Rezulin: how effective and efficient diabetes treatment?
Nicholas C. GEORGANTZAS, Kelley TRILLING, and Yun CHOU
Management Systems ¢ GBA
Fordham University at Lincoln Center
113 West 60th Street ¢ Suite LL617-D
New York, NY 10023-7471, USA
Abstract: System dynamics simulation is used to assess the dynamic effects
of using the new drug «Rezulin» for the intensive treatment of diabetes, a
most prevalent disease among patients of the Veterans Integrated Service
Network (VISN 3) in the New York City metropolitan area. According to
the Brooklyn VA Medical Center data, one in five of its outpatient visits is
associated with diabetes and approximately 10,000 of the VISN 3 veterans
are diabetic. Although the effects of diabetes on the development and
progression of long-term complications have been established, about 90
percent of the VISN 3 diabetic veterans do not seek treatment and thereby
create a tremendous cost because of complications. These are mostly type 2
diabetes patients who prefer to defer insulin injection therapy for as long as
possible because of the inconvenience, the cost, and the pain of constant
blood monitoring and insulin injections. Those who seek treatment, diet,
exercise, and use oral drugs or insulin therapy, generally have less costly
complications, but there is still a high cost associated with their treatment
and, more importantly, they can also become resistant to insulin. Because
Rezulin directly treats insulin resistance, and it can also lower the HbA1c
level of patients (a clinical marker of disease severity assessed via a blood
test), this new drug is expected to increase the life time of diabetics by
more than 1.5 times. Our model incorporates real-life policy parameters
and NIH statistics in order to assess the potential effectiveness and
monetary benefits of switching from insulin to the Rezulin oral treatment.
Introduction
Diabetes is a deadly, insidious disease that consumes one out of every
seven health-care dollars spent in this country. It affects 16 million
Americans, only half of which are officially diagnosed. Among diagnosed
patients, insulin-dependent diabetes affects 8 to 10 percent of them, usually
under 30 years of age. It is a chronic condition caused by low or no insulin
production by the pancreas. In order to survive, patients must self-
administer insulin injections once or several times a day.
Among all diagnosed and undiagnosed diabetics, most suffer from
type 2 diabetes or adult-onset Non-Insulin Dependent Diabetes Mellitus
(NIDDM). The pancreas does produces insulin under this condition but,
over time, human cells fail to utilize the body’s insulin production to lower
blood sugar. They become «insulin resistant». Type 2 diabetes patients can
be treated with diet, exercise, drug and sometimes insulin therapy. Most
patients prefer to defer insulin-injection therapy as long as possible,
however, because of the inconvenience, cost, and pain caused by constant
blood monitoring and insulin injections.
Those suffering from diabetes also suffer a serious multitude of long-
term complications. Heart disease is probably the biggest killer. The death
fraction of middle-aged diabetes patients is twice as high and their heart-
disease death fraction two to four times higher than non-diabetic middle-
aged persons. Stroke risk is 2.5 times higher for persons with diabetes than
for those without, and high blood pressure affects 60 to 65 percent of
diabetes patients.
Blindness is another terrible complication. Diabetes is the leading cause
of new cases of blindness among adults 20 to 74 years old. Kidney disease
is yet another complication, where most diabetics suffer from end-stage
renal disease, and thereby must have kidney dialysis and transplantation.
Diabetes patients also suffer from mild to severe nerve damage that often
leads to amputations. More than half of lower limp amputations are
associated with nerve damage caused by diabetes.
In addition to suffering its complications, diabetics suffer financially
too because of the costs associated with the disease. First and foremost is
the direct medical cost of treating the complications. Second is the indirect
cost of disability, work loss, and premature mortality. Diabetes is the
fourth leading cause of death by disease.
The morbidity, mortality, and cost of diabetes can be greatly reduced,
however, if patients undergo treatment. This essay describes a system
dynamics simulation model which shows how therapy with a novel new
drug called Rezulin can profoundly improve the well-being of diabetics.
Rezulin is a new oral-antidiabetic drug which works directly on insulin
resistance and, thereby gets to the root of the disease process.
