The 33rd International Conference of the System Dynamics Society, Cambridge, Massachusetts, USA, July 19
~July 23,2015
A System Dynamics Model of the Kidney Transplants in the U.S.
A System Dynamics Model of the Kidney
Transplants in the U.S.
Ignacio Paricio'*, Jorge Figal’
' Faculty of Technology, Policy and Management, Delft University of Technology, the Netherlands,
Jaffalaan 5, 2626 BX, Delft, the Netherlands
Abstract
In one of the most developed countries inthe world, the United States of America, a crucial
problem for the population remains unsolved: the huge waiting list for patients waiting for
an organ transplant. On average, 21 people die each day while waiting for a transplant. The
purpose of this paper is to analyze the current kidney transplant situationin the U.S., as
well as to test by means of system dynamics modeling whether policies lead to an
amelioration of the waiting list problem. Two different policies that balance effectiveness
and social acceptance are implemented with the aim to improve the supply side of the
system, i.e. the donation of kidneys. The main conclusions are that the implication of such
policies are effective and necessary, but not sufficient. Further research in living donation
is required to guarantee a sufficient solution to the problem.
Word count: 6123
Keywords: kidney transplant, SD modeling, system dynamics, waiting list, organ donors
* Corresponding author. E-mail address: ignacio.paricio@outlook.com
1 Introduction
In one of the most developed countries in the world, the United States of America, a crucial problem
for the population remains unsolved: the huge waiting list for patients needing an organ transplant.
On average, 21 people die each day while waiting for a transplant (OPTN, 2012).
In particular, kidney transplants are among the most solicited interventions (see Figure 1). Even
though there exist alternatives such as dialysis for patients with end stage renal diseases, kidney
transplantations represent a significant improvement in quality of life and longevity of the patients.
However, the number of candidates fora kidney transplant on the waiting list continues to increase
each year, while organ donation numbers remain flat, as displayed in Figure 2.
The purpose of this paper is to analyze the current kidney transplant situation in the U.S.,as well
as to study policies to ameliorate the waiting list problem through system dynamics modeling.
Transplants in the U.S.
_- Lung
“1%
Kidney
82%
Kidney =Pancreas = Liver "Heart ™ Lung
Figure 1: Distribution of organ transplants in the U.S. in 2011 (OPTN, 2012)
The rest of the paper is organized as follows: section 2 describes in more detail the problem
situation, the usefulness of a system dynamics approach, and the boundaries of the analysis.
Section 3 describes the conceptualization problem, as well as the development of the model.
Different identifiable submodels are described in an attempt to present the modeling process in a
comprehensive way. Both a highly aggregated causal loop diagram and a Sector-Diagram are also
presented. Section 3 ends with a validation of the model. In section 4 four different KPIs are
introduced, and the behavior of the model in the base scenario is presented. Section 5 presents two
proposed policies that have the objective to revert the current tendency in which demand for
kidneys increasingly surpasses supply. Finally, section 6 presents the conclusions of this paper, as
well as suggestions for further research.
100,000:
75,000:
50,000-
25,000
——_— Living
OF So ae T Deaths on the
1990 2000 ‘10 waiting list
Source: The Organ Procurement and Transplantation Network The Wall Street Journal
Figure 2: Evolution of patients on the waiting list and transplants from 1990 to 2013
2 Problem formulation
2.1 PROBLEM SITUATION IN THE U.S.
As indicated in Figure 2, the waiting list for a kidney transplant in the U.S. continues to increase
every year. The real problem however lies in the consequences this rising tendency has for the life
quality of the patients. Although there are currently some alternatives to organ transplantations,
such as dialysis, and the future might bring extraordinary breakthroughs that could solve this
problem, such as stem cells (EuroStemCell, 2013), the truth is that nowadays the only long lasting,
quality solution for patients suffering from end stage renal diseases are kidney transplantations.
The scope of this paper is limited to this issue.
Focused on the kidney transplantations, the size of the waiting list is one of the most controversial
issues, and also the one that media coverage targets. There are however several other implications
that need to be addressed if this problem is to be tackled. Hence, this paper divides the study and
analysis of the kidney transplantations problem into the following issues:
= Increasingly large waiting list for kidney transplants in the U.S.
= High mortality of patients waiting for a long time in the waiting list. On average, 12 people
die each day in the country while waiting for a life-saving kidney transplant.
= Many potential kidneys are lost every year because more than half of the population are
not donors upon death.
= It is not easy to increase society's awareness of the problem.
= Unequal distribution of the problem according to the blood type of the patient.
= There are many ethical and moral issues, especially when it comes to encouraging
donation.
2.2 SYSTEM DYNAMICS APPROACH
System dynamics is the main methodology used in this paper to study the problem of the kidney
transplants in the U.S. because of three reasons. First, studying the global picture of kidney
transplantations is a complex issue of dynamic nature.
