53). Although previous large observational studies in the USA17 and England19 found no increase in the rates of NNU admission following waterbirth, neither reported on the important outcome of treatment with therapeutic hypothermia.
A large cohort study published in 202119 used NHS routine data relating to births during the financial year 2015–6 to compare outcomes between 6284 waterbirths and 39,824 SVBs to women without medical complications regardless of water immersion use during labour. The POOL Study found that rates of Apgar score of < 7 at 5 minutes among babies born in water were very similar to those previously reported: POOL Study rate 0.5%, previously reported rate 0.57%.19 Among babies born out of water, the POOL Study found higher rates of babies with an Apgar score of < 7 at 5 minutes (0.9%) than those reported as associated with SVBs out of water (0.57%).19
The rates of Apgar score < 7 at 5 minutes and of therapeutic hypothermia among babies born out of water following water immersion support the fact that women who use water immersion during labour, but who leave the pool before birth, have higher rates of complications occurring during labour than women who remain in the pool for birth, or among the whole population of women having a SVB out of water.
Cord snapping
Consistent with previous studies, a higher rate of cord snapping prior to clamping occurred during births in water compared to births out of water. The rate of umbilical cord snapping prior to clamping in the POOL Study prospective cohort was 1.0% among births in water and was 0.4% out of water. This rate compares to 0.35% among 5192 waterbirths in an observational study, including sites across England, Scotland and Northern Ireland at different time periods between 2000 and 2008;18 and compares to 0.57% among 17,530 among waterbirths included in a large cohort study of births outside of hospital conducted in the USA.17 The largest previous cohort study of waterbirths in England (N = 6264) was unable to report rates of cord snapping, as this outcome is not captured routinely in all MISs or exported to centralised NHS data.
The complication of cords snapping before being clamped is uncommon but potentially serious for the baby. Among the 122 cases of cord snapping prior to clamping in the POOL Study occurring in births in and out of water, there were increased rates of NNU admission, respiratory support and antibiotic administration within 48 hours of birth. There were no cases of death or treatment for hypoxic ischaemic encephalopathy. We did not have data relating to neonatal blood transfusion, and neonatal anaemia needs to be considered as a potential consequence when the cord snaps prior to clamping. The mechanism behind cord snapping in water warrants further investigation as does the increased rates of respiratory support and antibiotic administration. The importance of having an awareness of tension in the cord when births occur rapidly in water, and when lifting the baby following birth, should be included in midwifery and obstetric training and parental preparation, particularly when waterbirth is planned or being considered.
Study strengths and limitations
The POOL Study is the largest global study of waterbirths conducted to date and provides the best available evidence in relation to outcomes associated with waterbirths conducted by NHS-employed midwives.
The gold standard of study designs is the randomised controlled trial (RCT). Previous studies have concluded that a trial of waterbirth of sufficient size is not feasible;15,71 therefore, a well-designed observational study using routinely collected data was needed. However, all study designs have inherent strengths and weaknesses and to interpret the POOL Study results, it is necessary to explore the potential biases and what impact these may have on the findings.
Non-inferiority margins were predefined with a difference of ≤ 1%, and ≤ 0.6%, for OASI taken as clinically non-significant among nulliparous and parous women without antenatal complexities, respectively. Similarly for the infant outcome, a non-inferiority margin of ≤ 1.0% was taken as clinically non-significant. These margins were informed by patient and public involvement (PPI) input and clinical judgement. Estimates of the background rates of the maternal and infant primary outcomes were based on a robust and timely source, and these were close to the rates found in the study.
Another key strength of the study was, although observational, methods followed recommendations and reporting guidelines for non-inferiority RCTs.43,74 Non-inferiority margins were justified in advance; for transparency, presentation of the CI was consistent with the type I error rate used in sample size calculations, missing data were handled appropriately in sensitivity analyses, and for transparency, the protocol and analysis plan were published in advance of use.
Obtaining pilot data in advance enabled a prior understanding of the robustness of the fields to be used for the outcomes and the extent of missing data. This meant that the proposed outcomes were, overall, measured robustly. However, some outcomes on inspection did not have the granularity of data required (e.g. re-admissions to hospital for infants).
The strengths of the study include identification of women who did not receive additional monitoring, obstetric or anaesthetic interventions before birth. This was important to answer the specific study objective of assessing whether there are differences in birth outcomes among women who used a pool during labour and remained in the water for birth and those who used a pool and gave birth out of water but had no contraindications to remaining in the pool. A previous large study conducted in England exploring maternal and neonatal outcomes associated with waterbirths, occurring between April 2015 and March 2016, used centralised maternity audit data to compare outcomes for women who gave birth in, or out of, water.19 The study included all women without labour complexities who gave birth out of water as the control group, as the study team, using centralised records, were unable to identify women who used a pool during labour but left the water prior to birth.
