Zhang BMC Health Services Research (2018) 18:687

https://doi.org/10.1186/s12913-018-3456-4

RESEARCH ARTICLE Open Access

Application of discrete event simulation in

health care: a systematic review
Xiange Zhang

Abstract
Background: The objective was to explore the current advances and extent of DES (Discrete Event Simulation)
applied to assisting with health decision making, as well as to categorize the wide spectrum of health-related topics
where DES was applied.
Methods: A systematic review was conducted of the literature published over the last two decades. Original
research articles were included and reviewed if they concentrated on the topic of DES technique applied to health
care management with model frameworks explicitly demonstrated. No restriction regarding the settings of DES
application was applied.
Results: A total of 211 papers met the predefined inclusion criteria. The number of publications included increased
significantly especially after 2010.101 papers (48%) stated explicitly disease areas targeted, the most frequently
modeled of which are related to circulatory system, nervous system and Neoplasm. The DES applications were
distributed unevenly into 4 major classes: health and care systems operation (HCSO) (65%), disease progression
modeling (DPM) (28%), screening modeling (SM) (5%) and health behavior modeling (HBM) (2%). More than 68% of
HCSO by DES were focused on specific problems in individual units. However, more attempts at modeling highly
integrated health service systems as well as some new trends were identified.
Conclusions: DES technique has been an effective tool to approach a wide variety of health care issues. Among all
DES applications in health care, health system operations research occupied the most considerable proportion and
increased most significantly. Health Economic Evaluation (HEE) was the second most common topic for DES in
health care, but with stable rather than increasing numbers of publications.
Keywords: Systematic review, Discrete event simulation, Health care management

Background an operational research technique can be more advanta-

DES is a form of computer-based modeling methodology geous to model complex systems at the individual level
that provides an intuitive and flexible method [1], char- instead of cohort level. The contexts simulated are usu-
acterized by the ability to simulate dynamic behaviors of ally viewed as queuing networks where individual en-
complex systems and interactions between individuals, tities pass through a series of discrete events one by one
populations and their environments [2]. The goal of at discrete intervals, between which they have to wait in
such a model is to comprehensively compare potential queues due to the constrained availability of resources
practices or strategy options, so as to identify the most [4]. DES is also able to allow decision makers to conduct
efficient and effective ones, especially in situations where “what if” analyses by changing the operational scenarios
it may be infeasible to carry out required surveys or and rules, to predict the possible impacts resulting from
comparative experiments in practice. a variety of policy alternatives before truly translated
In comparison with aggregate models without inter- into practice without any alteration in present systems.
action [3] like decision trees or Markov models, DES as However, admittedly DES models are also simplified pre-
sentations of reality, just as the portraits of other model-
Correspondence: [email protected] ing techniques indicate. Typically, patients or medical
Department of Health Care Management, Institute of Public Health and
Nursing Research, Health sciences, University of Bremen, Grazer Str. 2a, 28359 care providers are referred to as entities, an essential
Bremen, Germany

© The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to
the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver
(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Zhang BMC Health Services Research (2018) 18:687 Page 2 of 11

