Effects of facilitated family case conferencing for advanced dementia: A cluster randomised clinical trial

Background Palliative care planning for nursing home residents with advanced dementia is often suboptimal. This study compared effects of facilitated case conferencing (FCC) with usual care (UC) on end-of-life care. Methods A two arm parallel cluster randomised controlled trial was conducted. The sample included people with advanced dementia from 20 Australian nursing homes and their families and professional caregivers. In each intervention nursing home (n = 10), Palliative Care Planning Coordinators (PCPCs) facilitated family case conferences and trained staff in person-centred palliative care for 16 hours per week over 18 months. The primary outcome was family-rated quality of end-of-life care (End-of-Life Dementia [EOLD] Scales). Secondary outcomes included nurse-rated EOLD scales, resident quality of life (Quality of Life in Late-stage Dementia [QUALID]) and quality of care over the last month of life (pharmacological/non-pharmacological palliative strategies, hospitalization or inappropriate interventions). Results Two-hundred-eighty-six people with advanced dementia took part but only 131 died (64 in UC and 67 in FCC which was fewer than anticipated), rendering the primary analysis under-powered with no group effect seen in EOLD scales. Significant differences in pharmacological (P < 0.01) and non-pharmacological (P < 0.05) palliative management in last month of life were seen. Intercurrent illness was associated with lower family-rated EOLD Satisfaction with Care (coefficient 2.97, P < 0.05) and lower staff-rated EOLD Comfort Assessment with Dying (coefficient 4.37, P < 0.01). Per protocol analyses showed positive relationships between EOLD and staff hours to bed ratios, proportion of residents with dementia and staff attitudes. Conclusion FCC facilitates a palliative approach to care. Future trials of case conferencing should consider outcomes and processes regarding decision making and planning for anticipated events and acute illness. Trial registration Australian New Zealand Clinical Trial Registry ACTRN12612001164886


Background
Dementia is a terminal disease. Care for people with advanced dementia requires a palliative approach that is targeted to the illness trajectory and tailored to the needs of each individual and his/her family. Currently, the quality of care in residential aged care (RAC) is compromised by lack of staff expertise and poor communication between staff, family and health professionals. Residents suffer unnecessary hospitalisations and aggressive treatments, while symptoms often go unmanaged.
Facilitated case conferencing (FCC) is an approach that brings together RAC staff, health professionals and families to plan person-centred management based on best practice. FCC has improved outcomes in other palliative settings but evidence is lacking for RAC residents with advanced dementia.

Objectives
This phase III cluster RCT aims to: 1) compare the efficacy of FCC with Usual care in improving end of life (EOL) outcomes for residents with advanced dementia living in RAC; 2) provide insights into facility-and staff-related processes influencing the implementation and sustainability of FCC; and 3) evaluate the cost-effectiveness of FCC versus Usual care.

Study setting
Twenty RAC facilities (RACFs) in Sydney & Brisbane will be recruited.

Design and Methods
The proposed research will use a phase III parallel cluster RCT design, conducted in 6 stages over 3 years. A cluster RCT design was chosen to minimise contamination between arms when taking a systems-based approach to intervention. Activities for each stage will be as follows:  (e.g. GPs) involved in case conferencing for participating residents will also be eligible to participate.
Outcomes: Primary: Family-rated EOL outcomes will be measured using the End of Life in Dementia (EOLD) Scales.
The EOLD Scales will be used to measure symptom-related comfort during the last 7 days of life validated for rating either by family or nurses, family ratings will be our primary outcome because family perceptions of EOL suffering & its management are important outcomes for palliative care & our intervention's staff focus may lead to response bias in nurse ratings. All scales will be rated 4 weeks following resident death.

Secondary:
a. Nurse-rated symptom-related comfort & symptom management will be measured using the CAD- Aim 3 -Economic evaluation: The incremental benefit of FCC will be calculated in terms of QALYs gained. The cost of the intervention will be measured along with any cost-savings due to avoided healthcare utilisation. Results will be presented in terms of the incremental cost-effectiveness ratio (ICER). Many of these model parameters will not be powered for statistical significance. Therefore, mean estimates of resource utilisation will be used & confidence intervals will be generated by bootstrapping the data. Uncertainty will be explored using probabilistic sensitivity analysis.

