Mifepristone and misoprostol versus placebo and misoprostol for resolution of miscarriage in women diagnosed with missed miscarriage: the MifeMiso RCT

Existing evidence

We conducted a systematic review of trials investigating the use of mifepristone and misoprostol in women with miscarriage to gain a better understanding of studied interventions, outcomes and resolution rates. These studies are listed in Table 1. In addition to the publication by Stockheim et al. ,17 one more trial was identified,18 which compared MifeMiso combination with misoprostol alone in the missed miscarriage population. Together, the two trials included a total of 242 patients treated with MifeMiso or misoprostol alone. Meta-analysis of the results showed no significant difference between the two trial groups for outcome of resolution of miscarriage [risk ratio (RR) 0.97, 95% confidence interval (CI) 0.82 to 1.13] (Figure 1). However, given the imprecision that inevitably accompanies such small sample sizes, it is not possible to draw any firm inferences.

FIGURE 1.

Forest plot of completed trials comparing MifeMiso combination with misoprostol alone for the medical management of missed miscarriage, for the outcome of resolution of miscarriage.

Primary objective

The primary objective of the MifeMiso trial is to test the hypothesis that treatment with mifepristone plus misoprostol is superior to misoprostol alone for the resolution of miscarriage within 7 days in women diagnosed with missed miscarriage by pelvic ultrasound scan in the first 14 weeks of pregnancy.

Key secondary objective

To test the hypothesis that the addition of mifepristone reduces the need for surgical intervention to resolve the miscarriage up to discharge from early pregnancy unit (EPU).

Other secondary objectives

• To evaluate if the addition of mifepristone reduces the need for further doses of misoprostol.

• To evaluate if the addition of mifepristone improves other clinical outcomes, including surgical intervention up to and including 7 days post randomisation and after 7 days post randomisation, duration of bleeding, infection, negative pregnancy test at 21 days post randomisation, time from randomisation to discharge from EPU care, side effects and complications.

• To evaluate if the addition of mifepristone improves patient satisfaction.

• To assess the cost-effectiveness of the combination of mifepristone and misoprostol in the medical management of missed miscarriage.

Sequence generation and minimisation

Participants were randomised online via a secure internet facility. This third party independent ITMS was designed, developed and delivered by MedSciNet^® (London, UK) according to standards of the International Organisation for Standardisation (ISO) 27000 and the requirements of the US Food and Drug Administration CFR21:11.

Participants were randomised to receive mifepristone or placebo in a 1 : 1 ratio. A ‘minimisation’ procedure using a computer-based algorithm was used to avoid chance imbalances in important stratification variables. The stratification variables used for minimisation are listed below:

• Maternal age (< 30 or ≥ 30 years). The aim was to achieve a balance in the number of participants from younger and older age groups. However, there is no known biological reason why the intervention should lead to differing efficacy in women aged < 30 or ≥ 30 years.

• Body mass index (BMI) (< 35 or ≥ 35 kg/m²). It was postulated that the pharmacokinetics and metabolism of mifepristone and misoprostol may be different in women with a BMI that is < 35 or ≥ 35 kg/m².

• Previous parity (nulliparous, parous women). It was postulated that parous women may be able to complete their missed miscarriage at a higher resolution rate than nulliparous women. The aim was, therefore, to achieve balance in the participants according to their previous parity.

• Gestational age (< 70 or ≥ 70 days). It was postulated that women with a gestational age of < 70 days may have a higher resolution rate than women with a gestational age of ≥ 70 days.

• Amount of bleeding [Pictorial Blood loss Assessment Chart (PBAC) score: ≤ 2 or ≥ 3]. It was proposed that women with a PBAC score of ≥ 3 may achieve a higher resolution rate than those with a score of ≤ 2, as an increased amount of bleeding would suggest that the miscarriage was more likely to resolve with the allocated intervention.

• Randomising centre. It was hypothesised that different centres may respond to patients with different ultrasonography findings in different ways. For example, some centres may have a quicker recourse to miscarriage than others. Therefore, the aim was to achieve balance in the number of participants randomised to mifepristone and misoprostol and placebo and misoprostol at each recruiting centre.

Allocation

When all the eligibility criteria and baseline data items were entered online, the ITMS generated a trial number that took into account the minimisation variables recorded for the individual and was linked to a specific trial intervention pack. The pack number was revealed via e-mail to the local principal investigator (PI), the relevant trial pharmacist (see Blinding) and the research nurse or midwife performing the randomisation. The trial intervention pack was dispensed to the patient by the clinical trial pharmacist at the randomising hospital. Each trial intervention pack contained either mifepristone or an identical-looking placebo tablet.

Mifepristone tablets

The IMP was mifepristone at a single dose of 200 mg, taken following randomisation. Mifepristone was based on licensed and commercially available Mifegyne (Mifepristone) 200-mg tablets.

Placebo tablets

Placebo tablets were oral tablets, composed of Pearlitol 200SD Dev, Microcrystalline Cellulose NF, Ph. Eur.; DAB, Crospovidone, NF/Ph. Eur.; Magnesium Stearate, Ph. Eur.; and Quinoline Yellow produced in the same form as the IMP, and identical in colour, shape and weight, for use in the control group of the MifeMiso trial. The dose, route and timing of administration were also identical to those in the active mifepristone group of the study.

Dose

The dose of 200 mg of mifepristone was chosen as it is the most commonly used dose for the medical management of miscarriage when used with misoprostol and it is the most commonly studied dose in published trials investigating its efficacy in the medical management of miscarriage.

Concomitant non-trial treatments

Concomitant therapy was provided at the discretion of the care-providing clinicians, and all concomitant treatment and medications were documented via the ITMS. Other than identified contraindicated drugs (see Participants), the initiation of treatment for another indication did not necessitate withdrawal from the MifeMiso trial.

Compliance monitoring

The dispensing of the MifeMiso trial drug was recorded in the pharmacy drug accountability log. Ingestion of the drug was observed by a health-care professional (HCP) and documented in the patient’s notes. Compliance to trial treatment was defined as taking the allocated mifepristone or placebo tablet on day 0 (day of randomisation) and subsequently misoprostol on day 2 unless the gestational sac had passed before the scheduled time for misoprostol; in the latter case, the patient was deemed to be compliant to the trial medication as long as the allocated mifepristone or placebo tablet was taken on day 0.

Primary outcome

Failure to spontaneously pass the gestational sac within 7 days after randomisation.

Secondary outcomes

Key secondary outcome: surgical intervention to resolve the miscarriage (collected up to discharge from EPU care).

Additional secondary outcomes are as follows (as a proportion of those randomised unless stated):

• Surgical intervention to resolve the miscarriage up to and including day 7 post randomisation.

• Surgical intervention to resolve the miscarriage after day 7 post randomisation to discharge from EPU care.

• Need for further doses of misoprostol up to day 7 post randomisation.

• Need for further doses of misoprostol up to discharge from EPU care.

• Overall patient satisfaction score [measured using the Client Satisfaction Questionnaire (CSQ-8) questionnaire and collected at discharge from EPU care].

• Patient quality of life (index value and overall health status measured using the EuroQol-5 Dimensions, five-level version (EQ-5D-5L), questionnaire and collected on date of randomisation, day 6–7 post randomisation or day of follow-up ultrasound scan if different from day 6–7 and day 212 days post randomisation. If a woman obtains an initial positive pregnancy test result at day 21 ± 2 days post randomisation, then a further EQ-5D-5L questionnaire is collected (at discharge from EPU care).

• Duration of bleeding reported by woman (days) (collected up to discharge from EPU care).

• Diagnosis of infection associated with miscarriage requiring outpatient antibiotic treatment (collected up to discharge from EPU care).

• Diagnosis of infection associated with miscarriage requiring inpatient antibiotic treatment (collected up to discharge from EPU care).

• Negative pregnancy test result 21 days (± 2 days) after randomisation.

• Time from randomisation to discharge from EPU care (described using summary statistics only).

Safety outcomes:

• blood transfusion required (collected up to discharge from EPU care)

• side effects (collected up to discharge from EPU care)

• death (collected up to discharge from EPU care)

• any serious complications (collected up to discharge from EPU care).

