Presented at the 30" annual conference of the International System Dynamics Society
St. Gallen Switzerland, July 22-26, 2012
Deceased Donor Potential for Organ Transplantation: A System Dynamics Framework
By Gary B. Hirsch’, Khalid Saeed’, Karl J. McCleary’, and Kevin A. Myer®
Abstract
Organ transplantation is a lifesaving procedure for many people. However, the lack of organs
from deceased donors makes it unavailable for many additional people who need it. A
commissioned study was undertaken to estimate deceased donor potential in the US. Organ
procurement and transplantation take place in the context of a complex system of
organizations and policies. This system can both constrain and enhance the realization of
deceased donor potential. A system dynamics model is being developed to help identify how
that system’s behavior affects the availability of deceased donor organs and how particular
strategic policy options might increase the number available for transplantation. The version
described in this paper utilizes data for kidney procurement and transplantation for the entire
US. The structure and data sources for the model are described along with illustrative tests of
those strategic options.
1. Independent Consultant and Creator of Learning Environments, Wayland, Ma.
2. Department of Social Science and Policy Studies, Worcester Polytechnic Institute,
Worcester, Ma.
3. Center for Transplant System Excellence, United Network for Organ Sharing, Richmond,
Va.
Acknowledgements: We report on system dynamics models created as part of the Deceased
Donor Potential Study, a commissioned study funded by the Organ Procurement and
Transplantation Network (OPTN). The OPTN is supported by Health Resources and Services
Administration (HRSA) contract 234-2005-370011C. The models and policy analyses described
here are the responsibility of the authors alone and do not necessarily reflect the views or
policies of the Department of Health and Human Services, the OPTN, or UNOS; nor does
mention of trade names, commercial products, or organizations imply endorsement by the US
Government.
The authors would like to express appreciation to Dr. Laura Siminoff of Virginia Commonwealth
University for her help in creating and administering the stakeholder survey. They would also
like to thank Kris Wile, Leah Edwards, and John Rosendale for their assistance in gathering
information and data for the system dynamics model and Dr. Anjali Sastry and Dr. Hazhir
Rahmandad for their helpful comments.
Deceased Donor Potential for Organ Transplantation: A System Dynamics Framework
1. Introduction
According to the National Center for Chronic Disease Prevention and Health Promotion, chronic
diseases are the leading causes of death and disability in the US (CDC/NCCDPHP, 2012). Seven
out of ten (7 of 10) deaths among Americans each year are from chronic diseases. Heart
disease, cancer and stroke account for more than 50% of all deaths each year (Kung et al.,
2005). Diabetes continues to be one of single largest determinants of kidney failure, non-
traumatic lower-extremity amputations, and blindness among adults (CDC, 2008). For many
patients in the final stage of these diseases, organ failure, transplantation may be the only
option for remaining alive. Even for organ failure where there are alternatives such as dialysis
for end stage renal disease, transplantation represents a significant improvement in quality of
life and longevity. Once an infrequent event, transplantation has now evolved into everyday
procedure supported by an elaborate system that includes interaction of the following
elements: organizations seeking organ donations; patient waiting lists; people signing up at
motor vehicle and/or state registries to allow use of their organs after death; families reached
by other means who consent to the recovery of organs from of loved ones; transplant programs
specializing in a reliable and relatively safe transplant procedure; and government agencies that
regulate the system to assure efficacy, fairness and promotion of public interest in organ
allocation and transplantation.
Some of the transplants performed each year utilize organs from living donors (kidney and
liver), but the majority come from patients who are declared dead by either neurologic criteria
(brain dead) or circulatory criteria (cardiopulmonary arrest). For kidneys, the organs
transplanted in the largest numbers, there were a total of 11,042 transplants from deceased
donors and 5,771 from living donors in 2011. (OPTN, 2012). Only a small fraction of the 2.5
million deaths in the US each year occur in a manner that lends itself to retrieval of organs for
transplantation. As a result, there are many more organs needed than available and long
waiting lists of patients who would benefit from a transplant. At the end of 2011, there were a
total of 90,468 patients waiting for kidneys with 62% of those in “active” status ready to receive
atransplant. The number of transplants that can be performed is naturally limited by the
number willing to be living donors and the number of deceased donors which are, in turn,
limited by the number of deaths that take place in settings where organs can be retrieved in a
timely manner consistent with clinical requirements.
A study was undertaken in 2010 by the United Network for Organ Sharing’s Center for
Transplant System Excellence, under contract to the US Health Resources and Services
Administration’s (HRSA) Division of Transplantation, to estimate deceased donor potential for
the US and to examine ways in which this potential could be expanded. Recognizing that
transplants in the US occur in a complex system, UNOS contracted with two system dynamics
modelers to create a model of the donation and transplant system that could be used to
understand better
e how that system functions in the context of influencing estimates of deceased donation
potential, and
e how various strategic policy options could be employed to enhance the projected
availability of deceased donor organs over time.
This application to organ procurement and transplantation was also expected to benefit from
experience with extensive system dynamics work in chronic illness and health care delivery.
(Homer and Hirsch, 2006; Homer et al, 2004; Hirsch et al, 2010; Homer et al, 2010)
This paper will describe the model and how it was developed, sources of data used, results of
initial policy tests’ of those strategic options, and suggestions for work that could be done in
the future.
1.1 The Organ Procurement and Transplantation System
The system in which organ transplants occur has two key parts: organ procurement and
transplantation as shown in Figure 1.1. These components work together to determine the
numbers and types of transplants done. In the US, there are 58 Organ Procurement
Organizations (OPOs) covering defined geographical catchment areas called Donation Service
Areas (DSAs). The OPO’s primary responsibility is to work with hospitals within their service
area to identify potential donors and arrange for the efficient and safe retrieval of organs in a
timely manner. OPO hospital development and clinical procurement staff members are in
constant communication with hospitals, are notified when there is a brain death or impending
death, work with families to obtain consent for organ donation, and arrange for organ retrieval
with transplant program physicians. As indicated in Figure 1.2, OPOs effectively manage a flow
of referrals that looks like a funnel, starting with a large number of deaths and shrinking as
donors are excluded for various medical reasons. OPOs inform transplant programs of the
availability of organs, and participate in an allocation and distribution processes defined by the
Organ Procurement Transplant Network (OPTN).
