- Top of page
Objective To compare an in-hospital birth centre with standard maternity care regarding medical interventions and maternal and infant outcome.
Background The birth centre care was characterised by comprehensive antenatal, intrapartum and postpartum care with the same team of midwives, restricted use of medical technology, and discharge within 24 h after birth.
Methods Of 1860 women meeting low risk medical criteria in early pregnancy and interested in birth centre care, 928 were randomly allotted birth centre care and 932 standard maternity care. Data were collected mainly from hospital records, and analysis was by intention-to-treat.
Results Of the women in the birth centre group, 13% were transferred antenatally and 19% intrapartum. No statistical differences were observed in maternal morbidity or in perinatal mortality, neonatal morbidity, Apgar score or infant admissions to neonatal care. Perinatal mortality, defined as intrauterine death after 22 weeks of gestation and infant death within seven days of birth, occurred in eight cases (0.9%) in the birth centre group and in two cases (0.2%) in the standard care group (OR 4.04, 95% CI 0.80 to 39.17; P= 0.11). Subgroup analysis showed that a larger proportion of first-born babies in the birth centre group (15.6%) were admitted for neonatal care than in the standard care group (9.5%) (P= 0.003), whereas the converse was the case for the newborns of multiparous women: 4.7% and 8.4%, respectively (P= 0.04). The overall rates of operative delivery (e.g. caesarean section, vacuum extraction and forceps), 11.1% in the birth centre group and 13.4% in the standard care group, did not differ statistically (P= 0.12), but obstetric analgesia, induction, augmentation of labour and electronic fetal monitoring were less frequently used in the birth centre group. Labour was 1 h longer in the birth centre group.
Conclusion Birth centre care was associated with less medical interventions than standard care without any statistically significant differences in health outcomes. However, the excess of perinatal deaths and of morbidity in primigravidas’ infants in the birth centre group gives cause for concern and necessitates further studies.
- Top of page
Birth centres offer an alternative form of maternity care for pregnant women who have been assessed as having low risk of obstetric complications and who want continuity of care and a family-oriented, homelike environment when giving birth. Birth centres encourage natural childbirth, limiting the use of medication and technology during labour.
A descriptive study of nearly 12,000 women admitted for labour to free-standing birth centres concluded that birth centres offer a safe and acceptable alternative to hospital confinement for selected low risk women, particularly multiparous women1–4. A retrospective follow up study of more than 5000 women booked with an in-hospital birth centre drew the same conclusion5, and still another study comparing the outcome of nearly 3000 women at an in-hospital birth centre with those receiving standard care found no difference regarding safety6.
No randomised controlled study of comprehensive birth centre care has yet been published, but a midwife-managed delivery unit was studied in one randomised controlled trial comprising 2844 women7, while in another trial, including 3510 women, midwife-led maternity care was studied, including antenatal visits and delivery in birthing rooms adjacent to a delivery suite8. No statistical difference was observed regarding maternal or infant outcome in either of these trials, though in the latter the numbers of stillbirths and early neonatal deaths were slightly higher in the midwife-led group than in standard care.
The safety of birth centres, including the antenatal care, needs further study in randomised controlled trials analysed by the intention-to-treat model. This paper describes medical interventions, and maternal and infant outcome in one such study.
Characteristics of care
Birth centre care in this study included integrated antenatal, intrapartum, and postpartum care. The parents were cared for by the same team of midwives, and in the same premises, from the outset of pregnancy, throughout the birth, and up to the final visit two months after the birth. The women gave birth at the centre in gestational weeks 37 to 43. If they went into labour before or after these weeks, they were transferred to standard care. Pharmacological pain relief, induction, augmentation of labour and electronic fetal monitoring (EFM) were available only after transfer to the hospital's standard delivery ward, located one storey above the centre. During pregnancy a woman could be referred for EFM or ultrasound scanning on specific medical grounds and then continue with birth centre care, provided the unit's medical criteria were still fulfilled. The midwives assisted women in labour without the presence of a doctor. They made their own decisions about transfer in labour, according to medical guidelines set up by the obstetrician having medical responsibility for the centre. Standard care in this study was the usual form of public maternity care offered to women in the Greater Stockholm area, with approximately 75 community centres providing antenatal care (two of which were private) and seven hospitals providing intrapartum and postpartum care. Women could make their own choice regarding antenatal clinic and hospital for the birth, but they usually attended the agency located nearest their homes. In standard maternity care different teams of midwives took care of women during pregnancy, labour and birth and postpartum, and in different premises. In Sweden midwives in standard care also handle normal deliveries, although an obstetrician is usually available on the delivery ward. The main differences between the two models of care are presented in Table 1.
