Living and dying with advanced dementia: A prospective cohort study of symptoms, service use and care at the end of life

Setting

We recruited people with advanced dementia from May 2012–December 2014, from six Clinical Commissioning Groups (CCG; National Health Service (NHS) organisations overseeing healthcare delivery in England) across Greater London. To reflect the estimated place of residence of people with advanced dementia in the United Kingdom, we aimed to recruit 100 people with advanced dementia, 70 from nursing homes and 30 residing in their own homes.¹ We purposively selected 14 nursing homes ensuring a representative range of

• Care Quality Commission (CQC; public body of the Department of Health which regulates and inspects health and social care services in England) ratings;

• Gold Standards Framework (GSF; national training programme enabling staff to provide generalist care at the end of life) implementation;²²

• Number of beds;

• Ownership (private company or state services);

• Areas of socio-economic and ethnic diversity.

The majority (90%) of UK nursing homes are privately owned, but residents should have equitable access to all core NHS services, such as primary and secondary care (including mental health), and palliative care. However, access to chiropody, opticians and physiotherapy is less consistently provided by statutory services and nursing homes, and residents or their families may purchase this care privately.

To recruit people with advanced dementia residing in their own homes, we used general practices in five CCG areas of greater London; practice staff screened patient lists to identify potential participants. To ensure access to primary care records, we recruited participants (both those residing at home and those in nursing homes) who were registered with practices linked to the Primary Care Research Network which works with general practitioners (GPs) to support research.

Inclusion and exclusion criteria

Nursing home and primary care staff identified potential participants meeting the following criteria: clinical diagnosis of dementia aged over 65 years and at Functional Assessment Staging (FAST) grade 6e and above²³ (one of: doubly incontinent and speaks only a few words, unable to walk, loss of intelligible speech, unable to smile, unable to hold their head up). For recruitment and consent procedures, see Jones et al.¹⁷

Data collection

At study entry, we documented demographics, Charlson Comorbidity Index (CCI)²⁴ (19 diseases weighted on their association with mortality); prescribed medications;²⁵ advance directives, advance statements and lasting power of attorney. Research staff assessed participants using validated scales, through direct observation or in consultation with carers and nursing home staff. Study measures (see Table 1) included physical symptoms: pain (Pain Assessment IN Advanced Dementia (PAINAD) Scale,³⁰ scores of 2 or greater indicate clinically significant pain),³¹ pressure sore risk (Waterlow Scale)²⁸ and pressure sore severity (Stirling Scale);²⁹ neuropsychiatric symptoms: Neuropsychiatric Inventory (NPI)³³ and Cohen Mansfield Agitation Inventory (CMAI; scores of over 39 indicate clinically significant agitation),³² comfort (Symptom Management at the End of Life in Dementia (SM-EOLD)),³⁴ quality of life (Quality of Life in Late-Stage Dementia (QUALID))³⁵ and service use with the Client Services Receipt Inventory (CSRI) and the Resource Utilization in Dementia-Lite (RUD-LITE) Scales.^36,37

Participants were assessed every 4 weeks for a maximum of 9 months. In addition, we documented events from the prior month:

• Painful or burdensome interventions, for example, enteral feeding tubes, venous or arterial blood tests and mechanical ventilation.³⁸

• Symptoms documented as being common or problematic in advanced dementia: dehydration, constipation, difficulty swallowing, insufficient oral intake, weight loss, aspirating/choking on swallowing and difficulties breathing.¹³

• Acute clinical events: seizures, fever, urinary tract infection, pneumonia and septicaemia.³⁹

• After death, we collected information from staff and/or family carers on whether the death had been ‘expected’, whether there was a ‘do not attempt resuscitation’ (DNAR) order at the time of death, place and primary cause of death.

Data were collected prospectively on paper case-report forms. A new form was used at each visit so raters would not be influenced by previous scores.

Analysis

Participant characteristics were described using mean values or medians for continuous variables (with standard deviation (SD) or inter-quartile range (IQR)) and frequencies (percentages) for categorical variables at four time points: (1) study entry, (2) for those who died at the final visit prior to death, (3) for those who were still alive at the end of the study at the final visit and (4) for the whole cohort at the final visit. If more than 20% of questionnaire items were missing, the total was set as missing for the individual as this may have implied non-completion of the whole questionnaire. If less than 20% of items were missing per individual, mean scores were imputed.⁴⁰ Analyses were performed using Stata 13.

Ethics

Our approach was based on the UK Mental Capacity Act 2005, and a personal or a nominated ‘professional’ consultee gave their informed written agreement for the person with advanced dementia to participate¹⁷ (National Research Ethics Committee East of England approval 12/EE/0003).