RATIONALE / BACKGROUND
More than half (53%) of all Australians in RAC have a dementia diagnosis, 1 and most people with advanced dementia live in RAC. 2 Dementia is a terminal disease, and a palliative approach is fundamental to best practice care for people in the advanced stages. [2][3][4][5] According to the World Health Organisation, palliative care should "improve the quality of life of individuals and their families … by means of early identification and impeccable assessment and treatment of pain and other problems, physical, psychosocial, cultural and spiritual needs". 6 Unfortunately, delivery and quality of palliative care provision for people with advanced dementia in RAC is often suboptimal. 3,[7][8][9][10][11][12][13][14][15][16] What makes a dementia-specific palliative approach critical? Advanced dementia requires a palliative approach targeted to the specific needs of the illness trajectory as well as tailored to the needs of the individual and his/her family. 16 People with advanced dementia are usually unable to communicate their needs or participate in care decisions. Symptom assessment and management is in the context of severe cognitive impairment, symptom profiles are unique, and deterioration is often complicated by other life limiting conditions. 17 Pneumonia and eating problems are more common than in non-demented RAC residents and require focused management and planning. 18 Family members have defined a 'good death' for RAC residents with advanced dementia as requiring symptom management, clear decision-making, preparation, completion, and affirmation of the whole person. 18 Current deficiencies in RAC dementia palliation: While there is substantial evidence to guide dementia-specific palliative care, many RAC staff and other health professionals lack awareness that a palliative approach is required, or else find it difficult to apply to dementia care. 11,13,19,20 Major barriers include deficiencies in expertise, 12,21 poor communication between services and between RAC staff and families, and inadequate planning and/or inconsistency in decision-making. 11,12 Symptoms such as pain often go unrecognised and unmanaged. 12,15,22 Medical problems may be inappropriately managed with aggressive treatments (e.g. intravenous antibiotics/hydration, tube feeding), 13 This pilot has tested the key components to be utilised in the proposed intervention and has demonstrated: 1) acceptability of the advanced dementia-specific CC toolkit, the list of core skills for the coordinator role and associated training materials; 2) feasibility and appropriateness of dementia-specific "trigger points" based on resident characteristics and/or clinical events after which a CC is recommended; 3) acceptability by RAC staff of a palliative care framework that enables them to map the trajectory of residents in specific domains of disease progression (functioning, swallowing, weight loss, continence, level of consciousness, communication, and factors related to comorbid illness) and identify the issues a related CC should address; and 4) feasibility of collecting descriptive data and outcome measures in the proposed evaluative study.
Baseline data collected in the pilot confirms previous findings that, without facilitation, CC does not occur; 44 over a 6-month observation period, less than 20 CCs took place, despite executive support.
Staff communication with families was limited to exchange of information rather than shared decisionmaking. The pilot will be completed in June 2011, with thus far 60 residents meeting the triggers for CC and 10 facilitated CCs successfully undertaken utilising the framework in a 2 month period.

Approach in the proposed research:
This application will bring together the team's experiences in palliative care CC (Agar/Mitchell/Phillips) and PCC-related RAC cultural change (Chenoweth/Beattie/Luscombe) to implement and evaluate a model of sustainable FCC for residents with advanced dementia in RAC. Training and materials in CC and palliative care described above 40,43 will be implemented together with an adapted version of the PCC training and support package evaluated by Prof Chenoweth. 36,37 Our pilot work highlights that a "bottom up" approach utilising purely transient and less skilled staff is not effective. A more practical approach is to include responsibility for CC within the core, paid duties of specialised staff trained for the role -"Palliative Care Planning Coordinators" (PCPCs) -with continuous training for other RAC staff. The pilot work referred to above found use of a specially trained coordinator increased referrals to specialist palliative care by 90%. 42 FCC for residents with advanced dementia in RAC is a complex intervention, as defined by the UK Medical Research Council (MRC). 45 Complex interventions require tailored implementation to local contexts and monitoring of processes to assess fidelity of implementation, clarify causal mechanisms, and identify influential contextual factors to inform ongoing development and evaluation. As noted above, a development phase has been completed over a number of projects and a feasibility and piloting phase will end in June 2011. The current project is the evaluation phase aimed at determining efficacy, understanding change process within the RACFs and evaluating cost effectiveness. This will inform dissemination, surveillance and monitoring, and longer-term follow-up in a future implementation phase for which further funding will be requested. The intervention will be implemented using the UK's National Institute for Health and Clinical Excellence guidance on behaviour change, which advocates partnerships with those involved to ensure interventions are appropriate to local need. 46

Aims
This phase III cluster RCT aims to: 1) compare the efficacy of FCC with Usual care in improving end of life (EOL) outcomes for residents with advanced dementia living in RAC; 2) provide insights into facility-and staff-related processes influencing the implementation and sustainability of FCC; and 3) evaluate the cost-effectiveness of FCC versus Usual care.

Hypotheses
Primary hypothesis (Aim 1): Compared with Usual care, FCC for residents with advanced dementia will achieve better family-rated EITHER result in an overall reduction in health care costs with no loss to resident health OR be costeffective in terms of health gained at acceptable additional cost.

PARTICIPATING SITES
Twenty RACFs in Sydney and Brisbane will be recruited (10 per arm). RACFs likely to meet the eligibility criteria will be identified from lists and approached in random order.: 1) ≥100 beds, 2) ≥50% residents with dementia, 3) identified as high care. RACFs meeting these criteria will be approached by letter and phone to canvass interest. Informed consent for facility participation will then be obtained from facility managers.