Table 1. Characteristics of care.
| ||Birth Centre care||Standard care|
|Premises for antenatal, intrapartum and postpartum care||Same||Different|
|Continuity of care from early pregnancy to postpartum check-up||Same team of midwives||Different midwives|
|Recommended visits to midwife||8–11||10–12|
|Recommended visits to doctor||On medical indication||2|
|Doctor at hand during labour||No||Yes|
|Pharmacological pain relief (epidural, pethidine, N2O)||No||Yes|
|Induction or augmentation of labour||No||Yes|
|Electronic fetal monitoring||No||Yes|
|Transfer in event of medical complications||Yes||No|
|Length of postpartum stay||24h||3–5 days (mean)|
|Postpartum home visits offered||Yes||By 2 of 7 hospitals|
- Top of page
The birth centre opened in October 1989 as the first of its kind in the Greater Stockholm area. From the outset, care at the centre was integrated in the present trial. Leaflets describing the birth centre and the trial were distributed at all the antenatal clinics in Greater Stockholm to inform expectant mothers, who might also have learned of the centre through friends or the mass media. At the women's first telephone contact with the birth centre they were already self-selected, and very few were turned away at this point. An appointment was made for a meeting at the centre, where a research assistant gave additional information if required.
Women's understanding of the random assignment was checked, and they were then asked to pick an opaque envelope from a box. Letting the women themselves select the envelope was considered to increase their involvment and understanding of the randomisation procedure. In order to obtain groups of similar size, 100 envelopes were prepared at a time, 50 containing a paper strip with the text ‘Birth Centre Care’ and 50 with ‘Standard Care’. A new batch of 100 envelopes was added when only a few remained in the box. The envelopes were mingled, and it was not possible for any member of the research team, or the woman herself, to predict the group allocation.
Criteria for joining the birth centre trial included willingness to participate in the research project with random allocation and to answer questionnaires. The women had to be residents of Greater Stockholm and at least one partner in each expectant couple had to be Swedish-speaking. Women with a complicating general condition (e.g. diabetes or hypertension), drug abusers, and women who continued to smoke during the present pregnancy were excluded from the trial. A history of low birthweight, preterm birth, perinatal death, or a difficult vaginal delivery did not preclude participation. Women with a previous caesarean section were accepted if their last delivery was vaginal. There were no preconditions regarding maternal age or height. Women were encouraged to enrol in the trial as early in pregnancy as possible, although they were allowed to join throughout pregnancy. A minimal requirement for inclusion was one antenatal visit.
The project was initially funded for a two-year period. In order to increase the power of detecting differences in medical interventions (such as caesarean section and epidural rates and rates of newborn transfers to a neonatal care unit) the trial was allowed to continue for another 1.5 years, but not longer. Altogether 1860 women were included in the final sample, 928 women allotted to the birth centre group, and 932 to the standard care group. This sample provided power (confidence 95%, power 80%) to detect a reduction in the caesarean section rate from an expected 10% in the standard care group to 6.3% in the birth centre group, a reduction in the rate of epidural anaesthesia from an expected 16% to 11.4%, and an increase in the rate of neonatal transfers from an expected 10% to 14.4% in the respective groups. The expected date of birth among the 1860 women (1869 infants) in the sample ranged from the initiation of the trial in October 1989 to the end of June 1993.
Two women in the standard care group were lost to follow up: one emigrated and the other withdrew from all participation in the trial. In addition, eight hospital records were missing, four because of home births. In these cases essential information was collected by telephone interviews with the mothers, from the National Birth Register and from the delivery ward register. Data on interventions and maternal and infant health were extracted from the case records, and information about the participants’ background characteristics from self-completed questionnaires filled out on the women's first visit to the birth centre before the randomisation. A detailed description of the participants in the trial has been published elsewhere9. This report covers the period from early pregnancy to the end of the first week after the delivery. A follow up of infant health, one year after the birth, is in progress.