concept of DES, which go through the whole process  Studies had to be carried out in health care delivery
simulated. Attributes of target entities, which can affect or public health scenarios, which can be related to
or even determine entities’ responses to events, include any specific context of health service systems.
age, sex, health status, illness history, duration of disease  Discrete event simulation frameworks should be
and other demographics, the values of which are updat- pointed out as the main modeling technique and
able as models are running [1, 5]. Events can be viewed independently formulated and structured.
as a wide range of things which may happen during  How exactly work flows were structured and
simulation, like the occurrence or recurrence of a dis- simulated should be clarified and clearly
ease, admission to health facility, delivery of some med- demonstrated in the form of either flow charts or
ical treatment or transitions between health states. More descriptive texts, according to the recommendations
detailed characterizations of DES are available from the of the ISPOR-SMDM modeling good research
studies by Caro and Karnon et al. [5, 6]. practice [1].
There are two relatively extensive reviews on DES  Only peer-reviewed original research was included
applications in the health care sector [7, 8], which for further review. Papers published as literature
were restricted to specific sub-systems or contexts of reviews, editorials, conference proceedings or
health care systems. The earlier one was published in methodological guidelines were excluded.
1999, surveying and classifying the DES literature  Only English articles were taken into account.
which focused on single health care units [7]. Around
ten years later, the second one sought to analyze DES Selection of publications
applications in hospital care [8]. To the best of our After removing duplicates, all studies were screened and
knowledge, the overall picture of how DES has been reviewed following two main phases. The initial stage of
applied to health care is still unclear. this review was conducted independently by XG and FG
The aim of this review is to provide an up-to-date following the predefined criteria mentioned above, at
overview of a considerable body of academic journal lit- title and abstract level, with the aim to identify studies
erature relevant to the application of DES in health care, which are appropriate for further full-text review. For in-
thereby further gaining deep insights into the extent of clusion criteria 1, 2 and 4, titles and abstracts screened
DES used to assist with health decision making and cat- in the Microsoft Excel table provided adequate informa-
egorizing the wide array of health-related topics where tion for reviewers to make decisions, but for criterion 3,
DES was applied. full texts of records were to be retrieved, when neces-
sary. In case of discrepancies in assessment results be-
tween the two reviewers, a third reviewer (WR)’s
Methods opinion was taken into consideration.
Search strategy In view of the emphasis of this review laid on the
For this systematic review, two main databases, PubMed overview of application advances of DES in health
and Web of Science, were systematically searched as the care, therefore only basic information hints of each
main sources for publications of interest. The last search paper retrieved were deemed necessary to be ex-
was performed on March 31, 2017, based on broad search tracted and screened on a standardized Excel form.
terms: ‘discrete event simulation’ or ‘(discrete event) AND All papers were stored and managed via EndNote X7.
(computer simulat* OR model*)’, accompanied by terms The following items were extracted and summarized
like ‘health service*’, ‘Patient*’, ‘healthcare’ or ‘health care’. from the literature: Title, author, publication date,
Papers which brought up discrete event simulation and journal, medical conditions, perspective for cost meas-
health-related topics in the title, abstract or keywords and urement, specific settings analyzed, setting classifica-
were published over the last 2 decades were searched for tion and application category, on the latter two of
without any restrictions on setting and country of origin which the two precedent similar reviews were focused
applied. As a supplement, the reference lists of relevant respectively [7, 8]. In this sense, this work could be per-
reviews previously published and the latest studies in- ceived as an update for our precedents. Nonetheless, more
cluded were scanned and hand-searching of pertinent detailed insights can be gained herein, for example, this
open-access journals was also conducted in order to re- review presented the physical contexts simulated more
trieve additional potential peer-reviewed publications. specifically instead of only roughly classifying them as in-
patient facilities, outpatient clinics and emergency depart-
ment. Additionally, a more comprehensive taxonomy of
Inclusion criteria application areas was introduced, which was built on re-
To identify studies pertaining to our intentions, the in- views by Mielczarek et al. [9] and Fone et al. [10]. To indi-
clusion of papers was based on the following criteria: cate the breadth and scope of DES applications in health
Zhang BMC Health Services Research (2018) 18:687 Page 3 of 11

care, medical conditions surveyed were also indicated and before 2010 (except for 2008) to almost 35 in 2016, with
categorized based on ICD-10 codes (International Statistical 73% of the included papers published after 2010. This
Classification of Diseases and Related Health Problems indirectly reflects a constantly rising popularity of DES
10th Revision) endorsed and recommended by WHO. modeling in health care management.

Results Categorization of applications

Search results Based on the nature and inherent properties of the DES
A total of 479 citations were yielded in the first step. After models selected, they can be arguably classified into 4
removal of 123 duplicates, 356 papers were screened and major categories, which were inspired by previous tax-
reviewed at title and abstract level. This first round of onomy [9, 10]: health and care systems operation
reviewing resulted in the exclusion of 136 articles which (HCSO), disease progression modeling (DPM), screening
failed to meet the inclusion criteria: 8% of articles ex- modeling and health behavior modeling (HBM) (Fig. 3),
cluded were not related to health care domain, 38% were which are explained further in the subsequent sections.
not original studies, 34% did not perform DES independ- Figure 3 illustrates that health and care systems oper-
ently, 19% did not or not clearly present DES structures ation accounts for the most considerable proportion
and entity flows and only one paper was not in English. (65%) of all modeling studies throughout the whole time
The remaining 220 articles were further reviewed on the span, followed by disease progression modeling (28%),
full text level. During this phase, the final 211 articles suf- relative to the limited proportion of screening (5%) and
ficiently met the eligibility criteria and 9 were excluded health behavior modeling (2%). In addition, the number
mainly due to the fact that they were methodological ana- of health and care systems operation modeling increased
lyses instead of original research, 2 of which actually fo- by the largest margin compared to other topic areas and
cused on other modeling techniques, not obviously disease progression modeling remained at around 5 pub-
related to DES. The review process and results are illus- lications level per year. What is also worth noting is that
trated as a PRISMA diagram in Fig. 1. modeling screening programs and health behavior
changes by DES seems to have drawn the attention of
Growing trend of DES applications in health care health care modelers in recent years.
The last two decades witnessed a noticeable scale of ex-
pansion of publications in this arena as illustrated in Disease distribution
Fig. 2, especially after 2010. The number of publications Research which specified a naturally developing history of a
increased from less than 10 papers annually published specific disease and an according course of treatment was

