FENTANYL ALONE PLUS BUCCAL MIDAZOLAM: AN OPEN-LABEL, RANDOMISED, CONTROLLED FEASIBILITY STUDY IN THE DYING

To establish how best to conduct a definitive, randomised controlled trial (RCT) to determine whether NF and BM administered by families, for patients dying at home, lead to faster and better symptom control and fewer community nursing visits than standard breakthrough medication by healthcare professionals. This open-label mixed methods feasibility RCT compared the efficacy of NF and BM by family members to standard breakthrough medication by nurses for the terminally ill in a specialist palliative care unit. Partway through the study, a third observational arm was where BM alone was used. The primary outcomes were whether recruitment and randomisation possible, assessment of withdrawal and drop-out, and whether the methods acceptable and appropriate.


INTRODUCTION
People with terminal illnesses need timely symptom control and should be able to die in their 'preferred place of care' 1 (usually home 2) . Dying patients are often too weak to take oral medication. The mainstay in the United Kingdom (UK) is subcutaneous infusions by syringe driver and top-up medication by subcutaneous injection 3 . Family carers can be trained to give injections [4][5][6] . In the UK, this is less common. Usually, when the terminally ill experience symptoms, a carer calls a community nurse for an injection. It can take hours for nurses to arrive 7 . This is often distressing for patients and families. There can be disquiet about injections 9,10 . There are alternative preparations given more rapidly and easily -(both fast acting) nasal fentanyl (NF) and buccal midazolam (BM). Research has examined NF for breakthrough pain 11 and BM for seizures 12 but not in the dying. In preparation for a community randomised trial of these administration modes, we assessed study methodology.

METHODS
We started recruitment to a randomised controlled trial from December 2016 but gained approval to recruit an additional third observational arm of BM alone from October 2017. All patients could receive standard as needed medication. Hospice inpatients and carers fitting inclusion criteria were identified as possible participants by the clinical team. If the patient had capacity, the study was discussed with them first and then their carer. Where they lacked capacity, the study was discussed with carer only.

Abbreviated Inclusion Criteria:
• Hospice inpatients • Terminal cancer; estimated prognosis 1-2 weeks • Carer/family member willing to give medication AND likely to be at the hospice at least 25% of the time

Study Procedures
Three patient information leaflets AND consent forms were used for: 1) Patients with capacity; 2) Carers; 3) Carers willing to consent on behalf of patients lacking capacity. Potential participants were given information leaflets. After as much time as they wished, they were asked to sign appropriate consent forms. Patients eligble for Groups A and B were randomised via telephone by the sponsoring hospital Research Support Service: • Group A -NF replaced subcutaneous opioids for pain and BM subcutaneous benzodiazepine for agitation. Group A could receive NF four hourly; up to four times daily on a titration schedule. Once effectively titrated, carers could also administer BM four hourly; up to four times daily.
• Patients eligible for Group C -BM replaced subcutaneous benzodiazepine for agitation. Group C could receive BM four hourly; up to four times daily.
Nursing staff could administer trial medication if carers were not present OR not confident.

Adverse Events
As a study in the terminally ill, it was expected death would be frequent. It was reported to the sponsor, but not considered a serious adverse event if, in the Chief Investigator's opinion, it was a natural conclusion to the illness. Deaths did not 8 require immediate reporting to the Medicines and Healthcare products Regulatory Agency (MHRA) or Ethics Committee.

Statistical Analysis
For all quantitative outcome measures, the main aim was feasibility of intervention delivery to help design a main trial. Indicative outcomes were underpowered for statistical interpretation.

RESULTS
There were 337 hospice admissions during the study period (320 individuals). 308 did not meet inclusion criteria. Main reasons not terminally ill; not on a high enough background opioid dose; family not present 25% of the time. Of the 29 eligible patients/carers approached, 9 declined. Of the 20 patients enrolled, 3 completed the study, 8 died, and 9 withdrew (family request -4; adverse event -2; unable to titrate NF -2; discharged home -1). Of 9 in Arm A, 1 died before study drug and 2 withdrew because they could not be titrated on NF (i.e. pain uncontrolled 30 minutes after 800mcg dose). All 9 in Arm B received symptom-relieving medication. Of the 2 in Arm C, 1 did not receive study drug. There were 308 breakthrough episodes requiring medication: Arm A -165; Arm B -125; Arm C -18. There were 85 doses of experimental drug given; 41 post-titration.
Median time from recruitment to death was 7 days; 1 patient lived 119 days. In Arm A, the successful NF dose was 100mcg for 2; 200mcg for 3; 400mcg for 1. Of the 3 given BM (Arms A and C), all responded to 2.5mg. There were missing data for outcome measures but none for dose timing. Results are in Table 1 for the 6 titratable patients in Arm A, 9 Arm B, and 1 in Arm C who received trial medication.
Those successfully titrated on study drugs had faster and longer lasting symptom control than standard medication.

Adverse Events
There was one serious adverse event (wrong dose of study drug).

DISCUSSION
It was possible to conduct a feasibility study in a single hospice. When we planned this study, carers administering symptom medications at home were rare. It was thought safer to conduct a feasibility study in a controlled environment. In the years it took to finalise the protocol, secure funding and approvals, a multi-centre, feasibility, community randomised trial has been conducted comparing subcutaneous medication administered by families versus healthcare professionals 14 .
It was expected many admissions would be ineligible. Half of the unit's admissions go home. Often the families of those admitted are unable to provide care (or would not be present 25% of the time). Patient/family distress or likely inability to safely administer drug were issues in a few. Families approached often wanted the opportunity to give medication. A future community study would only approach terminal patients wanting to stay at home (perhaps making recruitment easier).
Recruitment time was limited by a short expiry date for BM. A substantial amendment was approved for an observational Arm C (BM alone as experimental drug). This led to two patients recruited and while one did not receive study medication, the other did receive symptomatic benefit. There was much missing data. We were wary of burdening patients, families, and busy staff. As the study required caregivers to be present 25% of the time, we anticipated missing data for family assessments. There was much missing nursing assessment data despite research team support including training sessions and 24-hour research team advice.

11
The drugs were largely well tolerated. Of concern was the wrong dose of NF on three occasions by nursing staff; in one incident, four times the correct dose. We classified this as a serious adverse event. The patient was sleepier but otherwise unharmed. No errors were made by families and these incidents confirm how important training and 24-hour support would be in a potential community study. For a future community study, dose timing, number of doses used, and need for rescue medication from community nurses would be the best outcome measures.

CONCLUSIONS:
We hope to use lessons from this research to plan studies to investigate how best to support patients dying at home and their families. One would expect those in a specialist palliative care unit to have the most complex symptoms and families struggling to cope with home care. A future community study would likely recruit more 'normal dying' with easier to treat symptoms and families more able to help.
Our study showed that even amongst the most complex illnesses, patients and families are happy to participate.