The cost of providing end of life care for nursing care home residents: A retrospective cohort study

Abstract

Background

The proportion of people dying in long-term care institutions is predicted to increase in future years. Establishing the costs associated with the provision of such care is important as it represents a potentially increasing burden.

Aim

This study describes the end of life healthcare costs for nursing home residents. The study also explores the effects of resident characteristics on costs, through regression modelling.

Setting

This study took place in south-east England.

Participants

Thirty-eight nursing care homes took part in the study, comprising 2444 individual residents.

Methods

Using a retrospective cohort design, end of life service use was recorded from residents’ nursing care home notes. In this study, end of life was defined as the last six months of life, or from time of residency if this was less than six months. Costs were calculated assuming a healthcare payer perspective.

Results

The total mean healthcare cost per resident was £3906. Hospital stays accounted for two-thirds (67%) of these costs. Fifty-six percent of these hospital stays occurred in the final month of life. Death in hospital vs. in the nursing care home was associated with an average increase in costs of £4223.

Conclusions

Death in hospital is costly, and is seldom identified as a preferred place of death. Therefore, interventions are needed which help nursing care home staff to identify when an individual is dying, and have the skill and confidence to make difficult decisions regarding care provision at the end of life.

Keywords

costs end of life care long-term care nursing homes palliative care

Introduction

Internationally, the population is increasing in size and number.¹ This is giving rise to concerns not only about meeting future demands for the delivery of care, but also concerns about the cost of providing such care. End of life care costs are already substantial in the UK. For example, the annual cost to health and social care services is estimated at £1.8 billion for cancer patients alone.² More precise estimates of end of life care costs are important in light of the predicted rise in the proportion of older people within the UK population. Data regarding end of life care costs are also relevant to other nations whose aging populations are similarly increasing.³

In the UK, Laing reported that of people aged 65–74 years, 0.7% were living in a long-stay hospital setting or a care home.⁴ This rose to 15.8% in the population, of people aged 85 years and over. The Department of Health estimated that 19% of all deaths in the UK occurred in care homes.⁵ As the UK population ages, there will potentially be an increase in the proportion of people who die in long-term care institutions such as care homes. It is perhaps not surprising then that Gomes and Higginson anticipate a 20% increase in the number of people dying in institutional care by 2030 in England and Wales.⁶ Costs associated with end of life care in care homes clearly represent a significant and increasing economic burden.

Recently, studies have begun to calculate the cost of end of life care provision in care homes.^7,8 The present study takes this further using a larger sample, with more detailed information about the services used. This study is based on resource use data collected during a cluster randomised controlled trial. Briefly, the trial investigated the effect of using high facilitation when implementing the Gold Standards Framework in Care Homes programme.⁹

Aims

The aims of this article are to:

Describe the service use and associated costs of healthcare at the end of life for residents of nursing care homes

Describe the inpatient costs of residents:

in the four months prior to the penultimate month of life

the penultimate month of life

and the final month of life

Identify the impact of resident characteristics on costs, through regression modelling.

Methods

Participants

Thirty-eight nursing care homes in South East England took part in the study. The study took place over three years and aimed to use data from all residents who died over that period. Each nursing care home contained between 5 and 143 residents who died during the study period. They were aged between 33 and 107 years (the mean number of residents per nursing care home was 64; the mean age of residents was 84.9 years).

Inclusion criteria

The nursing care home:

completed an application to take part in the Gold Standard for Care Homes (GSFCH) programme at St. Christopher’s Hospice

registered to provide care for frail, elderly residents and

manager and owner were willing to take part in the trial.

Exclusion criteria

The nursing care home was:

caring for mainly young, chronically ill individuals and

outside the private sector.

There were no exclusion criteria for individual participants in the study.

Identification of resources

Costs were calculated from the health and social care resources utilised by the deceased residents. While we did not take a societal perspective, the only major exclusions were the use of medication and the cost of unpaid informal care from family members/friends.

Resource use included residents’:

use of hospice, accident and emergency department, and hospital (inpatient and outpatient) – whether there was a stay, and if so how many days were spent there;

contact with primary care professionals, visiting specialists, and contacts with auxiliary health professionals such as social workers, podiatrists and opticians – whether there had been a contact, and how many contacts there had been and

use of equipment, which represented a one-off costs, such as the use of pressure cushions, syringe drivers, standing hoists, etc.

