Ondansetron and metoclopramide as second-line antiemetics in women with nausea and vomiting in pregnancy: the EMPOWER pilot factorial RCT

Main trial aims

This trial aimed to determine which hospital-prescribed therapy [metoclopramide (metoclopramide hydrochloride, Actavis UK Ltd, Barnstable, UK; IV Ratiopharm GmbH, Ulm, Germany) and ondansetron (ondansetron hydrochloride dehydrate, Wockhardt UK Ltd, Wrexham, UK; IV Hameln Pharma plus GmbH, Hameln)], in addition to i.v. rehydration, should be adopted as the mainstream second-line treatment of severe NVP by the UK NHS when standard first-line treatment has failed. To achieve these aims, we set the following objectives.

Primary objectives

To determine whether or not, in addition to i.v. rehydration, ondansetron compared with no ondansetron and metoclopramide compared with no metoclopramide reduced the rate of treatment failure up to 10 days after initiation.

Secondary objectives

To determine whether or not, in addition to i.v. rehydration, ondansetron compared with no ondansetron and metoclopramide compared with no metoclopramide:

• improved symptom severity at 2, 5 and 10 days after intervention initiation

• improved quality of life at 10 days after intervention initiation

• had an acceptable side effect and safety profile.

In addition, we also aimed to estimate the incremental cost per treatment failure avoided and the net monetary benefits from the perspective of the NHS and women.

Internal pilot and qualitative trial objectives

Owing to uncertainty around the willingness of pregnant women to be recruited into a complex drug trial, an internal pilot phase was incorporated alongside a qualitative substudy. The main objectives of the pilot phase were to assess the feasibility of recruitment and rates of retention in those randomised. The qualitative component aimed to improve our understanding of why women did or did not consent to participation in the trial.

Trial design

This was a multicentre, double-dummy, randomised, double-blinded, dummy-controlled 2 × 2 factorial trial with an internal pilot phase. Women attending hospital with severe NVP who experienced little or no improvement while taking first-line anti-sickness medication were invited to participate. Participants were randomised in a 1 : 1 : 1 : 1 ratio to receive active metoclopramide and dummy ondansetron; dummy metoclopramide and active ondansetron; active metoclopramide and active ondansetron; or dummy metoclopramide and dummy ondansetron. The trial aimed to compare whether or not, in addition to i.v. rehydration, ondansetron compared with no ondansetron and metoclopramide compared with no metoclopramide reduced the rate of treatment failure up to 10 days after initiation of treatment. The trial included an economic evaluation and a qualitative component.

Internal pilot

It was planned that there would be an internal pilot phase involving up to eight sites (subsequently extended to up to 15 sites) followed by a main phase with an additional seven to nine sites (22–24 sites in total).

To progress from the pilot trial to the main trial, 59 participants needed to be recruited within 6 months, with ≥ 40 participants retained to day 10.

Participants

The trial recruited women aged ≥ 18 years who were < 17 weeks’ gestation and had attended a secondary care service suffering from severe NVP or HG, and who had received previous treatment with a first-line antiemetic therapy in the current pregnancy with little or no sustained benefit. Potential participants were first approached by the clinical staff on duty when they attended hospital for treatment. Participants presented at gynaecology/early pregnancy clinics, maternity assessment units (MAUs) or accident and emergency (A&E) departments. The trial was introduced to the participant and permission was sought for someone in the research team to approach them and provide more information about the trial. Patients who expressed an interest in taking part were provided with a copy of the participant information sheet (PIS). Those still interested after reading the information provided were then assessed for eligibility.

Screening

Screening logs were completed at recruiting sites to record reasons for patient ineligibility and for declining participation. In August 2018, the screening logs were extended to collect information on prior use of trial drugs in the current pregnancy. Confirmation of eligibility was performed by a General Medical Council (GMC)-registered medically qualified doctor who was trained in good clinical practice (GCP).

Eligibility

The eligibility criteria that were initially used during the trial are listed below.

Consent

A copy of the PIS was provided to eligible women by the local site team. Patients were first provided with a short PIS to provide them with an overview of the trial. Those interested after reading the short PIS were provided with a copy of the full PIS. Potential participants were given a minimum of 1 hour to review the PIS. During this time, an i.v. cannula could be sited, blood could be taken for necessary clinical tests and i.v. fluids could be commenced. Consent was received by a GMC-registered medically qualified doctor trained in GCP. Women who agreed to participate were asked if they would also be willing to take part in an optional qualitative interview.

Women who declined to participate in the trial itself were still asked if they would like to be approached to participate in the qualitative interviews.

Once consent had been received, randomisation was undertaken.

Schedule of events

The schedule of events for the trial is shown in Table 1 and a participant’s journey through the trial is shown in Figure 1.

FIGURE 1.

Trial flow chart. EOI, expression of interest; EPDS, Edinburgh Postnatal Depression Scale; IMP, investigational medicinal product; PUQE, Pregnancy Unique Quantification of Emesis; STAI, State–Trait Anxiety Inventory; VAS, visual analogue scale.

Once a participant was consented and randomised, the trial medication was dispensed by the research team and i.v. medication commenced as soon as possible. Baseline questionnaires were also completed by the participant.

While in hospital, participants were provided with a trial pack that included the questionnaires to be completed at 2, 5 and 10 days, a trial diary to record adverse events (AEs) and concomitant medications, and a trial identification (ID) card, including emergency contact details, for participants to carry with them while participating in the trial. All participants, unless they had been discharged from hospital, were reviewed at 12 hours to assess responsiveness to trial medication. If the participant had not responded to the trial medication within the first 12 hours of treatment, trial medication was stopped and the participant was placed on further treatment outside the trial.

Follow-up with the participants took place at 2 days post first dose of trial medication, at 5 days and at the final follow-up questionnaires at 10 days. These follow-ups were completed via telephone call by members of the research team. Participants were also followed up at 20 weeks’ gestation and post birth, but via medical chart review.

Participant expenses

No expenses were incurred by participants travelling for trial visits. Follow-up was completed via telephone or via review of the participant’s medical chart.

Details of the trial interventions

Participants were randomised to one of four groups. The groups that a participant could be allocated to are outlined below, along with the dose and route of administration:

• metoclopramide [10 mg three times daily (i.v. and p.o.)] and dummy ondansetron (i.v. and p.o.)

• ondansetron [4 mg three times daily (i.v. and p.o.)] and dummy metoclopramide (i.v. and p.o.)

• metoclopramide [10 mg three times daily (i.v. and p.o.)] and ondansetron [4 mg three times daily (i.v. and p.o.)]

• double dummy three times daily (i.v. and p.o.).