Study design
The proposed research will use a phase III parallel cluster RCT design, conducted in 6 stages over 3 years. A cluster RCT design was chosen to minimise contamination between arms when taking a systems-based approach to intervention.
Activities for each stage will be as follows: 1. Set-up: Project staff recruitment; application for ethics approval; identification of eligible RACFs and randomised approach; RACF level consents; RACF staff consents; recruitment and training of research staff (including inter-rater reliability). 6. Close-out: Data checking, analysis and reporting; feedback to participants; analysis and reporting.

Conditions:
The Intervention and Usual care to be compared in the proposed study will be as follows.

INTERVENTION:
We will use a train-the-trainer model, whereby a Palliative Care Planning Coordinator (PCPC) is trained and then supported to train other nursing and personal care staff at each RACF.
Palliative Care Planning Coordinators (PCPCs): RACFs randomised to the Intervention arm will be invited to discuss allocation of PCPC responsibility best-suited to local conditions. The PCPC will be the team leader in each RACF responsible for implementing the CC model as per the protocol and providing ongoing education and mentoring to other RACF staff (see below). Our pilot experience suggests the role will best suit a permanent, full-time, senior nurse. (S)he will require communication skills and clinical expertise appropriate to dementia and palliative care. PCPCs will be expected to commit 0.5FTE to the role, paid at their usual rate. Selection will be via interview and, once appointed, relevant duties will be added to the staff member's job description. PCPCs will attend one week full time equivalent of training at a central location. Where a trigger for FCC has been identified, the point of disease progression will be mapped onto relevant domains to indicate whether survival time will likely be measured in months, weeks or days.
The "map" for each resident will highlight issues for discussion at CC, including preemptive planning for predicted deterioration. The PCPC will receive training in organising CCs and receive specific guidance on when and how to initiate advance care planning regarding do-not-resuscitate (DNR) and

do-not-hospitalise (DNH) orders and artificial nutrition/hydration. 48
Training in facilitating general practitioner involvement: Strategies will include using a direct telephone approach with an offer of a follow-up visit to discuss, assistance with obtaining remuneration for time spent, and ensuring that FCCs are as time-efficient as possible. Wherever possible, GPs will be given advance notice of FCCs and asked to book them in as appointments. FCCs will be arranged at times to suit GPs (e.g., lunch times), and GPs will be given the opportunity to join by teleconference only for those CC contents most relevant to them (e.g., symptom management). These approaches have led to impressive GP attendance rates in a previous RCT by the team. 38 Palliative environment: Finally, training will assist RACF staff to provide a "palliative" space (an area which provides the resident and family a quiet, homelike private environment 24 hours per day. 49 PCC training will make use of experiential and adult learning approaches, guidance on how family members can be involved as much as possible in decision-making about care planning, implementation and monitoring, and support reconceptualisation from a "service focus" to personcentred focus. 50,51 Train-the-trainer training will provide PCPCs with the skills and resources to deliver one-day training and mentoring in advanced dementia-specific CC and PCC palliative care to other nurses and care staff. Ongoing support in the Intervention arm will be provided following initial training and for the rest of the study period. The team will liaise monthly with each PCPC to discuss progress and difficulties and plan support of staff training and CCs where monitoring data (see below) indicate this is required.

USUAL CARE:
There will be no additional education, training or support provided to Control RACFs but there will also be no restriction on service-provider education and training where this is current practice. The level, duration and type of such initiatives will be documented.
Randomisation and blinding: Block randomisation will occur after initial consent & baseline data at RACF level using two stratification factors (dementia specific unit or not; part of organisation or stand alone facility) using a computer generated allocation sequence. The statistician allocating sites will be blind to their identity. Research assistants will be blinded to the purpose of the study, & each will visit only RACFs in either the Intervention or Control arm. Project managers will need to be unblinded in order to liaise with PCPCs at Intervention RACFs; they will not collect process or outcome data after baseline.

Data sources/Collection
Baseline data: Variables that have potential to influence outcomes will be tracked to inform further development and implementation. 45 Descriptive data will be collected as follows using adapted proformas from the CADRES 36 and PerCEN 37 studies. Family member: Self-reported relationship to resident, age, gender, education, occupation, dependents, frequency/duration of visits, person responsible status and prior involvement in decision-making.
Process and sustainability data: This will be collected to determine 'dose' and 'duration' of the study intervention and describe facilitative or obstructive mechanisms that can be used to inform future implementation. 45 Process data will be collected at Intervention arm RACFs using an adapted version of the PerCEN study process data schedule. Where relevant, data will inform the need for trouble-shooting. Data will include: PCPC and other staff attendance at training; PCPC and other staff learning outcomes (pre-/post-training and 6-month follow-up qPAD); CC numbers, triggers, duration and attendance; care plan content and adherence; quality of documentation. Data will be collected via reviews of nursing and CC records, interviews with PCPCs and observation of CCs. Assessment of adherence will be undertaken by a blinded, expert panel (including a palliative care physician, GP and RN) who will review documentation for a random 10% of residents at each RACF. The first 15 CCs observed by each research assistant will be audio-taped and independently analysed to validate observation-based coding. Sustainability data will be collected via exit interviews with all PCPCs and a random sample of up to 30 families, RAC staff and community health professionals caring for enrolled residents in each arm (precise numbers will be determined by saturation). 56 Interviews will be semi-structured and will explore perceptions of PCC and palliative care for advanced dementia and the importance of multidisciplinary and family involvement in care planning. In the Intervention arm, interviews will also concern perceptions of the intervention's usefulness and facilitators/barriers to its implementation.
PCPCs will also be asked to give examples of how they have used the training/materials to improve the care of individual residents as well as to describe their approach to a range of hypothetical scenarios Outcome data: Intervention and Control arms will be compared on the following outcomes: Primary (Hypothesis 1): Family-rated EOL outcomes will be measured using the End of Life in  70 and our intervention's staff focus may lead to response bias in nurse ratings. All scales will be rated 4 weeks following resident death.

