Balancing key stakeholder priorities and ethical principles to design a trial comparing intervention or expectant management for early-onset selective fetal growth restriction in monochorionic twin pregnancy: FERN qualitative study

Study design

The FERN study9 involved three work packages, including (1) prospective UK multicentre observational study, (2) qualitative study, (3) international survey10 and consensus meeting. This paper presents the findings of the qualitative work package 2 phase of the study. The aim of work package 2 was to explore parents’ and clinicians’ perspectives on how the future clinical trial should be designed (including recruitment and consent approaches and the design of research materials), the factors that influence parents’ and clinicians’ decision-making and the acceptability of a future clinical trial.

Following ethics approval (REC reference: 20/SW/0156), we conducted online or telephone interviews with English-speaking women and their partners (where applicable) in the UK who were experiencing early-onset sFGR in MC twin pregnancy (or with experience in the last 3 years), and English-speaking clinical staff involved in the management of MC twin pregnancies in the UK and Europe, to explore their views on the feasibility, acceptability and design of a proposed RCT, with one ‘watch and wait’ expectant management arm, and two intervention arms: (1) selective termination of the sFGR twin and (2) laser treatment, for early-onset sFGR in MC twin pregnancy. Interviews were conducted between September 2022 and March 2023.

We used previous research11 12 to develop the parent, partner and clinician participant information sheets (PIS) (see online supplemental files 1-3), while ongoing study findings were used to develop parent and clinician interview topic guides (see online supplemental files 4-5) as part of an iterative process. Interview topic guides included questions on the experience of management of MC twin pregnancies that were complicated by sFGR, decision-making processes, proposed trial design, information materials, trial acceptability, willingness to randomise/be randomised, prioritised outcomes and clinician training needs. The consolidated criteria for reporting qualitative research (COREQ) checklist13 was used to aid reporting (see online supplemental file 6).

Patient and Public Involvement and Engagement (PPIE)

Our PPIE members include six coapplicants: Michelle Watson, Jessica Mendoza, Danielle Harding and Joel Marsden (two with personal experience of sFGR in MC twin pregnancy) and Natasha Fenwick and Shauna Leven from Twins Trust (https://twinstrust.org), which is a registered charity who support parents through every milestone of their journey with twins, triplets or more. The PPIE members were involved in the grant development, design, recruitment for, conduct, progress and/or findings of the FERN study; and/or as members of the study oversight/steering committee; and/or attended the work package 3 Key Stakeholder meeting in London (3 July 2023) and/or reviewing and providing input on draft research information materials for this qualitative study and reviewing drafts of this manuscript.

Recruitment and sampling procedure

Based on previous qualitative feasibility studies,11 14 we anticipated that we would need to interview 15–25 parents and clinicians to reach information power,15 which is the point at which data address the study aims; sample variance13 (eg, parents offered expectant management or intervention, bereaved and non-bereaved parents and clinicians in favour of intervention and expectant management); our reflexive and interpretive approach to theory and analysis16 17 and sufficient quality of interview dialogue.15 We planned to hold additional focus groups if divergence in opinion was observed in interview data, but these were not required. Parents were recruited via work package 1 hospital sites and social media (Facebook and Twitter, with the support of Twins Trust).

Eligibility screening and conduct

Research midwives at hospitals (n=5/17) involved in work package 1 checked eligibility and approached parents with FERN study information, which included details of the qualitative study. MP (female, Social Scientist) or TKM (female, Social Scientist) contacted parents who had registered interest to participate in an interview to arrange a convenient time for an online or telephone interview (according to their preference). For social media recruitment, MP and TKM responded to parents’ expressions of interest to take part in an interview in sequential order. Once eligibility had been confirmed, parents were emailed a copy of the Parent PIS which explained what would happen during their interview (see online supplemental file 7). Once parents confirmed their continued interest, they were then sent a proposed trial PIS and the core outcome measures list (see online supplemental file 8), derived from a review of the literature and Core Outcome set for this population.18

TKM and MP contacted work package 1 site clinicians and attendees at the International Society of Ultrasound in Obstetrics and Gynaecology World Congress 2022 to invite them to take part in an interview. Clinicians who expressed an interest in taking part were sent the Practitioner PIS, proposed Inclusion and Exclusion Criteria (see online supplemental file 9), and the same outcome measures list that was sent to parents before interview.

TKM and MP facilitated parent and clinician interviews using the topic guides.19 Interviews stopped when information power15 was reached. Parents then received a £30 Amazon voucher via email to compensate them for their time.

