Residential aged care residents and components of end of life care in an Australian hospital

In Australia, a residential aged care facility (RACF) is a special-purpose facility which provides accommodation and support, including assistance with day-to-day living, intensive forms of care, and assistance towards independent living, to frail and aged residents (predominantly 65 years or over) [1]. Internationally, RACFs are variously known as nursing homes, long-term care facilities or other names depending on the country. The term RACF is used interchangeably in this article. With population ageing in Australia, the numbers of residential aged care (RAC) residents is rising. In 2016, about 3.7 million Australians were aged 65 years or over [2]. In 2011, about 4% of Australians aged 65 years or over resided in RACFs [3].

Studies of hospitalisation of RAC residents internationally have shown that rates vary from 9 to 60% annually [4]. Predictors of hospitalisation have been studied [5, 6] and reviewed [7]. Factors increasing RAC residents’ likelihood of hospitalisation were socio-demographics (older, male, shorter length of stay), health characteristics (higher burden of chronic disease, polypharmacy, functional disability and various acute problems, particularly infections), certain care practices (physical restraints, pressure sores, feeding tubes, urinary catheters), RACF provider attitudes (overburdening of staff) and financial reimbursement issues. Factors decreasing hospitalisation were presence of dementia and neurological disorders [6, 7], advance care directives (ACDs) and use of hospice, a specific model in the United States of America (USA). Availability of certain medical interventions (intravenous therapy, respiratory therapy) and support services (on-site radiology, laboratory testing) had conflicting associations with hospitalisation. Greater staff resource, in terms of nurses and physicians, showed conflicting data in the review [7] but was associated with decreased hospitalisation in other studies [5, 6].

For RAC residents, hospital transfer decisions are complex. Studies have attempted to measure whether hospitalisations were inappropriate or avoidable [8–11]. Estimated rates of avoidable hospitalisation vary from 19 to 67% [4]. Reviews of measures of appropriateness of hospital transfer have described studies using various tools and their underlying frameworks [12, 13]. These studies had methodological shortcomings, including being based on expert opinion and retrospective in nature (not tested at point-of-care), and the tools lacked comprehensiveness and generalisability.

To the best of our knowledge, there is no literature that directly focuses on the benefits of hospitalisation for RAC residents. However, there is recent literature on the benefits of hospitalisation for those with palliative care need [14, 15]. Conversely, the documented harms of hospitalisation include delirium, falls, worsening function, pressure sores, hospital-acquired infections, medication misadventure and care inconsistent with patients’ wishes, not to mention the cost to the health system [4, 12, 16, 17]. Observational studies show better outcomes treating infections in the RACF compared to hospital [18, 19]. For residents with advanced dementia, compared to a strategy of staying in the RACF, hospitalisation was not cost-effective, especially in the setting of pneumonia [20], and treatment of pneumonia with antibiotics was associated with worse comfort [21].

A systematic review of interventions to reduce hospitalisation of RAC residents [4] showed positive results for advance care planning [22, 23], pneumonia assessment/treatment pathway, mobile geriatric outreach, Hospital in the Nursing Home (HITNH) in combination with advance care planning [23], influenza immunisation and hospice referral (specific to USA). In the Australian advance care planning study, the intervention group showed a statistically significant reduction in mortality [23]. An Irish study showed that advance care planning reduced hospitalisation and overall cost [24]. Three Australian studies have used emergency department (ED) nurses or HITNH to support RACFs [25–27]. Hospital in the Nursing Home has evolved from Hospital in the Home, which are nurse-led programs designed to avoid hospitalisation. These studies have shown reduced ED visits and/or hospital admissions. An advance care planning study in USA, using a video decision-aid in patients with advanced dementia, showed improved quality of communication and reduced hospitalisation [28]. The Aged Residential Care Healthcare Utilisation Study (ARCHUS), a nurse-led multidisciplinary geriatric intervention study in New Zealand, showed no effect on hospitalisations [29]. The Regular Early Assessment Post-Discharge (REAP) study, a predominantly physician-led multidisciplinary geriatric intervention study, showed reduced hospital readmissions, ED visits and overall cost [30].

