Kidney transplantations in the United States:
A System Dynamics approach to reduce the waiting
list and illegal trafficking
Sebastiaan Fakkert!, Philipp Schwarz’, Erik Pruyt?
Delft University of Technology - Faculty of Technology, Policy and Management,
Jaffalaan 5, 2626 BX, Delft, The Netherlands
Abstract — The increasing need for kidney transplants and the
structural gap between kidney donation and demand is virtually
a universal problem, and has been challenging policy makers in
the US and worldwide for years. Although improvements for
kidney procurement have been made over the years, still, on
average 13 people on the waiting list die every day while awaiting
a kidney transplant. At the same time, illegal kidney transplants
are flourishing. In this study, an attempt was made to explore the
various dynamics involved in the kidney transplant system by
means of System Dynamics simulation, to identify leverage points
for policy interventions and explore various policies and their
effectiveness under highly uncertain conditions. Key focus was on
the transplant waiting list, donor registrations, and both legal
and illegal transplants performed. The model was validated in
accordance with Organ Procurement and Transplantation
Network Data. Simulation results showed that without
intervention, problems are expected to worsen. Complementary,
uncertainty analysis confirmed these results. The only policy
which was found to have large enough impact to reverse the up
going trend of the waiting list, is to provide financial
compensation for unrelated living donors. However this policy is
accompanied with many ethical issues.
Index Terms—Kidney transplantation, illegal organ trafficking,
deceased and living donor, system dynamics, uncertainty analysis.
Word count: 5152
I. BACKGROUND
idney shortage is a growing problem for patients in need
for kidney transplants. In the United States alone, nearly
100,000 patients with end-stage renal disease (ESRD)
were registered on the national waiting list for kidney
transplantations in 2012. As Fig. 1 illustrates, the shortage for
kidneys has been increasing every year. Demand structurally
exceeds the supply by far; each year there are more waiting
list additions than removals. For example, in 2012, about
34,000 people were added to the waiting list, but only 11,033
patients actually received a kidney (Organ Procurement and
Transplant Network 2013). Consequently, currently waiting
patients on the list wait on average (median) more than four
years before they receive a kidney transplant (OPTN, 2012).
More transplantations are not possible due to insufficient
donors. The consequences of this shortage are dramatic; every
*Main author names listed in alphabetic order
} Sebastian Fakkert, MSc student Engineering and Policy Analysis
Email: S.C.M.Fakkert@ student.tudelft.nl
Philipp Schwarz, MSc student Engineering and Policy Analysis
Email: p.schwarz@ student. tudelft.nl
Fig. 1: The growing shortage
THE GROWING SHORTAGE
1998-2012
60000
40000
20000
0
1998 2000 2002 2004 2006 2008 2010 2012
== Waiting list | Deceased donor
— living donor = Transplants
year more than 4,000 patients in the US die while waiting fora
kidney transplant and many more suffer from degrading health
condition and are therefore dropped from the list, unlikely to
ever be resumed (United States Renal Data System 2010).
Legal kidneys are obtained from both living donors (people
willing to donate one of their kidneys by life) and deceased
donors (people willing to donate their organs after death).
Many potential kidneys of deceased persons are lost for
mainly the following reasons; 1. The person did not register in
the national donor register (currently only 38% of US citizens
are registered (U.S. Department of Health & Human Services ,
2015)), 2. The circumstances of death were not suitable for
donation, 3. When the conditions for donation are optimal, but
the decedent is not registered as a donor, the related family
often decides to decline donation, since the persons lack of
registration is in case of doubt rather seen as an objection.
Meanwhile, the failure of effective organ procurement
policies to meet the rising demand for kidneys in the US,
causes flourishing in illegal kidney trafficking (Glazer, 2011,
pp. 341 - 366). Desperate waiting patients in need for a kidney
are tempted to participate in illegal organ transplantations
provided by international organized crime. So far, little is
known about the size and the structure of this sordid business
which typically takes place in less developed countries.
However, patients are reportedly willing to pay up to $200,000
for an illegal kidney transplant, a kidney for which the donor
often receives as little as $1000 to $5000 (World Health
3Co-author: Erik Pruyt
Delft University of Technology, Delft, The Netheriands.
Email: E.Pruyt@ tudelft.nl
Organization, 2015). Often transplantations takes place under
bare, suboptimal conditions (World Health Organization,
2015) and endangers the health of donor as well as recipient.
Some previous System Dynamics research with regard to
organ shortage has been performed. A similar scope as ours
was chosen by (Hirsch, McCleary, Saeed, & Myer, 2012),
who modelled the (kidney) donor potential in the US and
identified in particular feedback loops within the organ
procurement and transplant center capacity. However, their
approach differs, as they have a particular strong focus on data
calibration rather than on policy exploration and behavioral
change. Moreover, they focus solely on the donor potential of
deceased donor, overlooking the potential of living donors.
Another study by (Azar, McDonnell, & White, 2013)
modelled the dynamics of the renal system as a whole,
incorporating many aspects relevant regarding dialysis.