Donor
Potential
Organ
Transplantation}
Procurement
_—|
Figure 1.1: Overview of donor potential, organ procurement, and transplantation system
+ Policy tests in this paper refer to the exploration of strategic options that may be considered by the transplant
network in setting strategic priorities or for planning purposes. Therefore, policy in this context does not denote
specific directives or implications for formulating policies and bylaws via traditional OPTN policy development
processes (private rulemaking) or the adoption of additional federal regulations by government agencies with
regulatory oversight like HRSA and the Centers for Medicare and Medicaid (CMS).
Population Distributed by Age
Deaths fom All
Causes
Deaths Occurring in
Hospitals Referred to
OPOs
Medically
Suitable Deaths
Selected
(Authorized)
Donors
Organs Available for
fo) Recovered Isls,
gens per ee oe Transplantion Transplant Rate
Organ Procurement Transplantation
Figure 1.2: Donor potential and organ procurement characterized as a narrowing funnel of deaths
Only 35-38% of deaths occur in hospitals (CDC/NCHS, 2010). All (hospital) deaths are referred
to the respective DSA organ procurement organization (OPO). After a thorough, rule-based
screening process, only a fraction of those referred will be further evaluated as potential
donors. Only a fraction of those referred will be deemed medically suitable and a fraction of
those will be selected as donors once OPO’s receive consent from donors’ families and/or find
the donor in a state organ donor registry. The number selected as donors together with the
number of organs per donor will determine the number of organs that are available to be used
in transplants.
Transplant programs evaluate candidates for transplantation, manage waiting lists of those
evaluated as medically acceptable candidates, perform transplants as organs become available,
and provide follow up care for those who receive transplants. Some deal with only a single
organ such as a kidney while others are able to transplant multiple types of organs. Other solid
organs that are transplanted include the liver, lungs, heart, pancreas, and intestine. Figure 1.3
shows the critical flows of people through the transplant process. There is a population of
patients who develop chronic disease leading to end stage organ failure. (A few patients
require transplants as a result of acute conditions.) Some fraction of those patients move on to
waiting lists. Some die while waiting for a transplant, some get too sick to have the surgery,
and some receive transplants. The rate at which transplants are performed depends on the
availability of organs and patients and capacity of the transplant programs. A large fraction of
transplants are successful initially, but some patients have grafts that do not survive each year
and some die despite having the transplants. Patients whose grafts do not survive may rejoin
waiting lists for re-transplantation.
Deaths =
People with
——— tnd Siege -~+2— People on }—x—— Transplanted
New Cases of Organ Failure = Waiting Lists Patients
Organ Failure Transplant
Rate’
wee Capacity
ee
ae
Oo Availabl el 5
TOONONANE S| Transplantation
for Transplantion
Organ Procurement
Figure 1.3: Flows of People Associated with Transplantation
The description of the system so far suggests that there is a straightforward flow of organs and
patients to transplantation limited only by the number of available deceased donors. However,
conversations with those in the transplant world, on both the organ procurement and
transplant sides, suggest that there are important feedbacks in the system that can constrain or
enhance the numbers of transplants performed. For example, as shown in Figure 1.4,
transplant programs decide whether organs from particular donors are acceptable for one of
their patients. The volume of transplants done may affect the average medical quality of
donors and organs from those donors which can, in turn, affect outcomes such as graft and
patient survival. Concern about organ quality and outcomes can then affect criteria for organ
acceptability. One transplant program may deem an organ from a particular donor acceptable
while another that is more risk averse will not want to accept the organ, even if it means a
longer wait for their patients. On the other hand, a more conservative approach may provide
more consistent outcomes that encourage more patients to seek transplants and also help to
maintain the survival and potential growth of the transplant program. It is important to
understand how these feedbacks modify donor potential.
Figure 1.5 presents some additional feedback loops that can constrain organ availability. One,
indicated as loop B, suggests that transplant programs have particular goals and they may
become more selective about organ acceptance once those goals are met. This loop can work
in concert with loop A to make transplant programs more risk averse once their goals have
been met. Loop C similarly suggests that OPO criteria for the types of donors they will attempt
to obtain will depend on their goals and how their performance is measured. Measurement
based on variables such as organs per donor and conversion rates may cause them to avoid
certain potential donors who are likely to yield fewer organs3or be more difficult to convert
from potential to actual donors. Past experience with transplant programs not accepting
organs from certain types of donors may also discourage them from pursuing those donors.
Outcomes: Patient
and Graft Survival Og
Average Quality of
Transplanted Organs
Organ
Acceptance Rate
Transplant Rate
Organs Available
for Transplantion
Figure 1.4: Feedback loop affecting organ acceptance through perceived quality
OPO
Measurement
OPO Efforts and
Criteria Outcomes: Patient
and Graft Survival
‘a Average Quality of
Selected Transplanted Organs
Donors c ™ Organ A
Acceptance Rate
Transplant Program
Capacity and Goals
Transplant Rate
Organs Offered
Organs Available
for Transplantion
B
Figure 1.5: Additional feedback loops constraining organ availability and acceptance
The next section describes sources of data used to quantify and validate the model.
2. Sources of Insights and Data for the Model
Forrester (1980) indicates that the causal structure of a system dynamics model should be
based in large part on behavioral information that resides in people’s experience. The behavior
of such a model depends largely on the structure that is elicited from this experience, especially
where the structure contains feedback loops that drive a system in a particular direction
(reinforcing loops) or that resist change (balancing loops). Behavior patterns produced by the
model should hold true over a wide range of different input parameters. We have made a
concerted effort to elicit this experiential information as it relates to organ procurement and
transplantation and supplement it with quantitative information that allows the model to be
validated against historical data.
