Cervical ripening at home or in hospital during induction of labour: the CHOICE prospective cohort study, process evaluation and economic analysis

Preplanned primary comparison

The planned primary comparison was home dinoprostone (a prostaglandin) cervical ripening (exposure) versus in-hospital dinoprostone cervical ripening (comparison) from 39 weeks of gestation. A secondary exploratory comparison was planned to be undertaken to explore home cervical ripening with balloon catheter (exposure) versus home cervical ripening with dinoprostone (comparator).

Additional supplementary research questions to be applied to the cohort study, should numbers suffice, were:

1. How do IOL rates, methods and outcomes vary across the UK?

2. What are the outcomes of IOL in different subgroups of women (e.g. women with multiple pregnancy, women with IOL after a previous caesarean section) and at each week of gestation (37, 38, 39, 40 and 41+ weeks)?

3. Can we predict which women will have caesarean section after IOL?

However, these supplementary questions are not addressed in this report.

Revised primary comparison

Following preplanned pilot analysis at 156 site months of data collection, the primary comparison was changed to ensure feasibility and to reflect current practice. The revised primary comparison became ‘Home balloon cervical ripening (exposure) versus in-hospital prostaglandin cervical ripening (comparator) from 37 weeks gestation’.

Throughout the report, these main two comparator groups are referred to as ‘home balloon’ and ‘hospital prostaglandin’, respectively.

Study design as planned at study outset

We planned to perform a prospective multicentre observational cohort study with an internal pilot phase. 19 We set out to obtain data from electronic health records from at least 14 maternity units offering only in-hospital cervical ripening and 12 offering dinoprostone home cervical ripening. The expected sample size was 8533 women with singleton pregnancies undergoing IOL at 39 + 0 weeks’ gestation or more. To achieve this and to contextualise our findings, we planned to collect data relating to a cohort of approximately 41,000 women undergoing IOL after 37 weeks. We planned to use mixed-effects logistic regression for the non-inferiority comparison of NNU admission and propensity score matched adjustment to control for treatment indication bias.

A secondary exploratory comparison was planned if sufficient cases allowed: home cervical ripening with balloon catheter (exposure) versus home cervical ripening with dinoprostone (comparator), to explore whether there are any indications of different safety profiles of these two methods of home cervical ripening.

Changes to study design

Results of the planned interim analysis after 156 site months of data led to a change in study design and expected sample size. Three key changes were made:

1. Data would be obtained from a total of 26 NHS maternity units, of which 18 offered only in-hospital cervical ripening (predominantly with prostaglandin) and 8 offered home cervical ripening using balloon catheters, meaning that around 25,000 women would be in the initial data extract from which the analysis sample would be drawn.

2. Instead of including a population of women from 39 weeks’ gestation onwards, women were included if they underwent IOL at 37 completed weeks of gestation onwards.

3. Instead of comparing home cervical ripening using dinoprostone with in-hospital cervical ripening using dinoprostone, home cervical ripening with balloon would be compared with in-hospital cervical ripening using any prostaglandin.

4. Planned analysis would use multivariable logistic regression without propensity score matching for the primary outcome. All other outcomes would be reported using descriptive statistics.

Participants

Data on all women having IOL at 37 + 0 weeks’ gestation or more were extracted. Women who opted out of data provision were excluded from the data set by the BadgerNet (System C, Stratford-upon-Avon, UK) maternity system analysts before data were extracted, so the number of such women is not known to the research team.

More stringent inclusion and exclusion criteria were applied at the analysis stage. These are summarised in the flowchart in Figure 1. In the primary analysis, a cohort of women was created with broadly comparable level of risk in pregnancy (i.e. those without key risk factors for adverse maternal or perinatal outcomes defined below) in whom there was no contraindication to home cervical ripening, who had singleton pregnancies with IOL at 37 weeks’ gestation or more. This group included women having IOL for post dates, because of maternal or clinician preference, for maternal age, for discomfort or social indications, for hypertension, diabetes, fetal concerns (oligohydramnios, reduced liquor volume, macrosomia, intrauterine growth restriction, static growth, polyhydramnios), fetal movement concerns, obstetric cholestasis, antepartum haemorrhage, previous obstetric history, prelabour rupture of membranes (ROM) documented as primary or other indication for IOL (prolonged ROM, spontaneous ROM and suspected spontaneous ROM) and other maternal reasons, such as raised body mass index (BMI), in vitro fertilisation pregnancy, musculoskeletal pain, thrombophilia. Exclusion criteria consisted of previous caesarean section, antepartum stillbirth (before cervical ripening initiated), congenital fetal anomaly, abnormal cardiotocograph (CTG)/Doppler, breech presentation.

FIGURE 1.

Identification of CHOICE observational cohort study populations and analysis cohorts by setting and method of cervical ripening. a, Some participants may have more than one reason for exclusion. CS, caesarean section.

Participants were identified from data recorded in specified fields in BadgerNet electronic maternity records. We used data fields indicating IOL, estimated due date and date of IOL to identify women having IOL at 37 weeks of gestation or more.

Women were made aware of the CHOICE study through posters in participating sites, business cards, information leaflets, online adverts on hospital/maternity websites and relevant social media sites, and information in maternal electronic maternity records.

Women were able to opt out of data provision by notifying their clinician or midwife or e-mailing the study research midwife at the local site, following which their opt-out status was recorded on their electronic record. This became a conditional field, where there must be no opt-out selected for the data to be extracted for the study. There was no restriction on co-enrolment in other studies.

Study settings

The study was performed in 26 UK obstetric units, 8 of which offered cervical ripening both in-hospital (mix of methods) and at home (predominantly balloon method) and 18 of which offered exclusively in-hospital cervical ripening (predominantly prostaglandin method). All included sites used the BadgerNet maternity record system.

The maternity units included were selected to represent the diverse range of maternity service settings in the UK, and included urban tertiary referral units, mid-sized urban district general hospitals and small, more isolated, rural units.

Participating trusts/boards/units included:

• NHS Borders

• NHS Fife

• NHS Grampian

• NHS Highland

• NHS Lanarkshire

• NHS Tayside

• Ashford and St Peter’s Hospitals NHS Foundation Trust (Maternity)

• Birmingham Women’s and Children’s NHS Foundation Trust

• Darlington Memorial Hospital

• University Hospital of North Durham (Maternity)

• Dorset County Hospital

• Epsom Hospital (Maternity)

• St Helier Hospital (Maternity)

• Hereford County Hospital (Maternity)

• Princess Royal University Hospital

• King’s College Hospital (Maternity)

• Chorley and South Ribble Hospital

• Royal Preston Hospital

• Cumberland Infirmary (Maternity)

• West Cumberland Hospital (Maternity)

• Queen Elizabeth Hospital, Gateshead (Maternity)

• Warwick Hospital

• Queen Elizabeth Hospital Kings Lynn (Maternity)

• New Cross Hospital, Wolverhampton (Maternity)

• Walsall Manor Hospital

• Worcestershire Royal Maternity Hospital

No data were collected from each site for the first 14 days after their ‘go live’ date to ensure that women had the opportunity to see and read study materials such as posters and study cards, and to opt-out of data collection if preferred.

Clinical and sociodemographic characteristics

The following demographic and clinical characteristics were described across cohorts: number of women undergoing IOL by study site; age of women in years; BMI of women; class III obesity at booking (≥ BMI 40 kg/m²); marker of deprivation [index of multiple deprivation (IMD)]; parity (number of parity 0, 1, 2, 3 +); gestation at IOL (completed weeks) plus number at each completed week; birth weight (grams) and indication for IOL.

Indications for IOL (as defined by study sites) were categorised as follows: post dates; hypertension (pregnancy-induced hypertension or pre-eclampsia), antepartum haemorrhage; diabetes; obstetric cholestasis; obstetric history; maternal age; maternal request (including social indications); other maternal reasons (including raised BMI, in vitro fertilisation; musculoskeletal pain, thrombophilia); fetal concern (excessive growth, polyhydramnios, reduced liquor volume, small for gestational age, static growth, suboptimal growth, abnormal dopplers); previous precipitate labour; ROM (preterm, prolonged, suspected, confirmed); reduced fetal movements (isolated or recurrent episodes).

Exposures

The exposure group consisted of women who, at the start of the cervical ripening process, planned to have home cervical ripening and who underwent this procedure using a balloon catheter as the first method of IOL. This group is referred to as the ‘home balloon’ group in the report.