Whereas a purely forecasting approach was performed by
(Prasanna Devi, Suryaprakasa Rao, Krishnaswamy, & Wang,
2010), who modelled the development of corneal transplants
for a transplant center in India. Finally, kidney transplantation
system. characteristics were explored regarding the optimal
ion to recipients across the country with
both a system dynamics approach by (Fernandez, 2014) and
with a purely discrete modeling approach by (Davis,
Mehrotra, & John Friedewald, 2013).
For our study we chose a holistic approach to identify
II. METHODOLOGY
In this study, a model was made by the use of System
Dynamics, with the purpose of exploring different policy
options to decrease the kidney transplant waiting list and fight
the illegal kidney trafficking business. A simulation time
frame from 2012 until 2030 was applied.
System dynamics is a methodology for studying complex
feedback systems and makes use of stock flow diagrams to
simulate system behavior over time (Forrester, 1961; Sterman,
2000). SD models allow to identify generic structures and
behaviors that can be used to search for policies that influence
the systems behavior in different scenarios (Azar, McDonnell,
& White, 2013).
First, a conceptual model was made by means of a Causal
Loop Diagram (CLD), in which the main feedback loops and
the different subsystems were identified. In addition, a bull’s
eye diagram was created that expresses which variables were
included or excluded from the model and to which detail
modelled (Pruyt, 2013). The actual modelling of the system
was started with several Stock Flow Diagram (SFD) which are
explained in later paragraphs. After developing enough
confidence in the model in an iterative process, policies were
developed and implemented, and as a result the SFD’s
structure and/or parameter values were adjusted. This included
an extensive uncertainty analysis to test the explored policies
and the models parameters for deep uncertainty.
The model initial values, stocks and fractions are calibrated
dynamics in the kidney antation sector, 1 by
numerous interrelated feedback loops. The model incorporates
also the interactions between the illegal kidney
transplantations and the legal kidney transplantations system.
The main purpose of the model, is on exploring the relative
performance of diverging policy options that influence the
systems’ behavior. Though, in this study, OPTN and other
data source were used for quantification of the model’s initial
values, flow rates and fractions, the main focus is on system
behavior.
In the remainder of this paper, a description of the model is
given, along with the chosen policies for increasing the legal
transplants performed along with a reduction of the waiting
list and illegal kidney trafficking. First, the methodology that
was applied to obtain the results is addressed. Second, the
| model is p which i the modelers’
line of thought and ‘important modelling decisions. Third, a
detailed description of the model and its sub-systems is
provided. Then, base case simulation results are discussed for
this model. These results express the behavior of the
hypothetical model. Following, different policies are
addressed that were explored for this model. Finally, an
uncertainty analysis was performed to address the deep
uncertainty in the models’ parameters, delays and explored
policies, These results are discussed in the last chapter.
ding to OPTN data, which contains data of every organ
donation and transplantation event in the US since 1987 and
other sources: (OPTN, 2012), (Centers for Disease Control
and Prevention, 2015) and (National Kidney Foundation,
2015). These extensive databases provided all reference data
needed for donor characteristics, transplantations, waiting list
size, and survival rates etc.
IIL. MODEL DESCRIPTION
A. Model boundaries
During the modelling process, decisions were made for the
level of detail on which certain parameters/issues should be
modelled. The bull’s-eye diagram in Fig. 2 provides a general
overview of the elements that were modelled thoroughly
endogenously, superficially endogenously, exogenous or
deliberately omitted from the model. Only the deliberately
omitted elements and the exogenous elements will be
addressed in this section.
An important exclusion was made for the donor-recipient
matching process that is essential in real-life organ
transplantation. In this process, important characteristics
(mainly blood type and human leukocyte antigens) for both
the donor and the possible recipient are taken into account in
order to determine the best match. However, since the amount
of patients on the waiting list is so large, a match is always
found. Thus, this process is not of particular interest in this
research, as the amount of kidney donors are the main
problem, rather than their specific matching characteristics.
Fig. 2: Bull's-eye diagram
yi" Mortality rates,
Active/
sp
aan Ye °
swraling tet / Birth rate trafficked
passive
Kidney qual
nservation
opulation kidney \
growth \
Matching / Deceased donation
proces | / Trnssanatons\
| \ \
| Price Graft failure Organ \ \
| trafficke procurement \ \ |
couse of | mec | legal kidney Tansey
ause o |
kidney | \ transplants Transplant patients | Senter |
\ \ : / cspacty
failure | \ on waiting list
\ ar \ Living donation
\ itlegat \ THOROUGHLY MODELLED
\ ENDOGENOUS VARIABLES
\market
Geographical \ ESRD-
aspects “SUPERFICIALLY MODELLED so, /
ENDOGENOUS VARIABLES
Technological
developments
DELIBERATELY OMITTED
IAL
Another element that was deliberately omitted from the model
are possible future technological improvements that might
affect the kidney transplantation system. For example, the
creation of kidneys with use of genetic modification. Although
such developments could radically change the models
behavior, they are not included within the timeframe of the
model.
Moreover, as the authors applied a holistic approach to
understand the general behavior for the US as a whole,
geographical aspects, such as transplant center locations or
kidney distribution networks, are excluded from the model as
well.
Furthermore, the model omits the difference between passive
and active waiting list, as only the later people are eligible to
receive transplants.
