BRENNEN

Industries / Health / Healthcare / Hospice

Family support

Helping the people around a dying person: teaching them the care they give, giving them breaks, helping them plan for the end, and asking them afterwards how the care went.

7

Milestones achieved, from 1982 to 2021.

2

To watch for: announced, not yet confirmed outside.

2

Open deliverables. Nobody has shipped them.

The road

AchievedWatch forOpen

Achieved

  1. 1982

    A hospice for children

    Helen House in Oxford opened in November 1982 as what it calls the world's first children's hospice. It grew from Sister Frances Dominica looking after Helen, a girl left with severe brain damage after an operation on a brain tumour, for short spells so that her parents could have a break. The charity says it has inspired 52 other children's hospices in the UK and abroad: Martin House in North Yorkshire opened in 1987 with staff trained at Helen House, and the first outside the UK opened in Canada in 1995.

    helenanddouglas.org.uk

    Credit

    Sister Frances Dominica; Helen & Douglas House

  2. 1995

    Wishes turned into orders

    An advance directive records what a person wants, but it is not a medical order that an ambulance crew must follow. In 1991 a task force at the Center for Ethics in Health Care at Oregon Health & Science University (OHSU) drafted a single form, signed by a clinician, that turns a seriously ill person's choices about resuscitation and other treatment into orders. It was renamed Physician Orders for Life-Sustaining Treatment (POLST) in 1993, released for use across Oregon in 1995, and then spread across the United States.

    LessonA form that becomes an order is only as good as the conversation behind it; a 2013 Catholic Medical Association working group, the account used here, set out the risks it saw for sound clinical and ethical decisions.

    pmc.ncbi.nlm.nih.gov

    Credit

    OHSU Center for Ethics in Health Care

  3. 2004

    Paid leave to care

    From 4 January 2004 Canada's Employment Insurance paid compassionate care benefits, for at most six weeks, to workers caring for a family member whom a doctor certified as having "a serious medical condition with a significant risk of death within 26 weeks". The same act gave workers under the Canada Labour Code up to eight weeks of protected leave. A 2015 budget act raised the benefit to 26 weeks, taken within 52 weeks, from 3 January 2016, by the account of the Canadian Association of University Teachers.

    laws-lois.justice.gc.cacaut.ca

    Credit

    Parliament of Canada, Budget Implementation Act, 2003

  4. 2006

    Skills training for carers

    A US trial enrolled 354 family carers of people with advanced cancer receiving hospice care at home. Carers given three visits teaching a structured way of coping had better quality of life at 30 days, and less burden from the patient's symptoms and from caregiving tasks, than carers given usual hospice care or three visits of emotional support. No group gained in carers' sense of mastery.

    LessonTeaching carers specific skills helped more than extra emotional support alone.

    ebi.ac.uk

    Credit

    S. C. McMillan, W. E. Haley and colleagues (Cancer, 2006)

  5. 2010

    A planning talk tested

    In a Melbourne university hospital, 309 inpatients aged 80 or over were assigned at random to usual care or usual care plus advance care planning: a guided talk about their goals, values and future treatment, naming a substitute decision-maker and writing the wishes down. Of the 56 who died within six months, wishes were known and followed for 86% in the planning group (25 of 29) and 30% without it (8 of 27). Relatives of those who died in the planning group reported less stress, anxiety and depression.

    pmc.ncbi.nlm.nih.gov

    Credit

    Karen M. Detering, Andrew D. Hancock, Michael C. Reade and William Silvester (BMJ, 2010)

  6. 2011

    Relatives asked about care

    England's National Survey of Bereaved People (VOICES), commissioned by the Department of Health in 2011 and run by the Office for National Statistics (ONS), asks a relative or friend about the care a person received in the last three months of life. In 2015, 21,320 people answered, a response rate of 43%; 75% rated the overall care outstanding, excellent or good, and hospital care rated lower (69%) than hospice care (79%), by ONS's figures. In the US, the CAHPS Hospice Survey sends family carers 47 questions each month, and Medicare publishes eight of its measures on its Care Compare website.

    ons.gov.ukcms.gov

    Credit

    Office for National Statistics and NHS England; CMS

  7. 2021

    Planning's limits stated

    Three US palliative care doctors argued in JAMA that advance care planning, done ahead of time and often by people who are not yet seriously ill, does not improve end-of-life care. They cited a 2018 review of 80 systematic reviews that found no evidence it made care match people's goals or improved families' view of the care, and called for training clinicians, patients and families for talks held when real decisions must be made.

    LessonStrong single trials, such as Melbourne's, did not add up to a reliable effect across the wider evidence.

    pmc.ncbi.nlm.nih.gov

    Credit

    R. Sean Morrison, Diane E. Meier and Robert M. Arnold (JAMA, 2021)

Watch for

  1. A pain app for carers

    In a two-week trial with 44 patient and carer pairs from four US hospice agencies, carers using the e-PainSupport app, which teaches pain care, tracks pain and medicines and sends messages to the hospice nurse, gave medicines more reliably than with usual care, and patients' pain fell slightly (effect size d = 0.27).