The comparator group included women who planned to have in-hospital cervical ripening from maternity units not offering home cervical ripening, and who underwent this using prostaglandin methods. This design minimised potential bias arising from the fact that, in maternity units which offer both home and in-hospital cervical ripening, the risk of complications in the babies of women having home cervical ripening (lower-risk pregnancies) is inherently different from that of babies of women having in-hospital cervical ripening (higher-risk pregnancies).

A further two cohorts were identified (in-hospital cervical ripening using prostaglandin where the hospital also offers home cervical ripening, and at home cervical ripening using dinoprostone – the original exposure group), to provide context to the primary outcome, although the latter group was not used in any of the secondary outcome analyses. Figure 1 outlines the derivation of each cohort.

Outcomes

Changes to outcomes

The primary outcome remained the same, although following the pilot analysis, it was agreed that it was unlikely that there would be sufficient power to achieve our primary objective of assessing safety of home versus in-hospital cervical ripening. To compare outcomes across groups, a simplified analysis with two non-matched cohorts replaced the original analysis plan, which had involved a propensity-score matching process.

Sample size

The original sample size required 1920 home IOL with dinoprostone to be matched with 1920 in-hospital IOL with prostaglandin from hospitals not offering home IOL. However, at the time of the pilot study, only 150 women had been given prostaglandin for IOL at home, with even fewer having received dinoprostone. Our stop–go criteria at this point required us to have achieved at least 400. With the changes in practice necessitated by the COVID-19 pandemic, 1920 was not going to be achievable in a practicable time frame. The requirement to match was dropped to increase the power from the in-hospital cohort.

Overall, the lower numbers of women in the home setting than expected meant that the study was deemed to be underpowered to assess the primary outcome in any other way than exploratory. The primary end point was presented with a 95% confidence interval (CI), but should only be considered as hypothesis generating rather than hypothesis testing.

Data sources

Data were collected directly from electronic maternity and neonatal records for participants who had babies cared for by neonatal staff. These data are recorded by clinical staff (midwives, doctors and neonatal nurses) while providing antenatal, intrapartum and postpartum care. Existing data fields were used. Data were assumed to reflect the understanding of the clinical staff who entered the data. Diagnoses were assumed to reflect national guidance.

Unless women opted out of secondary data use, deidentified data were transferred from BadgerNet participating sites to a secure University of Edinburgh server for analysis.

No personal data were collected. Potentially identifiable data, such as the date and time of birth, date of events such as commencing cervical ripening and hospital discharge, were converted into gestation at birth (weeks + days); and antenatal and postnatal events into ‘t – x’ and ‘t + x’ minutes, respectively.

Data from a total of 40 tables in the BadgerNet system were extracted. These included details of the following:

• neonatal care admission

• critical incident affecting baby

• birth details of baby

• discharge details of baby

• baby examination details

• baby feeding details

• operative birth details

• internal transfer of baby

• medications administered to baby

• microbiological tests performed in relation to baby

• baby patient index (demographic and clinical details)

• risk assessment relating to baby

• transfer of baby’s care

• maternal admission

• maternal analgesia

• maternal antenatal assessment

• maternal care plan

• maternal communication details

• maternal critical incident

• maternal discharge details

• maternal perineal tears or trauma

• maternal health history

• maternal induction of labour details

• maternal induction of labour booking details

• maternal internal transfer

• maternal labour and birth details

• maternal lifestyle

• maternal death

• maternal observations

• maternal partogram

• maternal patient index

• maternal pre-operative checklist

• maternal previous pregnancy

• maternal risk assessment

• maternal ROM

• maternal sepsis pathway

• maternal transfer of care

• maternal transfusion

• maternal vaginal examination

• summary of neonatal care.

Algorithm to determine setting of cervical ripening

Women in the study data set were categorised as having either had home cervical ripening, in-hospital cervical ripening or unclear setting of cervical ripening. The categorisation took place using the following variables: mother_induction (first induction), mother_admission (admissions after IOL), mother_patientindex, mother_discharge (excluding discharge before IOL), mother_communication (phone call between ‘patient’ and other), mother_previouspregnancy and baby_deliverydetails.

The main analysis population was identified following the exclusion of cases listed below:

• Remove if fetal death is reason for IOL.

• Remove if IOL time < 14 days before birth (likely coding error).

• Remove if IOL time is after birth (likely coding error).

• Remove multiple births (population 1 in flowchart).

• Remove if reason for IOL is:

  • congenital fetal abnormality

  • abnormal CTG

  • breech.

• Remove if any previous pregnancy involved caesarean section.

• Remove if not confirmed IOL.

• Remove if gestational age at IOL < 37 weeks (population 2 in flowchart).

The algorithm used to determine induction of labour setting (‘IOL type’) was as follows (where 'y' = yes and 'n' = no):

• If the hospital does not offer home cervical ripening, then IOLtype = ‘In hospital only’.

• If the hospital is a mixed hospital (allows both at home and in-hospital cervical ripening) AND dischargedhome = ‘y’, then IOLtype = ‘home’.

• If the hospital is a mixed hospital (allows both at home and in-hospital cervical ripening) AND dischargedhome = ‘n’, then IOLtype = ‘inhospital mixed’.

• For those still undefined at this point:

  • If baby born outside hospital, then IOLtype = ‘home’.

  • If the hospital allows home IOL AND there is no admission time, then IOLtype = ‘inhospital mixed’ (data sets only include admissions after IOL, so mother must already be in hospital if there is no admission time after IOL has commenced).

  • If place of birth in hospital that allows home IOL AND there is no discharge note AND first admission > 120 hours, then IOLtype = ‘uncertain’.

  • If place of birth in hospital that allows home IOL AND there is no discharge note AND first or second admission is 6–120 hours after birth, then IOLtype = ‘home’.

  • If place of birth in hospital that allows home IOL AND there is no discharge note AND only one admission AND admission ≤ 6 hours after birth, then IOLtype = ‘uncertain’.

  • If place of birth in hospital that allows home IOL AND there is no discharge note and time from IOL to discharge > 24 hours, then IOLtype = ‘inhospital mixed’.

  • If place of birth in hospital that allows home IOL AND IOL time < discharge time AND discharge ≤ 24 hours after birth, then IOLtype = ‘inhospital mixed’.

  • If place of birth in hospital that allows home IOL AND IOL time < discharge time AND discharge ≤ 24 hours before birth, then IOLtype = ‘home’.

  • If place of birth in hospital that allows home IOL AND first admission < 6 hours after birth, then IOLtype = ‘home’.

  • If place of birth in hospital that allows home IOL AND first discharge is after birth, then IOLtype = ‘inhospital mixed’.

  • If there is a record of a phone call between patient and other during period between IOL and birth, then IOLtype = ‘home’.

Definition of cohorts

• If IOLtype = ‘home’ and induction method = ‘cervical balloon’, then cohort = 1.

• If IOLtype = ‘In hospital only’ and induction method = ‘prostaglandins’, then cohort = 2.

• If IOLtype = ‘inhospital mixed’ and induction method = ‘prostaglandins’, then cohort = 3.

• If IOLtype = ‘home’ and prostaglandin type = ‘propess’, then cohort = 4.

Interim analyses and stopping guidelines

As proposed, we carried out a pilot phase analysis to determine the parameters of the primary outcome and feasibility of obtaining the planned sample size for the original preplanned analysis at 156 site months, using criteria as shown in Table 1. This was based on an evaluable comparison group of 1920 women in each arm, so acted as an inherent check on home cervical ripening eligibility and uptake rates. We assessed variation of the primary outcome at the pilot stage, along with that of other measures of neonatal morbidity included as secondary outcomes (e.g. any NNU admission). We retained the preplanned parameters of NNU admission used in the primary outcome after analysis of pilot data. We redefined the comparison groups following assessment of the pilot analysis data, which demonstrated much lower uptake of home cervical ripening using dinoprostone than expected, higher uptake of home cervical ripening using balloon, a much wider range of indications of IOL in home cervical ripening groups and earlier gestation at IOL in both settings. The decision to redefine the comparison groups was made in consultation between the expert project management group, the study steering committee and the funder.