Second, structural changes are required in order to solve the problem, which are relatively easy to
test by means of system dynamics.
Finally, organ transplantations is also a field that, because of the big ethical implications and moral
dilemmas implied, is not acceptable for real life testing. Studying this kind of system by means of
system dynamics has the advantage that policies can be tested without actually needing to consider
these kind of human issues: “system dynamics allows to identify desirable system changes and test
them in a virtual laboratory “ (Pruyt, 2013).
There already exist research in the field of chronic diseases (Homer, et al., 2004), addiction
treatment policies (Nielsen & Wakeland, 2013) and organ donation (Hirsch, et al., 2012) using
system dynamics. The approach followed in this paper is however different in that intends to
capture the problem in a purely endogenous way, using the age chain of the U.S. as a starting point
and projecting it over a long timeframe, 35 years. Furthermore, its scope is limited to kidney
transplants.
Anyhow, the authors believe that different formulations of the same problem capture
characteristics from different angles, give room for different policies, and eventually enhance the
support in the decision making-process.
3 System Dynamics Model: Conceptualization, Development
and Validation
Section 2 discussed the motivations to use System Dynamics in the study of the problem field and
the testing of policies. This section presents the approach followed while modeling the problem
situation, from the model conceptualization and to the model validation.
3.1 MODEL CONCEPTUALIZATION
Figure 3 summarizes the conceptualization of the problem. It includes the boundaries of the
system, as well as the different subsystems that will be considered during the model development.
The socio-technical complexity of the problem is represented by a supply and demand subsystem
embedded ina social context characterized by ethical dilemmas and society sensitization.
SYSTEM BOUNDARIES
REGULATIONS
MEDICAL
PATIENT/DONOR BREAKTHROUGHS
CONDITIONS BLOOD. RE [TRANSPLANTS | .
' _ SOCIO-MEDICAL
“ INDICATORS
ae DEMAND~
\ 2
PEDIATRIC
AGE CHAIN WAITING LIST CASES
\ 4
HOSPITALS SUPPLY
NETWORK ss IMMUNOLOGICAL
ADVERSIONS
LOGISTICS a MATCHING
SYSTEM POVARENESS ECHANISMS:.. GEOGRAPHICAL
UMITATIONS
DIALYSIS TREATMENTS
Figure 3: Conceptualization diagram
Incompatibilities among blood types add further complexity to the problem. Due to the fact that
transplants between different blood types are not symmetric, chances of an individual to find a
suitable kidney depend on the blood group she/he belongs to. The matching system of organ
(kidney) transplantations according to the blood type of donors and recipients is shown in Table
5
Table 1: Influence of blood types in the donor/recipient matching process
| TETO CAN REC
DONOR RECEIVER RECEIVER DONOR
A A, AB 0 0
B B, AB A A,0
AB AB B B,O
0 A,B, AB, 0 AB A,B, AB, O
The timeframe considered is 35 years, allowing to explore the problem and to study the implication
of policies on the system without having too much uncertainty regarding future technology
developments and societal changes.
Table 2 contains data included in the modelas starting point, corresponding to the year 2010. This
will allow some preliminary validation using historical data.
Table 2: Initial values and constants in the model.
| Parameter Value Units Source
Initial waiting list for an organ transplant 83900 person (OPTN, 2012)
Initial number of children under 12 5,08E+12 person
Initial number teenagers (under-age) 3,15E+12 | person (CDC, 2015)
Initial number of adults 1,89E+13 person
Initial number of retirees 4,47E+12 person
Base net annual immigration 1E+09 person (OIS, 2012)
Percentage of teenagers that need a kidney 1.91E-05 1/year
transplant
Percentage of adult people that need a kidney 1.51 E-04 1/year (OPTN, 2012)
transplant
Percentage of retirees that need a kidney 5.66 E-05 1/year
transplant
Teenagers death rate 3.93 E-04 1/year
Adult death rate 0.0033 1/year | (Sherry, et al., 2013)
Retirees death rate 0.04 1/year
Percentage of adult deaths in hospitals 0.27 Dmnl (NVSS, 2013)
Percentage of families that approve a donation 0.365 Dmnl (Kolata, 1995)
Fraction of adults donating upon death 0.4 Dmnl (Makely, 2013)
alive adults donation rate 3.2 E-05 1/year (OPTN, 2012)
Kidneys transplanted per organ recovered 0.85 Dmnl
Average time patients live before dying 8 year (Graham, 2013)
Fraction of patients that accept the organ 0.8 Dmnl (OPTN, 2012)
Fatality ratio of rejections 0,1 Dmnl
. (Hampton, et al.,
Average sphere of influence 25 person 2016)
3.2 MODEL DEVELOPMENT
The highly aggregated causal diagram of Figure 4 presents the essence of the model developed in
this paper.