A large potential to tackle the root cause of the problem are
the causes for kidney failure, particularly obesity and
hypertension. Nevertheless they were not included in the
model.
B. Conceptual Model
The model is centered around the transplant waiting list and
focuses on the identification of reinforcing and balancing
feedback loops that influence its length. Fig. 3 provides an
overview that shows, which loops drive the system and which
loops are rather balancing.
The flow of renal patients in need of a kidney transplant
entering the waiting list is indicated by the arrow at the right
towards the waiting list. Over the last decades the number of
ESRD has shown a rising rapidly trend.
It is triggered by the dynamics of the population in terms of
age structure, death rates and the fractions of new cases of
end-stage renal disease. In addition, the population variables
originates also how many kidneys are available from both
deceased and living donor. Notice, the development of the
population is incorporated in the model and described in more
detail in the next chapter, but not displayed in the highly
aggregated causal loop diagram.
Fig. 3: Highly aggregated causal loop diagram
Demand for
trafficked kidneys
| (7) Waitinglist =
+\ controls demand Supply of
trafficked kidneys
wells © |
GY
‘drives supply
= [__ @2)lllegal me
(Transplants reduce
waiting list
4% Renal patients
Transplant patients 4
Removals from on waiting list /
o (3) Removals
waiting list control growth of
ting list
sale Social pressure to
(1) Transplants | + become donor
reauee wating
*
— () Social pressure ,
‘ang ak deceased Registered
t, donors \
Hospital
capacity
(4) TFafsplants
limited by capaci
wih pressure
controls living dong,
Living donors
Furthermore, the CLD provides an overview of the eight main
feedback loops that can be identified:
(1) Transplants reduce waiting list: patients are removed from
the waiting list after transplantation;
(2) Illegal transplants reduce waiting list: patients are
removed from the waiting list after transplantation with
trafficked kidney;
(3) Removals control growth of the waiting list: the growth of
the waiting list is dampened by the patients that are removed
from the waiting list, because their health condition has
deteriorated or they died while waiting to receive a transplant.
(4) Transplants limited by capacity: transplant capacity
restraints the legal transplants that can be carried out. This is
not problematic under normal development, since capacity is
delayed adaptive to the legal transplants performed. However
it is expected to be a limitation if overall kidney supply
increases suddenly, for instance because of a policy action.
(5) Social pressure controls deceased donors: increasing
social pressure stimulates the number of transplants from
deceased donors;
(6) Social pressure controls living donors: increasing social
pressure stimulates the number of legal transplants performed
with kidneys from unrelated or related living donors. In turn
the length of the waiting list depends on the number of kidney
donors.
(7) Waiting list controls demand: increasing length of the
‘ith
percentage of 2.1% of living donors is at the time donating
aged 65 and older (National Kidney Foundation, 2015).
B. Sub-system Transplantation
Fig. 5 provides an overview of the stocks and flows within the
transplantation sub-system. Patients only leave this system
when they die, for example due to renal failure or as
of s
waiting list triggers demand for wi
trafficked kidneys;
(8) Demand drives supply: the waiting list or more precise the
average waiting time triggers the demand for trafficked
kidneys. Criminal organizations respond to this demand for
trafficked kidneys and expand supply channels.
In the next chapter the model structure is presented in more
detail. Four main subsystems have been modelled (1)
(2) T: (3) Illegal
kidney trafficking and (4) Kidney transplant capacity.
IV. DETAILED MODEL STRUCTURE
A. Sub-system Population
As can be seen in the stock flow diagram in Fig. 4, the
population is classified broadly into three age categories;
pediatric, adult and senior population. In addition, two other
stock variable were added to distinguish living donors (both
related and unrelated) from the rest of the population.
At first, population aging, was modelled with regular first
order material delays. However, since aging dynamics are
actually rather discrete events, in Vensim they are often
modelled by using conveyors. This function was implemented
in the model for maturing and retiring, and it was found that
with the same set of parameters the behavior correspond closer
with far more complex population projection models such as
those published by the U.S. Census Bureau (United States
Census Bureau, 2009).
Finally, there is also no link between the senior population
and the senior living donors since, only a negligible
The inflow to the transplantation sub-system are new cases
of ESRD, which accumulate in a stock. At present, in the US
alone, over 700,000 patients are affected by ESRD (United
States Renal Data System, 2012), which require compulsory
routine kidney dialysis or transplantation. The main causes for
ESRD are diabetes and hypertension (OPTN, 2012; United
States Renal Data System, 2012), however in this model only
simplified represented as fractions of new cases per year of the
population.
A fraction of ESRD patients move further each year to the
waiting list as their health status is acute and requires as soon
as possible transplantation. Anyhow, since the United States
lacks a universal health insurance for all citizens, access to the
waiting list is further restricted and the real scarcity of kidneys
is rather underestimated. Since transplantation and post-hoc
medication is extremely expensive it can usually not be
afforded by uninsured patients. Consequently out of fear that
that their financial circumstances will cause kidney failure,
uninsured persons are likely to be declined by transplant
centers (Laurentine & Bramstedt, 2010); among others,
possible changes at this leverage point were addressed in the
uncertainty analysis.
