Integrating specialist palliative care into residential care for older

people: a stepped wedge trial (INSPIRED trial)

INSPIRED: INtegrating Specialist Palliative In REsiDential care

Public title: specialist palliative care in residential aged care

Investigator team

Chief investigator: Prof Liz Forbat

Co-investigators: Nikki Johnston, Michael Chapman, Wai Man Liu, Clare Lovell, Prof Meera Agar,

A/Prof Deb Parker, Prof Lawrence Lam.

Protocol version 2. 7th December 2016

Contents

Executive summary 1
1 Introduction 2
1.1 Pilot work 3
1.2 Policy Context 4
2 Aim 5
2.1 Context and setting 6
2.2 Theoretical framework 7
3 Intervention components 8
4 Methods and analysis 10
4.1 Design 10
4.1.1 Blinding and randomisation 11
4.2 Setting, sample and recruitment 13
4.3 Sample size calculation 14
4.4 Outcome measures 17
4.5 Data collection 18
4.6 Analysis plan 22
4.7 Ethics and dissemination 23
4.8 Steering committee 24
4.9 Discussion 24
5 References 26
6 Appendices: Outcome measures 28
6.1 Quality of Death and Dying Index 28
6.2 Capability of Adopting Palliative Approach (PEPA questionnaire) 29
6.3 Brief assessment scale for caregivers 30
6.4 Simple data extraction 32
Executive summary

This project aims to address core goals of the National Palliative Care Strategy, by integrating

specialist palliative care into aged care. The intervention is a new model of care: whereby specialist

palliative care nurses will help facilities prioritise residents who have palliative care needs. The

prioritisation will occur through ‘needs rounds’ which are staff-only meetings to discuss residents

who may die in the next six months. Education to staff will also be provided.

Following pilot work in 2014-5, this stepped wedge trial will seek to establish whether a new model

of care will:

a) reduce length of stay in hospital, leading to significant cost savings

b) improve residents’ ability to die in their preferred place

c) improve care staff understanding of death and dying, and staff capacity

d) reduce symptom burden at end of life

e) reduces relatives' distress.

Page|1
1 Introduction

Residential Facilities for older persons (hereafter “residential facilities”) are increasingly involved in

supporting older people at end of life; consequently, adopting a palliative approach is required (Teno

2003, Miller 2015), and indeed is considered a basic human right (Ahmedzai et al.). In Australia,

improving the quality of life for residents approaching end of life has focused on staff providing

primary palliative care. The Government funded Palliative Approach Toolkit, rolled out in October

2013, provides resource materials designed specifically for the residential facility setting to support

staff in a number of key end-of-life processes including advance care planning, and palliative care

case conferencing (Parker & Hughes 2010).

Despite this appropriate focus on supporting a palliative approach to care, some residents will

require specialist support (Quill & Abernathy 2013). How to access such support is complicated by a

lack of clear data on the appropriateness of different models of specialist palliative care provision.

A recent rapid review found that specialist palliative care models broadly lacked uniformity (Luckett

et al. 2014). Although models varied, frequently cited elements included case management, shared

care, specialist outreach services, managed clinical networks, integrated care and volunteers.

Luckett’s review suggests that models integrating specialist palliative care into primary and

community care should be needs-led, with a particular focus on transitions across care settings, and

adopting dynamic approaches flexible to the changing needs of patients.

However, there is an absence of clear and tested models of how to provide efficient quality specialist

palliative care services to residential facilities and significant variation in practice exists both within

Australia and internationally (Kaasalainen et al. 2013, Luckett et al. 2014).

Page|2
The elements and outcomes of specialist palliative care integration into residential facilities remain

unknown. The importance of staff education in improving palliative care in residential facilities

suggests its utility as a component of any proposed model (Miller 2015). Increasing death literacy is

an important component, in supporting staff, relatives and residents to engage in conversations

about death and dying in order to develop informed plans for end of life care (Noonan et al. 2016).

This is underpinned by a public health approach to palliative care (Kellehear 1999, Kellehear &

Sallnow 2012), whereby there is provision of education and information for death/dying to

‘promote optimal health even in the presence of incurable disease’ (Rosenberg & Yates 2010 p206).

A recent systematic review identified a range of positive impacts from specialist palliative care

involvement in residential facilities including an increased incidence of advance care planning,

decreasing hospital admissions and improved symptom management (Cimino & McPherson 2014).

specialist palliative care services are also demonstrated to increase in-facility deaths for residential

facility residents (Rosenwax 2006). Although models involving a nurse practitioner providing

palliative care consultation support within residential care have reported sustained benefits

(Finucane et al. 2013), in practice the professional mix of clinicians varies.

1.1 Pilot work

Developing models of care that provide effective and sustainable access to specialist palliative care

for residents whose needs require them is therefore a matter of urgent clinical need. In 2014-2015,

we developed and tested a new approach to integrating specialist palliative care into residential

care. The approach led to decreased length of hospitalizations and increases in residents dying in

their preferred place (Chapman et al. 2016). The approach was found to also assist staff in

normalizing death and dying, while providing essential anticipatory prescribing and better decision-

making leading to planned care for residents (Johnston et al. 2016). However, this study focused

only on four facilities, and applied a quasi-experimental design with decedents from the previous

Page|3
year acting as the control group. The current study seeks to expand the provision of the model and

produce an evidence base for its impact.

1.2 Policy Context

All four goal areas identified in the National Palliative Care Strategy (Commonwealth of Australia

2010) are relevant to providing specialist palliative care in residential care and can be addressed

through the study described in this protocol. Specifically:

 Awareness and Understanding. There is a need to significantly enhance the understanding of

dying, death, grief, bereavement and loss in the residential care sector. Notably, this is in
need of development for staff and relatives. Death literacy, as noted above is a component
of this alongside more sophisticated understandings of disease process at end of life.
Advance care planning is also key to the goal of increasing awareness and understanding.