Fig. 1 Flow chart of this systematic review

Zhang BMC Health Services Research (2018) 18:687 Page 4 of 11

Fig. 2 Number of studies included in each year of publication

classified in terms of medical conditions in this section. ear and mastoid process), are included in Fig. 4 to varying
Out of the 211 papers included, there are a total of 101 pa- degrees.
pers (48%) which stated explicitly that specific disease areas
and corresponding treatment flows were targeted and mod- Disease progression modeling
eled. The most frequently analyzed medical indications are In regard to disease progression modeling, DES is com-
related to circulatory system, nervous system, Neoplasm monly employed to transparently conceptualize and con-
and musculoskeletal system diseases, occupying 18%, 15%, struct the course of diseases, health states transition and
15% and 13% respectively. The detailed distribution of dis- disease-related events patients will go through under dif-
eases focused on is clearly displayed in Fig. 4. ferent interventions [11, 12]. This application field is
ICD-10 is divided into 22 chapters from chapter I to mainly intended to assist in making well-informed med-
chapter XXII, each of which can be further divided into a ical decisions through comparing different treatment al-
number of sub-categories, and chapters I-XVII are associ- ternatives at medical level in terms of resources
ated to specific morbid conditions and different body sys- consumed, health outcomes or both.
tems. This distribution of diseases presented in Fig. 4 Forty six (77%) of all 60 disease progression modeling
covers more than half (14) of all 22 chapters of ICD-10. compared various medical interventions from a cost-ef-
Furthermore, it is worth noting that 13 (76%) of all the 17 fectiveness perspective. Ten (16%) were undertaken in
chapters I-XVII of ICD-10 and nearly all categories related order to find the most beneficial treatment options,
to body systems, except for chapter VIII (Diseases of the based only on either economic cost or health impact

Fig. 3 Categorization of applications of DES in health care

Zhang BMC Health Services Research (2018) 18:687 Page 5 of 11

Fig. 4 Distribution of disease areas analyzed in studies

dimension. Most of the remaining 7% papers were pre- investigate complex relationships among different sections
dictive models for future health service demands. In within a system with the aid of DES, so as to make opti-
total, the most (56) of disease progression models are re- mal operational and administrative decisions [9, 13]. As
lated to health economic evaluations. has been presented in section “Categorization of applica-
Of the 51 papers which collected and analyzed cost tions”, DES is most commonly utilized for operations re-
data, 21 (41%) employed the health care payer perspec- search purposes in health care, with the most significant
tive. 15 (29%) and 14 (28%) were conducted from the upward trend in contrast with other application areas.
perspective of the health care system and society, re-
spectively, with only one analysis taking the point of Distribution of settings for HCSO
view of a health care provider. It has been repeatedly pointed out that the applications of
DES were highly confined to specific single health care units
Health and care systems operation with rare attempts to model broader scenarios [7, 8]. The re-
Health and care systems operation is meant to enable sults of this review are also in support of this conclusion. As
health care managers to better understand the underlying is presented in Table 1, more than 68% (93) of HCSO con-
mechanisms of how a system operates, comprehensively centrated on single micro-systems. Furthermore, consistent

Table 1 Distribution of settings of HCSO by DES

Hospital-based outpatient clinic 26/19.1% Inpatient department 7/5.1%
Orthopedics 6/4.4% Orthopedics 4/2.9%
Pain 3/2.2% Cardiovascular 2/1.5%
Endoscopy 2/1.5% Obstetrics 1/0.7%
Obstetrics 2/1.5% Multi-facility health care provider 18/13.2%
Procedure center 2/1.5% Hospital 12/8.8%
Physiotherapy 2/1.5% Primary care clinic 3/2.2%
Cardiology 1/0.7% Blood center 1/0.7%
Dermatology 1/0.7% Obesity care center 1/0.7%
Vaccination 1/0.7% Sleep center 1/0.7%
Neurology 1/0.7% Emergency department 32/23.5%
Fetal diagnosis and treatment 1/0.7% Intensive care unit 10/7.4%
Phlebotomy service 1/0.7% Operating theater 7/5.1%
Dental clinic 1/0.7% Auxiliary department 11/8.1%
Ophthalmology 1/0.7% Multi-departmental network 7/5.1%
Teaching ambulatory care 1/0.7% Health care system 18/13.2%
n/%:Number of studies/proportion of studies in all HCSO papers
Zhang BMC Health Services Research (2018) 18:687 Page 6 of 11

with previous reviews and surveys [8, 14, 15], the major at- Screening modeling
tention of health care modelers was mainly focused on mod- A total of 10 screening modeling papers by DES were
eling workflows of emergency departments (32) and identified in this review, most of which were published
intensive care units (10) relative to other health care contexts after 2010.
modeled. Among all DES screening models, cancer screening
Nevertheless, the potential of DES for representing broader modeling has been most frequently embarked on. It is
continuums of health care delivery system was increasingly noteworthy that breast cancer screening dominates this
taken advantage of. Multi-facility health service providers type of modeling, accounting for 50% of cancer models.
and even health care systems as a whole were more fre- DES was applied to investigate the costs and health out-
quently modeled by DES than reported before, both taking comes of different mammographic follow-up schedules
up around 26% (36) of all HCSO papers. The remainder 6% [16, 17], alternative breast cancer screening programs
examined coordination effectiveness among departments or [18, 19] as well as the routine performance of a mam-
micro-systems within a certain health service provider. mography facility under different operational conditions
[20]. Health benefits were commonly measured as gains
in QALY (quality-adjusted life years) and costs were
Sub-categories of HCSO mainly measured from the perspective of the health care
The DES approach has been used to address a wide payer and system.
range of operational challenges encountered by health Other screening studies reviewed incorporate two
care stakeholders, as is depicted in Fig. 5. This tax- models which analyzed the effects of distinct colorectal
onomy was derived from the proposal of Lagergren cancer screening strategies on patient outcomes and sys-
[13], which was extended in this review by adding the tem performance [21, 22], with the remaining evaluating
category of “health economic evaluation”. This classi- different screening practice options for tuberculosis [23],
fication method distinguishes six major operations Niemann–Pick type C disease [24] and diabetic retinop-
research issues consisting of patient scheduling (in- athy [25].
cluding appointment and discharge scheduling for
both outpatient and inpatient care), resource alloca- Health behavior modeling
tion, capacity planning and management, staff sched- In marked contrast to other industries modeled by DES,
uling, system diagnosis and evaluating the effects of the health care domain is deemed more complex to
operational changes or reconfigurations, the last one simulate. One important reason, among others, is that
of which occupies the most substantial proportion entities such as patients or health service providers
(25%) of all HCSO papers and all others stand at a modeled are individuals characterized by complicated
very similar level of around 12%. 19 (14%) health eco- psychological and behavioral attributes. This attribute
nomic evaluations were identified in HCSO category. makes health care more difficult to be represented by