Delivery of the trial interventions

Participants initially received trial medication intravenously; trial medication could be administered intravenously for up to 4 days. Once the participant was able to tolerate fluids, they were converted to oral tablets. The oral medication was taken at the same frequency and dosage as the i.v. formulation. Both the i.v. and the p.o. medication were supplied in the same trial medication kit to prevent the necessity for a double prescription. The total duration of treatment for each participant was 10 days.

Treatment compliance

Participants took trial medication for a maximum of 10 days. Participants were contacted at 2, 5 and 10 days and compliance regarding taking oral medication was checked with participants at these time points.

Participants were asked to return all unused oral trial medication and all packaging (even if empty) to the trial team at each site. A stamped, addressed envelope was provided as part of the trial pack for participants to return medication after they had taken their final dose at day 10. Participants were reminded by the research team to post trial medication back to the site, but if they forgot there was the option to bring it with them to their next clinical appointment. The site team counted all returns (both tablet returns received from the participant and leftover ampoules returned to the pharmacy) and recorded these on the trial database.

Withdrawals

Participants had the right to withdraw from the trial at any time without having to give a reason; however, if provided, the reason for withdrawal was recorded. Participants remained in the trial unless they withdrew consent or the local principal investigator (PI) felt that it was no longer appropriate for the participant to continue. For participants who chose to withdraw, permission was sought to allow the research team to collect routinely collected data from their medical records. If the participant did not agree to this, only the data collected up to the point of withdrawal were retained. Participants who withdrew from the trial were not replaced.

Sites

The EMPOWER trial was conducted at 12 NHS sites in England. Site set-up commenced in October 2017.

Initially, seven sites were set up:

1. Newcastle upon Tyne NHS Foundation Trust

2. City Hospitals Sunderland Hospitals NHS Trust (now South Tyneside and Sunderland NHS Foundation Trust)

3. South Tees Hospitals NHS Foundation Trust

4. Bradford Teaching Hospitals NHS Foundation Trust

5. Leeds Teaching Hospitals NHS Trust

6. Birmingham Women’s and Children’s NHS Foundation Trust

7. Guy’s and St Thomas’ Foundation Trust.

Five further sites were set up during the extended internal pilot:

1. Pennine Acute Hospitals NHS Trust

2. University Hospitals of North Midlands NHS Trust

3. St George’s University Hospitals NHS Foundation Trust

4. Nottingham University Hospitals NHS Trust

5. Portsmouth Hospitals NHS Trust.

Primary outcome

The primary end point was the number of participants experiencing a treatment failure. Treatment failure was defined as the need for further treatment if a participant’s symptoms had worsened between 12 hours and 10 days post initiation of treatment. Where further treatment was required, a participant should have been placed on third-line antiemetic treatment (i.e. high-dose ondansetron or corticosteroids) unless the clinician considered further second-line treatment more appropriate.

Secondary outcomes

The following secondary patient-reported and clinical outcomes were collected:

• Symptom severity, as measured by the Pregnancy Unique Quantification of Emesis (PUQE) score19–21 (see Appendix 1), collected at baseline and at 2, 5 and 10 days post initiation of treatment.

• Severity of nausea, as measured by a visual analogue scale (VAS)17 for nausea, collected at 48 hours and at 5 and 10 days post initiation of treatment.

• Quality of life, as measured by the health-related nausea and vomiting during pregnancy quality-of-life (NVPQoL)22 questionnaire (see Appendix 2), collected at baseline and 10 days post treatment commencing.

• Depression, as measured by the Edinburgh Postnatal Depression Scale (EPDS)23 (see Appendix 3), collected at baseline and 10 days.

• Anxiety, as measured by the six-item State–Trait Anxiety Inventory (STAI),24 collected at baseline and 10 days [see https://oml.eular.org/sysModules/obxOml/docs/ID_150/State-Trait-Anxiety-Inventory.pdf (accessed September 2021); the six items used were items 1, 3, 6, 15, 16 and 17].

• Social support, as measured by the Maternity Social Support Scale (MSSS)25 (see Appendix 4), collected at baseline.

• Clinical indicators of antiemetic effectiveness:

  • the number of participants experiencing a treatment failure at 2 days

  • relapse rate at 5 and 10 days (defined as a PUQE score of ≤ 6 at 2 days followed by an increase to > 12 at 5 or 10 days)

  • remission rate at 10 days [defined as a PUQE score of ≤ 6 at 2 days with return to persistent symptoms (PUQE score of ≥ 7) at 10 days]

  • readmission rates (the number of participants readmitted with NVP within 10 days of recruitment and between 10 days of recruitment and 20 weeks of pregnancy)

  • total inpatient days related to NVP between recruitment and 20 weeks of pregnancy and between 20 weeks of pregnancy and delivery

  • additional antiemetic use.

• Side effects and AEs – participants were asked about side effects/AEs that they experienced at 2, 5 and 10 days post treatment commencing. Severity was graded using the Common Terminology Criteria for Adverse Events (CTCAE). 26

• Pregnancy and neonatal outcomes – the numbers of women experiencing miscarriage, TOP and stillbirth were recorded. For each infant born, the mode of delivery, gestational age at delivery and birthweight were collected. Information on congenital anomalies detected prior to discharge was also collected. In the case of multiple births, this information was collected for each infant.

• Key economic outcomes:

  • costs to the NHS and to participants (see Appendix 5)

  • willingness to pay (WTP) (see Appendix 6) for the benefits of antiemetic treatment

  • incremental cost per treatment failure avoided and net monetary benefits.

All data fields were locked on 5 December 2019, with the exception of fields relating to pregnancy outcome data, AEs/serious adverse events (SAEs), deviations/violations, investigational medicinal product (IMP) accountability and concomitant medication, which were locked on 9 April 2020.

Sample size calculation

As per the 2 × 2 factorial design, the intention was to have two main comparisons:

1. active metoclopramide (with either active or dummy ondansetron) and dummy metoclopramide (with either active or dummy ondansetron)

2. active ondansetron (with either active or dummy metoclopramide) and dummy ondansetron (with either active or dummy metoclopramide).

An antiemetic failure rate of 10% was assumed based on prior ondansetron studies (with rates of 13% at 48 hours27 and 7% at 5 days28) and our pilot RCT of midwife-led outpatient care,29 in which 10% of women had failure of a second-line antiemetic (primarily metoclopramide). With 266 participants in each comparison (i.e. 133 in each of four treatment combinations, 532 in total) and assuming no interaction between the two active treatments, it was estimated that we would have 90% power to detect an increase in failure rate from 10% in either antiemetic group to 20% in either dummy group (based on a two-sided test at the 5% level). This effect size was felt to be of clinical relevance to applicants and sufferers.

As the primary outcome was treatment failure between 12 hours and 10 days post treatment initiation, we anticipated obtaining outcome data from at least 90% of participants. Therefore, to account for a 10% dropout prior to assessment of the primary outcome, it was planned to recruit 600 patients (150 in each randomised group). Sample size calculations were performed using power twoprop in Stata^® version 14 (StataCorp LP, College Station, TX, USA).