 Appropriateness and Effectiveness. Care should be informed by the right approach, in the
most suitable setting in a timely manner. Notably this means providing care within facilities
and reducing unnecessary admissions to acute care. Achieving this goal is predicated on the
accurate completion of advance care planning documentation, and discussing goals of care,
to ensure that residents’ wishes inform what care they want and where that should be
delivered. A key action area is to support the evolution and evidencing of innovative models
of service provision, alongside operationalising the PA Tookkit approach to triggering
specialist palliative care referrals.

 Leadership and Governance. Collaborative, proactive, effective governance is required. This

includes facilities developing standard operating procedures which support residents to die
in their preferred place, and access high quality care from specialists when required. It also
means a joined-up approach between the providers of primary palliative care (facility staff
and GPs) and specialist palliative care.

 Capacity and Capability. Appropriately skilled people in the right place with adequate
systems to support safe, quality care are required. This includes specialist palliative care
clinicians to deliver specialist interventions for people within residential care. The strategy
indicates that there may be enhanced roles for aged care providers in palliative care, and
Page|4
 this project operationalises that by enhancing aged care providers knowledge/skills and
providing an in-reach model from the specialist team including education and support to
provide primary palliative care. Hospital avoidance is also named as a core outcome, and this
project seeks to provide robust evidence for reducing hospital length of stay.
A further connection with the national strategy from this work comes from the ability to
action the goal around enhancing and legitimising the role of specialist consultancy services.
The pilot work was able to evidence this goal in action and we anticipate the proposed trial
will provide further evidence on how the model operates and its outcomes.

2 Aim

This study seeks to establish evidence for the effectiveness of the model of integrating specialist

palliative care into residential facilities, using a prospective stepped wedge design.

The model as tested in the previous study will be delivered to a greater number of facilities with

additional outcome measures, in order that the trial is fully powered.

Research questions:

1. Does integrating specialist palliative care into residential care reduce hospital length of stay?

2. Does integrating specialist palliative care into residential care increase residents dying in

their documented preferred place of death?

3. What is the impact of integrating specialist palliative care in residential care have on

residents’ quality of dying?

4. What impact does the integration of specialist palliative care in residential care have on staff

death literacy?

5. What impact does the integration of specialist palliative care in residential care have on

complaints?

6. What impact does the integration of specialist palliative care in residential care have on

relatives’ self-reported distress?

Page|5
Null hypotheses:

 Integrating specialist palliative care into residential care does not reduce hospital length of

stay

 Integrating specialist palliative care into residential care does not improve residents dying in

their preferred place.

 Integrating specialist palliative care into residential care does not improve quality of dying.

 Integrating specialist palliative care into residential care does not improve staff death

literacy.

 Integrating specialist palliative care into residential care does not reduce the number of

complaints.

 Integrating specialist palliative care into residential care does not improve relatives’ distress.

2.1 Context and setting

The intervention will be delivered in the Australian Capital Territory (ACT). Specialist palliative care

staff will include a nurse practitioner and a clinical nurse consultant, supported by a palliative care

medical specialist. All team members are employed by local government health funding. The staff

are based in the specialist palliative care unit. There are currently 26 facilities in the territory, with a

further facility due to open in early 2017. Four facilities were involved in the pilot work and

consequently will not be included in this study as collection of uncontaminated baseline data will not

be possible. A further site will be used as a training site whereby a clinician not currently delivering

the model in full will be supported to learn and implement the approach.

Facilities range in size, from 20-120 residents (mean: 94) with a total of 2,247 beds. Based on 2015

data, the number of deaths of residents referred to specialist palliative care anticipated per 12

months is 205 and number of admissions to specialist palliative care per 12 months is 277.

Page|6
2.2 Theoretical framework

The service is provided on the basis of risk stratification and specialist need. Following service

organisation and delivery models in other disciplines, we adopted a case-finding (Lewis et al. 2011)

stepped-care model (Richards et al. 2012). Residents with more complex needs receive higher levels

of specialist input. Complex needs in this context are defined as residents with a high degree of

uncontrolled symptoms, risk of unmanaged symptoms, risk of unnecessary hospital admissions in

the last six months of life due to a lack of planning conversations about goals of care, relatives who

are highly distressed. With approximately 2,200 residents and only two staff, identification of those

most in need of specialist care is essential in the equitable and efficient distribution of their time and

skills.

As illustrated in Figure 1, the majority of residents’ care needs will be met by the registered nursing

staff and carers with Cert 3 or 4 in aged care, team leaders, enrolled nurses within the facilities, with

input from general practitioners. A smaller number will require primary palliative care which can be

provided by general practitioners and facility nurses focusing on life-limiting illnesses and the frail

elderly. The smallest cohort will require input from specialist palliative care, involving nurse

practitioners, clinical nurse consultants and medical consultants with expertise in palliative care. Risk

prediction is built into the model being trialled in this study (see section ‘description of the

intervention’ for further details).

Page|7
 High risk:
Direct support
from specialist
palliative care

Moderate risk: Indirect support

from specialist palliative care

Low risk: Primary palliative care

Figure 1: Stratified model of clinical need for residents

3 Intervention components

The intervention model has four components. As Figure 1 indicates, there are both direct and in

direct supports, with the majority being indirect.

Indirect support

1. Needs rounds. Needs rounds are monthly 60minute triage (and risk stratification) meetings,
where up to ten residents with a short prognosis and high symptom burden are presented
by facility staff to the specialist palliative care clinician. Residents do not attend needs
rounds. Discussions focus on treatment/symptom management planned. Case-based
education is integrated into the needs round, with each resident’s bio-psycho-social status
discussed to promote symptom management, and identify opportunities to extend and
reinforce staff knowledge. The markers for a prognosis of six month or less, taken from the
Palliative Approach toolkit (Parker & Hughes 2010), are used to identify residents to be
discussed in needs rounds:
 Answering “no” to the question would you be surprised if the resident died within
the next 6 months
 Significant functional or medical decline

Page|8
  Conflict concerning whether goals of care should be curative or palliative in intent
(e.g. after an acute event)
 If the resident is transferred or admitted to the facility specifically for comfort or
palliative care
Needs rounds frequently lead to initiating the other elements of the model.