The literature on care provision of RAC residents, in the last months of life and at the time of death, does provide data on hospital use (frequency and duration of hospitalisations, and rate of death in hospital) and associated clinical problems (underlying and necessitating hospital admissions) [31–37]. In a South Australian (SA) study of a general sample of RAC residents admitted to a metropolitan hospital over 6 months, 4% died during the admission [31]. In a German study in the last 12 months of life, about 58% of RAC residents were admitted to hospital at least once [32], compared to about 40% in a Canadian study in the last 6 months of life [33], and about 34% in an English study in the last 1 month of life [34]. The frequency of death in hospital of RAC residents is reported to be about 29% in Germany [32], 27% in England [34], 19% in Canada [33] and 9% in New Zealand [35]. In Australia during 2010–11, about 18% of RAC residents died outside of the RACF, i.e. the majority in hospital [36]. Measurement of care provision by specific health care providers has been undertaken in the English study [34] and a Norwegian study, which looked at RAC residents who died within 48 h of hospitalisation [37]. This Norwegian descriptive study was the first to measure care provision by providing a detailed account of specific care interventions (diagnostic tests, life prolonging and palliative interventions). Our study adds to this by providing a more detailed description of the specific palliative end-of-life (EOL) interventions.

The main aim of this study was to describe a sample of RAC residents who died in hospital to quantify easily measurable components of end-of-life-care (EOLC). By describing the extent of palliative care required, it may provide some information that assists with RAC resident transfer decisions between the RACF and hospital. For this study, EOLC refers to care in the last few days to weeks of life. To test the hypothesis that the advance care planning and components of EOLC may be different between residents with and without advanced dementia, comparisons were conducted.

The total number of hospital attendances (excluding outpatient visits) over 12 months was 42,142. The number of RAC resident attendances was 1545 (3.7%). The number of RAC residents who died in hospital was 109 (7.1% of RAC resident attendances). The number of RAC residents who died within 1 week of discharge to the RACF was 66 (4.3% of RAC resident attendances).

For overall resuscitation orders (for/against), 70 (64%) had documented preferences. Comparing residents with/without advanced dementia, those with advanced dementia had more overall resuscitation orders (p = 0.019) and more not-for-resuscitation (NFR) orders (p = 0.008). For overall hospitalisation orders (for/against), 27 (25%) had documented preferences. Residents with advanced dementia had no difference in overall hospitalisation orders but had more do-not-hospitalise (DNH) orders (p = 0.019) compared with residents without advanced dementia.

Whilst at the RACF, 100 residents had potential for GP input in the decision about hospital transfer. In the hospital record, there was documentation of GP input in 30 (30%) transfers. Of these, 22 (73%) had input from the usual GP and seven (23%) from a locum GP. General practitioner assessments were face-to-face in 19 (63%), telephone in six (20%) and not documented in five (17%).

All residents were transferred from RACF to hospital by ambulance. In metropolitan Adelaide, extended care paramedics (ECPs; ambulance officers with expanded scope of practice) are available to provide assistance with hospital avoidance when appropriate. Only one resident had ECP input.

In Australia, the Australasian Triage Scale rates priority and waiting time from one to five [42]. For the 99 residents who attended ED, ten (10%) were category one, 28 (28%) were category two, 60 (61%) were category three and one (1%) was category four. For category three, patients should be seen within 30 min. In this study, the mean time to physician assessment was 10.3 min (range 0 to 44 min). The mean length of stay (LOS) in ED was 7.7 h (range 1.1 to 47.7 h).

The mean number of documented GOC discussions with family/caregivers was 1.7. There was no difference between residents with/without advanced dementia.

For EOL medication use in the last 3 days of life, charting of opioids was the most common, followed by benzodiazepines, anti-secretory agents, anti-emetics and anti-psychotics. The mean daily number of doses of medications was highest for opioids, followed by benzodiazepines, anti-secretory agents, anti-psychotics and anti-emetics. The mean daily number of doses of all EOL medications was 4.3. A continuous subcutaneous infusion (CSCI) was commenced in 38 (35%) cases. There was no difference in the overall administration of EOL medications between residents with/without advanced dementia.

In terms of advance care planning, participants in this sample appointed legal proxies for financial decisions in 50% and medical decisions in 45%. Documenting wishes for EOLC (Anticipatory Direction) was present in 11%. In the advance care planning study of hospitalised elderly (80 years or over) by Detering et al. [43], of the 154 intervention participants, 70 (45%) completed a written “ACP” (equivalent to ACD). This suggests that documenting wishes for EOLC can be improved. Advance care plans were present in 60% and resuscitation orders in 64% of the sample. Hospitalisation orders (an order detailing a preference about hospitalisation) were present in 25%. It is possible to improve rates of DNH orders as shown by Lamberg et al. [44]. Their retrospective cohort study of patients with advanced dementia in USA found at time of death, rate of DNH orders was 83.8%. This suggests that discussions and recording of hospitalisation orders can be improved. Although this study shows higher rates of ACDs and ACPs compared with an earlier study [38], improving the quality and frequency of advance care planning remains important because it is known to improve the rates of EOL wishes being respected [43], (i.e. improved decision-making), and reduce hospitalisation [22–24, 28].