Mixed-methods evaluation resource use outcomes

These are detailed in Chapter 4.

Resource use outcomes

These are detailed in Chapter 5.

Future outcomes

Women’s consent was sought for future evaluation of themselves and any subsequent pregnancies using their health records. Although long-term follow-up was outside the scope of this trial, we plan to conduct further studies on outcomes such as subsequent successful pregnancies post-miscarriage resolution.

Outcome generation

Details of how outcome measures were generated are given in Table 2. The ITMS was utilised to capture baseline and outcome data, and to maintain an audit trail. Relevant trial data were transcribed directly into the ITMS. Source data comprised the research clinic notes, hospital notes, self-reports, the EQ-5D-5L questionnaire completed at 21 days post randomisation and subsequently at discharge (for women who obtained an initial positive pregnancy test result), the patient satisfaction survey completed at discharge, and the transcript of the in-depth qualitative interviews. All case report forms are provided in the additional documentation (see Report Supplementary Material 4).

Definition of the end of the trial

The interventional phase of the trial ended when the last woman recruited had taken her last dose of the trial treatment. The observational phase of the trial to assess clinical outcomes ceased and the primary analysis was scheduled to occur when the final outcome of the last woman recruited had been completed and data had been entered onto the database and validated as being ready for analysis.

Notes on adverse events and serious adverse events

All the trial participants were asked to report any hospitalisations, consultations with other medical practitioners, disability, incapacity or any other AEs to their local research team; if the local study nurse or midwife was unavailable for any reason, the participants were able to report the events to the trial manager via telephone at any time. Moreover, at the time of each outcome assessment, investigators and research nurses and midwives at each study centre proactively asked each participant about any AEs in the preceding weeks. AEs were assessed by clinical investigators, further reported as appropriate, and recorded on the ITMS.

Serious AEs and serious adverse reactions (SARs) were recorded on a purpose-designed SAE form. Local investigators notified the trial manager within 24 hours of the local investigators becoming aware of these events. In addition, local investigators were responsible for reporting SAEs to their host institutions, in accordance with local regulations, and instituting supplementary investigations as appropriate based on clinical judgement of the causative factors. SAEs and SARs were followed up until resolution even if this was beyond the participant’s discharge date from EPU care. The trial manager reported all SAEs to the Data Monitoring Committee (DMC) approximately once every 6 months. The DMC viewed data blinded to treatment, but was able to review unblinded data if requested.

Research questions

• Why do women choose medical management?

• Why do women choose to participate in the trial?

• What is their lived experience of the process of treatment and care?

• What factors influence the acceptability of and satisfaction with the treatment they have received?

• Does receipt of mifepristone with misoprostol influence women’s satisfaction with and acceptability of medical management?

• What implications does this have for the care offered to women and their partners?

Study design and rationale

To address the identified aim and research questions, we undertook a qualitative study nested within the MifeMiso trial, which sought to develop an understanding of:

• women’s experience of the trial treatment, considering the structures, resources and processes occurring during their participation as well as their perceptions concerning the treatment itself

• the contextual factors influencing women’s experience of the trial treatment and their satisfaction with it. 57,58

In conducting this study, we therefore sought to contribute an in-depth understanding to the overall findings of the trial, by addressing aims that illuminate aspects of women’s experience that are currently less well understood but that may influence their satisfaction with medical management and its acceptability. 57 Results such as this can make substantial contributions to the implementation of findings of the trial in the future. 57,59

The primary source of data collection was qualitative, semistructured interviews. Semistructured interviews are a widely used method of data collection in studies that explore people’s personal experiences and which aim to develop a deep understanding of their attitudes, perceptions and beliefs. Interviews are particularly suited to discussing sensitive topics when an open-ended approach is needed. 60

Two additional sources of data collection were utilised. All participants taking part in the MifeMiso trial were asked to complete the CSQ-8. 61 This questionnaire was designed for use in health studies and has been used recently in treatment studies for miscarriage. 42 The CSQ-8 has eight questions, each of which offers responses ranging from quite dissatisfied (scoring 1 point) to very satisfied (scoring 4 points). A total score of 32 can, therefore, be achieved. The higher the total score, the more satisfied the woman is with the treatment she has received. In this study, the CSQ-8 was utilised as a sampling frame when selecting participants for interview. In addition, participants completing the CSQ-8 were invited to provide additional thoughts concerning their satisfaction in a free-text box at the end of the questionnaire. These free-text comments were used as an additional source of data.

Methodological approach

We have framed this qualitative study within an epistemological position of constructivism62 and an ontological position of critical realism. 63 Constructivist theory proposes that there is no single unassailable truth; rather, meaning is constructed by individuals to make sense of their world, and these meanings are constantly being revised and re-interpreted. 64,65 By using a constructivist epistemology, we seek to understand the multiple realities and meanings that individuals create, by asking questions that relate to complicated features of human activity. 57,65 In applying constructivism in this study, we seek to understand how medical management of missed miscarriage is perceived by women and why they experience the treatment as acceptable or unacceptable. Although constructivism is grounded in relativism, it is argued that qualitative health-care research is commonly based on assumptions about the objective existence of specific conditions, such as, in this research, miscarriage. 63 Willig63 therefore suggests that it is appropriate to adopt a critical realist ontology alongside a constructivist epistemology, in which it is accepted that miscarriage exists, but that what we are concerned with is how medical management of miscarriage is constructed and understood by the women who experience it.

Recruitment

Recruiting HCPs provided eligible women with a PIS and informed them about the opportunity to participate in a qualitative interview when they were approached to take part in the trial. A subsequent invitation was provided at day 21, when they were asked to complete the CSQ-8 satisfaction survey. At this point, women self-selected to be approached for participation in the interviews by providing their contact details and indicating their agreement to be contacted by the qualitative research team. Satisfaction survey questionnaires were returned directly to the trials office along with signed permission to be contacted about an interview. These contact details were then passed onto the qualitative research team (JLFP and LJ).

Satisfaction scores were used as an initial sampling frame. We purposively selected participants from those who were satisfied/found the treatment acceptable, and those who did not. We also sought to increase the diversity by purposively recruiting a maximum variation sample of participants according to trial site, age, BMI, gestational age, parity and ethnicity. Within this context we sought to achieve diversity in the sample to ensure that a broad range of views, characteristics and experiences of participation were obtained.

We recruited women for interview between January 2018 and March 2019. Of the 238 women agreeing to be approached, we purposively sampled 85 (35.71%) for follow-up. After completion of the trial protocol and discharge from the service, we contacted the women using their stated preferred method (telephone, e-mail or post) and provided them with another PIS. If no response had been received, we contacted women again between 1 and 2 weeks later. Over the 15-month period, 50 (58.82%) women initially responded positively to contact and agreed to participate, although eight (9.4%) of these women subsequently withdrew before an interview could take place. All interviews were conducted within 6 weeks of completion of the trial protocol. Figure 7 provides a summary of the recruitment process and identifies reasons for non-participation where available.

FIGURE 7.

Summary of recruitment January 2018 to March 2019.

Consent

Women who responded positively to our initial approach were offered further information concerning the purpose of the interviews and their rights regarding participation and withdrawal. We gave them the opportunity to ask questions and time to consider whether or not they wanted to take part, although the majority indicated that they did not require this. If women agreed to participate, we set an interview date and consent forms were posted along with a prepaid reply envelope. Completed consent forms were returned to the trial office. Consent was discussed again at the agreed interview date for all women and any further questions were addressed before commencing the discussion.

Data collection

Semistructured interviews were the primary source of data. We developed the interview guide through a process of reflection and discussion, considering relevant literature and the aims of the research. During our discussions, we also explored any bias we might hold by exploring our preconceptions and personal experiences that could influence our thinking in this research. The interview guide explored the main areas outlined in Box 1.