Relevant costs were identified from previous literature assessing the costs of care at the end of life.¹⁰

Measurement of resources

Client Service Receipt Inventory

The Client Service Receipt Inventory (CSRI) is a questionnaire used for retrospectively identifying resource use of an individual and has been used extensively in previous health economic studies.¹¹ For this retrospective cohort study, a modified version of the CSRI was used, in order to capture costs which are relevant to the nursing care home setting.

Data were collected from the deceased residents’ notes and measured end of life resource use. We defined end of life as the last six months of life in the care home, or from the time of residency if this was less than six months. All deaths were included. Information about each resident’s death was obtained directly from the care homes residents’ healthcare records. These details were then checked against all other available sources of information. These included the Care Quality Commission (the UK’s health regulator) death notification record, the care home admission and discharge book, all GP and hospital discharge letters, and any supplementary records kept to record external healthcare professional visits such as the care home diary. Two researchers visited each care home to ascertain type of death, and any uncertainties were resolved through joint discussion.

We calculated costs of hospital bed use separately for the following time periods:

The four months prior to the penultimate month of life

The penultimate month of life and

The final month of life.

These were recorded separately so as to identify if the associated costs with providing end of life care changed over time. The focus of this study was on services used in the homes rather than on the costs of residency. Consequently, non-healthcare costs were also omitted from our analysis.

Valuation of resources

The costs of hospital stays were taken from the NHS Reference costs 2011–2012 and inflated to 2013 levels.¹²

Unit costs of other services were obtained from a national source (Unit Costs of Health and Social Care; UCHSC).¹³ The cost of one hour of the professionals’ face-to-face contact time was multiplied by the average length of time of an appointment. Previous versions of the UCHSC manual were consulted if data were unavailable in the 2012 version. These were then scaled up to 2013 costs. Where costs of home visits were included, these were used, as nearly all professional contacts with nursing care home residents took place in the nursing care home. Market values were used where unit costs were not available. This was only the case for alternative therapists, dentists and opticians. For dentistry, the cost of NHS band two dental treatment was used. For opticians, the cost of an eye test and NHS grade B glasses was used.

With regards to equipment costs, the UCHSC manual was used where possible.¹³ In most cases, however, it was necessary to use market values. In all cases, six-month equipment costs were calculated using Equivalent Annual Cost factors obtained from HM Treasury’s Green Book.¹⁴ A life span of five years was assumed in the cases of equipment such as wheelchairs, whilst for modifications such as hoists, a life span of 10 years was assumed.

Time frame

Healthcare costs were calculated for a maximum of six months prior to an individual’s death, and so discounting – the process of downwardly adjusting future costs and benefits in relation to present ones – was not used.

Other measures

A number of other variables were also obtained from the residents’ notes. These were:

– demographic variables (age, gender)

– all documented diagnoses

– date of admission to nursing care home and date of death

– place of death

– evidence of:

○ a resuscitation decision in place

○ use of an end of life care plan

○ anticipatory prescribing

○ advance care planning.

Statistical analysis

Service use was analysed using descriptive statistics. All healthcare costs were summed in order to create an overall total. The total cost variable formed the dependent variable in a regression model which included the following independent variables: age, gender, all documented diagnoses, days spent in the nursing care home, place of death (nursing care home, other), type of death (sudden, acute, terminal illness dwindling^a), evidence of dementia, evidence of cognitive impairment, enactment of anticipatory prescribing, use of an end of life care plan, advance care planning and resuscitation decision. Confidence intervals were generated using bootstrapping (with 1000 repetitions) in order to account for skewness in the data.

Data were analysed using SPSS v21.0, and Stata v11 IC.

Results

Sample characteristics and end of life care

Unsurprisingly, the majority of the residents were females (61%) and elderly (mean age 85 years). Seventy-nine percent of the residents had dementia or cognitive impairment. Details of their type of death, place of death and the use of end of life care tools are provided in Table 1. Nearly half the sample died following a dwindling trajectory.

Table 1.

Place of death, type of death and use of end of life care tools.