Secondary (Hypothesis 2):
a. Nurse-rated symptom-related comfort and symptom management will be measured using the CAD-EOLD and SM-EOLD Scales, and rated 4 weeks following residents' death by the nurse who has provided most care in the last 7 days and 90 days of life respectively. Nurse-ratings have been included because they may offer an alternative and more 24-hour perspective 60 and comparison with family ratings will allow exploration of potential bias from both kinds of respondents.
b. Nurse-rated resident QOL will be measured second weekly using the Quality of Life in Late-stage and CASCADE 55 will enable comparison with large Australian and US samples. For the purposes of economic evaluation, QOL will also be assessed by proxy using the EQ-5D-5L, 65 a new version of the world's most widely used multi-attribute utility instrument EQ-5D 65 that has been revised to include a larger number of severity levels among its response options (therefore giving greater sensitivity). The EQ-5D has been widely used in dementia studies. 66 c. A palliative approach to care at the facility level will be measured using the following indices: rates of acute care episodes and length of stay (including ED presentations with admission and actual admission) for residents with advanced dementia; rates of inappropriate acute care episodes, as judged again by an expert review of 10% of admissions at each RACF. Documents reviewed will be nursing records, transfer letters, discharge summaries and medical records written on admission and discharge; rates of non-palliative interventions, defined as ventilation, resuscitation, nasogastric/PEG feeding and non-palliative pharmacological treatments; rates of inappropriate, non-palliative interventions, as judged by expert review of nursing records; rates of adverse events, identified via a review of nursing records and incident charts and coded as: falls with/without injury, skin tears, injuries during care and medication incidents; and number/type of complaints from families regarding the quality of care.
d. RACF person-centred approach to care will be measured by the 'Care and Activities, and e. RAC staff's attitudes to, knowledge of and confidence in providing palliative/EOL care to residents with advanced dementia will be evaluated using the Palliative Care for Advanced Dementia (qPAD). 67 This 35-item scale has demonstrated satisfactory factor structure and internal consistency (Cronbach's alpha 0.58 -0.90). The qPAD will be administered before and after training to PCPCs by research staff and to other RAC staff by the PCPC delivering training.
Responses from RAC staff will be anonymous to reduce response bias.

Secondary (Hypothesis 3):
The economic evaluation will take a cost-utility approach, in which health benefit will be estimated in terms of quality adjusted life years (QALYs) gained. This will capture expected intervention effects on morbidity within the context of any unexpected change in survival. Resident QOL will be estimated using nurse-ratings on the EQ-5D-5L introduced above, for which preference weights have been valued by a representative sample of the general population.
Healthcare costs from a societal perspective will include: training costs (materials, trainer's time and opportunity cost of trainee time); CC costs (session time, travel time and distance travelled by attendees); routine healthcare costs (use of GPs, specialists and pharmaceuticals).

Family members (ideally the person legally responsible)
Visits the resident at least once a fortnight; knew the resident prior to their dementia diagnosis; is willing to be involved in decisions about the resident's care; English proficiency sufficient to complete outcome measures.

RAC staff
Permanent nursing, assistant in nursing and careworkers employed at a participating facility.

Health professionals
Health professionals (e.g. GPs) involved in case conferencing for participating residents.

Sample size
In the absence of established minimal clinically important differences on the EOLD Scales, we will follow a common rule-of-thumb and assume 0.5 standard deviation (SD). 71

Expected duration of study and start times
This 3-year study is expected to begin in September 2012 and be completed by June 2015.

