The Arabin pessary to prevent preterm birth in women with a twin pregnancy and a short cervix: the STOPPIT 2 RCT

Inclusion criteria

The inclusion criteria for the screening and the treatment phase comprised all of the following (i.e. all criteria had to be fulfilled):

• twin pregnancy (monochorionic or dichorionic)

• known chorionicity (as defined by first trimester ultrasound screening)

• gestation established by scan at ≤ 16 weeks’ gestation (in accordance with NICE guidelines4)

• aged ≥ 16 years

• willingness to participate in both the screening and randomisation phase of the study.

Exclusion criteria

Patients were ineligible for the screening phase if they fulfilled any of the following criteria:

• unable to give written informed consent

• known significant congenital structural or chromosomal fetal anomaly at the time of inclusion

• existing or planned cervical cerclage in the current pregnancy

• existing or planned (prior to 20^+ 6 weeks’ gestation) treatment for twin-to-twin transfusion syndrome in the current pregnancy

• suspected or proven rupture of the fetal membranes at the time of recruitment

• bulging fetal membranes at the time of recruitment

• singleton pregnancy or higher-order multiple pregnancies

• women who have had any fetal death (i.e. fetal heartbeat previously detected) in the index pregnancy (prior to randomisation)

• known sensitivity, contraindication or intolerance to silicone

• involved in a clinical trial of an investigational medicinal product, a phase 1 study or an investigation of a treatment for the prevention of preterm birth

• monochorionic, monoamniotic pregnancy

• heavy bleeding due to a low-lying placenta prior to randomisation.

Inclusion criteria: screening phase

• Cervical length of ≤ 35 mm at 18^+ 0–20^+ 6 weeks’ gestation confirmed by an accredited clinician.

In addition, patients were ineligible for the treatment phase of the study if they fulfilled any of the following:

• cervical length of > 35 mm at 18^+ 0–20^+ 6 weeks’ gestation

• cervical length not measured at 18^+ 0–20^+ 6 weeks’ gestation

• bulging fetal membranes at the time of pessary insertion

• suspected or proven rupture of the fetal membranes at the time of pessary insertion.

Screening phase

To assess eligibility for the treatment phase, all women who wished to participate and who fulfilled the inclusion criteria for the screening phase were invited to undergo a transvaginal ultrasound measurement of cervical length between 18^+ 0 and 20^+ 6 weeks’ gestation by an accredited operator (see bullet list). If convenient, cervical length measurement was performed at the time of the routine fetal anomaly scan. In centres where cervical length measurements were part of routine clinical care, women could be consented after the cervical length and were not asked to have a further transvaginal scan, provided:

• the transvaginal measurement was performed between 18^+ 0 and 20^+ 6 weeks’ gestation by an accredited operator [i.e. someone who had completed the Cervical Length Education and Review programme (Oklahoma City, OK, USA) or Fetal Medicine Foundation (London, UK) cervical length training] who had been delegated by the principal investigator

• the woman was eligible for participation (as per the above inclusion/exclusion criteria) and the pregnancy was not ≥ 21 weeks’ gestation.

Images of the cervical length measurements were anonymised and uploaded to the electronic case report form (eCRF). Women with a cervical length of ≤ 35 mm were invited to continue to the treatment phase and randomised to either the intervention or the comparator group, as described below.

Intervention group (treatment phase)

The intervention was the Arabin pessary, which was inserted through the vagina and around the cervix between 18^+ 0 and 20^+ 6 weeks’ gestation by a trained member of the team. The Arabin pessary was given in addition to all standard-care measures. Unless indicated earlier, the pessary was removed at 35^+ 0 to 36^+ 6 weeks’ gestation.

Video instructions for pessary insertion were circulated and formed part of the study protocol training given to clinicians. Meetings about the trial gave clinicians the opportunity to practise pessary insertion on anatomical models. There were no prohibited cotreatments.

Comparator group (treatment phase)

The comparator was standard care only. There were no prohibited cotreatments. The Arabin pessary was not in general use in women with twin pregnancy in participating hospitals and so use of the pessary outside the trial (and outside allocation to the intervention group) was not anticipated.

Baseline assessment

In addition to eligibility criteria, the following baseline information was collected on all women: age, ethnicity, employment status, years of full-time education, date of last menstrual period, estimated date of delivery, chorionicity, cervical length, current smoking (tobacco and e-cigarettes), current recreational drug use, current alcohol use, obstetric history (parity, miscarriage) and ongoing medical conditions (e.g. hypertension, insulin-dependent diabetes, respiratory disease, cardiac disease, neurological disease, skin condition, thrombophilia). We also collected information on the current pregnancy, including the results of any fetal anomaly scans for each twin and the results of amniocentesis on each twin. All baseline assessments were conducted between April 2015 and February 2019.

Follow-up

Follow-up visits to review pregnancy and fetal well-being were conducted at approximately 4-weekly intervals (plus or minus 1 week) following randomisation until pessary removal at around 36 weeks’ gestation. Women randomised to the pessary and those undergoing standard care were all invited to participate in study visits. These study visits could be face to face, by telephone or by other means of communication (e.g. by letter or e-mail). At each study visit, we collected information on any pregnancy complications (i.e. serious adverse events, including bleeding), experience of the pessary during the previous 4 weeks (if applicable) and reason for and duration of any hospital admissions.

Primary outcome

There were two primary outcomes for women in the treatment phase of the study: (1) an obstetric primary outcome and (2) a neonatal primary outcome.

The obstetric primary outcome was all births before 34^+ 0 weeks’ gestation following the spontaneous onset of labour. Preterm prelabour rupture of membranes at < 34 weeks’ gestation with or without contractions was included in this definition of spontaneous onset of labour. Iatrogenic delivery due to maternal or fetal conditions was not considered to fulfil the criteria for the primary outcome.

The neonatal primary outcome was a composite of adverse outcomes, including stillbirth or neonatal death, periventricular leukomalacia, early respiratory morbidity (defined as any need for supplemental oxygen > 30%, continuous positive airway pressure or intratracheal ventilation or surfactant replacement therapy within the first week of life), intraventricular haemorrhage, necrotising enterocolitis or proven sepsis, all measured up to 28 days after the expected date of delivery.

Secondary outcomes

Key obstetric secondary outcomes were mean gestation at delivery, incidence of all births before 37^+ 0 weeks’ gestation, adverse events (including infection and cervical trauma) and acceptability of the pessary, as determined by participant questionnaire and experience of the device throughout the trial and pessary removal.

Key neonatal secondary outcomes were incidence of each of the individual components of the primary neonatal outcome, median weight (in grams) of the newborn at birth, any deaths of liveborn babies within the first 28 days after birth and discrete episodes of bloodstream or central nervous system infection (i.e. positive blood or cerebrospinal fluid culture), categorised by timing (either within the first 72 hours or between 72 hours and discharge).

