Conceptualization and formulation of a UK health and social
care system using System Dynamics
S.Sardiwal!
Information Management and Modeling research Group (IMMaGe)
London South Bank University, 103 Borough Road, SE1 0AA
0207 815 8229
sardiws@ Isbu.ac.uk
Paper for presentation at the 25" annual conference of the SD Society
50" Anniversary Celebration, Boston, July 29"-August 2™ 2007
Abstract
The UK health and social care systems are continuously changing over time. Other
authors have previously put a strong case for usage of system dynamics (SD) in this area
largely because SD address issues of system complexity and identification of feedback
loops, resulting in a greater insight into this problem situation.
This paper presents research carried out in two areas of SD, firstly the conceptualizing of
a problem and secondly the building of a SD model related to the dynamic problem of
“bed blocking’’ in the UK health and social care domain. A case study approach has
been applied to a hospital discharge department and elderly wards in a main UK
hospital.
This paper provides a useful insight into issues that have occurred when conceptualizing
and formulating a health and social care SD model. System behavior has been discussed
as has the use of causal loop diagrams and stock and flow diagrams. Causal loop
diagrams and stocks and flows have shown to play a useful part in overcoming SD
difficulties. SD has proved to be a useful method in helping to gain an insight into the
dynamics of a health and social care system. This is a preliminary paper, future papers
will expand on this to look at policy experiments and sensitivity tests.
Keywords
SD, model conceptualization and formulation
' Views and comments are welcomed and should be sent to Sangeeta Sardiwal via the
email address sardiws@Isbu.ac.uk
1, Research context and structure
This paper is about using SD to conceptualize and formulate a UK health and social care
system. Essentially SD promotes the usage of a whole system perspective and is highly
advantageous in helping to understand the past as well as the future.
Holmer and Hirsch (2006) defines SD as ‘involving the development of computer
simulation modeling that portrays processes of accumulation and feedback and that may
be tested systematically to find effective policies for overcoming policy resistance’ .
This paper provides insights into the dynamic problem of ‘‘bed blocking’. A dynamic
problem of bed blocking exists when patients occupy beds unnecessarily. Bed blocking
can take place among different population groups whether its children, young people or
the elderly. Bed blocking is a serious problem; it results in operations being cancelled as
new patients are not able to be admitted and millions of pounds being wasted (BBC,
2002a).
However in practice bed blocking is predominately a bigger problem among the elderly
population. The National Health Service (NHS) and Social Services are responsible for
making hospital discharge arrangements for elderly people. These hospital discharge
arrangements typically involve making sure that the right care plan is in place,
monitoring the needs of patients and identifying the help that patients will need when
they leave hospital.
There is a clear process that exists and must be followed between Social Services and the
NHS. For a hospital discharge arrangement to exist firstly the NHS must inform Social
Services that the patient is ready for discharge in order for Social Services to go ahead
with carrying out a patient assessment. This is outlined by the Community Care Act 2003
(Department of Health, 2003), which encourages joint working between the NHS and
Social Services, referred to as a Section 2 notification. There is later notification by the
NHS and Social Services once the patient’s discharge date is agreed in the form of a
section 5 notification.
In reality elderly people are waiting in NHS hospitals for assessments to be carried out by
social services in order for them to be discharged from hospital. Bed blocking often
occurs because of firstly the lack of care homes available and secondly an information
delay existing between social services and the NHS, causing patients to experience
delayed discharge. Bed blocking is an important problem to investigate, as bed blocking
is a problem that has not gone away. The government introduced a fining policy to social
services, stating social services departments are to be fined for patients that wait in the
hospital unnecessarily (BBC, 2000b).
While the research area is focused on the health and social care domain this paper is a
contribution to how real world problems and dynamic systems can be conceptualized and
formulated in the public policy arena.
This paper starts by outlining the background to the UK health and social care system
relevant to the problem situation. Following from this the research genesis, objectives and
methodology have been outlined. Given the dynamic problem that exists (Department of
Health, 2002) the SD approach is outlined and justification is given to using SD as an
appropriate method for addressing the problem. Work is then presented on the application
of SD modeling carried out by well-known educators, consultants and practitioners and
on the conceptualization and formulation of the health and social care model. A
discussion follows regarding issues that have arisen during the formulation and
conceptualization the system. Finally conclusions, recommendations and policy
implications of the research are outlined.
2. Background to the UK Health and Social Care System
The focus of healthcare is changing. There is greater emphasis on early intervention of
healthcare in which people are treated and helped long before patients are admitted into
hospital, to help decrease hospital intakes and prevent likely illnesses from occurring. There is
an emphasis on reducing the inequality that exists between different people (or groups of
people) with more social inclusion and strategic needs assessments (Integrated Care Network,
2007). It is recognized that a holistic approach is needed in which there is better
commissioning of health and social care services, citizen involvement and greater
opportunities for partnerships to exist.