Randomisation

Randomisation was carried out centrally by the Newcastle Clinical Trials Unit (NCTU) using a secure web-based system. Patients were randomised on a 1 : 1 : 1 : 1 basis using computer-generated random permuted blocks of size four and stratified by site to receive (1) metoclopramide and dummy ondansetron; (2) ondansetron and dummy metoclopramide; (3) metoclopramide and ondansetron; or (4) double dummy. Randomisation was performed by local research staff, who were delegated this task via the delegation log. The randomisation system generated a unique participant trial identifier and allocated a pack number to be dispensed to the participant. Pre-populated prescriptions were generated by the randomisation system.

Unblinding

The EMPOWER study was a double-blind trial. It was not planned that participants would be routinely unblinded once they had completed trial treatment.

However, participants could be unblinded in a valid medical emergency or for safety reasons where it was necessary for a treating clinician to know which treatment the participant had been receiving. Emergency unblinding could be carried out by the site PI (or another member of the team delegated this responsibility via the delegation log) by accessing the 24-hour randomisation system. A set of back-up sealed code-break envelopes were also held at sites in case of system failure or lack of availability of a delegated individual.

Following discussions with the Trial Steering Committee (TSC), a decision was made to fully unblind all trial medications after the trial was stopped, allowing PIs to notify participants of the medication that they received during the trial.

Statistical analysis

A statistical analysis plan was drafted and finalised prior to data lock. Given that the trial closed to recruitment at the end of the pilot phase, only descriptive summaries of the data have been reported. No efficacy analyses have been undertaken owing to the small sample size. Had the internal pilot trial progressed to a full trial, the group comparisons would have been used as originally intended for this factorial trial. However, given that the pilot data are being reported, the focus of the results is on feasibility outcomes and, therefore, data have been summarised by randomised treatment group.

The progress of all participants through the trial is presented using a Consolidated Standard of Reporting Trials (CONSORT) flow diagram, including the number of screened and eligible patients. Rates of eligibility and recruitment are summarised by site. Reasons for ineligibility or non-participation have been tabulated.

Baseline data are summarised descriptively. Numbers (with percentages) for categorical variables and medians (with ranges) for continuous variables were summarised by randomised treatment group and overall.

A summary of trial medication received is reported using numbers (with percentages) by randomised treatment group and overall. Reasons for premature treatment discontinuation are described.

Primary and secondary outcome measures were summarised descriptively by randomised treatment group and overall. Data are presented following the intention-to-treat principle, with participants retained in the groups to which they were randomised. The primary outcome measure of treatment failure between 12 hours and 10 days was tabulated. The rate of treatment failure [with 95% confidence intervals (CIs)] was calculated out of those evaluable; participants choosing to withdraw from trial treatment before 12 hours and participants choosing to withdraw from the trial without allowing further use of routine data, between 12 hours and 10 days or without prior documented treatment failure were not considered evaluable. Readmissions for NVP, inpatient days related to NVP and additional antiemetic use were tabulated out of those evaluable. Relapse at day 5 or day 10 and remission at day 10, as defined by changes in the PUQE score, were tabulated out of those with completed PUQE questionnaires at day 2 and also at days 5 and 10. The completeness of participant-reported questionnaires was tabulated for each measure. Questionnaires were considered to be completed only if responses were available for all items. Questionnaire scores at each assessment point were summarised descriptively using numbers (with percentages) for categorical variables and medians (with ranges) for continuous variables. For each questionnaire, a score was calculated only if all items were completed. No imputation of missing data were used.

Pregnancy and neonatal outcomes were summarised descriptively, by randomised treatment group and overall, as numbers (with percentages) for categorical variables and medians (with ranges) for continuous variables. For multiple births, neonatal outcomes were summarised for each infant born.

Analyses were conducted in Stata^® version 16.

Sources of bias

The primary outcome measure of treatment failure, defined as the need for further treatment because participants’ symptoms had worsened between 12 hours and 10 days, is likely to be partially subjective. This risk of ascertainment bias was reduced by the trial being double blinded, so neither the participant nor the clinical team were aware of the allocated treatment. Secondary outcome measures, including participant-reported questionnaires, were also completed or collected by participants, clinicians or research staff blinded to treatment allocation.

As far as possible, the reasons for eligible patients declining participation were recorded; however, patients were free to decline without giving a reason. It is possible that those who were too unwell to consider the trial information were more likely to decline than those with less severe symptoms; however, randomisation should have ensured that the treatment groups were balanced with respect to severity of NVP at baseline. The PIS illustrated the uncertainty and lack of high-quality evidence around the effectiveness of the trial drugs for women with NVP, and women had to be willing to be allocated to any treatment group, including double dummy.

Patients were not considered evaluable for the primary outcome measure if they withdrew from trial treatment prior to 12 hours or fully withdrew from the trial between 12 hours and 10 days without prior treatment failure. This could have introduced bias if participant withdrawal related to the effectiveness of trial treatment. Reasons for withdrawal were obtained where possible; however, participants were free to withdraw from the trial at any time without giving a reason.

Missing outcome data at follow-up, particularly for participant-reported outcome measures, could have introduced bias if data were not missing at random. Had the trial progressed to a main trial, patterns of missing data would have been explored and multiple imputation methods considered.

Health economic evaluation

The health economics analysis plan was finalised and approved before data lock. Given that the trial closed after the pilot phase, the proposed evaluation was no longer appropriate and the economic analysis was confined to two primary components:

1. Presentation of health service resource use data in the form of summary statistics.

2. Information on health-care resource use was collected via electronic case report forms (eCRFs) (for concomitant medication) and participant-completed health-care utilisation questionnaires (A&E attendance, receipt of primary care and other NHS/Personal Social Services care services), and was completed at 10 days post randomisation. Data on participant-related costs were also collected via the health-care utilisation questionnaire at the day 10 follow-up assessment. Participants were asked to provide information relating to the purchase of private health care or personal care. 30

An analysis of the completeness of the resource use data collected was conducted using descriptive statistics to calculate the number and percentage of participants, with data available for each resource item.

1. Presentation of WTP data in the form of summary statistics.

2. Participants were asked to directly express their WTP for a good service using a contingent valuation (CV) questionnaire. To ascertain their preferences, participants were presented with hypothetical scenarios and were asked how much they would be willing to pay to improve symptom severity for a 10-week period. The hypothetical scenarios were informed by input from women with experience of NVP, the literature and other experts. For a given level of income, a higher WTP value indicated that individuals derived greater benefit from a reduction in symptom severity.

Information gathered via the CV questionnaires is presented as mean [standard deviation (SD)] and median [interquartile range (IQR)] WTP amount to improve symptom severity over a 10-week period. The maximum and minimum WTP values for each of the four randomised treatment groups have also been reported.