Average time people
live while waiting for a
transplant
Teens! Adults! _—__» WAITING LIST FOR A
Retirees Necessity for * KIDNEY TRANSPLANT a
7 kidney transplant > f Deaths from people
+ —_ waiting too long
increase in awareness through deaths
i m mM
Children —! {-—» —t—r Retirees Cases ofrecovery SSM
+ z ¥ People aware of
birth loop the problem
fraction of patients that
accept the organ
TRANSPLANTATIONS
n + + increase in awareness
eat a through sucessful cases ‘
~
+ ? Sensitization
AVAILABLE + with the problem
KIDNEYS
Alive adults +
donation rate sod
)
increase in awareness
affecting decision of families
‘Adults willing to donate
their organs upon death
Potential kidneys
u Percentage of families that +
4 POTENTIAL 3 — aprove a donation
KIDNEYS LOST
Figure 4: Highly Aggregated Causal Loop Diagram of the model
The variables of interest, which later will define the main KPIs, are the waiting list for a kidney
tr I the tr I ions, the ilable kidneys and the potential kidneys lost. Note that,
according to the conceptualization of the problem presented, these variables are influenced by
demographic aspects (age chain, immigration etc.), medical aspects (organ rejections, blood type
etc.) and social aspects (people aware of the problem, adults willing to donate their organs upon
death, alive donors, etc.).
Taking the highly aggregated causal loop diagram as a starting point, the complete model was
developed. The final model contains over 160 variables (over 360 if subscript equations, which will
be introduced later, are considered), and is presented in Figure 5 in the form of a Sector-Diagram.,
where the main submodels that conform the overall model have been identified. These submodels
are described in subsequent sections. The complete Stock-Flow diagram can be found as an
appendix.
Age chain
Transplantations (regular + urgent)
opt wing aoe
‘Sy rc
ene we
‘oa
Sensitization
with the problem
“om oa }
Waiting list |
pat
| "
Policy
-| Media
Kidneys
supply
Policy Allocation
1
"iets pe
‘anya
Figure 5: Sector-Diagram of the kidney transplants in the U.S. Different submodels are identified.
Understanding the different submodels presented in the Sector-Diagram of Figure 5 is an easy
and comprehensive way to understand the modeling approach followed in this paper. These
different submodels are described below.
3.2.1 Waiting list, kidneys supply and transplantations
The issue of the increasing waiting list for a kidney transplant is modeled as a supply and
demand problem, as depicted in Figure 6.
Supply for kidneys Demand for kidneys
Figure 6: Waiting list as a function of the demand and supply for kidneys
The supply of kidneys comes from three major sources (Figure 7). These sources are the people
willing to donate upon death, the donations approved by families (people that were not originally
donating upon death but their relatives decided to approved a donation), and alive donors.
Supply of Kidneys
i
f \
People willing to Donations approved Alive donors
donate upon death by families
Y
Deceased donors
Figure 7: Main sources of the supply of kidneys present in the model
As for the demand part of the system, it has also been modeled as the sum of three sources:
teenagers, adults and retirees (Figure 8).
Demand for Kidneys
[Waiting List]
A
Teenagers needing a Adults needing a Retirees needing a
kidney transplant kidney transplant kidney transplant
Figure 8: Breakdown of the demand for kidneys
Note that it has been assumed pediatric cases (children under 12 years old) are uncoupled from
the rest. Kidneys coming or going from children are not compatible with the rest of the groups
because of size reasons, and thus they have been excluded from the model. A dedicated model
would be necessary to study pediatric cases.
3.2.2 Urgent cases
Just as in reality, urgent cases have been included in the model. Patients in the urgent waiting
list get priority over those in the regular one. Thus, if there is an available kidney, it will only be
used to satisfy the demand of the regular waiting list if there are no urgent cases. In the model,
urgent patients are assumed to be limited to those that suffered some kind of non-fatal
complication or problems during their transplant, and hence require a new intervention.
3.2.3. Age chain
The age chain sub-model is formed by 4 age groups: children under 12, teenagers (12-17),
adults (18-65) and retirees (66 and over). A yearly net inflow of immigrants, which has been
assumed to be constant, also forms part of the model.
This submodel is important because it pulls the dynamic of the general model. If the population
grows, the necessity of transplants will grow and the available kidneys will grow too. In other
words, both the supply and the demand for kidneys introduced in the previous section are a
function of the population.
In the supply part of the model the three sources of kidneys depicted in Figure 7 are defined
as fraction rates of the population. These rates were obtained from the Kidney Annual Data
Report of the U.S. (OPTN, 2012), and are not fixed over time, as they change with the society's
awareness of the problem (see 3.2.4 Sensitization with the problem).
The demand for kidneys is also specified by using a fraction of teenagers, adults and retirees
needing a kidney transplant in the U.S. These values are also obtained from the same source,
and their tendencies are determined using historical records.