We used the interview guide flexibly to respond to the women’s pace and experience and encouraged each participant to tell her story in the way that best suited her. The guide was subject to development over the duration of the interviews as we moved iteratively between analysis and interviews to ensure that early themes achieved saturation and no new themes were interpreted. 66

Interviews lasted on average 59 minutes (range 28 to 89 minutes). Thirty-eight (90.5%) of the interviews were conducted by telephone, three (7.1%) face to face and one (2.38%) by Skype. Although it was possible to offer (where the researcher was able to travel) a face-to-face interview as an alternative to 10 women, seven chose telephone as the medium they preferred. There may be a number of reasons influencing the choice of telephone interviews; for example, some of the women participating had returned to work and/or were caring for their children. As a result, interviews commonly took place in the evening or during women’s days off from work. It is also possible that the anonymity offered by telephone interviews was valued, given the sensitivity of the research. 67 Where interviews were face to face, two took place in the participants’ own home at their request. One face-to-face interview occurred in a private space at the University of Birmingham.

Ethics

For many women, the experience of miscarriage represents an unanticipated loss that has significant implications for psychological and emotional well-being. 7,56,68–70 Consequently, research with women at such a vulnerable time required us to be reflexive and sensitive to the potential distress that revisiting the experience might cause, particularly given that the interviews were occurring very soon after their miscarriage. Furthermore, some of the women had identified difficult experiences during previous miscarriage or birth-related trauma. To address potential difficulties we implemented a number of strategies:

• All women who were approached were provided with a list of support services (via the PIS) they could access to address their emotional and informational needs.

• We employed ethical mindfulness during the recruitment and interview process,71,72 considering women’s experiences and addressing needs as these arose. This included explaining what would be involved in the interview during initial discussions, stating that they could pause or stop the interview as required and maintaining awareness of the woman’s emotional well-being during the interview.

• Any questions regarding additional support needs were addressed at the end of the discussion following completion of the interview.

• At the end of the interview, we sought feedback from women about the interview process and used their feedback to adapt our approach as appropriate.

• Reflexive notes were maintained for each interview.

• We engaged in regular reflexive discussions following interviews.

Reflexive statement

I [JLFP] am a research fellow and trained as a mental health nurse, although I worked in a different area of health during my career as a nurse. I have an interest in women’s health and particularly the psychological and emotional needs of women in health care. I am also a woman and have one child of my own, two step-children and five grandchildren. I have experienced miscarriage and have undergone fertility treatment, both of which contributed to my interest in this research study.

My experience as a mental health nurse and my work in research with vulnerable populations has sensitised me to the potential challenges and sensitivities that may be present when conducting research with women in this context. However, I also appreciate that this and my personal experience can lead to assumptions about how a woman might view her miscarriage. For example, there is a risk that I uncritically apply my own experience and values and, therefore, assume that miscarriage is only a negative experience, which could influence the way questions were phrased, how I responded to women during interviews and how I analysed the data.

Consequently it was very important that I engaged in reflexive practice involving ethical mindfulness71,72 throughout the period of planning the interviews, recruitment, interviewing and analysis. In addition, my work as a health-care professional sensitised me to the potential challenges both women and health-care services might experience but I needed to ensure that these experiences informed rather than dominated my thinking during the research and analysis. Research supervision and reflection with LJ was, therefore, an essential process to ‘check in’, explore and challenge perceptions, as well as reflect on my own emotional well-being.

Analysis

Following completion of the interviews, all recordings were transcribed to a clean verbatim standard by a GDPR compliant transcription company. JLFP checked transcripts for consistency with the recordings and anonymised them, removing identifying details including names, locations and other details as appropriate. Interviews were then uploaded to NVivo qualitative software version 12 (QSR International, Warrington, UK) to support data analysis.

We used qualitative thematic analysis (QTA),73 informed by our theoretical position, to analyse the 42 interview transcripts. QTA is described as a flexible approach to analysis that has theoretical freedom; thus, it is possible to apply a constructivist paradigm and an inductive approach to arrive at a detailed interpretive account of the data. 73 These authors describe the analytical approach as involving six steps.

Step 1 involves becoming familiar with the data. JLFP read transcripts repeatedly alongside listening to the associated recordings, identifying emotions and emphasis. JLFP took notes to identify areas of interest and potential recurring messages that were discussed with LJ. LJ read a sample of the transcripts and compared thoughts and ideas on an initial coding structure with JLFP. At this point in the analysis of early interviews, an interpreted analytic memo reflected the variation in women’s experience of the scan and the sonographer’s approach when the diagnosis of missed miscarriage was made. JLFP then carried out steps 2 and 3 in tandem, involving developing initial codes and beginning to group nodes into low level themes. In developing the analytical memo further as other interviews contributed to these initial interpretations, two nodes were identified, reflecting different dimensions of the interaction between the woman and the sonographer (Box 2).

Given that part of the focus of the research was to explore women’s experience of the process of delivering the trial, these nodes were then collected together under the theme ‘the experience of the scan’. As interviews were conducted over a prolonged period of time, this step involved coding new interviews at nodes and grouping nodes into themes, which then informed subsequent interviews in an iterative process. These were regularly reviewed in analytical meetings between LJ and JLFP. Steps 1, 2 and 3 were completed following the end of data collection. Step 4 then involved exploring, reviewing and refining the content of the themes to determine if they belonged together or were more appropriately situated elsewhere or in new nodes, followed by a review of the whole set of themes. This process involved revisiting transcripts and holding analytical meetings to discuss progress and coding.

We determined that specific features of these and other nodes grouped under the theme ‘experience of the scan’ needed to remain within the context of a descriptive superordinate theme named ‘the pathway of care’ (not reported here). However, other aspects of the theme were more appropriately located in a separate theme that addressed the contextual factors influencing women’s satisfaction with medical management. This theme was labelled ‘empathic’, as women explicitly referred to the extent to which the HCPs involved in their care were empathic towards themselves and their partners. At step 5, JLFP further developed and refined the themes through discussion with LJ, revisiting data as appropriate to develop an overall thematic map. In this context, ‘empathic’ was determined as a subtheme of ‘alongside me’, reflecting the relative importance women placed on the extent to which HCPs were able to be present, empathic and responsive to their needs and enabled women to feel supported during their care. Analysis of the free-text comments was carried out using the themes derived from the interviews as a framework. No additional themes were identified from these comments.

Sample

In total, 42 women participated in the qualitative interviews. Table 11 provides a summary of the demographic characteristics of the women as a total figure and by group allocation. Twenty-three (55%) women were randomly allocated to the placebo and misoprostol group and 19 were randomly allocated (45%) to the mifepristone and misoprostol group. Participants were geographically distributed across the UK regions. Thirty-one (73.8%) participants were aged > 30 years. Thirty participants (71.4%) identified themselves as white British, with the remaining women coming from a range of different ethnicities. It is notable that it was not possible to recruit women from South Asian communities into the qualitative interviews, despite making contact with a number of women from this ethnic group who had participated in the trial.

Of the women participating, 24 (57.1%) were nulliparous (in the context of the trial, this was defined as women who had never carried a pregnancy beyond 24 weeks). Seventeen women (40.47%) had children (range 1–3), of whom one (2.4%) had previously experienced still birth. One other woman had experienced still birth and had no children. Sixteen (38.1%) of the women had experienced between one and three miscarriages in addition to the loss being treated within the trial. One further woman had experienced more than four miscarriages in addition to the loss treated within the trial.

Table 12 provides a summary of the satisfaction scores for the 42 women participating. As can be seen from the table, 37 (90%) of those who fully completed the CSQ-8 (n = 41) scored above 21, indicating satisfaction with the trial and service they had received.

In addition, 233 comments were analysed from the free-text box of the CSQ-8. These comments included the perspectives of women who chose not to agree to an interview, some of whom were not satisfied with the care they received. The demographic data for these respondents is not shown and their quotations are not included in the presentation of the findings.

Findings

We begin the presentation of our findings by exploring women’s satisfaction, discussing the perspectives of women on a continuum ranging from those who would choose medical management again to those who would not. Five superordinate themes are discussed within this, representing the factors influencing satisfaction and acceptability of medical management: (1) alongside me, (2) feeling informed and in control, (3) timely resolution to the physical process, (4) treatment experience and (5) treatment success. Following this, we briefly describe views on participation in the trial.