Cohort characteristics

The majority were women (median age of 85 years, IQR of 81–90). Median length of nursing home residence was 3 years (IQR of 1–5). For the 53 (62%) participants alive, at end of follow-up, a median number of 9 (IQR of 8–9) assessments were completed. Mortality at 9 months was 37%, 32 participants died, with a median of four assessments (IQR of 2.5–6.5). At study entry, 76% of the cohort had a DNAR order, 30% had an advance care plan, 40% a documented preferred place of death and 40% a lasting power of attorney.

Diagnoses and comorbidities

Alzheimer’s disease followed by vascular dementia were most common. The mean Bedford Alzheimer Nursing Scale (BANS) score was 21 (SD = 3.4), indicating participants were severely impaired. Participants had an average Charlson comorbidity score of 6 (SD = 1.5), and 38% were taking 6–9 different classes of drugs (Table 2).

Clinical symptoms

At study entry, 53% were at very high risk of pressure sores (Waterlow Scale). On the Stirling pressure sore severity scale, 22% had early signs of pressure sores and 16 participants (19%) had partial-to-full skin thickness loss. Pressure sore risk and prevalence of pressure sores remained constant throughout follow-up and there were no differences in pressure sore risk between those who died and those who were still alive at the final assessment (Table 3 and Figure 1).

The proportion of participants observed to be in pain at study entry, both at rest (11%) and during movement (61%), changed little during follow-up (Figure 1). Over half of participants had clinically significant agitation on the CMAI at study entry (54%) with a slight reduction at the final assessment for those who died (45%) compared to those still alive (56%). Other common psychiatric symptoms at study entry were depression (36%), anxiety (35%), apathy (53%), motor disturbances (33%) and night-time behaviours (waking during the night or sleeping excessively during the day (44%)). The prevalence of these remained unchanged throughout follow-up for all participants, apart from depression which increased to 42% of those alive at the end of the study and 48% who died and motor disturbances which increased to 48% of those who remained alive and 50% of those who died (Table 3).

The commonest physical symptoms at study entry were difficulty swallowing (42%) and weight loss in the prior month (34%). At the final study visit, a higher proportion of those who died experienced aspiration on swallowing (20% compared to 4% those who remained alive) and breathing difficulties (47% compared to 21% of those who remained alive).

Acute clinical events

Urinary tract infection was the commonest acute medical event at study entry (14% of participants in the prior month). Septicaemia and pneumonia occurred in 17% of participants who died compared to those who were alive at the final study assessment (septicaemia 2% and pneumonia 4%).

Quality of life and comfort

Quality of life (QUALID Scale) remained constant from study entry (24.4, SD = 6.8) until final study assessment (24.3, SD = 5.3) deteriorating slightly towards death (25.8, SD = 7.0) (higher scores indicate lower quality of life) as did scores for comfort on the SM-EOLD (Table 3 and Figure 1).

Service use

General (acute) hospital admission was infrequent in the month prior to study entry (0.18 admissions per participant) and remained low during follow-up (Table 4). Use of outpatient services was more frequent. In the month prior to study entry, there were 0.31 contacts per participant with paramedics or emergency ambulance services and 10.3% saw a paramedic. In the month prior to death, there were 0.75 paramedic contacts per participant and 19% were seen by a paramedic at the nursing home.

Participants had no contacts with community nursing services, although contact with ‘specialist’ nurses increased to 12.5% of those who died (from 7.7% at baseline) likely due to palliative care nurses being called in. Contact with other professionals such as chiropodist, optician and dentist was more frequent. The majority of participants were seen by a GP during the study and 96% of those who died saw a GP in their last month. There was little contact with geriatricians, neurologists, psychiatrists or community psychiatric nurses.

Care received at the end of life

The most frequent interventions in the month prior to death were blood tests (14%) and urinary catheters (14%). Death was ‘expected’ in 90% of cases; 81% died in the nursing home and one person died in their own home. Of those who died, 34% were referred to and 28% were seen by the palliative care team. A DNAR was in place for 93% of those who died, and the commonest cause of death was aspiration pneumonia (Table 5).

Course of physical and psychological symptoms

Mortality at 9 months (37%) was comparable to 6-month mortality rates in people with advanced dementia from the United States (18%), Italy (20%) and The Netherlands (37%).^41–43 The commonest physical symptoms at study entry were difficulty swallowing and weight loss. Pain on movement, agitation and apathy were also frequent and persisted at clinically significant levels throughout the study. Aspiration and breathing difficulties increased in the month prior to death, and the commonest acute events prior to death were septicaemia and pneumonia.

