Supporting Material is available for this work. For more information, follow the link from
the Table of Contents to "Accessing Supporting Material".
Table of Contents
Assessing the impact of a care innovation: telecare
Steffen Bayer, James Barlow, Richard Curry
Tanaka Business School
Imperial College London
South Kensington Campus
London, SW7 2AZ
UK
Telephone: +44 20 75945928; Fax: +44-20 75945915
s.bayer @imperial.ac.uk; j.barlow @imperial.ac.uk; richard.curry @ imperial.ac.uk
Abstract
The provision of health and social care for an increasing elderly population is a challenge
facing many societies. Telecare, the delivery of health and social care to individuals within
the home or wider community, with the support of ICT systems, has been advocated as an
approach to reduce the rise of the number of elderly people in institutional care and to
contain costs. A dynamic model is required to understand the systemic impact of telecare
implementation over time. The presentation cautions against overoptimistic expectations of
the impact of telecare in the short term and emphasises that the benefits of implementation
will only become fully effective with a significant delay.
Key words: telecare, elderly care, social care, health care, implementation
Introduction
The provision of health and social care for an increasing elderly population is a challenge
facing many societies. Policy makers and health professionals are seeking to develop
innovative approaches to the delivery of care services. Elements of new approaches include
an increased emphasis on rehabilitation, the delivery of care in new settings (such as in
intermediate care facilities and the client/patients’ own home) and the support for care
delivery using information and communications technologies (e.g. telecare systems).
Studying such service innovations in pilot projects gives important insights into their cost and
benefits. However, these pilot projects are limited in the extent to which they can give
reliable indications of the systemic impact of the implementation of the service innovations.
Moreover, the introduction of a service innovation in care delivery is likely to only gradually
change the system — sudden complete change from one system to another is neither feasible
nor likely. Understanding the time dimension of implementation and the time required for
changes to become effective is therefore important. Similarly, the distribution of cost and
benefits between different institutions and individuals, and influences on other parts of the
care system (and at later times), cannot be captured if innovations are studied in isolation.
To understand these consequences of service innovations on the level of a local care economy
over time we have used a system dynamics approach.
Background - telecare
Telecare involves the delivery of health and social care to individuals within the home or
wider community, with the support of systems enabled by information and communication
technologies (ICT). It can be distinguished from telemedicine, which involves applications to
support the exchange of information between health care professionals. Telecare is based on
the premise that people in need of care should be able to participate in the community as
much as, and for as long as, possible. Care should therefore be deliverable where it is most
appropriate and potentially anywhere in normal physical environments. At different times this
may be at home, in ‘lower intensity’ residential care settings or on a mobile basis in the
normal daily living environment.
A typical telecare service involves a system connecting sensors in dispersed homes or worn
by the user to a call centre. The sensors are activated in case of need, for example following a
fall or an overflowing bath, and elicit a response by the alarm service, involving informal
carers or mobile staff as appropriate. In some ways, telecare is simply a development of the
existing community alarm services, currently covering over 1.5 million people in the UK,
into a more proactive form of monitoring by incorporating passive alarms and sensors which
can alert the call centre automatically when hazards arise or if there are other indications for a
problem such as a significant lack of activity of the user. More advanced telecare services
might be linked to electronic assistive technology (e.g. environmental controls). Other forms
of telecare include the provision of health information.
Safety and security
monitoring, e.g. bath
overflowing, unlit gas left on, =
door unlocked
Mitigating risk
8 XX
Social inclusion
XX
Information ye praia i Personal
provision e.g. fs ome Of, monitoring:
health advice, a ys amet - physiological signs
triage, access to environme! - activities of daily
self-help groups living
‘>
Electronic assistive
technology, B e
e.g. environmental controls, doors
opening, control of beds
Improving
functionality
xX
Telecare therefore provides a service that can be remotely delivered and is location
independent. It is flexible and expandable to meet changing individual needs. It fits into a
care package derived from an individual assessment of need which can include domiciliary
care, assistive technology and home nursing care.
A major function of telecare is to enable older and disabled people to remain in their own
homes by providing increased safety and reassurance to them and their carers. Telecare thus
has important implications for the future location of care delivery because it can potentially
transform a previously unsuitable environment into one that is sufficiently safe for a patient
to be discharged to or remain in as their condition changes. Proponents of telecare have high
hopes of the potential of ICT supported care delivery to transform the care system, slow the
increase in the cost of care delivery while preserving the independence of elderly people in
their own homes. Particular expectations include a reduction of the cost of care provision, the
prevention of unnecessary admission to institutional care as well a beneficial effect on the
maintenance of health (through the earlier detection of signs of deterioration and appropriate
action). It is not only hoped that telecare will reduce the need for nursing and residential care
provision, but also that it will reduce the need for hospital care by avoiding hospitalisation
(through the early detection and management of crises in the community) and through a
speedier, but safe discharge back into the community.
In the UK, government policy is playing a direct part in stimulating interest in telecare. Of
particular significance are the moves since the early 1990s towards increased care provision
outside the institutional context. This is partly a result of a concern to provide people with
greater choice over their care pathways and partly because of moves to contain the cost of
providing care. The majority of older people prefer to remain in their own homes for as long
as possible and residential or nursing home care is viewed very much as the choice of last
resort (Henwood ef al., 1998; Smith et al., 1993). The typical care and housing pathway for
older people — from mainstream housing to sheltered housing or residential care and
eventually to hospital and/or nursing care — has been widely criticised (Royal Commission,
1999). A move away from this approach is seen as desirable.