3.2.4 Sensitization with the problem
When studying the evolution of the supply side of the problem, i.e. the kidney donors, it is
necessary that the model capture the social essence and dynamics of the issue. A
transplantation is an event that not only affects an individual, but also his sphere of influence:
relatives and friends. Thus, in order not to miss a big part of the conceptualization of the
problem, these extended social influence need to be modeled.
In the model, every time a patient receives a kidney, a fraction of the people on his/her sphere
of influence are sensitized with the problem, becoming thus donors upon death. This positive
feedback loop contributes to reduce the waiting list because more people become donors every
time there is a successful transplant. The same happens when someone dies. Because of the
traumatic situation of a relative or friend dying, some people get sensitized with the problem
and become donors, creating a positive feedback loop that contributes to reduce the waiting
list.
3.25 ducing blood types: subscript formulati
Although the subscript formulation forms part of the equations of the model rather than of its
structure, it is described here because of the impact it has on the behavior dynamics.
10
As explained during the model conceptualization (see 3.1 Model Conceptualization),a more
representative model of the problem of the kidney transplants in the U.S can be obtained if the
different blood types of individuals and their implications on transplants are included in the
model.
Data presented in Table 1 showed that people belonging to blood type O are universal donors,
ie. they can donate to all the other groups. Similarly, blood type AB represents the universal
receiver. An important conclusion that can be drawn from this table is that donors of type 0
have the greatest potential to reduce the waiting list, as they have the potential to satisfy the
demand ofall four blood types.A question that follows is what percentage ofthe population has
blood type O. Statistics for the U.S. are shown in Table 3.
Table 3: Blood type distribution of the population of the U.S.
| BLOOD TYPE DISTRIB IN THE
BLOOD TYPE PERCENTAGE IN POPULATION
A 42%
B 10%
AB 4%
10) 44%
This blood type distribution, as well as the implications it has in the supply-demand model, are
introduced in the model by the means of subscript equations, with the objective to capture the
extra complexity caused by organ incompatibilities.
3.3. MODEL VALIDATION
Before analyzing the behavior of the main model, a validation process is carried out to
determine how useful itis in representing the kidney transplantation system in the U.S. under
the system boundaries described in 3.1 Model Conceptualization. Four validation tests are
performed (Sterman, 2000):
= Reproduction of past real data: the tendencies of the waiting list can be confirmed for
the period 2010-2015 by using past real data. According to the U.S. Department of
Health & Human Services, the number of people waiting for a kidney transplant as of
March 13 2015 is 101,637, while the number provided by the simulation model is
roughly 97,500.
= Behavior anomaly test: This test analyzes whether changing or deleting assumptions
leads to anomalous behaviors. The following assumptions were tested:
1A.
o Constant Immigration: Was replaced by a lookup function in which immigration
rose exponentially from 1 million people a year in 2010 to 5 million people a
year in 2030, and remained constant from there on.
o Blood type of the immigrants is equal to that of Americans: These hypothesis
was substituted, changing the blood type distribution of immigrants to an
average of the blood type distributions in Europe.
o New people gets sensitized with the problem when close relatives/friends
undergo a transplantation process: This hypothesis was deleted.
Results from this test are plausible and match the logics that the modification of these
assumptions would imply in reality.
Extreme conditions behavior test: The purpose of this test is to analyze whether the
model responds plausibly under extreme conditions. The fraction of patients that
accepts a kidney during the transplant and average sphere of influence (used to model
the new people sensitized with the problem) are set to extreme values in four different
experiments. Results are also satisfactory in this case, as the behavior is plausible given
the starting conditions.
Sensitivity analysis: A sensitivity analysis was carried out and proved that the model
does not present behavior sensitivity under small variations in eight of the main
parameters. These parameters and the range of variations used in presented in Table
4.
Table 4: Ranges of variation for the sensitivity analysis
Lower Current Upper
sineiaaneil bound value bound
Probability to recover an organ from a person dead in a
hospital 0.25 0.268 0.3
Probability to recover an organ from a person not dead
ina hospital 0.006 0.008 0.01
Base percentage families that approve a donation 0.33 0.365 0.39
Base fraction of adults donating upon death 0.36 0.4 0.44
Alive adults’ donation rate 3.15E-05 | 3.19E-05 | 3.22e-05
Average percentage of adult people that need a kidney 1.4 E-04 1515-04 | 1.6E-04
transplants . . .
Average percentage of retirees that need a kidney 5.15 E-05 | 5.663E-05 | 6.15E-05
transplants . . .
Average percentage of teenagers that need a kidney 1.7 E-05 1.91E-05 | 2.1E-05
transplants . . .
4 Model Behavior: Base Scenario and Sensitivity Analysis
Simulating the base case scenario shows the behavior of the system if no policies were applied.
Four main KPIs are defined as indexes to quantitatively assess the evolution of the simulation:
1. KPI 1 - Total waiting list for a kidney transplant: the number of people waiting for
a kidney transplant in the U.S.