Other outcomes

In addition to the primary and key secondary outcomes, we collected information on the obstetric outcomes of incidence of all births before each of 28^+ 0, 32^+ 0 and 34^+ 0 weeks’ gestation, incidence of all births before 28, 32, 34 and 37 weeks’ gestation preceded by the spontaneous onset of labour, incidence of preterm birth before 34 weeks’ gestation preceded by preterm premature membrane rupture, method of delivery (in three categories: spontaneous vaginal delivery or vaginal breech, forceps or ventouse and caesarean section), duration of labour overall and of each of the first and second stages of labour, duration of stay in hospital, other adverse events (including haemorrhage, tachycardia, vaginal injury and other trauma) and serious maternal adverse events up to 28 days after discharge from the hospital.

We also collected information on key neonatal outcomes, including birthweight centile (for gestation) within 4 weeks after expected date of delivery, death of liveborn babies within the first 28 days after estimated date of delivery, cord pH and Apgar score at 1 minute and 5 minutes.

Outcomes for screening phase only

For women recruited to the screening phase but who were not randomised, the primary outcome measure was delivery before 34 weeks’ gestation.

Original sample size justification

We aimed to randomise women with a cervical length of ≤ 35 mm, which we believed to be around the 30th centile. We originally calculated that we needed a sample size of 1850 women recruited to the screening phase, which would allow us to randomise 500 women in the treatment phase (250 in each group), assuming a 15–20% drop-out between screening and randomisation. We assumed a RR reduction of 0.6 in the primary obstetric outcome (i.e. spontaneous preterm labour leading to preterm birth before 34 weeks’ gestation) in the intervention group. We considered our estimate of RR reduction to be conservative, given a RR of 0.49 for delivery before 32 weeks’ gestation and 0.47 for delivery before 34 weeks’ gestation (21% vs. 42%) in the ProTWIN trial12 (S Liem, Academic Medical Center, University of Amsterdam, 2013, personal communication). We anticipated that 35% of women in the control group would deliver before 34 weeks’ gestation. Again, we believed that this was a conservative estimate, given that a systematic review indicated that 34.9% of women with a cervical length of ≤ 35 mm (when scanned at 20 weeks’ gestation) will deliver preterm before 32 weeks’ gestation. 16 Assuming a baseline rate of 35% and a RR of 0.6, we calculated that a sample size of 500 women would provide 94% power to detect a difference at the 5% significance level. If the preterm birth rate before 34 weeks’ gestation was only 30%, the power would drop to 88%. Both estimates allowed for losses to follow-up and imperfect compliance. For the primary neonatal outcome, Liem et al. 12 showed an effect size of 0.42 at the baby level and an incidence of 24% in the control group. We powered our study for a RR of 0.6 for the primary neonatal outcome. Assuming prevalence rates as in the ProTWIN trial,12 our study would have 97% power. In practice, our post-screening groups of women with cervical length of ≤ 35 mm are probably at lower risk than women in the ProTWIN trial12 group (cervical length of < 38 mm), given comparisons of the rate of preterm delivery in each of the control groups. Therefore, if we assume a lower rate of the neonatal primary outcome of (say) 18%, we still have 88.4% power to detect a RR of 0.6 in the intervention group. Such a calculation assumed that analysis at the baby level was appropriate for the neonatal outcome. For the subgroup of women with cervical length of ≤ 25 mm, the anticipated rate of the primary obstetric outcome in the control group was 51% (82/159). 15 The study had 85% power to detect a RR of 0.6 in this group, with a likely sample size of 234, given that 25 mm was likely to be around the 14th centile. 15

Revised sample size justification

Following a reanalysis of the data (masked to treatment allocation) after 29 months of screening, with data on 1214 women screened (September 2017), we estimated that to randomise 500 women the sample size for screening needed to be increased to 2500.

Obstetric primary outcome

In total, 46 out of 250 women in the intervention group and 52 out of 253 women in the control group had preterm birth before 34^+ 0 weeks following the spontaneous onset of labour. The adjusted OR of the treatment effect was 0.866 (95% CI 0.546 to 1.375; p = 0.542) (Table 6).

In total, 67 out of 500 babies in the intervention group and 76 out of 506 babies in the control group had the composite neonatal primary outcome of stillbirth, neonatal death, periventricular leukomalacia, early respiratory morbidity, interventricular haemorrhage, necrotising enterocolitis or proven sepsis. The unadjusted OR for the primary neonatal outcome was 0.88 (95% CI 0.61 to 1.25; p = 0.46). Our planned three-level linear regression model for the primary neonatal outcome (babies nested within mother nested within centre, adjusting for chorionicity) failed to converge. We therefore used standard logistic regression for the neonatal primary outcome, adjusting for chorionicity and clustering at the mother level, giving an adjusted OR of 0.86 (95% CI of 0.54 to 1.36; p = 0.52). The rates of components of the neonatal primary outcome and the adjusted ORs of the treatment effect are shown in Table 7. As described in the statistical analysis plan, the level of statistical significance was set at 1% for subgroup analyses, but the adjusted OR is presented as 95% significance levels.

Subgroup analyses

There was no evidence that the presence or absence of any of the subgroup criteria (i.e. monochorionicity, dichorionicity, cervical length of ≤ 25 mm or cervical length of ≤ 28 mm) had any impact on treatment effectiveness for either the primary obstetric or the primary neonatal outcome.

Obstetric secondary outcomes

Obstetric secondary outcomes are shown in Tables 8 and 9. There are no differences between the groups for any of the obstetric secondary outcomes.

Neonatal secondary outcomes

Neonatal secondary outcomes are shown in Tables 10 and 11. Again, there are no significant differences in outcomes between the two treatment groups.

Device fitting

Questionnaire responses on women’s experience of having the device fitted are shown in Table 13. The majority of women (158/234, 67.5%) found pessary fitting either painless or only slightly uncomfortable, and clinicians found device fitting easy or moderately easy (202/234, 86.3%).

The majority of women who provided a response about the experience of the pessary said that they ‘never’ or ‘only a few times’ felt it and that it was ‘never’ or ‘only a few times’ uncomfortable (Table 14). Experiencing vaginal discharge was common, but bleeding was uncommon.

Device experience throughout the trial

Women’s experiences of the device during the trial are shown in Table 14.

Experience of device removal

Questionnaire responses on women’s experience of having the device removed are shown in Table 15. Of those who responded, the majority of women found pessary removal painless or only slightly uncomfortable and the majority of clinicians found it easy or fairly easy.