Analysis

MP and TKM conducted the analysis with oversight from KW (female, Social Scientist) and RA (male, Ethicist). Digital audio recordings of interviews were transcribed verbatim by a professional transcription company (UK Transcription, Brighton, UK). Transcripts were checked for accuracy and identifiable information were anonymised before being imported into NVivo V. 12 Plus software,20 which was used to assist the organisation and coding of data. Reflexive thematic analysis was broadly interpretive and inductive.17 MP, TKM and KW met regularly to discuss interpretation and develop the coding framework. Outcome measures prioritised as being most important were given a score of 13, second most important a score of 12, third most important a score of 11 and so on down to a score of one. Outcomes were then ranked. Findings were considered against the Adapted Theoretical Framework of Acceptability (ATFA) for paediatric trials11 21 and Principles of Biomedical Ethics22 23 (in particular, autonomy, justice, beneficence and non-maleficence) and synthesised using a symbiotic empirical ethics approach24 to produce normative conclusions (eg, should a randomised controlled trial be conducted?)

Participant recruitment

Seventy-three parents registered interest in taking part in an interview (figure 1). Recruitment was closed at the point of information power.15 Nineteen parents (14 mothers, 5 partners representing 28 babies and 14 families) took part in an online (n=11) or telephone (n=8) interview. Characteristics of the 19 parents and their pregnancy are shown in table 2.

Figure 1

Participant recruitment. MC, monochorionic.

Sixty-seven clinicians were invited to an interview. Recruitment was closed when 14 clinicians had taken part in an online (n=10) or telephone (n=4) interview (see table 3). Interviews with clinicians lasted between 53 and 83 min (mean=62 min), whilst parent interviews lasted between 47 and 106 min (mean 68 min).

Six interlinked themes will now be presented, which highlight the importance that parents and clinicians place on answering the research question and considering the practical and ethical challenges of conducting the proposed clinical trial.

An important question to answer

Parents and clinicians indicated that the proposed trial would answer an important research question to guide clinical practice and discussions with parents:

I think it’s great that you’re doing something to help parents make decisions … there wasn’t really a lot of information … that me and my wife could find. (P14, partner, social media)

(Parents require) figures, so percentiles … because the science that is available at the moment is a bit contradicting and, in some aspects, also not always fitting to current practice. (C14, doctor)

Participants spoke of the need for evidence to alleviate the psychological distress that comes with making the ‘traumatic’ (P2, bereaved mother, social media), ‘impossible decisions’ (P1, bereaved partner, social media) about whether to go down the expectant management or intervention route, and for clinicians to confidently counsel parents about which route to take (C3, doctor). Clinicians stated that they ‘sometimes counsel too pessimistic’ and are ‘ashamed’ that they cannot provide parents with the right information specific to their pregnancy:

I sometimes am ashamed of that I have to say to parents, in this modern, developed world in which medicine can treat, let’s say, metastasis of melanomas with immune therapy, we cannot predict what the outcome is of their specific pregnancy complication. (C14, doctor)

Pregnancy management decision-making as ‘traumatic’ when outcomes include death or serious disability of one or both twins

We began parent interviews exploring clinical practice and the pregnancy management route experienced. All parents had their pregnancy managed expectantly. Nine parents (six families) had considered selective termination and laser treatment options, yet described how their decision to decline intervention was informed by clinicians explaining how there were positive indicators of life (e.g., blood flow), which gave parents hope for the survival of both babies.

She (clinician) said, You can terminate little twin and focus on just having one baby, but because she had also said that all the internal blood flow was normal, that was an option which we said we weren’t going to take. (P12/13 joint interview, pregnant mother, site)

I don’t see why if we’ve got this far we can’t get further (…) I only need to make it to 28 weeks for them to be able to be born, even if it is very prematurely … I kind of felt like I didn’t want to do too much action. I felt protective of the pregnancy in a sense I didn’t want to do anything to upset it. (P11, mother, social media).

Although clinicians believed that parents should lead decision-making about whether to intervene, it was clearly evident in parent interviews (and compounded by whether their twins lived or died) that such decisions were traumatising, causing much distress and burden. They spoke of feeling disconnected from their surroundings, with bereaved parents stating ‘that some people end up being diagnosed with PTSD after having to make difficult decisions around their babies’ (P8, bereaved mother, social media). One mother who went on to have two healthy babies described her smaller twin as having ‘a life that wouldn’t need to be lost’ (P9, mother, social media).