2. Case-based education for staff. There are multiple opportunities for educating staff within
the model that the specialist palliative care team engage in. Specifically, education may
focus on communicating with residents, relatives and GPs, symptom management
approaches, identifying deteriorating patients.

3. Case conferences. These meetings between facility staff, resident, relatives and relevant
health care providers (including, but not limited to the GP, geriatrician, dementia services)
and are facilitated by either the GP or facility staff. Case conferences are attended and
facilitated by specialist palliative care staff at the beginning of the intervention to model
how to conduct the meetings. As the facility staff gain confidence, they take over the chair
function, and specialist palliative care are involved only when resident/family have complex
physical, psychological or social needs. The agenda focuses on resident and/or family
concerns, moving to discussion of how quality of life could be improved, goals of care
(including discussion about hospitalization), and completion of an advance care plan.
Medications and care plan may be changed as a consequence of these discussions. The
facility staff explain their capacity to achieve the goals which have been described.

Direct support
4. Clinical work with relevant residents. Both needs rounds and case-conferences may lead to
referrals for direct specialist palliative care clinical work with residents. Clinical referrals may
also arise from other residents who have not been discussed in needs rounds or case
conferences. Direct clinical work involves: symptom assessment, diagnosis, symptom
management. At times, direct clinical work with residents also included Chairing case-
conferences. However, as the model proposes that these are primarily driven and led by the
facilities, they are primarily conceptualised as indirect care. Only the most complex case-
conferences are Chaired by the specialist palliative care team, for example where staff
report that family conflict over goals of care requires specialist intervention to explain how
symptoms and end of life can be supported.
Page|9
Usual care: Usual care involves access to the specialist palliative care team’s two nurses who work in
residential aged care. Their role includes elements 3&4 of the model described above. Critically
though there is no embedded ‘triage’ element in the form of the needs rounds and limited case-
based education for staff. Essentially, the current model of care (usual care) is reactive, whereas the
trial intervention is proactive and anticipatory.

4 Methods and analysis

4.1 Design

This is a stepped wedge trial with an embedded process evaluation. Stepped wedges are recognised

as a beneficial design for trials involving a palliative population, where there are no clinical equipoise

considerations, and where sequential roll-out of the intervention is beneficial. An integrated process

evaluation will also be conducted, to elicit perceptions of the model from the point of view of

residents and relatives.

The trial will be conducted with six steps, with a cluster size of two facilities in each step, as

illustrated in Figure 2. The staggered commencement timings and intervention length (8–60 weeks),

supports feasibility while maintaining the rigour of the study. This design allows research staff to

collect robust data in all sites, and the clinicians to change over from usual care to the new model

with facilities. This maximises consistency of implementation of the model and avoiding managing

change at 12 sites across the ACT simultaneously. This design also avoids having control facilities that

get no intervention, which could make it difficult to recruit facilities, and ensures equity.

P a g e | 10
4.1.1 Blinding and randomisation

This trial will not be blinded to facilities and clinicians, but outcome assessment will be blinded, as

the research assistant responsible for data inputting and the statistician will not be aware as to

resident and facility status. Simple randomisation will be used, with facilities allocated a unique code

at the outset of the project. Allocation concealment and sequence generation will be managed

through an internet based programme (www.sealedenvelope.com) which will randomly select

facilities for each step. Once randomisation has been conducted, facilities will be informed of the

timing of their facility’s migration from control to intervention condition. An independent data

monitoring committee will be not be convened as the primary outcomes are related to quality of

death and staff education rather than reducing risk of major adverse events (FDA. 2016).

A process evaluation will also be conducted to understand the impact of the trial on usual care,

management and internal processes. The process evaluation will be qualitative and conducted

toward the end of the trial.

P a g e | 11
 Randomisation of all sites

Step 0 Step 1 Step 2 Step 3 Step 4 Step 5 Step 6

Intervention: 0 Intervention: 2 Intervention: 4 Intervention: 6 Intervention: 8 Internvention: 10 Intervention: 12
Control: 12 Control: 10 Control: 8 Control: 6 Control: 4 Control: 2 Control 0

Commence Commence Commence Commence Commence Commence

establishment in establishment in establishment in establishment in establishment in establishment in
Step 1 sites Step 2 sites Step 3 sites Step 4 sites Step 5 sites Step 6 sites

Lost to follow-up Lost to follow-up Lost to follow-up Lost to follow-up Lost to follow-up Lost to follow-up Lost to follow-up
n=x n=x n=x n=x n=x n=x n=x
analysed: n=x analysed: n=x analysed: n=x analysed: n=x analysed: n=x analysed: n=x analysed: n=x

Process evaluation → → → → → → → →

Figure 2: Participant flow (2 staff over 22 months shown)

P a g e | 12
4.2 Setting, sample and recruitment

Recruitment will proceed through residential aged care facilities in the Australian Capital Territory.

There are 26 such facilities, four of which will be excluded as they were involved in developmental

work and are therefore considered contaminated for the purposes of measuring outcomes from a

true baseline. A further facility will be excluded, as it will be a training site, where clinicians not

currently using the model will be trained to apply it (e.g. through modelling needs rounds, being

observed running needs rounds and engaging in case-based education).

The specialist palliative care team provides both in-patient and outpatient services to residents

across the Territory. It is staffed by a multi-disciplinary team including medical, nursing, allied health,

social work and pastoral care practitioners. Recruitment will proceed through the community

palliative care team who specialise in aged care work, who provide specialist palliative care to

patients and carers where their symptoms cannot be managed by primary palliative care.