Statistical analyses
Aim 1 -Efficacy: Primary and secondary analyses will be performed on an intention to treat (ITT) basis. Descriptive statistics will be used to compare characteristics at individual-and facility-levels in each group at baseline, accounting for the clustering effect in the former. Mixed or 'multilevel' modelling will be used to determine the effects of FCC and Usual care on primary and individual-level secondary outcomes (i.e. all secondary hypotheses except d). Mixed models allow adjustment for both individual (i.e. resident, family or staff) and cluster-level covariates as well as adjustment for the inherent correlation within clusters. Cluster level analyses will be implemented to determine the effect of the intervention on facility-level outcomes (i.e. hypothesis 2d). Analyses will be weighted by cluster size as required. Results will be interpreted and generalised accordingly.
Aim 2 -Process and sustainability: Bayesian inference networks analysis (BINA) 73,74 will be used to further model the influence of resident, family, staff, RACF and intervention variables on processes and outcomes. This analysis will overcome two limitations of traditional statistical analyses, namely: 1) inability to detect intervention effects due to smaller than expected effect size or heterogeneity in participants or facilities, and 2) relevance of results only to groups with certain 'average' characteristics that cannot be easily translated to each individual. BINA overcomes these problems by linking individuals' characteristics (e.g., severity of dementia) and events (e.g., change in clinical status) to estimate likelihoods of future improvement or deterioration. These models enable initial assumptions regarding efficacy to be modified and probabilities recalculated to accommodate new observations. Network analysis will also be used to model relationships between family and staff EOLD Scale ratings in each arm to provide insight into potential direction and degree of response bias. BINA is widely applied in information technology (e.g., the Google search engine) but use in health research and dementia care is only just emerging and observational rather than evaluative data have been used. 75,76 BINA will be carried out by Dr Victor Vickland, from the Dementia Collaborative Research Centre (DCRC) (UNSW). Dr Vickland is CIA on an NHMRC-funded project using BINA to develop computer-assisted clinical guidelines of anxiety and depression management. Qualitative analysis of interview data will be supervised by CIs Chenoweth, Beattie and Phillips and will use standard coding and classifying techniques and N-Vivo software. 56 Aim 3 -Economic evaluation: The incremental benefit of FCC will be calculated in terms of QALYs gained. The cost of the intervention will be measured along with any cost-savings due to avoided healthcare utilisation. Results will be presented in terms of the incremental cost-effectiveness ratio (ICER). Many of these model parameters will not be powered for statistical significance. Therefore, mean estimates of resource utilisation will be used and confidence intervals will be generated by bootstrapping the data. Uncertainty will be explored using probabilistic sensitivity analysis.

Recruitment and selection of participants
Aged care facilities will be selected for approach from comprehensive lists of all facilities in each geographical area by means of a randomisation schedule intended to minimize selection bias. Facility managers will be approached via letter and telephone by the research team to canvass their initial interest. Managers expressing initial interest will be sent an information and consent form, followed by a follow-up phone call.
Facility managers and staff will be asked to identify residents who: 1) have dementia, in the advanced stages (with or without a confirmed diagnosis); 2) have a family member or friend who visits on a regular basis; and 3) have a person responsible to give consent to participate on their behalf, whether this be the regular visitor or another person. The person responsible and the regular visitor will be given an information and consent form by facility staff either in person or by post and invited to contact the research team by telephone if they would like more information. A cover letter from the facility introducing the study will be included in post-outs, which may also be followed up by telephone. Upon contact, research staff will go through the procedure of informed consent, making it clear that participation is voluntary and will not affect relationships with the facility staff or researchers. Persons legally responsible who wish to will give informed consent on behalf of the resident with advanced dementia while regular visitors will be invited to give informed consent on their own behalf. Note that, in many cases, we expect the person responsible and regular visitor to be one and the same person so that they will give two consents -one on behalf of the resident with advanced dementia and one on their own behalf.
Facility staff with care responsibilities (i.e. nurses and assistants in nursing) will also be eligible to participate. Staff will be told about the study verbally at staff meetings, provided with information and consent forms and invited to contact the research team if they have further questions or would like to participate.
In facilities randomized to receive the intervention, a key participating staff member will be a palliative care planning coordinator (PCPC) who will be employed 0.4FTE to coordinate case conferencing at their facility. Candidates for the PCPC position will, wherever possible, be identified by managers from existing senior nursing staff using standardised criterion. Potential candidates will be given a duty statement by the manager and invited to contact the research team if they would like more information.
Finally, other health professionals who take part in case conferences for participating residents will also be eligible to take part. Their decision whether or not to participate in the evaluation will be quite independent of their decision to contribute to case conferences. The PCPC, who will have liaised with these health professionals to invite them to case conferences, will provide them with information and consent forms and invite them to contact the research team if they are willing to participate.

Informed consent
Informed consent will be collected from persons responsible (on behalf of residents), family members, RAC staff and health professionals. Persons responsible will receive and sign a study-specific patient information sheet and consent form (PIS/CF). A waiver will be sought from the Human Research Ethics Committee (HREC) for informed consent relating to summary data at the facility level used to monitor the quality of care (e.g. number of admissions, accidents, complaints).

Confidentiality and Privacy
Participant names will be replaced by ID numbers at the sites soon after data has been collected. A file linking ID numbers with names will be stored in a password protected folder at the study sites.
Consent forms will be kept in a separate locked filing cabinet also at the study sites. The electronic file linking ID numbers with names will not leave the study sites and will be destroyed at the end of the project period once the need to locate data for removal in the case of withdrawal has passed.
All reports will be at the summary level and will not refer to individuals in any way that would enable their identification.