Because of its ability to mitigate the risk of individuals living in the community when they
would otherwise be in institutional care, there is substantial interest by the UK government in
the opportunities telecare may offer (Department of Health, 2001a-c). Even though the
expectations for the benefits of telecare are high, actual experience (Roush, 1995; Wootton et
al., 1998; Balas and Iakovidis, 1999) of (mainstream) service delivery through telecare
remains scarce (Barlow ef al., 2004). While trials of telecare services suggest substantial
benefits at the individual level, the understanding of the effect of the introduction of telecare
on the system level is more limited. Telecare has generally been evaluated with a view to
exploring clinical outcomes rather than systemic impacts. Available data from trials of
specific telecare services in specific situations with specific (and sometimes only vaguely
defined) patient groups might not give more than an indication for the potential effects of
telecare in other settings and even less so its systemic implications. In addition, the base line,
i.e. the current system of service delivery is also only partially understood. In particular,
evidence of the change in health status of people in the community over time is mostly
anecdotal.
Research approach
This paper aims to examine these expectations for the impact of telecare. In particular we
want to investigate how realistic the hopes placed in telecare are in the short-term to medium-
term (i.e. over 5 years).
Clearly the data available to answer these questions is not as plentiful and of as high a quality
as would be desirable. However, system dynamics has often succeeded in gaining valuable
insights even in situations where the available data has been quite limited. Indeed, its ability
to develop an understanding of the behaviour of a system based on the structure of the system
and only approximate parameter values is one of the key strengths of this approach. However,
it is necessary to be careful not to turn this methodological strength into an application
weakness by inappropriate neglect of data. Homer (1996, 1997) rightly emphasises how
structure and policy conclusions can depend on adequate use of data.
We have therefore to establish to what extent in our case the limited data allows a judgement
on the systemic impact of telecare to be made. Sensitivity analysis is required in order to
ensure that insights into fundamental relationships, system behaviour and policy conclusions
derived from our model are robust.
Previously, system dynamics has been used for the study of policy issues in the provision of
health and social care for the elderly (e.g. Wolstenholme, 1993 and 1999, Linard, 1996,
Walker and Haslett, 1999). In some of this work the availability and quality of data was
severely restricted. In their system dynamics work on care services for the elderly,
Wolstenholme (1993 and 1999) and Linard (1996) have shown that insights can be gained
from structural analysis and estimates of some parameters. Related work on care for patients
with chronic illnesses is also relevant (Homer et al., 2003)
We have carried out this modelling work as part of a larger research project (funded by the
UK Engineering and Physical Science Research Council) on telecare implementation. This
project brought together a local health trust, a social services department and a range of
industrial partners. The project evaluated the implementation of a telecare scheme for frail
elderly people, examined organisational barriers to telecare implementation and explored the
wider implications of mainstream telecare deployment using a combination of scenario
building and simulation modelling. Primary data as well as the views of the various
stakeholders for a (potential) telecare implementation process were collected through a range
of workshops and in interviews.
A system dynamics model of care delivery
The starting point of our system dynamics work is a highly aggregated, simplified description
of the main patient/client flows in the system before the implementation of the telecare
innovation.
Elderly people enter the model as they reach the age of 65 (first stock in the model). After
some time, some of these elderly people will become frail and require low intensity home
care (HC fL); with increasing frailty more intensive home care packages will be required (HC
fM, HC fH). Some of the clients of the two higher intensity home care packages are referred
on to institutional care (Inst) representing both nursing and residential homes. As the
capacities of care homes are limited, a waiting list for institutional care exists. In our model,
clients wait in their previous home care setting for new places to become available.
Clients/patients leave the system mainly as a consequence of death. The provision of services
for clients in different settings incurs costs for the system. Among the clients receiving
homecare the model assumes that both deterioration and improvement in frailty is possible.
(Figure shows only stock — flow structure; influences on the rates are omitted for readability).
The categorization of elderly people according to their frailty is based on an indicator
developed by Arber (1991) and applied to data from the 1997 Elderly Follow-Up Study to
British General Household Survey! (Office for National Statistics). (The medium frailty
category comprises two of Arber's categories). The possibility of improvement in frailty was
included since the available data from the 1997 Follow-Up Study of Elderly respondents to
the 1994 General Household Survey shows this to be an important feature. Service use of
clients in different frailty categories was derived from the 1994 General Household Survey
and costed according to typical 2003 prices (Netten/Curtis, 2003).
from HC to waitin
Inst entry 3 a le f2tofL death rate Inst
p y waiting entry {M
Inst M_F7; Ting | Inst Mt a
death rate HC priiioetle
death rate hi fh, 7
from HC fLto} fromHCfLto | from HC tH to to waiting Inst
HG IM ™ from HC fH from waiting to death atest
Cog fhealty Hot | HC IM en waling [stent 4 —_entey ;
ging from healthy to from he (M fo7—Trom He tht Inst fH
HC IL iL death rate HC [death r w Inst
death Fate HC HOM I pe a
iM
4
Building on this model, we can then examine the systemic impact of the introduction of
telecare. The structure representing telecare is parallel to that of standard home care. We can
use the model to explore the consequences of different assumptions about the differences of
telecare and standard care provision concerning
- the cost of care provision;
- the impact on the prevention of institutionalisation; and
- the impact on disease progression.