Satisfaction with and acceptability of medical management for missed miscarriage

Having completed the pathway, women were asked to consider and reflect on their overall satisfaction with the treatment they had received. This is now explored, along with the contextual factors influencing their perceptions of the satisfaction and acceptability of medical management. Women’s experiences suggested that satisfaction with and acceptability of medical management need to be considered on a continuum, as represented in Figure 8.

FIGURE 8.

Satisfaction with and acceptability of medical management of missed miscarriage.

Figure 8 shows that some women reported being entirely satisfied and would choose medical management again if this was needed, or would recommend medical management to someone they knew who was experiencing a miscarriage (note that all names used are pseudonyms):

For me, if it’s something that you know is going to happen anyway then for me speeding up that process is a lot better. It helps you then be able to grieve a lot quicker and then get back to normality a lot easier I think, so I would definitely recommend it. . . . Oh I’m really happy with everything that’s happened.

Jacinta

Other women reported unequivocally that they would not choose medical management should they experience a further missed miscarriage:

So you go through . . . we went through a transition of 8 weeks thinking this is the weekend, it’s all going to be over and each time something else was thrown at us. . . . If I knew then what I know now I would have done surgery in the very first instance, there would have been absolutely no doubt in my mind I would have just went for the surgical option.

Gemma

In addition, some women, despite reporting that they were satisfied, appeared from our interpretation of their responses to have reservations about medical management. For example, Florrie reported that in the context of recurrent miscarriage, it was the most appropriate option, but not one she was completely comfortable with:

If like me it’s becoming quite a recurrent thing then I would suggest medical management, even though it’s quite a horrible experience I think I would still have to choose that if it happened again.

Florrie

All three women quoted above gave specific answers to questions about satisfaction. However, their responses also illustrated that there are contextual factors underpinning their satisfaction that had an impact on their experience of the treatment, both positively and negatively. Although these may not result in women refusing medical management, they nevertheless have a significant influence on women’s views on the acceptability of medical management. Five main subthemes were identified, as illustrated in Figure 8. These contextual factors reflect both the experience of the treatment itself and the care environment in which medical management took place.

Women’s personal journeys, experiences and needs

Although not directly influencing women’s satisfaction with medical management, our interpretation of our participants’ experiences suggest that there are additional elements that are important context when understanding women’s experience of medical management of missed miscarriage. These elements are (1) women’s personal journey, (2) women’s desire for active management of missed miscarriage and (3) women’s experience of missed miscarriage and its influence on treatment choice.

Women’s perspectives on what should be offered in miscarriage care and treatment in the future

Women identified two main areas that they felt were needed in miscarriage care and treatment in the future: (1) the need for an accessible, woman- and partner-centred pathway of care and (2) personal support needs throughout the experience of pregnancy loss.

Strengths and limitations

There are strengths and limitations to this aspect of the study. We have considered the perspectives of 42 women taking part in the MifeMiso trial. We were able to recruit a range of participants representing different characteristics and demographics and achieve saturation in the thematic structure that we developed. This is therefore a large qualitative sample that has resulted in new insights and a depth of understanding of women’s experiences of medical management of missed miscarriage. These women’s experiences are likely to be transferable to women in similar contexts and we believe we have provided sufficient detail from which transferability to different contexts can be assessed. Nevertheless, this study relates to women’s experience of medical management for missed miscarriage in the UK, where EPUs have been in place for more than 20 years. This needs to be considered when exploring transferability of the results in countries with health-care systems that differ from this provision.

Our study has focused on women’s perspectives of medical management of missed miscarriage. Although it is a strength of this research that it provides an in-depth exploration of women’s experiences, it would have been further strengthened by considering the perspectives of professionals and partners alongside the women. This would have provided a broader, systemic consideration of the experience of medical management of missed miscarriage and the contextual factors influencing satisfaction. Considering these different perspectives may have contributed towards the development of a model of best practice for women and their partners. However, given that patient experience is considered to be a central tenent of clinical safety and effectiveness,75,86 our findings offer an important insight into women’s experience of medical management of missed miscarriage. In particular, we have been able to highlight the factors that positively influence satisfaction, a finding that has been largely absent from existing research.

One limitation of the study is that it only compares two different regimens for the medical management of missed miscarriage, and therefore cannot compare this against expectant management. However, our research group are currently undertaking a network meta-analysis that will estimate the relative effectiveness and safety profiles for all methods of management of miscarriage and rank the available methods according to their effectiveness and safety profile. 97

Another potential limitation is that the observed effect size was smaller than that considered in the sample size calculation. The observed difference in primary outcome rates between the two treatment groups was 7.6%, which is lower than the 10% difference that was deemed important by the survey of clinicians prior to commencing the study.

One final limitation of the qualitative study is the absence of the perspective of South Asian women. Existing research findings98 emphasise that participation of South Asian people in health research is likely to be increased if opportunities are provided to develop trusting and personal relationships. It therefore seems possible that a reason for the non-participation of South Asian women in this study reflects constraints of time and frequency of contact, which limited opportunities to develop a relationship that might facilitate their participation. Solutions to this will need to be addressed in subsequent research to ensure that the perspectives of women from South Asian communities are heard.

Data collection

Data on resource use and outcomes were collected during the MifeMiso trial using researcher-recorded trial collection forms. Data on health utility were collected using the EQ-5D-5L questionnaire. 101 As in standard practice, for the base-case analysis, only resource use incurred because of the trial intervention was considered. To calculate the total cost of the treatment, resource use data were multiplied by relevant unit costs.

Outcomes

The primary outcome of the economic evaluation, based on the main outcome of the trial (failure to spontaneously pass the gestational sac within 7 days post randomisation) was the number of successfully managed miscarriages. In keeping with the recommendation of NICE,101 an additional outcome of the economic evaluation was the QALYs gained. The QALY is a single index measure incorporating both quantity and quality of life using health utilities. 101

Some secondary clinical outcomes explored in the economic evaluation were the number of women who needed surgical intervention to resolve the miscarriage and the number of women who needed additional doses of misoprostol.

Resource use and costs

Resource use data within the NHS (secondary care costs) were collected to estimate the overall cost of drug administration, management of miscarriage and follow-up care (until discharge from EPU care) associated with the two groups of the trial. For each resource item, unit costs (see Table 13) were identified from established national sources such as the NHS Reference Costs 2018–19102 and the Personal Social Services Research Unit (PSSRU) costs. 103 In cases where there are different categories associated with resource use, weighted averages were used (see Table 13). All costs were reported in 2019–20 Great British pounds values and cost estimates from earlier years were inflated using the NHS Cost Inflation Index (NHSCII). 103

The main resource categories related to miscarriage that were collected included:

• the quantity of medication administered.

• the management of miscarriage in terms of the number of visits to secondary care facilities, such as outpatient visits, emergency visits and hospital admissions until final discharge, for example, if surgery is needed to resolve the miscarriage.

• The frequency of AEs and complications, such as blood transfusion and infection.

Drug administration (medication cost)

The trial medication was a single dose of 200 mg of mifepristone, taken orally after confirmation of missed miscarriage by pelvic ultrasound scan. The cost of mifepristone (Mifegyne,^® 200 mg) was identified from the British National Formulary (BNF)104 as £52.66 for a pack of three oral tablets, equivalent to £17.55 per 200-mg tablet (Table 13). As every woman in the trial received an initial dose of vaginal, oral or sublingual 800 μg of misoprostol, this cost was not included in the analysis.

However, for women who required additional doses of misoprostol, these were costed. The BNF provides the cost of 400 µg of oral misoprostol (Topogyne^®, 400 µg) as £128.00 for a pack of 16 tablets; therefore, a 400-µg tablet costs £8, which is equivalent to a cost of £16 for two tablets (800 µg). The routine practice within the NHS is to use oral tablets for vaginal and sublingual administration. Hence, the same cost was applied for all misoprostol administration irrespective of the route. A cost was not available in the BNF for sublingual misoprostol; however, the cost of misoprostol for vaginal administration (Mysodelle^®, 200 µg) was provided as £465 for a pack of five tablets, which is equivalent to £372 for 800 µg. The impact of the cost of vaginal misoprostol was explored in a sensitivity analysis.

