Pharmacoeconomics 2009; 27 (2): 159-165

ORIGINAL RESEARCH ARTICLE 1170-7690/09/0002-0159/$49.95/0

2009 Adis Data Information BV. All rights reserved.

Comparison of Markov Model

and Discrete-Event Simulation
Techniques for HIV
Kit N. Simpson,1 Alvin Strassburger,1 Walter J. Jones,1 Birgitta Dietz2
and Rukmini Rajagopalan3
1 Medical University of South Carolina, Charleston, South Carolina, USA
2 Abbott GmbH and Co KG, Ludwigshafen, Germany
3 Abbott Laboratories, Abbott Park, Illinois, USA

Abstract Background: Markov models have been the standard framework for pre-
dicting long-term clinical and economic outcomes using the surrogate marker
endpoints from clinical trials. However, they are complex, have intensive data
requirements and are often difficult for decision makers to understand.
Recent developments in modelling software have made it possible to use
discrete-event simulation (DES) to model outcomes in HIV. Using published
results from 48-week trial data as model inputs, Markov model and DES
modelling approaches were compared in terms of clinical outcomes at 5 years
and lifetime cost-effectiveness estimates.
Methods: A randomly selected cohort of 100 antiretroviral-naive patients
with a mean baseline CD4+ T-cell count of 175 cells/mm3 treated with lopi-
navir/ritonavir was selected from Abbott study M97-720. Parameter esti-
mates from this cohort were used to populate both a Markov and a DES
model, and the long-term estimates for these cohorts were compared. The
models were then modified using the relative risk of undetectable viral load as
reported for atazanavir and lopinavir/ritonavir in the published BMS 008
study. This allowed us to compare the mean cost effectiveness of the models.
The clinical outcomes included mean change in CD4+ T-cell count, and
proportion of subjects with plasma HIV-1 RNA (viral load [VL]) <50
copies/mL, VL 50400 copies/mL and VL >400 copies/mL. US wholesale
acquisition costs (year 2007 values) were used in the mean cost-effectiveness
analysis, and the cost and QALY data were discounted at 3%.
Results: The results show a slight predictive advantage of the DES model
for clinical outcomes. The DES model could capture direct input of CD4+
T-cell count, and proportion of subjects with plasma HIV-1 RNA VL <50
copies/mL, VL 50400 copies/mL and VL >400 copies/mL over a 48-week
period, which the Markov model could not. The DES and Markov model
estimates were similar to the actual clinical trial estimates for 1-year clinical
results; however, the DES model predicted more detailed outcomes and had
160 Simpson et al.

slightly better long-term (5-year) predictive validity than the Markov model.
Similar cost estimates were derived from the Markov model and the DES.
Both models predict cost savings at 5 and 10 years, and over a lifetime for the
lopinavir/ritonavir treatment regimen as compared with an atazanavir
regimen.
Conclusion: The DES model predicts the course of a disease naturally, with few
restrictions. This may give the model superior face validity with decision makers.
Furthermore, this model automatically provides a probabilistic sensitivity anal-
ysis, which is cumbersome to perform with a Markov model. DES models allow
inclusion of more variables without aggregation, which may improve model
precision. The capacity of DES for additional data capture helps explain
why this model consistently predicts better survival and thus greater savings
than the Markov model. The DES model is better than the Markov model in
isolating long-term implications of small but important differences in crucial
input data.

