A comparison of intrapartum interventions and adverse outcomes by parity in planned freestanding midwifery unit and alongside midwifery unit births: secondary analysis of ‘low risk’ births in the birthplace in England cohort

The population for this study consisted of ‘low risk’ women in the Birthplace national prospective cohort study who planned birth in an AMU or an FMU. The Birthplace cohort study setting, participants, and study data have been described in detail elsewhere [2, 3]. Briefly, the Birthplace study collected data on 79,774 ‘low’ and ‘higher risk’ births between April 2008 and April 2010 from 142 NHS trusts, 53 FMUs, 43 AMUs and a stratified random sample of 36 OUs. Women with a singleton pregnancy were eligible for inclusion if they planned a vaginal birth and received some labour care from an NHS midwife during established labour in their planned birth settings. Women who presented in preterm labour (<37 weeks’ gestation), who were ‘unbooked’ (received no antenatal care) or experienced a stillbirth prior to the onset of labour were excluded.

Planned place of birth was defined as the woman’s intended place of birth at the start of care in labour. Women were classified as ‘low risk’ if, before the onset of labour, they were not known to have any of the medical or obstetric risk factors listed in the NICE intrapartum care guideline [4].

Maternal characteristics, medical or obstetric risk factors known prior to the onset of labour, ‘complicating conditions’ noted by the midwife at the start of care in labour (for example, prolonged rupture of membranes), intrapartum interventions and adverse outcomes were recorded on a study-specific data collection form by the midwife attending the birth. Maternal and neonatal outcomes were recorded on or after day five by the midwife attending the woman.

When data for the birth episode indicated that an adverse outcome had occurred or that the baby or mother had been admitted for higher level care, additional neonatal and maternal morbidity data were extracted from the maternal and neonatal records by Birthplace local coordinating midwives using follow-up morbidity forms.

For this analysis we considered the original Birthplace primary perinatal outcome (a composite measure designed to capture adverse perinatal outcomes that may be related to the quality of intrapartum care [2, 3]) and a range of outcome measures capturing maternal interventions and outcomes:

Except where indicated below, statistical methods for this study followed those in the primary Birthplace analyses [2, 3]. Logistic regression was used to calculate the odds ratios and confidence intervals for each outcome. As in previous analyses we adjusted for maternal age, ethnic group, understanding of English, marital or partner status, body mass index (BMI) in pregnancy, index of multiple deprivation (IMD) score, parity (where appropriate), and gestational age at birth. In this study we additionally adjusted for complicating conditions identified at the start of care in labour, which are associated with an increased risk of transfer [5]. Analyses were stratified by parity. The Wald test was used to test for an interaction between planned place of birth and parity. For each outcome, we calculated the number of events, the weighted incidence with confidence intervals (CIs), an unadjusted odds ratio (OR), an adjusted odds ratio (aOR) controlling for potential confounders except complicating conditions, and for the main analyses, a fully adjusted odds ratio additionally controlling for ‘complicating conditions’ as a single composite variable. Robust variance estimation and probability weights were used for reasons explained elsewhere [2, 3].

The main methodological differences between the Birthplace primary analyses [2, 3] and the analyses presented here were as follows: analyses were conducted using the AMU as the reference group and were stratified by parity; analyses were adjusted for both maternal characteristics (as before) and the presence of complicating conditions identified by the midwife at the start of care in labour; as in the primary analysis we used 95% confidence intervals for the ‘Birthplace primary perinatal outcome’ and, because Birthplace had multiple secondary outcomes, we used 99% confidence intervals for all secondary outcomes to reduce the chances of ‘false positive’ findings due to multiple comparisons. However, we also present p-values for each comparison in addition to confidence intervals.

Stata version 13.1 was used for all analyses (StataCorp LP, College Station, TX, USA).

Table 1 shows the characteristics of ‘low risk’ women who planned FMU or AMU births by planned place of birth and parity. Amongst nulliparous women, compared with women planning to give birth in an AMU, women planning birth in an FMU were more likely to be white, have a fluent understanding of English, and live in a more socioeconomically advantaged area. There was little difference in the distribution of nulliparous women’s age, marital/partner status, BMI, gestation or baby’s birthweight. Similar differences were observed amongst multiparous women (Table 1).