Management of miscarriage (secondary care cost)

Resource use data were collected on the number of visits to the hospital and the emergency department as well as the number of outpatient admissions, inpatient admissions (a hospital stay of < 24 hours) and nights of admissions (a stay of > 24 hours). The cost of a regular hospital visit was obtained from the PSSRU. 105 For the cost of an emergency visit, the weighted average of emergency medicine for patients who require a category 0–2 treatment and a category 0–1 investigation were obtained from the NHS Reference Costs 2018–19. 102

Outpatient admission in the trial literature involved procedures such as an additional ultrasound and was, therefore, costed as an antenatal visit with a non-routine ultrasound. 102 Inpatient admission was described in the trial literature as the admission of a participant because of vaginal bleeding or any complication for observation. The trial literature further assumed that bleeding ceased and the participant was discharged on the day of admission and, hence, this situation was costed as a day-case management of medical/surgical, abortion or miscarriage care102 (Table 14). Nights in the hospital were recorded as inpatient night admissions costs within the trial. The weighted average cost per non-elective long-stay day was obtained102 to arrive at a mean cost per patient night of admission (see Table 14).

Further non-medical intervention for the miscarriage was categorised within the trial as manual vacuum aspiration (MVA) and surgical management (dilatation and evacuation) of miscarriage. These costs were obtained from the corresponding HRGs in the NHS cost schedule 2017–18. 106

Measuring health-related quality of life

To calculate QALYs for the CUA, health utility data were collected using the EQ-5D-5L questionnaire. 101 The EQ-5D-5L questionnaire is a self-reported preference-based measure of health-related quality-of-life index, which assigns a weight to different health states. It consists of five dimensions: mobility, self-care, usual activities, pain or discomfort, and anxiety or depression. Each dimension has five levels of functioning: no problem, slight problems, moderate problems, extreme problems and unable to. The EQ-5D-5L measure produces a possible 3125 distinct health states ranging from 11111 (full health) to 55555 (worst health).

Responses to the EQ-5D-5L were collected at baseline and at 6 or 7 days and 21 days post randomisation and at discharge from EPU care. For women in the trial that had a negative pregnancy test following the intervention, day 21 was the point of discharge. Baseline EQ-5D-5L data were collected immediately after randomisation. Follow-up EQ-5D-5L data were also collected at different clinical contact points.

Serious adverse events

Data on SAEs were collected using SAE forms. The trial’s literature defines a SAE as an untoward event resulting in maternal death, immediate threat to life, hospitalisation or persistent or significant disability. Only clinically specified SAEs deemed to have arisen from the trial intervention were considered to be relevant to the economic analysis. As there were no SAEs that were clearly related to the administration of mifepristone, we did not include such costs in the analysis.

Missing data

Multivariate regressions and t-tests were used to explore whether or not resource use and QALY values missingness could be predicted by other variables in the existing data. 107 If the associations between variables were not statistically significant at the 5% level, we assumed that data were missing completely at random. Based on this assumption, missing costs and QALY values were imputed using multiple imputations108 by applying chained equations with predictive mean matching across 25 imputations. 109

Primary cost-effectiveness analysis

A within-trial incremental CEA was conducted to estimate the costs and benefits of a combination of mifepristone and misoprostol (the MifeMiso intervention) compared with misoprostol alone. The CEA, which is the base-case primary analysis, was expressed in terms of the cost per successfully managed miscarriage. The analysis focused on the secondary care costs for participants incurred within the trial period.

The total cost during the trial period was estimated by multiplying the resource items used by the corresponding unit cost and summing these costs. We calculated the mean total cost and mean total resource use for participants across trial groups. Given the skewness inherent in most cost data and the concern of economic analyses with mean costs, using the bias-corrected and accelerated (BCa) non-parametric bootstrap method110 95% CI around mean differences were estimated by analysing 1000 resamples.

Multivariate cost analyses were conducted using seemingly unrelated regressions111,112 to assess heterogeneity in the trial population. Seemingly unrelated regression is an accepted method that factors in any congruity in the error terms between costs and outcomes113 and is robust to skewed data. Based on the minimisation variables for the trial, model covariates included baseline data on maternal age (< 30 or ≥ 30 years), BMI (< 35 or ≥ 35 kg/m²), gestational age (< 70 or ≥ 70 days) and the quantity of bleeding (PBAC score: ≤ 2 or ≥ 3). All results were presented as mean values with standard deviations (SDs) and, when applicable, as mean differences in costs and effects with 95% CIs.

Incremental cost-effectiveness ratios (ICERs) were calculated by dividing the difference in mean total cost by the difference in the number of successfully managed miscarriages between the trial groups. The ICER is a measure that depicts the additional cost ascribed to an additional effect. We used the formula below to calculate ICERs. Here C represents cost and E represents effects.

To quantify the uncertainty that is often attributable to sampling variations, we resampled the joint distribution in the mean cost and outcome difference using non-parametric bootstrapping. 114 The distributions were simulated 5000 times to generate paired estimates of incremental costs and outcomes, which were plotted in a cost-effectiveness plane. 115 A cost-effectiveness plane comprises four quadrants: north-east, north-west, south-east and south-west. The scatterplot that is produced on the plane represents the simulations. Depending on the location of the scatterplot dots on the quadrant, an intervention may be judged as more effective and more costly (north-east), more effective and less costly (south-east), less effective and less costly (south-west) or less effective and more costly (north-west) than the alternative intervention.

Cost-effectiveness acceptability curves (CEACs) were generated to depict the probabilities that the combination of mifepristone and misoprostol for the medical management of miscarriages is a cost-effective intervention compared with misoprostol alone across a range of values representing the decision-maker’s willingness to pay (WTP) for an additional benefit. 116

Cost-utility analysis

The EQ-5D-5L values were converted to utility scores using the interim crosswalk value set for the UK population. 117 This involved mapping the EQ-5D-5L scores back to the EQ-5D-3L valuation set, as recommended by NICE. 101 At each time point, participants’ responses to the EQ-5D-5L questionnaires were converted to utility score ranging from –0.594 (for worst health description) to 1.000 (for full health). Multiple imputations as described earlier were used to estimate missing EQ-5D-5L scores.

The number of QALYs gained by each participant was calculated with the area under the curve method using the trapezoidal rule. This method links the utility scores of each participant at different time points. To avoid bias, QALYs were estimated using multiple linear regressions with baseline EQ-5D-5L utility (in addition to other minimisation variables) as a covariate to adjust for any difference between the trial groups. 118

An incremental CUA was conducted to estimate the differences in costs and QALYs between the trial groups. The mean total QALYs across participants in the trial was calculated. We divided the cost difference between the two groups by the difference in QALYs gained to estimate the ICER and reported the outcome in terms of additional cost per additional QALYs gained. As the distribution of QALYs is typically skewed, 95% CIs around mean values were calculated based on 1000 replications using the bias-corrected and accelerated (BCa) bootstrap method. A CEAC was constructed to show the probability of the MifeMiso intervention being cost-effective across a range of possible values of WTP for an additional QALY.

Typically, ICERs are compared against the benchmark thresholds for cost-effectiveness in the NHS context of £20,000 to £30,000 per QALY gained, as recommended by NICE. 116 An ICER below £20,000 per QALY suggests that the intervention is a cost-effective alternative. An ICER above £30,000 per QALY suggests that the intervention is not cost-effective. For values between £20,000 and £30,000, the result is indeterminate.

Sensitivity analysis

One-way and/or multiway sensitivity analyses were performed, depending on the results of the deterministic analysis. These comprised an exploration of alternative unit costs applied to the different resources used or the variation of some parameters while leaving others at their baseline value. When appropriate, these analyses were combined with the stochastic sensitivity analysis. Sensitivity analysis was used to quantify the inherent uncertainty relating to assumptions and sampling variations in the methods used for the economic analysis and ultimately assess the generalisability of the results.

Sensitivity analyses for the CEA included:

• Exploring an alternative cost for additional misoprostol via the vaginal route – for the base-case analysis, as per the practice within the NHS, the additional doses of misoprostol were costed as the same irrespective of the route of administration (oral, vaginal or sublingual). The cost of 800 µg of vaginal misoprostol is £372 whereas the equivalent dose in the form of oral tablets costs £16. We explored the impact of assigning different costs (according to the route) on the analysis.