While waiting to receive a transplant a fraction of patients is
removed from the list, usually because their health status
deteriorates and make ible. As d in
the introduction the average waiting time for a kidney
transplant is more than four years, and in 2012 only 11,033
patients received a kidney transplant. Transplants are limited
by the availability of kidneys and at least theoretically also by
the transplant center capacity.
Fig. 4: Stock Flow Diagram for Sub-system population
living donor with senior living donor
with ESRD
os Living donor Senior living |
feat ashe age 18-64) retiring diving cE (age>65) |death of senior
Ara living donor
ee death of adi diving donor
Pediatric at Senior ee
population
pects} maturing Gee 18:88) | ~etrng adult Pee death of
pediatric with
ESRD
adult with
ESRD
senior with lor
ESRD
The transplants performed each year utilize organs
approximately half each from living and from deceased
donors. The number of transplants from living donors that
can be performed is naturally limited by the number of people
willing to become living donors. The motivation to become a
Fig. 5: Stock Flow Diagram Sub-system T'
Little is known about illegal transplantations since of its very
nature. However, it is estimated that around 5% of all
recipients, obtain a transplant from grey sources each year,
usually transplantation takes place outside the country
(Shimazono, 2007). The simplified underlying mechanism is
presented in the next paragraph. Anyhow, if the
a
New chses of
ESRD
Patients with end
stage renal disease
waitilig list patients Femoved
registfation from list
Illegal
‘ Transplant patients |szp-death
patients
on waiting list
Tegal
Transplantation
fatal Graft failure
Transplanted
is not-successful the patient may remain on the
waiting list and does not have to re-register.
C. Sub-system Illegal kidney trafficking
Because the organ supply cannot meet the rising demand a
flourishing global black markets for illegal kidney trafficking
has emerged. Potential recipi are out of |
willing to take the risk and travel abroad to obtain kidneys
through commercial transaction, the price of a renal transplant
ranges from US$ 70 000 to 160 000 (Shimazono, 2007).
In the model, the demand for illegal trafficked kidneys is
believed to be proportional to the transplant patients on the
waiting list, and the reference price paid for the illegal kidney
is assumed to be US$ 100,000.
The stock “supply of trafficked kidneys” is fed by the
supply response to markets (organizations that are active in
illegal organ trading react to demand from the ESRD patient
side). This supply response is controlled by the relative
profitability of organ trafficking which captures the supply
and demand mechanism based on an average kidney price, the
effective chance of being caught for trading, and the response
time of the supply to react on market demand.
The supply and demand ratio determines the average price
patients
living donor, is influenced by the social pressure to become a
donor, triggered by the length of the waiting list and may be
also influenced by policy making.
In terms of deceased kidney donation, the underlying
reasons for the supply shortage is that only a small fraction of
deaths each year occur in a manner that the kidney can be
obtained before kidney taint, since removed kidneys require
immediate and adequate conservation. If the death does not
occur in hospital there is little chance that the organ can be
transplanted. Furthermore, only a small fraction of potential
donors is generally medically suitable, which further
diminishes with ageing (OPTN, 2012).
Only in addition to this the fraction of people registered as
donors and the consent from donors’ families may pose two
leverage points for policy makers.
Transplantation is nowadays initially almost always
successfully. However transplanted patients face the risk that
the foreign organ is rejected and the graft fails, though the
patient intake of immunosuppressant a medicines that lower
the body's ability to reject a transplanted organ. In case of
graft failure, the patient moves back to the stock of patients
with end stage renal disease and may again register on the
waiting list for re-transplantation.
price for trafficked kidney. This
effect is reflected by a lookup function following an
exponentially declining graph. The smaller the supply relative
to the demand, the higher is the average price paid and vice
versa. This market supply demand effect is smoothed to take
into account information delay time to price changes.
This average price has a twofold effect. First, it drives the
demand from the ESRD patient side by an exponential
declining relation between the average price paid and the
demand, reflected by a lookup function: demand for trafficked
kidney is higher when the average price is low and vice versa.
Second, it influences the relative profitability of organ
trafficking which controls the response of supply to market.
Finally, demand for trafficked kidneys is affected by both the
risk of illegal transplantations and the average price paid for a
trafficked kidney. Hence, mutual interaction between supply
from kidney supply channels and demand from the ESRD
patient side determine market prices and thus the
transplantations performed with trafficked kidneys.
V. BASE CASE SIMULATION RESULTS
The base case is a hypothetical scenario, simulated with the
models default values and assumptions. The base case
behavior displayed in Fig. 6 suggests that the transplant
patients on the waiting list more than double until 2030. This
waiting list growth corresponds well to comparative System
Dynamics studies, such as (Hirsch, McCleary, Saeed, & Myer,
2012).
Fig. 6: Base case simulation result for waiting list
Transplant patients on waiting list
300,000
225,000
i 150,000 Leer
75,000
{tee
art
0
Fig. 8: Base case simulation result for transplants
Legal transplants per year
20,000
17,500 eee
j 15,000 pe: eta
et
10,000
2012 2014 2016 2018 2020 2022 2024 2026 2028 2030
Legal transplants per year :BaseCase #432333
Fig. 8 shows the base case behavior of the legal transplants
performed per year. A comparison of both Fig. 6 and Fig. 8
clearly shows the gap between the demand for kidney
transplants and the supply. As expected, illegal kidney
increase in line with the waiting list, see Fig. 9.