Recorded details of end of life care	n (Valid %)
Place of death (%)
Nursing care home	1768 (72.8)
Other	660 (27.2)
Type of death (%)
Dwindling	1192 (50.3)
Terminal condition	621 (26.2)
Acute	454 (19.2)
Sudden	102 (4.3)
Use of end of life care tools (%)
Evidence of anticipatory prescribing (%)	727 (31.0)
Use of end of life care plan (%)	341 (14.5)
Advance care planning in place (%)	1496 (63.6)
Resuscitation decision in place (%)	1365 (58.2)

Service use and cost data

Table 2 shows the proportion of residents who used each resource. Only resources contributing >0.5% to the total healthcare cost are included for brevity, but all services are included in the total costs.

Table 2.

Service use and associated mean costs of items contributing >0.5% to the total mean healthcare cost/resident in the last six months of life.

Resource	Percentage of using item	Mean contacts/stay	SD	Mean contacts/stay (for those using service)	SD	Mean cost	SD	Total cost (%)
Hospital use
Medical ward	13.8	1.60	5.49	11.59	10.14	683	2348	17.5
Unknown ward	14.6	1.36	5.37	9.32	11.13	582	2298	14.9
Geriatric ward	8.0	1.12	5.09	13.94	12.02	479	2178	12.3
Other ward	7.6	0.77	4.48	10.12	13.08	328	1916	8.4
Emergency ward	39.3	0.52	0.77	1.33	0.67	224	331	5.7
Surgical ward	3.5	0.41	3.60	11.72	15.60	174	1541	4.5
Medical assessment unit	7.6	0.30	1.60	3.94	4.41	129	686	3.3
Outpatient appointments	23.4	0.68	2.99	2.88	5.65	94	416	2.4
Hospice	0.5	0.07	1.21	14.55	10.29	31	538	0.8
Visiting professionals
GP	95.7	7.11	5.58	7.43	5.49	280	220	7.2
Psychiatrist	3.8	0.12	1.22	3.20	5.41	47	468	1.2
Physiotherapist	12.3	0.70	3.53	5.70	8.58	40	202	1.0
Palliative care nurse	20.4	0.71	2.08	3.46	3.42	29	85	0.7
Social worker	9.2	0.10	0.35	1.13	0.40	23	76	0.6
Equipment
Hoist	57.9	–	–	–	–	93	79	2.4
Specialist bed	70.3	–	–	–	–	72	47	1.8
Manual wheelchair	65.7	–	–	–	–	57	41	1.5
Specialist mattress	67.1	–	–	–	–	22	16	0.6
Total healthcare cost						3906	5261

Note: All costs shown in GBP (£).

Of the residents, 96% saw a GP at least once, making this the most commonly used resource. The mean number of GP visits was 7.11. GPs were therefore also the professionals who were seen most intensively. Specialist hospital beds were used by 70% of the sample, which is higher than for any other equipment. The greatest hospital use was attendance at emergency wards which were used by 39% of the residents. The mean stay on emergency wards was notably shorter than on any other listed ward.

Table 2 also shows the average costs associated with use of each resource, and the percentage contribution of each item to the total mean healthcare cost per resident. Nursing care home costs varied from £87 to £204 per night (mean = £118, SD = £19). The median total cost per resident for end of life care was £1767 (interquartile range = £3775) which differs substantially from the mean of £3906 indicating skewness in the cost distribution. The single greatest contributor to total cost of providing end of life care was time spent on medical wards, which accounted for 17% of the total healthcare cost. The four largest contributors to the total healthcare cost were all inpatient ward stays, together accounting for 53% of total healthcare cost. GPs accounted for the largest proportion of total healthcare cost in relation to other professionals (7%). Hoists accounted for the greatest contribution of equipment to overall healthcare cost, although only amounting to 2%.

Table 3 shows the mean in-patient ward use costs accrued during the three time periods of the study. The greatest proportion (56%) of total ward use costs were accrued in the final month of life. In total, ward use accounted for two-thirds (67%) of all resource use costs. Over a third (38%) of total resource use costs were attributable to ward use in the final month of life.

Table 3.

Mean costs of in-patient ward use in the four months prior to the penultimate month of life, the penultimate month of life, and the final month of life.