We went away for a couple of hours, sat in the car … crying … because we had been presented with these options and had no idea, and no one seemed to have any idea what was the best one to do. That was quite traumatic. (P2, bereaved mother, social media)

I feel like when it hit 24 weeks and I was being asked to make decisions about whether to keep a baby or not, I almost wasn’t able to clearly… I felt very separate from my…. almost disassociated from myself. I felt very separate from what was going on just because I’d been so detached. It was a really weird experience. (P11, mother, social media)

Clinicians highlighted how the timing of intervention is difficult due to the changing and unpredictable nature of sFGR in MC twin pregnancy:

We can only offer selective reduction (termination) up to a certain gestation, usually 24 weeks. It’s a tricky one, because you think, ‘Well, in two weeks’ time, it is highly likely that the ductus venosus is going to become abnormal … By then, I may not have the option of offering them a selective reduction. And then, what if it dies?’ … And so, you can end up in this very difficult situation. If you don’t offer it early enough … but then you end up in this situation where you’re 24 to 25 weeks, you can’t offer a selective reduction, because technically, it’s not possible … I think that is the really difficult decision for them, because then it is an impact on their life or the life of that child. I mean, accepting death is so difficult, but looking after a child with a disability is a different ballgame altogether. So, it is those things that I think are very difficult for parents to weigh up. (C8, Doctor)

As well as the long-term impact of having conversations about selective termination, clinicians and parents were worried about the ‘devastating’ scenario where ‘it’s possible that one could’ve made a wrong decision … based on a worst-case scenario’ (P17, mother, site).

The challenge of diagnosing an unpredictable condition in the context of a trial

Clinicians emphasised that every pregnancy is different and that there are many factors that determine their decision-making to recommend expectant management or active intervention. Clinicians’ decision-making was described as being ‘on a case-by-case basis’ (C12, doctor) and informed by multiple factors, including the severity and type of sFGR (as determined by Dopplers), as shown in Gratacós et al (2007) three types classification of sFGR25 26 (see table 4).

Other factors that determine clinicians’ decision-making to recommend expectant management, active intervention or early delivery were: the gestational age at diagnosis; the ‘size of the’ affected twin (C12, doctor); ‘the speed of growth decreases’ (C14, doctor) ‘where the placenta is’/placental vessels are (C14, doctor)/‘accessibility to the smaller twin’ (C7, doctor); and the perceived acceptability of intervention for parents. Furthermore, intervention in sFGR pregnancy ‘really varies according to … the centre where the patient is seen’ (C1, doctor), ‘the culture … (and) general consensus … of the population (area or country) and how they see things’ (C10, doctor), the knowledge and experience of, and how the individual clinician articulates the benefits and risks of pregnancy management options. These factors also contributed to clinicians’ suggestions for additions or amendments to the proposed inclusion and exclusion criteria (see online supplemental file 10).

While there was some variance in clinician’s preferred management approach, most clinicians stated they would not discuss intervention options early in the pregnancy, particularly for type I sFGR, and would ‘see again always … in one week’s time, just to see how things evolve’ (C2, doctor).

Parents described being told that their condition could correct itself or change from a more severe type to one with a more favourable outcome:

Like my consultant said, it’s a weekly thing. Today’s appointment is good, but we don’t know if it’s going to get worse or better next week. So, that was made very clear. (P17, pregnant mother, site)

When parents reflected on their discussions in these situations, they said they had relied on the clinician’s advice. However, this posed difficulties for parents who were faced with contradictory advice from different clinicians:

He (consultant) reached out to one guy in Germany and one guy in the USA, and he said that one of them went, ‘Why on earth would you intervene? There’s no proof that this works. Why would you do that?’ The other guy went, ‘Why on earth wouldn’t you do it? There’s something available to you. Why wouldn’t you intervene if you think there’s a big problem?’ (P1, bereaved partner, social media).

Parents require clear information about risks and potential outcomes to make informed decisions as to whether to take part in the proposed RCT

When asked what would make an RCT like FERN more acceptable, parents said that they would want reassurance from their consultant that (hypothetically) taking part would ‘have no greater adverse outcome if we do this than if we didn’t do it?’ (P1, bereaved partner, social media) ‘because it’s such a … not invasive. Invasive is the wrong word. But it’s not, like, an observational study. It could actually affect what happened’ (P2, bereaved mother, social media).

Parents would require information and statistics about potential outcomes for each trial arm to decide whether or not to take part:

I don’t even know whether this exists, but potentially, statistics of how successful expectant management would be, how successful interventions are. So, the science behind each of the choices in simple numbers, so that it’s in black and white, easy to see how positive each of the outcomes are. (P4, mother (one twin with neurodevelopmental disability), social media).