The model allows us to explore the effect of the introduction of telecare under different
scenarios, particularly the effect on the number of clients cared for in institutional care and
overall cost. We have chosen these two indicators as the most relevant to current care policy
and care provider concerns.
' Disclaimer note: The original data creators, depositors or copyright holders, the funders of the Data
Collections and the UK Data Archive bear no responsibility for their further analysis or interpretation of
this data.
a
from HC to waitin
Inst entry 3 ° He eet death rate Inst
fwatng Inst entry tM
from waiting to aN 2
death rate HC Inst entry 3
death rate hi fl, "
from HC fL to from HC 1H to
iM to Sai rt from waiting to death rate Inst
Hom [= Wartng inst entry 4 entry tH
fom Rc IMT [from HO Wut
io HC th
death rate HC
iM
fay
effect of TC on frac
share to
rate to inst care entry
fl
effect of TC on frac
Rompe to \ rate to inst care entry
g fH
ito waiting Inst 4
Tom Tp tl to from TC {M CtH to
‘waiting Inst
from TC fMto | from TC (Mto
. fl TOtH
—!
co
death rate To] TOL TC mM TotH
fl from TH fL to fron TC fH to
Tow death ratdtc pea me TS.
\ / 1M 3
So Bh
effect of TC on fty
progression
The 1997 Follow-Up Study to the General Household Survey 1994 has been used to estimate
the size of the stocks of people of different frailty in the community. This national data was
then scaled down to the population size of a particular local area (Surrey). However, many
parameters in our model are based on estimates. The share of people in each frailty category
moving to other categories remains constant. Transition rates have been chosen such that in
absence of the introduction of telecare (and without taking into account demographic change)
we are in a steady state: elderly people enter the system at the same rate as people leave the
system due to death, and the size of the frailty groups at home, as well as the institutional
population, remains constant over time. In a first approximation demographic change and the
growing number of elderly people can be ignored for the time horizon of 5 years in which we
are interested”. Nevertheless, it would be desirable to improve the data basis of our model.
The representation of the system with stocks and flows was also helpful to communicate with
project partners the implications of different patient routes. To understand the system at the
level of patient flows is in principle comparatively easy; in practice the collection of accurate
data is a significant challenge.
The dynamics of telecare implementation
Cost-benefit studies of proposed care innovations typically compare the impact of
implemented (and fully effective) changes with the current system of service delivery. This
state is however not reached immediately. For telecare this transition period is likely to be
2 Demographic change would change the rate with which elderly people enter the system. The
increasing life-expectancy after 65 indicates that frailty progression and/or fractional death rates will
also be adapted in the long-term to reflect this.
long. For two main reasons many of the benefits of telecare will only become fully effective
many years after its implementation.
A complete switch from the current status to a radically changed mode of service delivery
cannot happen over night. It is undesirable for existing, particularly frail clients or those
already in institutional care to suddenly adapt to a telecare enhanced service delivery in their
own homes. Moreover, organisational barriers to implementation have to be overcome;
training of staff and development of infrastructure are required, and can only take place over
time. Only a gradual shift towards a telecare supported care delivery system is therefore
realistic and desirable. This shift will ideally start with those elderly people requiring social
services homecare for the first time. For this reason our model allows for a switch to telecare
only at this stage.
Telecare changes the rates of entry to institutional care homes, where residents will remain on
average for several years. The full impact of the changed admission rates on the size of the
institutional care population is therefore only realised long after the switch to the new model
of care delivery has been made.
Simulation experiments
Our aim is to examine the likely effect of the implementation of telecare on the population in
institutional care homes in the short to medium term in an UK locality. Telecare is expected
to change the entry rates to institutional care from community care (from medium and high
frailty) and on the progression of frailty. However, as telecare services are still under
development, their precise effects on the fractional entry rates to institutional care are
uncertain. Similarly it is unclear which people requiring care for the first time will be suited
to a care package which includes telecare. This will depend on the telecare services offered in
a specific locality. In order to deal with this uncertainty we have performed a range of
sensitivity simulations.
We first look at the sensitivity in the different parameters in isolation and then run a
combined sensitivity analysis. Subsequently, we examine the likely financial impact of the
introduction of telecare under different assumptions of the cost differences between telecare
and conventional forms of care delivery.
Telecare and the entry to institutional care
Telecare helps individuals to avoid the admission to institutional care. The percentage of
individuals of medium and high frailty entering institutional care will therefore be lower for
those receiving a package including telecare than for those receiving a standard home care
package. It can be expected that telecare will be more effective in preventing admission to
institutional care from the medium frailty category than from the high frailty category since
very frail patients are likely to need either more frequent hands-on help or a quicker response
in case of a crisis than can be provided in a home setting without the permanent presence of a
carer. For many individuals telecare will therefore postpone the admission to institutional
care rather than making it entirely unnecessary.
We have simulated the size of the institutional population over 5 years depending on the
effect telecare has on the entry to institutional care. We have done this under the assumption
that telecare has no impact on frailty progression and that 50% of all people requiring care for
the first time receive a telecare package.