All information was summarised by randomised treatment groups and overall. The analysis was conducted using Stata^® version 15.

Qualitative evaluation

As part of the protocol for the EMPOWER clinical trial, an evaluation of women’s views of the acceptability of the trial design was carried out with the aim of improving our understanding of why women did, or did not, consent to randomisation and participation in the trial.

Recruitment to the qualitative study, as with the pilot trial, progressed slowly. The eligibility criteria were broadened in February 2019, as many women were found to be ineligible because of prior use of trial medication. As a result, the number of women available to approach for the qualitative study, particularly of decliners, was limited. In consultation with the TSC, it was therefore agreed to extend the qualitative component to include research staff to undertake a wider exploration of factors influencing trial recruitment.

Patient recruitment

Women who had been screened and were found to be eligible, and had accepted or declined to participate in the trial, were approached for recruitment to the qualitative substudy. Potential recruits were given a copy of the separate PIS and expression-of-interest (EOI) form. To be considerate to patients who were possibly still suffering from NVP, attempts by the qualitative researcher to contact and interview women were confined to the period between the 14th and the 28th day after the initial approach to participate in the EMPOWER trial. Brief 10- to 15-minute interviews were conducted with women who returned the form and were contactable (see Appendix 7). An option of an e-mail interview was also provided in case women were too ill to converse. Consent was obtained over the telephone, recorded and transcribed. A copy of their consent form, filled in by the researcher on their behalf, was sent to participants by e-mail or by post.

Research staff recruitment

Research staff contact details were accessed via NCTU. E-mails about the staff trial (with the PIS and consent form attached) were sent to staff who had screened or recruited women to the trial. Research midwives/nurses directly involved in trial recruitment, and site PIs who were interested in participating in the interviews, returned the consent form by e-mail or agreed to consent taken over the telephone. The large majority who were contacted were willing to participate and the telephone interviews lasted between 20 and 30 minutes (see Appendix 8).

Clinical indicators of antiemetic effectiveness

Readmissions [including assessment unit (AU) visits] for NVP up to 20 weeks’ gestation are summarised in Table 9. Of those evaluable readmissions (n = 31), nine (29%) (including for a day visit) occurred within the first 10 days. Between day 10 and 20 weeks’ gestation, seven (23%) participants were readmitted: one allocated to active metoclopramide and dummy ondansetron, three allocated to active ondansetron and dummy metoclopramide, three allocated to active metoclopramide and active ondansetron, and no one allocated to double dummy. Between 20 weeks’ gestation and delivery, three participants were readmitted to hospital for NVP: one allocated to active ondansetron and dummy metoclopramide, and two allocated to active metoclopramide and active ondansetron.

Further antiemetic treatments prescribed up to 20 weeks’ gestation and between 20 weeks’ gestation and delivery are summarised in Table 10. Up to 20 weeks’ gestation, further antiemetic treatment was prescribed to 25 (81%) participants: four (67%) allocated to active metoclopramide and dummy ondansetron, seven (88%) allocated to active ondansetron and dummy metoclopramide, eight (89%) allocated to active metoclopramide and active ondansetron and six (75%) allocated to double dummy. Twenty (65%) participants were prescribed ondansetron, 10 (32%) cyclizine, eight (26%) prochlorperazine, seven (23%) metoclopramide and one (3%) promethazine. Between 20 weeks’ gestation and delivery, 12 (43%) participants received further antiemetic treatment. Further information on relapse and remission rates, as defined by changes in PUQE score, can be found in Appendix 12, Table 21.

Participant-reported outcomes

The number of fully completed participant-reported questionnaires at days 2, 5 and 10 is shown in Table 11. At day 2, of those evaluable (n = 30), 24 (80%) participants had completed a PUQE questionnaire; by day 5 and day 10 this figure had fallen to 22 (76%) and 15 (52%) of those evaluable (n = 29), respectively. Of the 15 PUQE questionnaires completed at day 10, three were from participants who had met the primary outcome of treatment failure, meaning that 12 participants in whom treatment had failed did not have a completed questionnaire. These figures were similar for other questionnaires (see Table 11).

Summaries of the participant-reported outcome measures at each time point can be found in Appendix 10, Table 19 and Appendix 11, Table 20.

Pregnancy and neonatal outcomes

The pregnancy and neonatal outcomes are summarised descriptively in Table 12. Of those participants with follow-up data available (n = 31), 28 (90%) participants had a live birth and three (10%) participants had a TOP prior to 20 weeks’ gestation. One participant, who was allocated to active ondansetron and dummy metoclopramide, had a TOP because of fetal abnormalities (see Safety) and two participants, who were allocated to active ondansetron and active metoclopramide and double dummy, had a TOP for other reasons. The median gestation at delivery was 39^1/7 weeks and the median birthweight was 3.0 kg. All infants with a 5-minute Appearance, Pulse, Grimace, Activity, Respiration (Apgar) score available scored ≥ 7. Three infants (triplets born prematurely at 27^6/7 weeks gestation) did not have an Apgar score available. One infant, born to a mother who was allocated to receive double dummy, was reported to have a congenital abnormality at birth.

Health economic outcomes

The response rates relating to participant-completed health economic questionnaires are shown in Table 13. An equal number of the participants (n = 16, 48%) did not complete any part of the health-care utilisation and CV questionnaires. The pattern of non-response was similar across the four randomised treatment groups.

The main results of the WTP analysis are shown in Table 14. Overall, 13 participants (39%) provided an answer stating the maximum amount that they would be willing to pay to improve symptom severity for a 10-week period. The mean value across all groups was £674, with the double-dummy group reporting the lowest mean amount (£123). Participants across the other three randomised groups reported similar mean values. The minimum amount that a participant was willing to pay was £20 (reported by a participant in the double-dummy group) and the maximum stated amount was £2000 (reported by participants in the active metoclopramide and active ondansetron group, and the active ondansetron and dummy metoclopramide group).

Safety

The three SAEs that were reported during the trial are shown in Table 15. The TOP owing to fetal abnormalities was classed as a suspected unexpected serious adverse reaction (SUSAR), as the participant was identified as having received ondansetron following emergency unblinding. The causality of the event was indeterminable in relation to the IMP (ondansetron) and unexpected. The SUSAR underwent expedited reporting to the MHRA and REC and all site PIs were informed of the event.

Three further events, in two participants, were reported as being possibly related to trial treatment: one participant allocated to active metoclopramide and dummy ondansetron reported both constipation and diarrhoea; and one participant allocated to active metoclopramide and active ondansetron reported drowsiness. None of these events was serious.