Nevertheless, most parents and clinicians agreed that selective termination as a trial arm was not acceptable. While parents said that a trial comparing laser treatment with expectant management would be more acceptable because ‘you weren’t necessarily selectively choosing which baby would have to be terminated’ (P9, mother, social media), some clinicians said that they were ‘not a fan of laser because I think we have to be honest that we don’t have any pathophysiological argument to say that it will improve the outcome’ (C2, doctor). Another clinician said that it would be unethical not to provide parents with the evidence about outcomes of laser treatment in sFGR:

It would be unethical … (to) not provide parents with the … current evidence (that) shows that they’re more likely to take a baby home if they have a selective reduction, compared to if they had a laser. (C8, doctor)

The proposed RCT was viewed by parents and clinicians as ‘ethically murky’

While recognising the importance of answering the question about which management option is most effective for sFGR in MC twin pregnancy, our findings suggest that conducting an RCT comparing expectant management to active intervention would not be acceptable to parents, who view the proposed study as ‘ethically murky’ (P3, pregnant mother, site). Most parents clearly stated that they would not participate in an RCT with active intervention and expectant management as trial arms because they would not want the fate of their babies’ lives being left to a randomisation process:

If we were approached, I would be a straight ‘No!’ straightaway… Just reading the treatment sections (in the proposed participant information sheet), like the options of treatments, the termination treatments were definitely, immediately, I was like, ‘Okay, no!’ It was an immediate ‘No!’ That was it!. (P18, partner of pregnant mother, site)

Parents were clear that they would drop out of the proposed trial if they were randomised to a trial arm that they were not comfortable with, particularly if their pregnancy course was type I:

If it were us, I would’ve gone with it but if I was put into a category that I didn’t agree with I would’ve pulled out. Going back to the whole severity level of the pregnancy on a scale of one to ten, then being randomly put into category B which is termination, if you were on the less severe end of the spectrum you might look at that and think, ‘I’m not happy with that!’. (P6, partner, social media).

Indeed, clinicians raised ethical concerns about randomising women to a trial arm, especially selective termination, that might not be appropriate to their individual case. Decision-making was informed by the severity and type of sFGR (see table 4):

It very much depends on the type of selective fetal growth restriction, whether it’s Type I or Type II … or Type III … We know that outcome for Type I is good without intervention. Outcome for Type II with abnormal ductus venosus is bad without intervention … Type II … deteriorates much faster and in a predictable way. And Type III can go on for a long, long time, but it’s an unpredictable, sudden loss. (C8, doctor)

‘Live birth’, ‘childhood disability’ and ‘neurodevelopmental impairment’ were the most important outcome measures for parents and clinicians

Parents and clinicians were asked to consider a list of 13 potential outcomes sent prior to interview and were asked if there were any additional outcomes that they felt were missing.

The ranking of outcomes prioritised by parents can be seen in table 5, and the ranking of outcomes prioritised by clinicians in table 6. Outcome measures prioritised as being most important were given a score of 13 and those ranked least important were given a score of one.

The most important outcome for the proposed trial for parents and clinicians was live birth which, for clinicians, meant the survival of one of the twins and for parents meant the survival of both twins, even if it meant that the twins had neurodevelopmental impairment:

I remember saying I would have preferred to have two alive children with … a bit of cognitive impairment … (or) disabilities than two dead ones or one dead one (P11, mother, social media).

Neurodevelopmental impairment and disability, which were the next most important outcome measures for parents and clinicians, had been presented to the participants as two separate measures. However, participants spoke of how these outcomes, together with the child’s quality of life (which was an important outcome measure to parents), overlap and could be measured together:

They (outcomes) all come under the same umbrella for us. Because as a result of his neurological impairment, he has got childhood disabilities and then that affects his quality of life. (P4, mother (one twin with neurodevelopmental disability), social media)

Parental stress was ranked as one of the least important outcomes (weighted 11 most important for parents and clinicians). Participants spoke of how stressful going through a high-risk pregnancy was for families, and as stress is ‘almost like a given’ (P15, pregnant mother), they would not consider this an important outcome to be measured in the proposed RCT.

Almost half of clinicians, unprompted, said that ‘neonatal morbidity’ meaning ‘all of the complications that can arise in the neonate, while the neonate did not actually die’ (C6, doctor) was a missing outcome, with one clinician saying that that neonatal morbidity is an outcome that is included in other UK sFGR and neonatal trials.

In other UK trials of sFGR and neonatal trials, there is actually quite a long list of neonatal morbidity outcomes or indicators that should be included in this list. (C1, doctor).