The figure below shows the size of the institutional population after the introduction of
telecare, assuming that the fractional entry rate of medium frail people into institutional care
is reduced to 0%, 10%, 20%, 30%, 40% or 50% of the fractional entry rate for non-telecare
recipients. The coloured lines assume that telecare has no reducing effect on entry from the
high frailty category. The intermediate black lines show the size of the institutional
population if telecare also reduces the fractional entry rate from the high frailty category to
50%, 60%, 70 %, 80% or 90% of the fractional rate for non-telecare recipients’.
sensvay etry to inst care
ry M=0 5%: =100%
base long
500)
45
450
400
o 5 0
“Tine (Mons)
Under the given assumptions telecare reduces the size of the institutional population 5 years
after implementation from 467 to between 435 and 455, i.e. a drop of between about 2.5%
and 7%. The effect of telecare on entry from the medium frailty category is far more
important for the size of the institutional population than the effect on entry from the high
frailty category. This is because there will be virtually no effect on the size of the institutional
population until the low frailty patients which first receive telecare become frailer.
° The number given in graph refers to the fraction of people of one fragility group with telecare
entering the service compared to the standard homecare case. "0.5" for the variable telecare effect on
entry FM means therefore that the fractional rate of entry to institutional care from the medium frailty
group is in the telecare case half what it was in the standard homecare case.
Frailty progression and its effects
It is hoped that telecare will have some effect on the progression of frailty since an earlier
indication of the first signs of deterioration might allow measures to slow this process to be
put in place. Following this argument we have conducted a range of experiments varying the
effect of telecare on frailty. To do this we assumed that the share of new clients receiving
telecare will be 50% and that telecare reduces the fractional entry rate to institutional care
from the medium frailty category to 20% and from the high frailty category to 80% of the
non-telecare case.
The figure below shows simulation runs with the inclusion of an effect which reduces the rate
of progression by 10%, 20% and 30%. These experiments show, however, that the effect on
the institutional population will be small (less than 1% of the institutional population).
inst care
500
475
450
o 3 6 9 2 I IS 21 2 oF 30 33 36 99 42 45 48 SI s4 57 60
‘Time (Month)
inst care : base long people
inst care : frailty prog 70% people
inst care : frailty prog 80% people
inst care : frailty prog 90% people
inst care : frailty prog 100% people
This explicit effect of telecare on frailty progression is not the only effect it could have on the
composition and size of the elderly population. This is because different fractional death rates
are associated with the different settings and in particular the death rate in institutional care is
higher than the death rate at home. It is unfortunately impossible to say whether the
underlying assumptions leading to this effect are correct. This is not only due to the lack of
suitable data on telecare, but also due to a lack of understanding of the base case. The data on
whether and how morbidity changes in individuals and particularly in an aging population is
sparse. While the related debate on the "compression of morbidity" (Fries, 1980; Schneider
and Brody, 1983) is inconclusive, there are indeed indications that individuals not only live
longer, but also enjoy a healthy life for a longer time (Hessler et al., 2003; Doblhammer and
Kytir, 2001). This would therefore lead to a very small increase of the elderly population
even if, ignoring demographic change, the number of people reaching retirement age
remained constant.
The share of telecare recipients
In a further set of experiments we varied the share of first time care recipients receiving
telecare between 0% and 100%. For this set of simulation runs we assumed that telecare
reduces entry to institutional care from the medium frailty category by 80% and from the
high frailty category by 20%. We further assumed that telecare has no effect on frailty
progression.
If we simulate the care system over a period of five years, the model shows that under the
given assumptions the drop in the number of people in institutional care is small. Even in the
most extreme case, assuming 100% of those now requiring home support will get a telecare
enhanced package, the drop in institutional population after 5 years is only about 12%.
Given that not all clients might be suitable, assuming a share of telecare referral of 50%
might be more realistic. This leads to a reduction in the institutional population by 6%. The
limited effect of telecare on the institutional population is due to the time required for
substantial numbers of this cohort of elderly people becoming so frail as to require high
intensity care.
However, this discussion has not yet taken into account supply side delays in implementation.
The build up of infrastructure, training requirements, local policy and other factors might
further reduce the number of people to which telecare services can be delivered initially. This
would then further delay the impact of telecare implementation on the institutional care
population.
inst care
soo
450
a
Tine (Mow
ist cate share to TC L005 people
int cat share to Te people
inst cate share o TC People
int cat hare to Te People
inst cate base long People
These simulation experiments suggest that an argument for telecare based on the expectation
of a short-term reduction in the institutional care population has to be treated with some
scepticism. This discussion also shows that static considerations comparing the present state
with a future state of care delivery (ignoring the transition time) are less informative for
health and social care planning than a more dynamic point of view.
10
The combined effect
In order to assess the combined effect of different assumptions for all the parameters
characterising telecare, we have performed a sensitivity analysis varying the effect on frailty
progression between 70% and 100%, the effect on entry from high frailty between 50% and
100% and the effect on entry from the medium frailty category between 0% and 50%. We
further varied the share of first time care recipients receiving telecare between 25% and 75%.
Under the most optimistic combination of assumptions the institutional care population in
month 60 will drop by 11% to 417, compared to the non-telecare case. Under the most
pessimistic combination of assumptions the drop will be less than 1%.
combined fleet
base ng
S00
450
2s
00
° is w ws 0
Tine Mont)
The long-term effect
In the long-term, the effect of telecare will be far more pronounced. A simulation over twenty
years, allowing the care delivery system to get closer to a steady state again, shows a very
substantial drop in institutional population. The uncertainty in the parameter values
characterising telecare is however such that estimates of the long-term effect on the
institutional population are virtually useless. Moreover, as telecare is based on evolving
technology its characteristics are likely to change substantially over such a long time period.