The POOL Study was required to be specifically designed to compare outcomes between women who gave birth in water and those who gave birth out of water following water immersion. An important bias to overcome when designing the study was the potential for women and babies to be overexcluded from the primary analysis where concerns were recognised, or complications occurred prior to or during birth. To overcome this, women and babes were only excluded from the primary analysis, where there was a record of additional monitoring, obstetric or anaesthetic care prior to birth. To reflect clinical practice and reduce the potential influence of retrospective reporting bias in the event of adverse outcomes, where midwives recorded that concerns had been present prior to birth, but without recorded obstetric involvement in the woman’s care, women and their babies remained in the primary analysis.
A separate important question was whether birth in water, in the absence of any midwife concerns, impacts on maternal and neonatal outcomes. Sensitivity analysis restricted to women in whom there had been no midwife concerns prior to birth was performed. This also found waterbirth to be associated with no increase in the primary maternal or neonatal outcomes.
The use of routine data facilitated generating a large data set with required power to answer the study objectives of exploring relationships between uncommon but important adverse maternal and neonatal morbidity and birth in water. Limitations exist around the nature of routine data captured primarily for clinical purposes. High levels of missing data were seen for some fields, with some units not entering any data into specific fields, and some data fields being not as specific or detailed as if designed specifically for research use. Examples in the POOL Study included that some study sites did not enter data relating to postnatal care, and maternal assessment in the hospital could not be distinguished from postnatal re-admission.
The POOL Study was the first to adapt electronic MISs at individual NHS site level for the purpose of collecting research data. Using and adapting existing data systems in this way created an efficient and novel data collection system that maximised the use of existing routinely collected data fields while minimising additional or repetitive data collection for NHS staff.32 Furthermore, the study matched data held in electronic maternity records with that captured in the NNRD to enable an analysis of outcomes for babies admitted to a NNU following pool use in labour, without the need for any additional data entry or collection by NHS staff.
The study has a low level of inclusion bias. The study included multiple study sites across varied geographical locations, had a prolonged recruitment period and the model of opt-out resulted in the exclusion of few eligible births. Despite these strengths, there remains the possibility that unidentified confounders may exist between groups, which could not be adjusted for, including reason for getting out of pool, which were associated with rates of the primary and secondary outcomes.
Using routine National Health Service data for research
While using routine NHS data for research has advantages, it continues to hold limitations. Accessing NHS data on scale, was possible, but this was a time-consuming and a technically challenging exercise associated with potential methodological limitations.
The study experienced a protracted set-up duration and had to overcome considerable governance and technical hurdles to obtain data from 26 NHS organisations. Timely approvals were granted by the Health Technology Assessment and CAG with minimal changes to the proposed opt-out model. These approvals did not delay study progress. Site set-up incurred many, and for some sites, lengthy delays due to the series of steps required for approval and amendment of individual site systems. The first contract with study sites was signed in January 2019, and the last was signed in August 2020. The complexity of data extraction from NHS sites reduced the number of data extractions possible and limited the opportunity to review data quality, including completeness, and to provide timely feedback to study sites. Routine NHS MISs hold enormous research potential. Governance processes that span multiple organisations, and MISs designed for research as well as clinical use, would be required if their potential is to be realised.
The use of pseudonymised routine data meant that data captured in maternity records could not be verified by the research team. On occasions, data were clinically implausible, for example, placental delivery was described as ‘removed at caesarean section’ following a spontaneous vaginal waterbirth. In such a situation, where evidently incorrect, data were omitted or corrected, but in such a data set with many records and fields, it cannot be assumed that all such inconsistencies were identified. In other situations, data were accepted as provided, for example, where waterbirth position was ‘lithotomy’. This was not excluded as some free text comments described a ‘lithotomy-like’ position in the pool, suggesting midwives to use the description for women other than those with their legs elevated in supports, and it could not be determined how frequently such descriptions also applied to births out of water. While for most fields, data entry options were restricted to tick-box, some included an option for free text comments. Where these were provided, they were coded, but the amount and quality of information provided in open text varied, and some inconsistencies in the interpretation of comments may have also occurred.
Another limitation of the use of routine data was the chance that when midwives were entering data following a birth, they may not have recalled pool use some time previously, or in the event of transfer to an OU, prior to transfer. This may have led to water immersion not being included in the records of some women and to an underestimate of the proportions of women who used a pool and received obstetric or anaesthetic care prior or during birth.