Suggestions for longer term (and missing) outcome measures were made by both clinicians and parents, such as, importantly, including ‘some sort of measurement of parental experience or any regret or anything to do with their decision-making’ (C3, doctor) for trial participation, parent emotional well-being and living with the choices made (e.g., post-traumatic stress disorder and suicide risk), as demonstrated by this mother’s powerful question:

Did they (parents) survive emotionally … after the decisions … to terminate one of these kids? … (Did) parents … go and kill themselves because they have made the wrong decision? How the hell do you, as a mother, cope? (P16, pregnant mother, site)

Clinicians spoke of how ‘parents will always remember that they were offered a termination’ (C8, doctor) and ‘are (still) traumatised by mentioning the option of cord occlusion’ when their child is followed up at age 8 years:

It is so difficult, and that still, at the age of eight years, they (parents) look at the twins and they think frequently about one of them that they had, that they could end up in a situation that they had chosen cord occlusion …. (C14, doctor)

The proposed RCT is ‘like mission impossible’ and an alternative study design is required

After considering the proposed trial, participants stated that the FERN study would not be acceptable nor practical to conduct ‘in a randomised fashion’ (C2, doctor) as the risk of distress and burden for parents and harm or death to one or both twins would be too great. Applying our findings to the Principles of Biomedical Ethics22 23 (see online supplemental file 11) and ATFA for paediatric trials11 21 (see online supplemental file 12) and synthesising them using a symbiotic empirical ethics approach24 clearly support our findings that the proposed RCT would not be ethical or acceptable for clinicians or parents. As one parent said: ‘You can’t be ethical basically, I don’t think … It is almost like the mission impossible and you just need to find a way to kind of … There will be damage basically, you can’t avoid it, there is no way, there is no other way’ (P18, partner, site).

Some parents and clinicians suggested consideration should be given to other study designs that are more acceptable and still scientifically valid. One parent asked, ‘Is there another way of doing it? … Perhaps, for example, could those who are already going to have these managements, then base it on that, instead of selecting at random?’ (P5, mother, social media). Cohort studies were proposed by clinicians as an option ‘where patients are counselled in a similar way and then depending on what’s technically possible and also on patient preferences, you document an outcome in a uniform way. I think that’s the only way to do that’ (C2, doctor).

Conclusion

Our findings have shown that parents and clinicians do not consider an RCT comparing active intervention versus expectant management to be acceptable or ethical for the management of MC twin pregnancies complicated by sFGR. Drawing on findings from the wider FERN study, as well as the barriers identified in this study to both recruitment and retention, alternative study designs such as an international multicentre observational cohort study or propensity score matching should be considered to address this important research question. Parents value clear information about potential risks and outcomes to make better informed decisions and clinicians wish to be in a better position to counsel parents appropriately. As we have shown, care should be taken when counselling parents as the impact of such clinical discussions can have long-lasting effects on parents, regardless of outcome.

Strengths and limitations of this study

This study provides insight into how parents and clinicians would respond to being invited to participate in an RCT investigating active intervention versus expectant management of MC twin pregnancies complicated by sFGR. The primary strength of the study is the recruitment of parents whose experiences varied in terms of pregnancy outcomes. The sample included bereaved families who had lost both twins, those who had two healthy twins, a case where the co-twin had neurodevelopmental impairment and women who were currently experiencing an MC pregnancy complicated by sFGR and their partners. This last group, in particular, provided insight into how women and their partners might respond to being asked to participate in a definitive trial. Another strength of this study was the involvement of an ethicist in the analysis of findings. Several factors that would help with the design of a future study that does not include randomisation were identified. Our findings can help clinicians reflect on how best to carry out pregnancy management conversations with women to ensure that they are aware of the risks and benefits of each option.

One of the study limitations was that only parents who had their pregnancy managed expectantly took part in an interview (although six families had been offered active intervention). Thus, we cannot conclude that our results accurately reflect the views of women who experienced selective termination or laser treatment intervention. The sample consisted primarily of families of White British ethnicity and all interviews were conducted with English-speaking participants. The views, experiences and understanding of the decision-making parents are asked to make should be explored with families via an interpreter. Although research samples should always reflect the diversity in the population studied, this is particularly important in this study where parents will have to make decisions that include the termination of one of their twins. As demonstrated by our findings, pro-life or pro-choice views on selective termination can be influenced by cultural and religious backgrounds.36 37 Interviews were conducted over the telephone or online which may have impacted on possible eye contact or development of rapport. However, our previous research and other studies have reported that telephone interviews are preferred over face-to face interviews when discussing delicate topics or balancing childcare responsibilities.38