Measurement and valuation of hospital resource use

The costs of hospital stay and intervention-related costs are presented in Table 17. Resource consequences included length of stay within antenatal, postnatal and neonatal specialties (e.g. special care baby units or neonatal units), as well as any spells in intensive care for either mother or baby. The length of stays was valued using a per diem unit cost derived from NHS reference costs. 19 Costs of maternal hospital stay (per day) on different wards (i.e. labour, antenatal or other) were derived as a weighted (by activity) average from the NHS reference costs. 19 For instance, labour–delivery ward day unit costs were based on Healthcare Resource Group (HRG) codes of stays of ≥ 1 day [NZ30A to NZ51C (non-elective long stay)]. The costs of neonatal care unit admissions were also determined as weighted averages from NHS reference costs (critical care service codes CCU13–CU14, HRG codes XA01Z–XA05Z). 19 Costs related to the delivery of the Arabin pessary intervention were based on trial expenditure records for pessary device and clinician training and for outpatient attendances for pessary insertion/removal from NHS references costs [HRG code 501 (obstetrics outpatient attendance)]. 19 All costs were expressed in Great British pounds (£) at 2017/18 prices. Length of stay for the study population did not exceed 1 year and, as a result, no discounting of hospital costs was undertaken. To calculate the hospital cost of the babies, we used to two methods. Costing method 1 used the weighted average daily cost for neonatal care unit admissions across different levels of care (£725.65). Costing method 2 used the specific daily cost by level of care (i.e. normal care, special care, high-dependency care or intensive care) and assigned these to the babies’ individual number of days spent under various level of care.

Information on babies’ hospital length of stay (number of days) in the trial was recorded (1) based on the total number of days (i.e. the difference between the recorded hospital admission and discharge dates) and (2) from the number of days babies spent receiving different levels of care (i.e. intensive care, high dependency, special care or normal care). The second method of calculating the hospital costs of babies was performed as a sensitivity analysis and was undertaken to try to take into account how the distribution of total days was spread over different intensity/complexity levels of care and to examine whether or not this changed the cost estimates.

Recruitment and sampling: pregnant women

All of the women who took part in the interview study had previously agreed to being approached by a qualitative researcher when they were recruited into the main trial.

Seventy women were invited to take part by letter, with up to two follow-up telephone calls. Two women declined to participate, two women were hospitalised and so were not followed up and 20 women could not be contacted. Forty-six women agreed to be interviewed and 41 were successfully interviewed (five women could not be contacted on the agreed day). Women who took part in the interview were sent two consent forms, one to keep and one to return.

The women were recruited from 14 of the 57 UK sites taking part in the trial. These sites were chosen to reflect the diversity of hospital settings and UK regions. The sample included women from each arm of the trial [i.e. women who were screened but not eligible to take part (n = 11), women randomised to standard care (n = 10) and women on the intervention group (n = 20)]. The sample included women across a wide range of cervical measurements, ranging from 10 to 60 mm.

Many of the women who took part in the qualitative study were highly motivated to become pregnant and had also experienced multiple disappointments in attempting to conceive and carry a fetus to term. Table 20 details the sample characteristics.

Recruitment and sampling: health-care professionals/trial staff

Seventeen health-care professionals involved with the trial were interviewed, either face to face or by telephone, from 10 trial sites with differing resources and services, for example dedicated twin clinics. Sites also varied in the configurations of trial teams and with different levels of recruitment. The research midwives (RMs) (n = 8) who took part had the most contact with women and were involved in trial recruitment, consent processes, administering the survey and follow-up of women during the trial. We interviewed two consultant obstetricians (COs) and one clinical research fellow (CRF) who were involved in transvaginal scanning and pessary fitting and one midwife who had conducted transvaginal scanning. One consultant midwife (CM) was interviewed. The senior RMs (n = 5) had additional responsibilities in relation to trial administration and the work allocation of RMs (Table 21).

The interviews: pregnant women

The interviews took place between 24 and 34 weeks’ gestation and, wherever possible, before 28 weeks’ gestation. A semistructured topic guide combined consistency and flexibility. The topic guides were designed to explore how women experienced trial participation in the context of their pregnancies to develop a nuanced understanding of the experience of the trial in relation to a pregnancy at risk of preterm labour. The use of prompts within topic areas aided the generation of data from the woman’s point of view and her experience. [For more information please see the project web page: www.journalslibrary.nihr.ac.uk/programmes/hta/130422/#/ (accessed 22 April 2021).]

The women were offered face-to-face or telephone interviews; however, all the women preferred to take part in a telephone interview. These interviews were timed to fit around the women’s family and work commitments, and it was clear that some of the women who took part had gone to some lengths to fit in the interview around their busy lives. Five interviews took place during the pilot phase of the study late in 2015, and these interviews informed the topic guides for the main qualitative study. The remaining interviews were conducted between November 2017 and May 2018. Interviews lasted between 24 and 70 minutes and were fully transcribed. Follow-up interviews were conducted with five women in the intervention group after the removal of the pessary. Three of the women interviewed from the intervention group requested early pessary removal. Therefore, these interviews also captured the experience of having the pessary removed.

The interviews: health-care professionals

Three interviews were completed during the pilot phase of the trial in October 2015. The remaining interviews were carried out between January and September 2018. These interviews lasted between 19 and 83 minutes and were fully transcribed. The timing of the interviews allowed early interviews to inform the later interviews, and the analysis of both women’s and health-care professionals’ data sets to inform each other as the study progressed. As with the women’s interviews, the health-care professionals’ interviews took a semistructured approach, enabled by a topic guide. [For more information please see the project web page: www.journalslibrary.nihr.ac.uk/programmes/hta/130422/#/ (accessed 22 April 2021).]

The topic guides were designed to explore health-care professionals’ experience of administering trial processes, according to individual roles and responsibilities. The RMs’ interviews tended to be longer than those with the COs and clinicians, who were mainly involved with pessary fitting rather than more extensive contact with the women. Many of the RMs were involved at each stage of women’s journey through the trial and they were also involved in the ongoing support and follow-up of women. The topic guide included exploration of professionals’ experience of working within clinical research, moving to a more specific focus on the trial itself, and also incorporated each aspect of the trial process as relevant to individual professional experience.

The majority of these interviews were conducted by telephone; however, some were conducted face to face at two of the study sites. Again, the health-care professionals who took part did so in the context of busy work schedules, but they were keen to share their thoughts and experiences of the trial.