Background model structure has the advantage over a Markov

Modelling is an analytic methodology that
accounts for events over time and across popu-
lations, that is drawn from primary and/or sec-
ondary data sources, and whose purpose is to
estimate the effects of an intervention on valued
health consequences.[1] There are three major
approaches used to design models, including de-
cision trees, Markov models and discrete-event
simulations (DES).
A DES model is a mathematical structure that
is able to use a stochastic process to simulate
outcomes for a synthetic or theoretical group
of patients that replicate the statistical charac-
teristics that are specified in the model. A popu-
lation-based state-transition or Markov-type
model simulates what may be expected to happen
to groups of individuals over time, based on their
occupation of a set of mutually exclusive and
jointly exhaustive health states, which are linked
with a fixed set of time-dependent probabilities of
transmission from one health state to another.
See Stahl[2] for an excellent, in-depth description
of different model structures.
DES models have often been used in health-
care to simulate queuing problems, but lately
have also been applied to predict outcomes for
complex chronic conditions.[3,4] This type of
model, in that it represents the course of a
disease naturally, with few restrictions. It does
not require mutually exclusive branches or rigidly
defined health states with fixed cycles. Its only
real limitation is usually related to lack of in-
formation on distributional characteristics of the
data used to develop risk measures, as well as lack
of information on the correlation between spe-
cific risk factors. These are some practical chal-
lenges to using this type of model in healthcare
research since to date it has mainly been em-
ployed in healthcare for predicting outcomes of
care delivery processes.[5]
This study serves two purposes: (i) it uses
48-week input data for one treatment, and com-
pares the 5-year estimates from the Markov model
and the DES model with the actual outcomes for
the population on which the model estimates are
based; (ii) it compares how the two models perform
when they use 48-week input data from a clinical
trial to estimate the incremental cost-effectiveness
ratio (ICER) for two competing treatment regi-
mens over patients lifetimes. In this case we illus-
trate how well the models perform when they must
capture key differences in the results for lopinavir/
ritonavir and atazanavir/ritonavir to estimate an
ICER for these antiretroviral treatment regimens
based on published data.[4-6]

2009 Adis Data Information BV. All rights reserved. Pharmacoeconomics 2009; 27 (2)
Markov Model vs Discrete Event Simulation for HIV
Methods
We compared the performance of a newly de-
veloped DES model and a previously published
Markov-type state-transition model for HIV with
the results of a clinical trial[7] in patients treated with
lopinavir/ritonavir. We also estimated the out-
comes for two different antiretroviral therapies
(lopinavir/ritonavir vs atazanavir/ritonavir), based
on data from clinical trials reported in the litera-
ture,[6,8,9] and compared the ICERs that the
models estimated for these two treatment regimens.
For the comparison of the models with clinical
trial data, we generated a synthetic cohort of 100
antiretroviral-naive patients with a mean baseline
CD4+ T-cell count of 175 cells/mm3, based on
data from the Abbott M97-720 study.[10,11] For
the comparison of the ability of the models to
estimate ICERs, we used the relative risk of
achieving viral load (VL) suppression <50 copies/mL
reported for lopinavir/ritonavir and atazanavir
from the BMS 008 study.[6]
The clinical outcomes compared across the
two models included the mean CD4+ T-cell
count, proportion of subjects with plasma HIV-1
RNA VL <50 copies/mL, VL 50400 copies/mL
and VL >400 copies/mL. The economic outcomes
estimated by the two models for the two com-
peting antiretroviral regimens were as follows:
(i) 5- and 10-year estimates of cost; (ii) estimates
of total lifetime QALYs; (iii) mean cost per QALY
for each model estimate; and (iv) mean cost-
effectiveness ratios comparing the two regimens
(ICERs cannot be calculated because one regi-
men was dominant).
US wholesale acquisition drug costs (year
2007 values) were used in the cost-effectiveness
analysis and the cost and QALY data were ad-
justed by an annual discount rate of 3%. The
model input parameters are provided in tables
A13 of the supplementary material (see Article-
Plus at http://pharmacoeconomics.adisonline.
com).
Markov Model
The Markov Model has been used several
times to estimate the cost effectiveness of anti-
2009 Adis Data Information BV. All rights reserved.
161
retroviral therapies.[12-15] It simulates outcomes
(in terms of QALYs) and costs for a cohort of
patients starting on one drug regimen and com-
pares them with those for a cohort of patients
starting on another regimen. The Markov model
structure is explained in the supplementary
material.
Discrete-Event Simulation (DES) Model Design
Considerations
The DES model structure, parameters and
output are explained in detail in the supplemen-
tary material. The DES model has a large set of
automatic outputs and can accommodate a table
of user-defined outputs. A natural-progression
DES model was first constructed using Abbotts
data from the M97-720 study.[7] The model input
parameters in this model were then adjusted by
relative risk estimates to reflect the parameter
values from the clinical trial.[6,8,9]
The structure of the DES model is designed to
be able to capture two important parameters,
which may be expected to make significant differ-
ences in long-term outcomes for the two regi-
mens: differences in virological outcomes and
specific changes in laboratory results. For lopi-
navir/ritonavir, a greater proportion of patients
have VL suppressed to <50 copies/mL than ataza-
navir patients. For atazanavir, lower serum cho-
lesterol levels were observed at the end of the
clinical trial than for lopinavir/ritonavir.[8]
The same clinical trial reports[6,8,9] were used
to estimate the parameter values for updating the
Markov model. This consistency in data sources
allowed us to compare the Markov and the DES
model estimates.
Results
Clinical Outcomes
Results of the DES and Markov models are
similar when the timeframe of reference is short
(1 year). However, results at the end of a longer
duration (5 years) show that the DES model has a
slightly better predictive ability than the Markov
model. The 1-year estimates of CD4+ T-cell
count and 1- and 5-year VL estimates from both
Pharmacoeconomics 2009; 27 (2)
162 Simpson et al.