Statistical methods

All analyses were fully specified in an updated comprehensive statistical analysis plan (version 2.0) and agreed by the steering committee. Analyses were carried out in accordance with relevant guidance, including Strengthening the Reporting of Observational Studies in Epidemiology. 15

The original design, which informed the sample size calculation, was a non-inferiority design, chosen with a non-inferiority margin of 4% (deemed as likely to be an important difference on consultation with women and clinicians) for the primary outcome of NNU admission within 48 hours of birth for 48 hours or more.

The non-inferiority margin was established based on a combination of what is acceptable to women and partners and what costs are acceptable to healthcare providers. However, following the pilot phase and the expected inability to reach the required sample size, the principal analysis as defined originally, was deemed unlikely to be sufficiently powered, and the primary analysis was changed to an estimation problem, rather than a hypothesis testing problem.

For the principal analysis of the primary outcome, we used mixed-effects logistic regression for the comparison of NNU admission within 48 hours of birth for 48 hours or longer (yes/no). Within the mixed model, site has been included as a random effect and all other factors are fixed effects. A random intercept for each site was included, rather than a common intercept to allow for baseline differences, with the assumption of no correlation between sites.

Potential confounding variables were identified from the pre-planned list including:

• Gestational age at commencement of IOL (in weeks).

• Maternal age at delivery (in years).

• BMI at booking (or earliest record) in kg/m².

• Maternal medical conditions such as diabetes (pre-existing or gestational).

• Hypertensive disorder (proteinuria, hypertension, pre-eclamptic toxaemia, pregnancy-induced hypertension, pre-eclampsia).

• Other maternal risk factors (obstetric cholestasis, thrombophilia, in vitro fertilisation, past obstetric history).

• Fetal concerns (fetal growth or liquor concerns, reduced fetal movement, other fetal reason).

• ROM.

• Smoking.

• Deprivation.

• Parity.

Descriptive statistics were used to describe all secondary outcomes. We described the duration of hospital stay during IOL, time spent at home, total hospital stay and time to birth using medians and interquartile ranges, while categorical outcomes such as mode of birth were described using proportions and percentages.

As outlined in the exposures section, data were reported for cohort 1 (women who had balloon cervical ripening at home), cohort 2 (women who had in-hospital cervical ripening using prostaglandin) and cohort 3 (women who had in-hospital cervical ripening using prostaglandin in a hospital that also offers home cervical ripening). Data from the larger cohort of women having IOL from 37 + 0 weeks’ gestation (population 2) and those relating specifically to in-hospital cervical ripening in units also offering home cervical ripening (cohort 3) were used to contextualise the findings on the background of unit practices and populations undergoing IOL. This was due to the considerable inter-unit variation in both the rates of IOL and the risk profile of women giving birth that needed to be considered. This helped to demonstrate the generalisability of the findings.

Recruitment context

Unexpectedly, the CHOICE study took place in an NHS under pressure to manage high IOL rates (range 31–49% of all births in study sites) during the COVID-19 pandemic, with long delays during the IOL process. Home cervical ripening using balloon was performed in women for a wide array of indications (low to moderate-risk groups) and from 37 weeks’ gestation onwards.

Participant flow

Participants were identified from the data transferred from BadgerNet as illustrated in Figure 1. Data for a total of 25,260 women were extracted.

Losses and exclusions

As outlined in the flowchart, seven women were excluded due to fetal death occurring before IOL started. A total of 212 women with multiple pregnancies were excluded. Additional exclusions (7284 in total) were those women where home cervical ripening would be contraindicated (previous caesarean birth, congenital fetal anomaly, abnormal CTG, breech fetal presentation, gestational age < 37 completed weeks) and those where there was no documented evidence that any cervical ripening had taken place. Further exclusions involved women where there was assumed to be a higher level of clinical risk than those who had cervical ripening at home since these women had in hospital cervical ripening in hospitals that also offered home cervical ripening (n = 7397, cohort 3). The final exclusions from the primary analysis were those women who had cervical ripening using dinoprostone at home (n = 141, cohort 4) given that these were no longer included in the primary comparison.

Data on study population

Following initial exclusions, the total number of women who were coded as having had an IOL at each site is shown in Table 3 (population 1 as per Figure 1). Following exclusion of those in whom the method of IOL could not be established, the numbers of women with confirmed method of IOL at each site are shown in Table 3 (population 2). The IOL rate at each site (calculated as a proportion of all births at each site during the study period) is shown in Table 3 where denominator data were provided by sites.

Following further exclusions (see Figure 1), the final comparison groups for the main analysis comprised 515 women who underwent home cervical ripening using balloon methods (cohort 1), 4332 women who underwent in-hospital cervical ripening using prostaglandin methods (cohort 2) and 7397 women in the additional comparison group (providing background context) who underwent in-hospital cervical ripening with prostaglandin at a site that also offered home cervical ripening (cohort 3) as shown in Table 4.

Not all units offered cervical ripening using balloon method. The number of women undergoing balloon cervical ripening at any point during IOL by site is reported in Table 5.

Cervical ripening using cervical dilator was performed at two sites as shown in Table 6.

All sites used prostaglandin as a method of cervical ripening in some cases, as is shown in Table 6.

Of those in the exposed group (‘home balloon’) around 1 in 20 (24/515, 4.7%) received prostaglandin for cervical ripening after their balloon method was used.

The total number of women undergoing confirmed IOL (population 2) by parity and gestation are provided in Report Supplementary Material 1, Tables 1 and 2, respectively. The number of women undergoing IOL by setting (whether home or in-hospital, regardless of method) is provided in Report Supplementary Material 1, Table 3. Further breakdown of first method used in IOL (including but not exclusive to cervical ripening) by site is provided in Report Supplementary Material 1, Table 4. Median number of IOL attempts made by site, parity and gestation are provided in Report Supplementary Material 1, Tables 5–7. Median number of different methods used in IOL are provided in Report Supplementary Material 1, Tables 8–10. Number of women having balloon method for cervical ripening at any point by site, parity and gestation are provided in Report Supplementary Material 1, Table 11–13. Number of women having prostaglandin method for cervical ripening at any point by site, parity and gestation are provided in Report Supplementary Material 1, Tables 14–16. Number of women having non-ripening IOL methods at any point by site, parity and gestation are provided in Report Supplementary Material 1, Tables 17–19. Number of women having any cervical ripening method at any point by site, parity and gestation are provided in Report Supplementary Material 1, Tables 20–22. Subsequent method used in IOL after balloon or prostaglandin by site, parity, gestation and setting are provided in Report Supplementary Material 1, Tables 23–40. The proportion of women using more than one method of cervical ripening by site, parity and gestation are provided in Report Supplementary Material 1, Tables 41–43.

In all groups, the median cervical dilatation at onset of IOL was 1 cm, with an interquartile range of 0–1. A full breakdown of cervical dilatation at onset of IOL is provided in Report Supplementary Material 1, Tables 44–46.

Study design

The process evaluation comprised a questionnaire-based survey and case studies nested within the CHOICE observational cohort study. Case studies involved semistructured interviews with women and birth partners, and interviews and focus group discussions with professionals and key stakeholders.

An additional survey of all UK maternity units was undertaken to determine impact of the COVID-19 pandemic on IOL policy and practice.

Participants

1. Postnatal questionnaire-based survey.

Anyone who had undergone an IOL at 37 weeks or more at the 26 CHOICE study sites was initially eligible to participate. Those who suffered pregnancy or neonatal loss were excluded.

1. Case studies.

Participants in the case study interviews and focus groups were all drawn from the five case study sites and included:

1. A purposive sample of healthcare professionals and stakeholders, including: clinical directors, heads of midwifery, maternity commissioners, representatives from Maternity Voices Partnerships, obstetricians, midwives and other healthcare professionals involved in IOL policy development and/or clinical practice.

2. A purposive sample of women who had completed the questionnaire-based survey, given birth in one of the case study sites and had given consent for further contact on the survey.

3. A purposive sample of birth partners of the women who took part in the interviews. Birth partner was defined by the woman and included co-parents, friends and other relatives who had supported them during their IOL.

Study settings

1. Online postnatal questionnaire-based survey available initially in all CHOICE study sites.

2. Case study sites.

Five NHS trusts and health boards in England and Scotland were selected for variation in service context and configuration, setting and method of cervical ripening.

1. COVID impact survey.

All NHS trusts and boards in the UK that offer maternity services as identified through the NHS service directories for the four UK countries were invited to take part (n = 157).