2012 2014 2016 2018 2020 2022 2024 2026 2028 2030
Time (Year)
Tr list : BaseCase
Moreover, this behavior met the expectation, since it follows
the current trend and is driven by the rising number of ESRD
patients each year (United States Renal Data System, 2012).
The ESRD growth is a result of the US population dynamics
combined with an unhealthy nutrition and lifestyle. While the
number of donations remains relatively constant and does not
rise along with the demand, see also Fig. 1.
Fig. 7: Base case simulation result for registered donors
Percentage of registered donors in the US
40 ee
[Pt ery
30
2012 2014 2016 ©2018 +2020 +2022 ©2024 +2026 ©2028 2030
BaseCase
Fig. 7 shows the base case behavior for the potential donor
population that is registered as a donor, which is only 38
percent at the model’s initial time. A slight increase is seen
within the simulation timeframe, driven by the increasing
social pressure to become a donor. In general, this social
pressure restrains the increase of the waiting list to some
extent, however the balancing effect is not enough to flatten
the waiting list growth.
More specific, the length of the waiting list is the main driver
for the illegal transplants sub-system, as this is the only link
with the rest of the model. Initial interpretation showed that
any effective policy against illegal kidney trafficking should
come from the demand side, rather than the supply side.
Fig. 9: Base case simulation result for illegal transplants
illegal kidney transplants per year
a
3750 Lae
5
F 2500 {heey
0
2012 2014 2016 2018 2020 2022 2024 2026 2028 2030
Ilegal kidney transplants per year : BaseCase. —3—+—z—3- 2 —3-—
VI. POLICY EXPLORATION
The main purpose of the model is to identify which
interventions by policy makers are likely to be the most
effective, taking into account the deep uncertainty of many
parameters, functions and model structure. The status quo
simulation showed that the available donated kidneys limit the
transplants performed. Consequently, increasing transplant
program capacity is only secondary relevant.
Policies are derived from those currently debated or ones that
already have been implemented in other countries. For better
comparability, it is assumed that all policies are implemented
in 2015. The first two to be discussed policies aim to increase
the deceased kidney donation, the third policy encompasses
lowering of kidney quality criteria and the last two policies
describe measures to increase kidney donations from unrelated
living donors (Becker & Elias, 2007).
Some systems feedback loops form resistance to change and
impede policy making. First, since the social pressure is
related to the waiting list, there is a balancing effect: if the
waiting list is reduced, less people are willing to become a
donor. Second, the kidney transplant capacity limits the
possibilities of implementing any policy immediately
stimulating more transplants. Therefore, the effectiveness of
the policies are for now only tested under the assumption that
the transplant capacity is unlimited, assuming that adapting
this capacity to the changed conditions, is not the core
challenge.
A. Policy I: Incentives for Registered donors
As the first country in the world, Israel passed a legislation in
2008 to reward cadaveric donation with financial
such as reimk of funeral costs, to
encourage people to register as deceased donors. For the
purpose of this model, it is assumed that this policy increases
the percentage of registered kidney donors from 38% to more
than 50% (Fig. 10). However the effect on the waiting list is
very small, as the red lines in simulation results presented
show (Fig. 12-13).
B. Policy II: Default opt-out
Implementing this policy means that every citizen will be by
default registered as a donor and has to explicitly recall his/her
registration. In Belgium, where this policy was implemented,
the fraction of registered kidney donors doubled (Erasmus
School of Economics, 2014). This percentage has been also
assumed for the model. In Fig. 10, 12, and 13 the turquoise
and green lines, represent the effect after the policy was
activated. However, despite raising the number of registered
donors, the waiting list does not shrink significantly; the effect
is only slightly bigger than for the first policy. Extreme
boundary assessment tests show that even under assumption
that everyone becomes a deceased donor, the supply of proper
kidneys is insufficient. Suggesting policy makers to explore
the potential of living donation.
Fig. 10: Percentage of registered donors after implementation of
policies
fr registered kidney donors
8645 Pa a Ps a
i
7226 Se shag 2
5806 al : zt st
4387 |
a el fi pi ta
2968
2012 2014 2016 2018 2020 2022 2024 2026 2028 2030
Time (Y ear)
4 4 4 4
regjtered kidney donors: Policy 1- Imentives for Registered donors 2 2
1
UL 4
WV. 5 5
y r
C. Policy III: Increase Acceptance Rates of Organs from
Less-Than-Optimal Donors
Another often discussed policy to tackle the challenge of
kidney shortage is to accept kidneys from less-than-optimal
donors by lowering quality standards. Fig. 11 provides an
overview of the impacts such a policy may have.
Fig. 11: Causal loop Diagram kidney quality
strictness of
acceptance criteria
deceased donor
kidney transplants y
~ fr adult/senior
donor
death of
transplanted patients
+ LY d
we
* ta —
graft failure ESRD (dialysis)
+ patients not registered
onbtte | fA
control \ |
Transplanted Transplant patients
patients, , on waiting list
The quality of organ transplanted is influenced by three
drivers. The first driver is the ratio of adult to senior donors,
since with ageing the quality of the kidney generally
deteriorates. Second, the ratio between living donors and
deceased donors, as living donors provide generally kidney
with higher medical quality. Lastly, the policy parameter,
strictness of acceptance criteria, which in addition may
influence the two first mentioned factors.