Study generalisability
Twenty-six NHS study sites were eligible for study participation, and all agreed to participate. The partnered MIS provider was, at the time of study set-up, the most used in the UK. Records were received from all 26 sites relating to births that had occurred over a period of 8 years, from January 2015 to the end of June 2022.
Study sites covered a large geographic spread across England, with one site in Wales, and ranged in size from < 1000 to 15,000+ births annually. All sites provided care in obstetric-led units; 23 had an alongside MLU and 6 had a freestanding MLU. Eleven sites provided level 3 (intensive) neonatal care.
The mean age of women giving birth, in 2019, and, in 2020, in England and Wales was 30.7 years75 compared to the study population of 29.7 years among women who used a pool, and it was 30.3 years among women who did not use a pool. In 2017, in England and Wales, 2.1% of women gave birth at home,76 a figure that has remained relatively stable over several years. Among the POOL Study populations, the highest rates of home birth were 5.0% among women without complications in pregnancy or labour who gave birth in water and were 5.4% among women with risk factors at the time of pool entry.
The generalisability of the POOL Study findings to settings outside of the UK, or outside of NHS practice, is not known. In England and Wales, waterbirth is a commonly provided midwifery service. Within our study, care during water immersion analgesia and at waterbirths was provided by NHS-employed midwives. Although they have varying levels of experience of waterbirth, all are required to adhere to national18 and locally approved guidance. There are clear criteria for pool use and preagreed referral pathways to obstetric and neonatal care in the event of complications or should a woman desire analgesia requiring anaesthetic care. Our findings may be less applicable to countries where care is provided differently, to births unattended by a midwife outside of hospital, including freebirths and BBAs, or to births in the UK but outside of the NHS maternity system.
Equality, diversity and inclusion
The 26 eligible sites in England and Wales using the partnered MIS participated, representing a wide geographical spread. Selection bias in participation was reduced through the inclusion of all births, where care had been provided by the participating NHS organisation regardless of place of birth, the age, ethnicity or any other characteristic of the mother.
Although it was intended that all women using a pool during labour should be informed of the study and their right not to have their data included, it is not possible to determine the proportion of women who received the information as intended, or the extent to which information about the study was translated into another language when required.
Patient and public involvement
The aim of PPI in the study was to provide public and patient involvement during every stage of the research from development of the funding application to report writing and dissemination.
Two PPI coinvestigators have been involved throughout the study as equal members of the SMG and the SSC. One represented the National Childbirth Trust (NCT), the largest parent and childbirth organisation in the UK, and one had expertise in the public communication of health messaging. The NCT representative lives, works and volunteers in a deprived local authority in West Yorkshire, where a significant proportion of families are of South Asian descent. She has also worked as a trainer for a major supermarket chain and was a HR director in a haulage company, giving her extensive experience in listening to and communicating with a variety of audiences. The SSC was chaired by a PPI representative in the latter stages of the study. Our approach to PPI involvement was aligned with the six UK Standards for Public Involvement: communications, impact, governance, working together, inclusive opportunities, and support and learning.77
Changes from published protocol
The main change to the published protocol was the inability to use data held in the ‘medical and surgical’ history of electronic maternity records to identify women with risk factors to inform analysis of the characteristics and outcomes of women with complicating conditions using a pool during labour. In the protocol and SAP, it was intended to identify women with complicating factors at pool entry from two sources: from data held in the medical and surgical menu of electronic records, and, among women giving birth after site opening, from the additional fields of ‘risk factors at pool entry’, which was completed by midwives providing intrapartum care. The latter of these fields were unaffected. Prior to awareness of the unreliability of data held in the ‘medical and surgical’ menu, 13,811 (15.9%) of the cohort had been identified as having a risk factor. Once the identification of women with risk factors at pool entry was reduced to those in the prospective cohort with data captured in the new fields completed during intrapartum care, this figure was reduced to 2725. The necessity to restrict analysis to women with identified risk factors to those identified within the prospective cohort seriously impacted the study’s ability to meet an important objective.
Other analyses were not performed
Several planned analyses could not be performed. Apgar scores recorded at 10 minutes of birth were not routinely collected by sites. It was not possible to determine if the records within the data set relating to women returning to hospital following birth related to assessment only or resulted in an admission. It was also not possible to determine the duration between birth and postnatal return to hospital. All records of maternal return to hospital, including assessment or admission, were therefore included in this outcome. It was not possible to identify babies re-admitted following discharge, as many will be admitted to paediatric wards. Whether or not the baby was breastfed within an hour of birth was poorly captured, and this was done only in a small number of sites.