Model Description
Specifically, we looked at a patient population where diabetes is a
prevalent disease: the Veterans Integrated Service Network (VISN 3) in the
New York City metropolitan area. One in five outpatient visits at VISN 3 is
associated with diabetes. Although about 10,000 diabetic veterans belong
to VISN 3 (DVs in VISN 3, Fig. 1), because many of them are not in
treatment, they pose a tremendous financial burden with a 15-year life
expectancy (years to death from diagnosis). The diabetic veterans on the
other hand who do seek treatment (DVs in Treatment, Fig. 1), diet and
exercise counseling, and other drug or insulin therapy experience less
costly complications and live longer (T gain, Fig. 1). There is a cost (T cost,
Fig. 1), however, to that treatment.
Fig. 1
Veterans Integrated Service Network (VISN 3) diabetes management model
annual revenue annual cost
Cash
funding fr
complications cost per DV
annual gov't funds
death in treatment
deathonRezulin avg lifeonR
The new drug Rezulin works directly at treating the insulin resistance
that human cells develop through time and lowers diabetics’ HbA1c level
(a clinical marker of disease severity determined via blood test). Therefore,
those on Rezulin should live a 1.5 times longer (R gain, Fig. 1). This value
(s.a. Appendix for model parameters and equations) was derived from the
Diabetes Control and Complications Trial (DCCT), which showed that
lowering diabetics’ blood glucose and HbA1c reduces the microvascular
and neuropathic complications risk. Although the trial was performed with
type 1 diabetics, medical researchers feel that the results apply to type 2
diabetics too. The model also incorporates government's funding for VISN
3 ($1,009,714,793), with 15 percent ($151,457,218 per NIH) spent on
diabetes.
Simulation Results
In order to see what the implications of the VISN 3 situation and data
are, we run simulations from zero to twenty years, with a computation
interval dt=0.125, using the Runge-Kutta 4 integration method. Initializing
at steady state produced the results of Fig. 2. Although the number of
diabetic veterans (DVs) in VISN 3 remains constant throughout (line #1,
Fig. 2a), when the dummy variable of Fig. 1 changes from zero to one after
year 5, diabetic veterans in treatment migrate (line #2, Fig. 2a), and thereby
DVs on Rezulin rise (line #3, Fig. 2a).
Even in this ideal steady-state run, diabetics who seek no treatment die
off quickly because of complications (line #1: avg life, Fig. 2b), while those
in treatment live longer (line #2) and those on Rezulin live the longest (line
#3). These morbidity-defying results assume that a constant 50 percent of
DVs migrate from conventional to Rezulin drug therapy. Figure 2c,
however, shows how the annual cost improves markedly when the DV
migration fraction (fr) to Rezulin rises from 25 to 50 to 75 percent. The
increased life span of diabetic veterans on Rezulin reduces complications
cost, which in turn causes annual cost to decline.
Figure 3 shows the results of a more pragmatic current-state system
initialization. Again, the Rezulin option does not exist before year 5
(dummy=0), so the number of DVs in VISN 3 (line #1, Fig. 3a)) declines as
they seek treatment, thereby causing the number of DVs in Treatment to
rise (line #2, Fig. 3a). Once Rezulin comes into play after year 5 (dummy=1),
however, the new drug wins diabetic veterans over (line #3, Fig. 3a),
causing the number of those in conventional treatment to decline (line #2,
Fig. 3a). To get a better feel of what might transpire in reality, life expectancy
is permitted to vary, hence the jagged graphs of Fig. 3. This pragmatic run
shows Rezulin to be an effective and efficient diabetes treatment: effective
because it extends the average life of diabetic veterans (Fig. 3b), and efficient
because it lowers the network’s annual cost (Fig. 3c).
Although the life expectancy variability is removed to smooth its graphs,
Fig. 4 moves on to a most pragmatic run with Rezulin being available now,
Fig. 2
Simulation output time-series graphs with steady-state initialization
(a) 1500000] 1:DVs in VISN 3
2: DVs in Treatment
3: DVs on Rezulin
+-—2- ———
7500.00 4 \ a
<
Ly. 7! 72: ‘1 2-
0.00 3 f : ; ‘Years, 1
0.00 5.00 10.00 15.00 20.00
(b) 40.00] 1: avg life
2: avg life in T
3: avg life onR
| 3 3
20.00
1 1 1 1
0.00 Years,
0.00 5.00 20.00
(c) 3.50e+08'] annual cost
1—2—3
\,
2.00e+08 4
2
J oe
AY ‘1
cae 23
5.00e+07 1 : : Years, 1
0.00 5.00 10,00 13,00 20,00
at t=0, not 5 years from now. As diabetic veterans who seek no treatment
(line #1, Fig. 4a) begin to do so, and those in treatment migrate to Rezulin
Fig. 3
Simulation output time-series graphs with current-state initialization and variable dummy = 0, 1
(a) 15000.00 7] 1: TVs in VISN 3.