2.1 Field visits and stakeholder meetings
The modeling effort began with visits to a small number of OPO’s and transplant programs to
get a basic understanding of the processes involved in organ procurement and transplantation.
These were supplemented by interviews with UNOS staff and extensive conversations to get a
good overview of organ transplantation in the US. These were supplemented by further visits
and phone conferences with additional OPOs and transplant programs to validate the
completed model and help apply it in several Donation Service Areas with different
demographic characteristics.
Additional inputs about elements to include in the model came from two stakeholder meetings
held in March, 2011 and March, 2012 that included a cross-section of 40-50 people from OPO’s
and transplant programs as well as individuals from academic institutions and government
agencies with expertise relevant to questions of donor potential. The Stakeholder Committee
consists of thought leaders and key stakeholders from both the transplant and non-transplant
community. These stakeholders, selected with input and approval from HRSA, were a diverse
group representing the following constituents: OPO leaders and procurement professionals;
transplant clinicians (surgeons and physicians), and other clinicians with expertise in critical
care, emergency medicine, palliative care, and transplant nursing; and researchers with subject
matter expertise in epidemiology, geography, public health, health economics, health services
research and statistics, and system dynamics, many of whom hold an interest in transplantation
among other health care issues. Exercises at the first meeting identified key variables. Part of
the second meeting was devoted to reviewing a draft model and identifying additional concepts
to be represented. Validation of the model by these meetings will be supplemented by the
small group sessions mentioned above.
2.2 OPTN data base
The Organ Procurement and Transplantation Network has a very rich data base of donor and
recipient characteristics including data on outcomes (graft and patient survival rates). These
data were used in the quantification of the model for many of the initial values of stocks and
fractions of patients flowing from one status to another each year. As an initial test, the model
was parameterized for kidney transplantation for the US as a whole, beginning in 2001 and
running through to 2021. The model closely approximated recorded data during the 2001 to
2011 time period. Model parameters were further refined based on more extensive analyses of
OPTN data, and supplemented with other data sources that incorporate important population,
demographic, epidemiologic and geographic factors from sources such as the US Census and
the CDC’s National Center for Health Statistics.
2.3 Stakeholder input and causal factors survey
Initial identification of variables to include in the model came at the first stakeholders meeting
in March of 2011. As indicated above, stakeholders were presented with simple templates
showing flows of patients and organs through the transplant system and asked to identify
variables that had an influence on those flows. This process led to the identification of a
number of causal factors. It did not, however, provide a sense of the relative influence of these
factors. The next step was to do a survey of the stakeholder group to discern relative influence
and identify any new factors that the earlier exercise may have missed. There were 37
responses to the causal factors survey, balanced between OPO and transplant program
respondents plus third smaller group of knowledgeable observers from transplant community.
The survey yielded a number of useful insights and support for adjusting certain model
variables as well as some new variables to consider. Analysis of responses by those in the
survey revealed some interesting differences in perception between those working in OPO’s
and those in transplant programs.
Key insights from the stakeholder survey included the following:
* Agreement that transplant capacity is not a fixed constraint and can be somewhat
flexible.
* Imports (organs supplied to a DSA from a distant DSA) and exports (organs
provided to another DSA from an originating or “local” DSA) are not a large
component of available organs in most DSA’s.
* Volumes of transplants required to meet program goals, regulatory requirements
to maintain certification by Federal agencies and private payers are not as
important in determining transplant program capacity and volume as number of
qualified surgeons and medical support available
* Differences in perceptions between OPOs and transplant programs in at least two
areas:
— Perceived importance of organs per donor available and the numbers of people
on waiting lists on transplant rates
— Effect of concern about poor outcomes on transplant program capacity and
volume in a counterintuitive direction
More extensive analyses of the data will be performed to derive more insights and value from
the survey. This effort will be pursued concurrently with other proposed model validation
activities. .
3. Detailed Model Structure
The deceased donor model is organized into several subsystems. Figure 3.1 provides an
overview that shows how these subsystems relate to each other.
Demographics affect both the donor potential and people entering the patient (transplant
candidate) queue. Potential donors become actual donors through a conversion process that
depends on the OPO’s ability to gain the consent of donors’ families if the deceased donor is
not already registered as an organ donor via various state-based donor registries. Its success is
influenced by OPO capacity and effectiveness and results in organs being available for
transplantation. OPO capacity can respond to multiple factors that affect the demand for
organs including the length of the patient queue (on waiting lists) and transplant rate. An
OPO’s revenues affect its ability to expand its own capacity to obtain organs.
Available organs and, to a lesser extent, transplant program capacity, determine the transplant
rate and the wait to receive a transplant. Available organs may not be recovered. This is often
true for organs other than kidneys. Most organs recovered are transplanted and the fraction
not transplanted are discarded or used for other purposes such as research. Average organ
quality is affected by the volume of organs accepted by transplant programs. A larger pool of
organs may imply a larger fraction with less-than-ideal characteristics. Accepting more organs
may imply less selectivity over a continuum about donors’ condition and medical history and
the condition of organs to be transplanted. Centers with greater tendencies to accept organs
may be responding to longer wait lists and waiting times and competition for organs within a
region. (Garonzik-Wang, James, Weatherspoon et al., 2012) Alternately, accepting fewer
organs may denote surgeon/clinician preferences and comfort (experience, established
protocols, etc.) in dealing with suboptimal organs in their transplant market areas. Current
evidence suggests underutilization of the organ acceptance criteria system that could make
matching of donors and recipients more uniform. Many transplant centers are thought to use
the same overly broad criteria for almost all of their waitlist registrants, causing them to
overestimate their actual use of these organs in their clinical practice. (Massie, Stewart, Dagher
et al., 2010). Organ quality has a strong influence on transplant success and, in turn, on risk
tolerance by surgeons and patients’ willingness to enroll in waiting lists for transplants.