The strategy of analysis was by intention-to-treat (i.e. all women were included in their allotted group). Consequently, those women who were allocated birth centre care but were transferred antenatally, intrapartum, or postpartum, or who withdrew from birth centre care at their own request were included in the birth centre group.
The sample size was too small to reveal differences in perinatal mortality or serious perinatal asphyxia (national figures being 0.6% and 1.2%, respectively)10. The individual cases of perinatal death, and some of serious morbidity, are described hi detail together with our comments on possible avoidable factors in a separate technical report that can be requested from the authors. The cases in which possible avoidable factors were identified are briefly presented in this paper. Two independent doctors (a paediatrician and an obstetrician) evaluated our conclusions concerning avoidable factors, and whether outcome was related to the model of care or not. They compared the original case records with our summary in the technical report.
All statistical tests were two-sided. Percentage differences were estimated by the normal approximation to the binomial function, with Yates’ correction for continuity. Student's t test with unpooled standard deviation was used for comparison of means. Some of the distributions were very skewed, and we therefore checked these t test results with a computer-intensive method for hypothesis testing, the Randomisation Test11,12. In no case were the results statistically significantly different from those of the Student's t test.
Results are usually presented as risk differences except when numbers are very small, such as the perinatal mortality rate which is presented as a ratio measure. Percentage differences were considered easier to understand by the less experienced reader; 95% confidence intervals are given for major outcomes. The study was approved by the Ethical Committee at South Hospital, Stockholm.
- Top of page
The baseline characteristics of age, education, civil status, ethnic background, parity, and length of gestation of women randomised to the birth centre group and the standard care group were very similar when entering the trial (Table 2). Primiparous women were over-represented in this study: 57% when the trial groups were combined, compared with the national figure of 41% for 199213. The average length of postpartum stay was 1.7 days (median 1.0) in the birth centre group and 2.8 (median 2.6) in the standard care group (P < 0.001). The outer centiles were the same in the two groups: less than one day (10th) and five days (90th), respectively.
Table 2. Women's background. BCG = Birth Centre group; SCO = Standard Care group.
| ||BCG n = 928||SCG n = 932|
|Age at confinement (years): mean (SD)||29.9(4.5)||29.9(4.3)|
|Education* (years): mean (SD)||14.2(2.6)||14.2(2.4)|
|Married or cohabitingf: n (%)||845(93.4)||831(94.6)|
|Native SwedesJ: n (%)||785(86.9)||767(87.4)|
|Parity: n (%)|| || |
|Gestation at randomisation|| || |
|(weeks): mean (SD)||20.1(8.1)||20.2(7.9)|
Withdrawals and transfers in the birth centre group are shown in Table 3. The transfer rate differed considerably between primiparous and multiparous women, especially during labour. Only 51% of primiparous women allotted birth centre care actually gave birth at the centre, compared with 80% of multiparous women. Reasons for transfer are listed in Table 4. Antenatal transfers were 14% of the 890 women who remained in the birth centre group after exclusion of early miscarriages and withdrawals. Intrapartum transfers were 23% of those 762 women who started labour at the birth centre, and postpartum transfers were 2.8% of the 586 women who gave birth at the centre. There were 16 miscarriages before the 23rd gestational week in the birth centre group and 14 in the standard care group. Gestation lasted 40.2 weeks (mean) in both groups, excluding miscarriages, and the proportions of preterm and post-term pregnancies did not differ statistically (< 37 weeks: 4.0% in birth centre group and 3.1% in standard care group, P= 0.36; > 42 weeks: 2.9% and 2.1% in the respective groups, P= 0.36).