2. KPIZ - Ratio of total transplants over new people in the waiting list per year:
Total transplants [kidneys/year]
New people inthe waiting list [kidneys /year]
This KPI measures the evolution of the waiting list in a period. While the first KPI
represents the total cumulative waiting list, this one indicates variations of that
cumulative number. If KPI2 < 1, there is more demand than supply for kidneys. The
opposite happens when KPI2 > 1.
Note that the waiting list can decrease even if this parameter is less than 1, because
people not only leave the waiting list when they receive a transplant, butalso when they
die.
The optimal situation would be a waiting list of 0 and this KPI equal to 1.
3. KPI3 - Percentage of kidneys lost over total transplants per year:
Kidneys lost [kidneys/year]
Total tansplants [kidneys/year]
Kidneys lost are defined as those potential transplantable kidneys that are not donated
because the deceased person was not a donator upon death, and his relatives decided
not to donate his/her organs.
Thus, the higher this KPI, the higher the fraction of potential kidneys lost in year.
Conversely, the lower this KPI is, the less room for improvement, as there are fewer
potential transplantable kidneys lost.
4. KPI 4 - New people aware of the magnitude of the problem: cumulative number of
people that becomes sensitized with the problem, as explained in 3.2.4 Sensitization
with the problem.
The choice of KPI 2 and KPI 3 is motivated by the fact that the population size in the U.S.is going
to change over the timeframe considered (35 years), and these dimensionless KPIs avoid
misinterpretations that other, simpler KPIs, such as the total number of transplants in a year,
might bring. Figure 9 displays the trends of all the defined KPIs in the base scenario.
Furthermore, confidence in the simulation results can be increased if a sensitivity analysis is
performed. The ranges of variation of the parameters used for this analysis are the same to that
presented in Table 4. Results are shown in Figure 10. It can be seen that even considering
variations in 8 parameters, the behavior of all KPIs is only numerical sensitive, except for
extreme cases in which the waiting list might maintain a steady state, where a slight behavioral
sensitivity is shown.
413;
Total waiting list for a kidney transplant Ratio of total transplants over new people in the waiting list per year
140,100
125,100
110,000
95,020
KPI1
1
KPI2
po pg
o
2010
2015 2020 2025 2030 (2035 2040 2045
Time (Year)
Percentage of the kidneys lost over total transplants per year
2010 2015-2020» 2025-2030 »«2035 20402045
ime (Year)
New people aware of the magnitude of the problem
1 6M
KPI3
75 45M at
SS aM
25 15M
5 5 KPI4
2010 2015 2020 2025 2030 2035 2040 (2045 2010 2015 2020 2025 2030 (2035 2040 2045
Time (Year) ‘Time (Year)
Figure 9: Behavior of the model in the base scenario
Base
50% Hoo
Total waiting fist for a kidney transplant Ratio of total transplants aver new people in the waiting list per year
200,000 8
KPI1 KPI2
170,000
140,000
110,000
80,000, ©
2010
5
2010 2018 2028 2036 2035
4
2010
Tame (Year)
of the kidneys lost over total wansplants per year New people aware of the magnitude of the problem
™
KPI3
525M
35M
175M
KPI4
2028 2045
So10 2019 2028 2036 2045
Time (Year)
Time (Yeer)
Figure 10: Behavior of the model under sensitivity analysis
14
5 Policy Analysis and Implementation
Results from the sensitivity analysis for the base scenario in Figure 10 indicate that the issue
of the huge waiting list in the U.S. is going to become bigger in the next decades. This tendency
is also confirmed by external reports (OPTN/SRTR, 2013). Policy interventions are designed
and tested in this paper in order to study alternatives that can potentially ameliorate this
problem.
Two different policies are implemented. The first one has to do with a priority in allocation
system, which would give advantages to close relatives of organ donors. The second studies the
effects that a massive media campaign would have in the current donation ratio.
5.1 PRIORITY IN ALLOCATION POLICY
The basis of this policy is giving priority in the waiting list to the relatives of the people who
were organ donors upon death, and whose organs could be used to transplant to other people.
The authors hypothesized that such policy would largely increase the number of people
donating upon death, while the ethical and moral consequences that it implies are not as
negative as those from other incentives, such as fiscal or monetary (Garcia-Gallont, 2014).
In other to exemplify how this policy would work, a hypothetical example case is considered:
John, a 40-year family guy lives together with his wife and 2 daughters in the U.S. After finding
out about this new policy, he decides to become an organ donor, and is asked to select three
people that would benefit from this policy. He selects his wife and his two daughters. Sadly, after
two years, John dies in a hospital after a complication during an appendicitis surgery. His organs
are used to save 5 lives. Furthermore, if either his daughters or his wife need an organ
transplant in the future, they will have priority in the waiting list.