Table I. Markov and discrete-event simulation (DES) model clinical value and the corresponding average patients
predictive validity for the lopinavir/ritonavir patient sample[16] survival estimates are plotted together.
Parameter Actual Markov DES range
clinical value prediction prediction
Probabilistic Sensitivity Analysis (PSA) and
CD4 count at 258 NA 24770 Evaluation of Parameter Relationships
1 year (mean)
VL <400 copies/mL 65 65 6166 A DES model can be programmed to perform
at 1 year (%)
100 iterations, each based on a randomly selected
VL <50 copies/mL 54 NA 5153
at 1 year (%)
combination of the model variables, as specified
CD4 count at 841 NA 82258
by the values and distributions of the parameters
5 years (mean) of the model. Each estimate can be set for a spe-
VL <400 copies/mL 56 50 5760 cific cohort size. This allows the user to perform
at 5 years (%) probabilistic sensitivity analysis (PSA), as well as
VL <50 copies/mL 54 NA 5556 examine the effects of applying model decisions
at 5 years (%) to small and large practice settings.
NA = not available (the Markov model is not designed to produce
In addition, this probabilistic feature allows
these estimates); VL = viral load.
the user to plot the relationships between key
model outputs to get an understanding of how
the Markov and DES models are very stable, and outputs and variable values move together, as
lie very close to the values actually observed in shown in figure 1. This plot makes it very clear that
the cohort underlying the model (see table I). these two outcomes are correlated as expected.

Cost Effectiveness Discussion

Similar cost estimates are derived from Markov
models and DES (table II). Both models predict
cost savings at 5 and 10 years, and over a lifetime
for the lopinavir/ritonavir treatment regimen.
The values output of the DES model depends on
how many patients are specified for the simula-
tion, and how many iterations the model runs.
The larger the number of estimates (i.e. entities;
e.g. n = 2000), the less variation may be expected
between each model run. However, each time a
model is saved with a new name it will draw from
a new random set of starting values, and the re-
sulting estimates will differ from the estimates in
the previous model iteration. Figure 1 illustrates
this variation where the estimated 48-week VL
This paper presents results from a traditionally
structured HIV model, and compares these re-
sults with those of a more innovative healthcare
application,[17] a DES model. The estimated
outcomes for the two structures are very close,
but not identical. The DES model has slightly
better predictive ability over 5 years, and is able
to provide many more details about what may be
expected to happen to a population than the
more traditional Markov model. However, the
performance of the Markov model is quite good.
Its results fall clearly within the 3% margin of
error (see table I) defined by other modellers
as the acceptable level of variability for a point
validation of an economic model.[18,19]

Table II. Comparing the cost-effectiveness estimates of the Markov model (MM) and discrete-event simulation (DES) model
Model/regimen Costs ($US, year 2007 values) Total QALYs Mean cost per QALY
5-year 10-year total
MM: lopinavir/ritonavir 105 808 189 704 310 194 10.55 29 402
DES: lopinavir/ritonavir 90 335 152 912 340 022 12.40 27 421
MM: atanazavir 119 551 207 649 318 882 10.11 31 541
DES: atanazavir 97 283 163 288 352 843 12.11 29 136