Table 3. Birth Centre group. Values are given as percentages.
| ||Primiparas n = 544||Multiparas n = 384||P||All n = 928|
|Birth Centre births||51.1||80.2||< 0.001||63.1|
|Maternal postpartum transfer||2.2||1.3||0.45||1.8|
Table 4. Reasons for transfer of women in the Birth Centre group. Total number = 890, excluding withdrawals and miscarriages.
|Antenatal (14%)|| || |
|High blood pressure, toxemia||20||(2.2)|
|Post-term (> 42 weeks)||20||(2.2)|
|Preterm birth (< 37 weeks)||11||(1.2)|
|Signs of fetal distress||3||(0.3)|
|Intrapartum (20%)|| || |
|Failure to progress in labour*||88||(9.9)|
|Postpartum (2%)|| || |
Women in the birth centre group reported fewer health problems during pregnancy than women in standard care, but the difference was only statistically significant for symptoms difficult to diagnose in an objective way, such as low back pain, early uterine contractions, itching and headache. Medical problems necessitating hospital admission were similar in both groups. Forty-four women in the birth centre group spent 3.6 days (mean) in hospital during pregnancy, while the corresponding figure for 43 controls was 5.2 days (difference -1.6 days; P= 0.16).
Interventions during labour, such as obstetric analgesia, induction, augmentation of labour and electronic fetal monitoring, were less frequent in the birth centre group, but the rate of surgical procedures, such as caesarean section, vacuum extraction, episiotomy and manual removal of the placenta, did not differ statistically between the trial groups (Table 5).
Table 5. Medical interventions and outcome of labour. Values are given as n (%), unless stated otherwise. Key as for Table 2.
| ||BCG||SCO||Difference: % or mean||95% CI||P|
|Analgesia*†|| || || || || |
| Epidural block||108(12.1)||135(15.1)||−3.0||−6.2 to 0.1||0.07|
|Pethidine||33 (3.7)||120(13.4)||−9.7||−12.3 to -7.-2||< 0.001|
|Nitrous oxide||128(14.3)||417(46.6)||−32.3||−36.3 to -28.-3||< 0.001|
|Pudendal block||29(3.4)||47(5.6)||−2.2||−4.2 to -0.2||0.04|
|Local analgeisa postpartum||215(25.4)||295 (35.4)||−10.0||−14.4 to -5.6||<0.001|
|General anaesthetic||59(6.5)||56(6.1)||0.4||−1.9 to 2.6||0.83|
|Induction*||25 (2.7)||42 (4.6)||−1.9||−3.6 to -0.1||0.05|
|Augmentation of labour†|| || || || || |
|Araniotomy||209(23.3)||351(39.3)||−16.0||−20.1 to -11.7||< 0.001|
|Oxytocin 1st stage||140(15.6)||223 (24.9)||−9.3||−13.0 to -5.6||< 0.001|
|Oxytocin 2nd stage||160(17.9)||264 (29.5)||−11.6||−15.6 to -7.7||< 0.001|
|Monitoring of labour†|| || || || || |
|Electronic fetal monitoring||275 (30.7)||759 (84.9)||−54.2||−58.0 to -50.3||< 0.001|
|Internal monitor||165(18.4)||283(31.7)||−13.3||−17.2 to -9.3||< 0.001|
|Fetal scalp blood sampling||20 (2.2)||37(4.1)||−1.9||−3.5 to -0.3||0.03|
|Intrauterine pressure catheter||55(6.1)||51 (5.7)||0.4||−1.7 to 2.6||0.77|
|Surgical procedures*|| || || || || |
|Caesarean sections||65(7.1)||82 (8.9)||−1.8||−4.3 to 0.7||0.18|
|Elective||17(1.9)||22(2.4)||−0.5||−1.9 to 0.8||0.53|
|Emergency||48 (5.3)||60 (6.5)||−1.2||−3.5 to 0.9||0.29|
|Vacuum extraction||36(3.9)||40 (4.4)||−0-5||−2.3 to 1.4||0.74|
|Forceps|| ||1(0.1)||−1.0|| || |
|Episiotomy‡||66(7.8)||69 (8.3)||−0.5||−3.3 to 2.0||0.71|
|Manual removal of placenta‡||12(1.4)||9(1.1)||0.3||−0.6 to 1.3||0.65|
|Length of labour 100: mean (median)|| || || || || |
|Start of contractions to birth||15.0(12.1)||14.0(11.7)||1.0||−0.03 to 2.0||0.05|
|Postpartum hemorrhage‡|| || || || || |
|>600 mL||106(12.5)||106(12.7)||−0.2||−3.0 to 3.4||0.96|
|Blood transfusion||6(0.7)||5(0.6)||0.1||−0.7 to 0.9||0.98|
|Birthweight (g): mean (SD)||3563(532)||3531(529)||32.0||−17 to 81||0.20|
|Apgar score < 7 at 5 min||11(1.29)||10(1.1)||0.1||−0.9 to 1.1||0.99|
Labour was one hour (mean) longer in the birth centre group than in the standard care group. No difference was observed in postpartum haemorrhage. There was no maternal mortality and no statistical difference in maternal morbidity rates. One case of severe maternal morbidity needing intensive care treatment was observed in each group, one condition resembling water poisoning (Minor's cramp) with electrolyte imbalance in the birth centre group (see Case 2 below), and one case of severe toxaemia in the standard care group. Both women were discharged in good health.