The question that follows is: does this process actually improve the system? Would this policy
lead to a decrease in the waiting list?
In order to prove the effectivity of the policy, the acceptance it would have is to be tested. For
that purpose, a survey was carried out to find out what percentage of the non-organ donors
would change their minds if this policy was implemented.
The survey introduced this policy to the respondents, and clarified it with a simple example
similar to that previously presented. After that, it contained basically two main questions, plus
some socio-demographic data. These two main questions were:
- Are you an organ donor?
- [If notan organ donor] Would you start donating if the policy was implemented?
The total sample consisted on 131 individuals. Results indicate that almost a 70% of non-
donors would start donating if this policy was implemented, as illustrated in Figure 11.
15
Would you start donating if the
Are you an organ donor? policy was implemented?
, but I
considered that
option
Yes
Y
26% 4406
=Yes =No_ #No, but! considered that option
= Yes #No
Figure 11: Main results of the survey regarding the allocation policy
Results in Figure 11 show a potential of up to (0.26 + 0.3) * 0.69 = 46.7% new organ donors.
But are these new donors really beneficial for the system? Or are the benefits from new donors
offset by the new priority that some people would be given?
It is easy to verify that this policy would improve the system if the probability that an adult
needs a kidney transplant, roughly 15 in 100,000 in the United States, is considered to be an
average of the whole population. In that case, given that a person can select up to three other
people who can benefit from the policy, chances that any of these people needs to be placed in
the high priority list are only 0.045%. Thus, benefits from a new organ donor coming from this
policy outweigh the complications that giving priority imply. These numbers are improved even
further if overlaps in beneficiaries of this policy are considered, as probably two parents would
both indicate their children.
Figure 12 displays the effects of the policy in a visual way.On the one hand, the implementation
of the policy would boost the organ donors upon death. On the other hand, the waiting list itself
would be affected due to the priority allocation some people would have. As it has been
demonstrated that chances that a close relative need a kidney transplant are low, benefits from
the supply side cover effects on the demand. It is important to note that this applies to cases in
which a person’s organs are actually used. If this is not the case, i.e. a person becomes an organ
donor but his/her organs are not transplantable upon death, then neither the supply nor the
demand would be affected.
16
ONLY 0,0045% OF THE POPULATION NEED
‘A KIDNEY TRANSPLANT
cee.
fefe)
{e)
DEMAND
the priority
waiting list
Extra supply to
| PRIORITY W.L
EXTRA SUPPLY
FROM THE
Extra supply to POLICY
cover normal
waiting list
NORMAL PEOPLE
SUPPLY OF ON THE WAITING
ORGANS BEFORE usT
THE POLICY
Figure 12: Visual scheme of the Priority in Allocation Policy
Some other complications have also been raised, as reported by some respondents during oral
interviews about their opinions on the policy. Finding a compromise between urgent patients
and patients that would be subject to this policy is one example, and would require further
analysis.
When this policy is implemented in the model, the behavior is that shown in Figure 13, where
the base behavior is compared to that in which the policy is implemented. Note that results
from the survey, i.e. around 47% new donors, have been supposed to take place only years after
the implementation of the policy, since a transition time is expected.
In order to test the robustness of the policy, the same sensitivity analysis that was performed
before is repeated. Results are presented in Figure 14.
17
Pant
35,000
KPI1
118,800 Base
102,500
Vin
86,250 Allocation Poti
icy
2010 2015 = 2020.» 2025-2030 2035 = 2040-2045 2010 2015 ©2020» 2025-2030 «2035 = 20402045,
Dent
Total waiting list for a kidney transplant
Ratio of total transplants over new people in the waiting list per year
Priority
8
KPI2
728
Dat
bay
pp pe
ime (Year)
Percentage of the kidneys lost over total transplants per year
Time (Yer
New people aware of the magnitude of the problem
3
i el ioe ee
Ads
, | KPI3
ane
2010 2015-2020» 202520302035 2040 20
Time (Year,
100M
KPI4
75M
50M
25M
0 = <a
2010 2015 2020 2025 2030 2035 2040 2045
Time (Year)
Figure 13: Comparison of the behaviors of the base and policy models
50% 75% 95° ff} 100%
Total waiting lst for a kidney transplant
Ratio of total transplants over new people in the waiting lit per year
200,000
KPI1
150,000
190,000
50,000
0 § 2
2010 2019 2028 2036 2045 at 20r o2u8: 2058 2085.
Time (Year) ‘Time (Year)
‘New people aware of the magnitude of the problem
Percentage of the kidneys lost over total transplants per year
6
100M.
KPI4
73M
SoM
0
2010 2019 2028 2036
Time (Year)
0
2045 2010 2019 2028 2036 2045
Time (Year)
Figure 14: Sensitivity analysis of the priority in allocation policy. The base behavior is displayed as the
red, thick line.