The health and social care process is recognised as a dynamic process for those that need and
want public health and social care. There is a clear link between public health and social care
and integration, in which certain policies must be shared. There is a need for multiple services,
in which it is often difficult for health and social care services to work together. The
integration of services is thus seen as a mainstream focus, in which a whole way of working
will be achieved through the joint needs strategic assessment. In achieving integration there is
a need for joint appointments between the Department of Health and local authorities. For
example Tower Hamlets, a London borough an administrative unit of the govemment and the
Department of Health responsible for the overall health of the community has worked towards
some joint human resource policies (Integrated Care Network, 2007).
History has shown that previously (pre 1997) people in the UK were very much responsible
for there own health, however post 1997 the govemment have played more of an enabling
role, in which a health and social care environment exists for people to make their own
informed decisions. Post 1997 there was the development of service agreements in the NHS,
in 2004 there was the Choosing Health policy that was to address rising inequalities by
outlining key principles for supporting the public to make healthier and more informed health
choices. This led to the commissioning of services in the NHS as an attempt to sort out the
acute sector.
Successes in public health have shown that in 2007 there are the lowest ever infant mortality
rates. There are still many public health challenges. Health inequalities are increasing in some
cases, with increased obesity and teenage pregnancy. However it is evident that the services
that are needed by the public and the services that are demanded need to be distinguished
(Integrated Care Network, 2007).
The diagram below highlights the different types of health and social care services that are
provided. Services fall under the categories of services that are provided to the public, those
that are needed and those services that are demanded. The diagram highlights areas where
unnecessaty services are being used, where services are undervalued and where the frustrated
public exist as the result of the public not being able to obtain the services that they need. It is
important to think of these different elements of services to see if resources are being used in
the best way.
Unnecessary services being used
Services that
are provided
Services that
are demanded
Undervalued
services
Services that
are needed
Unmet public needs
Frustrated public
Diagram 1: Integrated Care Network (2007), Public Health and Integrated Services
diagram
In the UK there have been continuous changes in government policy to push towards the
integration of health and social care systems. At present these systems act as separate
entities and there is little evidence of full integration between them, rather there are
instances of joint working, co-location of services and some integration of specific
services.
Integration has been defined as ‘where organisations or services integrate into single
entities allowing for greater transparency between partners and enhanced benefits for
users’ (Oxfordshire County Council, 2004).
Integration is seen necessary as a means of improving services. It should provide patients
with more of an opportunity for patients to receive improvements in care and gain better
access to services.
3. Research genesis, objectives and design/methodology
The main aim of this paper is to present the work and thinking that has been involved in
the conceptualization of the well known health and social care problem, ‘bed blocking’,
and the issues that have arisen when building the SD model.
The research objectives are:
Objective A: To document knowledge and understanding of the health and social care
system and to capture mental models. This fits in with the purpose of the SD model (see
section 5).
Objective B: To conceptualize the problem of bed blocking.
Objective C: To formulate and build a SD model that represents the problem of bed
blocking.
Objective D: To highlight important aspects that needs to be considered during the
conceptualization of the problem and formulation of the SD model.
The research methodology has been to use a case study approach of a large NHS hospital
and has involved interviewing a range of stakeholders including hospital discharge
managers, an IT training staff member and a social worker. In-depth interviews were
conducted over a 2-month period from June-July 2006, involving a sample size of 4 staff.
The problem situation of bed blocking was brought to light in an interview with a
hospital discharge manager (Local Authority Social Services Hospital discharge manager,
2006b). The interviews have proved useful in helping to obtain the necessary qualitative
data to help conceptualize and formulate the SD model. This includes gaining
information on the structure of the health and social care system, where the delays exist
and what the patient process is.
Quantitative data was taken from published reports, largely involving patient data such as
the number of elderly people situated in that particular borough and policy data. Policy
data has included patient waiting times such as the time patients wait for a Social
Services assessment to be carried out. The time patients wait must be 24 hours from when
the notification is sent from the NHS to Social Services to inform Social Services that the
patient is ready for discharge.
4. Appropriateness of SD and explanation of the SD approach
In real world systems, particularly social systems like health and social care, we have
seen that interventions in complex systems have resulted in short term success and long-
term failure. With complex systems there are low-leverage policies, where by the
apparently influential polices have little effect (Sterman, 2000).
The health and social care system is not in equilibrium, it is continuously changing and
intemal system feedback structures often experience extemal policy intervention. This is
5
due to dynamic complexity existing where by there is policy resistance from system
stakeholders that are overwhelmed by complexity and fail to understand the system.
These systems are govemed by feedback, where by actions feedback on themselves and
the behavior of complex systems is counterintuitive and cause and effect are distant in
time and place.
