1. Top of page
  2. Abstract
  8. Acknowledgements
  9. References

Objectives  To assess the feasibility of undertaking an adequately powered multicentre study comparing waterbirth with land birth. To assess whether women are willing to participate in such a trial and whether participation has a negative effect on their birthing experience.

Design  A randomised controlled trial (RCT) with ‘preference arm’.

Setting  District general hospital with 3600 deliveries annually.

Population  Women with no pregnancy complications and no anticipated problems for labour/delivery.

Methods  Women were recruited and randomised between 36 and 40 weeks of gestation. Comparison of randomised and ‘preference arm’ to assess any impact of randomisation on women's birthing experience.

Main outcome measures  Data were collected at delivery concerning the labour, the pool water and baby's condition at birth and six weeks of age. The main outcome measures are means and standard deviation of cord O2, CO2, haemoglobin, haematocrit and base excess; medians and ranges of time to first breathe and cord pH; bacterial growth from pool water samples and neonatal swabs; and maternal satisfaction.

Results  Eighty women participated—60 women were randomised. Twenty women participated in a non-randomised ‘preference arm’. The babies randomised to a waterbirth demonstrated a significantly lower umbilical artery pCO2 (P= 0.003); however, it is recognised that this study is underpowered. Women were willing to participate and randomisation did not appear to alter satisfaction.

Conclusion  This small study has shown that a RCT is feasible and demonstrated outcome measures, which can be successfully collected in an average delivery suite.


  1. Top of page
  2. Abstract
  8. Acknowledgements
  9. References

One method of pain relief many women have chosen during childbirth is the use of warm water. There is anecdotal evidence that warm water provides benefits to women in labour by aiding relaxation and reducing the need for pharmaceutical methods of pain relief.1,2 When using a birthing pool, where a woman is immersed in warm water, there are additional benefits because the water supports the woman's body, making it easier for her to remain upright, move around and change position.

However, there is little evidence available concerning the effect on the baby of delivery into water. There have been concerns regarding feto-maternal thermoregulation,3 inhalation of water at delivery,4 infection,5 jaundice,6 snapping of the umbilical cord7,8 and death.9,10 The only randomised controlled trials (RCTs) conducted have looked at the use of water during labour, not birth.11,12

The United Kingdom Central Council for Nurses, Midwives and Health Visitors (UKCC) has recognised ‘that waterbirth is preferred by some women as their chosen method of care and management in labour’ and states that it falls within the duty of care of a midwife.13 The Royal College of Obstetricians and Gynaecologists published guidelines stating that there appears to be no difference in the outcome for the baby born in water compared with the baby born on land.14

There have been calls for ‘scientific’ research evidence to be obtained before healthcare professionals endorse waterbirths.10,15,16 RCTs are thought to be the ‘Gold Standard’ method to obtain an objective evaluation of different treatments or care. Midwives state women would not participate in a trial that would allocate them to either a water or land birth.17 It has also been stated that RCTs are unethical as randomisation would reduce a woman's choice and therefore affect their childbirth experience.18 However, evidence suggests that women have been willing to participate in other RCTs during pregnancy and labour19 without any suggestion that their childbirth experience was adversely affected.

This pilot study was undertaken to investigate these issues to assess whether it would be feasible to attempt a larger multicentre RCT, which is needed to provide substantive evidence concerning the effect on the baby of a waterbirth compared with a land birth. This article considers data obtained concerning the condition of the baby at birth, women's willingness to participate in the study and satisfaction with their birthing experience. Ethical approval was obtained from the Northampton Medical/Research Ethics Committee.


  1. Top of page
  2. Abstract
  8. Acknowledgements
  9. References

The trial randomly allocated 60 women who consented to participate in the RCT to either a waterbirth or a land birth. As our experience was that approximately 50% of women in the waterbirth arm of the trial may not deliver in the pool, it was decided to randomise at a ratio of 2:1 waterbirth to land birth. Forty were randomised to a waterbirth and 20 to a land birth.

The trial design also incorporated a non-randomised ‘preference arm’. Consent was obtained from 20 women—10 who definitely wanted a waterbirth and 10 who definitely wanted a land birth.