The graph below shows a set of simulations over twenty years for different assumptions
regarding the two key parameters. The effect of telecare on the entry to institutional care from
telecare enhanced homecare is varied while the other parameters are kept constant. The
coloured lines indicate runs for different values for the parameter describing the effect of
telecare on entry from medium frailty categories varying from a reduction to no entries from
this category (0%) to a reduction to 50% of the fractional entry rate without telecare.
1
Nevertheless, the simulation experiments might give an indication in which direction telecare
should be developed to have the biggest impact. Even for this very long time period the
impact on the institutional care population of reductions in the fraction of entry of telecare
recipients to institutional care, is considerably stronger for the case of medium frail telecare
recipients than for very frail telecare recipients. This suggests that — at least from this
perspective - the focus of the development of telecare services and technology should be on
reducing the entry to institutional care from medium, rather than high, frailty groups.
5%; 1H=100%
base long
inst care
600
500
400
0 60 120 180 240
‘Time (Month)
Financial considerations
In a further set of simulation experiments we investigated the cost of care service provision
under several scenarios of telecare effectiveness and cost. We assume here that 50% of
clients receiving care for the first time receive a package including telecare. We further
assume that telecare has no direct effect on frailty progression, but results in a 20% reduction
of entry into institutional care for high frailty clients.
We have run this simulation for 60 months for different combinations of telecare cost
(compared to standard homecare) and telecare effectiveness (in terms of its effect of entry
into institutional care from the medium frailty category). As can be seen from the table
below, under these assumptions telecare is only cost saving over 5 years if it is not more
expensive than standard homecare. (Any set-up costs are excluded; including these would
result in an even more pessimistic assessment.) Hopes that highly effective, but also more
expensive telecare services might be cost-saving, appear therefore - at least for the time
horizon of most care providers - questionable.
12
Cost over 5 years telecare case compared with standard care delivery system
telecare costs as share of standard homecare costs
effect 05 [06 10.7 [08 09 ]1 [14 [12 [13 [14 ]15
of TC [0.0 | -5% | -4% | -3% | -3% | -2% | -1% [0% |1% | 2% [2% | 3%
aa 0.1 | -5% | -4% | -3% | -2% | -2% | -1% | 0% |1% [2% [3% | 3%
rate [02 | -5% | -4% | -3% | -2% | -2% |-1% [0% |1% [2% |3% | 3%
to 0.3 | -5% | -4% | -3% | -2% | -1% | -1% | 0% |1% | 2% [3% | 4%
inst [0.4 | -5% | -4% | -3% | -2% | -1% | -1% [0% [1% | 2% [3% | 4%
care [0.5 | -5% | -4% | -3% | -2% | -1% | 0% [0% |1% | 2% [3% | 4%
entry [0.6 | -5% | -4% | -3% | -2% | -1% | 0% [0% |1% | 2% [3% | 4%
M07 [-4% |-4% | -3% | -2% |-1% [0% [0% |1% [2% [3% |4%
0.8 | -4% | -4% | -3% | -2% | -1% [0% |1% |1% | 2% [3% | 4%
0.9 | -4% | -3% | -3% | -2% | -1% [0% |1% [2% [2% [3% | 4%
1.0 | -4% | -3% | -2% | -2% |-1% [0% | 1% [2% [2% | 3% | 4%
Conclusions
The limited trials of telecare services have generally been evaluated with a view to exploring
their individual clinical outcomes, rather than systemic impacts. The work reported here
suggests that an increased emphasis should be put on understanding the systemic impacts of
the implementation of service innovations over time. While clearly more research and better
data are needed, this paper has given some indications that overly optimistic assessments of
the systemic effects of telecare in the short to medium term should be avoided — in the short-
term focusing on medium frailty clients and lower cost solutions might be more viable from a
policy and clinical effectiveness perspective. The importance of the time dimension of
implementation is so strong in the case of telecare for two reasons. First, as we argued earlier,
the particular characteristics of this type of care service and its users prevent rapid change.
Second, the effects of telecare can only have a delayed impact on the size of the institutional
care population since this population will for a long time largely comprise cohorts of elderly
people who entered institutional care before the telecare was introduced.
The availability of data remains a limiting factor for not only the model presented here but
also for the development of service and business models for telecare for three related reasons.
Firstly, there is little reliable data on the scale of the care need in the areas that telecare
addresses. Secondly, because of the paucity of evidence for the impact of telecare on resource
flows within the care system as a whole. Thirdly, the current lack of rigorous data on the
system-wide distribution of costs and benefits inhibits the development of sustainable
telecare business models, since potential telecare service providers are unable to make
accurate pricing decisions. As more trials are conducted and more evidence is gathered, it
should become easier to define more detailed business models.
The data we had available for the modelling in this paper was probably too limited to
adequately understand the long-term consequences of telecare implementation. Nevertheless,
important insights into the short-term behaviour of the changing system and more generally
into the importance of the dimension of time in the implementation process were derived.
Since only gradual shifts in the model of care delivery are possible (because of the limited
ability of the system to change and the limited abilities of the clients/patients to adapt to
13
change) an essentially static comparison of different models of care delivery is not enough.
