Transcript Disability, Dementia and the Future Costs of Long-Term Care

Disability, Dementia
and the Future Costs of
Long-Term Care
Adelina Comas-Herrera
In collaboration with Raphael Wittenberg, Linda Pickard, Derek
King, Juliette Malley and other colleagues
Personal Social Services Research Unit
London School of Economics and Political Science
Contact: [email protected]
Projecting the costs of long-term care
into the future
• We know we will get the wrong answer, unless we
manage to develop perfect foresight!
• But it is still useful to make projections:
– To understand the drivers of change.
– Not all variables involve the same level of uncertainty.
• Sensitivity analysis is vital to understand the
robustness of projections. This is particularly
important when making projections about different
ways of financing LTC that may affect policy
decisions.
Funnel of doubt:
Long-term care expenditure in the UK as % of GDP
Comparative base case
4
Using marital status
projections
Low Eurostat
3.5
3
High Eurostat
2.5
0.5 years delay dep.
2
1.5
1 year delay dep.
1
Grow th w ith GDP diff.
0.5
0
2000
Wages 0.5% slow er GDP
2050
Wages 0.5% faster GDP
Rise in formal home care
Determinants of future LTC costs
•
•
•
•
•
•
•
•
Demographic changes.
Trends in functional dependency/cognitive impairment.
Availability (and propensity to provide) informal care.
Structure of the LTC system and patterns of care.
Financing system.
Relative price of LTC and other goods and services.
Economic growth and other macroeconomic factors.
Values and public expectations about the quality, range
and level of care.
• Other factors? Quality/adaptability of housing, pensions…
Making projections of future LTC expenditure:
the PSSRU aggregate model
• The PSSRU aggregate model aims to make projections of:
–
–
–
–
Numbers of disabled older people
Long-term care services and disability benefits
Long-term care expenditure: public and private
Social care workforce
• The results are highly sensitive to changes in the assumptions
made about future disability rates.
• What should we assume about future disability rates?
The PSSRU LTC CI model
• Based on the England PSSRU aggregate LTC model
• Data from MRC-CFAS and PSSRU surveys of residents in care
homes is used to estimate the prevalence of CI and the
proportion of care users who have CI.
• Older people and users of services are divided into four
disability groups:
– No CI or functional disability (FD)
– FD but no CI
– CI only (used as proxy for mild stages of dementia)
– CI and FD (proxy for moderate/severe dementia)
• Simulates the impact on demand of specified changes in
demand drivers, or specified changes in patterns of care.
• Makes projections on the basis of specific assumptions about
future trends.
See, for example, Comas-Herrera et al., 2007
Key base case assumptions
• Numbers of older people change in line with GAD 2006-based
principal population projection.
• Age/gender-specific prevalence rates of cognitive impairment
and of functional disability remain unchanged.
• Marital status rates change in line with GAD 2003-based
marital status and cohabitation projections.
• Constant ratio of single people living alone to single people
living with others.
• Proportion of older people receiving informal care, formal
community care services and residential and nursing home
care remains constant for each sub-group by age, disability,
household composition and other needs-related
circumstances.
• Health and Social care unit costs rise by 2% per year in real
terms. Real Gross Domestic Product rises in line with HM
Treasury assumptions.
Probability of being in an institution for people
aged 75 to 84, by gender, household type and
dependency group.
Group probability of being in an institution
CI only
75-84 male
75-84 female
ADL only
ADL & CI
Alone
27
38
73
With others
37
49
72
Married
8
7
38
Alone
9
26
87
13
36
86
4
3
74
With others
Married
Source: PSSRU CI LTC model estimates (using MRC CFAS data).
Dementia
Expenditure projections for people
withUK
dementia 2002 to 2031
Projected total LTC
expenditure, at 2002 prices
LTC expenditure as % of Gross
Domestic Product
Red – older people with cognitive impairment; Blue - not
2.5
40
2
30
1.5
20
1
10
0.5
0
0
2002
Comas-Herrera et al, IJGP 2007
2031
2002
2031
Assumptions about future disability when
projecting future LTC expenditure:
1. Assuming unchanged disability rates
2. Extrapolating from past trends
3. Projections based on hypotheses linked to changes
in life expectancy (for example Brookings scenario).
4. Asking the experts for their view about the future.
5. Projections using epidemiological models of
chronic conditions and their mortality and
disabling outcomes.
1. Assuming unchanged age-specific
disability rates
• Seems a reasonable assumption in the context of uncertainty.
• Often criticised by policy-makers who argued this was a
pessimistic assumption in the context of increasing life
expectancy (this criticism involves an implicit belief that there
will be compression of disability).
• Sensitivity analysis showed that LTC expenditure projections
were very sensitive to small changes in disability rates.
2. Extrapolating from past trends
• A few studies have projected expenditure assuming that past disability
trends would continue into the future.