The pilot study was undertaken to assess how long it would take to randomise 60 women and determine whether randomisation was detrimental to participants' birthing experience.

Women were recruited through their community midwife, posters in antenatal clinics and by publicity in the local newspaper, radio and television. Women who were interested either contacted the researcher themselves or gave permission for their contact details to be given to the researcher. Women were consented and randomised between 36 and 40 weeks of gestation.

A randomisation schedule was provided by the National Perinatal Epidemiology Unit. A person unconnected with the study prepared sequentially numbered, opaque envelopes containing cards with the allocation to either ‘land’ or ‘water’. When a woman had given written consent, she opened one of the consecutively numbered randomisation envelopes. These were opened in the correct numerical order as the women were recruited. Women were aware that participation in the study was voluntary and that they were able to withdraw from the study at any time without affecting their care.

Women were eligible to join the study if they were aged between 18 and 50 years, had no complications during pregnancy and no envisaged problems for labour/delivery. Women were excluded if labour occurred before 37 weeks of gestation, they had a previous caesarean section, confirmed intrauterine growth retardation or any medical condition or complication of pregnancy that excluded birth in the pool.

The following data were collected on maternal outcomes: age, social history, length of first and second stage of labour, hourly maternal pulse and temperature in labour, analgesia used in labour, whether labour was augmented, any instrumental/surgical intervention and indication, condition of perineum after delivery and length of time in the water. Maternal satisfaction with their birthing experience was assessed by a numbered scale (0 =meaning not at all satisfied and 6 =meaning extremely satisfied) completed before leaving the labour ward and women's thoughts about participating in the trial were obtained six weeks after delivery by postal questionnaire.

The following information was collected concerning the birthing pool water: hourly water temperature and checking for any signs of meconium. A sample of pool water was obtained before mother entered pool and another sample collected at the end of labour for microbiological analysis.

The following data were collected to assess the condition of the baby at birth: cord arterial and venous pH, O2, CO2, cord haemoglobin and packed cell volume, Apgar score—timed with stop watch at 1, 5 and 10 minutes, time to first respiration, incidence of torn cord, rectal temperature at birth, swabs from baby's ear, umbilicus and mouth for culture, sensitivity and microscopy, method of feeding, date and time of first breast or bottle feed. After birth information was obtained concerning incidence of admission to neonatal intensive care and interventions undertaken, neonatal jaundice requiring treatment, infection requiring antibiotic therapy, the condition of baby six weeks after delivery and any problems since discharge, including mortality.

Statistical analysis was performed using Statistical Analysis for Social Scientists (SPSS version 10, SPSS UK, Working, 1999). Analysis of the randomised arm of the trial was undertaken on an ‘intention-to-treat’ basis. Means were compared using independent sample t test or Mann–Whitney U test/Wilcoxon's signed rank test. Nominal data were compared using χ2 or Fisher's Exact Test. The ‘preference arm’ of the trial was then compared with the randomised arm by one way between subjects ANOVA for continuous measurements and Kruskal–Wallis test for non-parametric data.


  1. Top of page
  2. Abstract
  8. Acknowledgements
  9. References

Recruitment was carried out over 11 months. Sixty women were randomised and 20 women joined the ‘preference arm’. Information concerning how many women were aware of the project or how many women were eligible to join the study during the recruitment period was not collected. Reasons for not participating in the study are given in Fig. 1.


Figure 1. Flow diagram of trial participants.

Download figure to PowerPoint

Sixteen (40%) women randomised to the pool did not actually use it for their labour or baby's birth. Thirteen (32.5%) women left the pool during the first stage of labour. The reasons are listed in Table 1. There were 10 actual waterbirths out of a possible 40 women randomised to a waterbirth. Because of an administrative error, one woman randomised to a land birth laboured and delivered her baby in the pool. Another woman who was randomised to a land birth used the pool during the first stage of labour.