Each of these subsystems is designed to function in a certain way. However, the interactions
among major subsystems can lead to outcomes that may not be intended by the bounded
rational decisions made in each subsystem as outlined in Morecroft (1985). Each subsystem is
described in the following sections.
OPO capacity V7
donor potential 7
demographics ‘7
act
9
quality of organs
patient queue 7
ry
| transplantation 7
organ
transplant
ystem
transplant center capacity 7
fansnnnnnonn
Figure 3.1: Subsystems in the model and their interplay
3.1 Donor potential
Figure 3.2 illustrates how donor potential is determined. It begins with a fraction of all deaths
occurring at the medical centers. The fraction currently used in the model is a placeholder
pending more accurate numbers awaited from the data subcommittee of the project. A
fraction of the potential donors become referred deaths, a fraction of which are deemed
medically suitable/eligible deaths depending on OPO screening process and expediting efforts.
10
donor potential Ags
Fraction hospital (jp >) hospill imely
timely referred 4 referred deaths
deaths .
deaths in med )«——) ft deaths in
unsigned peers med centers
population
deaths
signed
donors fr
deaths of people
with
chronic desease’.,
Figure 3.2: The donor potential sector
3.2 Conversion to actual donors
The conversion sector, shown in Figure 3.3, calculates the fraction of potential donors who will
become actual donors. It contains the multistage process outlined earlier in Figure 1.2
conversion Ag
UPUTMAncral : OPO ancl :
drivers for effort fr authorized effect of OPO thivers for efort {_) OPO financial
r medically suitable staff adequacy on condition
authorization
medically eo
normal f reered 5 Signed
effect of availability authorized ~” donors fr
on organ recovery
perceived 7
ea sisal ease ofconsent psa
; acceptance
normal donor donation suioble + rate
selection fraction vale donors ©fctofperceived acceptance
: on authorization
A authorized C
perceived 7 donor selection dence oe
donor need “--’ pressure f potental donors hospital timely
referred deaths
referred by
med centers
Figure 3.3: Conversion sector
A fraction of referred deaths are deemed potentially suitable for donation and become
authorized donors once the consent of families is obtained and/or donors are found to be
enrolled in registries. The fraction that become authorized depends on OPO staff capabilities
and incentives, OPO perceptions of donors likely to be accepted by transplant programs, and
the ease of obtaining consent which is affected by the fraction of the population signed up on
donor registries. Medically suitable authorized donors become selected donors at a rate based
on anormal fraction and a donor selection pressure that reflects perceived need.
11
3.3 Demographics
The demographic sector of the model keeps track of the general population in terms of its
distribution between healthy people who are not registered for organ donation, healthy people
who are registered, and the population afflicted with chronic disease. In the current version of
the model, the chronic population represents those with chronic kidney disease. The
subsystem representing this sector is shown in Figure 3.4. Deaths occur from each of the three
categories and their sum constitutes total deaths in the donor potential sector. Births and
immigration add to the healthy population not yet signed up on donor registries. This inflow is
assumed to be exogenous and in our tentative model is taken from the US Census data.
“Unsigned” healthy donors can sign up at motor vehicle registries and other sites to become
“signed” healthy donors. Currently, over 100 million individuals are registered as organ donors
in the U.S. with efforts underway to register an additional 20 million individuals in 2012 (Donate
Life America, 2012). Even with increasing registrations, a time lag between registration and
eventual death results in a delay before benefits are realized in the form of increased numbers
of organs available.
Both the unsigned and signed populations are disaggregated into six age groups: 0-6, 7-14, 15-
35, 36-59, 60-69 and 70+. Both groups of healthy people can become afflicted with chronic
disease, chronic kidney disease in the current version. This chronic population feeds the
patient flow that ultimately develops end-stage renal disease (ESRD) and requires kidney
transplants. (A small fraction of patients requiring transplants may come directly from the
healthy population as a result of an acute illness or injury.) The model can accommodate other
types of organs and transplant procedures either by creating arrays or creating separate models
as the one presented here for kidneys.
12
demographics Agi
word of ft deaths for people
mouth «. with chronic disease
insigned healthy pop
deaths of people
unsigned wih
f aficton chronic desease
ff wih ESCD
time trend
s
a
S-)
&
5,0
&
a
Q developed |
f healthy signed § i J renal ESRD
donors dying ae eee “in
age related prevalence
of chronic disease
population with
renal disease
Signed populaiof increase
Ey
‘Signed Immigraton
ft of chronically
affliction rate afflicted population
ft afficton rate with renal disease
Total signed donors
Figure 3.4: The demographic sector
3.4 Patient flow
Figure 3.5 shows the subsystem representing the flow of patients to waiting lists for transplants
and transplant procedures. New cases of ESRD (in the current version) flow into the stock of
people with ESRD. Patients removed from the waitlist are added to this stock, which is
depleted by deaths and referrals to waitlist. The longer patients have to wait for a transplant,
the more likely they are to develop other conditions (co-morbidities) that lead to their deaths
or removal from the wait lists. Wait lists are fed by new people entering the wait lists as well as
by graft failures. Patient decisions to join the wait list may be affected by expected wait and
likelihood of a successful outcome or may simply reflect the difficulties and poor outcomes of
remaining on dialysis. The model focuses on the active wait lists since these are the people
eligible to receive transplants. Wait lists are depleted by transplant rate and deaths of patients
waiting for transplants. Patients who have received successful transplants enter a stock of
transplanted patients that is depleted by deaths and cases of graft failure. Graft failure rates
can increase if acceptance of a larger fraction of potential donors causes more organs to be
transplanted from donors with less-than-ideal characteristics.