There is a clear need for a holistic, systems approach to problem solving, which is often
the exact opposite to the way most organizations solve problems. This is where SD is
useful as a method of enhancing leaming in complex systems (Sterman, 2000). This view
was supported by the Department of Health at a recent UK conference on integrated
health and wellbeing (Integrated Care Network, 2007). Public health recognizes the need
for a dynamic, interactive tool to look at different facets around heath needs. Govemment
finds it difficult conceptualising the interconnectness between social systems, which has
resulted in numerous policy problems occurring. Testing policies in the real world is
expensive (Coyle, 1996).
Using SD to model the complex interface between health and social care and to carry out
‘what if’ and ‘why’ analysis and for scenario planning would be particularly
advantageous. Given this SD would be highly beneficial in helping to address the
dynamic complexity that characterizes many public health issues (Homer and Hirsch,
2006).
Simulation is seen as necessary in situations like this where relying on feedback through
the real world could be too slow and ineffective, illustrated by the time delays that exist
that impact patient services. SD helps to deal with these issues of dynamic complexity
and policy resistance that the health and social care domain are experiencing. Thus SD is
used as a suitable approach to modeling a dynamic health and social care system.
The SD approach consists of two main steps:
a) Conceptualization of the problem
b) Formulating and testing the model
These two main steps consist of the following smaller steps.
a) Conceptualization of the problem
1. Define the problem
2. Identify the Stocks and Flows
3. Sketch the dynamic hypothesis
b) Formulate and test the model
4. Formulate the model and simulate
5. Conduct sensitivity tests
6 Conduct policy experiments
5. The application of SD modeling carried out by well known educators,
consultants and practitioners
System dynamics has occasionally influenced both govemment policy at national level
and health and social care organisational policy at local level. This approach enables
policy makers to understand why policies fail and what can be done differently to yield
better policy making. System dynamics has been influential in the health and social care
context in developing an organization’s way of thinking and the way that health and
social care organisations conceptualise the whole of the health and social care system.
SD has been widely used to model health and social care problems by a range of people
from educators, consultants to practitioners. Examples of some current applications of
system dynamics and the impacts on health and social care organisations are highlighted
below.
Case 1: Systems dynamics influencing the G overnment’s reimbursement policy
System dynamics was used by the NHS and Local Govemment Association to test out the
reimbursement policy, which is part of the delayed hospital discharge bill. This was in
order to shed light on the complexities involved with fining Social Services over ‘bed
blocking’ patients in hospitals in the areas of acute and post acute care.
The govemment were able to see from the systems dynamic model that an increase in
hospital capacity during times of high demand in the acute sector would result in an
increased number of people in hospital as empty hospital capacity will only be filled.
This would lead to the unintended consequence of increasing hospital discharge delays.
In contrast the policy of increasing post acute capacity would be advantageous to both the
post acute and acute sectors. This policy decision would result in lower reimbursement
fines being received by Social Services, as there is greater provision available to place
discharged patients into care homes. Hospitals would experience a reduction in waiting
times for patient admission as patient tumover is increased as more people are discharged
from hospital on time (The NHS Confederation, 2005).
The effect of the system dynamics model on the govemment was to delay the
implementation of the reimbursement policy and to increase funding to Social Services to
increase their care capacity before this policy was enforced. The system dynamics model
highlighted to the goverment that the most suitable policies and solutions do not always
translate into the desired outcomes expected and maximum benefit to the health and
social care system. The goverment was able to develop their thinking skills (their
‘dynamic thinking’) by conceptualising the behavior of health and social care
organisations over time.
Case 2: System dynamics influencing a hospital’s Accident and Emergency (A& E)
waiting time policy
System dynamics was applied to an A & E department to ascertain why delays in A & E
waiting time admissions were occurring. The behavior over time of an A & E department
was investigated. The results of the system dynamics model highlighted that there was a
7
knock-on effect of there being reduced numbers of A & E doctors during the early
mornings and evenings. This led to a 2-hour delay of patients being seen throughout the
day. The results highlighted that the Patient’s Charter on waiting times was unrealistic
(Lane et al, 1998). If a major incident took place there would be major disruption to the
utilization of staff and bed stock at hospital. The effect of the system dynamics model
was to influence the A & E staffing policy by addressing the need to improve the number
of specialist A & E doctors.
Case 3: System dynamics influencing the use of community matrons in the NHS
Existing Leicestershire NHS strategy has been to reduce the numbers of GPs and increase
the number of community nurses. Leicestershire health community used system
dynamics to investigate how best to use community matrons as a resource and how many
community nurses Leicester NHS should have. These decisions are important as
developing community matron competences are time consuming and expensive (Lacey,
2006). Effective usage of community matrons will ensure a higher standard of patient
care as community matrons importantly have the role of self-managing people’s
conditions and giving people altemative health choices.