Table 1.  Reasons why women did not use the pool or left the pool during labour (may be more than one reason per person).
 RCT waterbirth armWater preference arm
Reasons for not using pooln= 16n= 4
Meconium-stained liquor71
Prelabour/prolonged rupture membranes40
Raised blood pressure on admission30
Delivered before entered pool51
Pool unavailable20
Undiagnosed breech11
Fetal distress on admission11
Did not want to use pool when in labour10
Maternal pyrexia at start of labour10
Maternal HVS cultured group B strep prior to labour20
Maternal vulval warts noticed in labour—midwife felt inappropriate for mother to use pool.10
Reasons for women leaving the pooln= 13n= 1
Failure to progress in labour51
Requested more pain relief30
Meconium-stained liquor20
Left pool for assessment—cervix fully dilated did not want to re-enter pool10
Not comfortable in the pool20

The baseline characteristics of the women who randomised are shown in Table 2. Women randomised to water left education at a younger age than those allocated to land birth. Labour complications were experienced by 23 (57.5%) of the waterbirth women compared with 12 (60%) of the women delivering on land (χ2= 0.34, df= 2, P= 0.85). The short term outcomes are shown in Table 3. Women's response to the six-week questionnaire is shown in Table 4. Five women noted that midwives did not appear to support their decision to use water as a method of pain relief and that midwives did not seem willing to collect the research samples.

Table 2.  Characteristics of the women at trial entry. Values are given as n (%) or mean [SD].
 Random allocationRandomised arm PPreference armANOVA
Water (n= 40)Land (n= 20)Water (n= 10)Land (n= 10)
Age left full time education (years)17.2 [1.6]18.5 [2.6]0.0518.0 [2.2]18.2 [2.9]0.12
Professional qualifications9 (23)9 (45)0.135 (50)4 (40)0.19
Age at delivery (years)28.4 [5.4]28.7 [4.2]0.8429.5 [4.5]27.9 [4.7]0.90
Primiparous24 (60)15 (75)0.395 (50)3 (30) 
Gestation (completed weeks)39 [1.1]39 [1.1] 39 [0.6]39 [1.0] 
White39 (97)19 (95) 9 (90)9 (90) 
Black1 (3)1 (5) 1 (10)1 (10) 
Table 3.  Women's outcomes: randomised arm. Values are given as n (%) or mean [SD].
 Random allocationP
Water (n= 40)Land (n= 20)
  • *

    Did not deliver in pool.

  • Measured on a scale of 0–6 (0 =not at all satisfied, 6 =extremely satisfied).

Length of first stage (minutes)420.8 [225.4]409.4 [265.0]0.87
Length of second (minutes)47.3 [46.8]58.7 [44.3]0.38
Method of delivery
Spontaneous vaginal34 (85)14 (70)0.17
Forceps/ventouse4 (10)3 (15)0.68
Caesarean section2 (5)3 (15)0.38
Water21 (55)2 (10) 
Entonox33 (83)14 (70) 
Meptazinol7 (18)4 (20) 
Pethidine2 (5)5 (25) 
TENS3 (8)2 (10) 
Epidural6 (15)7 (35) 
None (except water)4 (10)1 (5)0.18
Maternal temperature (°C)36.9 [0.5]36.7 [0.8]0.72
Length of time in pool (minutes)146 [133]  
Intact9 (23)6 (30)0.54
Episiotomy9 (23)*6 (30) 
Labial tear5 (13)3 (15) 
Vaginal tear2 (5)0 
First degree tear7 (18)2 (10) 
Second degree tear8 (20)3 (15) 
Third degree tear1 (3)0 
Maternal satisfaction†4.32 [1.20]4.29 [1.26]0.94
Table 4.  The responses given by women to explain why they would participate in a similar trial again and why participating did not affect their birthing experience.
 Number of women*
  • *

    Some women gave more than one answer.