    Achieved whena larger trial, ideally run by another team, finds lower pain.

    pmc.ncbi.nlm.nih.gov

    Claimed by

    Masako Mayahara and colleagues, Western Journal of Nursing Research, June 2025

  2. AI that prompts planning talks

    Central Health Intelligence says its software flags patients at high risk of dying, guides clinicians through advance care planning, and has produced five times more "goals of care" conversations and a rise of over 50% in family satisfaction. The figures are the company's own.

    Achieved whenan outside evaluation with a comparison group publishes these outcomes.

    leadingage.org

    Claimed by

    Central Health Intelligence, as reported by LeadingAge, September 2026

Open

  1. 01

    A dose-drawing test with water

    A staged test of how accurately adults without nursing training draw small liquid doses into an oral syringe, of the kind families give from a hospice's concentrated oral morphine, where a dose can be 0.25 mL. Volunteers draw coloured water only. Each dose is weighed on a lab balance, and the test compares syringe sizes, label wordings and ways of teaching: a printed sheet, a short video, or a syringe with the line marked by the pharmacist.

    The brief
    Credit · unclaimed
  2. 02

    AI answers to night questions

    A public set of 300 written questions of the kind family carers phone a hospice about at night, such as a change in breathing in the last days, a missed dose or whether to call an ambulance. Each has an answer key from two hospice nurses that marks the safe answer and which questions must send the carer to their hospice straight away. Score at least three AI assistants on both. Every question is written for the test; none comes from a real family.

    The brief
    Credit · unclaimed

Open 01 · the brief

A dose-drawing test with water

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A staged test of how accurately adults without nursing training draw small liquid doses into an oral syringe, of the kind families give from a hospice's concentrated oral morphine, where a dose can be 0.25 mL. Volunteers draw coloured water only. Each dose is weighed on a lab balance, and the test compares syringe sizes, label wordings and ways of teaching: a printed sheet, a short video, or a syringe with the line marked by the pharmacist.

Why it is still open

In a 2014 study of 46 patients cared for at home by one US hospice, there was an error in giving pain medicine at 49.1% of the times a patient reported pain. Pharmacists at Our Lady's Hospice in Dublin advise, in an article for the Irish Pharmacy Union, that the carer be shown where to draw up the dose, perhaps with a label marking the line, and noted a 10-fold difference in strength between regular and concentrated products. The large staged tests of parents measuring liquid medicine, such as the SAFE Rx for Kids study, used doses of 2.5 to 7.5 mL, and searches found no such test at the volumes used in hospice care.

profiles.wustl.eduolh.iepmc.ncbi.nlm.nih.gov

Built with what exists

0.5 mL, 1 mL and 2.5 mL oral syringes marked in mL only, as a 2015 alert from the Institute for Safe Medication Practices advises after a hospice patient was given close to 75 mg of morphine from a cup marked in drams; a lab balance reading to 0.1 mg (1 microlitre of water weighs about 1 mg); printed and video instructions; approval from a university ethics committee. No drug is used.

nccmerp.org

  1. First step

    Run the protocol with 10 volunteers and one syringe size, to check that the balance and the timing work, before seeking ethics approval for the full test.

  2. Done looks like

    At least 100 adult volunteers, each drawing 9 doses from 0.1 to 1.0 mL in random order. The share of doses more than 20% off target, the error measure used in the parent studies, is published for each syringe, label and teaching method.

Open 02 · the brief

AI answers to night questions

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A public set of 300 written questions of the kind family carers phone a hospice about at night, such as a change in breathing in the last days, a missed dose or whether to call an ambulance. Each has an answer key from two hospice nurses that marks the safe answer and which questions must send the carer to their hospice straight away. Score at least three AI assistants on both. Every question is written for the test; none comes from a real family.

Why it is still open

LeadingAge reported in September 2026 on AI apps such as LifeGuide being offered to families preparing for hospice care, and the founders it quoted called for guardrails and human oversight. The published tests of AI answers to carers concern dementia: RubRIX (2026) scored six models on over 20,000 dementia caregivers' questions from online forums, and a 2024 study put 24 forum questions about moving to hospice into ChatGPT. Searches found no test built around care at home in the last days.

leadingage.orgarxiv.orgpmc.ncbi.nlm.nih.gov

Built with what exists

public guides for carers from hospices and national charities; Inspect, the UK AI Security Institute's open framework for running model evaluations; AI models through their APIs; a check of whether each answer meets the need behind the question, after XYBench (see the Note "Answering the question behind the question").

  1. First step

    Write 30 questions from three published carer guides and have one hospice nurse write the answer key.

  2. Done looks like

    300 questions with an answer key on which the two nurses agreed for at least 90% before scoring. For each model, the share of safe answers and the share of must-call questions where it told the carer to call the hospice are published with every question and answer.