Nulliparous women were slightly more likely than multiparous women to have complicating conditions noted by the midwife at the start of care in labour, but in both groups (nulliparous and multiparous) the proportion of women with complicating conditions at the start of care in labour was similar in the two settings (Table 2).

The absolute incidence of the outcome events considered varied markedly depending on the outcome (Table 3). For example, amongst nulliparous women, the incidence of the ‘Birthplace primary perinatal outcome’ was around 0.5% (i.e. 5 events per 1000 births), while maternal outcome rates ranged from 0.2–1% for maternal admission for higher level care through to 11–16% for instrumental delivery. Around 20–30% of nulliparous women and 3–5% of multiparous women experienced a birth that was not straightforward as defined in this analysis.

Amongst nulliparous women, those who planned birth in an FMU had highly significantly reduced odds of instrumental delivery (10.8 vs 16.3%, aOR 0.63, 99% CI 0.46–0.86, p < 0.001) and highly significantly increased odds of having a ‘straightforward vaginal birth’ (78.8 vs 71.5%, aOR 1.47, 99% CI 1.17–1.85, p < 0.001) compared with those who planned birth in an AMU (Table 4). Although not significant at the 1% level, nulliparous women who planned birth in an FMU had reduced odds of being admitted for higher level care compared with women who planned birth in an AMU (0.2 vs 1.0%, aOR 0.28, 99% CI 0.07–1.10, p = 0.016). None of the other outcomes, that is the ‘Birthplace primary perinatal outcome’, intrapartum caesarean section, third or fourth degree perineal trauma and blood transfusion, differed significantly between the two settings (p = 0.907, 0.147, 0.129 and 0.063 respectively).

Use of epidural or spinal analgesia (18.9 vs 24.4%), augmentation with syntocinon (13.9 vs 18.0%), episiotomy (16.0 vs 22.1%) and active management of the third stage (79.8 vs 87.2%) were significantly less common in nulliparous women who planned birth in an FMU vs an AMU. Breastfeeding initiations (84 vs 83.7%) and use of immersion in water for pain relief (51.9 vs. 37.1%) did not differ significantly between the two settings at the 1% level of significance (Additional file 1: Table S3).

Amongst multiparous women, those who planned birth in an FMU had highly significantly reduced odds of instrumental delivery (1.1 vs 2.5%, aOR 0.41, 99% CI 0.25–0.68, p < 0.001) and of third or fourth degree perineal trauma (0.9 vs 1.6%, aOR 0.60, 99% CI 0.36–1.00, p = 0.010) and highly significantly increased odds of having a ‘straightforward vaginal birth’ (97 vs 94.6%, aOR 1.86, 99% CI 1.35–2.57, p < 0.001) compared with those who planned birth in an AMU (Table 4). Although not significant at the 1% level, multiparous women who planned birth in an FMU had reduced odds of being admitted for higher level care compared with women who planned birth in an AMU (0.1 vs 0.4%, aOR 0.30, 99%CI 0.07–1.20, p = 0.025). None of the other outcomes, that is the ‘Birthplace primary perinatal outcome’, intrapartum caesarean section and blood transfusion, differed significantly between the two settings (p = 0.745, 0.224 and 0.052 respectively).

In multiparous women most other interventions were significantly less common in multiparous women who planned birth in an FMU vs an AMU (Additional file 1: Table S3): epidural or spinal analgesia (3.5 vs. 5.9%), augmentation with syntocinon (1.4 vs. 2.4%), episiotomy (2.3 vs 3.7%) and active management of the third stage (76.2 vs. 84.6%). Immersion in water for pain relief was used significantly more often by multiparous women who planned FMU birth (40.6 vs 23.2%), Breastfeeding initiation did not differ between the settings (78.2 vs. 78.6%).

Detailed results are tabulated in full in the Additional file 1.

When stratified by parity we did not find a significant difference between the two birth settings (FMU and AMU) in the odds of caesarean section, third or fourth degree perineal trauma (significant only for multiparous women) or blood transfusion. However, odds ratios were in the same direction and of broadly similar magnitude in nulliparous and multiparous women, and there was no strong evidence of heterogeneity (Wald test: intrapartum caesarean section p = 0.558; third or fourth degree perineal trauma p = 0.184; blood transfusion p = 0.506). We therefore conducted a post hoc combined analysis for these three outcomes, including all women (nulliparous and multiparous) and additionally adjusting for parity.