Data storage and Record retention
All data will be stored on password protected computers at the University of Technology Sydney's Faculty of Midwifery and Health and/or uploaded onto the CareSearch (www.caresearch.com.au) website research data management system which is password protected and encrypted, specifically developed for multicentre clinical trials. Access will vary according to the roles of individuals involved in the project. A/Prof Meera Agar, Dr Tim Luckett, Dr Georgina Luscombe, Dr Victor Vickland, a health economics research fellow, research assistants and a data manager will have access to all data either because they have collected and uploaded it, to monitor data quality or to prepare the data for analysis. Other members of the research team will have access only to summary data to contribute to interpretation.
The files linking ID numbers with personal details will be destroyed at the end of the project period after the need to locate data for removal in the event of participant withdrawal has passed. Other data will be maintained for 7 years after study completion. This meets national requirements for record retention of research materials. Once the waiting period is complete, we will erase all electronic files (including those on the CareSearch website) and shred any paper copies.

OUTCOMES AND SIGNIFICANCE
Dementia is now Australia's third leading cause of death and is increasing exponentially. 77 In RAC, dementia is the greatest single contributor to the cost of care. 78 The total cost of dementia in 2002 was $6.6 billion -over $40,000 per annum per person with dementia; this is expected to grow to 3.3% of GDP by 2050. 79 Hospital stays for people with dementia may be twice as long and ED costs 2.5 times greater than for non-demented elderly. 80 Research suggests that many hospitalisations occur for nonpalliative, unnecessary reasons. 20 The draft Productivity Commission report Caring for Older Australians 81 also suggests an increase in transfers of RAC residents with dementia to hospital for pain management and EOL care due to insufficient staff expertise and limited GP involvement (p.260).
The Commission calls for more palliative care in RAC and the education and resources to support this.
By addressing this recommendation, our intervention will reduce healthcare burden while improving care quality, resident wellbeing and family satisfaction. Although CC in RAC has been widely promoted, uptake is fragmented and the impact on outcomes poorly understood. Our intervention is novel in that it is targeted to the dementia trajectory, tailored to individual RACF, resident and family needs, and is strongly aligned to RAC business processes. These qualities will deliver an important contribution to the Government's aged care reform agenda.

TIMELINES / MILESTONES
The timeline for the 6 project stages is summarised in Table 1.

PUBLICATION POLICY
Any Intellectual Property arising from this study shall be jointly owned by the collaborating research team. There have been no limits placed on publication of results.

1.
Australian Institute of Health and Welfare. Residential Aged Care in Australia 2007Australia -2008 Specific objectives or hypotheses Cluster randomised trial addition: Scientific background and explanation of rationale, including the rationale for using a cluster design Specific objectives and hypotheses and whether they pertain to the individual level, the cluster level, or both

Background
Dementia is a terminal disease. Care for people with advanced dementia requires a palliative approach that is targeted to the illness trajectory and tailored to the needs of each individual and his/her family. Currently, the quality of care in residential aged care (RAC) is compromised by lack of staff expertise and poor communication between staff, family and health professionals. Residents suffer unnecessary hospitalisations and aggressive treatments, while symptoms often go unmanaged. Facilitated case conferencing (FCC) is an approach that brings together RAC staff, health professionals and families to plan person-centred management based on best practice. FCC has improved outcomes in other palliative settings but evidence is lacking for RAC residents with advanced dementia. A cluster RCT design was chosen to minimise contamination between arms when taking a systems-based approach to intervention. Hypothesis 2 (Aim 2): Compared with Usual care, FCC for residents with advanced dementia will achieve: a) better nurse-rated symptom-related comfort over the last 7 days of life and symptom management over the last 90 days of life; b) higher scores on resident quality of life (QOL) at last measurement prior to resident death; c) a palliative approach to medical care of residents with advanced dementia, as evidenced by reduced acute care episodes, non-palliative interventions and adverse incidents at the facility level; d) a person-centred approach by RACF staff to all aspects of care; e) improvements in RAC staff attitudes to, knowledge of and confidence in providing palliative care to residents with advanced dementia and dementia care more generally. f) improvements in family QOL.

Hypothesis 3 (Aim 3):
Compared with Usual care, FCC for residents with advanced dementia will EITHER result in an overall reduction in health care costs with no loss to resident health OR be cost-effective in terms of health gained at acceptable additional cost.

Trial design 3a
Description of trial design (such as parallel, factorial) including allocation ratio 3b Important changes to methods after trial commencement (such as eligibility criteria), with reasons Cluster randomised trial addition: How participants were allocated to interventions (e.g. random allocation, randomised, or randomly assigned), specifying that allocation was based on clusters The proposed research will use a phase III parallel cluster RCT design, conducted in 6 stages over 3 years, with a 1:1 RACF allocation to either FCC or UC. Activities for each stage will be as follows: -availability of a person legally responsible to give consent on their behalf; -informed consent from a family member or other who knows the resident well & will participate as the second member of the dyad. The FAST has been chosen because of its utility in the advanced stages of dementia; a FAST stage 7c & functional dependency (measured here by AKPS) is predictive of <6 months survival.
Eligibility for family/friend "informant" (ideally the person legally responsible): -visits the resident at least once a fortnight; -knew the resident prior to their dementia diagnosis; -is willing to be involved in decisions about the resident's care; -English proficiency sufficient to complete outcome measures (below).