The trial question and adherence to the Royal College of Obstetricians and Gynaecologists guidelines

For the trial to succeed as intended, the rationale for the research needed to be understood and accepted by all those involved in delivering the trial. As the trial progressed, however, two emerging issues were identified by some site PIs: first, that the trial question had lost its relevance and second, that NVP had a lower priority than other pregnancy complications:

. . . if I’m honest with you I think the general feeling was that practice has moved on, and that the question that this trial was asking was now probably a historical question . . . You know we’d almost like we’d missed the boat.

TS19

. . . when we’re coming to nausea and vomiting the perceived, the sort of the perceived issues or problems coming out of it will not sort of rank alongside stillbirth or neonatal death.

TS18

However, other PIs, patients and midwives/nurses were of the opinion that research in this area was much needed and that the trial question remained important. This was in spite of the issue of randomisation to double dummy being controversial at the initial stages among patient and public involvement advisors:

I thought it was a well-designed study and it was worth definitely worth, a question worth asking . . . I don’t think the study design was a problem which was what we initially thought was going to be a problem.

TS18

There was evidence from both PIs and research midwives/nurses that there had been a change in the practice of prescribing antiemetic drugs since the decision was made to conduct the trial (February 2016), which impacted negatively on recruitment during the course of the trial:

I think because it took us a long while to get it going and I think the whole landscape changed almost just before we started recruiting [. . .] and my, my belief is that was largely to do with the guideline the RCOG guideline . . . the trial was an independent endeavour but meanwhile there were lots of us working in the background to improve care for women with hyperemesis, which meant enabling GPs and hospital doctors to prescribe appropriately for it, and ironically our success in doing that has led to the failure of the trial.

TS16

This shift in practice was evident in both primary and secondary care in spite of RCOG step-up guidelines12 recommending the use of the trial drugs as a second-line medication and confirmation from all trial sites that local guidelines on antiemetic use were consistent with RCOG guidelines. 12 However, participants indicated that not all clinicians at the study sites and in general practices in their localities adhered to these guidelines when they were treating women with severe NVP because they had become used to prescribing the trial drugs:

It was not easy a trial to start with because of all the bad practice which is all very well set where people would prescribe whatever antiemetic they would feel like at that time rather than follow RCOG standardised step-up protocol.

TS17

Some of the research staff at sites were unaware of the antiemetic-prescribing culture within which the trial was attempting to operate:

. . . we just hadn’t realised how many women were already on those study medications from GPs and things so it’s been a big eye opener to us about what kind of prescribing practices are out in primary care.

TS20

As a result of this culture of prescribing in primary and secondary care, women were frequently presenting with severe NVP having already been treated with one of the trial drugs, rendering them ineligible for the trial. In some regions, this practice was widespread among GPs, whereas in others the problem was in secondary care:

I did think initially when the trial started that a lot will be ineligible because the GPs prescribing. But I was surprised actually we could catch a lot who were eligible and it was, we were messing up their eligibility at secondary care level which was sorted later part. So I don’t think GPs were a major problem here.

TS17

Organising the delivery of the EMPOWER trial

The successful delivery of the trial required organisational co-operation from clinical staff across departments, that is liaison between maternity/obstetrics, gynaecology, emergency and pharmacy departments. Staff at one trial site were more familiar with trials that were based solely in maternity services, which meant additional effort working with gynaecology colleagues:

This trial was a little bit different in that it was more gynae[cology] than obstetrics which gave I think the trial team in every unit a little bit of challenge whereas we’re used to approaching women in pregnancy further on in pregnancy.

TS16

Nevertheless, there was evidence of good working relationships between clinical staff in maternity and clinical staff in gynaecology day units, and research staff, to facilitate recruitment:

We work with quite closely with the early pregnancy unit and also gynae[cology] emergency, and what we decided to do was get the clerical staff that work on reception to contact us, bleep us when the women were coming in. Cos we’ve got quite a good relationship with that area.

TS21

Where there were staff with both a clinical and a research role, this facilitated the recruitment process for the EMPOWER trial:

One of our research nurses works there in her substantive role as well. So she does some research and some early pregnancy and, so that’s brilliant because she knows all the systems she knows all the staff and consequently they’ve got to know us as well through her. We do, they are very good at referring people to us.

TS02

In some sites, the responsibility for the care of the patient was restricted to unit or ward staff, with research staff providing a minimal amount of help. This may be because the number of research staff was limited. In other sites, however, research staff contributed to the care of eligible women while they were considering the trial:

. . . what I’ve tended to do is not just give the information sheet and come away but I’ve usually kind of helped with starting i.v. fluids and you know making sure that if the patient has any questions that I’m there to answer, you know.

TS04

Good working relationships with the NCTU staff at Newcastle University were critical to the operational delivery of the trial. In the initial stages, there were several glitches, mainly to do with the unavailability of trial staff when they were urgently required:

. . . we’ve tried to call the EMPOWER office and there’s never been anybody who actually works on the trial available to talk to us.

TS05

. . . you’re usually be able to ring by phone if not but there was no back-up there was no way of getting a hold of anybody so we just had to scrap it which is not, not great.

TS20

Some staff commented that the trial database was not user-friendly and that it was unhelpful having separate log-ins to recruit and randomise women:

. . . the whole recruitment process and the databases, having separate databases for randomisation and things like that it can be very confusing.

TS08

Nevertheless, as the trial progressed, and as lines of communication were more firmly established, sites reported how helpful the NCTU staff were in supporting the management of the trial:

Yeah we’ve never had a problem we’re in contact regularly we get a lot of e-mails from them if there’s anything on the database that we haven’t done right, or if there’s any discrepancies they’re quite good at liaising with us and managing like that, so yeah all our contact with the team has been really positive, I think it’s a lovely trials team to work with.

TS11

Demands of the EMPOWER trial protocol

The trial protocol stipulated that women had to have taken first-line antiemetic treatment for a minimum of 24 hours before they could be considered eligible for the trial. Most site guidelines for the management of NVP did not recommend the admission of women, especially with moderate symptoms, who were started on first-line antiemetic therapy. Thus, trial staff reported that they were unable to easily determine if potentially eligible women responded to first-line treatment:

It’s, like we’ve got to get her in and out we can’t wait 3 hours and then see how she feels and then put her in EMPOWER because they don’t tend to like admitting them either.

TS03

So if they’ve never had anything before, usually in that time they’ll just have one dose of cyclizine and then go home. So it’s then waiting to see if they come back and then we can consider them.

TS05

Women could return at any time, including out of hours or at the weekend, when site research staff were not available. Nevertheless, sites were informed that women could be approached about future consideration of the trial when they were on first-line treatment. This advice was followed at some but not all sites:

We’ve also seen a few people who have come in and maybe just having their first-line treatment and just mentioned the study to them and just said we have a study running in the unit you wouldn’t be eligible for it now but bear it in mind if this treatment doesn’t work for you we may come and see you again. So just to raise a bit of awareness with these women who might be coming in.