Inclusion criteria for facilities in the trial:

 Residential facility operating in the Australian Capital Territory from January 2017-
December 2018.
 Senior management agreement to access residents’ records.
 Staff working at the collaborating facilities who are registered nurses or assistants in
nursing (notwithstanding the important role of the whole multi-disciplinary team, these are
the staff groups who provide most direct care).
 Residents whose data is collected must have the facility as their main residence (thereby
excluding those on respite stay).

Inclusion criteria for staff completing outcome measures

 Staff who by virtue of their role, are invited to Needs Rounds meetings. This will include:
team leaders, registered nurses, enrolled nurses, site managers, and some carers.
 Over 18 and willing/able to give informed consent.

Inclusion criteria for relatives completing outcome measures

 Relative of a resident in one of the participating facilities.

P a g e | 13
  Over 18 and willing/able to give informed consent.

Inclusion criteria for the process evaluation:

Recruitment for the process evaluation will proceed through sampling of residents or relatives

whose care was considerably shaped and changed by the palliative care team. Five staff, patients or

relatives will be recruited per site. Inclusion criteria are:

 Resident able to provide informed consent, as assessed by the clinician, using the mini-
mental state examination, cut-off score of ≥20 (Pachet et al. 2010) to inform eligibility.
 Relative or enduring power of attorney of a resident, able to provide informed consent.
 The resident must have had their clinical care discussed at a needs round.
 Care of the resident was considerably shaped and changed, operationalised as the provision
of staff education (for example on death, dying, setting goals of care), or providing advice on
symptom management.

Process evaluation will also include data collection from facility staff. Building on data collected in

the pilot work, senior staff within the facility will be identified for individual interviews to identify

the impact of the intervention on their staff and working practices. Inclusion criteria are:

 Care managers, educators or general managers who have attended needs rounds.
 Willing and able to give informed consent.
 Aged 18 or over.

Participant timelines and flow through the trial component of the study are illustrated in Figure 2.

Baseline data collection will commence following ethical approvals, with recruitment commencing

two months later. An establishment phase is included in the flow of the trial to allow for initial

meetings to be set-up for commencing the intervention (e.g. scheduling needs rounds).

4.3 Sample size calculation

The sample size has been estimated taking into consideration of the study design as a stepped-

wedge randomised trial and the primary outcome of the study is the length of hospitalisation when

P a g e | 14
participants are admitted to acute care. Results obtained from the pilot study suggest that the

intervention could achieve a moderate effect size of 0.6 with a means difference in length of stay of

1.8 days (pooled S.D.=2.9). The calculation is also based on the conventional two-arm RCT for

estimating the unadjusted sample size. The calculation indicates that the study requires an

unadjusted sample of about 41 patients in each arm to provide 80% power at a 2-tail significance

level of 5% with an intervention effect size of 0.6 and the ratio between the intervention and control

groups is 1 to 1. It is further assumed that there will be an attrition of 10%, thus to total unadjusted

sample size required for the study is about 90.

Taking into consideration that this is a stepped-wedge randomised trial, the sample size calculated

above will be adjusted with the stepped-wedge design effect (DEsw) as proposed by Woertman et al.

(2013). 1 + ρ (ktn + bn -1) 3(1 – ρ)

The design effect (DEsw) = ●
1 + ρ (1/2ktn + bn -1) 2t (k- 1/k)

Where, k = 6 (number of steps)

n = 20 (the average number of patients within each cluster)
ρ = 0.05 (Intra-Class Correlation within cluster)
b = 1 (number of baseline assessment)
t = 1 (number of treatment measurement after each step)

Hence, the design effect is calculated as 4.55, thus the total sample required for the study is about

410, but a greater number offers greater power for the statistical analysis.

The following tabled provides information on the estimated number of patients available for

recruitment. As shown there will likely be a minimum of 720 potential residents for recruitment and

thus providing more than sufficient individuals in the pool (minimum numbers have been used to

ensure there is no inflation to the power calculation). The overall sample size is higher to account for

relatives’ responses to a questionnaire and staff completing education outcome measures.

P a g e | 15
Table 1: Sample size estimates for the trial and total resident population

Population Likely minimum total for

(all facilities) the trial1
(% of all facilities)
Facilities 22 10 (45%)
Patients within facilities 2500 920 (37%)
Range of patients per facility 20-160 60-140
Deaths/annum 205 78 (38%)
Number of needs round being 22 (one per facility) 10 (one per facility) (45%)
conducted/month
Number of residents discussed 150 60 (40%)
at needs rounds monthly
Number of residents discussed 1800 720 (40%)
at needs rounds across all sites
annually

At current rates we anticipate 78 deaths per annum within the ten sites of residents discussed at

needs rounds. However, as the study progresses this is predicted to increase, as the number of

residents referred to palliative care increases as people near end of life are given appropriate

support which they are not currently accessing.

With no pilot data for the relatives’ involvement in completing outcome measures, there is no data

from which to estimate completion rates (see item 5 in Section 4.4 for further details on this data

collection).

Sample size for the process evaluation are 15 relatives and 15 residents across all participating

facilities. This sample size will allow for theoretical sufficiency, and be adequate to identify and

saturate core themes and findings (Baker & Edwards 2012).

1
This Table, along with Figure 2 illustrating participant flow is based on 12 sites commencing the
trial, in order to build in some flexibility. We anticipate 12 sites commencing the trial and 10
completing, and hence have over-recruited to allow for withdrawals.

P a g e | 16
4.4 Outcome measures

The primary outcome measure assesses quality of death and dying. The outcome measures follow,

where relevant, recently published European Association of Palliative Care White Paper reporting

outcome measure use in palliative care (Bausewein et al. 2016).