Implications for policy and practice
The POOL Study’s results support a policy of offering healthy nulliparous and multiparous women with uncomplicated pregnancies the option for using water immersion during labour and birth.
Among nulliparous women without complicating conditions at pool entry, 29% received care prior to birth, which mostly would have necessitated transfer to an obstetric care if birth had been planned at home or in MLU settings. Rates of caesarean section and instrumental births were low and similar to those observed in a cohort of women planning birth in MLU settings in 2008–10.8
Rates of instrumental and operative births were low among women in spontaneous labour. It is important that these findings are shared with women when they are making decisions around planned place and mode of birth, use of water immersion for labour and birth or method of preferred labour onset.
There remains widespread scepticism among neonatal and paediatric healthcare professionals in the UK and internationally about the safety to babies of birth in water.26 This has been reinforced by case reports of adverse neonatal outcomes, following and often attributed to birth in water,26 and the absence of more robust recent evidence describing neonatal outcomes in this group.78 The POOL Study was designed specifically to address these neonatal concerns by including neonatal outcomes, such as any treatment for suspected early onset infection (even in the absence of confirmatory blood culture or other laboratory evidence of infection), which are potentially subjective but remain important to women and families and have important health economic implications. It is very reassuring, therefore, to see that in this large, appropriately powered, cohort of babies born to women deemed to be low risk at onset of labour, birth in water was not associated with a higher risk of NNU admission for respiratory support or of treatment for suspected early onset neonatal infection when compared to pool use in labour with birth out of water.
Obstetric anal sphincter injury was detected no more frequently for primiparous or parous women who used water immersion when they remained in the water to give birth. This finding will form important evidence on which professionals can base informed opinions on maternal outcomes associated with waterbirth. Antenatal information should be specific to women’s parity, given the different rates of obstetric intervention, regional analgesia, modes of birth and OASI.
Healthcare professionals should support women to make evidence-based decision when planning the place and mode of birth, or method of preferred labour onset.
Findings of the qualitative work have several implications for practice: increased exposure to waterbirth is vital to improve the confidence of midwives working in OUs; awareness among obstetricians and neonatologists on the practicalities of pool use, and associated outcomes, could increase the support for water immersion; and improved access to antenatal information would help increase awareness of the option to use a pool. We recommend that OUs increase pool provision, ensure birth room allocation, maximise the use of unit resources, design pool room environments that encourage use and ensure that all midwives take responsibility for being familiar with water immersion equipment and unit guidance.
For policy-makers, the results are important to inform recommended care, decisions about facilities and information provided to prospective families. It is important that the results of the study are interpreted and presented as representing outcomes observed, where care was provided by NHS-employed midwives.
Implications for future studies using an opt-out model
During the period of prospective data collection, the intention was for all women giving birth at study sites to be informed about the study through methods selected by individual sites. At the study end, sites were requested to provide the number of women who had opted out, and 24 sites were able to provide these data. A total of 65 women (range 0–16 women per site) opted out of the study after site opening, with 11 (42%) sites having recorded no opt-outs. It cannot be determined if the low level of opt-out reflected women’s willingness to be included, that some women did not receive the information as intended or that the methods used did not facilitate informed choice. With increasing use of routine data and opt-out models, where identifiable information from individual NHS sites is used in research, research is required to better understand how such systems are utilised in practice, understanding by potential participants and how best to facilitate informed choice around inclusion.
Research priorities
Further studies are recommended into the following:
- interventions to reduce rates of OASI during SVBs
- women’s experiences of use of water during labour and birth, including impact on potential short-term and longer-term impacts, for example, maternal–infant attachment, postnatal depression, sense of control and self-efficacy
- impact of waterbirth on neonatal physiology and transition to extrauterine environment
- blood loss measurement in water
- midwifery care during waterbirth, including in the event of non-spontaneous birth of the fetal shoulders
- care of babies following cord snapping
- teaching of waterbirth skills to student midwives and midwives.
Conclusion
The POOL Study established that among nulliparous and parous women, without antenatal complicating conditions, who used water immersion during labour, and who did not receive additional monitoring or interventions prior to birth, remaining in the water to give birth was not associated with an increase in the incidence of OASI, adverse neonatal outcomes or other measured morbidity.
Current NHS midwifery practice relating to labour and birth in water is safe for women and their babies. Women, parents, families, practitioners and policy-makers should be reassured that birth in water, in the context of NHS care, is not associated with increased risks for mothers or their babies. Women considering or using water immersion during an uncomplicated labour should be informed that remaining in the water to give birth is not associated with an increased risk to themselves or their baby, and they should be supported to make evidence-based individualised decisions on their care choices.