Outcomes

The planned outcomes were to describe women, birth partner and stakeholder experiences of IOL and cervical ripening. Specifically, the intention was to identify barriers and enablers to offering home cervical ripening, and aspects of positive and negative experiences in relation to IOL and cervical ripening. Use of questions based on the Labour Agentry Scale were planned as a primary outcome when comparing home and hospital cervical ripening. 23

Sample size

Data collection

The following methods were used to gather data.

Analysis

Data collected

Losses and exclusions

In the postnatal questionnaire-based survey, a total of 320 responses were received. Nine responses were excluded as respondents had not had an IOL and a further two because their IOL happened prior to the CHOICE study commencing. A total of 309 responses were included in analysis.

Findings

Findings by qCHOICE study objectives

1. In what ways does the service context influence cervical ripening approaches in hospital or out of hospital settings?

Women described variation in policy and practice across all sites. It was common to hear of care being planned around service capacity rather than in line with guidance/policy. At times, this proved detrimental.

Had balloon induction 8am Monday. Balloon out 8am Tuesday and was 2–3 cm. However, was sent home as there were not enough midwives to induce me further … was taken back in on Thursday 4pm. 7am Friday taken to the delivery room [for ARM] … at that point was then back to 1 cm.

Survey participant 89753808

There were significant delays to the IOL process. Of those who remained in hospital and had cervical ripening (n = 230), the range was 0–260 hours, with a mean stay prior to transfer to labour suite of 31.5 hours.

Staff were pushed to the brink, which is why I was in hospital for 11 days before my waters were broken.

Survey participant 89021852

Some women reported unsafe physically and emotionally unsafe situations caused by delays.

I was told I couldn’t have [an epidural] until I moved to labour ward but I couldn’t move to labour ward as it was full. I was only moved when I was pushing.

Survey participant 88910706

The case studies revealed variation in policy and practice across all sites (Table 31). Increased IOL rates and staffing shortages were the primary contextual influences shaping cervical ripening approaches to in-hospital and at home.

[I]t’s the induction of labour numbers … they’re just so high now. You know, that is one of our biggest pressures. We look at it every month on the dashboard and it just seems to be going … I mean God, I think at the end of last year, we hit a 58% induction rate. It was just chronic and the pressure on the service …

Site 3 midwife interview 063

[W]e date, we incident report every single delay we have, we put them through and you see the trends and when you have more staff, you see it go down and it’s just so hard.

Site 1 midwife interview 055

COVID-19 influenced approaches to cervical ripening, but little clarity or consistency rationality around decisions regarding changes in practice. One case study site increased home cervical ripening, while another completely suspended it.

[W]hen we got to March and COVID arrived and we didn’t want people in the hospital and women didn’t want to be in the hospital, that was the environment for change was there.

Site 4 midwife interview 075

Home cervical ripening was offered at just over half (54%) of trusts and boards represented in the COVID-19 impact survey responses. Criteria for offering home IOL varied considerably; often, this severely limited eligibility and meant that very few people were offered the option to return home.

Respondents at most maternity units reported that the COVID-19 pandemic did not impact IOL and cervical ripening practice: 75% no change in method used for cervical ripening; 93% no change in approach for ‘post-dates’ IOL; and 65% no change in criteria for offering home cervical ripening. The pandemic had only a small impact on the numbers of women being offered home cervical ripening: 61% of services reported no change; 28% reported more and 5% reported fewer women offered the choice to return home.

1. 2. What is the acceptability of home or hospital and different methods of cervical ripening to women (and their birth partners) and implication for their experience of IOL and care?

Attitudes towards home cervical ripening varied among those who completed the postnatal survey. While attitudes were more positive among those who took part in interviews, especially towards the idea of home cervical ripening, experiences of those who actually went home were mixed. Positive experiences were most often associated with being in the comfort of one’s own environment.

I think for both of us, rather than be in hospital, we just got on with our normal day. I think that was definitely better for both of us.

Site 3 birth partner 027

Some, however, felt home to be risky and preferred to stay in hospital. Safety appeared to be the biggest concern for those who said they would not want to return home.

[C]oncerns about safety of induction and distance from hospital to home meant staying at home for induction wasn’t an option I would consider.

Survey participant 78644478

I wanted to be in the safety of the midwives, doctors, nurses and all the care team there … if I had questions, I know that the midwives were there immediately to hand.

Site 3 service user 082

[T]he care you’d getting in hospital, they’ve done it a million times, so I just feel a bit better when they’re on hand.

Site 5 birth partner 108

Crucially, women and birth partners wanted the choice to go home or stay in hospital. Lack of choice about place of cervical ripening was evident in the survey comments and in the case study data.

I was told I could have balloon induction and go home at consultant appointment, then when I attended hospital was told this wasn’t actually something I could have, and I would need the gel induction and would need to stay in – this made me feel disappointed and anxious.

Survey participant 88158381

[T]he nurse, said to us, I’m looking at probably five days in hospital. I wasn’t happy about that. I would rather have been at home with my husband. But no, I was told that there was no way that I would be able to go home.

Site 1 service user 028

I would have preferred to be monitored in hospital, but that wasn’t an option. So it was just, you know, I would be at home with it.

Site 4 service user 062

Women who had home cervical ripening were more likely to choose that option again (64%) and recommend it to others (61%), than those who stayed in hospital (55%, 54%).

There were mixed experiences of different cervical ripening methods. Level of discomfort and ability to cope with discomfort varied between different cervical ripening methods.

It wasn’t that pleasant experience for me at all. I was in a lot of pain having it put in, which I did say to them at the time. After the 24 hours when they checked they said I actually had their reaction to it. So I was very, very sore and swollen.

Site 4 service user 030 – pessary

When I had it done, it was as bad as I thought it was going to be. I mean, it wasn’t horrendous, but it wasn’t as breezy as they kind of made out to be. And it was certainly worse than getting the pessary.

Site 4 service user 071 – balloon

Women who had a balloon catheter inserted reported more discomfort than those who had a pessary or osmotic dilator, although ability to cope with discomfort was similar across all cervical ripening methods.

Attentive care and access to pain relief were important to women and birth partners.

I felt it was ok. I just relaxed, the midwives that were dealing with me were just so friendly and they really made you feel comfortable. So, I didn’t find it too unpleasant.

Site 5 service user 076

1. 3. What is the acceptability of home cervical ripening from the perspective of clinicians and health professionals?

Attitudes towards home cervical ripening were generally positive.

Our understanding and perception is that it will be a lot easier to manage the stress and the anxiety associated with that delay if somebody is at home and in their own environment and relaxed rather than stuck in a hospital.

Site 2 midwife interview 109

Many saw home cervical ripening as a potential solution to workload and capacity issues.

We do have at the staffing issues, I’m sure quite a lot of hospitals do. I mean, I’m down to one ward at the moment, normally have two so if these women can stay at home safely, then it’s much better for us than have them just sitting here looking at us thinking, is it me next?

Site 4 midwife interview 090

It frees up hospital resources as well. So they’re not just having to stay in a hospital room by themselves in the hospital just waiting for things to happen.

Site 5 health professional interview 064

However, while there were no negative attitudes to home cervical ripening expressed, some did not perceive it to be the straightforward solution that others expected as there were unintended and unmonitored consequences for workload, such as additional scheduling work and management of re-admissions.

[I]t is acknowledged as a risk because with the staffing levels as they are at the moment, we can only get a woman to a certain point in that induction process, if we haven’t got a midwife available to do that and give them the one-to-one care on Labour Ward … it’s not uncommon to see delays of 24 hours plus with women just sitting in a ward waiting for that to happen, which I’m aware of the fact that with our staffing levels it may not completely remove that risk by doing an outpatient induction.

Site 2 midwife interview 109

There were concerns about appropriate method for home cervical ripening; many healthcare professionals perceived mechanical methods to be the safest for home cervical ripening.

Home induction with Propess makes me a little bit twitched as to what, if anything, goes wrong and they haven’t realised, and they’re not being monitored, are they?

Site 1 midwife interview 020

1. 4. What are the cost implications from the service user perspective?

See health economics analysis in Chapter 4.

1. 5. What information and outcomes are important for pregnant women and their partners?

Having information concerning cervical ripening setting, methods, length of time, and pain and having choice about cervical ripening were important to women and their birth partners.