Fig. 5 Distribution of sub-categories of HCSO

Zhang BMC Health Services Research (2018) 18:687 Page 7 of 11

computer simulation than to purely model production model to estimate the number of new pediatric HIV infec-
machines or materials. Incorporating human behavioral tions through Mother-to-Child Transmission [32].
factors into modeling practice is a newly emerging con-
cern for health care modelers. Most studies of this type More efforts towards modeling integrated multi-facility
were published after 2011. providers were identified
One attempt for health behavior modeling was to view The majority of modeling studies included are unit spe-
a certain individual behavior as the object to be mod- cific, focusing on individual micro-systems such as emer-
eled, in others words, to model the trajectories of indi- gency department and intensive care unit, consistent with
vidual behavioral changes over time under different previous reviews [7, 8]. Nonetheless, there are an increas-
interventions in order to determine the most influential ing number of studies modeling complex integrated
and efficient mechanisms to encourage desirable health healthcare providers, which were published particularly
behaviors. This category was highly dominated by after 2010. Certainly, these independent multi-facility pro-
smoking-cessation behavior. Smoking behaviors, quit at- viders modeled as a whole tended to be simulated at an
tempts, relapses and sets of events corresponding to aggregate and abstract level, because it appeared infeasible
smoking-cessation behaviors were simulated by DES to model in great detail—only a sequence of essential ac-
structures with the intention to identify the most tivities occurring during patients’ stay from admission to
cost-effective smoking-cessation strategies for diverse discharge were represented by DES frameworks.
populations [26–28]. Health care management could benefit a lot from DES
There were other attempts to take human behavioral used to model integrated providers as a whole rather than
factors into account while modeling health care delivery, just limiting to single units. Given the intricate inter-
so as to better reflect what really happens in reality. The dependencies of modern health care services, a more real-
latest example is the model developed by Brailsford et al. istic representation of the workings of a real system will
[29], which represented screening policies for breast can- be achievable from this integrative perspective. Instead of
cer via DES with the addition of behavioral factors to simplifying the multi-faceted treatment pathways of acute
each simulated patient’s characteristics. The earliest case bacterial skin and skin structure infections (ABSSSIs),
was published in 2003 and presented a screening model Revankar et al. [33] developed a DES framework which
for diabetic retinopathy focusing on attendance behavior captured the complete variations in locations of care from
through embedding individual emotion, cognition and emergency departments, through to outpatient and in-
social status elements [30]. patient units, which properly reflected the real-world
complexities related to such treatment administrations.
Discussion High-level decision supports can be provided via DES
This systematic review was undertaken to summarize models for hospital managers in terms of the diagnosis
the advance status and new trends of the application of of system inefficiencies and evaluation of alternative sys-
DES in health care over the last two decades and provide tem configurations [34–37]. Patient flow through a
a more quantitative overview of a larger number of DES multidisciplinary sleep center was constructed by Pend-
models compared to precedent reviews. harkar et al. with the support of DES [36]. Multiple sys-
tem constraints resulting in treatment access delays
Rarely applied at policy strategic level were recognized and alternative reconfigurations aimed
The extent to which DES modeling has been employed to at improving access to sleep care were tested.
deal with issues at the strategic level is relatively disappoint- Patient satisfaction lies in their comprehensive experi-
ing, which is possibly due in part to the intrinsic attribute ence throughout a whole integrated system. Consequently,
of this modeling technique. According to Mielczarek et al. to model an integrated provider as a whole is considered
[9], the DES framework lends itself better to simulate oper- necessary for improving patient overall experience [38].
ational details and physical mechanism of a specific process Bard et al. [39] applied DES to mimic patient flow through
instead of health policy evaluation and planning. Thus, a series of activities as a network in a Family Health Cen-
DES is a suitable tool when it comes to concrete oper- ter and gained deep insights into the influential factors
ational issues of health care delivery. Regarding public jeopardizing patient satisfaction such as excessively long
health policy evaluation, in contrast, System Dynamics (SD) hospitalization and high provider utilization.
is viewed as a more preferred modeling technique. Certainly there are a few pitfalls for DES modeling at
However, there are still some efforts of applying DES on this higher lever in health care. The most frequently
this highly aggregate level. One of these is the study con- mentioned would be the generalizability issue. These
ducted by Deo et al. [31], which evaluated the long-run model frameworks for integrated care providers were
merits and drawbacks of integrating outpatient and HIV usually populated by data collected from specific institu-
services in sub-Saharan Africa. Nguefack et al. built a DES tions during a certain historical period, which in this
Zhang BMC Health Services Research (2018) 18:687 Page 8 of 11