2009 Adis Data Information BV. All rights reserved. Pharmacoeconomics 2009; 27 (2)
Markov Model vs Discrete Event Simulation for HIV
64.5
Percentage of patients with
64.0
VL <400 copies/mL
63.5
63.0
62.5
62.0
61.5
12.4 12.5 12.6 12.7
QALYs
Fig. 1. Correlation between model predictions of 48-week viral load
(VL) and QALY estimates. Each point is one model run.
Decision-analysis models provide an im-
portant mechanism for translating the results of
clinical trials into measures of quality of life
(QOL) and costs. This has especially been the
case in HIV since the mid-1990s, when the ac-
cepted clinical trial outcomes changed from dif-
ferences in progression to AIDS or death, to
differences in surrogate markers,[20] usually VL
suppression and CD4+ T-cell count increase.
The most common analytical structure used
for HIV economic models has been the Markov,
or state-transition model.[21-26] One such model,
the Cost Effectiveness of Preventing AIDS
Complications (CEPAC) model, has been used
extensively to estimate outcomes expected for
treatments for individuals with HIV.[25,26] In only
rare cases are these models ever compared with
the estimations derived from differently struc-
tured models[23] or with actual clinical trial
results.[27] This is because the models are pro-
prietary, and even well described and extensively
published proprietary models (such as the
CEPAC) rarely have enough data in a manu-
script to allow others to replicate their complete
structure. However, the structure of a model is an
important component to assess in the evaluation
of the quality of a modelling estimate.[28]
Each of these two model structures has
advantages and disadvantages that must be
considered before choosing one over the other
(table III). In many cases, the choice must be
guided by the data available for the modelling.
The transition matrix in the Markov model
requires large amounts of data, especially if the
2009 Adis Data Information BV. All rights reserved.
163
model contains many health states, and these
data must be available as individual observa-
tions to allow the analysis required to estimate
transitions. However, once well documented
transitions for a disease are available for a
specific severity group, they can form the basis
for model adaptations. The DES model may
not need a large amount of data, and some of
12.8 12.9 the models parameters can come from com-
monly reported outcomes or risks, but it requires
data that are recorded over a much longer time
because it does not have the repetitive assump-
tion of a Markov model. Thus, as is often the
case in science, an aspect that is a strength in
one case may become a weakness in the next
application.
Furthermore, a Markov model uses mean
values for all the variables in the model to calcu-
late outcomes, while a DES model uses a random
draw from the distributions depicted by mean
values and their measures of variation to estimate
outcomes. Thus, DES outcomes are always esti-
mated a number of times and the mean of these
estimates are reported. This provides some
information about the effect of the statistical
uncertainty in the model input values. Stable
models that are based on strong statistical para-
meters will show little variation in the estimates,
while unstable model estimates will vary greatly.
Examining the high and low values for each
parameter provides a good representation of the
stability of a model.
In this study, the older Markov approach re-
quired substantially fewer data parameters for
estimation and much less time for implementa-
tion. The extra time requirement for the DES
model might be because of our relatively lesser
experience in working with the Arena soft-
ware.[29] However, the relatively large data re-
quirement may also be an issue in the use of DES
models in the estimation of outcomes and costs
for HIV treatments, because a structure of a
model may be substantially influenced by the
data available for modelling.[30] The issue of
which structure is best for HIV has not yet been
determined. It will depend on which of the two
models is most transparent to decision makers,
since model transparency is a critical character-
Pharmacoeconomics 2009; 27 (2)
164
Table III. Summary of strengths and weaknesses of the Markov and discrete-event simulation (DES) models for use in estimation of the cost
effectiveness of antiretroviral therapies in HIV
Markov model
Computationally simple, can be implemented in Microsoft Excel
Cumbersome to modify when the lack of memory assumption does
not hold
Uses the raw data frequencies from the clinical trial and may
therefore account for all differences between treatment groups
Does not lend itself easily to probabilistic sensitivity analysis
Can be visually inspected for programming errors, and is easy to test
for technical replication
A small increase in the number of health states requires a large