The system dynamics model highlighted that there is a lack of slack in Leicestershire’ s
local health and social care services to cope with unscheduled care events. The model
highlighted that Leicestershire NHS should have 30 more community matrons then they
currently have. Leicestershire NHS’s current decision to have 34 matrons was the result
of the NHS not considering the effect of the change in demographics of the local
population on the health and social care system (Lacey, 2006).
The effects of system dynamics influencing Leicestershire NHS’s policy decisions are yet
to be seen, partly due to the reason that existing decisions on community nurses have
already been made. Operational managers may fear that the system dynamics model
reflects their decision making in a bad light and thus do not want to be seen to enforce the
recommendations from the system dynamics model. Nevertheless system dynamics has
had a high impact in getting Leicestershire NHS to look at the wider health and social
care system and to extend their own limited mental models.
System dynamics was useful in getting Leicestershire NHS to understand the complexity
associated with community nurses and to help reduce their fear as Leicestershire NHS
were able to see changes in the system dynamics model that will move Leicestershire
NHS forward.
Case 4: System dynamics influencing the rollout of a Chlamydia screening
programme
System dynamics has been used in disease modeling to model Chlamydia, a sexually
transmitted disease. System dynamics has been useful in modeling the flow of the
population that are infected and are in recovery from Chlamydia and thus in modeling the
dynamics of disease transmission.
The results of the model have aided decision making in deciding who should be screened
and how often people should be screened. System dynamics was useful in demonstrating
the feedback between the infected and non-infected population and helped inform policy
making. From the results of the model it was decided that Chlamydia screening should be
targeted at females aged 16 to 20 (Brailsford, 2002). The system dynamics model was
influential in identifying ways that screening could be improved at health centers which
would lead to cost savings and delivery of a more effective screening programme
(Toohill, 2002).
Case 5: System dynamics influencing a Mental Health Trust (MHT)
System dynamics was used by a Mental Health Trust to aid the trust in achieving the
government’s agenda for modemisation and the national service framework for mental
health.
System dynamics was advantageous in helping the Mental Health Trust to understand the
different patient flows, care pathways and policy drivers in different areas of mental
health. The system dynamics model helped to introduce new ways of thinking and has
implications for investment in mental health services (Smith et al, 2005). The results are
significant as investment in mental services in the northwest has been below the national
average (Guardian, 2004).
Repper, a system dynamics consultant, highlighted the advantages of system dynamics in
this case by saying ‘This modelling will show up what you can do by making changes,
and reinvesting in different services, but it will also show the limits of that and what you
can not do’ (Guardian, 2004).
Case 6: Using system dynamics to investigate the effects of telecare
Telecare is a community alarm service that enables elderly people in their homes to
contact an emergency response service via the use of sensors to receive emergency care
services when needed. Usage of telecare has been promoted as part of the govemment’s
Care Services Improvement Partnership policy (CSIP).
System dynamics was used to investigate the effect of elderly people using telecare in the
provision of care services. System dynamics was useful in modeling issues such as the
numbers of clients that are receiving telecare services and the overall costs involved over
time. The results of the system dynamics model illustrated that the effect of telecare
would reduce the number of people in institutional care (Bayer et al, 2005). System
dynamics was useful in providing a systematic view fora health trust and Social Services
department, looking at the effect of telecare in the health and social care sector.
In other business areas Coyle has made significant contributions to the application of SD.
Case 7: Using system dynamics to investigate the Domestic Manufacturing
Company’s manufacturing problem
Coyle addressed a manufacturing problem for a company producing washing machines,
in which the company was unable to forecast the inflow of new machine orders. As a
result it was difficult for the company in particular the raw materials department to cope
with unpredictable order pattems, over and under ordering was experienced (Coyle,
1996). A SD model was constricted with the aim of analyzing the robustness of the
overall policies for nnning the business.
The policy of introducing an information system to control the quantity of raw material
was tested, as the raw material manager saw the information system as a good idea as
information would be provided on how much raw materials would be needed to fill
orders. The model illustrated that there were more sustained oscillations in the backlog of
ordered machines and the desired backlog of machines, which were closely aligned.
However the simulation illustrated that there was not great control of the raw material
stock and of the desired raw materials as the policy for the number of weeks of average
production was kept the same. The company produced at the same rate regardless of the
fluctuations in raw materials. Coyle highlighted that implementing new policies may not
be enough to be effective in any system, as they change the whole system, and other
existing policies might need to be changed (Coyle, 1996).
6. Conceptualization of a UK health and social care system
Conceptualization has centered on defining the purpose of the model, the problem
situation, identifying key variables, drawing the reference mode and identifying the basic
mechanisms, feedback loops. These aspects are drawn upon below.
a) Purpose of the model
The purpose of the SD model to date has been to document knowledge and understanding
of the health and social care system and to capture mental models.
b) Problem Situation
The problem description is that there is an information delay between the NHS and Social
Services. The NHS thus does not inform Social Services in a timely manner that patients
are ready to be assessed.