	Mean cost	SD	% total ward use costs	% total resource use costs	Cumulative %
Six to twelve months prior to end of life	689	2626	26.2	17.7	17.7
Penultimate month of life	458	1792	17.4	11.7	29.4
Final month of life	1481	2895	56.4	37.9	67.3

The results of the regression analysis are shown in Table 4. When death occurred in the nursing care home, costs were £4223 lower than death in the hospital. This makes place of death the variable which is associated with the largest potential cost differences. Importantly, the confidence interval around this estimate is also relatively narrow, and the lower bound is a great distance from zero. This indicates that death in the nursing care home is very likely to result in lower costs. Death following a dwindling trajectory resulted in a higher cost than any other trajectory. On average, death following a dwindling trajectory incurred £1533 higher costs than a sudden death, £1047 higher costs than death due to acute illness, and £657 higher costs than death due to terminal illness. Whilst all these variables were significant predictors of cost, the confidence intervals indicate that the cost differences are likely to be more variable and less extreme than those due to dying in the nursing care home. As age at death increased by one year, the mean total cost of care decreased by £65. As the number of systems affected by illness increased by one, there was an associated cost increase of £24. The predictive power of the model was significant (F (30, 2185) = 12.33, p < .001) and explained 13% of variance in total healthcare cost (adjusted R²= 0.133).

Table 4.

Regression results showing predictors of total healthcare cost.

Predictor	Beta	SE Beta	95% CI
Place of death
Hospital	Reference
Nursing home	−4223	344	−4862 to −3500
Type of death
Dwindling	Reference
Sudden (unexpected)	−1533	378	−2322 to −859
Acute	−1047	325	−1676 to −439
Terminal illness	−657	293	−1243 to −113
Characteristics of the resident
Age	−65	14	−90 to −35
Gender: Male	34	218	−364 to 496
Diagnoses
Number of systems affected	24	151	20–423
Dementia: Absent	Reference
Dementia: Mild cognitive impairment	−166	321	−906 to 346
Dementia: Positive	67	304	−589 to 586
Heart Disease	1	2	−5to 3
Stroke	0	2	−3 to 3
Diabetes	544	290	−142 to 1029
Respiratory disease	1	71	−14 to 7
Upper GI problem	−192	448	−1173 to 593
Lower GI problem	−75	332	−838 to 447
Musculoskeletal disorder	−2	56	−7 to 4
Skin disorder	−272	499	−1280 to 698
Urinary and Gynaecological	0	288	−3 to 1059
Neurological disorder	−0	270	−7 to 998
Organic mental disorder	−1	1	−4 to 2
Functional mental disorder	1	370	−1 to 1108
Other	3	63	−4 to 149
Elderly/frail	−3	311	−1156 to 2
Cancer	−3	2	−7 to 2
Cognitive Impairment	3	3	−2 to 8
Use of end of life care tools
Anticipatory prescribing	345	238	−101 to 836
End of life care pathway	160	313	−440 to 779
Anticipatory prescribing	433	239	−83 to 851
Resuscitation decision in place	292	324	−229 to 689
Other
Days in nursing home	0	0	0 to 1
Intercept	12021	1310	8968 to 13900

Note: Significant predictors are highlighted in bold. All costs shown in GBP (£).

Discussion

This study presents the resource use and healthcare costs associated with end of life care for 2444 people who died in 38 nursing care homes in the UK. So far as we know, this is the largest study of its kind reported on to date. Studies of this kind are important as they allow policy-makers to identify costly resources. The causes of such resource use can then be identified as potential areas for intervention and enable managers, commissioners and policy-makers to forecast budgets. Finally, this study has a more general appeal in that it highlights the large healthcare costs associated with end of life care.

Unfortunately, the level of detail reported in the two most similar studies is not sufficient to enable direct comparisons in relation to the use of all resources.^7,8 In this study, combined inpatient admission accounted for two-thirds (67%) of total healthcare cost. This is very similar to the estimate produced in a Belgian study,⁸ which reported that 66% of total healthcare costs were attributable to hospital costs (non-intervention group). Therefore, we might tentatively extrapolate this proportion to other countries comparable to the UK and Belgium.