I think they were very, very good with the literature that was provided and explaining the options when we came in for the induction. We spoke to both the midwife and a consultant who actually inserted the rod. They explained everything and how it would go.

Site 3 service user 003

[P]robably have anxiety just because it was overdue, but nowhere near as bad because they were keeping us informed of what the options are, what they’re going to do and what might work, and if not, then what would happen.

Site 4 birth partner 084

Half of survey respondents (50%) did not feel that they had enough information to fully understand what to expect during IOL; 53% reported having easy access to information about what to do; 52% of those who remained in hospital and 56% of those who returned home during cervical ripening had access to information about different types of IOL. More women who remained in hospital (66%) than who went home for cervical ripening (47%) felt that they were given clear information about IOL. However, women often reported in free-text comments that they lacked information about what to expect during cervical ripening.

I don’t think I was given enough information … I wasn’t told that this [painful CR] was normal.

Survey participant 88910706

This was also reflected in the interviews with women and birth partners, who reported receiving limited information on IOL more generally, including risks and benefits.

In terms of like the actual process of what was going to happen, she did go over it vaguely, but it was very much like ‘when you go in for your induction, they’ll explain everything once you get there’.

Site 4 service user 031

There is no risks or benefits discussed, it was pretty much just ‘this is what we would do in this circumstance. It’ll probably be really quick for you because of X, Y and Z’. And it wasn’t.

Site 4 service user 074

Over half of the women (57%) felt that they either had no choice or no alternative option when deciding whether to have an IOL. Women in the survey and interviews reported feeling ‘pushed’ into the choice or perceived it as a one-way choice where IOL was the only option for a safe labour and birth.

It was never something I had a choice in and I was told if I didn’t get induced there was a high chance of my baby being stillborn because I was almost 42 weeks so this scared me.

Survey participant 89942117

There was no kind of discussion as to what I wanted or what I was comfortable with. It was more ‘this is what’s going to happen’.

Site 3 service user 080

Some felt it was not possible to make a truly informed choice about IOL despite their own push for this.

I was induced because of my age. Whilst it was made clear that the decision was my choice, I also felt a lot of pressure from health professionals to be induced. I read up on the subject to inform my decision and asked a series of questions but felt a strong push to be induced quickly and before my due date which I was not comfortable with. It made my last couple of weeks of pregnancy quite stressful.

Survey participant 74256136

Many women stated they were not given a choice of method.

It was gel. Yes, I didn’t have a choice. They never gave me a choice, but I knew that my friend did the pessary.

Site 5 service user 016

1. 6. What are the psychological correlates of cervical ripening setting?

It was not possible to report correlates of home and hospital cervical ripening due to the small numbers of respondents who returned home; however, other data informed the understanding of psychological impact and aspects of cervical ripening. The short WEMWBS was used to measure psychological well-being. The mean score was 24.5, above the population average of 23.2; 102 women (33%) scored below the population average; and 2 women had a score of just 7.

Forty-one women who left free-text comments described their experience of IOL as difficult or traumatic and/or having caused significant long-term negative impact on their physical and/or mental well-being.

It was all so horrendous I will never have another child. It gives me anxiety thinking about it all. Before this experience I did want more than one child.

Survey participant 89649436

Participants in the case studies reflected on their experiences psychologically, and some reported difficult or even traumatic experiences related to the providing and receiving cervical ripening care both at home and in-hospital.

I think I was just worried that I was going to be in absolutely horrific pain for hours on end and it was going to be the worst thing ever. That’s still how I remember it, but I think probably I’ve blurred my memory a little.

Site 5 service user 033

I would never have an induction again. I wouldn’t do it. If I was overdue like I was before to the point where it would need an induction, I would opt straight for a C-section … [I]f I was to get pregnant again, I would probably need counselling before I gave birth because it was such a horrific experience that I know I would need support around it.

Site 2 service user 093

I found it traumatic witnessing it because being completely helpless in a scenario, just watching somebody suffer for that length of time, trying your best to try and do anything to try and help them out and alleviate that pain in any way you can.

Site 1 birth partner 040

This also applied to maternity professionals, some of whom reported moral distress in relation to the process.

I feel like we had quite a few recently where it’s just not worked and it just feels a little bit like you’re playing into assault in some ways and playing into trauma and being a part of something that you know could well lead to trauma, but not really knowing how to get off the train because that’s what you do in a hospital.

Site 2 midwife interview 041

1. 7. What potential factors mediate women’s experience, for example, rurality, distance from hospital, information provision, professional support?

There were several factors that mediated women’s experience: healthcare professional support, birth partner’s presence, information provision and having choice; privacy and having their own space; delays. Factors prevalent in poor experience were: lack of support; separation from birth partner; lack of privacy; inadequate pain relief; delays impacting progression of IOL process.

For women who completed the survey, the most important factor in a positive experience was support from kind, caring staff and feeling safe.

My midwife was incredibly supportive throughout labour and birth. We felt safe and well cared for.

Survey participant 83384325

In the interviews, most women and their birth partners spoke positively of the care they received. They highly valued healthcare professionals’ support throughout their IOL care, and this mediated whether an experience was positive or not, especially during CR.

Staff were amazing and I think I was on the induction ward for two days and every midwife I came into contact with was just a lovely. So, it was really nice that the staff are just so good with the patients.

Site 5 service user 076

Birth partner’s presence was one of the most important mediating factors for women.

I’m quite an anxious person anyway. It’s just having that partner to hold your hand and support you through and reassure you, tell you that things are going to be OK and that sort of thing. But I think the main thing for me was being that second voice.

Site 2 service user 096

Birth partners were an important source of support for women during their cervical ripening, but they often could not be present as much of the time as wanted and sometimes missed the birth. This was a source of anxiety for women and denied them important support during cervical ripening.

He [birth partner] had nowhere suitable to wait as we live 3 hours away from the hospital and this made him anxious as he wasn’t there to support me. Also made me anxious as I knew he was waiting for news and worrying while he was not with me.

Survey participant 87927535

Those who had their own space appeared to have more positive experiences, whether this was their space at home or a space in a separate room in hospital.

I went for the outpatient one. It was kind of positioned to me that that was the better option because you get to go home and be in your own house … It just seemed more relaxed and organised to me … those two combined were really big things for me: going home and passing in my own time.

Site 3 servicer user 023

I think the fact that I had a private room made it much better.

Site 5 service user 016

The physical environment of IOL suite/pre labour ward was often inappropriate and inadequate. Many women who stayed in ‘bays’ in the antenatal ward during cervical ripening spoke about the lack of privacy and how this sometimes negatively impacted their experience.

I was labouring behind a curtain, no privacy, others all around me … it was really hard to focus and stay calm and relax with no privacy of my own, no pain relief and no food.

Survey participant 82998991

[Redacted] was trying to sleep, and it’s difficult to sleep when people are constantly coming in and out of the room. You know, we’ve only got a curtain between you and next door, and people are in pain, screams. You can hear family on the phone, you know, there’s always some noise in the background.

Site 5 birth partner 097

Receiving information and being offered genuine choices about IOL contributed to positive experiences. Many women, however, reported not having genuine choices during IOL, as previously highlighted.

I really struggled with the pregnancy, and I’ve struggled since, but I think it’s just the lack of information … If I could have just sat down, had a conversation with somebody and asked what I needed to ask, I think it would have been different.

Site 2 service user 111

They [the staff] were open to questions. I know that they’re very busy, but I never felt that they were rushed with me. If I had anything to ask, they answered.

Site 4 service user 030

Delays in the IOL care, especially moving from having cervical ripening to having ARM in the labour ward, shaped women’s experiences. This was experienced by women receiving care in hospital and by those who went home.

I suppose the only major worry I had was how long it was taking them to get me into the hospital and getting bumped every day because I was a healthy, pregnant woman is awfully anxiety inducing.

Site 4 service user 019

1. 8. What are the service barriers and enablers of adoption of home cervical ripening?

Staff shortages was one of the key barriers identified during the case studies. Delays during IOL were frequently reported, primarily connected to unit capacity, staffing and workload.

There will be delays unfortunately. I think just to manage it as best we can, but then the unpredictability of our workload hugely impacts on the ability to offer induction.

Site 4 midwife interview 067

Others touched on the complexity of implementing a new cervical ripening approach into practice.