• Removing the cost of additional doses of misoprostol – in a situation where initial treatment was successful, there will be no need for additional doses of misoprostol. Hence, we explored the impact of removing these costs.

• Removing the cost of surgery – again, given a successful initial treatment, there would be no need for surgery.

• Imputation of missing secondary care costs – an analysis was carried out to assess the impact of missing outcome data on the analysis for all women with primary outcome data.

Sensitivity analyses for the CUA included:

• In the trial protocol, the plan was to derive utilities from EQ-5D-5L data using the English values. 119 However, following a recent recommendation by NICE,101 we used the crosswalk value set for the base-case analysis. For the sensitivity analysis, therefore, we explored the impact of using the English values.

• Some authors recommend adjusting for the baseline utility when calculating QALYs. In the sensitivity analyses, we assessed the effect of not adjusting for baseline utility values.

Outcomes

Of the 696 women with primary outcome data (Table 15), 289 (83%) in the mifepristone plus misoprostol group and 266 (76%) women in the placebo plus misoprostol group had a complete resolution of the miscarriage by day 6–7 post randomisation. This translates to an effect difference of 6.6% (95% CI 0.7% to 12.5%).

A breakdown of the resource use data by the trial group is presented in Table 16. For all resource use items with the exception of nights of admissions, women in the placebo group utilised, on average, more resources than women in the intervention group. These differences are also reflected in the costs, as shown in Table 17, which presents the mean secondary care costs per woman by trial group.

Mean total costs

The mean total cost by trial group for different variables is presented in Table 18. The mean hospital costs per woman for the trial period was £621 in the intervention group and £803 in the placebo group, generating a mean cost difference of –£182 (95% CI –£338 to –£26).

Cost-effectiveness analysis

The primary cost-effectiveness outcome of the MifeMiso trial was the cost of a successfully managed miscarriage as evidenced by the passage of the fetal sac by day 6 or 7. The analysis showed that the combination of mifepristone and misoprostol was more effective than misoprostol only, with an additional gain in seven successfully managed miscarriages per 100 women (Table 19). The use of mifepristone plus misoprostol resulted in a cost saving of £182 (95% CI £26 to £338) per woman.

The results of 5000 bootstrap replications plotted on the cost-effectiveness plane for the primary analysis is presented in Figure 9. Each point on the plane depicts a pair of incremental cost and incremental effectiveness estimates for the comparison between the trial groups. As shown in the figure, the majority of the scatter plots are in the south-east quadrant. This suggests that the MifeMiso intervention is dominant (i.e. less costly and more effective).

FIGURE 9.

Cost-effectiveness plane for the primary outcome.

Figure 10 provides the CEAC for the primary analysis, which illustrates the probability of the intervention being cost-effective at various values of decision-makers’ WTP per additional successfully managed miscarriage. The figure shows that for thresholds of WTP > £3000, the probability of the MifeMiso intervention being cost-effective is over 90%.

FIGURE 10.

The CEAC for the primary outcome.

Utility and quality-adjusted life-years

The response rates for the health economics participant-completed EQ-5D-5L questionnaires are presented in Table 20. There was a progressive decrease in data over the trial. Complete economic data were available for 593 (85%) women (intervention group, n = 296; placebo group, n = 297).

Of particular note is the data collected on discharge from the EPU, which were available for < 17% of the participants. The low level of data available for this variable is mostly because the last contact for women who had a negative pregnancy test following the intervention was on day 21. Owing to the paucity of data at discharge from the EPU, these data were not included in the analysis. In addition, the time points for discharge from EPU varied for each woman and it would be inappropriate to include the time points for discharge from EPU in the total QALYs calculation.

The level of missingness of the EQ-5D-5L data shows that, at day 21, complete data were available for approximately 85% of women in both groups. Comparisons were made between baseline covariates and the missing utility scores and t-tests were used to assess the rate of missingness. Although no relationship was found between age, BMI, gestational age or quantity of bleeding, missingness was related to baseline utilities. EQ-5D-5L scores were lower for individuals with missing data.

The EQ-5D-5L responses were converted into utility scores using the Crosswalk tariff. EQ-5D-5L scores at different trial time-points are shown in Table 21. For all time points except for discharge from EPU, the utility scores were higher in the intervention group than in the placebo group.

The mean QALYs attained over the trial period were estimated using the area-under-the-curve method (Table 22). These estimates are based on data for participants with complete EQ-5D-5L data from baseline to day 21. The CUA showed a slight effect in the intervention group compared with the placebo group with a QALY difference of 0.0004 (95% CI –0.0001 to 0.001). The combination of mifepristone and misoprostol remained a cost-saving intervention.

The cost-effectiveness plane for the CUA (complete-case analysis) is presented in Figure 11. As illustrated by the figure, the majority of the scatter plots are in the south-east quadrant, suggesting that the MifeMiso intervention is dominant (i.e. less costly and more effective). The CEAC (Figure 12) shows that for all WTP thresholds, the probability of MifeMiso being cost-effective is > 50%.

FIGURE 11.

Cost-effectiveness plane for the CUA.

FIGURE 12.

The CEAC for the CUA.

Similar analyses were conducted on the QALYs following the application of multiple imputations (Table 23 and Figure 13).

FIGURE 13.

Imputed QALYs: (a) cost-effectiveness plane; and (b) CEAC.

Principal findings

The results of the primary CEA suggest that the MifeMiso trial intervention (the combination of mifepristone and misoprostol) was less costly than the use of misoprostol only for the management of a missed miscarriage, with a cost saving of £182 (95% CI £26 to £338). The trial intervention is also more effective and led to an additional 6.6 (95% CI 0.7% to 12.5%) completely resolved miscarriages per 100 women. Hence, the combination of mifepristone and misoprostol is less costly and more effective, which suggests that ‘MifeMiso’ is a dominant intervention for the trial participants.

Similarly, the CUA showed that the intervention was slightly more effective for the management of a missed miscarriage with an additional QALY of 0.0004 (95% CI –0.0001 to 0.001). However, owing to limited data for the CUA, the results should be interpreted with caution. Sensitivity analyses were conducted to explore the robustness of primary analysis results to changes in the assumptions made. The conclusions drawn from the main analysis were shown to be robust to all sensitivity analyses.

Strengths and limitations of the trial-based economic analysis

A key strength of this analysis is that the economic evaluation was conducted in keeping with the recommended design and reporting guidelines. It was based on a multicentre randomised and controlled trial and provided the channel for prospective data collection. Data on outcomes and resource use were collected during the trial using CRFs and at specified points. Unit costs were drawn from established national sources and where variables were not clearly depicted by HRGs, we collaborated with the clinical teams to select the most suitable HRG. These are likely to enhance the generalisability of the study’s findings. The robustness of the main analyses as evidenced by the sensitivity analyses is a strength.

In addition, we carried out a CUA, thereby further measuring the effectiveness of the trial intervention in terms of QALYs as recommended by NICE. 116 The use of a preference-based measure of health outcome is more useful for comparative purposes. However, there were some missing EQ-5D-5L data in this trial, which we accounted for by imputing missing values. Although imputation is not ideal, the results are robust to these methods, as the complete-case analysis shows similar results. However, caution should be applied when making conclusions on the results of the CUA.

The trial protocol stated that we would collect data on primary care and from a societal perspective based on findings from preliminary data. There was no evidence in the trial or preliminary data to suggest that these data would have had an impact on the result. Therefore, to avoid unnecessary data collection and imposition on the women following a distressing event, this wider data collection was deemed inappropriate to pursue.

Comparison with the literature

To our knowledge, this is the first UK-based economic evaluation of the cost-effectiveness of mifepristone plus misoprostol versus misoprostol alone for the medical management of a missed miscarriage. A recent study in the USA120 assessed the relative cost-effectiveness of the two alternatives for the management of early pregnancy loss from the health-care sector and societal perspective and reported their results in terms of QALYs at 30 days post-intervention.