The quality of the kidney transplanted determines the
probability for graft failure. The lower the quality, the more
of the transplant list once they develop ESRD in their
remaining kidney and need a transplant. Though in the model
it is assumed that this measure would almost double the
motivation to become living donor, the effect on the waiting
list is however small (Fig.12).
F. Combined policies without financial compensation
Many people have ethical concems about financial
compensations for donation (Kelly, 2013). Therefore,
recipients will either decease as a c of the
transplantation or will return to the ESRD patients stock.
The effectiveness of this policy is limited, though the number
of transplanted kidneys raises slightly, but has a major
drawback. As Fig. 14 shows, the graft failures increase after
policy activation, when compared to the base case and other
policies.
On the contrary, policies that enhance the number of living
donors, do not only rise the number of transplants per year, but
also result in a decline of graft failure, due to the earlier
described dynamics.
D. Policy IV: Compensation for unrelated living donors
About half of all kidney transplants in the US come from
living donors. The advantage of policies stimulating living
donation, is that the additional kidney supply is not
marginalized by insufficient kidney condition. Furthermore,
living donor transplants have a lower probability for graft
failure than those obtained from deceased donors.
Compensation for unrelated living donors may take very
different forms, and does not necessarily has to be of monetary
value. Experiences from other countries show the
effectiveness of those policies. For example, in France,
transplant centers reimburse living donor travel and lodging
costs. In some Canadian provinces lost wages while
recovering from the surgery are reimbursed (Laurentine &
Bramstedt, 2010). The compensation for unrelated living
donors may only be the correction of disincentive, so should
the government protect donors from insurance companies that
charge higher premiums for living donors. Some say donors
should be d gover paid health i to
cover the risk of obtaining the kidney and later complications.
Though, the effect of such granting citizens a compensation
can only be estimated, the model shows that there is a lot of
potential in increasing living kidney donation. In the model it
is assumed that due to the incentive of US$ 15.000 the living
donors tenfold, see Figure 12 and 13. Thereby it is the only
policy which is found to be effective to reduce the waiting list
to a lower level, and which would significantly reduce the
average waiting time.
E. Policy V: Priority on waiting list of former living donor
The risk of a living donor is hardly significantly higher to
develop ESRD than a person with two kidneys (Muzaale, et
al.). Nevertheless it would be a powerful incentive to become
an unrelated living donor if former donors can jump to the top
lly a set of policies was created combining all
policies that do not include financial compensation and thus
are less controversial. Though the effect of the individual
polices add up, the joint impact is still relatively small Fig. 12
and 13.
G. All Policies
Last, the effect of all policies combined are examined. If all
policies are active the desired effect is greatest. Nevertheless,
as Fig. 12 shows, the trend can only be reversed until 2021,
from which on the growing number of ESRD patients drive
the waiting list up again.
Fig. 12: Graft failures after implementation of policies
Transplant patients on waiting list
217,200
178,500
g
i 139,800
101,100
62,370
2012 2014 2016 2018 2020 2022 2024 2026 2028 2030
Time (Y ear)
Fig. 13: Legal transplants after implementation of policies
Legal transplants per year
50,000
37,500
25,000
Person/Y ear
12,500
0
2012 2014 2016 2018 2020 2022 2024 2026 2028 2030
Time (Y ear)
Fig. 14: Graft failures after implementation of policies
post hoc test -graft failure
20,000
15,000
10,000
Person/Y ear
5000
0
2012 2014 2016 2018 2020 2022 2024 2026 2028 2030
Time (Y ear)
4
“post hoc test -graft failure" : Base Case. —4
“post hoc test -graft failure" : Policy I - Incentives for Registered donors
“post hoc test -graft failure" : Policy II - Default opt-out onoff +
“post hoc test - graft failure" : Policy without financial 7- #
“post hoc test -graft failure" : All policies &
VII. TESTING POLICIES UNDER UNCERTAINTY
The kidney transplant model described until now, incorporates
numerous socio-economic factors whose values, and feedback
loops, and strenght can only be estimated and not derived from
solid sources of data. Furthermore, many parameters and
fractions possibly change in the future, which may I
Fig. 15: Envelopes and kernel density estimates of registered donors
“APolicies Ty Pekeyt Poliey3
CombiPOLswoCompensation [= Policy2 El Falicy4
NoPolicies
reverese trends. For the purpose of this study most relevant are
dynamics, which change the effectiveness of policies.
(Bankes, 1993) was among the first suggesting to explore
with models possible scenarios and dynamics related to
uncertainty, rather than predicting future outcomes.
In this study, the approach of (Kwakkel & Pruyt, 2013;
Pruyt, Auping, & Kwakkel, 2015) was followed, who built on
this and d several cases to di the application
of policy exploration with System Dynamics.
In the following paragraph, the results taking into account
both structural and parameter uncertainty and their effect on
the policies are discussed. For the parameters used in the
uncertainty analysis, the uncertainty ranges as well as delay
order uncertainty are presented in the table in the Appendix.