[W]e just culturally have used inpatient induction for so long … I think the process is new and…I think new things scare people a bit.

Site 5 obstetrician interview 110

Enablers included cross-boundary collaboration, cross-trust/board knowledge sharing, consistent training and professional confidence.

It wasn’t a particular easy transition to out of hospital induction … If we offer it as a real option and a safe option with robust support and guidelines around about it, we can do this safely … [obstetrician co-lead on labour ward] was just so confident. She’s happy to look at the ones in the book that I look at am not sure why they’re coming in for an inpatient and say 'No she can go home' and it gives the confidence to the staff … But it was the coming together of the team that was really key to getting up [at home cervical ripening] and running.

Site 4 midwife interview 075

Response to the pandemic impacted offer of home cervical ripening in variable ways, acting as a facilitator at some sites but a barrier in others. Making home cervical ripening the default service also influenced the number of women who are offered and then go home for cervical ripening. However, interviews with women and birth partners revealed that home cervical ripening was not as acceptable when it was default as it would be if it were a genuine choice.

1. 9. Are there any unintended consequences associated with home cervical ripening, for individuals, families and or services involved?

For service users, lack of choice was a key issue associated with home cervical ripening and IOL more generally. Choice about place of cervical ripening is individual, and some women who returned home felt they did not have a choice to stay in hospital, others who stayed would have liked the option to return home.

I would like to have been given the choice to go home or stay at hospital. I didn’t know there was an option without losing my place in the queue for the labour ward.

Survey participant 80858153

I was not offered the choice to stay in hospital after my balloon was inserted. This would have made all the difference with a toddler at home.

Survey participant 89706212

For staff, potential workload increases were an unintended consequence; however, much of this workload is hidden. Home cervical ripening implementation highlighted that there is potentially more of this hidden work associated with this approach as there is additional administrative work and stress associated with prioritisation, discharge and re-admission procedures for women going home and returning to hospital and in telephone communication with women undergoing cervical ripening at home.

It’s obviously hugely disheartening for the people who are waiting around, but I think it also just creates really undue levels of stress on the clinicians who are trying to manage those lists and trying to keep on top of who’s coming through.

Site 1 midwife interview 021

All women have had delays across the board and it’s been quite difficult for medical professionals and midwife leads to try to prioritise long lists of women that are waiting. It’s quite difficult because in maternity care in general, there is an issue with burnout and keeping staff and I think the pandemic has made this worse.

Site 4 obstetrician interview 084

If the woman on top of the list, we ring her, and she doesn’t answer the phone at 1 o’clock in the morning, and you think 'OK, what if she rings me back at 3 o’clock and I’ve brought the next one in and she’s delayed again?'. But it’s a process – it’s a system, and we have to follow it. If I’ve got capacity right now, she needs to come in. I will ring her three times if I have to, and I’ll document that, and I’ll move on to the next one. If the next woman doesn’t pick up, then I’ll move on to the next one. I’ll keep going until I’ve got them all in.

Site 1 midwife, focus group 031

Finally, there were potential safety issues regarding IOL, whether cervical ripening was in hospital or at home, given the staff shortages and delays. During data collection, several of the accounts recorded by the research team triggered a formal reporting to the trial steering committee about safety concerns and communication to site 1 about them.

Study design as planned at outset

At study outset, we planned to undertake two economic evaluations alongside the CHOICE study: (1) to determine the cost-effectiveness of home cervical ripening with prostaglandin (exposure) versus in-hospital cervical ripening with prostaglandin (comparator) for pregnant women having IOL and (2) to determine the cost-effectiveness of home cervical ripening with balloon catheter (exposure) versus home cervical ripening with prostaglandin (comparator) for pregnant women having IOL. All costs were to be reported in UK pounds sterling (£) adjusted for inflation for the price of the year 2021.

The planned method was to undertake a within-study economic evaluation conducted alongside the CHOICE prospective cohort study utilising routine clinical electronic maternity (from BadgerNet) and neonatal data from the National Neonatal Research Database. The methods for analysis were in accordance with good practice guidelines for economic analyses based on observational data, using propensity score matching adjustment to control for treatment indication bias, which also aligned with the original statistics analysis plan. Extrapolation to the longer term via modelling was to be considered if the cohort data suggested a difference between arms in terms of the primary outcome – NNU admissions within 48 hours – or there were short- and medium-term complications for the mother or child, which could be important to capture over a longer time horizon.

A secondary analysis from the patient perspective was planned, using evidence on patient related costs and spill-over effects from the qCHOICE survey. As resource-use data from the patient perspective would not be available from the BadgerNet and NHS routine data sets, tailored questions related to economic resource use were embedded in the process evaluation survey questionnaire (qCHOICE).

A health economic analysis plan (HEAP) was developed in consultation with the CHOICE study statisticians and was reviewed regularly by the study management group. The HEAP was planned to be submitted for publication to a peer reviewed journal; however, following the internal pilot phase and revised analysis plans, publication plans were put on hold and, instead, a revised HEAP was developed alongside the revised main study analysis plan.

Changes to study design

In line with the key changes that were made to the CHOICE observational cohort study design and expected sample size, the HEAP was also revised and approved following the internal pilot. There are some maternity units in the UK who offer at-home IOL as default, and therefore the original question regarding cost-effectiveness of this approach remains valid, albeit amended to reflect current practice and the revised CHOICE primary study question (home cervical ripening using balloon vs. hospital cervical ripening using prostaglandin). Although the CHOICE study data set was underpowered to answer the primary efficacy question, a decision-analytic modelling approach (as opposed to an economic evaluation based solely on the observational cohort data set) has been used to estimate potential cost-effectiveness of at-home versus in-hospital cervical ripening, using as much of the CHOICE data as possible, and supplementing this with any additional evidence from the wider literature and clinical expert advice where needed. There were two main changes from the original protocol: (1) the analysis now focuses on the comparison of at-home cervical ripening with balloon catheter versus in-hospital cervical ripening with prostaglandin; (2) use of a decision tree model-based analysis (as opposed to individual patient data-based analysis), albeit using as many data from CHOICE as possible.

The revised key economic objectives are detailed in Chapter 4.

Health economic analysis overview

The economic evaluation explored the cost-effectiveness of home cervical ripening with balloon catheter (exposure) versus hospital cervical ripening with prostaglandin (comparator) for pregnant women having IOL. Cost-effectiveness analysis is a form of economic evaluation in which both the costs and effects of two or more health interventions are compared, and the results report the incremental difference between the alternatives under consideration as an incremental cost-effectiveness ratio (ICER). 30 The analysis was undertaken from the perspective of the UK NHS and PSS for price year 2021. Secondary analysis was undertaken from the patient perspective to account for IOL-related resource use incurred by women and their families. This analysis is informative for the home cervical ripening setting. The time horizon of the analysis was from the time of initial application of cervical ripening method up to postnatal discharge or 28 days post birth, to capture any cost and morbidity events incurred in the neonatal period. We used a short time horizon in line with the timeline and primary outcome of this study. Recent evaluations on IOL in other contexts have also used short-time horizon (from time of IOL to hospital discharge) for assessing cost-effectiveness of IOL at outpatient department using balloon catheter in Australia. 31 Costs and outcomes will not be discounted as the time horizon is restricted to < 1 year.

Data sources

Data used in the economic analysis are primarily from the CHOICE study sites consisting of 26 obstetric units across the UK (18 offered only in-hospital cervical ripening -predominantly with prostaglandin, and 8 offered home cervical ripening using balloon catheters). The maternity data were recorded by clinical staff (midwives, doctors and neonatal nurses) during the course of antenatal, intrapartum and postpartum care in the BadgerNet maternity system. Neonatal data were also obtained from BadgerNet. All data obtained for economic analysis were anonymised and accessed from the Edinburgh Clinical Trials Unit servers managed by the University of Edinburgh.