• This approach has been used particularly in countries like the US and
Sweden where disability rates had decreased overy time).
• BUT: Lack of good consistent time series data to estimate past trends in
most countries.
• Where consistent data are available, the trends that emerge are often
difficult to interpret:
– Different trends observed for different severity levels
– Changes in direction (reflecting epidemiological transitions?)
– And sometimes different surveys covering the same country and the same
period give contradictory answers (e.g. Great Britain between 1995 and 2001,
see LaFortune et al., 2007)
• Is the past necessarily a good predictor of the future?
3. Projections based on hypotheses
linked to changes in life expectancy
•
Modelling hypothetical assumptions that link the expected rise in life expectancy with
assumptions about changes in age-specific disability rates.
•
For example, the “Brookings scenario” shifts the rates of age-specific prevalence of disability
to higher ages as life expectancy increases. E.g. if life expectancy at age 65 is projected to rise
by 3 years between the year 2007 and the year 2030, then the disability rate of a person
aged 65 in the year 2007 would be applied to a person aged 68 in 2030.
•
More formally:
–
If D(x) t0 is the disability rate of a person aged x, in the year t0, the disability rate of a person aged x
at t1 would become:
D(x)t1 = D(x+)t0
(1)
Where = LE(x)t1 – LE(x) t0
With:
D(x)
= Disability rate for a person aged x
LE(x)
= Life expectancy at age x
t0, t1
= Points in time.
•
This method of modelling effectively decreases disability rates and, to a certain extent, can
compensate for increases in the numbers of older people when projecting long-term care
expenditure
4. Asking the experts about the future
• Normally involves using consensus building
methods (s.a. Delphi, focus groups…) to find a
set of assumptions about the future that a
group of experts agree with.
• Challenges:
– difficulties establishing how “representative” the
experts consulted are.
– Translating the experts views into future disability
or dementia prevalence rates may not be
straightforward
Expert (Delphi) panel on dementia futures
Expert panel: old age psychiatrists, geriatricians,
neurologists, public health doctors, basic
scientists, health economists, service
professionals … (n=26)
Considered various future scenarios. Panel was
‘moderately optimistic’ about impact of scientific
advances and changes in risk factors.
Overall scenarios chosen by the panel suggested:
– Small reduction in the prevalence of dementia.
– A freeze in the numbers of older people in institutions.
– An increase in the qualifications and pay of care assistants
that look after older people with dementia.
Comas-Herrera et al, Int Psychogeriatrics 2010
Translating the Delphi outcomes into
future scenarios
• Methodological issues: we were not able to model scenarios about
incidence or duration of dementia as we have a prevalence model.
• Where it was not possible to estimate the scenarios directly, we
approximated them as accurately as possible.
• Overall scenarios chosen by the panel suggested:
– Small reduction in the prevalence of dementia.
– A freeze in the numbers of older people in institutions.
– An increase in the qualifications and pay of care. assistants that look after
older people with dementia.
• We had to make assumptions about the size of some of the panel’s
suggestions. To illustrate the range of variation due to our interpretation
we produced two different interpretations of the panel’s views: a low and
a high expenditure scenario.
Impact of experts views on future dementia care
and costs projections
2002
2031
% increase between
2002 and 2031,
scenario
% increase between
2002 and 2031, base case
Low expenditure scenario combination
Numbers with CI only
280,000
590,000
111%
79%
Numbers with CI and ADL
275,000
400,000
47%
88%
All with CI
550,000
990,000
79%
83%
Numbers with CI receiving home-based services
110,000
250,000
124%
96%
Numbers with CI in care homes
205,000
205,000
0%
93%
5.4
14.3
163%
214%
0.60%
0.82%
Total LTC expenditure by people with CI (£bn)
LTC expenditure by people with CI as % of GDP
High expenditure combination of scenarios
Numbers with CI only
280,000
540,000
98%
79%
Numbers with CI and ADL
275,000
460,000
67%
87%
All with CI
550,000
1,000,000
82%
83%
Numbers with CI receiving home-based services
110,000
255,000
129%
96%
Numbers with CI in care homes
205,000
205,000
0%
93%
5.4
16.6
205%
214%
0.60%
0.96%
Total LTC expenditure by people with CI (£bn)
LTC expenditure by people with CI as % of GDP
5. Projections using the outputs of epidemiological
models of chronic conditions and their mortality and
disabling outcomes.
• As part of the MAP2030 project the PSSRU aggregate LTC
model was linked to the SIMPOPGENDER model (Carol Jagger,
Newcastle, Ruth Matthews and James Lindesay, Leicester).
• SIMPOPGENDER looks at the future pattern of disabling
diseases and the implications for the numbers of older people
with disability and disability-free life expectancy (DFLE) to
2030.