For each policy 500 simulations runs where performed
respectively with a different set of the possible input values
selected from their uncertainty ranges with latin hypercube
sampling. In addition, in order to test Policy V: “Priority on
waiting list for former living donors”, an extended model was
used to incorporate the additional stock flow structures and
dynamics.
The of the analysis is i with
visual ensemble inspection (left) and kernel density estimation
(right). Kernel density estimates are the continuous variant of
histogram representing the distribution of data probabilities.
The figures 15 to 17 present a comparative policy analysis
for the model with no priority waiting list for former living
donors. Because of the uncertainty ranges, the registered
donors (Fig. 16) are spread out, however repeat the base run
behavior.
On the contrary, Fig. 16 indicates that the waiting list is
behaviorrally sensitive. In the worst scenario 400,000 people
will be awaiting a kidney tranplant in 2030. In another
extreme scenario in which all policies are implemented and
other conditions are optimal, the waiting list will approach
zero already in 2020.
The kernel density estimation shows that there is a very clear
difference between all policies in which financial
compensations are povided to living donors, which confirms
the initial conclusions.
Finally, Fig. 17 provides an overview of the legal transplants
performed per year, with no surprising insights.
When the patients that have been living donors are prioritized,
the effect on the waiting list is marginal, which can be seen
from Fig. 18, comparing both models.
Te
Fig. 16: Envelopes and kernel density estimates of waiting list
[ia Aloficies Polga Polya
ymbiPOLSWoCompencation sl Policy2—‘Plicya
1 NoPolicies
2 Os ONS IS <0) —«moD~S«C«DAS«02G «OD «2030 80-05
Time
Fig. 17: Envelopes and kernel density estimates of legal transplants
iam AlPolicies Poly Policy
1 CombiPOLSwoCompensation = Policy? Policy
cm NoPolicias
¢ 0 700 i 700 30 0 5ee05
Fig. 18: Envelopes and kernel density estimates of priority policy
= ]
¥
g
g
g
“Peal renslant eters cn wating Sst
g
Ey
g
VIII. CONCLUSION & RECOMMENDATIONS
This study shows, traditional measures are not effective to
limit the up going trend of the kidney transplant waiting list in
the US. Providing financial compensations for unrelated
donation was found to be the only policy that has the potential
to reverse the trend and reduce the waiting list to a
significantly lower level. Nevertheless, for policy makers the
crux is whether it is moralistically justified to use market
incentives to increase living donation to save ESRD patients’
lives.
The persistent and increasing gap between kidney supply
and demand, leads to the conclusion that policies have to be
developed encouraging more donation. Simulation shows that
the waiting list will have doubled by 2030, if no actions are
taken.
Implementation imposes further hidden challenges, such as
the transplant center capacity, that only adapts slowly, and
thus limits the transplants that can be performed.
Consequently, any initiated policy has to be coordinated with
transplant center capacity. The waiting list length is the root
cause for illegal trafficking; any other policy regarding illegal
trading should tackle the demand rather than the supply side.
Efforts should be made on the prevention of end-stage renal
failure. Furthermore, socio-economic, legal and cultural
disincentives for organ donation should be eliminated, in order
to endeavor effective organ transplant procurement.
Future research may enrich the picture by incorporating the
root causes of kidney failure and its dynamics, along with
further socioeconomic and behavioral factors. Finally, a
participatory approach with stakeholders would provide the
opportunity to reflect upon the models underlying assumptions
and discuss the insights of the model.
REFERENCES
National Institute of Diabetes and Digestive and Kidney
Diseases (NIDDK). (2015).
http: //kidney.niddk.nih.gov/KUDiseases/pubs/kustats/
Azar, A. T., McDonnell, G., & White, J. C. (2013). Modelling
and Control of Dialysis Systems. Springer, 1275-
1320.
Bankes S. 1993. Exploratory Modeling for Policy Analysis.
Operations Research 41 (3):435-449.
Becker, G. S., & Elias, J. J. (2007). Introducing Incentives in
the market for Live and Cadaveric Organ Donations.
Journal of Economic Perspectives, 21(3), 3 - 24.
Centers for Disease Control and Prevention. (2015). Data
base.
Columbia University Deparment of Surgery . (2015).
Retrieved from
y.org/cond! -and-
treatments/compatible- living. donor-kidney-
transplant.
Davis, A., Mehrotra, S., & John Friedewald, D. L. (2013).
Characteristcs of simulation model of the national
kidney yt ion system. Proceedings of the
2013 Winter Simulation Conference (pp. 2320 -
2329). R. Pasupathy, S.-H. Kim, A. Tolk, R. Hill, and
M. E. Kuhl, eds.
Erasmus School of Economics. (2014, December 30).
Orgaandonatie: Opt-in of Opt-out? Geen van beide!
Retrieved from
http://www.eur.nl/ese/nieuws/economieopinie/artikel
en/detail/article/26097/
Ferandez, L. e. (2014). A System Dynamics approach to
optimizing resource allocation in kideny dialysis
versus transplantation. SMD M Society for Medical
Decision Making, 18 - 22.
Forrester, J. (1961). Industrial dynamics. Cambridge, MA:
The M.LT. Press.