Process of analysis

All interviews were digitally recorded and transcribed verbatim. Analysis began early into the data collection period to help to shape the interviews that were conducted in the later stages of the data collection period. This approach helped to explore the emergent themes in more depth before deciding on a final coding structure. An inductive, cross-sectional thematic approach was utilised in analysing the entire data set, whereby index coding was applied to reveal patterns and themes within the data (e.g. for each trial arm in relation to the women). 20 This method was selected as it was particularly suitable in identifying individual women’s experience of each arm of the trial in the context of their current and previous experience of pregnancy. A similar approach was taken in relation to the health-care professional data set, which facilitated an exploration of professionals’ accounts in conjunction with their individual roles and responsibilities.

NVivo Pro (version 11; QSR International, Warrington, UK) computer software was used to manage and support the coding of data. Initial codes were created around the main interview questions and comparison made across the women’s experience on each arm of the trial and health-care professionals’ accounts. The research team members were involved in ongoing discussions as data collection and analysis progressed. The use of computer software helped to facilitate a critical and rigorous approach to the assignment and with the interpretation of the codes and emergent themes. 21 A two-cycle approach to coding was undertaken. 22 First-cycle coding involved applying structural coding, which facilitated an approach whereby the data were coded according to broad topics [i.e. experience of transvaginal ultrasonography (TVU)]. The segments of data were then subject to further analysis and comparison. Analysis and interpretation were supported by analytic memo writing on emergent patterns and themes. Second-cycle coding further refined the original codes and memos into a smaller number of categories that were subject to further analysis and interpretation. This stage of the analysis involved a team approach, helping to improve the rigour of the findings. 23

Twin pregnancy

The backdrop to all the women’s experiences, as elicited in the interviews, was that of a twin pregnancy and the concerns this generates. Many had complex obstetric histories and, for some of the women, the current pregnancy represented what they saw as their last chance to become a parent. With such high stakes, many of the women encountered significant levels of anxiety about fetal development and the ability of their bodies to support a twin pregnancy.

Participants described a variety of emotional responses and different means of dealing with a pregnancy vulnerable to complications. Excitement was tempered by worry and apprehension and information was experienced as both reassuring and anxiety provoking:

I was very excited, I think I got less excited when I got home and had a little bit of the look at the information [. . .]. So, we were initially very excited, but then I got myself really quite worried.

NE6-Cx38

Many of the women described strategies with which they managed the anxiety they faced in the context of obstetric consultations that focused on potential complications of twin pregnancies. On the one hand, the women described feeling grateful for the additional monitoring their pregnancies incurred, whereas, on the other hand, some described having to manage high levels of anxiety due to continual monitoring, which provided a constant reminder of what might go wrong. Although notions of risk shaped the women’s pregnancies, managing their time served to control their anxiety and keep an emotional distance from the fetuses:

I just kind of live in those 2 weeks . . . and get to the scan, and have the usual anxiety, worry, before I go in there [. . .] and then I kind of move onto the next 2 weeks. So I actually, I never look too far ahead [. . .] [I’ve had] two unsuccessful pregnancies . . . in between . . . [. . .] I think it definitely has added to the anxiety and the worry [. . .] I’m just, I’m not waiting for something to go wrong, but I just feel nervous, to think too far ahead.

NE11-Cx38

The women were keen to learn the signs that might indicate pregnancy complications and, for some, this was also a means of gaining some control over uncertainty:

If we did have symptoms of things and I would know that it wasn’t necessarily normal and it was worth getting checked out. Quite a lot of the symptoms are normal pregnancy symptoms for twin-to-twin transfusion syndrome. So, the things like your tummy getting bigger, but a lot bigger, and your back hurting and needing to drink more. So, it’s quite, I found it . . . or at times I think is this a symptom? Is it not? [. . .] They will check you out and if it’s actually fine, that’s great, but it’s better to be precautious. So, it does give me an element of control because it means I know what I’m looking out for, rather than just blindly going between scans not knowing if everything is OK or not, erm . . . possibly.

NE6-Cx38

Women also drew attention to the role of midwives in relation to the emotional support they provided:

The midwives are lovely, ‘cause they’re very friendly and have got good people skills. All the ones I’ve seen, they’ve made me feel really comfortable, and it’s all normal and absolutely fine. They’ve been very good at making me feel comfortable with things.

P13-Cx33

Experience of transvaginal ultrasonography

Thirty-six of the 41 women who took part in the interview study had encountered TVU during assistive conception treatments or during early miscarriage, and those women expressed no concerns about the procedure itself. Many of the women framed vaginal medical procedures as ‘something you get used to’ or ‘to be expected in pregnancy’. Overall, women described the physical aspect of the procedure as ‘not as bad as a smear’ or ‘not uncomfortable’. One woman admitted that, although she had received information about the scan prior to the procedure, she ‘didn’t sort of think about in my own mind about an internal scan, let alone how they would do it, [. . .] probably naivety on my own part’ (P20-Cx25). From this participant’s description of the procedure, it was evident that trial staff offered further information and, aside from feeling a little embarrassed, she was happy to go ahead.

Although women were accepting of TVU as an aspect of the trial, partners were occasionally uncertain or embarrassed about the procedure, as the following quotation demonstrates:

He was the one in the room that was a bit like, oh, he didn’t know what do with himself. He was like, ‘do you want me to go and hide and stand behind the curtain?’ He said to the lady, she was like, ‘no just go up to the head.’ And I was thinking like, ‘Christ it’s not you that’s getting it.’ You know what I mean? I was like, ‘it’s fine,’ you know, I suppose you just have to get used to these things, don’t you?

P18-Cx25

Most women felt reassured that the procedure posed no risk to fetal development. However, a small number of women still felt worried on the day of their scan:

I was just thinking, ‘oh gosh, [. . .] I hope you don’t touch [my cervix] [. . .]’ because when you have your smear test, they always warn you if you’re pregnant, you know, this can lead to miscarriage, so you need to make sure you’re not pregnant; so those kind of warnings from before made me a bit more cautious.

NE5-Cx38

Overall, the women felt happy with the assurances they were given by trial staff and felt more confident that the procedure was safe:

. . . they talk to you about that [TVU] first, and you know, reassure you that it’s all safe.

P17-Cx28

I asked the question as to whether it was safe to have an internal scan at that point, and I think if the answer had been, ‘well we don’t really know,’ or ‘no’, then it would have influenced my decision.

NE9-Cx37

Some of the women who lacked experience of TVU felt concern about whether or not TVU would be ‘as rough’ as abdominal scanning. Some also made reference to ‘trust in the doctor’ and others remarked at how gentle the procedure was:

. . . [it] wasn’t an unpleasant experience either. It wasn’t like a smear test. It was much more comfortable.

SC9-Cx26

Other women described the TVU operator as ‘gentle and really kind’ (SC4-Cx29).