13
patient queue
organs for normal patient wait
transplant before transplant
waitbefore
referral
towaitist actual wait
entry to waits inane
ff waitisted dying
@) acceptance
efectof rate
acceptance rate on
graf failure
normal f removed
from waitist patents
fr death rate
of diagnosed patients
waitisted
patient
deaths OQ normal f transplanted
with failed graft
index
normal fr
sentio waitist
fraction of
transplanted dying
perceived *f
relative wait ad successful
transplants
normal transplants per
renal ESRD J unit transplant capacity
population with
al
patient waitrisk tolerance normal patent wait Patient renal disease
motivation bebe ranspiant |= monvaten
i = # transplant
yraft failure i
g lili acceptance pressure i + center
from patient queue capacity
Figure 3.5: The patient flow sector
3.5 Transplantation
The transplant rate shown in Figure 3.6 depends both on the supply of organs for transplant,
and, to a much lesser extent on the capacity of transplant programs. Capacity can be pushed a
bit if organs are available, but will be underutilized when organ supply is limited. In the case of
kidney transplants, organ supply is determined both by organs recovered from deceased
donors and those from living donors. Living donors may be motivated to come forward in
greater numbers if the average wait for deceased donor organs is longer.
14
ce) transplantation Ad
perceived
relative . .
wait impact of pekceived
relative wait on live\donations
accepted organs for
organs transplant
ff transplants living donor
successful organs
transplant
successful pressure
transplants poner transplant
center
capacity
'@)' capacity effect
trankplantrate on acceptance
average
transplant rate transplants
averaging time per year
Figure 3.6: The transplant rate
3.6 Organ Recovery
The organ recovery sector, shown in Figure 3.7, addresses the process that starts with selected
(authorized) donors and determines the number of organs available for transplantation. A
fraction of medically suitable donors are selected based on acceptability criteria utilized by the
OPOs, the perception of need based on length of the wait list, and an average volume
transplants based on past experience. OPOs will obtain consent for as many donors as possible
once potential donors are identified. Then the OPO gets more information on the donor and
offers organs to transplant programs based on wait list rank order (i.e., which patients have the
highest priority) Transplant centers accept or decline the “offer” based on their assessment of
the suitability of each organ for the intended recipient.
The normal rate of organs recovered per donor, known as “yield” (1.7 for kidneys, about 3 for
all organs), is modified by the perceived availability of organs. Lower perceived availability of
organs will create pressure to recover more organs per donor. The likelihood of organs being
accepted by transplant programs and volume of organs accepted are affected by the length of
waitlist and quality of the organs being offered, and, to a lesser extent, transplant program
15
capacity. A lengthy waitlist will make it more likely that organs will be accepted and that OPOs
will place more organs. Poor quality will limit the fraction of offered organs that become
accepted organs.
organ recovery as
capacity effect
on acceptance from patient queue effect of organ quality
on acceptance
acceptance
pressure from
( acceptance
“rate
normal
organ selector
recovered
organs
transplants
haar organs recovered
Pe ¢ donor
normal organs
per donor —
perceived
acceptance
perceived rate
donor need change in perceived perception
organ availability fime
perceived
organ availability
availabilty "gan
perception availability
delay
Figure 3.7 Organ Recovery Sector
3.7 Transplant Center Capacity
Transplant center capacity has been described to us as a flexible factor that can be adjusted as
needed in response to the availability of organs. However, over the long term, it can be an
important factor that ultimately affects the number of transplants. The capacity adjustment
process is therefore an important part of the dynamics we are trying to understand. Transplant
capacity has proven to be a hard concept to visualize. We have expressed it as the ability to
16
deliver a certain volume of transplantation and have represented the factors that impinge upon
the decision to change capacity over time. Figure 3.8 shows the transplant center capacity
sector of the model.
me) transplant center capacity Ag
ful
transplant center irwenesil
financial condi peryear — ‘ansplantrate
transplant center henspient preade. C a Aa
udget consraint bes transplants
per year
effect of fansplant center
financials on transplant capacity
B change in av
ig
‘ransplgnt ce! transplant Yansplent center trensplantrate
expenses ier revenues
financial Yegources deshed gransplant
i transplant — change in transplant inlas bares a
unit capacity lee cents cgpacity per transplant
intenance cost ree] al center capacity
capacity oe
O)_tansplent
transplant center ‘transplant center capacity goal
sired financial current budget |
ures waits!
4 = capacity
time to adjust goal i ;
transplant center Reepanicepecy})preinaton boas)
average budget O
need rnomal patient wait
pases ae orientation inward before transplant
9 orientation
Figure 3.8: Transplant center capacity
Transplant center capacity adjusts toward any of three targets that can be chosen by the user
of the model: 1) a desired value that is determined by past performance which is determined
both by the past capacity and organ availability, 2) the transplant need created by the waitlist,
or 3) exogenously determined targets. In all cases, this goal is modulated by financial
considerations that reflect past performance as well as the revenues derived from
transplantation and the costs of maintaining transplant capacity.
3.8 OPO capacity
OPO capacity affects donor selection and organ recovery. It is expressed in the model in terms
of OPO staff and is assumed to adjust towards a desired value through recruitment and attrition
processes. The OPO capacity sector is shown in Figure 3.9. The desired number of staff is
determined primarily by the DSA size and level of activity in terms of number of hospitals in its
DSA, numbers of potential organ donor referrals, other activities such as tissue and eye
donation (not addressed in this project), and geographic size of the DSA. OPO capacity can
respond to organ need and is further modulated by financial considerations. Rising or falling
revenues as a result of changes in the volume of organs procured can lead to adjustments in
OPO capacity.