TS13

One patient suggested that information should be provided about the trial at her first visit to the MAU:

I didn’t know whether you could like send something out after I’d been there in the first time, to say if you come back there is a trial. So I would have had prior knowledge.

EMP1003

The trial was perceived to be particularly challenging because it was a CTIMP with a complex design consisting of four groups, as well as requiring a trial population of pregnant women in a fragile physical and emotional state at the point of recruitment. For this reason, research staff were required to adopt a sensitive approach and give eligible women enough time (1 hour) to consider the trial information.

Women’s perspectives provided insight to the recruitment processes. Although one patient felt that she needed more time to think over her decision to participate, another patient declined because she thought that her wait for treatment would be lengthened because of the trial:

I think how it was explained was that obviously she would come back with, like in an hour, and then there was paperwork to be filled out so obviously that took, would have taken time and then that had to be uploaded in to the computer system to decide which treatment I would then go ahead with . . . I mean I think we’d had a chat about it so we’d pretty much made our mind up probably what 15 minutes after she’d left us. So had she maybe been around a little bit sooner than the hour that we were expecting, that might have helped a little bit.

EMP1003

Research staff reported that they felt that the recruitment process could be quite quick once women had agreed, whereas approaching the women and making sure that the information was understood took longer:

. . . obviously going to the patient making sure that they’re eligible, giving them the information sheet and obviously you want them to have the information sheet for the length of time so that they’re happy, that they understand and that I understand or know that they understand what we’re asking of them.

TS04

Some participants reported difficulties in accessing medical staff who were GCP trained. One group highlighted were GP trainees who were perceived to have less interest either in the specialty area or in contributing to the research:

I found I was quite successful with the doctors who were career trainees in obs[tetrics] and gynae[cology] but I didn’t have as much success with the GP trainees because they were here for 4 months or 6 months and I didn’t find that, they didn’t feel that for the time that it took that it was useful for their CVs.

TS19

On the other hand, other PIs reported that they had ensured that there were enough GCP-trained doctors available:

So most of our senior trainees have completed the GCP training and they are on the delegation log for a number of research trials.

TS22

If women agreed to participate in the trial, the protocol stated that an assessment of treatment efficacy had to be made at 12 hours to see if they were responding. However, some women felt that this time period was too long a time to endure:

I mean I do reassure them by saying after 12 hours if the medication wasn’t working you’d be withdrawn from the study and [ML – yeah] but I guess, you know, for the women who are feeling really bad, 12 hours probably seems like a very long time.

TS01

An additional issue was the chance that participants might receive the double dummy for 12 hours, an issue highlighted by both site staff and prospective participants:

And if they come in when they’re really, really ill when they come in, I could sort of see people not wanting to do that really just risking getting the placebo.

TS15

. . . I was really, really worried that I had two placebos.

EMP1702

The need for double blinding meant that participants were not aware of the trial drugs that they were taking. Concerns about unblinding in the event of treatment failure or trial withdrawal were dealt with through the initiatives of the research teams (e.g. by prescription of both trial drugs). However, this did not take away the desire of some participants to know what medication they had received after the trial was completed:

I just think sort of ethically I would although you sign up to do a trial you do have a right to know what you took because now I’m guessing I was on metoclopramide so in a way, or you know guessing that’s the best one for me to take erm yeah I would, I would like to know what I took.

EMP1702

Publicity and education about the trial

For the trial to be successful, relevant health-care professionals had to be made aware of the requirements of the trial so that appropriate referrals were made. Medical staff also had to be trained to approach patients, confirm eligibility, take written consent from the patients and prescribe the trial drugs. In addition, women with severe NVP needed to be made aware that there was an option for them to join a trial that might benefit them and other women with the same condition. If health-care professionals were not offering information about the trial and/or were not adhering to RCOG guidelines12 on antiemetic use, this could preclude potentially eligible women from participating in the EMPOWER trial.

The impact of the unpredictability of presentation to secondary care

From staff accounts, there did not seem to be predictable patterns of when patients presented to sites at which recruitment would take place. Women could present with symptoms out of hours, at weekends or on bank holidays, perhaps because they could not find the time during the week when they were at work or they had other family commitments. Women might also have to wait until evening for transport to the hospital by their partners who are at work:

. . . if they’re at home and then someone else brings them cos they’ve been at work all day and then they’ll bring them in the evening.

TS13

Women who attended at the weekend as an emergency were often given a weekday follow-up appointment at a specialist MAU or GAU (TS16). Some staff interviewees made reference to the maternity assessment centre (MAC) or the antenatal day unit, whereas others referred to the early pregnancy unit or the early pregnancy assessment unit.

In addition, care pathways differed from hospital to hospital depending on the gestational age at presentation. This influenced when women were referred to specialist assessment units (AUs) at which they could be approached by trial staff. At some sites, to ease the workload at their MAU, women at lower gestations were seen within the gynaecology department (either in an early pregnancy unit or on the gynaecological ward). This meant that research staff had to try to adapt their recruitment strategies accordingly:

I think it depends on the way that your trust is set up so we, we have the obs[tetrics] med team and they go and see the ward every single day to see if anyone’s been admitted with hyperemesis.

TS08

In some units, it worked well that the local A&E department policy was to transfer patients with NVP directly to the specialist MAU or GAU. Some sites provided out-of-hours or 24-hour access to specialist assessment, but several staff made reference to the issue of women presenting out of hours leading to their inability to recruit:

But it just, it’s out of hours you know they tend to come in, I think they battle on through the day and come in on an evening or a weekend or maybe they’ve battled on at work and they they’ve come in out of hours then, you know, so. But it’s like all trials you’re going to miss you cannot be there 24 hours a day, 7 days a week, so it’s hard.

TS09

However, there were also pathways in which women were encouraged to attend in the daytime if they were able:

But we have a system where, doesn’t always happen but if somebody’s got hyperemesis and they’re not too unwell I think certainly from evenings onwards and overnight the appointment is made for them to come in the morning and they can have their treatment in daylight hours and they can speak to the research midwives.

TS15

On the other hand, some women presented with such severe symptoms that they were admitted as inpatients rather than being seen in an AU, which meant that they could potentially be missed by research staff if they focused efforts on screening women referred to the AU. When the protocol was changed to allow trial drugs to be taken orally, inpatients treated intravenously were likely to be missed:

I think once they’ve got to the point where they’ve being admitted to hospital and they’re in and they’re vomiting and they’re being rehydrated they’ve got i.v. access so they tend to get given i.v. drugs.

TS20

The impact of the infrequency of eligible women

Clinicians reported that the number of women who required specialist assessment and treatment for severe NVP was small. Familiarity with trial processes took some time to achieve, and research staff commented that the small number of eligible cases meant that they gained limited experience with the recruitment processes:

And I think the relatively infrequent presentation of women requiring treatment for nausea and vomiting in early pregnancy again makes it quite challenging . . . And I think that made it difficult for people to, for the clinical staff to always be on the ground, and always be on the ball if you like.