It was also evident that, although most of the women were comfortable with the prospects of undergoing TVU, at the actual point of having the scan some women described feeling a sense of anticipation in relation to what it would reveal:

I just wanted to, almost get on with it, and have the scan, and know one way or the other.

NE11-Cx38

As noted earlier, anticipating knowledge of cervical length formed part of the process of decision-making regarding the trial. The possibility of entering the trial and receiving the pessary affected reaction to receiving the measurement, as outlined below.

Sexual relations after pessary fitting

Women varied in their experience of and approach to sexual relations after the pessary had been fitted. Only a few described no problems in maintaining their usual sexual relations after having the pessary fitted. One described a situation in which she had worried slightly about whether or not her partner would notice the pessary during sexual intercourse:

My husband said he thinks he’s noticed it, but he’s not . . . but again almost not sure. If he can feel it it’s not every time, and it’s not a barrier.

P3-Cx33

Other participants had tried to continue with sexual intercourse, but discontinued because of a fear that doing so posed a risk to the pregnancy:

I think we both were aware of it, and also I’m so fragile, the whole experience, that we wouldn’t risk it, I think, so we were again very careful.

P14-Cx33 removed

Others made their decision not to have sexual intercourse after they had the pessary fitted: some worried that it would induce labour, some decided that it was just something to avoid (a kind of sacrifice) and others had made a decision to abstain from sexual intercourse at the beginning of pregnancy for fear of inducing a miscarriage, as the following woman describes:

We haven’t . . . I’ve not had sex for. . . well since like 8 weeks pregnant or something.

P12-Cx23

On being fitted with the pessary later in her pregnancy, this participant goes on to say that, although the risk of miscarriage had passed, contemplating sexual intercourse was ‘just too scary. I just can’t . . . I couldn’t do it. I couldn’t . . . yes, it would just be weird I think. Oh no, there’s a big blue pessary in there . . . I’m like, oh, I couldn’t do it, it’s so weird’ (P12-Cx23).

Only one participant said that she had been advised to abstain from sexual intercourse:

I was actually told, because of my risks of preterm labour, and stuff, that I should try and abstain from it. So, from the beginning I have stopped because I didn’t want to risk, you know, going into preterm labour, or anything. So, it wasn’t only . . . it wasn’t the pessary that stopped me from doing that. It was from beforehand even where, you know, midwives, and even research that I’d done myself which said, you know, with a twin pregnancy they do, you know, recommend that you get advice about, you know, sexual activity, and stuff.

P2-Cx28

Some talked about the pessary affecting their or their partner’s experience of sex and a few said that more information would have been helpful, especially if they were trying to maintain their sex lives:

. . . the information does say that you can still have sex. But because it felt so low down I just felt like that wasn’t an option. So for quite a few months it just . . . I think it’s probably mind over matter. But . . . just didn’t want any intercourse at all to be honest. And then as I’ve settled down we have sort of managed to have . . . to do sort of a little bit. But it’s still different. It was still very conscious that there’s a pessary . . . if he knocks it or . . .

P7-Cx34

I think it’s incredibly important, throughout pregnancy we’ve been intimate, the whole way through, there’s not a moment where we haven’t been at all, or not felt like we wanted to, because yeah, I think it’s something that’s important to us, to stay close [. . .] [it] helps you through the harder times as well, when things are difficult. So you just remember that we’re doing this because we really want to do it, and we want it for our babies, but we were really happy before this as well, and going through it all, it helps if you feel close and can support each other.

P13-Cx33

Trial processes

Motivation and commitment are required to enable a trial to work and this commitment has to extend beyond the trial staff. As the following interviewee noted:

. . . it requires a sequence of events that probably . . . than you have to be really, really motivated to make it . . . to deliver the trial [. . .] we manage to do it because we have certain things that pull together, but we also have a number of staff who give their time, that are not paid [. . .] you know, they’re using that time . . . time that they could be doing other things.

CRF1

Much thought had gone into integrating trial protocols so that neither women nor centres had additional work/appointments. In the bigger centres, staff talked of carefully negotiated logistics, involving both the research teams and the clinical areas working together. Many of the RMs identified the successes they had in recruitment to be partly influenced by having the support of clinical staff who were keen to promote research.

There was also evidence that trial centres adapted care pathways to accommodate recruitment. For example, in the centre where RM8 worked, women with dichorionic diamniotic pregnancies were offered an obstetric appointment at 16 weeks rather than at 20 weeks. Behind-the-scenes planning involved high levels of communication and a commitment to making time for the trial in the context of other equally pressing commitments. Commitment to research and evidence-based practice seemed to shape the trial’s success, but access to resources was also crucial:

. . . there’s only two of us who do accredited cervical length scans, so actually that burden falls on people who aren’t actually directly having STOPPIT in their job description or in their job planning. So, there’s that motivation from them and that’s because the unit, as a whole, thinks that research is important and wants to take part [. . .] Then, the people involved in putting the pessaries in, there’s me and . . . all the other doctors who run the twins clinic [. . .] but when they’re in a busy clinic, they still . . . even if they’re over-running in their clinic and there’s a woman waiting for a pessary insertion, they don’t . . . they still come and do it, if that makes sense? They’re still committed to creating time or making space, for those women to have the process.

CRF1

Building a research culture requires understanding of the research process and roles of RMs in that process. Some RMs felt frustration that midwife colleagues misunderstood the role of the RM as ‘sitting up there and doing all research and statistics all day’ (RM6), whereas RMs saw their role as more clinical, dealing with the complexities of talking to people about research and adapting communications for each patient. As is evident from the accounts of women participating, the information and reassurance provided by health-care professionals was influential in decisions to participate and in women’s experiences of the trial process.

Recruitment of women to the trial

In most centres, the recruitment was led by the consultant and the initial introduction to the trial was during the first obstetric antenatal appointment. When recruitment was described as a smooth process, it was frequently reported by midwives to be because of ‘flexible consultants’ (RM7) and good research team communication and organisation [relating to identifying potential participants, accessing resources when required (i.e. scanning rooms) and the means to access clinical staff who could perform the scans and fit the pessary while women were attending antenatal appointments]. With the woman’s permission, a RM contacted a woman by telephone after she had considered participation:

[CO] is the one that approaches. She sees them at the, erm, our twin clinic. When they’re scanned, and they’re diagnosed, [CO] sees them. She speaks to them about the study, . . . gives them the patient information leaflet, then . . . tells them that I’ll follow it up with a phone call. She gives me a date and time, and I then phone, answer any more questions that need to be answered. And then, they, if they want more time to think about it, that’s fine. If they’re happy to go ahead, then I give them the date and time. If they decline, then that’s fine as well.