We estimated the total mean healthcare costs per resident in the last month of life at £3906. This is also comparable to the estimate produced by Simeons and colleagues,⁸ who report mean total healthcare cost in the final month of life as €3822 (non-intervention group). At the time of writing’s exchange rate of €1.23/£1, this would convert to £3110 per resident. This slight disparity could be due to differences in tariffs between the UK and Belgium. However, it might also be attributable to the very large variance usually associated with cost data. The confidence intervals of the two estimates would certainly overlap.

Over a third (38%) of total healthcare cost was attributable to ward use in the final month of life. Mean ward use costs in the final month of life were estimated at £1481. This accounted for over half (56%) of all ward use costs over six months.

Based on the above, we can conclude that (1) healthcare costs during the last six months of life are high; (2) the majority of these healthcare costs are attributable to ward use; and (3) the majority of ward use occurs in the final month of life. Hospital stays in the final month of life are therefore a clear target for cost-reducing interventions.

Our regression model showed that dying in the nursing home rather than hospital predicted a cost difference of £3500–4862. The model also showed that death following a dwindling trajectory resulted in higher costs than dying unexpectedly, or from an acute or terminal illness. The cost differences for the latter three types of death were in the order of £100–2322.

It is logical to suggest that the common factor linking place of death and type of death is hospital admissions. A death in hospital results in higher healthcare costs precisely because the individual is in hospital. The dwindling trajectory is recognised as one most commonly experienced by residents in care homes.^9,15–17 In this study, residents with a dwindling trajectory experienced a higher number of hospital admissions (and healthcare costs). A dwindling trajectory can make it more challenging to specifically predict when an individual is dying. However, failure to recognise this means that a resident might die in hospital. Given the large healthcare costs associated with deaths of this kind, and that individuals seldom identify hospital as their preferred place of death,^18,19 interventions are needed to reduce the proportion of residents with a dwindling trajectory dying in hospital.

Whilst the majority of the residents in this study had a dwindling trajectory, 19% died from an acute event. As 44% of these residents were documented to have heart disease and 33% a cerebral vascular accident, offering and undertaking future care planning discussions either with, or for, these residents may enable some of those admitted to hospital to more appropriately die in the care home. Cost is a considerable consideration in relation to place of care and better planning ahead for those residents whose medical condition may suddenly change should enable better use of all resources.

There are now a number of end of life care programmes targeting the structure for end of life care in care homes.^9,20,21 Implementing these programmes enables staff to develop specific skills to recognise and manage the care of all their residents. Examples include learning to plan ahead to meet their residents needs through advance care planning and inputting information onto electronic records that can be shared between health care providers.²² However, speaking with and getting permission from residents is not easy when 80% of care home residents have dementia or a severe memory problem.²³

Qualitative work supports the view that care home staff are sometimes unprepared in relation to their residents dying. Watson et al. identified poor knowledge of palliative medicine, failure to recognise the signs of dying, and a culture of rehabilitation as some of the barriers to implementing an end of life care pathway in nursing care homes.²⁴ With this in mind, we recommend the use of effective staff training interventions and greater GP collaboration to improve knowledge, and to provide the competence and confidence to make difficult decisions at the end of life stages. Emerging evidence demonstrates the importance of co-design between care home staff and external professionals when introducing end of life care programmes.^7,9

Whilst our study is the most detailed of its kind, outpatient medication costs were omitted. This may have artificially inflated the total healthcare cost share of inpatient services, because inpatient NHS Reference costs do include most medications.¹² However, we do not believe that this will have significantly affected our conclusions. Simoens et al. estimated pharmacy costs at €71 (£58) per resident in the final month of life.⁸ This equates to less than 2% of their total healthcare cost estimate, and would not have drastically altered the percentage share of inpatient costs in our study. A more significant limitation is that we adopted a healthcare payer perspective. Whilst a societal perspective is desirable, we suggest that in this instance the healthcare payer perspective gives a more accurate picture of societal costs than studies in other settings. For example, whilst informal care costs are often very significant,¹⁰ they are likely less significant in a population which is cared for in a nursing care home. Furthermore, costs such as those to the criminal justice system are unlikely to be relevant to a population which is largely elderly and frail. Productivity losses are also less significant to a population which is largely elderly and frail and very unlikely to return to work.