TS18

. . . it’s not like you can find your stride very quickly it takes, and then by the time you get another recruit you think what am I supposed to do again.

TS05

However, to some extent, features of help-seeking in the affected population offset this barrier. For example, some women attended the site several times, and staff reported that this gave opportunities for research staff to introduce the trial to them:

Yes to like attend you know the antenatal day unit because for 2 weeks I was constantly going in every other day because my, my, I was being dehydrated so nothing was working, sickness was at very peak. Then I kept going in to have some fluids because I was, my ketones in the urine was really high.

EMP1402

Women who were ineligible

Women who were eligible

Confidence and morale

The amount of paperwork, together with the slower than expected recruitment into the trial, had an impact on the confidence of research staff in carrying out trial processes to the standards required in a CTIMP. Accounts from staff on the delegation log indicated that they could be reluctant to take on the task of recruiting a patient because the process involved felt onerous:

. . . it sounds awful but you kind of dread actually recruiting someone because you know it’s so complicated, you know, it is really hard all the information you need to remember.

TS08

Failure to recruit was because of a number of administrative hurdles and in the early stages was a source of frustration:

And we had given our first recruit the medication orally because she asked for it that way and hadn’t realised that the first dose had to be i.v. so it was that, with that lady . . . and then there’s been some issues with sending things over different e-mail addresses . . . We have had to call to clarify at the point of recruitment on every single occasion.

TS02

Faced with the challenges of recruitment, enthusiasm for the EMPOWER trial waned. However, successfully recruiting new participants encouraged staff to press on:

. . . we’ve found, you know women, we were kind of getting a little bit, that we keep on looking and we can’t find women that are fitting the criteria but having the recent two recruits has been a real boost.

TS06

Morale was also boosted when comparisons were made with other trial sites through the regular updates provided by NCTU, and PIs reassured research staff that they were not alone:

. . . they feel better that it’s not just our site that isn’t doing as well. I mean they know that, you know, it’s a national thing it’s not just one particular site doing, not doing as well.

TS14

When staff encountered an AE, their morale was affected; however, when they saw participants benefiting from the care provided, they were motivated to carry on:

. . . the midwives have been on a shift where women have taken part it feels like they’ve got an enthusiasm for the study they can see, you know, these women kind of felt better in the short term. We know it might just be the fluids, we don’t know what really drug that we give them but it’s kind of a positive feeling which kind of helped, you know, remind the staff that it’s a good study.

TS06

Recruitment and trial processes

Recruiting vulnerable women suffering from severe NVP was frequently reported as challenging:

I would even say that like we’ve done studies on labour ward where you’re speaking to ladies in labour and sometimes that’s been easier than trying to speak to a lady who’s puking her guts up on the, on the assessment unit.

TS10

Trial staff were dependent on their clinical colleagues to assist, but they reported difficulties because of the amount of paperwork involved during recruitment that required them to be present. This made recruitment out of hours virtually impossible:

. . . it’s a little bit more time consuming that clinical staff aren’t able to fit that in. If it’s just a short period of time it’s not quite so bad and they will accommodate but this is a little bit more involved really I think.

TS04

. . . it’s always been really difficult to recruit patients out of hours when you’re relying on your medical teams unless it’s really straightforward and obviously there’s a lot of paperwork that goes along with EMPOWER recruitment isn’t there.

TS10

Staff also reported that data collection during the course of the trial could be demanding:

All the paper work you need to give, all the forms you need to fill out, you have to make sure the patients know, you know, when to fill it in and you know, you’ve got to call the patient, well if they’re discharged at 12 hours, they’re clearly fine at 48 hours, at you know 4 days, 10 days etc.

TS08

However, staff accounts indicated that good teamwork mitigated against the demands of the trial processes. Less confident staff reported that they felt supported by their colleagues:

But before I’d done those two I was a bit like if we get one can someone come with me so we can kind of go through it together.

TS03

Principal investigators in the trial valued their research midwives or nurses, especially if they did most of the liaising with fellow clinicians and co-ordinated the work of the trial:

I think the main staff on MAU they were aware of the study and I think I must give credit to [name 1] who was like the team leader and the co-ordinator there. So she used to liaise with [name 2] who was the research midwife and [name 3]. So [name 2] and [name 3] also co-ordinated closely with [name 1] and all the consultants too were aware of the study.

TS22

Patient counselling and staff empathy

There were reports describing how staff counselling about the risks involved in participating in the trial, particularly reassurance about patient care during and after the trial, was critical in the recruitment process:

. . . sometimes they worry about the placebo–placebo but once you explain to them that you are actually having the i.v. fluids they’re not leaving you with, you know, absolutely nothing and if it doesn’t work if you’re feeling unwell we can take you off straight away, they do feel more reassured.

TS08

The following observations from one of the PIs summarises the importance of staff counselling and care, not just from research staff but from clinical staff managing women at the point of admission:

. . . if, when a women came in with nausea and vomiting, the admitting midwife had a positive word to say for the study in ways, not just of course through the study, but other ways in which the woman would feel assured she was going to be taken care of, that makes it much easier for the recruiting team to come on board and make that, make that case effectively for taking part.

TS18

However, while providing information about the trial, convincing patients that rehydration with i.v. fluids was a form of treatment in itself could prove difficult.

At the same time, staff also expressed empathy for the women whom they were trying to recruit onto the trial, such as for the long wait that women had to endure for treatment:

I do sometimes feel that the women have a rough deal if you like, because they go to the GP and then they come to the hospital and it’s sometimes a while before they have, you know, the treatment.

TS04

If patients were given information about where they were in the queue to see a doctor, some participants felt that this could make women more open to considering the trial:

So we want to make sure that they know that their care isn’t being held up by us as soon as the doctor comes they’ll see them so, we just say we’re just seeing them while they’re waiting for the doctor anyway and I think that makes them feel a bit better cos they don’t feel like we’re like sort of delaying their treatment if you know what mean.

TS11

When approaching women, research staff had to be sensitive to how women were feeling at that point in time, and to approach them appropriately. One patient felt that the research was more important than the care that she was desperate to have:

But it was kind of, felt as though I’m waiting for some treatment which I’m not getting but the research information was more important, it felt to me, cos I had got a visit about three times.

EMP1401

On the other hand, many patients gave positive and appreciative accounts of how they were approached, whether they accepted or declined participation:

She explained everything in detail she gave me the time I needed to process. She, basically sat there and explained it all in all ways, in a way that I understood.

EMP1201

. . . when I spoke to them you know they said it was about my health and quality of life so they were very good at making sure I knew what my options were to withdraw from the trial if I needed to.