From the health-care sector perspective, the authors120 reported the trial intervention as minimally more costly, with a cost of US$2.87, and from the societal perspective they reported that the intervention was less costly with a cost saving of US$999.32. 120 Furthermore, the intervention group had higher QALYs with a difference of 0.0014 QALYs. From the health-care sector perspective, the use of mifepristone and misoprostol was cost-effective relative to misoprostol alone with an ICER of US$4225.43 (95% CI −US$195,053.30 to US$367,625.10) per QALY gained. Mifepristone and misoprostol also dominated misoprostol alone when the analysis was conducted from the societal perspective. When evaluating the cost-effectiveness by treatment success rates (complete expulsion of gestational sac after a single dose of misoprostol), the mifepristone and misoprostol combination was also found to be a cost-effective intervention in comparison with misoprostol alone.

Implications for policy

The within-trial economic evaluation found that the combination of mifepristone and misoprostol is likely to be considered by decision-makers for the medical management of women presenting with a missed miscarriage.

Objectives

The objective was to explore the relative cost-effectiveness of the combination of mifepristone and misoprostol (the MifeMiso intervention) versus all relevant alternative options for miscarriage management for which there are available and suitable data.

Methods

A decision analytic model was constructed to facilitate the comparison of all relevant alternatives for miscarriage management to determine the most cost-effective way of managing a missed miscarriage. The model was parameterised using a combination of evidence gathered in a clinical systematic review conducted by co-investigators of the MifeMiso trial and a systematic review of economic evaluations on early miscarriage management, which were supplemented by other published literature and expert opinion from within the research study team. The analysis was performed from the NHS perspective primarily because of the reliance on secondary data, which was limited from a wider perspective.

The primary outcome was the additional cost per additional successfully managed miscarriage, defined by the expulsion of the fetal sac within 7 days following randomisation. Given the short-term nature of the decision problem, a decision tree model was deemed appropriate. 121 The model was constructed and analysed in TreeAge Pro (2020) (TreeAge Software, Inc., Williamstown, MA, USA).

Extrapolation beyond the end of the trial using decision analytical modelling was not undertaken because the within-trial analysis found no demonstrable difference between groups in factors impacting long-term health outcomes.

Model structure

The model was developed to represent the alternative strategies for management of missed miscarriage in early pregnancy. The MifeMiso trial results were incorporated into the model structure to facilitate the comparison. The pathways of the model were finalised using expert opinion from within the research study team and, as far as possible, represent the practices followed in the UK in the event of a missed miscarriage. The decision tree is presented in Figure 14.

FIGURE 14.

Economic evaluation model pathway.

The model commences following the diagnosis of missed miscarriage in the first 14 weeks of pregnancy. Women can receive four alternative management strategies: surgical management, expectant management, current medical management or medical management with mifepristone plus misoprostol (medical management as proposed by the MifeMiso trial).

Surgical management refers to a procedure to remove the pregnancy tissue using a suction device. If, after the initial surgical intervention, the expulsion is incomplete, women will undergo a repeat surgical intervention.

Women receiving the current medical management could be given either mifepristone plus misoprostol, or misoprostol alone. It is assumed that women who were given mifepristone plus misoprostol and women given misoprostol alone follow the same pathway, which is equivalent to the pathway followed by participants of the MifeMiso trial. If, after the initial stage of medical management, the expulsion is incomplete, women undergo a second stage of medical management [additional dose(s) of misoprostol] or follow expectant or surgical management. If, after the second stage of medical management and surgical or expectant management, the expulsion is incomplete, women will undergo surgical intervention.

Expectant management refers to waiting for the pregnancy tissue to pass out naturally. If the initial stage of expectant management is unsuccessful, women will follow surgical management or medical management (with misoprostol) or undergo a second stage of expectant management. If, after subsequent surgical or medical management, the expulsion is incomplete, women will undergo surgical intervention. If, after the second stage of expectant management the expulsion is incomplete, women will undergo surgical or medical management. If the expulsion is still incomplete, women will undergo surgical intervention.

Model assumptions

Several pragmatic assumptions were required to develop a workable model. These are summarised and described below and divided into three categories: trial centres, model pathways and model inputs.

Model parameters

Costs and resource use data

The cost inputs for the model were estimated from the perspective of the UK NHS. Standard practice dosages were identified for mifepristone and misoprostol. With the exception of surgical intervention, the unit costs applied in the model are equivalent to those utilised in the within-trial analysis. The unit cost of surgical intervention in the medical management with mifepristone plus misoprostol branch was derived from the results of mifepristone plus misoprostol group of the MifeMiso trial, in which 22% of surgical interventions were performed using MVA and 78% of surgical interventions were performed using dilatation and evacuation.

The unit costs for surgical intervention in the branches populated by secondary source data were based on an assumption that 77% of surgical interventions are performed using dilatation and evacuation and 23% are performed using MVA. This assumption was derived from the results of both groups of the MifeMiso trial and is explored in the sensitivity analysis.

For surgical interventions performed in stage 1 of management, a gestational age of ≤ 14 weeks is assumed. For surgical interventions performed in the second or third stages of management, it is assumed that gestational age may be < 14 weeks or 14–20 weeks. The corresponding costs were obtained from the NHS Reference Costs 2017–18. 102 All costs sourced are reported in 2019–20 Great British pounds, having been inflated using the NHS Cost Inflation Index (NHSCII),103 if necessary. Table 29 presents the key costs for the model.

The assumed resource utilisation of the secondary care items for surgical management, expectant management and the current medical management were based on published literature identified in the systematic review of economic evaluations of early miscarriage management (Table 30). All items were agreed with the research study team before the model-based analysis was undertaken. Given a lack of studies assessing the cost-effectiveness of expectant management, where data were unavailable, the corresponding resource use for current medical management was applied. When data on medical management with mifepristone plus misoprostol or misoprostol alone were unavailable, the corresponding resource use for the alternative medical management regime was applied. In addition, data on nights of inpatient admission were not available; therefore, an assumption based on expert opinion from within the research study team was implemented. This inherent uncertainty is explored in the sensitivity analyses.

The resource use data for the mifepristone plus misoprostol branch of the model, based on the MifeMiso trial, is presented in Table 31. An additional hospital visit was added to the figure detailed in Table 4 of the trial-based analysis to account for the initial visit attended by each woman at randomisation. As all women in the model attend an initial visit, we consider only uncertainty for subsequent visits.

Outcome measure

The outcome in this model was measured as a successfully managed miscarriage. This is reported in terms of the ICER, which indicates the additional cost per additional successfully managed miscarriage.

Base-case analysis

Our base-case analysis was conducted from an NHS perspective in which direct health service costs are taken into account.

Deterministic sensitivity analyses

Additional scenarios were investigated following the base-case analysis, which tested the significance of the assumptions made in the model. Scenarios investigated included:

1. increasing the proportion of women undergoing a second stage of expectant management by 50%

2. reducing the proportion of women undergoing a second stage of expectant management by 50%

3. all surgical interventions performed by MVA

4. all surgical interventions performed by dilatation and evacuation

5. equal proportions of surgical interventions performed by MVA and dilatation and evacuation

6. increasing the effectiveness of medical management following an initial stage of expectant management by 10%

7. decreasing the effectiveness of medical management following an initial stage of expectant management by 10%

8. decreasing the nights of inpatient admission in all strategies populated by secondary source data by 50%.

Probabilistic sensitivity analysis

A probabilistic sensitivity analysis was conducted to explore the uncertainty of the model input data. For each uncertain model input parameter, distributions were assigned, from which a value was randomly drawn. In total, 10,000 Monte Carlo simulations were subsequently computed, which generated mean cost and effectiveness estimates by simultaneously varying all relevant parameters. These estimates were used jointly to form an empirical distribution of the differences in both the cost and effectiveness of interventions. Beta distributions were applied to probabilities where only two outcomes were possible. Dirichlet distributions, the multinomial extension of the beta distribution, were applied if three outcomes were possible. As there is no uncertainty around unit costs, no distribution was fitted. However, resource use is uncertain; therefore, gamma distributions were applied to costs including resource use. When resource use data were derived from alternative strategies or only point estimates were available, the widest possible uncertainty was applied.