The benefits (as well as the costs) of any change in care delivery can only be understood if
this time dimension is taken into account. Cost-benefit studies of proposed care innovations
in which the effect of implemented changes is compared with the current system of service
delivery therefore have to be complemented with dynamic considerations of the process of
change. In our example the time period of change is long and the time when the change is
completed and the benefits of telecare on the system are fully effective is likely to lie outside
the time horizon of policy and other decision makers. In the long term the effect of telecare
on the size of the institutional population might possibly be substantive, in the short to
medium-term this will not be the case. On a restricted time horizon telecare will possibly be
financially more viable when benefits in the health care sector (as opposed to the social care
sector) are included; telecare might help to reduce hospital admissions and save money in the
shorter term. The research reported here focused on social care costs, these considerations are
therefore beyond the scope of the research reported; the question of the effects of telecare
implementation on hospital costs should be the subject of additional research.
While the importance of the implementation process and its time dimension might seem
obvious, the expectations of proponents of innovations and the behaviours of decision makers
—and certainly not only in health and social care delivery — testify to the ease with which this
can be forgotten in practice.
Acknowledgements
This work was funded by the UK Engineering and Physical Science Research Council. We
thank our project partners and the project advisory group for their help and advice.
14
References
Arber S, Ginn J. 1991, Gender and Later Life. Sage Publications, Appendix B.
Balas, E. and Iakovidis, I. 1999. Distance technologies for patient monitoring, British
Medical Journal, 319: 1309.
Barlow, J.; Bayer, S.; Curry, R. 2004. Flexible homes, flexible care, inflexible organisations?
The role of telecare in supporting independence. Housing Studies (forthcoming)
Department of Health. 2001la. Building capacity and partnership in care. London,
Department of Health.
Department of Health. 2001b. National service framework for older people. London,
Department of Health.
Department of Health. 2001c. Community Equipment Services HSC 2001/008. London,
Department of Health.
Doblhammer G, Kytir J. 2001. Compression or expansion of morbidity? Trends in healthy-
life expectancy in the elderly Austrian population between 1978 and 1998. Social Science
and Medicine 52(3): 385-391.
Fries JF. 1980. Aging, natural death, and the compression of morbidity. New England
Journal of Medicine 303: 130-135.
Henwood M, Waddington E. 1998. Expecting the worst? Views on the future of long term
care. Help the Aged Report to the Royal Commission on Long Term Care. Leeds, Institute
for Health.
Hessler RM, Bo G, Eriksson BG, Dey D, Steen G, Sundh V and Steen B. 2003. The
compression of morbidity debate in aging: an empirical test using the gerontological and
geriatric population studies in Géteborg, Sweden (H70). Archives of Gerontology and
Geriatrics 37(3): 213-222.
Homer J. 1996. Why we iterate: scientific modelling in theory and practice. System Dynamics
Review 12(1): 1-19.
Homer J. 1997. Structure, Data, and Compelling Conclusions: Notes from the Field. System
Dynamics Review 13(4): 293-309.
Homer J, Hirsch G, Minniti M, Pierson M. 2003 Models for Collaboration: How System
Dynamics Helped a Community Organize Cost-Effective Care for Chronic Illness.
Proceedings International SD Conference, New York, USA.
Linard KT. 1996. System Dynamics Modelling and Aged Care, Modelling the Future -
Techniques & Directions, Australian Institute of Health & Welfare Workshop.
Office for National Statistics Social Survey Division, General Household Survey, 1994:
Follow-Up Survey of the Health of People Aged 65 and Over [computer file]. Colchester,
Essex: UK Data Archive [distributor], 2 September 1998. SN: 3903.
Netten A, Curtis, L. 2003. Unit Costs of Health and Social Care 2003. Personal Social
Services Research Unit, Canterbury.
Royal Commission on Long Term Care. 1999. With respect to old age. Long term care —
rights and responsibilities. Cmd. 4192. London, The Stationery Office Ltd.
Roush, R. et al. 1995. Impact of a personal emergency response system on hospital utilisation
by community residing elders, Southern Medical Journal, 1995, 88(9): 917-922.
15
Schneider EL and Brody J. 1983. Aging, natural death, and the compression of morbidity:
another view. New England Journal of Medicine 309 (14): 854-856.
Smith S, Alexander A, Hill S, McGuckin A , Walker C. 1993. Housing provision for people
with health problems and mobility difficulties. Housing Research Findings No. 86 York,
Joseph Rowntree Foundation.
Walker BC, Haslett T. 1999. System Dynamics and Action Research in Aged Care.
Proceedings International System Dynamics Conference.
Wolstenholme EF. 1993. A case study in community care using systems thinking. Journal of
the Operational Research Society 44: 925-934.
Wolstenholme EF. 1999. A patient flow perspective of U.K. Health Services: Exploring the
case for new "intermediate care" initiatives. System Dynamics Review 15: 253-271.
Wootton, R.; Loane, M.; Mair, F.; Moutray, M.; Harrisson, S.; Sivananthan, S.; Allen, A.;
Doolittle, G.; Mclernan, A. 1998. The potential for telemedicine in home nursing. Journal of
Telemedicine and Telecare, 4(4): 214-216.