EMP1703

Principal investigators mostly had complete confidence in the abilities of their research midwives/nurses to counsel patients about their options:

They don’t have to do obviously you know sort of, in terms of consenting and everything but they’re happy to do all the talking, and you know very objective and unbiased and stuff so.

TS15

From other feedback from interviews that described empathy for women with NVP, it appeared that staff commitment to the trial could be compromised if they felt that participation had a negative effect on the well-being of their patients. To illustrate this, when a SUSAR, a fetal abnormality that could have had a link to one of the trial drugs, occurred research staff reported feeling distress and a sense of responsibility for the patient’s treatment with the trial drugs and the subsequent termination of the pregnancy.

Staffing levels and support

Appropriate levels of staffing were also critical to the trial; recruitment processes could have been affected by staff being on leave:

I only work 15 hours and because we’ve been short of staff in the, in the team, really everybody, although we’ve got about, we had about three staff until somebody just came back off long-term sick. Because we’re so short when we’re at work we’re just mainly focusing on our own studies.

TS10

Seen from the perspective of one PI, as recruitment numbers were small, the workload for midwives or nurses was relatively lighter than for busier trials with a large number of participants:

. . . there were few patients, even if we’d recruited to target it wasn’t many patients per centre and, therefore, the workload for the midwives [they were] working with the women, supporting women and the questionnaire . . .

TS16

From the perspective of other staff, however, much of their work involved keeping the EMPOWER trial on the agendas of their colleagues. Maintaining their familiarity with trial processes that were complex was crucial, and this was helped by the co-operation and support from fellow staff:

I’ve got lots of support though we’ve got a really good research team here and everybody kind of helps out each other. So always kind of felt like we’ve had plenty of support during the whole thing.

TS05

At some sites, because the AUs were so busy, research staff were involved in administering i.v. fluids and obtaining the medication. Good relations were fostered when research staff understood the pressures that clinical staff were under and the added burden placed on them by participation in trials:

. . . they’re working flat out and we’re constantly asking them to remember something new and something extra and doing extra jobs. So the more you can kind of give back and take off them in terms of their workload then yeah like you say the better the relations between the two are.

TS20

There were accounts that when trial staff did manage to recruit a woman to the trial, there was a real sense of achievement, despite the fact that it required a lot of effort and teamwork:

The other day we did get the woman in the study there were three members of the research team that stayed 2 hours late and one of the consultants who stayed an hour over to finally get somebody in. So we certainly feel like we’ve gone above and beyond . . .

TS20

Trial-specific factors

The issue of adherence to the RCOG guidelines12 was raised as a hurdle to recruitment, but this was not necessarily the case if women were approached by research staff before second-line treatment was prescribed. The problem arose when women attended AUs providing out-of-hours or 24-hour access at times when research staff were not available. One solution was to raise awareness with clinical teams along the care pathway so that women could be signposted to the trial. However, publicising the trial among primary care clinicians was generally not felt to be the way forward because of the scale of the task.

Operational difficulties at the start of the trial were to be expected, but the smooth working of the trial was generally facilitated by good staff and professional relationships between maternity and gynaecology departments, and with the Clinical Trials Unit. At most sites, there was a dogged determination by lead staff to work hard at liaising with colleagues across departments and educating them about the trial and the need to follow RCOG step-up guidelines. 12 Nevertheless, the key issues for all research staff were the unpredictability of women presenting to services and the infrequency of their presentation. These barriers affected the efforts of research staff to recruit women in the best way possible.

The demands of the complex trial design added to the recruitment challenges faced by staff. Staff spent most of their time finding ways and means of maintaining the profile of the trial; reminding their colleagues about trial process requirements; and tracking, identifying and screening women. Recruiting women was made more challenging because women with severe NVP were physically and emotionally vulnerable; therefore, time and patience were required. Recruitment to the EMPOWER trial involved more than just paperwork; it required high levels of empathy and patient care. Time was needed for the patient to consider the information carefully before making an informed decision about consenting to participate or not. Continuity of care for trial participants was provided and, where appropriate, trial drugs were prescribed as patients requested. In this way, the issue of unblinding did not seem to be as problematic an ethics issue as when first considered.

Women who were eligible to participate were discouraged at the thought of their condition not improving because of the risk of receiving a double dummy. Their experience of severe NVP had an impact on their decision; all that they could think of was a quick, effective solution. The effort and time required for participation, to read and understand the information, and to wait to be entered into the trial were significant for several women. However, for altruistic reasons, there were women who willingly participated to help others.

Site-specific factors

The care pathways that women took in seeking help for NVP were dependent on the trial site, the catchment area and the health service provision therein. Ethnic composition, health behaviour and antidepressant use were factors at each site. Trial sites that recruited more women were characterised by primary care providers that tended not to prescribe trial drugs as first- or second-line antiemetics. Sites that were unsuccessful were those that had to work hard to change the culture of prescribing antiemetics within secondary care. Relationships with A&E departments varied across trial sites; A&E departments in which research was not a priority was a factor. The availability and engagement of GCP-trained staff also varied across sites. Staff confidence and morale varied depending on recruitment rate, recruitment success, which ranged from 11% to 63% of eligible patients.

Topic guide for interviews with clinicians

Interviewer outlines the purpose of the interview, which is to identify problems of recruitment into the trial.

Interviewer explains that the participant will receive a copy of the interview transcript and a summary of the results if they would wish to have one:

• Explain use of the audio-recorder – the interview is being audio-recorded to provide an accurate account of what the participant has said.

• Explain that interviews will be anonymised when they are typed up prior to analysis (i.e. names and any other information that could identify them are taken out and replaced with pseudonyms or anonymous identifiers).

• Assure confidentiality.

• Ask whether they have any more questions about study?

• Go through and sign consent form.

The semistructured interview approach adopted in this study focuses on encouraging the participant to explain the issues that they feel are relevant in an order that makes most sense to them. Apart from the introduction and conclusion, the questions listed below therefore act as an aide memoire rather than a definitive list of questions that are asked in a particular order.

Introduction

1. Can I just ask some background questions if you don’t mind? How long have you been in your present post?

2. Can you tell me about your experience of being involved in clinical trials?

3. What has your experience been of the set-up of the EMPOWER trial?

4. How does it compare with your previous experience of other trials?

5. Can you describe how you go about recruiting patients to this trial? (MAU, EPA unit, A&E department, pathway.)

6. What do you see are the main difficulties of recruiting patients in this trial?

7. Are there any improvements that you feel could be made to the trial processes?

8. What is your impression of patients’ understandings of a) the different treatment arms and b) being randomised in a trial?

9. How has the progress of the trial at your site been so far?

10. Have you had any feedback from patients about the trial?

11. Are there any particular cases/events you would like to highlight to the trial investigators?

12. Do you have anything else you would like to add?