2; DVs in Treatment to new drug fr=0.50
3: DVs on Rezulin
7500.00 4
Z.
NN _
0.00 3 : ' : Years. :
0.00 5.00 10.00 15.00 20.00
(b) 40.00 ] liavg life
2: avg life in T
3: avg life onR i
S)
3
4
20.00 44 ¥12
0.00 3
0.00
(c) 3.50e+08] annual cost
7 i 5
f
2.00e+08 4 /
|i
Y 3 OM anyon,
5.00e+07 T Y T ‘Years
therapy, DVs in Treatment initially rise but decline quickly because some
migrate to the new Rezulin drug (line #2, Fig. 4a). Their migration in turn
Fig. 4
Simulation output time-series graphs with current-state initialization and constant to new drug fr = 0.50
(a) 15000.00'] 1: DVs in VISN 3
2: DVs in Treatment
3: DVs on Rezulin
7500.00 4
= ‘I—
0.00 1 1 1 Ment 1
0.00 5.00 10.00 15.00 20.00
(b) 40.00 ] I: avg life
2: avg life in T
3: avg life onR
3 3
20.00 2
rl ‘1: ‘L ‘1:
0.00 iu
0.00 5.00 15.00 20.00
(c¢) 3.50e+08 ] annual cost
2.00e+08 4
5,00e+07
0.00 5.00 10,00 15.00 20.00
causes DVs on Rezulin to increase at a declining rate (line #3, Fig. 4a). Fig.
4b again shows the effectiveness of Rezulin therapy, attributed to the new
Fig. 5
Simulation output time-series cosi and Cash graphs with current-state initialization
(a) 1,00e+08 7] 1: complications cost
2: T cost
3:R cost anny
t 1 to new drug fr
5.00e+07 4
a |
4 2s,
3 3
3
0.0 —_—— : : ——Years___,
0,00 5.00 10.00 15.00 20.00
(b) 1.28e+09 ]
3.91e+08 4
-5.00e | 08
3.916+08 4
-5.00e+08
0.00 500 1000 1500 20.0
drug’s ability to prolong the diabetic veterans’ life by lowering their blood
glucose and HbA1c level. Concerning Rezulin’s efficiency, the shaded area
of Fig. 4c shows the potential annual cost benefit VISN 3 can enjoy now that
this new oral-antidiabetic drug therapy has become available. Without the
new drug therapy, the annual cost rises and stays high (line #1, Fig. 4c), but
with Rezulin, VISN 3’s annual cost stays under control (line #2, Fig. 4c).
Looking at the annual cost components, one can see why using Rezulin
yields a cost reduction. Figure 5a shows that while the complications cost
declines as diabetic veterans seek treatment (line #1), treatment cost rises
sharply at first precisely because more DVs move into treatment (line #2: T
cost). Gradually, however, as more and more diabetic veterans start using
Rezulin, T cost drops but it’s still higher than the Rezulin cost (line #3: R
cost, Fig. 5a). This does not happen simply because Rezulin-based therapy
costs less but, more importantly, because it also prolongs diabetics’ life and
thereby reduces the complications cost of the disease.
The last two graphs of Fig. 5 look at VISN 3’s Cash position through
time. Specifically, line #1 on the comparative graph of Fig. 5b shows that
ceteris paribus, without Rezulin (dunmiy=0), very soon VISN 3 may have no
Cash left for diabetes treatment. The broken-line segment of the #1 Cash line
follows a dire-straights trajectory very far below zero. Conversely, with the
use of Rezulin (dummy=1), the veteran’s network Cash position improves
through time (line #2, Fig. 5b).