• builds on a previous model developed for Wanless Review of
Social Care.
• The age-specific prevalence of disability from the
SIMPOPGENDER model are incorporated in the PSSRU model
for all projection years.
See: Jagger et al. Age and Ageing 2009;38:319–25
Epidemiological model scenarios
(provided by SIMPOPGENDER, based on literature review for specific
conditions)
Scenario 1: Central Health Scenario
•
•
Age-specific prevalence of diseases, incidence & recovery rates all remain the
same.
Mortality rates continue to fall according to levels set by GAD principal
projection
Scenario 2: Current trends in health continue
•
•
•
Prevalence of arthritis, stroke, CHD and cognitive impairment INCREASED by 2%
every 2 years from 2012
Onset of disability INCREASED by 10% from 2012 in those with arthritis, stroke
and CHD
Mortality from Stroke, CHD and mild cognitive impairment REDUCED by 5%
from 2012
Scenario 3: Improving population health
•
•
•
Prevalence of arthritis, stroke, CHD, and mild CI REDUCED by 2% every 2 years
from 2012
Onset of disability REDUCED by 10% in those with arthritis, stroke, CHD and mild
CI from 2012
Mortality REDUCED by further 5% in those with stroke, CHD and mild CI from
2015
Projections of future use of social care and associated
expenditure under different assumptions about disability.
England, 2007 - 2032
2007
2032
Constant
Disability
prevalence
“Brookings”
scenario
Constant
Illness
prevalence
Continuation
of previous
trends
Improved
health
Home care
298,000
514,000
413,000
543,000
530,000
560,000
Institutional
care
332,000
609,000
365,000
800,000
715,000
896,000
Total LTC
Expenditure
(£billion)
17.5
49.8
35.5
59.1
54.9
63.8
LTC
Expenditure
as% of GDP
1.4%
2.7%
1.93%
3.2%
3.0%
3.5%
Comparison of Base and
MAP2030 scenarios
Total expenditure on long-term care, England 2007-2032
60,000
55,000
50,000
Expenditure (£m)
45,000
40,000
Base
MAP2030
35,000
30,000
25,000
20,000
15,000
10,000
2007
2012
2017
2022
2027
2032
Public expenditure as % of GDP, England 20072032, under alternative socio-demographic
assumptions
Public expenditure as a percentage of GDP, England 2007 to 2032
2.80%
2.60%
Expenditure as a %age GDP
2.40%
2.20%
2.00%
1.80%
1.60%
1.40%
1.20%
1.00%
2007
2012
England regime current
2017
England regime current + MAP2030
2022
Comprehensive
2027
2032
Comprehensive + MAP2030
Conclusions
• Projections of the future affordability of alternative financing regimes are
very sensitive to the assumptions made about future disability rates.
• The PSSRU model’s former base case scenario of constant age-specific
disability prevalence requires (in the context of increased life expectancy)
a decline in the prevalence, the disabling consequences and/or duration
of chronic illnesses.
• Unless there is strong evidence that points in the direction of such
declines, constant age-specific prevalence of disability is an optimistic
(instead of neutral) assumption.
• Improvements in the future health of older people will have a substantial
impact on future care needs and associated expenditure.
• The underlying causes of disability are changing, and we know the care
needs of people with different chronic conditions are very different (even
if they have the same number of ADLs). We need to develop more
sophisticated models that link better care needs and use of services.
References
•
•
•
•
•
•
Comas-Herrera A, Northey S, Wittenberg R, Knapp M, Bhattacharyya S, Burns A. (2010).
Future costs of dementia-related long-term care: modelling scenarios about the future.
International Psychogeriatrics (first view available online).
Comas-Herrera, A., Wittenberg, R., Pickard, L., Knapp, M. and MRC-CFAS. (2007). Cognitive
impairment in older people: its implications for future demand for services and costs.
International Journal of Geriatric Psychiatry, 22: 1037-1045.
Jagger C., Matthews R., Comas-Herrera A., Wittenberg R., Pickard L., Malley J., King D., and
MRC CFAS. (2009a). Disability in later life: causes, consequences and future prospects.
Presentation given at a London School of Economics seminar, 15th September 2009.
http://www.lse.ac.uk/collections/MAP2030/Seminars/Disability_seminar/MAP2030%2015%
20Sept%20Handout.pdf
Jagger C., Matthews R., Lindesay J., Robinson T., Croft P., Brayne C. (2009b).The effect of
dementia trends and treatments on longevity and disability: a simulation model based on the
MRC Cognitive Function and Ageing Study (MRC CFAS). Age and Ageing 38 (3): 319-325.
Wittenberg R, Pickard L, Comas-Herrera A, Davies B and Darton R. (2001) Demand for longterm care older people in England to 2031, Health Statistics Quarterly 12, 5-17.