Infant birthweight and Apgar scores did not differ statistically between the trial groups (Table 5), nor did the principal newborn diagnoses (Table 6). The larger proportion of treated jaundice in the birth centre group, although not statistically significant (P= 0.16), was probably a result of the lower limit for phototherapy treatment at the neonatal unit to which birth centre babies were admitted (300 μmol/L), than in other neonatal units in the city (350 umol/L).
Table 6. Principal newborn diagnoses. Key as for Table 2; IRDS = infant respiratory distress syndrome.
|Diagnoses||BCG n = 933||SCO n = 936|
|Miscarriage (22 weeks)||16||14|
|Small for gestational age (< 25D)||8||7|
|Premature (< 2500 g)||10||12|
|Very low birthweight (< 1500 g)||2||5|
|Post-term and/or heavy for dates||22||21|
|Intrauterine hypoxia and birth asphyxia||9||9|
|Other respiratory conditions||19||16|
|Suspected or confirmed septicaemia||9||6|
|Fetal and neonatal haemorrhage||1||1|
|Jaundice and/or hemolytic disease (phototherapy)||39*||27|
|Lost for follow up|| ||2|
No statistical difference was observed in the proportions of admissions to neonatal care within the first week after birth: 102 newborns (11.1%) in the birth centre group and 83 (9.0%) in the standard care group (difference 2.1%; 95% CI -0.7 to 4.9; P= 0.13). When excluding infants with physiological jaundice as their principal diagnosis, the corresponding figures were 76 (8.6%) and 71 (7.9%) (P= 0.58). Physiological jaundice in standard care was normally treated on the maternity ward.
Subgroup analysis showed that the pattern of neonatal transfer differed between primiparous and multiparous women. First-born babies were admitted to neonatal care more often in the birth centre group (15.6%) than in the standard care group (9.5%) (difference 6.1%; 95% CI 2.1 to 10.1; P= 0.003), while the converse was the case for newborns of multiparous women, 4.7%versus 8.4% (difference -3.7%; 95% CI -7.1 to 0.2; P= 0.04). When newborns with physiological jaundice were excluded, the pattern remained, but the difference was no longer statistically significant: 11.8% and 8.4% of the first-born infants (P= 0.07), and 4.3% and 7.2% of the infants of multiparous women (P= 0.08) were transferred in the birth centre and standard care groups, respectively.
The average length of neonatal stay for the infants transferred was 9.6 days (median 3.8) in the birth centre group, and 10.2 days (median 4.4) in the standard care group (difference: -0.6 days; 95% CI: -4.7 to 3.5; P= 0.78).
Eight infants had some form of serious morbidity not caused by malformations or preterm birth (described in detail in the technical report). Six of these were in the birth centre group and two in the standard care group (OR 3.03; 95% CI 0.54 < OR < 30.72; P= 0.28). In three cases in the birth centre group we identified possible avoidable factors related to the care at the birth centre as well as to standard care after transfer. The details are as follows:
Case 1. The woman was transferred from the birth centre to the delivery ward after 8 h of labour with only slow progress (cervix dilated from 3 to 7.5 cm). Electronic fetal monitoring on the delivery ward was deemed normal, and the woman received an epidural block. Three hours after transfer vacuum extraction was started because of signs of fetal distress according to the EFM. The baby was bom 29 min later, after the cup had slipped twice. Apgar scores were 2, 3 and 6 at 1 min, 5 min and 10 min, respectively. The baby suffered neonatal seizures and showed signs of brain damage at discharge. Early fetal scalp blood sampling and caesarean section might have improved the outcome.