A dynamic problem exists in that this information delay results in ‘bed blocking’. Bed
blocking occurs when patients are occupying beds unnecessarily as patients must be
assessed by social services before being discharged. Thus the number of patients that are
recovered and awaiting assessments is greater than the number of assessments being
carried out. The problem of bed blocking is increasingly complex and dynamic and is a
long-term policy problem. The problem is dynamic as the problems are characterized by
variables that undergo significant changes over time.
10
From existing hospital discharge data the reference mode illustrates large oscillations in
the patients recovered and awaiting treatment by Social Services staff (see diagram 2).
WW
q 2 3 DAYS
36-e-MoUZseO
Diagram 2: Local Authority Social Services Hospital discharge manager (2006a),
Reference mode illustrating the oscillations in the patients recovered and awaiting
treatment
There is great motivation to address this problem. The problem is important to solve
because ‘bed blocking’ creates resource management problems. If a suitable policy
solution is not found to control the problem the result will be too many elderly people
waiting for an assessment to be carried out by social workers. The flow of patients in and
out of the health and social care system would be reduced. There would be continuous
bed blocking by elderly people and vital financial resources will be wasted if Social
Services are fined over delayed discharge.
c) Dynamic hypothesis illustrating the current model
The hypothesis is that an increase in the number of elderly people awaiting assessments
will lead to more patient delays, as the adjustment time that social workers respond to is
slower. This is due to the delay in Social Workers receiving the information from the
NHS. This will lead to oscillatory behavior, as there are larger delays.
The dynamic hypothesis is illustrated in a stock and flow diagram below (see diagram 3),
in which the black and white line illustrates the feedback loop between the NHS and
Social Services.
11
Social Services
social workers
adjustment time to close
the gap of social workers
average number of
social workers retiring
of social workers
desired numberfo
adjus fhent for
awaiting aksessments
percentage of elderly
peopje admitted
wp people
time to close patients
waiting for assessment
iS to retire
percentage target Wf people
waiting for assessment
gap in patients wan
for assessment
Nuctivity of|social workers
Health Sector
rate of patients ill
rate of patients
receiving treatment
@
© Sh. patients waiting for treatment Bape mS Tecovered
rate of patients
receiving assessment
Diagram 3: Author researcher (2007) A Systems Dynamics Model showing the
feedback loop between the NHS and Social Services
12
This is an exploratory model. For the purposes of this illustration I have made the
assumption that the number of social workers contribute to the patient delays. This is
illustrated by this simplified stock and flow diagram. This has shown the main
relationship between the availability of social workers and patients receiving
assessments, a core determinate of whether bed blocking occurs. Whilst this is an
important factor, other factors like information system flows, competing policies over
social worker time and political problems between health and social care are contributory
factors towards patient's delays.
d) Causal Loop diagram
The following things were considered when devising the causal loop diagram.
1, Exogenous variables. The exogenous variables are those that are not affected by the
system and are not included in the causal loop diagram. Only the desired goal that is
exogenous is included, such as the desired patients waiting for assessments goal. The
exogenous factors that are not included are:
1. The adjustment time to close the gap of social workers
2. The time to close patients waiting for assessments.
3. Patients ill
2. Feedback loops. Feedback loops illustrate circular causality over time.
The feedback loops that exist in the health and social care system are the following, in
which the causal loop diagram corresponds to the problem statement:
1. Reinforcing loop (R1). The more hiring of social workers the higher the retiring
of social workers that leads to an increase in the average number of social workers
retiring, which in tum leads to more hiring of social workers.
2. Counteractive loop (C1). A rise in the number of social workers will lead to
more patients receiving assessments, which will decrease the number of
recovering patients awaiting assessments. The gap in patients waiting for
assessments will decrease leading to smaller adjustments needed for awaiting
assessments, which will lead to a smaller gap in the number of social workers
needed to carry out patient assessments.
3. Counteractive loop (C2). The more patients that are recovered and are awaiting
for an assessment the higher the number of patients that receive assessments. The
higher the patients receiving assessments the lower the numbers of patients that
are recovered and are awaiting an assessment.
13
social workers
3
ti,
hiring and firing of PP retiring of social
social workers workers
XX RI y
average mmber of
social workers retiring
Gap in number of
productivity of
social workers social workers.