Base-case analysis

The results of the base-case analysis are shown in Table 32. Medical management with mifepristone plus misoprostol is the least costly strategy, with a mean cost of £761 per woman. The most effective strategy is surgical management, whereas medical management with mifepristone plus misoprostol is the second most effective strategy. Both the current medical management and expectant management strategies are dominated by medical management with mifepristone plus misoprostol, as they are more costly and less effective than medical management with mifepristone plus misoprostol.

However, compared with medical management with mifepristone plus misoprostol, surgical management is more costly but more effective. The estimated ICER for surgical management compared with medical management with mifepristone plus misoprostol is £6969 per additional successfully managed miscarriage. This means that every additional successfully managed miscarriage achieved using surgical management over medical management with mifepristone plus misoprostol costs an additional £6969.

Deterministic sensitivity analyses

The results of the deterministic sensitivity analyses are shown in Report Supplementary Material 7. Each scenario investigated made no substantial difference to the base-case results.

Probabilistic sensitivity analysis

Probabilistic sensitivity analysis using 10,000 Monte Carlo simulations showed moderate uncertainty.

Medical management with mifepristone plus misoprostol compared with surgical management

The cost-effectiveness plane in Figure 15 shows the modelled uncertainty in the cost and effectiveness between medical management with mifepristone plus misoprostol and surgical management. The ICER of each simulation is plotted on the cost-effectiveness plane, providing information about the joint density of the differences in cost and effectiveness between the strategies. It is shown that medical management with mifepristone plus misoprostol is highly likely to be less effective than surgical management while being less costly.

FIGURE 15.

Cost-effectiveness plane for medical management with mifepristone plus misoprostol relative to surgical management.

The CEAC for medical management with mifepristone plus misoprostol compared with surgical management is shown in Figure 16. Given an arbitrary WTP threshold of £5000 per additional successfully managed miscarriage, the probability that medical management with mifepristone plus misoprostol is cost-effective is 86% and the probability that surgical management is cost-effective is 14%. However, if the WTP threshold is higher at £10,000 per additional successfully managed miscarriage, the probability that medical management with mifepristone plus misoprostol is cost-effective is 15% whereas the probability that surgical management is cost-effective is 85%. As the WTP per additional successfully managed miscarriage tends to infinity, the probability that surgical management is cost-effective compared with medical management with mifepristone plus misoprostol tends to 99%. The difference in probabilities over WTP thresholds reflects uncertainty in the model.

FIGURE 16.

The CEAC for medical management with mifepristone plus misoprostol compared with surgical management.

Medical management with mifepristone plus misoprostol compared with current medical management

The cost-effectiveness plane in Figure 17 shows the modelled uncertainty in the cost and effectiveness between medical management with mifepristone plus misoprostol compared with the current practice of medical management. It is shown that medical management with mifepristone plus misoprostol is highly likely to be more effective than the current practice of medical management; however, it is uncertain as to whether or not it is less costly.

FIGURE 17.

Cost-effectiveness plane for medical management with mifepristone plus misoprostol relative to current medical management.

Figure 18 shows the CEAC for medical management with mifepristone plus misoprostol compared with current medical management. The CEAC shows that at any WTP threshold > 0, medical management with mifepristone plus misoprostol is the preferred strategy compared with the current medical management. As the WTP per additional successfully managed miscarriage tends to infinity, the probability that medical management with mifepristone plus misoprostol is cost-effective compared with the current practice of medical management tends to 99%.

FIGURE 18.

The CEAC for medical management with mifepristone plus misoprostol relative to current medical management.

Medical management with mifepristone plus misoprostol compared with expectant management

The cost-effectiveness plane in Figure 19 shows the modelled uncertainty in the cost and effectiveness of medical management with mifepristone plus misoprostol compared with expectant management. It is shown that medical management with mifepristone plus misoprostol is consistently more effective than expectant management; however, it is uncertain as to whether or not it is less costly.

FIGURE 19.

Cost-effectiveness plane for medical management with mifepristone plus misoprostol relative to expectant management.

Figure 20 shows the CEAC for medical management with mifepristone plus misoprostol compared with expectant management. The CEAC shows that at any WTP threshold > 0, medical management with mifepristone plus misoprostol is the preferred strategy compared with expectant management. As the WTP per additional successfully managed miscarriage tends to infinity, the probability that medical management with mifepristone plus misoprostol is cost-effective compared with the current practice of medical management tend to 99%.

FIGURE 20.

The CEAC for medical management with mifepristone plus misoprostol relative to expectant management.

Principal findings

The results of the model-based analysis show that for the management of an early missed miscarriage the strategy proposed by the MifeMiso trial (medical management with mifepristone plus misoprostol) is dominant when compared with expectant management and the current medical management strategy, as it is more effective and less costly than both strategies. However, medical management with mifepristone plus misoprostol is a less effective strategy than surgical management, though it is less costly.

The ICER for surgical management compared with medical management with mifepristone plus misoprostol compared with surgical management is £6969 per additional successfully managed miscarriage. The PSA suggests that at a WTP threshold below approximately £7000, medical management with mifepristone plus misoprostol is the preferred strategy relative to surgical management. The results of this analysis were shown to be robust over the full range of deterministic sensitivity analyses.

Strengths and limitations of the model-based economic analysis

A key strength of the model-based economic evaluation is that it is the first model-based economic evaluation to compare the cost-effectiveness of the three broad alternative management strategies. Furthermore, the model-based economic evaluation considers the cost-effectiveness of a management strategy proposed by a clinical trial in the context of current practice for missed miscarriage management. Being able to directly compare alternative management strategies and to rank them in terms of cost and effectiveness is especially useful for policymakers.

The principal limitation of the model-based economic analysis is that in the absence of a network meta-analysis on the management strategies for missed miscarriage over the relevant intervention period, the effectiveness data were based on the results of published clinical trials. Although the quality and relevance of the trials were stringently assessed, several biases may be attached to the trials that could compromise the accuracy of the data.

In addition, data on resource use items were not available for all management strategies. This meant that assumptions had to be made about the resource use from alternative management strategies. Similarly, no data were available on nights of inpatient admission for the expectant, surgical and current medical management strategies; therefore, assumptions were made using expert opinion from within the research study team. Attempts were made to make missing resource use data as accurate as possible, and the significance of the assumptions made was tested in the deterministic sensitivity analysis in an attempt to rectify this limitation.

No published studies analysed the effect of medical management with misoprostol following an initial stage of expectant management over the relevant period. Similarly, no data were available on the proportion of women undergoing various management strategies following an unsuccessful initial stage of expectant management. Therefore, assumptions based on expert opinion from within the research study team were made. The significance of these assumptions was tested in the deterministic sensitivity analysis where the results showed they made no substantial difference to the base-case result.

Information on the impact on quality of life was not available for all management strategies included in this analysis; therefore, the outcome for the model-based analysis was expressed in terms of clinical effectiveness rather than in terms of the standard unit of benefit, the QALY, which means that the meaning of the results is not as easy to interpret. A model-based analysis based on an outcome of additional QALYs at discharge may be useful for policy-makers.

The analysis did not examine how patient preferences inform decisions, although a previous study on stated preference using a discrete choice experiment found that women undergoing management for first trimester miscarriage expressed a general preference for surgical over medical management, and would value being offered alternatives to expectant management. 129 Furthermore, the model-based economic evaluation makes no comparisons for different dosages of mifepristone and misoprostol or different routes of administration. Comparing the effects of different dosages and routes of administration in the medical management of miscarriage may be useful for future research.

Comparison with the literature

To the best of our knowledge, there is currently no published evidence on the cost-effectiveness of the medical management with mifepristone plus misoprostol compared with alternative management strategies including surgical and expectant for the management of missed miscarriage.

Implications for policy

The results of the model-based economic analysis suggest that medical management with mifepristone plus misoprostol is likely to be recommended by decision-makers for the medical management of missed miscarriage than the current practice, which consists of a combination of mifepristone plus misoprostol or misoprostol alone. When alternative methods of miscarriage management are considered, the results suggest that the best choice is between medical management with mifepristone plus misoprostol and surgical management, but that medical management with mifepristone plus misoprostol is likely to be recommended by decision-makers ahead of expectant management.