17
wo OPO capacity ag
av expenses OPO
per staff expenses av length of
OPO staff OPO employment
effect of OPO adequacy
staff adequacy on (_} ¢
donor selection
staff per 3
needed ‘OPO staff
organ attrifions
effect of OPO 5
setae meee
OPO financial on aditions
drivers for hiring =e
+
o>) normal patient wait fi
PO financial before transplant J issue revenue
O desirea gonaiton living donor revenue per
recovered organ
aah organs rg)
average
ee ae change in tesplats
OPO budget OPO expense. | per year
538 85
#
esau OPO recovered
PO budget financial organs
resources
budget OPO
averaging expenses kidney revenues as jennie an
time ‘roftotal Renee on
Figure: 3.9 OPO capacity
3.9 Quality of organs
The average quality of organs recovered and implanted depends on a number of factors as
shown in Figure 3.10. A larger number of consented donors allows matches with recipients to
be made more readily and allows organs to be recovered and placed more quickly, helping to
assure higher quality. On the other hand, a higher donation rate may imply widening donor
selection criteria that may result in diminished quality. Similarly, an increase in organs
recovered per donor can also imply lower average quality if it results from loosening the organ
recovery criteria. Higher organ yield could result without lowering quality if a larger fraction of
donors are younger and healthier. A perception of quality is a key determinant of organs
accepted for transplant. Recovered organs not transplanted yield discards, which have been
rising along with the transplantation activity. Finally, the quality of the transplanted organs
affects the potential for graft failure and mortality rates of transplanted patients and, as
indicated above, risk tolerance of patients deciding whether to enter waiting lists.
18
i) quality of organs aa
donation
ease of consent rate _ffectof orgen quality
fom ro on transplant program
y ; nomal fraction of
organs transplanted dying
per donor “
normal organs organ quality fraction of
per donor transplanted dying
Figure 3.10 Quality of Organs
The next section describes the process of model validation and some initial strategic policy
experiments.
4. Model Validation and Policy Analyses
4.1 Validation
There are a number of methods for validating a system dynamics model. (Sterman, 2000) One
is reviewing the structure with people who are familiar with the real-world system and
ascertaining that it accurately represents the underlying causal structure responsible for a
system’s behavior of interest. As mentioned earlier, the model has been presented and
critiqued at two meetings of stakeholders representing a good cross section of the organ
procurement and transplant community. We have also had additional meetings to go deeper
and review the model in detail with smaller groups of stakeholders from the field. These were
meetings with staffs of OPO’s and a transplant programs separately and one meeting with both
represented. These were interesting, not only for model validation, but to see differences in
perceptions between OPO’s and transplant programs. It was possible to arrive at a shared
sense of how the system works.
Starting the model at an earlier point in time and comparing results to actual historical data is
another approach to validation. Data from the OPTN data base was used to evaluate the
model in this manner. The model was parameterized to represent kidney transplantation for
the US as a whole with initial values from 2001. Figures 4.1 and 4.2 show results of a baseline
simulation of the model with time zero corresponding to 2001 and the (annual) kidney
transplant rate and size of the active waiting list for kidney transplants (number of people
waiting) plotted against their historical values taken from the OPTN data base. The transplant
rate includes those done with both living and deceased donor organs. (By transplant rate, we
mean the number of transplants done per year which is different from how the transplant
community uses the term transplant rate.) Simulations over the 2001-2009 time period suggest
that the model tracks these historical values well in terms of similar growth observed over the
2001 to 2011 time period. The simulation does not reflect the relative change or increase that
19
occurred in the transplant rate around 2005-2006 which occurred as a result of a HRSA
sponsored Institute for Healthcare Improvement (IHI)-based national collaborative for
performance improvement effort to improve system performance from earlier efforts of that
initiative (Howard, Siminoff, McBride, and Lin, 2007). However, the model does duplicate the
longer term change in the real-world transplant rate that occurred over the period 2001-2009.
The size of the active waiting list is a stock variable that is affected by a number of inflows and
outflows. The size of the waiting list produced by the model tracks its historical value over the
2001-2011 time period.
® transplant rate Historical Kidney Transplant Rate
4] 20000 sssacn aa
2
= 150004
1:
3] 10000 r
2001.00 2003.50 2006.00 2008.50 2011.00
Page 1 Years 1:07 PM Tue, Aug 14, 2012
3 aF ? Kidney Transplant Rate vs Historical Rate
Figure 4.1: Kidney Transplant Rate vs. Historical Values
® © vaitist Historical Kidney Waiting List
4 790009 2s = ,
2
" 700004 E A
2001.00 2003.50 2006.00 2008.50 2011.00
Page 1 Years 1:07 PM Tue, Aug 14, 2012
3 aF id Kidney Waiting List vs Historical List
Figure 4.2: Kidney Waiting List vs. Historical Values
20
4.2 The Baseline Simulation
Results from the baseline simulation are shown in Figure 4.3. The simulation starts in 2001 and
goes to 2021. The general trends that persist in the 2011-2021 time period highlight the
concerns of the stakeholders of the system. The transplant rate (line 1-blue) is not rising in
proportion to the wait list (line 2-red), hence wait lists are expanding. (Note that wait list is on
a larger scale, 40,000 to 160,000, than the transplant rate, 10,000 to 20,000.) Also, while
organs recovered have increased over the past decade, the rate of increase has tapered off
and the gap between organs recovered and transplanted has widened, indicating that discard
rates will rise (line 4-green) in the face of declining average organ quality (line 3-pink). Organ
quality is an index that varies around a value of one and reflects the fraction of donors selected
from potentially suitable donors compared to the initial fraction selected. Having to select
more donors to support a higher number of transplants will result in lower average quality if it
means drawing on less-than-ideal donors.
&® i: teanspiant rate
200004
160000
1
7500
: waitlist 3; organ quality 4; discarded organs
iy
2:
3:
4
ayn
10000
40000
ayn
3
2500: 1
2001.00 2006.00 2011.00 2016.00 2021.00
Page 1 Years 1:24PM. Tue, Aug 14, 2012
aaF ? Baseline Simulation
Figure 4.3: Results of Baseline Simulation
Why is the growth in the transplant rate (line1) so constrained? Why would more organs be
discarded (line 4) even as the waiting lists are growing? The causal factor diagrams in Figures
1.4 and 1.5 indicated that the underlying system contains a number of balancing feedbacks
driven by performance and quality concerns that constrain growth.