18
Paes
Results obtained indicate that the base behavior would greatly improve. The waiting list
would revert its increasing tendency in about 5 years and the number of transplants would be
around 75% of the new people joining the waiting list.
5.2 MASSIVE MEDIA CAMPAIGN
This policy focuses on increasing the number of people aware through media campaigns.
Creating periodically TV, radio or on-line advertisements would increase the number of people
aware of the problem and consequently would increase the number of kidneys donors. This
media campaign is assumed to be carried out once a year during two months.
Currently there exist no reports or data concerning the implementation of such a policy when
it comes to organ donations. Instead, the effect that the implementation of this policy would
have are estimated by reviewing and extrapolating the effects of previous campaigns in general
road safety (Delaney, et al., 2004), in preventing alcohol-involved crashes (Elder, et al., 2004)
and in public health messages (Newbold & Campos, 2011).
Figure 15 shows the impact this policy would have on the behavior of the model. For
comparison purposes, this figure also includes the behavior of the policy described in the
previous section, as well as that of a combination of both policies. It can be concluded that
implementing both policies at the same time is advisable, as their overlap is rather low,
resulting in a further improvement of the problem situation.
Total waiting list for a kidney transplant Ratio of total transplants over new people in the waiting list per year
130,000 8
Base case KPI 1
112.500 725
aaa assive Media Poli
Comp Priotity poy;
elgg gee |
nn
On.
POlicéed He |
95,000
‘ KPI 2
040 204%
2010 ©2015 «2020 2025-2030 2035 2040 2045 2010 2015-2020 2025, 2030 2035 iS
Time (Year) Time (Yeas)
Percentage of the kidneys lost over total transplants per year New people aware of the magnitude of the problem
Den
5 200M
KPI 4
a ela |
aie ty
a ame
125 ie 50M wee |
Time (Year)
0 |
2010 © 20152020» 2028- 2030-2035 += 20402045 2010 2015-2020 202520302035 20402048
Time (Year)
Figure 15: Comparison of the behavior of the KPIs in the base scenario and after the implementation of
the policies
6 Conclusions and Further Research
6.1 CONCLUSIONS
This paper has descbried a system dynamics approach to explore and analyze the problem of
the kidney transplants in the United Stated. The dynamics of the model depend not only on the
population size, but also on social effects and blood type compatibilities. Results of the
simulation indicate that the enormous waiting list for a kidney transplant in the U.S. will further
worsen if no action is taken.
Two different policies have been implemented with the aim to improve the supply side of the
system. As the most advantageous policies usually imply some ethical dilemmas, the policies
chosen are a compromise between effectiveness and social acceptance. The compliance with
the first proposed policy, the priority in allocation, was tested through a dedicated survey,
which happened to be a great success. As for the second, the massive media policy, research
reports from public health messages and road safety campaigns were used to estimate the
impact and consequences this policy would have in the U.S. society. If treated separately, the
priority in allocation policy saves more lives than the massive media campaign. In order to fully
assess and compare both policies, an economic analysis should also follow. The costs of
implementing each of the policies come from very different sources: developing a legal
framework for the first one and massive advertisement for the second one. Further research is
needed to come with realistic cost estimations, which could then be compared to other courses
of action that are currently being undertaken, such as research in stem cells.
The most positive outcome is achieved when both policies are implementing at the same time.
Because of the low overlap that the individual policies have in the system, this scenario
outperforms those in which the policies are implemented separately. Note however that,
although a combination of these policies play a big role in improving the behavior of the model,
they do not guarantee a solution to the problem. More sources of organs are required if the
waiting list is to be brought to zero. Although there is still room for improvement in kidneys
coming from people donating upon death, organs coming from deceased donors are limited
because the people dying a year is limited. Hence, encouraging living donation is an issue that
should be further explored, as its potential is unlimited.
6.2 FURTHER RESEARCH
This paper has tried to use some innovative techniques in the exploration of the problem. The
inclusion of blood types is an example of this approach. There are however many areas in which
further research and research methods are needed. This section intends to summarize the main
points that the authors regard as crucial when it comes to expanding the knowledge in this field.
First, pediatric cases should also be analyzed and included. Even though they are uncoupled
from the rest of the transplantation cases, they still affect the social aspects of the model. In the
same line of thought, a detailed analysis of the implications that different ages have in the
waiting list need to be considered. In this paper, the waiting list has been treated as a whole.
This was possible due to the age chain, which allowed for instance to realistically model the
20
supply coming from retirees by using a much lower fraction of transplantable to donated organs.
The authors however encourage further research in modeling the kidneys compatibility prior
to a transplantation.
Second, as part of the everyday lives of thousands of people in end stage renal diseases, dialysis
should also be model along with transplantations. Modeling how a combination of age renal
disease stage and dialysis affect the allocation in the waiting list would further improve the
results.