16
Appendix: model (Vensim)
(001)
(002)
(003)
(004)
(005)
(006)
(007)
(008)
(009)
(010)
(O11)
(012)
(013)
(014)
(015)
(016)
(017)
(018)
(019)
(020)
(021)
(022)
(023)
(024)
(025)
(026)
(027)
(028)
(029)
(030)
(031)
(032)
(033)
(034)
(035)
(036)
(037)
(038)
(039)
(040)
(041)
(042)
(043)
(044)
(045)
"65+ frac death rate" = 0.00493476
aging = total stable deaths
average length of stay entry 3 = 35
average survival inst 2 = stable inst 2 / stable death inst 2
average survival inst c = 28
capacity Inst = stable inst
CC costs per time period = ( HC fL costs per client and month * HC fL + HC f{M
costs per client and month * HC fM + HC fH costs per client and month * HC fH ) *
SS share of community care costs
community = waiting in community + TC + HC
Community care costs = INTEG( CC costs per time period , 0)
community care no TC = HC fL + HC fM + HC fH
Community Care with TC costs = INTEG( TC CC costs per time period , 0)
community waiting for institutional care = waiting Inst f{M + waiting Inst fH
cost waiting in comm per time period = waiting at home costs per client and month *
( waiting Inst {M + waiting Inst fH ) * SS share of institutional care costs
death r w Inst fH = waiting Inst fH * frac death rate fH
death rate h = frac deaths rh * healthy
death rate HC fH = frac death rate fH * HC fH
death rate HC fL = frac death r fL * HC fL
death rate HC fM = frac death r f{M * HC {M
death rate Inst entry fH = Inst fH / average survival inst 2
death rate Inst entry f{M = Inst fM / average length of stay entry 3
death rate TC fH = frac death rate fH * TC fH
death rate TC fL = frac death r fL * TC fL
death rate TC fM = frac death r f{M * TC f{M
effect of TC on frac rate to inst care entry fH = 0.8
effect of TC on frac rate to inst care entry f{M = 0.2
effect of TC on fty progression = 1
FINAL TIME = 60
fit criteria : all h,all fL,all f2,all f{M,all fH,age cohort
frac death r fL = 0.003
frac death r fM = 0.004
frac death rate fH = 0.005
frac deaths r h = stable deaths h / stable h
frac from fH to inst = stable from fH to inst / stable fH
frac from fM to inst c = stable deaths inst 1 / stable {M
frac imp fH = 0.01
frac imp fL = 0.01
frac imp f{M = 0.01
frac prog fM to fH = stable from fM to fH / stable {M
frag prog fL = stable from fL to f{M / stable fL
frag prog h = stable from healthy to fL / stable h
free cap Inst = capacity Inst - used beds Inst
from HC fH to fM = HC fH * frac imp fH
from HC fL to h = HC fL * frac imp fL
from HC fL to HC fM = frag prog fL * HC fL
from he fM to fL = HC fM * frac imp {M
17
(046)
(047)
(048)
(049)
(050)
(051)
(052)
(053)
(054)
(055)
(056)
(057)
(058)
(059)
(060)
(061)
(062)
(063)
(064)
(065)
(066)
(067)
(068)
(069)
(070)
(071)
(072)
(073)
(074)
(075)
(076)
(077)
(078)
(079)
(080)
(081)
(082)
(083)
(084)
(085)
from HC fM to HC fH = frac prog fM to fH * HC f{M
from HC to waiting Inst entry 3 = HC fM * frac from fM to inst c
from healthy to HC fL = frag prog h * healthy * ( 1 - share to TC )
from healthy to TC fL = healthy * frag prog h * share to TC
from TC fH to fM = TC fH * frac imp fH
from TC fL to h= TC fL * frac imp fL
from TC fL to TC f{M = effect of TC on fty progression * frag prog fL * TC fL
from TC fM to fL = TC fM * frac imp {M
from TC fM to TC fH = effect of TC on fty progression * frac prog f{M to fH * TC f{M
from waiting to Inst entry 3 = MIN ( indicated waiting fM to Inst , max inflow Inst
entry 3 )
from waiting to Inst entry 4 = MIN ( indicated waiting fH to Inst , max inflow Inst
entry 4)
HC = HC fH + HC fL + HC {M
HC f2 to fL = waiting Inst fM * frac death r f{M
HC fH = INTEG( from HC fM to HC fH - death rate HC fH - to waiting Inst from HC
fH - from HC fH to fM, stable fH )
HC fH costs per client and month = 0.463
HC fL = INTEG( from healthy to HC fL - death rate HC fL - from HC fL to HC f{M
+ from he fM to fL - from HC fL to h, stable fL )
HC fL costs per client and month = 0.255
HC fM = INTEG( from HC fL to HC f{M - from HC fM to HC fH - from HC to
waiting Inst entry 3 - death rate HC fM - from he fM to fL + from HC fH to
fM , stable fM )
HC fM costs per client and month = 0.264
healthy = INTEG( aging - death rate h - from healthy to HC fL - from healthy to TC
fL + from HC fL to h + from TC fL toh, stable h )
indicated waiting fH to Inst = waiting Inst fH / minimum filling time
indicated waiting fM to Inst = waiting Inst {M / minimum filling time
INITIAL TIME = 0
inst care = Inst fH + Inst {M
Inst cost per time period = Inst costs per patient and month * ( Inst fH + Inst {M ) * SS
share of institutional care costs
Inst costs = INTEG( Inst cost per time period , 0)
Inst costs per patient and month = 1.79