RAC staff:
all permanent nursing/personal care staff.
Other health professionals (e.g. GPs) involved in case conferencing for participating residents will also be eligible to participate.

Settings and locations where the data were collected
Brisbane and Sydney, metropolitan The interventions for each group with sufficient details to allow replication, including how and when they were actually administered Cluster randomised trial addition: Precise details of the interventions intended for each group, whether they pertain to the individual level, the cluster level, or both.

INTERVENTION:
We will use a train-the-trainer model, whereby a Palliative Care Planning Coordinator (PCPC) is trained and then supported to train other nursing and personal care staff at each RACF.
Palliative Care Planning Coordinators (PCPCs): RACFs randomised to the Intervention arm will be invited to discuss allocation of PCPC responsibility best-suited to local conditions. The PCPC will be the team leader in each RACF responsible for implementing the CC model as per the protocol and providing ongoing education and mentoring to other RACF staff (see below). Our pilot experience suggests the role will best suit a permanent, full-time, senior nurse. (S)he will require communication skills and clinical expertise appropriate to dementia and palliative care. PCPCs will be expected to commit 0.5FTE to the role, paid at their usual rate. Selection will be via interview and, once appointed, relevant duties will be added to the staff member's job description. PCPCs will attend one week full time equivalent of training at a central location. To assess learning outcomes, PCPCs will complete pre-and post training administrations of the Palliative Care for Advanced Dementia (qPAD), a 35-item tool assessing health professionals' knowledge of advanced dementia care needs and attitudes to providing related care. Where a trigger for FCC has been identified, the point of disease progression will be mapped onto relevant domains to indicate whether survival time will likely be measured in months, weeks or days. The "map" for each resident will highlight issues for discussion at CC, including pre-emptive planning for predicted deterioration. The PCPC will receive training in organising CCs and receive specific guidance on when and how to initiate advance care planning regarding do-not-resuscitate (DNR) and do-not-hospitalise (DNH) orders and artificial nutrition/hydration.
Training in facilitating general practitioner involvement: Strategies will include using a direct telephone approach with an offer of a follow-up visit to discuss, assistance with obtaining remuneration for time spent, and ensuring that FCCs are as time-efficient as possible. Wherever possible, GPs will be given advance notice of FCCs and asked to book them in as appointments. FCCs will be arranged at times to suit GPs (e.g., lunch times), and GPs will be given the opportunity to join by teleconference only for those CC contents most relevant to them (e.g., symptom management). These approaches have led to impressive GP attendance rates in a previous RCT by the team. Palliative environment: Finally, training will assist RACF staff to provide a "palliative" space (an area which provides the resident and family a quiet, homelike private environment 24 hours per day.
PCC training will make use of experiential and adult learning approaches, guidance on how family members can be involved as much as possible in decision-making about care planning, implementation and monitoring, and support reconceptualisation from a "service focus" to person-centred focus.
Train-the-trainer training will provide PCPCs with the skills and resources to deliver one-day training and mentoring in advanced dementiaspecific CC and PCC palliative care to other nurses and care staff. Ongoing support in the Intervention arm will be provided following initial training and for the rest of the study period. The team will liaise monthly with each PCPC to discuss progress and difficulties and plan support of staff training and CCs where monitoring data (see below) indicate this is required.
USUAL CARE: There will be no additional education, training or support provided to Control RACFs but there will also be no restriction on service-provider education and training where this is current practice. The level, duration and type of such initiatives will be documented.

Outcomes 6a
Completely defined pre-specified primary and secondary outcome measures, including how and when they were assessed 6b Any changes to trial outcomes after the trial commenced, with reasons Cluster randomised trial addition: Report clearly defined primary and secondary outcome measures, whether they pertain to the individual level, the cluster level, or both, and, when applicable, any methods used to enhance the quality of measurements (e.g. multiple observations, training of assessors)