Why they would participate again
Happy to help provide health information14
Want to help future parents11
To support research10
 Research is good as it provides more information (n= 8) 
 Research improves care so mothers should join studies (n= 2) 
Did not interfere/change anything8
To help midwives6
Want to improve experience of childbirth for women in the future5
Fulfilling/exciting to participate4
Helped me get desired birth (trial midwife available to support waterbirth)3
To help future babies3
Still would not mind either method3
 Undecided which to choose (n= 2) 
 Would hope to get waterbirth next time (n= 1) 
Built up a relationship with a midwife who was present at the birth2
Nothing sinister/worrying about joining a research project2
No pressure to take part so feel happy about joining again2
Did not cost me anything1
No reason given6
Why joining did not affect birthing experience
Happy with either a waterbirth or a land birth15
Did not receive different care from other women6
Wanted water and randomised to water4
Happy to help provide information4
No pressure to go into study, discussed it thoroughly with midwife4
It depends on ‘circumstances on the day’ as to whether a woman can use the pool—not guaranteed so happy to participate4
All I cared about was a healthy baby—I was happy with trial care2
No woman can be sure of the outcome of labour1
No reason given22

The baby outcomes are shown in Table 5 and the results of the microbiology in Table 6. The only statistically significant result obtained was a higher land birth pCO2 than waterbirth group. However, it is recognised that this study is under powered.

Table 5.  Outcomes for babies (randomised arm). Values are given as n (%), mean [SD] or median {range}.
 Random allocationP
Water (n= 40)Land (n= 20)
  • *

    Number of samples obtained.

Weight (grams)3500 [414.3] 39*3468 [485.5] 20*0.78
Head circumference (cm)34.5 [1.31] 39*34.4 [1.38] 20*0.67
Length (cm)50.9 [2.6] 39*51.0 [2.30] 20*0.81
Sex of baby
Female23 (57)7 (35) 
Male17 (43)13 (65) 
Incidence of meconium stained liquor10 (25)4 (20)0.67
Apgar <8 at 5 minutes10 
Breast feeding31 (78)18 (90)0.47
Time to first feed (minutes)56.5 {23–833} 34*60.0 {17–909} 17*0.53
Cord arterial pH7.23 {7.037–7.403} 35*7.18 {7.045–7.260} 13*0.15
Cord venous pH7.32 {7.147–7.520} 36*7.33 {7.147–7.424} 16*0.59
Cord arterial oxygen (kPa)2.86 [1.13] 35*2.41 [0.74] 12*0.21
Cord arterial carbon dioxide (kPa)6.26 [1.49] 36*7.76 [1.24] 12*0.003
Cord haemoglobin (g/dl)15.86 [1.54] 30*16.96 [2.13] 9*0.06
Cord haematocrit0.514 [0.06] 28*0.546 [0.08] 9*0.15
Arterial cord base excess mmol/L)−8.80 [4.36] 35*−8.59 [3.98] 12*0.89
Time to first breath (seconds)6.10 (1.0–251) 39*4.5 (1.0–35) 16*0.60
Rectal temperature (°C)36.7 [0.467] 37*36.6 [0.528] 16*0.73
Antibiotics (number of babies)310.71
Incidence of jaundice10 
Admission to neonatal intensive care unit310.71
Length of hospital stay (hours)28.0 {3–720} 39*30.5 {11–97} 20*0.70
Readmitted to hospital after discharge10 
Table 6.  Results of the swabs taken from baby at delivery: organism cultured and number in each of the trial arms.
OrganismRandomised waterbirthRandomised land birthPreference arm waterPreference arm land
Mouthn= 36n= 15n= 7n= 9
Staphylococcus aureus1001
Staphylococcus epidermidis3011
Escherichia coli4100
Enterococcus faecalis1000
No growth261357
Earn= 37n= 15n= 7n= 10
S. epidermidis7101
E. coli4010
Candida species3000
Haemophilus influenzae0100
Ent. faecalis0110
β-Haemolytic group B strep0100
α-Haemolytic Streptococcus0010
Bacillus species0001
No growth241228
Umbilicusn= 36n= 15n= 7n= 8
S. epidermidis4010
S. aureus0001
E. coli4100
β-Haemolytic group B strep1000
Bacillus species0001
No growth251355
Water samples
Pre-use of the pool
 P. aeruginosa2 1 
Post-use of pool
 P. aeruginosa2 1 

The results of the comparison between the ‘preference arm’ and the randomised arm and shown in Table 7. There was no statistically significant difference between age at delivery (F3,76= 0.199, P= 0.897) and age of completion of full-time education (F3,76= 2.018, P= 0.118). There was also no statistically significant difference among the gestation of pregnancy, the babies length (F3,73= 0.136, P > 0.938), head circumference (F3,73= 1.959, P > 0.128) or weight (F3,75= 0.909, P > 0.441) (Table 8).