A rapid systematic review of economic evaluations of cervical ripening at home or in hospital was undertaken in August 2022 to explore: (1) a wider evidence base on the effectiveness and possible complications of home-based/outpatient cervical ripening and (2) cost-effectiveness studies and existing decision models of home/outpatient versus hospital IOL. 32 A detailed description of this systematic review is included in Appendix 3, with Appendix 3, Tables 119 and 120 containing the search terms used for the effectiveness and cost-effectiveness reviews, respectively. The papers from this review were used to inform, develop and potentially parameterise the CHOICE study economic model, as well as to aid reflection and comparison of results. We added an overview of the cost-effectiveness studies identified in the rapid systematic review in Appendix 3. A total of nine studies were included, conducted in various settings, such as Europe, USA, India and Australia. Most studies (seven of nine) used a short time frame, focusing on events from randomisation/admission/IOL to discharge. Two studies extended the follow-up period up to 4 weeks post delivery. Decision tree or decision-analytic modelling approaches were used in only two studies, while the remaining studies employed non-parametric bootstrap or logistic regression methods on study data sets. Three studies used NNU admission avoided as the primary outcome measure, and two studies estimated quality-adjusted life-years (QALYs) for the mothers (not babies). Other outcomes assessed in the selected studies included postpartum length of stay, asphyxia, postpartum haemorrhage and caesarean section rates. Notably, six of eight studies identified the use of Foley or Cook balloon in outpatient settings as a cost-saving strategy compared with inpatient settings. The other study did not use balloons. Further details of the rapid systematic review search strategy, databases and Preferred Reporting Items for Systematic Reviews and Meta-Analyses diagram results can be found in Appendix 3, Figures 14 and 15.

Decision model

The cost-effectiveness analysis was undertaken using a decision tree model to synthesise the parameter inputs of relevance for the economic analysis. Decision trees are economic models that illustrate alternative decision options and their possible consequences. The structure of the model was based upon the pathway that patients will follow in IOL in the UK setting, but is also similar to previously published decision models in this setting and patient group. 33 As per the CHOICE observational data set, the patient population for the economic model is women who had singleton pregnancies with IOL at 37 weeks’ gestation or more with broadly comparable levels of risk, such as those without key risk factors for adverse maternal or perinatal outcomes, and who had pregnancies in which there was no contraindication to home cervical ripening. Figure 5 illustrates the decision tree pathway. In the model, women either can receive balloon catheter cervical ripening and are then sent home for their cervical ripening stage of IOL (exposure arm) or can receive prostaglandin and are admitted to hospital into an antenatal ward. Women in the ‘home cervical ripening’ arm, can be admitted to antenatal ward at a later time, and may also call the maternity unit from home for advice and reassurance during their IOL period. The decision tree is split by type of delivery (vaginal, vaginal with instrumental, and caesarean section), as this impacts on the costs, duration of time in labour ward and possible complications. The primary outcome is NNU admission avoided, rather than NNUadmission, as this enables the outcome to represent a health benefit for easier interpretation of economic outcomes, rather than based on a detrimental effect of NNU admissions.

FIGURE 5.

Decision tree pathway displaying potential processes and outcomes of IOL using either balloon or prostaglandin cervical ripening.

Model parameters/data

Presentation of results and uncertainty

Incremental costs (ΔC) and effects (ΔE) were used to estimate the ICER (where ICER = ΔC/ΔE). Cost-effectiveness is typically expressed as an ICER, that is the incremental cost per NNU admission avoided; however, given the original study hypothesis of non-inferiority within a margin of 4% increase in NNU admissions (i.e. no difference between arms in terms of NNU admissions, and if any minimal difference is observed it would be acceptable up to a 4% increase in NNU admissions), it is likely that there will be no or minimal difference in effects between the home cervical ripening and hospital cervical ripening groups. The traditional ICER presents results as a ratio representing the relation between difference in costs and difference in effects, and in the case of a non-inferiority effectiveness measure (and depending on the cost outcome) could result in a ‘dominated’ or ‘dominant’ ICER in relation to the comparator.

A 1000-iteration Monte Carlo simulation was undertaken for probabilistic sensitivity analysis (PSA) to represent uncertainty in each of the model parameter inputs and to illustrate the impact of this on the overall cost and effect results. The stability of the 1000-iteration Monte Carlo simulation was tested and was found to be within reasonable bounds; 95% uncertainty intervals were presented with the mean difference in cost and effect estimates. 40 To visualise this uncertainty, the 1000 iterations from the PSA were plotted on a cost-effectiveness plane, with the non-inferiority margin for effects presented. Non-inferiority of the effect outcome could be illustrated by a narrow spread of cost–effect pairs very close to zero (and within the bounds of the non-inferiority margin) on the horizontal axis on a cost-effectiveness plane.

Subgroup and sensitivity analysis

As per the main study analyses, the economic analysis plan intended to explore potential subgroup analyses of: (1) nulliparous and parous women and (2) indication for IOL, however, due to little difference between groups and insufficient numbers to allow meaningful analyses, we did not conduct these in the final analyses.

Additional sensitivity analysis was undertaken to explore alternative IOL methods, taking into account possibility of variations in the utilisation rates based on clinical experts’ advice. The base case analysis assumed a 50% split between the Foley and Cook balloons. Four scenarios were generated, as follows: (1) Foley balloon (100%), Cook balloon (0%), (2) Foley balloon (0%), Cook balloon (100%), (3) Foley balloon (75%), Cook balloon (25%), (4) Foley balloon (25%), Cook balloon (75%). We conducted a separate sensitivity analysis considering societal perspective cost (i.e. including both NHS and patient’s costs).

Resource use

Table 34 presents the mean resource quantities per arm.

Table 34 presents the resource use in each group (based on the CHOICE study data set). The length of stay of the women who received balloon catheter at home was significantly lower than those who received prostaglandin ‘in hospital’ (mean 5482 vs. 7317 minutes; p < 0.01). We estimated the average length of stay in antenatal, labour and postnatal wards. Mothers who had home cervical ripening spent significantly less time in the antenatal ward and more time in the labour ward compared with mothers who had hospital cervical ripening. There was no significant difference in the length of postnatal stay between the two groups, which could indicate no or minimal difference in maternal complications that required a longer stay in hospital. For vaginal and assisted deliveries, mothers’ length of stay was significantly higher in the antenatal ward when they received the in-hospital cervical ripening method. Mothers who had caesarean delivery after home cervical ripening stayed significantly less time in each ward except labour ward than who had caesarean delivery after having hospital cervical ripening. The admission to and length of stay of newborns in the NNU was less in home arm compared with the hospital arm, irrespective of delivery types and in total.

Table 35 presents the proportion of mothers or partners who communicated with their midwife during the antenatal period. Higher proportions of mothers/partners in the home cervical ripening arm (75.4%) communicated with the midwife during the period than in the hospital cervical ripening arm (2.4%). Similarly, the number of calls and average duration of calls were higher for mothers who had home cervical ripening compared to mothers who had cervical ripening at the hospital.

Table 36 presents the mean cost (£) per service by delivery types and model groups. Overall, the home cervical ripening using a balloon catheter method reduced the average total cost by £1423 compared with prostaglandin at hospital (£2237 vs. £3660). The antenatal and postnatal ward stay cost was lower for home cervical ripening than hospital ripening. However, the costs of midwifery staff dealing with phone calls and maternal duration in labour ward were higher for the home arm than hospital arm. For both cervical ripening approaches, the delivery cost was the highest for caesarean section followed by assisted and vaginal delivery, respectively. The NNU stay cost was highest for newborns in the ‘in hospital’ group than newborns in the ‘home’ group for all types of deliveries except vaginal delivery. Overall, the cost of NNU admissions was greatest in the ‘in hospital’ group.

Effectiveness outcome

The base case economic analysis uses NNU admission (within 48 hours of birth) avoided, transforming the study primary outcome of NNU admission (a detrimental outcome, which the original cohort study aimed to show non-inferiority within a 4% margin) into a health benefit scale. Table 37 illustrates that home cervical ripening compared with hospital cervical ripening reduced NNU admission by 1.5% (p = 0.224), which is statistically insignificant, yet well below the non-inferiority margin of a 4% increase.

Base case economic analysis results

Table 38 presents the results of the economic analysis, showing the deterministic results and the probabilistic results, which account for the underlying uncertainty in the model inputs and how this impacts on cost-effectiveness outcomes.

The economic analysis found that home cervical ripening with balloon led to a cost savings of £993 (−£1198, −£783) per woman, with no difference in NNU admissions avoided (mean 0.005, 95% CI −0.05 to 0.013), and is therefore considered to be the dominant strategy compared with in-hospital cervical ripening with prostaglandin.