sense leads to limitations on the applicability of model- with DES through a health service system for speech
ing results into other settings or other periods of time language pathology and further tested the effectiveness
[34, 36]. In addition, lack of necessary data makes it in- of integrated policy changes which involved increasing
evitable to make assumptions about the systems mod- group treatments, paraprofessional utilization and so on.
eled or rely on expert estimates as well as data from Similarly to the shortcomings stated in section “More
literature instead of real observations. These alternatives efforts towards modeling integrated multi-facility pro-
naturally can help simplify the DES modeling practice, viders were identified”, these system-wide DES simula-
but, how and to what extent the validity and accuracy of tions also suffer limited generalizability and lack of
DES models will be affected should not be disregarded sufficient data needed. In addition, the construction of
[33, 40, 41]. macro-models is considered far more time-consuming
and data-intensive because of higher complexity and vast
Modeling of the continuum of entire health services for a numbers of activities involved [43]. Furthermore, synthe-
certain disease area emerged sizing of data from a range of distinct sources will lead
Health service pathways are a continuous process to additional uncertainty into parameters estimated [50].
throughout the whole health system involving a wide Another challenge mentioned by Katsaliaki et al. [51] is
range of care providers, from primary health care clinics the lack of consistent terminology. The same processes
(downstream), to highly specialized tertiary units (up- or services may be named quite differently across differ-
stream), caring for different phases of disease progres- ent parts of a health system, which can also give rise to
sion. Unexpectedly, a few DES models of this continuum some difficulties in modeling.
of care have been reported since 2005, imitating the in-
tegrated networks and interconnectedness among differ- Advantages of DES in health economic evaluation have
ent parts of the whole health care pathways for a certain been increasingly admired
disease area. According to the results of this review, health economic
The exploration of DES in this direction started in evaluation is the second most common theme, occupy-
2005, which, since then, opened up the possibilities of ing 36% (75) of all DES applications in health care, fol-
drawing the attention of health care modelers to a bigger lowing health operations research.
picture of the relationships between diverse components Although Markov models and decisions trees are still
of a health care continuum. These macro modeling en- the most frequently applied modeling techniques in
deavors are intended to deal with some critical chal- pharmacoeconomic evaluations, it is not hard to find the
lenges in which micro-system models can hardly play a arguments for the advantages of DES used in HEE that
role, for example, evaluating the cooperation and trans- highlight the valued potential of DES to overcome the
ferring efficiency [42, 43] and balancing the distribution inherent limitations of cohort Markov models and deci-
of scarce health resources within a health system [44]. sion trees.
Medical and administrative initiatives may be favorable One merit of DES is called “individual heterogeneity”,
for one specific section of a health system where they namely, each of the entities modeled is stochastically
will be conducted, but may conversely cause unintended assigned their own unique properties at the beginning of
damages to other parts, or lead bottlenecks to shift simulation [26, 52–54]. Moreover, these initial character-
within the system [42, 43, 45]. Therefore, applying DES istics can be updated according to events experienced
to evaluate the overall long-term performance of policy over time, which will in turn impact the possibilities of
changes from a scale perspective is considered appealing. the occurrence of following events [55]. Hartz et al. [52]
With the assistance of DES in modeling the care system argued that individual characteristics neglected by most
for Parkinson’s disease (PD) in the UK, Lebcir et al. [42] of cohort models might have profound impacts on the
reported the impact of an increased use of PD commu- long term outcomes predicted. The omission of indi-
nity services on the demand for hospital-based health- viduality could be viewed as an obvious deficiency for
care and resources. homogenous cohort models.
In addition, complex health care interventions now- Another noteworthy point of DES for HEE is its allowing
adays usually imply integrative programs, implementa- for “individual interaction”, which is valuable especially for
tions of which require the deployment of profound models involving analyzing the effects of treatment
organizational and systematic changes and improve- delays due to patient competition for constrained re-
ments. Accordingly, analyzing the potential benefits of sources [56, 57]. Standffield et al. [56] conducted a
such sophisticated interventions necessitates the em- cost-utility analysis using DES because of the need to
ployment of a systematic and holistic horizon [46–49]. consider the impact of introducing OPSC (Ortho-
As an example, a recent research presented by Wil- paedic Physiotherapy Screening Clinics and Multidis-
loughby et al. [48] mapped a system-wide patient flow ciplinary Services) on patients queuing time.
Zhang BMC Health Services Research (2018) 18:687 Page 9 of 11