There was no difference in adverse perinatal outcomes, as measured by the ‘Birthplace primary perinatal outcome’, between planned AMU and FMU births. The odds of an instrumental delivery were reduced in planned FMU births and the odds of having a ‘straightforward vaginal birth’ were increased in planned FMU births compared with planned AMU births. The odds of intrapartum caesarean section did not differ significantly between the two settings. The overall pattern of the findings suggested a trend towards lower intervention rates and fewer adverse maternal outcomes in planned FMU births compared with planned AMU births.

Strengths and limitation of the Birthplace cohort study are discussed more fully elsewhere [3]. In brief, strengths include the ability to compare outcomes by planned place of birth at the start of care in labour, the large sample size, the minimisation of bias through achievement of a high response rate and the absence of self-selection bias arising from non-consent, and the ability to control for a range of potential confounders. In this analysis we have additionally controlled for complicating conditions identified at the start of care in labour, such as prolonged rupture of membranes, meconium stained liquor and proteinuria.

The study has a number of limitations. First, as in all Birthplace analyses, it is possible that the use of a composite perinatal outcome measure encompassing events of varying severity may have masked important differences between settings in more serious outcomes such as stillbirth, neonatal death and neonatal encephalopathy. Second, although we were able to control for a number of potential confounders, because of the non-randomised nature of the study it remains possible that women in the two study groups may have differed in ways that we did not measure and which may be associated with differences in outcome. For example, women opting for birth in an FMU may have a different attitude towards interventions and ‘natural birth’ than women who opt for birth in a hospital with medical facilities available on site, which may in turn influence their chances of receiving some of the interventions that we studied. Related to this, because the AMU and FMU groups were ‘self-selected’ (i.e. in most instances women will have ‘chosen’ an AMU or FMU) and this was a relatively uncommon choice at the time of the study, we cannot be certain that the findings are generalisable to other groups of women who may differ from those in the study sample. Finally, the findings relate to services available during the Birthplace data collection period (2008–2010) at which time there were fewer midwifery units than today and most AMUs were relatively small. Since 2010 the number of FMUs has remained relatively static but the number of AMUs has increased and the characteristics of these units (size, staffing, and admission criteria) may well have changed [6]. The generalisability of these findings to current models of service provision, clinical practice and to current users of midwifery-led services is unknown. These issues, and the need to undertake monitoring and evaluation of current services, are discussed more fully elsewhere [5].

The findings of this study are broadly consistent with the unadjusted analyses conducted as part of the NICE evidence review for the 2014 intrapartum care guideline [1] and support their conclusions that perinatal outcomes are similar in the two settings and women who plan birth in an FMU are more likely to experience a spontaneous vaginal birth than women who plan birth in an AMU. Our analyses also confirmed a reduction in serious perineal trauma in women who planned birth in an FMU compared with women who planned birth in an AMU but we cannot rule out the possibility that differences could be partly attributable to different levels of ascertainment in births planned in the two settings.

We did not observe a statistically significant reduction in intrapartum caesarean section for either nulliparous or multiparous women, or overall, so our analysis does not confirm the statistically significant reduction (unadjusted RR 0.82, 95%CI 0.73–0.93) found by NICE in their analysis of aggregated data [1]. However, although we did not find a statistically significant reduction in the odds of caesarean section in planned FMU births (p = 0.093), the observed direction of effect (odds ratio 0.82) was not inconsistent with a possible reduction in caesarean section rates in planned FMU births.

Our results show a statistically significant reduction in instrumental delivery in births planned in an FMU compared with an AMU (10.8 vs 16.3% in nulliparous women, and 1.1 vs 2.5% in multiparous women). A number of factors might explain this, including possible differences in labour management, easier access to epidurals in births planned in an AMU (which increase the risk of instrumental delivery [7]) and possibly a higher threshold for transfer for failure to progress in the second stage of labour in births planned in an FMU (since ambulance transfer is required). However, it is also possible that women who opt for birth in an AMU differ in their attitudes towards medical interventions or in other attributes that may influence outcomes and differences in provider factors such as staff seniority and experience, and organisational culture may also play a part.

We did not find that adverse perinatal outcomes differed significantly between the two settings and for both nulliparous and multiparous women the odds of the ‘Birthplace primary perinatal outcome’ were close to one. As noted above, we cannot rule out a difference in serious adverse perinatal outcomes.