Primary outcome measure:

1. Length of hospitalisation for residents. This will be measured in days (with an

admission of <23 hours 59minutes constituting one day, and >24hours constituting

two days, and so on)

Secondary outcome measures:

2. Preferred place of death will be measured by recording recorded preference and actual

place of death. Data will be recorded as a binary of yes/no, regarding whether preferred

place of death was achieved.

3. A 9-item questionnaire measuring staff death literacy and knowledge will be used. This

validated scale is used routinely in PEPA (Progam of Experience in the Palliative Approach).

This widely used (though as yet unpublished) tool has been used with participants from

varied disciplines, including considerable use in residential care settings. The tool is formally

known as the Capability of Adopting Palliative Approach. Preliminary analyses of

psychometrics were conducted, identifying it as a uni-dimensional scale. Internal consistency

reliability is very high with a Cronbach’s alpha of 0.95, and Split-half reliability, via Spearman-

Brown split-half coefficient at 0.93 (He 2016). Appendix 6.2 shows this scale.

4. Based on the original 31-item version of QODD (Curtis et al. 2002, Curtis et al. 2013) we will

use the revised short-form version by (Downey et al. 2010). The 18 item QODD is a

questionnaire examining four correlated but distinct domains: symptom control,

P a g e | 17
 preparation, connectedness and transcendence. The decedent’s experience is rated on a 0-

10 scale, where higher scores indicate a better experience. Following correspondence with

the scale’s originator, one item has been excluded as it asks about access to assisted-suicide.

Appendix 6.1 shows this scale.

5. Family views on care will be collected using the Brief Assessment Scale for Caregivers (BASC)

of the Medically Ill. This is a 14-item measure, where items on the impact of caring and

caregiver distress are rated on a 5-point Likert scale. The scale has good psychometric

properties with a Cronbach’s alpha of 0.70, and 0.8 for the negative personal impact sub-

scale. Appendix 6.3 shows this scale.

6. A further bespoke measure will capture core data on facilities and residents, including the

Aged Care Funding Instrument (ACFI) score. The ACFI provides a rating (A-D) based on a set

of scores assessing the needs for ongoing complex nursing procedures and activities

(Australian Government Department of Health and Ageing). Charlson score will also be used.

7. Recruitment and retention rates will be reported.

4.5 Data collection

Outcome measures will be gathered at the start of each step in the roll-out. Figure 3 (Brown &

Lilford 2006) illustrates six of these steps; our study will also collect follow-up data from the final site

to receive the intervention which is not indicated in the figure below.

P a g e | 18
Figure 3: Data collection points (Brown & Lilford 2006 p2)

Stepped wedge designs place greater burden on data collection (Kotz et al. 2012). Consequently, the

measures have been selected to reduce participant burden, by capping the length and number of

measures used. We have generously estimated that measures may take a total of 30 minutes to

complete. These time estimates are based on the pilot work conducted in 2015 (Chapman et al.

2016, Johnston et al. 2016).

i. The primary outcome measure (hospitalisation length of stay) will be collected on all

residents during the course of the trial, as well preferred place of death.

ii. The 9-item PEPA measure will be sought from staff working in the facilities who by virtue of

their role would be expected to attend Needs Rounds. At the outset of each Needs Round

staff will be asked to complete the tool. Staff will only complete it once at baseline, but at

the start of all needs rounds, attendees will be asked if they have completed the tool, to

capture staff joining for the first time. Paper copies will be provided to staff, and returned to

P a g e | 19
 the clinician attending the Needs Round. A separate list of staff names who have completed

the tool will be kept in order to identify respondents to complete the tool again at the end of

the study. Paired data will be collected to facilitate analysis, though unmatched analysis may

be required if there is substantial changeover in staff and those who completed baseline

measures no longer work for the facility.

iii. The Quality of Death tool (QODD) will be collected on all residents who die during the course

of the trial. Though the tool is designed for relatives to complete, the burden which this

places on relatives (who may not have been involved and aware of the resident’s end of life

experience) was considered too great for this study. Registered nurses and experienced

team leaders in each facility will be provided with training on completing the tool from the

research team. Having the same staff complete the QODD is a strategy to reduce variability

(and increase reliability) in use of the measure. Between four and six staff per facility will be

trained to account for absences due to annual leave, sick leave and staff turnover. Training

on the tool will involve discussion of the items and then completing the QODD for three

case-studies devised by the specialist palliative care team of recent deaths, to provide

reassurance on reliability of completion. Further training will be provided to staff where

necessary.

iv. Family views on care will be collected within facilities on a paper-based survey. The survey

will be placed at reception, alongside the Family Survey Information Sheet. Families will be

invited to complete the survey by reception staff; the information sheet will be provided

next to the survey. A secure locked-response box will be provided in the reception area for

the collection of completed surveys. The limitation of this data collection method (bias

sample of those motivated to complete the questionnaire and those who visit) is

P a g e | 20
 outweighed by the potential benefit of learning what impact the model of care has on

relatives’ perceptions of caregiving.

v. Resident data will be collected from participating facilities’ records, Pal-Care, CRIS, ACT-PAS,

Concerto (for details of acute hospital admissions) and EDIS (for details of emergency

department admissions). Appendix 6.4 shows the data which will be collected from facilities

and local hospital services. This data will be collected by the study research team.

vi. Fidelity to the model will be monitored. Data will be collected on each clinician’s Needs

Rounds, randomly sampling 20% of Needs Rounds, which will be scrutinised for covering

core elements. Needs Rounds will be recorded for monitoring purposes. Written consent of

all staff participating will be gained. These central elements have been identified in a recent

ethnographic study of Needs Rounds and constitute the core active ingredients (Forbat et al,

forthcoming). This data will be collected by the study research team.