Table 7.  Outcomes for women in preference arm with ANOVA/Kruskal–Wallis significance comparing randomised and preference arm. Values are given as n (%) or mean [SD].
 Preference armANOVA/Kruskal–Wallis comparison with randomised arm
Water (n= 10)Land (n= 10)P
  • *

    Did not deliver in the pool.

  • Measured on a scale of 0–6 (0 =not at all satisfied, 6 =extremely satisfied).

Length of first stage of labour (minutes)294.3 [223.2]515.0 [280.6]0.31
Length of second stage of labour (minutes)35.1 [44.6]69.4 [80.1]0.46
Method of delivery
Spontaneous vaginal7 (70)9 (90)0.37
Forceps/ventouse1 (10)1 (10) 
Caesarean section2 (20)0 (0) 
Water6 (60)0 
Entonox6 (60)7 (70) 
Meptazinol2 (20)5 (50) 
Pethidine2 (20)2 (20) 
Epidural2 (20)1 (10) 
None (except water)01 (10)0.78
Maternal temperature (°C)36.9 [0.3]36.8 [0.6]0.27
Length of time in pool (minutes)82 [46] 0.27
Intact5 (50)3 (30)0.39
Episiotomy2 (20)*1 (10) 
Labial tear2 (20)1 (10) 
Vaginal tear1 (10)1 (10) 
First degree tear02 (20) 
Second degree tear1 (10)4 (40) 
Third degree tear00 
Maternal satisfaction&dagger;4.50 (0.55)3.78 (0.97)0.58
Table 8.  Outcomes for babies (‘preference arm’) with ANOVA/Kruskal–Wallis significance comparing randomised and preference arm. Values are given as n (%), mean [SD] or median {range}.
 Preference armANOVA/Kruskal–Wallis comparing randomised/preference arms
Water (n= 10)Land (n= 10)P
  • *

    Number of samples obtained.

  • One woman in each ‘preference arm’ delivered at home.

Weight (grams)3723 [384.3] 10*3464 [437.0] 9*0.441
Head circumference (cm)35.4 [1.67] 9*34.1 [1.80] 9*0.128
Length (cm)50.7 [2.56] 9*51.3 [1.94] 9*0.938
Sex of baby
Female5 (50)4 (40) 
Male5 (50)6 (60) 
Apgar <8 at 5 minutes00 
Cord arterial pH7.24 {7.16–7.37} 7*7.20 {7.12–7.30} 7*0.336
Cord venous pH7.28 {7.25–7.48) 5*7.33 {7.27–7.49} 10*0.727
Cord arterial oxygen (kPa)2.57 [1.41] 7*2.60 [0.62] 7*0.61
Cord arterial carbon dioxide (kPa)6.14 [0.81] 7*6.74 [0.77] 7*0.009
Cord haemoglobin (g/dl)14.7 [0.78] 3*16.3 [1.79] 8*0.067
Cord haematocrit0.489 [0.05] 3*0.526 [0.05] 8*0.092
Arterial cord base excess (mmol/L)−6.83 [4.49] 7*−8.39 [3.34] 7*0.730
Time to first breath (seconds)21.0 (6–38) 5*2.5 (1–33) 10*0.097
Rectal temperature (°C)36.8 [0.465] 10*36.5 [0.359] 9*0.696
Antibiotics (number of babies)01 
Time to first feed (minutes)50.0 {19–152} 9*50.0 {22–80} 9*0.757
Breast feeding8 (80)8 (80)0.781
Admission to neonatal intensive care unit01 
Incidence of jaundice00 
Length of stay in hospital (hours)19.5 {0–93}&dagger;20.5 {0–110}&dagger;0.366

The maternal satisfaction score demonstrated no statistical difference in scores between the randomised and ‘preference arm’ (P= 0.58). There were no statistical differences in maternal or child outcomes between the two trial arms.