Additionally, the advances of the DES-specific software increasing number of publications, particularly after
packages make it possible to communicate the frame- 2010. In addition, this extending trend could also been
works and results of HEE to lay decision makers in an proven by the fact that more than half of ICD-10 chap-
easier and more intuitive way, for example with different ters have been covered by DES studies. This could prob-
levels of animation [58]. The transparent presentation ably be viewed as an indication that DES models are
plays an important role in improving the understandabil- becoming more and more embraced in health care man-
ity and acceptability of model outputs to health care agement, but this point warrants further evidence of the
stakeholders [5, 59, 60]. value and extent of DES implementations in realistic
There are also other reasons for DES being chosen in health care decision making, which would be one of the
preference to other alternatives, such as less limitations on focuses of the following publication.
the number of events and health states to be represented In this survey, DES models were arguably classified into 4
[61, 62], higher flexibility, less computing intensity [63], main categories based on the nature of subjects modeled.
and event-based conceptualization attribute [55]. Taken This classification may need more discussions because of
together, DES models are increasingly considered as a the intrinsic complexity and overlapping of the applications
more practical and flexible modeling technique in case of of DES in health. Undoubtedly, health and care systems op-
complex and dynamic contexts in health care. erations were the dominant issues approached with the as-
sistance of DES, occupying a massive proportion and
Limitations expanding at the highest rate. The dominance of this lead-
There are several limitations of this review which need ing application area has also been demonstrated in previous
to be mentioned. Firstly, only PubMed and Web of Sci- comprehensive reviews on computer simulation in health
ence were searched as main databases. This seems to care, especially, the reviews conducted by Mielczarek et al.
limit the supply of target publications. However, given [9] and Fone et al. [10] which evidently presented quantita-
the significant number of studies included and no novel tive supporting evidence for the extensive use of DES in
application areas of DES in health care was further iden- health system operations.
tified in the later reviewing stage, it is reasonable to be- Although modeling of specific individual units still has
lieve that the majority of representative literature has a predominant role in all the HCSO areas, more scien-
been captured. tific endeavors to simulate more integrated health care
Secondly, the inclusion criteria could also be seen as a settings like hospitals or even the whole health service
limitation. Only peer-reviewed original DES studies in continuum have been identified than before, which im-
health care management, which unambiguously pre- plies the rich potential of DES to provide a broader pic-
sented model structures and mechanisms, were given ture of how health care systems behavior. Both of the
consideration. This set of criteria excluded a certain pro- precedent DES reviews by Jun et al. [7] and Günal et al.
portion of literature, especially papers compiled in pro- [8] also discussed the weaknesses of unit-specific piece-
ceedings of prestigious conferences. Nevertheless, these meal modeling as well as the significance of modeling at
criteria were deemed necessary in order to make sure the macroscopic and integrated level. Additionally, the
the modeling techniques employed in papers were in- causes and barriers of the limited number of higher-level
deed DES frameworks and also ensure consistency with DES models were also analyzed and delineated in both.
the suggestions by ISPOR-SMDM [6]. In addition, as Health economic evaluations via disease progression
discussed before, it was believed that the sufficient rep- modeling have become the second most common theme
resentativeness of the papers included was not affected among all DES applications in health care. This topic area
substantially in this review. was distinguished as one of major categories in the classifi-
Thirdly, this review is limited to a broad overview of cation systems proposed by Fone et al. [10] and Katsaliaki
applications of DES in health care delivery, but without et al. [64]. However, no significantly increasing trend in this
critically assessing their reporting quality. There is thus sphere has been discovered. The majority of these evalua-
a need for more detailed quantitative analyses which tions could be reasonably categorized as cost-effectiveness
were beyond the scope of this study. analysis, approximately half of which were carried out from
the perspective of health care payers.
Conclusion It is hoped that health policy developers could yield an
This systematic review has presented the versatility and up-to-date overall picture of the application of DES
new trends of DES models applied to deal with a broad across various health-related topic areas and get moti-
spectrum of issues facing health care administrators and vated to actively apply DES techniques to tackle the co-
stakeholders. A survey of 211 studies was conducted and nundrums they are facing. Regarding the quality and
the rapidly growing popularity of DES in health care sec- value of the research surveyed, a critical appraisal will be
tor could be reflected by the substantially and constantly demonstrated in the following publication.
Zhang BMC Health Services Research (2018) 18:687 Page 10 of 11

Abbreviations 9. Mielczarek B, Uzialko-Mydlikowska J. Application of computer simulation