vii. Process evaluation data will be collected via face-to-face interviews where possible, or

telephone (relatives only). Interviews with residents will take place in the facilities. Relatives

may choose if they wish to be interviewed at the facility, their own home or at the specialist

palliative care unit. Joint interviews will be conducted if requested by interviewees, but

recruitment will not proceed to purposefully identify dyads. Dyadic interviewing would only

be appropriate if adopting a case-study or triangulation approach to the data. Interviews

will be conducted when the resident’s facility has been in intervention mode for at least one

month. Interviews will focus on: views and experiences of the quality of care provided,

symptom management, access to prescriptions, quality of life and planning for end of life

care. These areas are derived from our pilot work as those topics most fruitful in

P a g e | 21
 understanding the experience of the new model. Interviews with facility staff will be

conducted at their place of work in a private room. Interviews will focus on perceived impact

of the intervention on death literacy, working practices, staff morale, quality of care, access

to specialist input and quality of death.

viii. Impact on conflict will be measured with reference to the number of complaints received,

and staff perspectives (as per data collection method above integrated into the process

evaluation). The pilot work indicates that the package of work around anticipatory care

(triaging, staff education and case conferences) reduces friction between staff and relatives.

Complaints are a proxy measure for conflict. Staff will be asked on the first of each month

how many complaints were received in the previous month. The definition of ‘complaints’

will be operationalised as a letter/email to the Facility about concerns relating to care

received within the site. Verbal or written statements made to staff indicating that a family

has written to a Minister about care received will also be included in this definition of

complaint.

4.6 Analysis plan

Stepped wedge designs have been proposed fruitful for reducing the necessary sample size

(Woertman et al. 2013).

Given the intrinsic limitation of the study with a high level of turnover for both staff and residents,

that staff and patients in each step at each site are to be considered as independent groups. Hence,

the comparison on the primary outcome, namely the length of hospitalisation, between the

intervention and control phases will adopt an independent two-group comparison approach for the

overall intervention effect. Based on the underlying distribution of the outcome measure, an

P a g e | 22
appropriate comparison technique will be applied parametrically or non-pararmetrically. A similar

approach will also be adopted for the analyses of secondary outcomes.

Economic analysis of anticipated cost savings will be derived from cost per bed day in local acute

hospitals, comparing the control and intervention time periods.

Qualitative data will be analysed adopting a five-stage process of familiarisation: identifying a

thematic framework, indexing the data, synthesising across respondents and data interpretation to

form key themes (Braun & Clarke 2006). A more fine-grained approach to analyzing data from

patients may be adopted, for example conversation or discourse analysis which has been used to

positive effect with adults with cognitive impairment (Pierce et al. 2015).

Members of the research team will have access to the cleaned data set.

4.7 Ethics and dissemination

Ethical permission will be sought through Calvary Public Hospital Bruce’s HREC, and Australian

Catholic University’s research ethics committee (ACU). To access data regarding hospital admissions

HREC approval will also be sought via ACT Health (for Canberra Hospital records) and National

Capital’s ethics committee.

Any required protocol amendments or adverse events will be reported to each HREC. The trial will

be registered with the Australian New Zealand Clinical Trials Group. Data will be held in locked

cabinets and password protected electronic files. All identifying details will be removed, and

replaced with alpha-numeric identifiers. Data in paper form will be stored in a locked cabinet in a

P a g e | 23
locked office and destroyed after 5 years. Computerised data will be stored in a password protected

file, destroyed after 5 years. No public access to the data is planned.

Criteria for discontinuing the trial are informed by Stallard (2001) whereby data will be monitored

and the trial ceased if there is evidence of adverse events or of a lack of efficacy.

Data entry and coding will be conducted by one research assistant with a sample of 20% checked for

accuracy by a second researcher. Security and storage of data will be managed by the principal

investigator.

Results will be presented to the ACT Clinical Network for Palliative Care, published in peer-reviewed

journals, presented at conferences and a lay summary sent to participating facilities for distribution

to families and residents. Authorship of publications arising from the study will be allocated using

the guidelines for authorship defined by the International Committee of Medical Journal Editors.

4.8 Steering committee

The study has an international steering committee comprising senior academics, clinicians and

Health Care Consumers’ Association. The committee will meet quarterly throughout the study via

email and Skype discussions. A-synchronous feedback will be used primarily, due to the different

time-zones of steering committee members.

4.9 Discussion

This trial seeks to establish whether a model of care integrating specialist palliative care into

residential facilities for older people reduces the length of stay in acute care, improves staff death

literacy, improves quality of death for residents and improves residents dying in their preferred

P a g e | 24
place. Data from the trial is intended to inform national and international governments to improve

the clinical support to residents dying in nursing and care homes.

P a g e | 25
5 References

Ahmedzai SH, Costa A, Blengini C, Bosch A, Sanz-Ortiz J, Ventrafridda V, Verhagen SC &