  1. Top of page
  2. Abstract
  8. Acknowledgements
  9. References

This is the first RCT to randomise women to either a waterbirth or a land birth. Recruitment was slower than wished as the grant paid for a midwife to work just 4 hours a week on the project.

Many of the uncertainties about undertaking a waterbirth RCT related to users of maternity services, for example, would women recruit to a study that would randomly allocate them to a waterbirth or a land birth? This pilot study indicates that women were willing to participate. Most, even if keen to use the pool appreciated that they could not be guaranteed use of it. Many, who had not thought about using the pool, were happy to participate to help produce evidence for healthcare professionals and other parents. There was a 100% response rate for the six-week questionnaire—this would seem to suggest that women continued to support the study, although the majority had not achieved a waterbirth.

There was also concern that the women who participated in the RCT may be different in some way from the ‘normal’ users of the pool and subsequently may have different outcomes. If so, a RCT would not provide evidence appropriate to the ‘normal’ pool users. The inclusion of a ‘preference arm’ in this pilot study was to compare the characteristics and the outcomes of the women who were randomised with those who joined the ‘preference arm’. There were no statistically significant differences between the women's characteristics and the outcomes of the two trial ‘arms’ for either the women or their babies. This suggests that randomisation does not affect the characteristics of women using the pool and, therefore, healthcare professionals could have confidence in the results obtained from a larger RCT.20

Another concern raised in relation to a RCT was whether women's satisfaction with their birthing experience would be affected by the fact that they themselves did not decide whether to use the pool.18 The results of this study would indicate that randomisation does not affect women's satisfaction as no statistically significant difference was found when comparing the scores of women in the two trial ‘arms’. It was expected that women who randomised to land might be disappointed and have a lower score. However, the mean score for women in the ‘preference arm’ for land was lower than women randomised to land indicating that feelings of satisfaction may involve more than just the ability to choose.

Women were randomised to the study after 36 weeks of gestation, so that they knew their allocation before they went into labour. The randomisation process was weighted towards waterbirths but there were still only 10 (25%) babies born in water in the waterbirth arm. However, when considering only those randomised women who entered the pool, then 44% actually achieved a waterbirth—which is near the expected figure. Unfortunately, many women were not eligible to use the pool when admitted in labour, which backs Brown's21 findings that only 50% of women who prebooked the pool prior to labour actually used it. If a larger study is organised, then one consideration must be when to randomise. One suggestion would be to consent after 36 weeks but delay randomisation until arrival on the labour ward to enable women to be reassessed for eligibility and this would also ensure that the pool was available for use.

There were concerns that data and sample collection would not be feasible due to poor midwifery staffing levels.22 Unfortunately, some of the midwives did not collect samples. The reason was often stated in the notes as ‘there was too much going on at the delivery’. There seemed to be a lack of appreciation that the babies needing resuscitation were the ones who required cord blood gas analysis to help assess their condition at birth. However, it is now routine at many maternity units to obtain cord arterial and venous blood for analysis especially when there is concern about a baby's condition at birth. If a similar study was repeated, cord blood samples would be obtained more effectively as midwives are now familiar with the procedure.

It has been speculated that a rise in maternal temperature, while using the pool, may have contributed to the deaths of two babies by affecting the fetal–maternal thermoregulation.3 The fetus is totally dependent on the mother for thermoregulation.23 Heat produced by the uterus must flow to the maternal body to be dissipated. For this physiological process to occur, a fetal–maternal temperature differential of 0.5°C is required. Therefore, an important factor for the wellbeing of the fetus is the maternal temperature in labour. There was no statistical difference in mean maternal temperature or for baby's rectal temperature at birth between the groups, which supports evidence that careful monitoring of both maternal temperature and water temperature maintains fetal–maternal thermoregulation.