RM1

In centres where women attended specialised twin clinics, RMs’ accounts suggested that recruiting from a specialist twin pregnancy clinic was more straightforward than trying to identify eligible women across several clinics. In centres with no such facility, analysis suggests that midwives experienced additional workloads, in that more clinicians were involved in recruitment and the RM spent time reminding clinical staff should an eligible woman be attending a clinic. RMs said that recruiting from a specific clinic per week also helped with the logistics of pre-screening potential participants prior to the clinic beginning, making them feel fully prepared with all the information they required to ensure that the process was as smooth as possible. One midwife contrasted STOPPIT 2 with another trial, saying:

. . . you feel like you’re ready . . . rather than getting a phone call saying, ‘can you come down now because this women’s here, we’re going to consent this woman’ . . . and you’re grabbing all your bits and pieces and running down and you’re think, ‘oh, I don’t know anything about her, I’ve not had a chance to look on the computer.’

RM6

Some of the data reflected a high level of concern for women’s feelings throughout the trial and particularly in relation to the type of information given to them in the early stages of pregnancy:

. . . we tried for a little while [to recruit] [. . .] at 12 weeks [. . .] but we felt in the end that that was a bit too early [. . .] Because it is quite a big thing to take in, isn’t it? [. . .] we don’t want them to Dr Google lots of stuff necessarily.

RM8

Despite the care some centres had given to ensuring that women were not inconvenienced by additional clinic attendance and the care given to considering when it was appropriate to discuss the trial, some RMs reported that it was a ‘hard sell’ (RM8) when trying to recruit women. Some noted that women were sometimes reluctant to increase their involvement with the clinical setting and/or found the notion of cervical length measurement and the fitting of the pessary unacceptable:

They absolutely cringe at the thought of the pessary but I would say probably most of those are probably primips. I think it’s an easier sell maybe to multips who have already had babies, probably already had a smear, you know, are used to that kind of gynae stuff, maybe. Whereas like an 18-year-old, first pregnancy, twins, I mean, firstly they’re in utter shock, aren’t they?

RM8

Many of the RMs felt that, as women had a high level of contact with maternity services because of the high-risk status of their pregnancy, some women were reluctant to participate as they were concerned that doing so would increase their level of clinical contact. Often this was expressed in simple terms, for example: ‘[women] don’t necessarily want extra visits’ (RM8). Some centres had taken significant steps to mitigate this issue by reassuring women that involvement in the trial would not incur additional appointments, although appointments may be extended because of screening and pessary fitting, etc.

Analysis suggested that one of the ways that some of the RMs helped to support the recruitment of women was by viewing participation incrementally, starting with screening. Some talked about offering support by stressing the ongoing voluntary nature of participation:

. . . we go through everything, and I say, we will talk about that again, if you have a short cervix, I make the point that we will address that again.

RM7

This particular midwife viewed the flexibility offered by the trial design as an important means by which to meet her own and the women’s emotional needs, as this quotation illustrates:

. . . it feels much more comfortable for me, being involved in this, that we can say – ‘yeah, you can see how you feel, you don’t know how you might feel until it happens, about having a short cervix.’

RM7

Relatedly, when considering issues to do with recruitment and retention of women, the trial was viewed by this midwife, and others, as particularly good in that there was no pressure to ensure continuing participation:

From a target point of view, if anyone agrees to have the initial scan, and then they chose not to be randomised, we still get the accrual for that, so we haven’t got the pressure of the figures, to encourage you.

RM7

The study design, therefore, seemed to support the midwives, who felt concerned about women’s reaction to cervical measurement results and ongoing participation. Cervical measurement was conveyed as providing useful information in itself.

Barriers to recruitment

Analysis has already alluded to why, from the RMs’ perspectives, women might not want to participate, and how both clinic factors and their own communicative practices could support recruitment. Focusing further on perceived barriers, RMs felt that they encountered barriers at the level of language, despite the availability of interpreters (sometimes a family member), understanding of trial processes and of the intervention itself:

I think that quite a lot of the women . . . hear the word research and then they think it’s going to be something that, ‘oh, I couldn’t possibly understand that.’

RM6

It was difficult to explain the concept of internal scanning and to ascertain whether or not women had understood the nature of the procedure. Some midwives felt that women chose not to participate because they had concerns about TVU scanning. One midwife describes the point of explaining the scan to women as the point that women say ‘no, I don’t want it’ (RM6). Others said that they suspected women were concerned about the effect of TVU on the pregnancy:

. . . maybe they worry that having a scan is going to stir things up and maybe cause them to miscarry.

RM8

I think it’s a pretty well-run trial. And, erm, it, it’s probably not the easiest to, erm, approach women on, because of the intimacy things . . . and things like that. But I, I do think, as time goes on, everyone’s gaining experience in it.

RM1

The size of the pessary was also considered a difficult aspect to manage:

I felt that people might be quite unwilling to accept the intervention [. . .] The pessary, it’s the pessary itself, um, I felt was quite big . . . and sort of bulky-looking, and I thought . . . I don’t . . . I don’t know whether to show it to people, whether not to show it to people. Um, did a bit of experimentation with that, er, and, I mean, I was still learning, I’m still deciding. But I think with the majority of people, it’s better not to show it at that initial chat.

RM3

The RMs worked hard to build their own confidence in talking about the pessary. After visiting a centre that was more familiar with the device, RM3 noted that:

. . . it’s interesting because you notice a spike in recruitment. And em, that’s because I was more confident then, telling people about the trial and selling the trial, and selling the . . . the Arabin pessary as a real potential for preventing preterm delivery.

RM3

Supporting women through trial processes

As noted, trial participation depended on several inter-related factors, including clinic logistics, health-care professional commitment and reassuring communication at every stage. Midwives also discussed how they found their own role in STOPPIT 2 enjoyable, as they had increased contact with ‘high-risk’ women over the course of their pregnancies:

. . . you do get continuity, because you’re at the antenatal clinics. So, when you recruit somebody, or you’ve given them some information, you, you see them quite frequently.

RM1

. . . you have more time to spend with women and . . . as . . . in a midwifery role, it’s always rushed and you spend your entire time saying, ‘I’ll be with you in a moment’, and apologising, but it’s nice in research to have, um, to spend the time . . . with the women, um, and have time to talk to them, and because obviously they always . . . they know you’re a midwife, they always talk to you about other issues as well and ask you questions that are midwifery related, um, so it’s nice to have that time to spend with them.

RM2

Overall, RMs’ accounts of communicating the results to women demonstrated that they felt concerned about how women would respond to the news. One way in which they supported women, should they be found to have a shorter cervix, was to prepare them in advance through a carefully planned dialogue, whereby the information that would reassure women was ‘planted’ at an earlier stage of the trial process, linking the meaning of the measurement to the trial, as the following quotation illustrates:

Even those that have been found to have the short cervix, none of them have been particularly fazed about it [. . .] I think the way they’ve been counselled – ‘So, yes, you will have a short cervix by definition for the study but actually if it’s not less than 30, do you know what I mean, we would consider that normal by our normal standards’. However for the study it’s considered short so I think maybe they’re a bit more reassured.