Oncology IwgcbtESo (2004): A new international framework for palliative care. European
Journal of Cancer 40, 2192-2200.
Australian Government Department of Health and Ageing (2012) Aged Care Funding
Instrument (ACFI): User guide. Commonwealth of Australia, Canberra, Australia.
Baker SE & Edwards R (2012) How many qualitative interviews is enough? Expert voices and
early career reflections on sampling and cases in qualitative research. National Centre for
Research Methods Review Paper, London.
Bausewein C, Daveson BA, Currow DC, Downing J, Deliens L, Radbruch L, Defilippi K, Lopes
Ferreira P, Costantini M, Harding R & Higginson IJ (2016): EAPC White Paper on
outcome measurement in palliative care: Improving practice, attaining outcomes and
delivering quality services - Recommendations from the European Association for
Palliative Care (EAPC) Task Force on Outcome Measurement. Palliative Medicine 30, 6-22.
Braun V & Clarke V (2006): Using thematic analysis in psychology. Qualitative Research in
Psychology 3, 77-101.
Brown CA & Lilford RJ (2006): The stepped wedge trial design: a systematic review. BMC Med
Res Methodol 6, 54.
Chapman MD, Johnston N, Lovell C, Forbat L & Liu WM (2016): Avoiding costly
hospitalisation at end of life: Findings from a specialist palliative care pilot in residential
care for older adults. BMJ Supportive & Pallative Care.
Cimino NM & McPherson ML (2014): Evaluating the impact of palliative or hospice care
provided in nursing homes. Journal of Gerontological Nursing 40, 10-14.
Commonwealth of Australia (2010) Supporting Australians to live Well at the End of Life.
National Palliative Care Strategy. Department of Health Canberra.
Curtis JR, Downey L & Engelberg RA (2013): The quality of dying and death: is it ready for use
as an outcome measure? Chest 143, 289-291.
Curtis JR, Patrick DL, Engelberg RA, Norris K, Asp C & Byock I (2002): A measure of the
quality of dying and death. Initial validation using after-death interviews with family
members. J Pain Symptom Manage 24, 17-31.
Downey L, Curtis JR, Lafferty WE, Herting JR & Engelberg RA (2010): The Quality of Dying
and Death Questionnaire (QODD): empirical domains and theoretical perspectives. J
Pain Symptom Manage 39, 9-22.
FDA. (2016) Guidance for Clinical Trial Sponsors Establishment and Operation of Clinical Trial
Data Monitoring Committees. U.S. Department of Health and Human Services Food
and Drug Administration.
Finucane A, Stevenson B, Moyes R, Oxenham D & Murray S (2013): Improving end-of-life care
in nursing homes: Implementation and evaluation of an intervention to sustain quality of
care. Palliative Medicine 27, 772-778.
He W (2016) Validation of the PEPA tool (Forbat L ed.), email.
Johnston N, Lovell C, Liu WM, Chapman MD & Forbat L (2016): Normalising and planning for
death in residential care: Findings from a qualitative focus group study of a specialist
palliative care intervention. BMJ Supportive & Pallative Care.
Kaasalainen S, Ploeg J, McAiney C, Schindel ML, Donald F, Martin-Misener R, Brazil K,
Taniguchi K, Wickson-Griffith A, Carter N & Sangster-Gormley E (2013): Role of the
nurse practitioner in providing palliative care in long-term care homes. Int J Palliat Nurs
19, 477-485.
Kellehear A (1999) Health Promoting Palliative Care. Oxford University Press, Melbourne.

P a g e | 26
Kellehear A & Sallnow L (2012) Public health and palliative care; an historical overview. In
International perspectives on public health and palliative care. (Sallnow L, Kumar S & Kellehear A
eds.). Routledge, London, pp. 13-29.
Kotz D, Spigt M, Arts IC, Crutzen R & Viechtbauer W (2012): Use of the stepped wedge design
cannot be recommended: a critical appraisal and comparison with the classic cluster
randomized controlled trial design. J Clin Epidemiol 65, 1249-1252.
Lewis G, Curry N & Bardsley MC (2011) Choosing a predictive risk model: a guide for
commissioners in England. Nuffield Trust, London.
Luckett T, Phillips J, Agar M, C. V, Green M & Davidson P (2014): Elements of effective
palliative care models: a rapid review. BMC Health Serv Res 14, 22.
Miller S (2015): Hospice and Palliative Care in Nursing Homes: Challenges and Opportunities
for Enhanced Access. Rhode Island Medical Journal 96, 26-29.
Noonan K, Horsfall D, Leonard R & Rosenberg J (2016): Developing death literacy. Progress in
palliative care 24, 31-35.
Pachet A, Astner K & Brown L (2010): Clinical utility of the mini-mental status examination
when assessing decision-making capacity. J Geriatr Psychiatry Neurol 23, 3-8.
Parker D & Hughes K (2010) Comprehensive Evidence-Based Palliative Approach in Residential
Aged Care. Final Report. In Final report, University of Queensland/Blue Care Research &
Practice Development Centre. .
Pierce S, Lamers C & Salisbury K (2015): Knowingly not wanting to know: Discourses of people
diagnosed with mild cognitive impairment. Dementia (London)
10.1177/1471301215600895.
Quill TE & Abernathy AP (2013): Generalist plus Specialist Palliative Care — Creating a More
Sustainable Model. The New England Journal of Medicine 368, 1173-1175.
Richards DA, Bower P, Pagel C, Weaver A, Utley M, Cape J, Pilling S, Lovell K, Gilbody S,
Leibowitz J, Owens L, Paxton R, Hennessy S, Simpson A, Gallivan S, Tomson D &
Vasilakis C (2012): Delivering stepped care: an analysis of implementation in routine
practice. Implementation science 7, 3.
Rosenberg J & Yates P (2010): Health promotion in palliative care: the case for conceptual
congruence. Critical Public Health 20, 201-210.
Rosenwax AJ (2006): Who receives specialist palliative care in Western Australia - and who
misses out. Palliative Medicine 20, 439-445.
Stallard N, Whitehead J, Todd S & Whitehead A (2001): Stopping rules for phase II studies. Br J
Clin Pharmacol 51, 523-529.
Teno JM (2003): Now is the time to embrace nursing homes as a place of care for dying person.
Palliative Medicine 6, 293-296.
Woertman W, de Hoop E, Moerbeek M, Zuidema SU, Gerritsen DL & Teerenstra S (2013):
Stepped wedge designs could reduce the required sample size in cluster randomized
trials. J Clin Epidemiol 66, 752-758.

P a g e | 27
6 Appendices: Outcome measures

6.1 Quality of Death and Dying Index

Instructions: Please allocate a score for each item for the deceased resident.
Scoring should be given from 0-10, with 0 indicating a terrible experience and 10 indicating an
almost perfect experience.