Case 2. The woman was transferred from the birth centre to the delivery ward after 13 h of labour, because a deceleration of the fetal heart rate was detected by auscultation. Labour had progressed very slowly (cervix dilated from 3 to 8 cm). The woman had spent 3 h in a hot tub bath and had vomited several times. Oxytocin augmentation of labour started 1.5 h after transfer. Two fetal scalp blood samples, motivated by decelerations of short duration, were normal. Four hours after transfer, vacuum extraction was started because of maternal fatigue and weak contractions. Extraction was abandoned after four unsuccessful attempts with maximal oxytocin stimulation and manual fundal pressure. Fetal heart rate decreased to 70 to 80 beats per minute (bpm) during the procedure. Ten minutes later the woman was delivered by caesarean section. Apgar scores were 2, 4 and 6 at 1 min, 5 min and 10 min, respectively. Mother and baby both had convulsions, probably because of low plasma levels of sodium. In addition, the baby was exposed to asphyxia during the second stage of labour. The mother recovered after three days in intensive care, the baby after one month in neonatal care. The outcome might have been otherwise if the woman had been transferred earlier, oxytocin stimulation started earlier, and if the staff in both units had been more aware of the woman's electrolyte imbalance. Earlier caesarean section might also have improved the outcome.
Case 3. A primigravida with a normal pregnancy started labour at the birth centre in gestational week 42+2 days. The fetal head was not fixed hi the pelvic inlet on arrival, but labour progressed normally. Fetal heart rate was deemed normal by auscultation. The baby was born with Apgar scores 8, 9 and 10 at 1 min, 5 min and 10 min, respectively. Neonatal convulsions were observed within 24 h of the birth, and computerised tomography showed a small bleed around the falx and subdurally at the tentorium edge. The baby was discharged after three weeks without signs of neurological damage. The pelvic inlet diameter was determined by postpartum X-ray as 9.4 cm. Electronic fetal monitoring might have given information (variability) leading to a change in labour management, and one can question whether this woman met the medical criteria for starting labour at the birth centre. The guidelines require no abnormality in gestational week 43. A nonfixed fetal head at onset of labour in a nulliparous woman is a sign requiring observation.
Perinatal mortality, defined as intrauterine death after 22 weeks of gestation and infant death within seven days of birth, occurred in eight cases (0.9%) in the birth centre group and in two cases (0.2%) in the standard care group (OR 4.04, 95% CI 0.80 < OR < 39.17; P= 0.11) (Table 7). Possible avoidable factors were identified in two of the stillbirths in the birth centre group. Case 4. The woman called the birth centre in the 40th gestational because she had not felt any fetal movements during the previous 2 h. She had no contractions or other pain, but her abdomen was slightly tense. The midwife's advice was to call again if she continued to feel no movements. The woman called 8 h later and was then advised to come without delay. Immediately after admission to the birth centre fetal bradycardia and a tense uterus were diagnosed, and the woman was promptly transferred to the delivery ward. The baby died during the caesarean section due to placental separation. The interval between arrival at the hospital and the delivery was 24 min, including 5 min at the birth centre. The outcome might have been otherwise if the woman had been advised to come to the hospital at the first telephone contact.
Case 5. The infant died after the onset of labour in the 42nd gestational week. The woman had come to the birth centre in early labour, and after 3.5 h, when the cervix was 4 cm dilated, the midwife could hear no fetal heart beat at auscultation. Autopsy revealed no malformations, but there was suspicion of amniotic folliculitis and urinary tract infection. Electronic fetal monitoring on arrival might have altered the outcome. However, as this mother did not want the birth centre midwife to use the hand-held ultrasound monitor during labour, she might have rejected EFM if she had been allotted routine care.