The fetal condition during labour is assessed by intermittent auscultation of the fetal heart. Due to abnormal observations, some women (n= 2) were advised to leave the pool due to anxieties about the baby. It is important that women appreciate the need to leave the pool, if advised, to prevent possible risk to the baby at birth. By being vigilant in assessing both mother and fetus, midwives hope to prevent a fetus that is not coping with labour being delivered in water. It has been reported that the baby appears to be calmer when delivered in the pool24 and few cry at birth. The longer median time to first breath in the waterbirth arm was not significant in this study and would be expected as a result of the short delay between birth and the baby coming to the surface of the water.

There was no evidence of clinical neonatal infection, which confirms Garland and Jones'25 findings. Brown21 reported that culture from ear swabs grew Pseudomonas aeruginosa on two occasions. It has been suggested that P. aeruginosa may be present in stagnant water remaining in the pool taps if the pool has not been used on a daily basis.26 Running the pool taps for 5 minutes each day may prevent this occurrence. Many of the bacteria cultured on the neonatal swabs were normal skin flora, probably due to contamination from early skin to skin contact in the waterbirth group, which caused no clinical problems.

Studies have noted an increase in incidence of snapped umbilical cord when a baby is born in water.5,7 but there was no such case in this pilot study. Another concern is increased risk of jaundice caused by the cord pulsating for longer due to the warm water, but there was no difference in haematocrit and haemoglobin levels found in this study between land or waterbirths and only one baby whose mother was randomised to water was treated for jaundice.

This was a pilot study, so a larger trial is required to assess more reliably the risks of waterbirth. To have a power of 80% of showing that the mean pH is equivalent in babies born in water or on land, it would be necessary to include at least 1220 births, equally randomised to the two methods, with the assumption that ‘equivalent’ is defined as having a mean pH within 0.02 of each other.

Other possible primary outcomes are percentage of babies with cord pH below 7.05, percentage with Apgar less than 7 at 5 minutes, and percentage admitted to a neonatal intensive care unit. These would all lead to larger studies for the same power, of the order of 1500–2500 births, taking equivalence to 0.5%.


  1. Top of page
  2. Abstract
  8. Acknowledgements
  9. References

This study demonstrates that women are willing to participate in a waterbirth RCT. Randomisation did not affect women's satisfaction with their birthing experience and most women responded positively when asked whether they would participate in a similar trial. The results of this study support the feasibility of organising a multicentre RCT, which could fully evaluate the differences between land and waterbirths on a large enough sample size to provide statistical significance.


  1. Top of page
  2. Abstract
  8. Acknowledgements
  9. References

We would like to acknowledge the help and support given to us by the midwives (M. Ashton, L. Bates, J. Bennett, A. Bithray, I. Blencoe, E. Brown, B. Boss, F. Bradbury, J. Buckler, J. Butler, J. Cale, S. Cleary, F. College, L. Cooper, S. Cooper, J. Clarke, R. Dempsey, D. Donaldson, L. Edwards, S. Edwards, L. Flemming, J. Franklin, W. Gadsden, H. Goode, M. Goodman, K. Grey, H. Groome, J. Harkness, E. Hawtin, V. Heslip, S. Hesselden, D. Honey, J. James, A. Jeffers, M. Josephs, T. Lodge, C. McGovern, Y. McKenzie, J. New, M. Perkins, J. Percival, D. Rapley, A. Richley, J. Romecin, C. Robins, C. Robotham, L. Ryan, C. Sullivan, A. Taylor, L. Taylor, L. Valentine, B. Verney, K. Wenman, C. White, B. Williams, J. Winwright) and the obstetrician (E.J. Shaxted). We also acknowledge the help we had concerning research support and advice (Oxford RDSC), statistical advice (R. Weatherall, MSc, NHS/CR-UK Centre for Statistics in Medicine, Oxford), database (H. Pepper, Northampton General Hospital NHS Trust), consent form (S. Gilling, Northampton General Hospital NHS Trust), randomisation schedule (National Perinatal Epidemiology Unit, Oxford), advice concerning collection of samples (M. Burgess-Wilson, T. Ward, T. Whitlock), analysis of cord blood samples (nurses/midwives in neonatal intensive care unit) and funding (partly funded by Getting Started in Research Grant from Northampton General Hospital NHS Trust).