RM8

Health-care professionals found that responding to women and discussing the results of cervical screening measurement involved a high level of emotional engagement and attention to an ongoing awareness of women’s emotional response to the information, as this quotation suggests:

I do it case by case really, see how they react, and how much reassurance they’re going to require.

CM1

Cervical length measurement provided reassurance to the midwives, as well as to the women:

. . . the vibe is after a scan when we have a long cervix, the consultants give a feeling of, yeah, there’s a reassurance, yeah. Although they do clearly say, as well, ‘we wouldn’t expect you to have a very early preterm delivery, but we can’t exclude it.’

RM7

Staff also felt that women sought participation as a means to gain reassurance:

. . . they want the reassurance of knowing that the cervix is alright.

CM1

Like the women, many of the midwives also expressed the notion that taking cervical measurements was a useful screening tool. On the one hand, taking a measurement meant that women could be reassured if they were found to have a long cervix, whereas, on the other hand, women with shorter cervices could be offered ‘serial scans or more regular appointments, or you know, you might be able to do something a little bit more to reassure those ladies who are found to have a short cervix’ (RM4).

There was also a perception that women were curious about their cervical length, but not as keen to be randomised to receive the pessary:

I think a lot of women, possibly, are interested in finding out their cervical length, but may not be quite so interested if they’re randomised to get the pessary.

RM1

Randomisation was identified as a point in the study when RMs felt that they had to exert caution in terms of the information and reassurance they offered women, not least because women with a short cervix would be worried if they were randomised to standard care:

. . . you have to be very careful how you word things with the ladies that don’t get randomised to the pessary, you know, because often they will be quite anxious and you have to reiterate the fact that the reason we’re doing the study is to find out if this pessary is effective.

RM4

Some of the RMs raised concerns about how they imagined women might feel about being randomised to the control group. When asked about how she would feel about a woman with a cervical length of 25 mm being randomised to standard care, one RM said that even if the woman was unconcerned she would feel compelled to check on the progress of the pregnancy:

I’d be thinking, ‘oh my God.’ I’d probably be thinking, ‘yes, I’m definitely going to be ringing you every 4 weeks,’ but you see if she hadn’t gone into the study we would never have known. Even if she did show signs of preterm labour at say 25 weeks, yes, we might discover she’s got a short cervix on speculum but potentially if it wasn’t for the study we wouldn’t know.

RM8

This account suggests that the RM draws on the logic that knowledge can potentially ‘make people extra vigilant if the woman felt something was amiss’ (RM8).

Reassurance could be offered in several ways, including giving direct emotional support:

The second lady who got randomised for control, um, she really wanted the pessary, she said, ‘um, so that wasn’t, you know, that wasn’t very nice.’ I think I gave her a hug and said . . . and made sure she was OK, and said, ‘do you feel OK,’ and she went . . . and she said that yeah, she felt really gutted that she didn’t get it, um, but she was OK and she under . . . and she went, ‘I know, I understand, you know, um . . . that that’s just the way it goes, I understand it’s, um’ . . . I think she, you know, she understood about, like, the randomisation, really, so she was . . . you know, she under . . . I think she said, ‘I understand that you know, it’s not your decision,’ um, so, um . . . yeah, that was OK as well [. . .] I was more worried how she would go home and react to it.

RM2

The RMs anticipated women’s concerns and sought to smooth their experience of the trial and their pregnancy. This required optimising their own strategies, learning how best to provide information, seeking support from the trial team, if necessary, and adjusting their practice accordingly. One RM noted that she altered what she said to women about pessary fitting based on her observations of the fitting procedure, from being ‘quick and easy’ to ‘it can be quite uncomfortable when it’s being fitted’ (RM4).

The RMs’ concerns extended to those who did not continue their participation, especially after the TVU had identified a short cervix, anticipating that the woman would feel guilty should anything happen, something some of the women interviewed noted as a reason to participate. RMs also expressed relief when women with shorter cervices were randomised to the intervention group, as this quotation suggests:

. . . some that have been around 24, 26, who have been randomised to pessary, so it’s kind of almost a relief when I’ve found that, and they have been randomised to pessary, you think, oh well that’s good.

CM1

Patient Advisory Groups

Following several face-to-face and teleconference meetings in December 2014, it was decided that the most efficient and cost-effective way of managing the PPI would be for Dr Andrew Gibson (collaborator) to run one PAG, consisting of approximately eight members and meeting in London every 6–9 months. However, Dr Gibson left the project team (April 2015) to take up a new post and Dr Kath Maguire of the PenCLAHRC PPI team agreed to take on the role.

Dr Maguire determined that she would use the PenCLAHRC group and set up a PAG using the existing members who were located throughout the south-west of England. Dr Maguire recruited six members to the STOPPIT 2 PAG and provided each member with training about trial processes to facilitate discussion of issues identified as being key to a particular stage of the project. For instance, contribution to development of the patient information sheet and consent form, with suggested strategies to improve the acceptability of the intervention. It was also agreed that the group would help develop lay summaries of the research and a plan for dissemination to service users and families. Between the face-to-face meetings, the group received e-mail updates about the progress of the project.

Trial Oversight Group patient and public involvement representation

Jane Denton (a study collaborator) was also the Multiple Births Foundation (MBF) (London, UK) representative on the TMG.

Independent patient and public involvement

The Twins and Multiple Births Association (TAMBA), now known as Twins Trust, provided Mr Keith Reed as a member of the TSC.

Members of our TMG, TAMBA and MBF offered to assist in nominating and identifying individuals for the TSC. Initially, four parents volunteered to join the TSC and each had a lived experience of either preterm birth or twin pregnancy within the last 10 years. It was clear that the commitment and training required was too much for many of the volunteers. Two volunteers declined to participate and after a year one member withdrew. However, Mrs Andrea Hall provided continued support throughout the project as a member of the TSC.

Jane Denton, Keith Reed and Mrs Hall each offered extensive lay and patient involvement experience, acting as the voice of women with experience of twin births. They were involved in the design of the study, including the protocol, patient information sheets and the development of strategies for participant communication.

Outcomes

The PAG met regularly between September 2015 and September 2017, contributing to the development of the participant information documents. [For more information please see NIHR Journals Library, URL: www.journalslibrary.nihr.ac.uk/programmes/hta/130422/#/ (accessed 22 April 2021).] During the process of determining the need for an extension to the study, Dr Maguire informed the TMG that she was no longer able to continue to support the PPI on this project. After moving to Cornwall, Dr Maguire had found it difficult supporting the PAG, but this became even more difficult as she was no longer working with the groups who previously attended workshops and the continuity of the relationship could not be maintained.