With the new oral-antidiabetic drug therapy enabled at time t=5 years,
VISN 3 might still run into financial trouble due to Cash shortage. The
veterans’ network may still recover, however, thanks to Rezulin. How fast
the recovery takes place will depend on the DV migration fraction (fr). The
faster the network’s diabetic veterans migrate from conventional insulin
treatment to Rezulin therapy (by 25 to 50 to 75 percent), the faster VISN 3’s
Cash position moves from being negative to being positive again (Fig. 5c).
Conclusion
The system dynamics modeling process used here aimed at showing
exactly how the new oral-antidiabetic drug Rezulin might prove itself to be
an effective and efficient treatment for diabetes. The model built depicts
relationships within VISN 3 among diabetic veteran sub-populations,
depending on whether or not DVs seek treatment and are willing to switch
from conventional insulin or other drug treatment to Rezulin therapy. The
model also incorporates real-life statistical data, and assesses Rezulin’s
effectiveness and efficiency via performance variables such as average life,
and annual cost and Cash position, respectively.
The simulation results show that indeed Rezulin is an effective drug. In
all simulation experiments, diabetic veterans’ average life is prolonged
once they move from conventional insulin or other drug treatment to
Rezulin therapy. Rezulin therapy is also an efficient anti-diabetic treatment,
not only because it costs less than insulin and other anti-diabetic drugs but,
more importantly, by prolonging life it also reduces the complications cost of
diabetes. Rezulin use will help diabetics battle health care’s rising cost.
Appendix: Model Equations
Level Variables
Cash(t) = Cash(t - dt) + (annual_revenue - annual_cost) * dt
INIT Cash = 151,457,218 {dollars}
DVs_in_VISN_3(t) = DVs_in_VISN_3(t - dt) + (wew_DVs - to_treatment - death) * dt
INIT DVs_in_VISN_3 = 10000 {persons (DV = diabetic veteran)}
DVs_in_Treatment(t) = DVs_in_Treatment(t - dt) + (to_treatment - to_new_drug -
death_in_treatment) * dt
INIT DVs_in_Treatment = 1000 {persons}
DVs_on_Rezulin(t) = DVs_on_Rezulin(t - dt) + (to_new_drug - death_on_Rezulin) * dt
INIT DVs_on_Rezulin = 500 * dummy {persons}
Rate Variables
annual_cost = complications_cost + T_cost + R_cost {dollars/year}
annual_revenue = funding_fr * annual_gov’t_funds {dollars/year}
death = DVs_in_VISN_3 / life_expectancy {persons/year}
death_in_treatment = DVs_in_Treatment / (avg_life *, T_gain {persons/year}
death_on_Rezulin = (dummy * DVs_on_Rezulin) / (avg_life_in_T * R_gain)
{persons /year}
new_DVs = 500 {persons/year}
to_new_drug = to_new_drug_fr * DVs_in_Treatment * dummy {persons/year}
to_treatment = to_treatment_fr * DVs_in_VISN_3 {persons/year}
Auxiliary Variables
avg_life = DVs_in_VISN_3 / death {years}
avg_life_in_T = DVs_in_Treatment / death_in_treatment {years}
avg_life_on_R = DVs_on_Rezulin / death_on_Rezulin {years}
complications_cost = complications_cost_per_DV * (death + death_in_treatment +
death_on_Rezulin) {dollars /year}
dummy = IF(TIME < 5) THEN(0) ELSE(1) {dimensionless, 0 or 1 variable}
R_cost = R_cost_per_DV * DVs_on_Rezulin {dollars}
T_cost = T_cost_per_DV * DVs_in_Treatment {dollars}
Constant Parameters
annual_gov’t_funds = 1,009,714,793 {dollars/year}
complications_cost_per_DV = 100,000 {dollars /person}
funding_fr = 0.15 {dimensionless}
life_expectancy = RANDOM(12, 18, 123) {years (Set to 15 years for smooth graphs)}
R_cost_per_DV = 1,077 {dollars/year}
R_gain = 1.5 {dimensionless}
to_new_drug_fr = 0.5 {dimensionless}
to_treatment_fr = 0.5 {dimensionless}
T_cost_per_DV = 25,242 {dollars/year}
T_gain = 4/3 {dimensionless}