ABSSSIs: Acute Bacterial Skin and Skin Structure Infections; DES: Discrete modeling in the health care sector: a survey. Simul-T Soc Mod m. 2010;
Event Simulation; DPM: Disease Progression Modeling; HBM: Health Behavior 88(2):197–216.
Modeling; HCSO: Health and Care Systems Operation; HEE: Health Economic 10. Fone D, Hollinghurst S, Temple M, Round A. Systematic review of the use
Evaluation; ICD-10: International Statistical Classification of Diseases and and value of computer simulation modelling in population health and
Related Health Problems 10th Revision; ISPOR: International Society for health care delivery. J Public Health Med. 2003;25(4):325–35.
Pharmacoeconomics and Outcomes Research; PD: Parkinson’s Diseases; 11. Caro JJ, Huybrechts KF, Xenakis JG, O'Brien JA, Rajagopalan K, Lee K.
PRISMA: Preferred Reporting Items for Systematic Reviews and Meta- Budgetary impact of treating acute bipolar mania in hospitalized patients
Analyses; QALY: Quality-adjusted Life Years; SD: System Dynamics; with quetiapine: an economic analysis of clinical trials. Curr Med Res Opin.
SM: Screening Modeling; SMDM: Society for Medical Decision Making 2006;22(11):2233–42.
12. Mielczarek B. Review of modelling approaches for healthcare simulation.
Acknowledgements Oper Res Decisions. 2016;26(1):55–72.
The author is grateful to Wolf H. Rogowski (WR) for the valuable feedback on 13. Lagergren M. What is the role and contribution of models to management
the draft as well as the suggestions provided during the reviewing process. and research in the health services a view from Europe. Eur J Oper Res.
The written comments of Fabia Mareike Gansen (FG) as well as the 1998;105(2):257–66.
assistance provided in the initial phase of literature selection are highly 14. Zhao YC, Peng QJ, Strome T, Weldon E, Zhang M, Chochinov A. Bottleneck
appreciated. detection for improvement of emergency department efficiency. Bus
Process Manag J. 2015;21(3):564–85.
Funding 15. Carmen R, Defraeye M, Van Nieuwenhuyse I. A decision support system for
The author is supported by China Scholarship Council (CSC). The Financial capacity planning in emergency departments. Int J Model Simul. 2015;14(2):
source was not involved in study design, review process, interpretation of 299–312.
results and writing of this manuscript. 16. Bessen T, Keefe DM, Karnon J. Does one size fit all? Cost utility analyses of
alternative mammographic follow-up schedules, by risk of recurrence. Int J
Technol Assess Health Care. 2015;31(5):281–8.
Availability of data and materials
The datasets used and/or analyzed during the current study are available 17. Stout NK, Rosenberg MA, Trentham-Dietz A, Smith MA, Robinson SM,
from the author on reasonable request. Fryback DG. Retrospective cost-effectiveness analysis of screening
mammography. J Natl Cancer Inst. 2006;98(11):774–82.
18. Lee CI, Cevik M, Alagoz O, Sprague BL, Tosteson AN, Miglioretti DL, et al.
Authors’ contributions Comparative effectiveness of combined digital mammography and
The sole author was responsible for all aspects of the study and manuscript. tomosynthesis screening for women with dense breasts. Radiology. 2015;
The author read and approved the final manuscript. 274(3):772–80.
19. Comas M, Arrospide A, Mar J, Sala M, Vilaprinyo E, Hernandez C, et al.
Ethics approval and consent to participate Budget impact analysis of switching to digital mammography in a
Not applicable. population-based breast cancer screening program: a discrete event
simulation model. PLoS One. 2014;9(5):e97459.
Consent for publication 20. Coelli FC, Ferreira RB, Almeida RM, Pereira WC. Computer simulation and
Not applicable. discrete-event models in the analysis of a mammography clinic patient
flow. Comput Methods Prog Biomed. 2007;87(3):201–7.
Competing interests 21. Berg B, Denton B, Nelson H, Balasubramanian H, Rahman A, Bailey A, et al. A
The author declares that he/she has no competing interests. discrete event simulation model to evaluate operational performance of a
colonoscopy suite. Med Decis Mak. 2010;30(3):380–7.
22. Campbell LA, Blake JT, Kephart G, Grunfeld E, MacIntosh D. Understanding
Publisher’s Note the effects of competition for constrained colonoscopy services with the
Springer Nature remains neutral with regard to jurisdictional claims in introduction of population-level colorectal Cancer screening. Med Decis
published maps and institutional affiliations. Mak. 2017;37(2):253–63.
23. Heim JA, Huang H, Zabinsky ZB, Dickerson J, Wellner M, Astion M, et al.
Received: 11 May 2018 Accepted: 8 August 2018 Design and implementation of a combined influenza immunization and
tuberculosis screening campaign with simulation modelling. J Eval Clin
Pract. 2015;21(4):727–34.
References 24. van Karnebeek CDM, Mohammadi T, Tsao N, Sinclair G, Sirrs S, Stockler S,
1. Karnon J, Stahl J, Brennan A, Caro JJ, Mar J, Moller J. Modeling using et al. Health economic evaluation of plasma oxysterol screening in the
discrete event simulation: a report of the ISPOR-SMDM modeling good diagnosis of Niemann-pick type C disease among intellectually disabled
research practices task Force-4. Med Decis Mak. 2012;32(5):701–11. using discrete event simulation. Mol Genet Metab. 2015;114(2):226–32.
2. Pidd M. Computer simulation in management science. 5th ed. New York: 25. Davies R, Roderick P, Canning C, Brailsford S. The evaluation of screening policies
John Wiley; 2004. for diabetic retinopathy using simulation. Diabet Med. 2002;19(9):762–70.
3. Brennan A, Chick SE, Davies R. A taxonomy of model structures for 26. Igarashi A, Goto R, Suwa K, Yoshikawa R, Ward AJ, Moller J. Cost-
economic evaluation of health technologies. Health Econ. 2006;15(12):1295– effectiveness analysis of smoking cessation interventions in Japan using a
310. discrete-event simulation. Appl Health Econ Health Policy. 2016;14(1):77–87.
4. Brailsford SC, Hilton NA, editors. A comparison of discrete event simulation 27. Getsios D, Marton JP, Revankar N, Ward AJ, Willke RJ, Rublee D, et al.
and system dynamics for modelling healthcare systems. Operational Smoking cessation treatment and outcomes patterns simulation: a new
Research Applied to Health Services (ORAHS). Glasgow: Glasgow Caledonian framework for evaluating the potential health and economic impact of
University; 2000. smoking cessation interventions. PharmacoEconomics. 2013;31(9):767–80.
5. Caro JJ. Pharmacoeconamic analyses using discrete event simulation. 28. Xenakis JG, Kinter ET, Ishak KJ, Ward AJ, Marton JP, Willke RJ, et al. A
PharmacoEconomics. 2005;23(4):323–32. discrete-event simulation of smoking-cessation strategies based on
6. Karnon J, Stahl J, Brennan A, Caro JJ, Mar J, Moller J, et al. Modeling using varenicline pivotal trial data. PharmacoEconomics. 2011;29(6):497–510.
discrete event simulation: a report of the ISPOR-SMDM modeling good 29. Brailsford SC, Harper PR, Sykes J. Incorporating human behaviour in
research practices task Force-4. Value Health. 2012;15(6):821–7. simulation models of screening for breast cancer. Eur J Oper Res. 2012;
7. Jun JB, Jocaboson SH, Swisher JR. Application of discrete-event simulation 219(3):491–507.
in health care clinics: a survey. J Oper Res Soc. 1999;25(4):325–35. 30. Brailsford S, Schmidt B. Towards incorporating human behaviour in models
8. Gunal MM, Pidd M. Discrete event simulation for performance modelling in of health care systems: an approach using discrete event simulation. Eur J
health care: a review of the literature. J Simul. 2010;4(1):42–51. Oper Res. 2003;150(1):19–31.
Zhang BMC Health Services Research (2018) 18:687 Page 11 of 11