Several loops are present to constrain growth. For example, transplant programs with long
waiting lists may accept more organs and do more transplants, but may have to do so by
expanding their criteria and accepting lower quality organs on average. Lower quality leads to
the balancing feedback (red loop) through shorter graft and patient lifespans and reduced risk
tolerance by the programs that result in reduced acceptance of organs and fewer transplants.
Figure 1.5 suggested that there are other balancing loops that involve the OPO’s. OPOs may
become more selective in the donors they pursue once they have generated enough revenue to
cover their expense budgets and meet the demand of transplant programs within their
21
Donation Service Areas (DSA’s). Similarly, OPO’s may respond to transplant programs
acceptance or rejection by adjusting their own criteria for the types of donors that are
acceptable and thereby limit the number of donors they pursue. The net effect of these
balancing loops is to constrain growth and maintain an equilibrium level of transplants that may
be somewhat below what could be achieved without those constraints (true donor potential).
The next section illustrates how the model can be used to assess how various strategic policy
changes would relax those constraints and move the US organ procurement and transplant
system closer to achieving that true donor potential.
4.3 Policy Analyses with the Model: Exploring Strategic Options in the Transplant Network
As stated earlier at the beginning of the paper, initial policy analyses done with the model
represent strategic options and are only illustrative, pending further model development.
However, these strategic options provide a good sense of the kinds of policies the model can
help to evaluate. Results of these initial analyses are presented in the following sections.
Results graphs show what would have happened relative to the baseline simulation if the
particular options had been in place.
4.3.1 Increasing Transplant Program Capacity
One way of increasing the number of transplants is to increase the capacity of transplant
programs. Figure 4.4 shows what the effects would have been if transplant capacity is based on
an exogenous goal for a 33% capacity increase. There is a small improvement in the transplant
rate (red line) as the transplant programs are willing to accept a wider range of organs in order
to utilize their additional capacity. However, limits on organs available constrain the growth.
Eventually, financial pressures on transplant programs created by having unutilized capacity
cause them to reduce capacity.
@ successful transplants per year: 1-2-
1 11000 pene es ieageee
1: 15500
1; 14000
2001.00 2006.00 2011.00 2016.00 2021.00
Page 1 Years 1:00PM Wed, Jul 11, 2012
Nee ? Untitled
Figure 4.4 Transplant rate as affected by policy to adjust transplant capacity
22
4.3.2 Increasing Sign-Up Rates on Donor Registries
In dynamic feedback systems, growth can be promoted by strengthening reinforcing loops that
promote growth or weakening balancing loops that constrain it. One of the reinforcing loops
goes through donor registries where more people signing up would result in a greater number
of consented donors, more transplants, greater awareness of the value of transplants, and
more people signing up at donor registries. Another intervention considered was to step up
efforts for people to sign up in registries as willing to be organ donors. Figure 4.5 compares the
baseline transplant rate of successful transplants with an intervention that increases the
population signed up on registries by 40% by 2021. Graph 1 (blue) shows base line behavior
and graph 2 (red) shows behavior with the intervention to double organ donor sign up rate.
@ successful transplants per year: 1-2 -
1 20000
al: 150004
1: 10000
2001.00 2006.00 2011.00 2016.00 2021.00
Page 1 Years 9:01 PM Tue, Jul 10, 2012
NeeF ? Untitled
Figure 4.5 Transplant rate as influenced by aggressive campaigns to sign up organ donors.
The impact is at best marginal due to the balancing feedback control processes represented in
Figures 1.4 and 1.5. While signing up more people on donor registries is always worth doing,
the results shown in Figure 4.6 suggest that these campaigns can have only a limited impact on
transplant rates. The limited impact can also be explained by where in the system the higher
sign up rate has its impact. Improving the consent rate comes at a point when the “funnel” of
deaths down to potential donors has already narrowed significantly and improvements due to a
higher sign up rate can produce only small increases in selected donors.
4.3.3 Expanding Entries to Waiting Lists
The other reinforcing loop is one through entries to waiting lists and demand for organs, more
accepted organs, more transplants and greater word of mouth about the potential benefits of
transplantation, and increased entry to waiting lists. The next option to try is one that
increases the number of people entering waiting lists by 50%. Figure 4.6 shows the effect of
this policy (red line) compared to the baseline simulation (blue line). (Note the expansion in
graph scale to 12,000-24,000.) It has a somewhat greater effect than strengthening the growth
loop through sign ups in registries because the larger waiting lists and longer wait times that
result create a greater pressure to accept more organs that permit more transplants to be
23
done. The waiting list is 32% longer at the end of this simulation compared to the baseline.
This pressure helps to partially overcome the resistance of the balancing loop through organ
quality and acceptance of organs. Larger numbers of people on waiting lists also increase the
likelihood that a match will be found for organs that are offered to transplant programs, leading
to fewer discarded organs. This effect, however, is only temporary and the rate of successful
transplants reverts to what is achieved in the baseline simulation.
@ successful transplants per year:
z 24000
1: 180007"
a 12000:
NaeF ?
Figure 4.6 Effects of increasing entries to waiting lists
4.3.4 Increasing Timeliness and Expanding Range of Referrals from Hospitals
A different approach would be to increase referrals from hospitals to OPOs. While hospitals are
required by law to make referrals, this could be accomplished by a combination of motivating
hospitals to make referrals of patient deaths in a more timely manner and expanding the
criteria by which people are considered potential donors or narrowing exclusionary criteria that
would keep them from being considered as donors. OPO’s can also make greater use of donors
declared dead by circulatory criteria (DCD), something most of their counterparts are already
doing. The graph in Figure 4.7 shows the effect of a 20% increase in timely referrals in
simulation number 2 (red line) compared to the baseline simulation (blue line). The change
produces an increase in the transplant rate despite the “push back” from the balancing loops in
the system. Eventually, the increase in transplants levels off.
24
@ successful transplants per yea
1 24000
1: 18000:
ar 12000
NaeF ?