Funding for the PAG was re-allocated to Edinburgh and the team determined that a new advisory group was required. The Edinburgh trial team implemented a number of strategies to engage the public and set up a new PAG. A request for volunteers was placed on the Edinburgh Tommy’s Midwives Facebook page [Facebook, Inc., Menlo Park, CA, USA, URL: www.facebook.com/tommysmidwivesedin/ (accessed 23 April 2021)]. TAMBA and the MBF were asked to see if any of their members would be willing to become PAG members. We initially received 13 responses; however, only three women engaged in further conversations and did not want to commit to anything on a regular basis, as they had young families and work commitments.

The STOPPIT 2 group felt that it was important that patients were supported to build confidence, facilitate effective communication and enable involvement to represent a true partnership between research and its participants. PPI is a crucial part of this support to ensure relevant and acceptable research goals. Funding was allocated for PPI from within STOPPIT 2 and from QUIDS [Health Technology Assessment programme (UK) REC 17/WS/0081 ISRCTN41598423, see URL: www.journalslibrary.nihr.ac.uk/programmes/hta/143201/#/ (accessed 23 April 2021)], a trial that was also struggling to have effective PPI representation and so resources were jointly pooled. The STOPPIT 2 trial administrator was asked to be the co-ordinator for the PAG within the Medical Research Council (MRC) Centre for Reproductive Health (Edinburgh, UK), assisted by the QUIDS study administrator. The group was renamed Patient Involvement in Clinical Trials Research (PICTR).

Recruitment for the proposed group took place at the Baby & Toddler Show, SEC Centre, Glasgow, UK, 27 April 2018. A total of 162 people agreed to sign up to become part of the PICTR group. This included 34 men, some of whom were very keen to be involved at all levels, including being part of a grant application. One man informed us that we were the first people to have asked for his opinion about any aspect of pregnancy and childbirth, as all the focus had been on his partner. The group seemed keen to share the knowledge they had gained from their own experience of pregnancy and birth, and were pleased to be asked to participate in an advisory capacity.

Initial contact was made with the PICTR group by e-mail and further correspondence was conducted in this way. This included asking for member input concerning various research questions for the study plus other projects. Regular newsletters were also sent. To engage with members of the group in person, it was decided that all the volunteers who had become part of the PICTR group should be invited to a meeting in Edinburgh Royal Infirmary, Edinburgh, UK, on Friday 17 August 2018 to ‘meet the researchers and their teams’.

From the substantial list of names, of which around 20% lived in, or around, Edinburgh, we had one definite acceptance. It was clear that this format for discussing research is not suitable for parents with small children. We were asking parents for a commitment at one of the busiest times of their lives (i.e. bringing up small babies and children). This initial meeting was organised for a Friday afternoon within a hospital setting. Owing to the poor response to our invitation, the proposed meeting was cancelled and we discussed further what other opportunities could we possibly offer parents with small children to facilitate easy attendance at any kind of organised event.

It was decided to hold a children’s party, with refreshments and children’s entertainment laid on, so that the study researchers could talk to parents directly. We also felt that this kind of event was giving something back to parents in return for their involvement in the PICTR group. The party was organised to take place on a Saturday afternoon at a local church hall with excellent facilities.

Patient Involvement in Clinical Trials Research party (Edinburgh, January 2019)

The invitation was extended to all members of the PICTR group and initial acceptance was good (14 responses). Actual attendance totalled 10 parents bringing seven children, who varied in age from 7 months to 7 years. Members of the research team attending were two professors, with a PhD student and medical student also attending. It was possible for each researcher to speak individually to the PICTR members.

All research staff involved in reproductive health were also invited to attend, but their acceptance was limited. It appears there is a disjunction between the ability (or willingness) of both parents and staff to attend arranged events, depending on the venue and timing. It became clear that the informality of a children’s party is a much more sympathetic space in which to discuss various reproductive health issues. One mother told us that she did not wish to come to a meeting to sit in a circle while being expected to speak, as this would be too intimidating. A meeting, possibly including presentations, on a weekday within a hospital setting is perhaps not ideal and too formal (although this suited the research staff). Parking at the hospital is also limited and expensive.

The parents who did attend the party were very keen to become involved in advising on our reproductive health research. We subsequently wrote to each parent individually with information about the ongoing trials in the Centre for Reproductive Health. We consulted with the group on several issues concerned with our research.

Overall, our resource has identified PPI representatives to act:

• on the TSC for Pravastatin For The Treatment Of Preterm Labour (PIPIN) study

• as a co-applicant on a grant to investigate whether or not delaying removal of long-acting reversible contraception in obese women may reduce weight prior to pregnancy

• on the Trial Oversight Committee for the CASSAVA trial.

Strengths

The strengths of STOPPIT 2 are that it is a large ‘real-world’ study and used a population threshold to define a short cervix group of women randomised to pessary of placebo. All of the ‘short cervix’ pessary studies of twins (including STOPPIT 2) were open label; however, to the best of our knowledge, STOPPIT 2 had a larger number of events than any of the other studies published to date. 11–14 Only the Goya et al. 13 and Liem et al. 12 studies (the two smallest studies) suggested benefit in the short cervix subgroup. Adherence was reasonably good. Ninety-two per cent of women in the intervention group had the pessary successfully inserted, although 10.4% of all women allocated to the intervention group subsequently asked for the pessary to be removed prior to the scheduled date, and in a further 5.2% the pessary fell out.

Limitations

The limitations of STOPPIT 2 are that it was underpowered to identify smaller treatment effects than originally anticipated (i.e. 40% reduction in preterm birth before 34 weeks’ gestation) and/or small subgroup effects. An additional limitation is the open-label design of this study. Although a single- or double-masked design would have been extremely challenging and our outcome is unlikely to be influenced by ascertainment bias, it is possible that the open-label approach led to unintended and undocumented differences in treatment between the two groups.

Implications for health care

• The Arabin pessary should not be used in women with twin pregnancy for the prevention of preterm labour.

• A ‘long’ cervical length measurement should not be used to reassure women with twin pregnancy about their risk of SPB.

Future research implications

• Women with twin pregnancies have high rates of preterm birth (19.4%), with many having at least one serious neonatal outcome. Further work is required to find effective therapies.

• Individual patient data meta-analysis of STOPPIT 2 and other published trials should be conducted to determine if there are (as yet unidentified) subgroups of women with twin pregnancy who benefit from Arabin pessary insertion.