\
desired number of
social workers desired patients waiting
+ for assessments
adjustment for
awaiting assessment .
gap in patients waiting
for assessments
,h
a patients receiving
ea + Patients recovered and C2
patients waitingfor © + Patients recieving awaiting for an Z assessments
treatment treatment sm
Diagram 4: Author researcher (2007), Causal Loop diagram of a health and social
care
6. Aspects considered during the formulation and conceptualization of the health
and social care model
The following aspects where considered during conceptualization.
a) Establishing a clear distinction between the causes of the problem from the actual
problem. This can be a particular problematic area experienced by modelers as these are
separated in time and space. For example it was important to distinguish the cause in this
case study as being an information delay between the NHS and Social Services from the
problem of bed blocking.
b) Illustrating the internal feedback loop between the health and social care system.
This was crucial as this feedback loop acts as a long-term driver of the system, in which
the loop provides the leverage for influencing the behavior of the system.
c) Formulating the dynamic hypothesis. The dynamic hypothesis was based on a
literature review and information gathered from hospital discharge staff. Formulation has
involved translating the feedback loops into equations, making assumptions and
14
estimating different parameter values. The following aspects were considered when
formulating the health and social care SD model.
a) Representing the necessary patients flow that we are considered with in regards
to the problem situation from the patient being ill to receiving an assessment. This
information was confirmed at an interview with a social care discharge manager. The
patients get admitted to hospital, they wait for treatment then receive treatment. Patients
then recover and wait for an assessment from social services, then receive an assessment
and then leave hospital.
b) Illustrating the main stocks. The main stocks considered are social workers, patients
waiting for treatment and the patients recovered. These stocks are seen to play a central
role in health and social care management problems and are the source of endogenous
dynamics.
c) Considering the parameters. When formulating the model parameters of the model
were either estimated, taken from secondary or primary data that was available about the
problem situation and case study.
d) The adjustment time. The adjustment times it takes to close patients waiting for
assessment and to close the gap in obtaining social workers to do assessments are
considered.
e) Illustrating the productivity of workers. The productivity of workers depends upon
the desired number of Social Workers and patients receiving assessments. The equations
in the stock and flow diagram correspond to the causal loop diagrams.
f) Illustrating the desired number of social workers. The desired number of workers
depends on the productivity of Social Workers and the patients receiving treatment.
g) Illustrating the delays. Delays such as the information delay in getting the average
number of social workers that are retiring have been important to consider as delays
intervene between causes and their effects.
h) Determining the gap in the number of social workers. The gap in the number of
Social Workers is equal to the desired number of Social Workers minus the number of
Social Workers currently employed.
i) Considering unit consistency. Unit consistency was maintained through out the
building of the model, for example in calculating the desired number of workers needed
to carry out assessments.
15
Desired number of = patients receiving treatment - adjustment for awaiting
Social workers assessments
productivity
Units as illustrated below for the desired number of social workers:
<<Social worker>> = patient/month - patient/month
patient/month/social worker
For the top half of the equation patient month- patient month= patient month. Then
patient/month cancels out with patient/month leaving the desired number of social
workers measured in ‘social worker’ unit.
<<social worker>> = patient/month_ —
Dele ten eel worker
<<social worker>> = <<social worker>>
Unit consistency can thus advantageously be used to help work out what variables are
needed in the model, as equations must be dimensionally consistent. All equations in the
health and social care model had real life meaning.
8. Research findings and discussion
The research findings are indicated below, which corresponds to the earlier indicated
paper objectives. This achieves the overall aim of presenting work and thinking that has
been involved in the conceptualization of the bed blocking problem and issues that have
arisen when building a SD model (see section 3).
Objective A: Documenting knowledge and understand of the health and social care
system and to capture mental models
A case study approach was useful in conjunction with SD as primary data could be
collected on the processes, delays and problem situation of the hospital discharge team.
Finding 1:
SD was helpful in helping to elicit and map my mental models of the health and
social care system.
Objective B: C onceptualization of the problem of bed blocking
Conceptualization of the bed-blocking problem was achieved through defining the
problem, identifying stocks; flows, sketching the dynamic hypothesis and drawing causal
loop diagrams to establish the main feedback loops that exist (see section 6).
Objective C: Formulation and simulation of a SD model that represents the problem
of bed blocking
16
Formulation of the SD model involved constructing the dynamic hypothesis, which
involved including the main stocks and flows and feedback loops in order to capture the
structure of the health and social care system (see section 6).
Finding 2:
When the model was simulated oscillatory behavior was experienced which was
similar to the reference model, in which there are oscillations in the patients
recovered and awaiting treatment.
Objective D: Aspects considered during the conceptualization of the problem and
formulation of the SD model
Finding 3:
Adequate time needs to be spent in conceptualizing the problem before formulating
and testing the model. Modeling has shown to be very iterative in which it has been
necessary to revisit and define the hypothesis and problem differently.
Finding 4:
Causal loop diagrams and stocks and flows have been useful when conceptualizing a
problem.
Discussions of the findings are provided below:
Objective A, Finding 1
SD has helped to expand my mental models and has brought to my attention other issues
of feedback, delays, nonlinearities that I would previously not have considered fully if I
was using another systems thinking approach such as soft systems methodology.