This study showed that end-of-life healthcare costs in nursing care homes were highest in the final month of life. These costs were largely driven by hospital admissions. Death in hospital predicted a very large increase in healthcare costs versus death in the nursing care home. Taken with other evidence, we suggest that staff uncertainty in recognising the dying process resulted in more hospital admissions, particularly for those residents who died following a dwindling trajectory. Implementing and importantly sustaining end of life care programmes in care homes alongside co-designed, staff-level interventions may show promise in reducing the proportion of inappropriate deaths occurring in hospital.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article:

We would like to acknowledge The Wives Fellowship and St Christopher's Hospice for the funding of the study and St Christopher's volunteers for their assistance with the study.

Note

References

Froggatt

Reitinger

. Palliative care in long-term care settings for older people EAPC taskforce 2010–2012 Report, Milan: EAPC, 2013.

National Audit Office. End of life care, London: The Stationery Office, 2008.

World Health Organisation. Good health adds life to years: Global brief for World Health Day 2012, Geneva: WHO, 2012.

Laing

. Care of elderly people UK market survey 2011/2012, 24th ed. London: Laing & Buisson Publications Ltd, 2012.

Department of Health. End of life care strategy fourth annual report, London, UK: Department of Health, 2012.

Gomes

Higginson

. Where people die (1974–2030): Past trends, future projections and implications for care. Palliat Med 2008; 22: 33–41.

Amador

Goodman

King

. Exploring resource use and associated costs in end-of-life care for older people with dementia in residential care homes. Int J Geriatr Psychiatry 2014; 29: 758–766.

Simoens

Kutten

Keirse

. Terminal patients in Belgian nursing homes: A cost analysis. Eur J Health Econ 2013; 14: 407–413.

Kinley

Stone

Dewey

. The effect of using high facilitation when implementing the Gold Standards Framework in Care Homes programme: A cluster randomised controlled trial. Palliat Med 2014; 28: 1099–1109.

10.

McCrone

. Capturing the costs of end-of-life care: comparisons of multiple sclerosis, Parkinson’s disease, and dementia. J Pain Symptom Manage 2009; 38: 62–67.

11.

Beecham

Knapp

. Costing psychiatric interventions, London: Gaskell, 2001.

12.

Department of Health. Reference costs 2011–2012, London: Department of Health, 2012.

13.

Curtis

. Unit costs of health & social care 2012, Canterbury: Personal Social Services Research Unit, 2012.

14.

Treasury

. The Green Book, London: The Stationery Office, 2011.

15.

Hockley

Watson

Oxenham

. The integrated implementation of two end-of-life care tools in nursing care homes in the UK: An in-depth evaluation. Palliat Med 2010; 24: 828–838.

16.

Sidell

Katz

Komaromy

. Death and dying in residential and nursing homes for older people: Examining the case for palliative care, London: Department of Health, 1997.

17.

Bowman

Whistler

Ellerby

. A national census of care home residents. Age Age 2004; 33: 561–566.

18.

Hays

Galanos

Palmer

. Preference for place of death in a continuing care retirement community. Gerontologist 2001; 41: 123–128.

19.

Meeussen

Van den Block

Bossuyt

. GPs’ awareness of patients' preference for place of death. Br J Gen Pract 2009; 59: 665–670.

20.

End of Life Care Network. Six steps to success. Six Steps to Success, 2011, http://www.endoflifecumbriaandlancashire.org.uk/six_steps.php (accessed 9 January 2015).

21.

GSF Centre CIC. Care Homes Training Programme, 2015, http://www.goldstandardsframework.org.uk/care-homes-training-programme (accessed 9 January 2015).

22.

Coordinate My Care. Coordinate My Care: a clinical service that coordinates care, giving patients choice and improving quality of life. 2012, http://coordinatemycare.co.uk/cmc/wp-content/uploads/2014/06/cmc-london-bulletin-sept-2012.pdf (accessed 9 January 2015).

23.

Alzheimer’s Society. Low expectations. Attitudes on choice, care and community for people with dementia in care homes, 2013, http://www.alzheimers.org.uk/site/scripts/download_info.php?fileID=1628 (accessed 9 January 2015).

24.

Watson

Hockley

Dewar

. Barriers to implementing an integrated care pathway for the last days of life in nursing homes. Int J Palliat Nurs 2006; 12: 234–241.