Primary:
Family-rated EOL outcomes will be measured using the End of Life in Dementia (EOLD) Scales (Volicer et al., 2001).
The EOLD Scales will be used to measure symptom-related comfort during the last seven days of life (CAD-EOLD), symptom management in the last 90 days of life (SM-EOLD) and family satisfaction with care during the last 90 days of life (SWC-EOLD). While the CAD-EOLD and SM-EOLD have been validated for rating either by family or nurses, family ratings will be our primary outcome because family perceptions of EOL suffering and its management are important outcomes for palliative care and our intervention's staff focus may lead to response bias in nurse ratings. All scales will be rated 4 weeks following resident death. a. Nurse-rated symptom-related comfort and symptom management will be measured using the CAD-EOLD and SM-EOLD Scales, and rated 4 weeks following residents' death by the nurse who has provided most care in the last 7 days and 90 days of life respectively. b. Nurse-rated resident QOL will be measured second weekly using the Quality of Life in Late-stage Dementia (QUALID) Scale 61 and EQ-5D-5L 66 for economic evaluation purposes. c. A palliative approach to care at the facility level will be measured using the following indices: -rates of acute care episodes and length of stay (including ED presentations with admission and actual admission) for residents with advanced dementia; and -rates of inappropriate acute care episodes, as judged again by an expert review of 10% of admissions at each RACF. Documents reviewed will be nursing records, transfer letters, discharge summaries and medical records written on admission and discharge; -rates of non-palliative interventions, defined as ventilation, resuscitation, nasogastric/PEG feeding and non-palliative pharmacological treatments; -rates of inappropriate, non-palliative interventions, as judged by expert review of nursing records; -rates of adverse events, identified via a review of nursing records and incident charts and coded as: falls with/without injury, skin tears, injuries during care and medication incidents; and number/type of complaints from families (e.g. quality of care). d. RACF person-centred approach to care will be measured by the 'Care & Activities, & Interpersonal Relationships & Interactions' domain of the Person-Centred Environment & Care Assessment Tool (PCECAT). e. RAC staff's attitudes to and knowledge of providing palliative/EOL care to residents with advanced dementia will be evaluated using the 35-item Palliative Care for Advanced Dementia (qPAD. 67 Sample size 7a How sample size was determined 7b When applicable, explanation of any interim analyses and stopping guidelines Cluster randomised trial addition: How total sample size was determined (including method of calculation, number of clusters, cluster size, a coefficient of intracluster correlation (ICC or k), and an indication of its uncertainty) and, when applicable, explanation of any interim analyses and stopping rules In the absence of established minimal clinically important differences on the EOLD Scales, we will follow a common rule-of-thumb and assume 0.5 standard deviation (SD) 71 . Sample size has been based on the EOLD scale with the highest intra-cluster correlation (ICC) (i.e., requiring the greatest number of clusters). To detect a change in CAD-EOLD of half a standard deviation (a change in score of at least 3), with a two-sided 5% significance level, power of 80% and an intracluster correlation coefficient of 0.05 (estimated from unpublished data sourced from Dutch RAC facilities), a sample size of 8 clusters per group (16 in total), with 15 residents per cluster, was necessary. Given an anticipated dropout rate (at the resident level) of 10%, a recruitment sample of 272 residents (approximately 17 per site) would be needed for a final sample of 240 residents at 8 facilities. An interim check conducted on blinded data as of 6 months data collection will be performed to assess whether any of these assumptions requires adjustment and a re-estimation of sample size is necessary. The feasibility of accruing 17 residents with advanced dementia per RACF with a life expectancy less than the study period (<18 months) is based on: 1) Protocol v1.1 25 th Feb 2013 Page 50 of 51 a review of death data at Hammond Care RACFs, and 2) research showing that 40% of people admitted to RAC with advanced dementia die within 1 year and our eligibility criteria predict further decreases in life expectancy. 70,72 Randomisation: Sequence generation 8a Method used to generate the random allocation sequence 8b Type of randomisation; details of any restriction (such as blocking and block size, stratification, matching) Allocation concealment mechanism 9 Mechanism used to implement the random allocation sequence (specifying that allocation was based on clusters rather than individuals), and clarifying whether the sequence was concealed until interventions were assigned Implementation 10 Who generated the random allocation sequence, who enrolled participants, and who assigned participants to interventions/their groups Block randomisation will occur using two stratification factors (dementia specific unit or not; part of organisation or stand alone facility) using a computer generated random allocation sequence developed by the study statistician. This method will facilitate balance in the allocation of RACFs to the FCC and UC groups. The size of the blocks will vary randomly between two, four and six. Facilities will be enrolled prior to assignment. Enrolment and assignment will be undertaken by the project manager who will in any case need to be unblinded in order to liaise with facilities in the FCC arm regarding implementation of the intervention. The random allocation sequence will be concealed until facilities are assigned.

Blinding 11a
If done, who was blinded after assignment to interventions (for example, participants, care providers, those assessing outcomes) and how 11b If relevant, description of the similarity of interventions Cluster randomised trial addition: Whether participants, those administering the interventions, and those assessing the outcomes were blinded to group assignment. If done, how the success of blinding was evaluated.
In order to avoid selection bias, recruitment of patients was by research assistants masked to the cluster allocation. The research assistants will be blinded to the purpose of the study, and each will visit only RACFs in either the Intervention or Control arm. The statistician allocating sites to FCC or UC groups will be blind to the identity of the RACFs. Project managers will need to be unblinded in order to liaise with PCPCs at Intervention RACFs; they will not collect process or outcome data after baseline.

Statistical methods 12a
Statistical methods used to compare groups for primary outcome(s) indicating how clustering was taken into account 12b Methods for additional analyses, such as subgroup analyses and adjusted analyses