Objective C, Finding 2
This system meets the minimum number of requirements for oscillatory behavior to
happen, in which there are more than two stocks and a balancing loop exists.
Thus oscillations occur in this health and social care system due to:
1. A major feedback loop existing between the health and social care systems.
2. Numerous delay times being present, which are the adjustment times. The delay times
are an indispensable factor.
If any of these parts of the system become constant then we would not experience these
oscillations. If the delays are not long enough then this will translate into smaller
overshoots and undershoots.
17
Objective D, Finding 3
It is important that adequate time is spent conceptualizing and formulating a SD model.
Otherwise this would lead to inadequate, wrong models being made that provide no
sound basis for managers to base their decisions on.
The difficulties and practicalities of conceptualizing a dynamic problem and formulating
aSD model cannot be underestimated. This may be a part of why SD has not taken off to
the extent other approaches have, because the complexity in conceptualizing problems
and formulating models are underestimated.
Objective D, Finding 4
With stocks and flows the feedback structures become more transparent in seeing how
the cause and effect feedback on each other (Haraldsson, 2006). Stocks and flow
diagrams have greater tendency to be more detailed than causal loop diagrams and have
allowed me to think more about system structure, in which the components are more
strictly defined than in causal loop diagrams.
Causal loop diagrams have aided brainstorming and model creation. Causal loop
diagrams are useful in helping to illustrate ideas from a model that have previously been
created. However causal loop diagrams have numerous disadvantages in that you cannot
determine the behavior of the system from the polarity of the feedback loops, as
dynamics behavior is not created. Thus causal loop diagrams cannot be used to predict
dynamic behavior. It has been harder for me to understand the causal loop diagrams
compared to the stock and flow diagrams as it is a less specific representation to the
health and social care system being modeled and the same level of detail as the equation
in the stock and flow diagram have not been shown. Simulation is crucial for gaining a
complete understanding and in determining the dynamics of a system.
However through SD emergence causal loop diagrams have helped to make SD more
accessible to a wider audience as they have brought about simplicity and can be used as a
tool for communication purposes (Richardson, 1986, Coyle, 1996). Thus there are
advantages and disadvantages of usage of causal loop diagrams and stock and flow
diagrams.
9. Conclusions, recommendations and policy contributions
From this modeling experience SD has provided the greatest leaming in terms of
understanding the problem situation better and helping me to organize my mental models.
The process has helped provide a clearer insight into how the structure of a system
translates into behavior. SD has proved to be a participative activity, where by one leams
by trial and error. This can be very powerful in changing mental models.
Whilst SD has been useful, the next step is to apply other systems thinking approaches
like Soft Systems Methodology (SSM) in conjunction with this research. SSM provides
the advantage over the SD approach in giving greater consideration to the ‘holons’. These
are human activity systems and are highly appropriate to consider, as these are
18
appropriate to the problem situation in the health and social care domain. It is very
difficult to incorporate soft aspects of the problem using the SD approach, where as SSM
goes further in understanding the different people involved, their conflicting objectives,
perceptions and attitudes. The focus of SSM is different in contrast to SD, where by
SSM’s main focus is is to look at the people involved with the problem and the secondary
focus would be to look at the problem. Where as the SD focus is very much vice versa
where by the problem is the primary focus and little focus is given to the people involved
with the problem.
Having experienced the systems dynamics process the following recommendations are
made in devising a SD model.
1. Timing. Adequate time needs to be spent in conceptualizing a dynamic problem
with the use of causal loop diagrams and stock and flow diagrams.
2. Development of the system dynamic model. To include the following points:
a) There should be continuous feedback between the client and an
experienced SD modeler in devising a SD model, in order to ensure that it
mimics reality sufficiently and produces the reference mode behavior.
b) A model must be continuously tested at each stage when additional
complexity is added in order to understand clearly the relationship
between structure and behavior of the system. This will add confidence in
the model when devising the most appropriate policy to solving the
problem situation.
c) The model should be kept as simple as possible, so to address the problem
situation only and unnecessary additional complexity should not be added
for any reasons.
Future papers will expand on this to look at policy experiments and sensitivity tests for a
particular policy of providing integrated information systems between the NHS and
social services. Little is known about what affects integrated information systems have on
service provision. This will be valuable to social services and health stakeholders
including hospital management staff and the government who will be concemed with
getting patients through the health and social care system as quickly as possible, in order
to save costs, improve patient services and improve policy making nationally and locally.
19
10. References
Bayer, S, Barlow, J, Curry, R (2005) Assessing the impact of a care innovation: telecare.