Our participants were similar in disease severity and demographics to cohorts from The Netherlands,⁴⁴ Belgium⁴⁵ (for whom data were collected retrospectively after death) and the United States,³⁹ which also found pain, agitation and dyspnoea to be the commonest symptoms.^39,46 As the end of life approaches, difficulties in swallowing, problems with eating and risk of pneumonia also increase.¹³ We report a comparable symptom profile in the United Kingdom and, in addition, demonstrate how these symptoms were chronic and unchanging. This is important as our participants were unable to express their needs and were, therefore, vulnerable to long-term suffering. However, despite concerns regarding suboptimal care in the United Kingdom,¹⁰ mean symptom management scores (SM-EOLD) were similar to those found in The Netherlands⁴⁴ and Belgium,¹⁵ but lower than found in the United States.⁴⁷

Service use

General Practitioners were the main providers of medical care with most participants having contact during the study. Despite the prevalence of chronic pain and psychiatric symptoms, little specialist healthcare was provided. Of 85 participants, only 5 saw a geriatrician and a psychiatrist with negligible contact with community psychiatric nurses. Paramedics were major providers of assessment and healthcare towards the end of life (although few contacts led to acute hospital admissions) suggesting a reactive response to needs. In the United Kingdom, studies have found that nursing home residents are often poorly served by existing healthcare arrangements⁴⁸ and receive less planned and more emergency care than those living in the community.⁴⁹ Use of social care services also increased in the month prior to death; 25% of participants who died having contact compared to an average of 4.6% of participants during follow-up. There was more contact with chiropodists, dentists and opticians, reflecting how these services are often purchased-in by care homes or by relatives.

Care received at the end of life

Apart from septicaemia and pneumonia, indicators of impending death were not clearly identifiable. Palliative care outreach teams saw 28% of those who died (34% were referred), predominantly in the month prior to death.⁴⁰ This proportion is higher than in the United States (25%)⁵⁰ and The Netherlands (2.5%);⁵¹ however, there are differences in healthcare provision between these countries. In The Netherlands, nursing homes have a dedicated physician who provides the majority of medical and palliative care ‘in-house’. Nursing homes in our study were poorly serviced by community medical services and did not have dedicated physicians, thus palliative care teams were called in at the end of life to assist with symptom management. Although 76% of our participants had DNAR orders, only 5% had an advance decision to refuse treatment, the same proportion as that found in The Netherlands,⁵² 40% had a recorded preferred place of death and 16% died in hospitals. Care planning focussed on resuscitation decisions and place of death, rather than the type of care that the person would like to receive. Numerous initiatives have tried to address a lack of advance care planning in nursing homes,²⁸ but there are systemic challenges to effective implementation.⁴⁴

Strengths and weaknesses

We did not reach our recruitment target despite extending our recruitment period and receiving support from the Primary Care Research Network; many GPs did not hold records of patients with advanced dementia, and it was difficult to locate people living in their own homes. We recruited from a range of nursing homes, ensuring these were representative by purposively sampling by size, CQC rating, ownership and local socio-economic factors to reflect the current UK nursing home market. In the United Kingdom, 25% of nursing homes are rated ‘inadequate’ or ‘requires improvement’; in London, 20% receive a similar rating and 28% of nursing homes in our study were in this category. Thus, by these measures, our study nursing homes are more representative of the United Kingdom in general, rather than of London. We cannot be sure that our participants are representative of people in the United Kingdom with advanced dementia as no comparative data are available. However, the demographic composition of our cohort is similar to that of UK⁴⁸ and international studies of advanced dementia.^13,46 Nursing home documentation was not standardised, particularly for comorbidity, service use and interventions. Previous studies have been mainly retrospective, relying on symptom recall. We used detailed prospective methods^44,45 and reliable validated tools to allow international comparisons.

Implications for clinical practice

Many current policies and tools focus on making an advance care plan, particularly regarding the preferred place of death. These can be beneficial but rely on identifying those who are approaching the end of life and, as our data suggest, this is challenging. Complex symptoms require active specialist intervention, multidisciplinary working and effective care coordination, but GPs feel they do not have the time or knowledge to do this.⁵³ Increased collaboration between geriatric medicine, palliative care and psychiatry may be helpful but should be coordinated and supported by adequate resourcing and service commissioning.⁵⁴ Although the evidence base for interventions in dementia end-of-life care remains limited,⁵⁵ considering data used in our development of a complex intervention to improve care in advanced dementia,²⁰ we suggest a coordinated needs-based longitudinal approach should be developed and tested.¹² A recent survey found ‘hospice enabled care’ is an emerging model for palliative care for people with dementia in the United Kingdom.⁵⁶ This involves palliative care specialists, predominantly nurses, providing care coordination and symptom management expertise to nursing home residents,¹² and this may warrant further formal evaluation.

Policy and end-of-life care interventions promote the importance of end-of-life care in dementia⁸ and have focused on the place of death; but to improve comfort and quality of life, better management of chronic symptoms is required.⁵³ This would enable people to live and die well with dementia.