Alternative outcomes: incremental cost per time in hospital

Reporting cost-effectiveness in terms of: (1) incremental cost per time in hospital and (2) incremental cost per time from IOL to birth can be important for consideration of alternative outcomes (beyond the CHOICE primary outcome) and also for comparison with other similar studies. Table 34 reports the resource use summary results from the CHOICE study, detailing both total time in hospital for each arm, and time from IOL to birth. In terms of time in hospital, the home cervical ripening arm has a total of 5482 minutes (3.8 days), compared with 7317 minutes (5 days) in the in-hospital cervical ripening arm, a reduction of 1895 minutes. In terms of an ICER, the home cervical ripening arm is the dominant strategy, both cost saving and reducing overall time in hospital.

When the outcome of interest is time from IOL to birth (regardless of where that time is spent), Table 34 shows that the home cervical ripening arm has the longer duration [3105 minutes (51.7 hours) compared with 2662 minutes (44.4 hours)]. This translates to an ICER of £137 per additional hour from IOL to birth. While this is not a large difference in time duration between arms, it does show that the at-home balloon catheter cervical ripening option is longer overall, which could impact on patients’ perspectives and preferences for method of cervical ripening. The longer duration may be due to the cervical ripening method of balloon catheter as opposed to prostaglandin, where IOL time is longer.

Sensitivity analyses – use of Foley and Cook balloons

Table 39 reports the sensitivity analysis of alternative IOL methods considering changes in utilisation rate from our assumption 50% split between the Foley and Cook balloons based on clinical experts’ advice. The increased use of the Foley balloon led to a lower cost estimate for the home IOL arm, as this balloon is less expensive compared with the Cook balloon. In all four different scenarios, home cervical ripening remained a cost-effective strategy. We considered two balloons equally efficacious in this analysis.

Sensitivity analyses – patient perspective costs

Table 40 shows the resource items of relevance from the patient perspective. While this is a small sample (the qCHOICE study surveyed a subsample of the larger CHOICE cohort study), this is indicative data of the resource use from the patient perspective and the point estimates (along with standard errors) can be useful inputs to the economic model. The data indicate that 74.1% and 97.2% of partner or caregivers stayed with the mother who had hospital and home cervical ripening, respectively. The percentage of mothers who relied on paid or unpaid child-care services was higher among mothers who had cervical ripening in hospital (30.4%) compared with at home (17.1%). Among the mothers who had home cervical ripening, 63.9% called the midwife for consultation. Around 47% of the mothers in the ‘home’ arm travelled to hospital and came back to home when they were waiting for initiation of labour at home. There was little difference between average transport costs, which overall were higher among mothers who had hospital cervical ripening (£23.1) than mothers who had home cervical ripening (£18.74).

Table 41 shows the cost from the patients’ perspectives in the home and hospital study arms. The total patient perspective cost was higher in the hospital arm (£954) compared with the home arm (£665); this is predominantly driven by the opportunity cost of partners’/other caregivers’ time supporting the mother either at home or in hospital ‘away from other activities’, There was little reported difference in terms of child-care cost (both paid and unpaid) and transportation costs between arms.

Table 42 reports the economic analysis outcomes, including the patient perspective costs in addition to the NHS costs; that is, the ‘societal’ perspective. The addition of patient perspective costs makes both arms more expensive, yet, overall, home cervical ripening remains the optimal strategy, saving costs to NHS and patients while showing little difference in terms of NNU admissions avoided.

Strengths

The CHOICE study has provided a unique multilayered insight into contemporary IOL practice in the UK that will inform multiple aspects of clinical practice going forward. These include the provision of evidence-based information on the process of balloon cervical ripening at home that can be shared with women and staff, and informing NHS trusts and boards of the potential cost savings involved in offering home cervical ripening with balloon as an alternative to the default approach of in-hospital cervical ripening with prostaglandin. The relatively early gestation (10% of home cervical ripening occurred at 37 weeks of gestation), the wide array of indications (only one in four home cervical ripening cases were for post dates) and the importance of how care is delivered to women undergoing home cervical ripening may each influence clinical guideline development going forward.

Limitations

Multiple elements of the CHOICE study were not conducted as planned and described in the study protocol. This largely related to the impact of the COVID-19 pandemic on site recruitment and IOL practice (setting) such that fewer women in the cohort and health economic analysis sample underwent home cervical ripening than anticipated. An attempt to continue the study to reach the original sample size was judged to be futile. Instead, the primary comparison was changed to ensure that the most common approach to home cervical ripening (balloon method) was compared with the most common approach to in-hospital cervical ripening (prostaglandin). However, the disadvantage was that both method and setting then differed between comparison groups, not just setting, making any conclusions about safety of setting alone impossible to reach. The smaller sample size than expected meant that the original sophisticated analysis plan could not be carried out. It is not possible to know whether the absence of evidence of any difference in safety outcome between home balloon cervical ripening and in-hospital prostaglandin cervical ripening would have been rejected in a larger sample size.

The cohort study analysis took the approach of including women where all relevant variables were recorded or were assumed not to have been relevant to the woman (e.g. no cervical ripening method recorded so assumed not to have had cervical ripening, no admission of baby to the NNU was assumed to mean that the baby was not admitted to the NNU). This ‘complete case analysis’ was performed as it was not possible to establish whether a variable was really ‘missing’ or simply did not apply to that individual woman/baby. Alternative options such as imputing a variable (inserting a value for the missing variable, estimating what that value should be) could have been considered if it had been clear which variables were ‘missing’. Similarly, if it had been known which values were missing, a sensitivity analysis could have been performed to look at the extremes of possible findings if all missing values were positive (i.e. the event did occur) or all missing values were negative (i.e. the event did not occur). As a result of the approach taken, it is possible that women who underwent cervical ripening but did not have full details recorded were inadvertently excluded from analyses, and that women who were included may have been inappropriately recorded as not having experienced an outcome that they did experience due to it not being recorded. The implications of this for the study are unknown, but could include higher incidences of certain outcomes than reported, or that a larger sample of women could have been included had the missing data been included. This issue does not apply to those variables where recording is virtually complete, for example, maternal age, postcode (used to assess deprivation status), mode of birth, blood loss.

Ultimately, the safety of home cervical ripening would ideally be assessed in a randomised trial to avoid the potential for confounding and reduce missing data, but the rare nature of key safety outcomes makes such a trial unfeasible due to the required sample size being excessive.

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 pretrial funding to report writing and dissemination.

From the start, two PPI co-applicants were involved in the study throughout the bid stage.

During the post-funding preparatory work, the PPI co-applicants were included in communications (e-mail, online meetings and face-to-face meetings) for development of the study protocol.

During the study, a PAG was established in July 2020, which included the two PPI co-applicants and two service users. A PPI framework was developed following the NIHR’s framework (INVOLVE) for public involvement in research. It outlined the values, roles, payment and recognition for PPI.

The PPI co-applicants on the team are involved in the production of the final report to the NIHR and publications of findings in peer-reviewed journals.

The PAG met several times through the study and provided feedback on patient-facing materials and ad-hoc consultation on specific aspects of the study, including:

• CHOICE information leaflet

• qCHOICE participant information sheets

• qCHOICE study leaflets and poster

• CHOICE BadgerNet PUSH notifications

• qCHOICE questionnaire-based survey

• qCHOICE social media strategy and Facebook ad campaign.

All documentation and media were amended in light of their comments, including changing pictures used in the Facebook advertising campaign to illustrations, reflecting one member’s experience that the women she works with respond better to these media types. The PAG was also consulted on widening participation in the study, the results of which are reported in the equality, diversity and inclusion statement.

The PPI positively influenced the development and progress of the study, in particular qCHOICE, and improved the quality and accessibility of the patient-facing study materials.

The PAG membership was small, and the study would have benefited from a larger group that was more representative and inclusive. More consultation with the PAG during data collection may have been beneficial, particularly during qCHOICE when survey recruitment was low.

The CHOICE team’s consultation regarding widening participation was a strong point during the study PPI, as it has produced key recommendations (outlined in the equality, diversity and inclusion statement) for researchers and future projects.

Study sites (Q8 what hospital did you give birth in?)

Of the responses, 50% came from women who had an IOL at one of the five qCHOICE sites, rising to 61% at the seven sites focused on for recruitment; 34 responses (11%) came from women who had had an IOL at NHS trusts and boards that were not participating in the CHOICE study.