Third, geographical limitations should be considered. In this model, it has been assumed that
85% of the available kidneys are used for transplants, while the remaining 15% are lost (OPTN,
2012). These losses account also for difficulties in finding a match within a limited geographic
area. Also, the cases of end-stage renal diseases are aggravated in some states, as shown in
Figure 16. The authors suggest further research in how different geographies within the
country affect the allocation and transportation of transplantable organs. Integrating the
interstate collaboration required to maximize the distribution of kidneys will also shed more
light into the problem.
(ess Mass MBean: Missesveo Ml. s0
Figure 16: Map of the adjusted! incidence rate of ESRD, per million/year, by state, in the US.
population, 2012 (United States Renal Data System, 2014)
Fourth, in an attempt to capture the complexity of the real system, the issue of urgent cases
requires more exploration. The model hereby presented only labels as urgent cases those in
which a previous surgery went wrong. Including other medical conditions is also necessary to
expand the boundaries of the problem formulation.
Finally, as explained in the conclusions, more research is needed in the field of living donation.
The social dimension of the problem will play an even bigger role in this case.
1 Data Source: USRDS ESRD Database. Adjusted for age, sex, and race.
21.
7 References
CDC, 2015. Centre for Disease Control and Prevention. [Online]
Available at: http://www.cdc.gov/nchs/fastats/deaths.htm
[Accessed 13 March 2015].
Delaney, A., Lough, B., Whelan, M.& Cameron, M., 2004. A Review of Mass Media Campaigns in
Road Safety, s.1.: Monash University Accident Research Centre.
Elder, R. W. et al., 2004. Effectiveness of Mass Media Campaigns for Reducing Drinking and
Driving and Alcohol-Involved Crashes. American Journal of Preventive Medicine, 27(1), pp. 57-
65.
EuroStemCell, 2013. Kidney disease: how could stem cells help?. [Online]
Available at: http://www.eurostemcell.org/factsheet/kidney-disease-how-could-stem-cells-
help.
Garcia-Gallont, R., 2014. A Market for Kidneys. [Online]
Available at: http://www.nytimes.com/roomfordebate/2014/08/21/how-much-for-a-
kidney/paying-organ-donors-would-set-us-back
Graham, J.,2013. Dialysis Raises Hard Questions for Older Patients. The New York Times, 4
December.
Hampton, K. N., Goulet, L. S., Rainie, L. & Purcell, K., 2016. Social networking sites and our lives,
Washington, D.C.: Pew Research Center’s Internet & American Life Project.
Hirsch, G. B., Saeed, K., McCleary, J. K. & Myer, K. A., 2012. Deceased Donor Potential for Organ
Transplantation: A System Dynamics Framework, St. Gallen: System Dynamics Society.
Homer, J. et al., 2004. The CDC's Diabetes Systems Modeling: Developing a New Tool for Chronic
Disease Prevention and Control, Oxford: System Dynamics Conference.
Kolata, G., 1995. Families Are Barriers to Many Organ Donations, Study Finds. [Online]
Available at: http://www.nytimes.com/1995/07/07/us/families-are-barriers-to-many-
organ-donations-study-finds.html
Makely,J., 2013. Transplant outrage has a solution: more organ donors. [Online]
Available at: http://vitals.nbcnews.com/_news/2013/06/09/18809532-transplant-outrage-
h lution-more-organ-d. s?lite
Newbold, B. & Campos, S., 2011. Media and Social Media in Public Health Messages: A
Systematic Review, Hamilton: McMaster Institute of Environment & Health.
Nielsen, A. & Wakeland, W., 2013. Modeling Opiod Addiction Treatment Policies Using System
Dynamics, Cambridge: System Dynamics Conference.
NVSS, 2013. National Vital Statistics. Deaths: Final Data for 2013. Centre for Disease Control
and Prevention.
22
OIS, 2012. Yearbook of Immigration Statistics, Washington, D.C.: Office of Immigration
Statistics.
OPTN/SRTR, 2013. 2012 Annual Data Report: Kidney, Richmond: Organ Procurement and
Transplantation Network / Scientific Registry of Transplant Recipients.
OPTN, 2012. Kidney Annual Data Report, s.1.:
Organ Procurement and Transplantation Network.
Pruyt, E., 2013. Small System Dynamics Models for Big Issues. Delft: TU Delft Library.
Sherry, L. et al., 2013. National Vital Statistical Reports, National Vital Statistics System, s.1.: U.S.
Department of Health and Human Services.
Sterman, J.D., 2000. Business Dynamics: Systems Thinking and Modeling for a Complex World. 1
ed. s.l.:McGraw-Hill.
United States Renal Data System, 2014. 014 USRDS annual data report: An overview of the
epidemiology of kidney disease in the United States, s.|.: National Institutes of Health, National
Institute of Diabetes and Digestive and Kidney Diseases, Bethesda, MD.
23