increase in the data input to get stable transitions
Sequences in treatment over time require the population to transit to
a new Markov model within the model structure
Requires mutually exclusive and jointly exhaustive health states,
which is an unnatural way for clinicians to think about a disease
Is limited to reporting mean values estimated across the specific
health states for the time increments in the model, but results are
easy to program and to check visually
All characteristics of individuals must be defined in relation to the
health states in the model, and sub-models may be needed to
capture important conditions
istic, and one that can only be determined by
exposing the model to critique from those who
consider using it to inform decisions.
Conclusion
The DES model predicts more detailed out-
comes and has slightly better long-term predictive
validity than the Markov model. It represents the
course of a disease much more naturally, with
fewer restrictions. This may give the model
superior face validity with decision makers. Most
importantly, this model automatically provides
a PSA, which is cumbersome to perform with a
Markov model.
The capacity of the DES model for additional
data capture helps explain why it consistently
predicts better survival and greater savings. The
DES model is better than the Markov model in
isolating long-term implications of small but im-
portant differences in crucial input data.
2009 Adis Data Information BV. All rights reserved.
Simpson et al.
DES model
Requires special, often expensive, software with archaic
programming conventions not commonly used in healthcare
Can easily accommodate risk changes, such as ageing, over time
Uses statistical parameters, such as relative risk of viral load
breakthrough, or mean CD4 increase per time period, which may not
capture subtle variations in distributions between the measurement
times
Probabilistic sensitivity analysis is inherent in the model
Parameter checks require the examination of hidden logic in nodes
and tables, which can be difficult to do
Very parsimonious data requirements, but may need populations
observed over long time periods to identify flow change points
New treatments are easily accommodated by simple flow
specifications
Creates many individual entities that represent patients. Entities can
carry permanent traits (sex) and changing traits (age) and contact
probabilistic events (AIDS or myocardial infarction)
Can provide a report for any variable at any time; however, reports can
be difficult to program
No limit to the characteristics of individuals that can be included, as
long as both parameter estimates and the correlation among the
variables included are available
Because of the limitations of the Markov
model, researchers end up using categorical
groupings for complex interacting continuous
measures, with a potential for short-term ag-
gregation bias leading to long-term prediction
errors. DES models allow inclusion of individual
variables without a need for creating compound
health states, thus improving the model precision
as demonstrated above. However, they may also
require much more data analysis and program-
ming time, and at this point it is not clear to us
which structure has the greater transparency.
Acknowledgements
This study was funded by a grant from Abbott Labora-
tories to the Medical University of South Carolina. Drs Dietz
and Rajagopalan are Abbott employees and as such are
eligible to receive stock options. Kit Simpson has received
consultancy fees from Abbott. The authors acknowledge the
assistance of Menaka Bhor, PhD, Abbott Laboratories, in
editing and organizing the manuscript.
Pharmacoeconomics 2009; 27 (2)
Markov Model vs Discrete Event Simulation for HIV
References
1. Weinstein M, OBrien B, Hornberger J, et al. Principles of
good practice for decision analytic modeling in health-care
evaluation: report of the ISPOR Task Force on Good Research
Practices-Modeling Studies. Value Health 2003; 6 (9): 9-17
2. Stahl JE. Modeling methods for pharmacoeconomics and
health technology assessment: an overview and guide.
Pharmaceconomics 2008; 26 (2): 131-48
3. Caro JJ. Pharmacoeconomic analyses using discrete event
simulation. Pharmacoeconomics 2005; 23 (4): 323-32
4. Patten SB. An animated depiction of major depression epi-
demiology. BMC Psychiatry 2007; 7: 23
5. Matta ME, Patterson SS. Evaluating multiple performance
measures across several dimensions at a multi-facility out-
patient center. Health Care Manag Sci 2007; 10 (2): 173-94
6. FDA Center for Drug Evaluation and Research. Applica-
tion 21-567 statistical review(s) 2002 [online]. Available
from URL: http://www.fda.gov/cder/foi/nda/2003/21-
567_Reyataz_Statr.pdf [Accessed 2006 Nov 7]
7. Hicks C. Long-term safety and durable antiretroviral
activity of lopinavir/ritonavir in treatment-nave patients:
4-year follow-up study. AIDS 2004; 18: 775-9
8. Johnson M, Grinsztejn B, Rodrigues C, et al. Atazanavir
plus ritonavir or saquinavir, and lopinavir/ritonavir in
patients experiencing multiple virological failures. AIDS
2005; 19 (7): 685-94
9. Johnson M, Grinsztejn B, Rodrigues C, et al. 96-week
comparison of once-daily atazanavir/ritonavir and twice-
daily lopinavir/ritonavir in patients with multiple vir-
ological failures. AIDS 2006; 20 (5): 711-8
10. Gathe J, Podzamczer D, Johnson M, et al. Once-daily vs
twice-daily lopinavir in antiretroviral-naive patients:
48-week results [poster #570]. 11th Conference on Retro-
viruses and Opportunistic Infections; 2004 Feb 8-11; San
Francisco (CA) [online]. Available from URL: http://
www.retroconference.org/2004/cd/PDFs/570.pdf [Accessed
2007 Oct 22]
11. Walmsley S, Bernstein B, King M, et al. Lopinavir-ritonavir
vs. nelfinavir for the initial treatment of HIV infection.
N Engl J Med 2002; 346 (26): 2039-46
12. Simpson KN, Voit EO, Goodman R, et al. Estimating the
social and economic benefits of pharmaceutical innova-
tions: modeling clinical trial results in HIV-disease. Res
Hum Cap Dev 2001; 14: 175-98
13. Simpson KN, Luo MP, Chumney ECG, et al. Cost effec-
tiveness of lopinavir/ritonavir versus nelfinavir as the first-
line highly active antiretroviral therapy regimen for HIV
infection. HIV Clin Trials 2004; 5 (5): 294-304
14. Simpson KN, Luo M, Chumney ECG, et al. Cost-
effectiveness of lopinavir/ritonavir compared to atazanavir
in antiretroviral-nave patients: modeling the combined
effects of HIV and heart disease. Clin Drug Invest 2007; 27 (1):
67-74
15. Simpson KN, Jones WJ, Rajagopalan R, et al. Cost effec-
tiveness of lopinavir/ritonavir compared with atazanavir
plus ritonavir in antiretroviral-experienced patients in the
US. Clin Drug Invest 2007; 27 (7): 443-52
16. Murphy R, daSilva B, McMillan F, et al. Seven year follow-
up of a lopinavir/ritonavir-based regimen in antiretroviral-
2009 Adis Data Information BV. All rights reserved.
165
nave subjects [poster PE7.9-7]. 10th European AIDS
Conference; 2005 Nov 17-20; Dublin
17. Weinstein MC. Recent developments in decision-analytic
modeling for economic evaluation. Pharmacoeconomics
2006; 24 (11): 1043-53
18. Weinstein MC, Coxon PG, Williams LW, et al. Forecasting
coronary heart disease incidence, mortality, and cost: the
coronary heart disease policy model. Am J Public Health
1907; 77 (11): 1417-26
19. Hunink MG, Goldman L, Tosteson AN, et al. The recent
decline in mortality from coronary heart disease,
19801990: the effect of secular trends in risk factors and
treatment. JAMA 1997; 277 (7): 535-42
20. Simpson KN. Design and assessment of cost effectiveness
studies in AIDS populations. JAIDS 1995; 10 Suppl. 4: S28-32
21. Schulman KA, Lynn LA, Glick HA, et al. Cost-effectiveness
of low-dose zidovudine therapy for asymptomatic patients
with human immunodeficiency virus (HIV) infection. Ann
Intern Med 1991; 114: 798-802
22. Simpson K, Andersson F, Shakespeare A, et al. Cost effec-
tiveness of antiviral treatment with zalcitabine in combi-
nation with zidovudine for AIDS patients with CD4 counts
300 per mm3 in five European countries. Pharmaco-
economics 1994; 6 (6): 553-62
23. Chancellor J, Hill A, Simpson K, et al. Cost effectiveness of
3TC and ZDV in patients with HIV-Disease. Pharmaco-
economics 1997; 12 (1): 1-13
24. Biddle AK, Simpson KN. Modeling the cost effectiveness of
nevirapine triple combination therapy and dual combina-
tion therapy for the treatment of HIV disease in the United
Kingdom. J Med Econ 1999; 2: 85-105
25. Schackman BR, Goldie SJ, Weinstein MC, et al. Cost-
effectiveness of earlier initiation of antiretroviral therapy
for uninsured HIV-infected adults. Am J Public Health
2001; 91 (9): 1456-63
26. Paltiel AD, Weinstein MC, Kimmel AD, et al. Expanded
screening for HIV in the United States: an analysis of cost-
effectiveness. N Engl J Med 2005; 352 (6): 586-95
27. Mauskopf J, Lacey L, Kempel A, et al. The cost effectiveness
of treatment with lamivudine and zidovudine compared
to zidovudine alone: a comparison of Markov model
and trial data estimates. Am J Managed Care 1998; 4 (7):
1004-12
28. Phillips Z, Bojke L, Schulper M, et al. Good practice
guidelines for decision-analytic modeling in health techno-
logy assessment. Pharmacoeconomics 2006; 24 (4): 355-71
29. Rockwell Automation. Arena. Warrendale (PA) [online].
Available from URL: http://www.arenasimulation.com/
[Accessed 2008 Nov 21]
30. Garrison LP. The ISPOR good practice modeling principles:
a sensible approach. Be transparent be reasonable. Value
Health 2003; 6 (1): 6-8
Correspondence: Prof. Kit N. Simpson, Department of
Health Administration and Policy, College of Health Pro-
fessions, Medical University of South Carolina, 151 Rut-
ledge Avenue, Building B, Charleston, SC 29425, USA.
E-mail: simpsonk@musc.edu
Pharmacoeconomics 2009; 27 (2)