Available from <http://www3.imperial.ac.uk/pls/portallive/docs/1/43011.PDF>
[Accessed 5th January, 2006]
BBC (2002a) Bed-blocking a massive problem. Available from
http://news.bbc.co.uk/1/hi/health/1935791.stm [A ccessed 20th April, 2007]
BBC (2002b) Bed-blocking: a fine approach? Available from
http://news.bbc.co.uk/1 /hi/health/1937221.stm [A ccessed 20th April, 2007]
Brailsford, S (2002) Health Systems for the NHS. Available from
<http://www.orsoc.org.uk/region/study/sdplus/report1.htm> [Accessed 5th January,
2006]
Coyle, G. R (1996) System Dynamics Modelling A practical Approach, Chapman and
Hall
Department of Health (2002) £200m allocated to councils to further reduce ‘bed-
blocking’. Available from
http://www.dh.gov.uk/en/Publicationsandstatistics/Pressreleases/DH_ 4011641 [Accessed
2" March, 2007]
Department of Health (2003) Community Care Plans (England) Directions 2003.
Available from
http://www.dh.gov.uk/en/Publicationsandstatistics/Publications/PublicationsLegislation/
DH_4083029 [Accessed Sth January, 2006]
Guardian (2004) New model services. Available from
<http://society.quardian.co.uk/managingnewrealities/story/0,14030,1165545,00.html>
[Accessed 5th January, 2006]
Haraldsson, H.V, Belyazid, S,Sverdrup, H (2006). Causal loop diagrams - promoting
deep learning of complex systems in engineering education. Available from
<htip://www.|th.se/fileadmin/Ith/genombrottet/konferens2006/Haraldsson_etal_bidrag.pd
f> [Accessed 9th April, 2007]
Homer, J, Hirsch, G (2006) System Dynamics Modeling for Public Health: Background
and Opportunities. American J ournal of Public Health, Vol. 96, No. 3 pp. 452-458
Integrated Care Network (2007) Integrating Services for Health and Wellbeing the Role
of public health. Available from
http://www. integratedcarenetwork.gov.uk/index.cfm?pid=17 &eventID =59 [Accessed 2"
March, 2007]
20
Lacey, P (2006) The use of qualitative and quantitative modeling in driving change by
showing policy outcomes in UK health policy. Proceeding of the 2006 8" System
dynamics Annual Gathering held at Harrogate, London
Lane, D, Monefeldt, C, Rosenhead, J (1998). Emergency -but no accident. Available
from <http://www.orsoc.org.uk/about/topic/insight/article_orinsight_emer.htm>
[Accessed 7th January, 2005]
Local Authority Social Services Hospital discharge manager (2006a), [Personal
Communication] June.
Local Authority Social Services Hospital discharge manager (2006b), Interview with
author in June 2006. London [recording in possession of author]
Oxfordshire County Council (2004). Integration of services for Older People and People
with Physical Disabilities in Oxfordshire. Available from
<http://www2.oxfordshire.gov.uk/hlpdownloads/EX 010205-06.htm> [Accessed 19"
February, 2005]
Richardson, G (1986). Problems with causal-loop diagrams. Available from
http://sysdyn.clexchange.org/sdep/Roadmaps/RM4/D-3312-2.pdf [Accessed 9th April,
2007]
Sterman, J (2000) Business Dynamics Systems Thinking and Modeling for a Complex
World, Irwin McGraw-Hill
Smith, G, Wolstenholme, E, Repper, D (2005) Initial Experiences of Introducing System
dynamics through a Mental Health project in North West England. Available from
<http://www.systemdynamics.org/conf2005/proceed/papers/SMIT H428.pdf> [Accessed
5th January, 2006]
The NHS Confederation (2005), The potential of system dynamics. Available from
<htip://www.symmetricsd.co.uk/files/le_systemdynamics_nov2005.pdf> [Accessed 5th
January, 2006]
Toohill, L (2002) Chlamydia Screening: Learning from Pilots and Rolling out to the
Nation. Available from
<http://www.orsoc.org.uk/region/study/sdplus/DoH% 20Chlamydia%20LT oohill.pdf>
[Accessed 5th January, 2006]
11. SD terminology
Reference mode- This is the behavior pattern that has been observed historically in the
system that we are studying and is what we want to replicate. It is a graphical
representation of the problem, in which are interested in looking at the dynamic problem
of the interplay between stocks and flows. The reference mode would illustrate anything
that is of interest to the client, for instance in a supply chain management scenario we
21
would draw the stocks such as the inventory, employees and also the flow variables, sales
and production.
Stock and Flow diagram- This is a map representation for the reason of the behavior
that you are seeing.
Dynamic hypothesis for the reference behavior- This is the hypothesis about the
dynamic behavior and is the theoretical explanation. It is represented in terms of the stock
and flow.
Characteristic behavior- We look for feedback loops to explain the characteristic
behavior, which is not driven by exogenous feedback. We are concemed with the systems
dynamics behavior that is endogenous.
22.