Table 7. Perinatal deaths. Key as for Table 2; FDIU = Fetal death in utero before onset of labour; ICH = intracranial haemorrhage grade II; IUGR = intrauterine growth retardation.
|Trial group||Parity||Reason for transfer||Gestation(week+day)||Birthweight(g)||Cause of death|
|BCG||0||Ruptured membranes||24+4||740||Immaturity, chorioamnionitis, ICH|
|BCG||0||FDIU||37||2250||Unknown, suspected IUGR|
|BCG||1||Bradycardia, tense uterus||39+4||3315||Placental separation|
|BCG||0||Intrapartum death in birth centre||41+6||3950||Unknown; suspected amniotic in|
| || || || || ||folliculitis, and urinary tract infection|
|SCO||3|| ||37+0||2470||Asphyxia (second twin), FDIU|
Women admitted for labour at the birth centre
Among women in the birth centre group who started labour at the birth centre (n= 762), there was one perinatal death (1.3/1000), and 31 (4.1%) caesarean sections. Three (0.4%) of the babies born at the centre had low Apgar score (< 7 at 5 min), 5.2% were transferred to the neonatal ward, and the average length of neonatal stay for these infants was 4.5 days (median 2.4 days).
- Top of page
This trial was not large enough to conclude whether or not the birth centre was as safe as standard care, nor did it have sufficient power to detect a small reduction in the rate of operative deliveries. With the relatively low Swedish caesarean section rate, a dramatic reduction among women receiving birth centre care was less likely than it might have been in a country with a higher national rate. The number of women in the trial was determined more by pragmatic reasons than by power calculation. It would have been difficult to extend the trial period over the 45 months during which it was conducted.
Women allotted to birth centre care had fewer interventions during their pregnancy, a longer labour, and a shorter postpartum stay than the women allotted to standard care. However, the rates of operative delivery and maternal or infant outcome were not statistically different. The results are valid only for women who fulfil birth centre criteria for care: that is, women whose background characteristics resemble those described in one of our previous studies9. These women were older, better educated, and more interested in being actively involved in their own care and in the psychological aspects of childbirth than were women choosing standard care. The transfer rate of newborns to neonatal care was close to the national figure of around 10%13 in both birth centre and standard care groups, despite the selection of relatively low risk women for this study. This may be explained by the large proportion of primi-gravidas, and maybe also by the rather liberal inclusion criteria.
The higher rate of neonatal transfers among primi-parous women in the birth centre group can be partly attributed to the low limit for phototherapy for jaundice on the neonatal ward close to the birth centre, since jaundice was more common in first-born infants. The early discharge from the birth centre is another explanation. In the standard care group minor neonatal complications were dealt with on the postpartum wards. The lower rate of transfers among the multiparous women's infants in the birth centre group, compared with the standard care group, is more difficult to explain, but may be the result of a more restrained disposition to intervene.
There were few perinatal deaths or cases of serious morbidity, demonstrating that birth centre care was reasonably safe. However, the excess of perinatal deaths and infants with serious morbidity gives cause for concern. The combination of these two groups yielded 14 cases in the birth centre group compared with four in the standard care group, which is a statistically significant difference (OR 3.55; 95% CI 1.11 < OR < 14.88; P= 0.03). These findings must be interpreted with caution considering the large number of tests in this study, which increases the risk that some might have been significant by chance alone. Our findings demonstrate the need for further studies, a conclusion supported by the fact that another randomised controlled trial presented findings regarding mortality which were similar to ours8. A meta-analysis could probably further elucidate the issue of safety.
The possibility of avoidable factors was identified in two cases of perinatal mortality in the birth centre group (see case reports), one perhaps more obvious than the other, namely the inappropriate telephone advice to the woman with reduced fetal movements. In the other cases of intrauterine death (Table 7), neither the authors nor the independent external experts could identify any signs of substandard practice. The two cases of early intrauterine death were assessed as unavoidable, and the women had only been in birth centre care for five and two days respectively. One infant had the cord looped tightly twice around the neck and a knot tightly drawn. Another infant had the cord five times around the body and twice around an arm. The woman with suspected intrauterine growth retardation, and the woman who was post-term were checked with CTG and ultrasonography according to guidelines generally accepted in Swedish maternity care. The intrauterine death of a second twin in the standard care group was also assessed as unpreventable.