1 How often did the resident appear to have her/his 1 2 3 4 5 6 7 8 9 10

pain under control?
2 How often did the resident appear to have control 1 2 3 4 5 6 7 8 9 10
over what was going on around her/him?
3 How often did the resident have control of her/his 1 2 3 4 5 6 7 8 9 10
bladder or bowels?
4 How often did the resident breathe comfortably? 1 2 3 4 5 6 7 8 9 10

5 How often did the resident appear to feel at peace 1 2 3 4 5 6 7 8 9 10

with dying?
6 How often did the resident appear to be unafraid of 1 2 3 4 5 6 7 8 9 10
dying?
7 How often did the resident laugh and smile? 1 2 3 4 5 6 7 8 9 10

8 How often did the resident appear to be worried 1 2 3 4 5 6 7 8 9 10

about strain on her/his loved ones?
9 How often did the resident appear to keep her/his 1 2 3 4 5 6 7 8 9 10
dignity and self-respect?
10 How often did the resident spend time with family 1 2 3 4 5 6 7 8 9 10
and friends?
The questions ask whether the event occurred. After each filter question, please rate the
decedent’s experience, using the same 0-10 scale, or chose ‘NA’ if it did not happen.

11 Was the resident touched or hugged by NA 1 2 3 4 5 6 7 8 9 10

her/his loved ones?
12 Were all of the resident’s health care costs NA 1 2 3 4 5 6 7 8 9 10
taken care of?
13 Did the resident say goodbye to loved ones? NA 1 2 3 4 5 6 7 8 9 10

14 Did the resident have one or more visits from NA 1 2 3 4 5 6 7 8 9 10

a religious or spiritual advisor?
15 Was a mechanical ventilator or kidney dialysis NA 1 2 3 4 5 6 7 8 9 10
used to prolong the resident’s life?
16 Did the resident have her/his funeral NA 1 2 3 4 5 6 7 8 9 10
arrangements in order prior to death?

For note to the HREC, the following item has been removed: ‘Have the means to end your life if you
need to.’ Following discussion with the scale developer, we are confident that removing this item
does not impact scale validity, and have been given guidance on how to score the adjusted scale.

P a g e | 28
 6.2 Capability of Adopting Palliative Approach (PEPA questionnaire)

Whilst in my care, I am confident Not at all Somewhat Moderately Very Completely

that I can: confident confident confident confident confident

1. Identify the needs of people

who have a life-limiting illness
2. Identify the interventions
required by people who have a
life-limiting illness
3. Discuss end of life issues with
people who have a life-limiting
illness and their families
4. Implement the interventions
required for people who have a
life-limiting illness
5. Assess the ongoing needs of
people who have a life-limiting
illness
6. Identify personal coping
strategies to effectively manage
personal issues associated with
caring for people who have a
life-limiting illness
7. Discuss the needs of people in
my care who have a life-limiting
illness with palliative care
specialists
8. Refer people with life-limiting
illness to a palliative care
specialist when needed
9. Apply principles of culturally-
centred care when caring for
people who have a life-limiting
illness

P a g e | 29
6.3 Brief assessment scale for caregivers

Please place a cross in the box, to show to what extent you have had any of the following feelings in
the past month because of your relative who lives at this care facility.

Your name: Your relative’s name: Today’s date:

Because of your relative’s health Not at all A little Some A lot Does
condition, how much have you not
apply
Worried about them even when you
are not with him/her
Been depressed because of their illness
Been upset about not having enough
time for yourself because of caring for
them
Been overwhelmed by the
responsibility of caring for them
Please rate your distress during this Not A little Some A lot of Does
past month distressed distress distress distress not
at all apply
Distress over seeing them in so much
pain or discomfort
Distress at not having enough time to
do your job, or other responsibilities
Distress about making decisions about
hospitalizing them
Distress over talking about what
medical procedures they would want
or would not want if their heart or
breathing were to stop
Distress of seeing how much their
illness has changed your relationship
Distress at having strained relationships
with other family members over taking
care of them
Please rate whether you agree or Agree a lot Agree a Disagree Disagree Does
disagree with each of the following little a little a lot not
statements apply
Taking care of them has drawn the two
of us closer together
Taking care of them has brought
meaning to my life
Taking care of them has drawn other
members of our family closer together
Taking care of them makes me feel
good about myself
PLEASE TURN OVER

P a g e | 30
 About this questionnaire

This questionnaire is part of a piece of research on providing specialist support to aged care facilities.

The research is being run by Prof Liz Forbat, who is in a joint appointment with Calvary HealthCare

ACT and Australian Catholic University.

Information collected on this form will be shared with the facility staff, but your details and those of

your relative will not be shared.

Once you have completed it please place it in the locked box by reception. One of the research team

will collect responses each week.

P a g e | 31
6.4 Simple data extraction

For each facility, we need to collect a panel data focused on the following information:

Facility level data

1. Total number of residents
2. Total number of needs rounds
3. Total number of residents discussed at needs rounds
4. Total number of case conferences are triggered by needs rounds
5. Total number of referrals to specialist palliative care that are being received outside of
needs rounds
6. Total number of complex patients that need specialist input
7. Total number of registered nurses
8. Total number of beds
9. Average length of stay
10. Resident turnover
11. Average care needs based on ACFI submission
12. Number of hospital transfers in past month
13. Number of deaths by preferred place
14. Workload change for staff
15. Number of complaints received in the last month

Demographic data for residents discussed at needs round

1. Age
2. Gender
3. Culture and linguistic background
4. Religion
5. Primary Diagnosis
6. Medical record number
7. Chief needs
8. ACFI
9. Charlson score
10. Required specialist input
11. In-round or out-of-round
12. Type of referrals made at needs round (e.g. DBMAS, Physio)
13. Referrals to specialist palliative care that are being received outside of needs rounds?
14. Case conferences held (n)
15. What case-based education is being done (key topics)
16. Place of Death

Acute admissions data (from Calvary Hospital, National Capital and Canberra Hospital)
1. Reason for admissions (DRG)
2. Length of stay

P a g e | 32