Glazer, S. (2011). Organ trafficking can the smuggling of
human organs be stopped? CQ Global Researcher.
Hirsch, G. B., McCleary, K.J., Saeed, K., & Myer, K. A.
(2012). Deceased Donor Potential for Organ
Transplantation: A System Dynamics Framework.
Annual conference of the International System
Dynamics Society. St. Gallen Switzerland.
E. (2013). International Organ Trafficking Crisis:
Solutions Addressing the Heart of the Matter. Boston
College Law Review, 1317 - 1349.
Laurentine, K. A., & Bramstedt, K. A. (2010). Too poor for
transplant: finance and insurance issues in transplant
ethics. Progress in Transplantation, 20(2), 178 - 185.
Muzaale, A. D., Massie, A. B., Mei-ChengWang,
Montgomery, R. A., McBride, M. A., Wainright, J.
L., & Segev, D. L. (n.d.). Risk of End-StageRenal
Disease Following Live Kidney Donation.
National Kidney Foundation. (2015). Retrieved from
\ttps://www.kidney.org/transplantation/beadonor.
OPTN. (2012). Annual report 2012.
Kelly,
Prasanna Devi, S., Suryaprakasa Rao, K., Krishnaswamy, S.
& Wang, S. (2010). System dynamics model for
simulation of the dynamics. Springer, 284-292.
Pruyt, E. (2013). Small System Dynamics Models - Triple
Jump towards Real-World Dynamic Complexity.
Delft: TU Delft Library.
Pruyt, E., Auping, W. L., & Kwakkel, J. H. (2015). Ebola in
West A frica: Model-BasedExploration of Social
PsychologicalEffects and Interventions. Systems
Research and Behavioral Science, 32(1), 2 - 14.
Shimazono, Y . (2007, December). The state of the
intemational organ trade: a provisional picture based
on integration of available information. Bulletin of
the World Health Organization, 85(12), pp. 901 -
980.
Sterman, J. (2000). Business Dyamics: Systems Thinking and
Modeling for a Compled World. New Y ork, NY:
Trwin/McGraw-Hill.
U.S. Department of Health & Human Services . (2015, March
11). Retrieved from.
http://www. organdonor.gov/about/data.html.
United States Census Bureau. (2009). United States Census
Bureau 2009. Retrieved from
http://www.census.gov/population/projections/files/a
nalytical-document09.pdf
United States Renal Data System. (2012). 2012 Atlas of CKD
& ESRD. Retrieved from 2012 Annual Data Report:
http://www.usrds.org/atlas12.aspx
WHO. (2004). Resolution on human organ and tissue
transplantation. Geneva: (WHA 57.18). Retrieved
from
http://www. who.int/transplantation/en/A57_R18-
en.pdf
World Health Organization. (2015). Retrieved from
http://www.who.int/bulletin/vol
Ajen/.
APPENDIX
Fig. 19: Parameters and calibration intervals used for Uncertainty Analysis
Parameter Units Value Interval Sources/remarks
Normal graft failure rate [1/Y ear] 0.289 (0.289,0.504) (OPTN, 2012)
Average response time to [Y ear] 1 (0.3,4) Assumed
illegal kidney demand
Impact of waiting list on [Dmnl] 0.0005 (0.0001, 0.001) Assumed
social pressure
Magnitude of priority waiting [Dmnl] 5e-005 (1e-006, 1e-005) Assumed
list advantage
Fraction of patients removed _[1/Y ear] 0.04 (0.01,0.1) (OPTN, 2012)
without transplants from
waiting list
Effectiveness of financial [1/Dollar] 3e-009 (1e-009, 7e-009) Assumed
compensation
Ratio that related donors [Dmnl] 0.12 (0.1,0.2) (Columbia University Deparment of Surgery
organ match , 2015)
Mortality rate of diagnosed [1/Y ear] 0.2 (0.15, 0.25) ( National Institute of Diabetes and Digestive
ESRD patient and Kidney Diseases (NIDDK), 2015)
Magnitude of influence of [1/person] 3e-007 (3e-008, 3e-006) Assumed
waiting time on health status
Magnitude of waiting list [1/person] 7e-007 (3e-007,1e-006) Assumed
influence on registration
Initial transplanted patients [Person] 172553 (172553, 185000) ( National Institute of Diabetes and Digestive
and Kidney Diseases (NIDDK), 2015)
Impact of kidney quality on. [Dmnl] 0.05 (0,0.2) Assumed
graft failure
Impact of social pressure on [Dmnl] 0.0005 (0.0001, 0.001) Assumed
living donors
Fraction of donor kidneys [Dmnl] 0.008 (0.006,0.01) (OPTN, 2012)
transplanted before taint
Rate of new adult ESRD 1/Y ear 0.0004 (0.0002, 0.0006) (United States Renal Data System, 2012)
Rate of new senior ESRD 1/Y ear 0.0016 (0.0014, 0.0018) (United States Renal Data System, 2012)
OrderSocialPressure [Dmnl] 3 (1,3,20) Assumed
OrderA veragePrice [Dmnl] 3 (1,3,20) Assumed
OrderGraftFailure [Dmnl] 3 (1,3,20) Assumed