  1. Top of page
  2. Abstract
  8. Acknowledgements
  9. References
  • 1
    Beech B. Water birth—a passing fad? Modern Midwife 1998;7(5):1114.
  • 2
    National Childbirth Trust. Labour and Birth in Water. London: NCT, 1995.
  • 3
    Rosevear S, Fox R, Marlow N, Stirrat G. Birthing pools and the fetus. Lancet 1993;342: 10481049.
  • 4
    Nugyen S, Kuschel C, Teele R, Spooner C. Water birth—a near-drowning experience. Pediatrics 2002;110: 411413.
  • 5
    Hawkins S. Water vs. conventional births: infection rates compared. Nurs Times 1995;9(11):3840.
  • 6
    Austin T, Bridges N, Markiewicz M, Abrahamson E. Severe neonatal polycythaemia after third stage of labour underwater. Lancet 1997;350: 1445.
  • 7
    Rosenthal M. Warm-water immersion in labor and birth. Female Patient 1991;16: 3547.
  • 8
    Gilbert R, Tookey P. Perinatal mortality and morbidity among babies delivered in water: surveillance study and postal survey. BMJ 1999;319: 483487.
  • 9
    Odent M. Birth under water. Lancet 1983;24(31):14761477(December).
  • 10
    Atalla R, Weaver J. Labour and birth in water safety has yet to be determined. BMJ 1995;311: 390.
  • 11
    Eckert K, Turnball D, MacLennan A. Immersion in water in the first stage of labor: a randomized controlled trial. Birth 2001;28(2):8493.
  • 12
    Ohlsson G, Buchhave P, Leandersson U, Nordstrom L, Rydhstrom H, Sjolin I. Warm tub bathing during labor: maternal and neonatal effects. Acta Obstet Gynecol Scand 2001;80: 311314.
  • 13
    UKCC Position Statement on Waterbirths. Annex 1 to Registrar's Letter 16/1994.
  • 14
    Royal College of Obstetricians & Gynaecologists. Birth in Water. Available: (Accessed 2 June 2001).
  • 15
    Aird I, Luckas M, Buckett W, Bousfield P. Effects of intrapartum hydrotherapy on labour related parameters. Aust N Z J Obstet Gynaecol 1997;37(2):137142.
  • 16
    Alderdice F, Renfrew M, Marchant S, Ashurst H, Berridge G, Garcia J. Labour and birth in water in England and Wales. BMJ 1995;310: 837.
  • 17
    Jenkins assessing the effect of a new health technology. In: BeechB, editor. Waterbirth Unplugged Proceedings of the First International Water Birth Conference. London: Books for Midwives, 1996: 5358.
  • 18
    Jowitt M. Problems with RCTs and midwifery. Midwifery Matters 2001;91: 910(Winter).
  • 19
    Rogers J, Wood J, McCandlish R, Ayres S, Truesdale A, Elbourne D. Active versus expectant management of third stage of labour: the Hinchingbrooke randomised controlled trial. Lancet 1998;351: 693699.
  • 20
    Fielding L, Grace R, Hittinger R. Patients who are eligible but not randomised should be included as additional comparative arm in study. BMJ 1999;318: 874875.
  • 21
    Brown L. The tide has turned: audit of water birth. Br J Midwifery 1998;6(4):236243.
  • 22
    Forde C, Creighton S, Batty A, Hawdon J, Summers-Ma S, Ridgway G. Labour and delivery in the birthing pool. Br J Midwifery 1999;7(3):165171.
  • 23
    Power G. Biology of temperature: the mammalian fetus. J Dev Psychol 1989;12: 295304.
  • 24
    Burns E, Greenish K. Pooling information. Nurs Times 1993;89(8):4749.
  • 25
    Garland D, Jones K. Waterbirth: updating the evidence. Br J Midwifery 1997;5(6):368373.
  • 26
    Robb E, Spiby H, Stewart P, Norman P. Hygiene in birthing pools. Nurs Times 1991;25(87):14.

Accepted 17 February 2004