The analysis of possible avoidable factors in the cases of mortality and severe morbidity did not reveal any specific pattern. Possible mistakes were related to both birth centre care and to the standard delivery care to which the women were transferred. Our data did not warrant any particular changes in the birth centre's routines or guidelines. However, some aspects of birth centre care should receive special attention. The focus on normality and the strong commitment to a common philosophy of natural childbirth within the birth centre staff and their patients may increase the risk of overlooking signs of complications. On the other hand, a focus on pathology in standard care may lead to unnecessary intervention and overuse of obstetric technology, which may cause other problems.
Another aspect of birth centre care is the risk of delay that can arise when women in labour are transferred. The new staff team then need to make their own assessment. Midwives and doctors in standard care need to appreciate that a woman in labour transferred from a birth centre is a woman at risk and no longer a natural childbirth candidate. One way of resolving this situation would be to let the birth centre midwife accompany the woman and continue her shift on the hospital's delivery ward.
A comparison of the birth centre group data with findings from other birth centre studies shows similar perinatal death rates, 0.7%5 and 0.8%6–8, although the use of different definitions can render conclusions uncertain. The figures for intrapartum transfers were also similar, around 20% of the bookings, in all the studies of in-hospital birth centres5–8. The National Birth Centre Study in the United States1–4, describing the outcome of free-standing birth centres, reported only 10% transfers in labour, a figure closer to those observed in home birth studies14–16. This difference between free-standing birth centres and home births on the one hand, and in-hospital birth centres on the other may derive from differences in parity and inclusion criteria. In-hospital units seem to admit more primi-gravidas and permit more liberal inclusion criteria because of the proximity to the hospital delivery ward.
Two important objectives of maternity care are safety and parental satisfaction. This study demonstrated that birth centre care was as safe as standard care for multi-parous women and their infants. The primiparous women faced more problems, demonstrated by higher rates of intrapartum and neonatal transfers. All six cases of serious infant morbidity in the birth centre group (and the two in the standard care group) were first-born infants.
Previous reports from this trial described how women in the birth centre group were more satisfied with their care than those in the standard care group17; they also had a more positive birth experience18. The difference in childbirth experience was most pronounced among first-time mothers. In particular, they were more satisfied with their own achievement and felt more involved in the process of giving birth, aspects likely to boost their self esteem as new mothers. Consequently, the issue that needs further study is infant safety among primiparous women, the group who seems to benefit emotionally the most from birth centre care.
The proportion of home deliveries in the standard care group was higher (3.4%) than the home delivery rate in the Stockholm area as a whole (0.5%), which indicates that some women saw the birth centre as an alternative to home birth. For those who believe that in-hospital birth centres are safer than home births, this aspect should be taken into account in future evaluations of alternative birth options.
This evaluation was instituted from the start of a new programme, and the birth centre was the first of its kind in Sweden. It is conceivable therefore that the results might have been influenced by a general atmosphere of pioneering work, such as the enthusiastic staff and much attention, but also by a lack of experience in this form of care. A study of the midwives’ experiences of birth centre care found that they were more confident in assessing the progress of labour when interviewed after 2.5 years of birth centre practice than at the commencement of their new posts (they all had several years experience of standard care prior to commencing at the birth centre)19. At the 2.5 year interview they also expressed a greater respect for the medical risks of prolonged labour and said they transferred women with long labours earlier than they did in the beginning of the trial. A follow up of birth centre care as an established option for childbearing women would be worthwhile.
We would like to thank Professor M. Bygdeman and Associate Professor R. Thunell, for their analysis of case records and for their expert opinions.
This study was supported by grants from The Swedish National Delegation for Social Research (F88/42:2), the Swedish Medical Research Council (B89–27X-8701–01A), research grants from the Karolinska Institute and from Sodersjukhuset, Stockholm. A technical report, including a detailed description of the cases of perinatal mortality and severe morbidity, and the medical guidlines of the birth centre, can be ordered from the authors.