Figure 4.7: Effect of higher rate of referrals to OPOs
4.3.5 Increasing Acceptance Rates of Organs from Less-Than-Optimal Donors
As indicated earlier, weakening balancing loops that constrain growth is another way to
promote growth. The balancing loops through organ quality and accepted organs area
significant constraint on growth. Quality in the real-world system is, to some extent, perceived
quality based on what is known about a donor when the decision is being made to accept an
organ. Donor age and health (e.g., presence of chronic conditions) will influence the decision.
The next option to be considered is one in which this constraint on perceived quality is relaxed,
making it possible to recover more organs from the same stream of potential donors. This
might be achieved by being less conservative about accepting organs from less-than-ideal
donors rather than actually sacrificing quality and using organs that would result in a higher
graft failure rate. As shown in Figure 4.8, weakening this constraint can result in a significant
number of additional transplants. New technologies for organ preservation may also help to
improve the range of organs that can be acceptable. Average quality goes down as a result of
relaxing this constraint and graft failures increase as a result, but more people are transplanted
than would have been if this constraint remained at its original strength.
25
@ successful transplants per yea!
1 24000
a 18000:
L 12000:
2001.00 2006.00 2011.00 2016.00 2021.00
Page 1 Years 7:03 PM Tue, Aug 14, 2012
Nee ? Untitled
Figure 4.8 Effect of relaxing (perceived) quality constraints on organ acceptance
4.3.6 Combined Strategies
One lesson learned from working with system dynamics models is that combined strategies
focused on different parts of a system will yield better results than strategies with a single
focus, no matter how much in the resources are devoted to that single strategy. Figure 4.9
shows the effects of combining the two previous strategies to expand referrals from hospitals
and relax quality constraints on organ acceptance.
@ successful transplants per year:
1: 24000
Tt 18000
1G 12000:
2001.00 2006.00 2011.00 2016.00 2021.00
Page 1 Years 7:05PM Tue, Aug 14, 2012
Nae ? Untitled
Figure 4.9: Results of combined strategy with increased referral from hospitals
and relaxed quality constraints on organ acceptance.
For comparison purposes, Line 2 (red) in Figure 4.10 represents the relaxed quality constraint
policy alone while Line 3 (pink) represents the combined strategy. The combination of an
increased referral flow and greater flexibility in which donors are accepted could produce an
additional number of transplants in the future.
26
Conclusions and Further Development
This paper has described a model of the US organ procurement and transplantation system
along with illustrations of how the model can be used for policy analysis. It provides a context
for understanding how various strategic policy options and other system characteristics can
constrain or enhance donor potential; and, in particular, how particular policies can affect
donor potential. Further development of the model is proceeding in several directions. The
structure itself is being extended in order to more closely tie the model to mortality data
coming out of other parts of the larger deceased donor potential study. We have already
developed the demographic sector further to include age cohorts and other demographic and
epidemiologic data that will drive mortality rates and donor potential. We will also be applying
the model at the DSA level, working with several OPO’s and transplant programs to validate the
model and possibly adapt the model for their use as a decision support tool.
27
References
Centers for Disease Control and Prevention, Chronic Diseases and Health Promotion Web Page:
http://www.cdc.gov/chronicdisease/overview/index.htm. Accessed, May 2012.
Centers for Disease Control and Prevention. National diabetes fact sheet, 2007. Atlanta, GA:
U.S. Department of Health and Human Services; 2008. Available from:
http://www.cdc.gov/Diabetes/pubs/factsheetO7.htm
Centers for Disease Control/National Center for Health Statistics, Health US 2010, Figure 33, P.
43
Donate Life America, 2012, Information available at www.donatelife.net
Forrester, J W. 1980. Information Sources for Modeling the National Economy, Journal of the
American Statistical Association, Vol. 75, No. 371 , 555-566.
Garonzik-Wang, JM, James, NT, Weatherspoon, KC, Deshpande, NA, Berger, JA, Hall, EC,
Montgomery, RA, and Segev, DL. 2012. The Aggressive Phenotype: Center-Level Patterns in the
Utilization of Suboptimal Kidneys. American Journal of Transplantation, Vol. 12, Issue 2, 400-
408.
Hirsch G., Homer J., Evans E., Zielinski A, A System Dynamics Model for Planning Cardiovascular
Disease Interventions, Am J Pub Health, 2010, 100: 616-622.
Homer JB, Hirsch GB, Minniti M, Pierson, M, Models for collaboration: how system dynamics
helped a community organize cost-effective care for chronic illness. Sys Dyn Rev, 2004, 20(3):
199-222
Homer JB, Hirsch GB. System dynamics modeling for public health: background and
opportunities. Am J Pub Health 2006; 96(3):452-58.
Homer J, Milstein B, Wile K, et al. Simulating and evaluating local interventions to improve
cardiovascular health. Preventing Chronic Disease 2010; 7(1).
Howard, DH, Siminoff, LA, McBride, V, and Lin, M. 2007. Does Quality Improvement Work?
Evaluation of the Organ Donation Breakthrough Collaborative. Health Services Research
Vol. 42, No. 6 (Part I), 2160-2173.
Kung HC, Hoyert DL, Xu JQ, Murphy SL. Deaths: final data for 2005. National Vital Statistics
Reports 2008; 56(10). Available from:
http://www.cdc.gov/nchs/data/nvsr/nvsr56/nvsr56_10.pdf.
Massie, AB, Stewart, DE, Dagher, NN, Montgomery, RA, Desai, NM, and Segev, DL. 2010.
Center-Level Patterns of Indicated Willingness to and Actual Acceptance of Marginal Kidneys.
American Journal of Transplantation, Vol. 10, Issue 11, 2472-2480.
Morecroft, JDW. 1985. Rationality in the analysis of Behavioral Simulation Models.
Management Science, V
28
Organ Procurement and Transplantation Network 2012, received via personal communications
from Leah Edwards and John Rosendale of UNOS during June and July, 2012.
Sterman, JD Business Dynamics: systems thinking and modeling for a complex world, 2000,
Boston: Irwin McGraw-Hill, Chapter 21
29)