In terms of participant pre-hospital resource use, GP input occurred in 30% of cases and ECP input in 1%. This data, showing that RAC nurses are responsible for most decisions about hospital transfer due to suboptimal access to physician input, is consistent with the literature [13, 45]. Interventions to improve RAC nurses’ knowledge and skills is associated with reduced hospitalisations [46]. There is interest in developing alternative models to support nurses in decision-making. Using ED nurses and HITNH have been shown to reduce hospitalisations [25–27] but they were only available during working hours in these studies. Extended care paramedics were introduced in metropolitan Adelaide in 2008 and may be of use, especially after-hours. Observational research is being conducted to measure ECP use and hospital avoidance [47] but there are limited intervention trials on the effectiveness of ECPs [48]. In this study, there were residents who could have received ECP care and whether RAC staff are using this option is not known.

In terms of participant in-hospital resource use, for residents admitted under non-palliative care teams, referral for SPC consultation occurred in 44% and phone advice was provided in 5%. In comparison, Le et al. evaluated EOLC and measured SPC consultations for all patients in a large hospital in Melbourne, Australia [49]. They found that, of the patients where death was anticipated, 42% were referred for SPC consultation, which is comparable with this study.

In terms of EOL medications, our study showed in the last 3 days of life, charting of opioids was 87%, benzodiazepines 73% and antipsychotics 43%, which compares favourably with a retrospective cohort study of three RACFs in Norway, that showed on the last day of life, Morphine was charted in 71.4%, Midazolam 55% and Haloperidol 46.9% [50]. In terms of administration of EOL medications, there have been two recent retrospective cohort studies from the Netherlands, one conducted in a PCU [51] and another conducted in hospital, hospice and home settings [52]. These studies of the last days of life, have measured the proportion of use of specific medications, types of medications (palliative vs preventative) and medication classes, the total amount given in milligrams each day and route of administration. Our study adds to this by measuring the actual number of doses of medications each day given in each class, in order to indicate the extent of palliative medication administration in the EOL.

To summarise the components of EOLC for the average resident in this sample, hospital LOS (admission to death) was about 5 days. Less than two GOC discussions with family/caregivers were documented. A little over four doses of medications were required per day in the last 3 days of life. A CSCI was needed in about one third. About half of the residents who were under non-palliative care teams needed SPC input. It is hoped that having specific information about the extent of palliative care required will assist with decision-making, including appropriateness of transfers between RACF and hospital, and feasibility of care in the RAC setting.

Although this study is thought to be the most detailed description of EOL palliative interventions of RAC residents in hospital, it has significant limitations due to study design (retrospective and case-note review). It focuses on care that has been delivered instead of measures such as care need, quality of care or quality of dying. The components of care measured are medico-centric and fail to capture the breadth and complexity of EOLC [53]. Data collection was from hospital records. Measurement of advance care planning was dependent on delivery of the RACF record to hospital. There may be underestimation of presence of advance care planning documents and GP input. The number of GOC discussions with family/caregivers may be underestimated due to inadequate documentation. The findings may be specific to RACFs in an Australian setting and not be generalisable to other locations.

Decision-making in the care of RAC residents is often complex and challenging. Resourcing RACFs adequately is important, especially for making decisions whether to transfer to hospital. This study reinforces that RAC nurses are mainly making these decisions as GP and ECP input were suboptimal. Ways to optimise decision-making may include quality improvement of RAC nurses and policy measures to maximise input from the usual GP. Alternative models to support RAC staff include greater collaboration with emergency staff and HITNH. Advance care planning plays an important role in decision-making and this study suggests there may be room for improvement, especially documentation of EOL wishes and hospitalisation orders.

This study has described the components of EOLC for hospitalised RAC residents. By describing the extent of palliative care required, it is hoped that providers and policy makers in primary care, RAC, ambulance and hospital services have more information to assist with making decisions about what is the most appropriate care for this population.