Inst fH = INTEG( from waiting to Inst entry 4 - death rate Inst entry fH , stable inst
2)
Inst {M = INTEG( from waiting to Inst entry 3 - death rate Inst entry fM , stable inst
1)
max inflow Inst = free cap Inst / minimum filling time
max inflow Inst entry 3 = max inflow Inst * share waiting Inst entry 3
max inflow Inst entry 4 = max inflow Inst * share waiting Inst entry 4
minimum filling time = 0.25
"pop 65+" = 9429.7
rate rel aging[fit criteria] = 0
SAVEPER = TIME STEP
share f2 = 0.15
share fH = 0.028
share fL = 0.23
share f{M = 0.07
18
(086)
(087)
(088)
(089)
(090)
(091)
(092)
(093)
(094)
(095)
(096)
(097)
(098)
(099)
(100)
(101)
(102)
(103)
(104)
(105)
(106)
(107)
(108)
(109)
(110)
(111)
(112)
(113)
(114)
(115)
(116)
(117)
(118)
(119)
(120)
(121)
(122)
(123)
(124)
(125)
share h = 0.47
share inst = 0.05
share inst 1 = 0.8
share to TC = 0.5
share waiting Inst entry 3 = IF THEN ELSE ( ( waiting Inst {M + waiting Inst fH) > 0,
waiting Inst fM / ( waiting Inst {M + waiting Inst fH ) , 0)
share waiting Inst entry 4 = | - share waiting Inst entry 3
SS share of community care costs = 0.5
SS share of institutional care costs = 0.5
stable death inst 2 = stable deaths inst - stable deaths inst 1
stable deaths fH = stable fH * frac death rate fH
stable deaths fL = stable fL * frac death r fL
stable deaths fM = stable f{M * frac death r {M
stable deaths h = total stable deaths - stable non healthy deaths
stable deaths inst = stable inst / average survival inst c
stable deaths inst 1 = stable inst 1 / average length of stay entry 3
stable fH = share fH * "pop 65+"
stable fL = "pop 65+" * share fL
stable f{M = "pop 65+" * ( share f{M + share f2 )
stable from fH to fM = frac imp fH * stable fH
stable from fH to inst = stable deaths inst - stable from fM to inst
stable from fL to fM = stable from healthy to fL - stable from fL to healthy - stable
deaths fL + stable from fM to fL
stable from fL to healthy = frac imp fL * stable fL
stable from fM to fH = stable deaths fH + stable from fH to inst + stable from fH to
f{M
stable from fM to fL = frac imp fM * stable {M
stable from fM to inst = frac from fM to inst c * stable {M
stable from healthy to fL = total stable deaths - stable deaths h + stable from fL to
healthy
stable h = "pop 65+" * share h
stable inst = share inst * "pop 65+"
stable inst 1 = stable inst * share inst 1
stable inst 2 = stable inst - stable inst 1
stable non healthy deaths = stable deaths fL + stable deaths fM + stable deaths fH +
stable deaths inst
sum of clients = community + inst care
sum of elderly = healthy + sum of clients
sum of shares = share fL + share f2 + share f{M + share fH + share h + share inst
TC =TC fH+TC fL+TC fM
TC CC costs per time period = ( TC fL costs per client and month * TC fL + TC f{M
costs per client and month * TC f{M + TC fH costs per client and month * TC fH ) *
SS share of community care costs
TC costs as share of standard HC costs = 0.9
TC fH = INTEG( from TC fM to TC fH - death rate TC fH - TC fH to waiting Inst -
from TC fH to fM , 0)
TC fH costs per client and month = HC fH costs per client and month * TC costs as
share of standard HC costs
TC fH to waiting Inst = TC fH * frac from fH to inst * effect of TC on frac rate to inst
care entry fH
19
(126)
(127)
(128)
(129)
(130)
(131)
(132)
(133)
(134)
(135)
(136)
(137)
(138)
(139)
(140)
(141)
TC fL = INTEG( from healthy to TC fL - death rate TC fL - from TC fL to TC fM -
from TC fL to h + from TC fM to fL , 0)
TC fL costs per client and month = HC fL costs per client and month * TC costs as
share of standard HC costs
TC fM = INTEG( from TC fL to TC fM - from TC fM to TC fH - death rate TC fM -
to waiting Inst from TC fM - from TC fM to fL + from TC fH to fM, 0)
TC fM costs per client and month = HC fM costs per client and month * TC costs as
share of standard HC costs
telecare = TC fL + TC f{M + TC fH
TIME STEP = 0.0625
to waiting Inst from HC fH = frac from fH to inst * HC fH
to waiting Inst from TC fM = effect of TC on frac rate to inst care entry fM * frac
from fM to inst c * TC {M
total costs = Community care costs + Community Care with TC costs + Inst costs +
waiting in comm costs
total stable deaths = “pop 65+" * "65+ frac death rate"
used beds Inst = Inst fH + Inst f{M
waiting at home costs per client and month = 1.79
waiting in comm costs = INTEG( cost waiting in comm per time period , 0)
waiting in community = waiting Inst fH + waiting Inst {M
waiting Inst fH = INTEG( to waiting Inst from HC fH + TC fH to waiting Inst - from
waiting to Inst entry 4 - death r w Inst fH , 0)
waiting Inst f{M = INTEG( from HC to waiting Inst entry 3 + to waiting Inst from TC
fM - from waiting to Inst entry 3 - HC f2 to fL , 0)
20