31. Deo S, Topp SM, Garcia A, Soldner M, Yagci Sokat K, Chipukuma J, et al. 53. Erenay FS, Alagoz O, Banerjee R, Said A, Cima RR. Cost-effectiveness of
Modeling the impact of integrating HIV and outpatient health services on alternative colonoscopy surveillance strategies to mitigate metachronous
patient waiting times in an urban health clinic in Zambia. PLoS One. 2012; colorectal cancer incidence. Cancer. 2016;122(16):2560–70.
7(4):e35479. 54. Fines P, Garner R, Bancej C, Bernier J, Manuel DG. Development and
32. Nguefack HLN, Gwet H, Desmonde S, Oukem-Boyer OOM, Nkenfou C, implementation of microsimulation models of neurological conditions.
Tejiokem M, et al. Estimating mother-to-child HIV transmission rates in Health Rep. 2016;27(3):3–9.
Cameroon in 2011: a computer simulation approach. BMC Infect Dis. 2016; 55. Simpson KN, Pei PP, Moller J, Baran RW, Dietz B, Woodward W, et al.
16(11):1–11. Lopinavir/ritonavir versus darunavir plus ritonavir for HIV infection: a
33. Revankar N, Ward AJ, Pelligra CG, Kongnakorn T, Fan W, LaPensee KT. cost-effectiveness analysis for the United States. PharmacoEconomics.
Modeling economic implications of alternative treatment strategies for 2013;31(5):427–44.
acute bacterial skin and skin structure infections. J Med Econ. 2014; 56. Standfield L, Comans T, Raymer M, O'Leary S, Moretto N, Scuffham P. The
17(10):730–40. efficiency of increasing the capacity of physiotherapy screening clinics or
34. Huynh N, Snyder R, Vidal JM, Sharif O, Cai B, Parsons B, et al. Assessment of traditional medical services to address unmet demand in Orthopaedic
the nurse medication administration workflow process. J Healthc Eng. 2016; outpatients: a practical application of discrete event simulation with
2016:1–14. dynamic queuing. Appl Health Econ Health Policy. 2016;14(4):479–91.
35. Jahn B, Pfeiffer KP, Theurl E, Tarride JE, Goeree R. Capacity constraints and 57. Cooper K, Brailsford SC, Davies R. Choice of modelling technique for
cost-effectiveness: a discrete event simulation for drug-eluting stents. Med evaluating health care interventions. J Oper Res Soc. 2007;58(2):168–76.
Decis Mak. 2010;30(1):16–28. 58. Caro JJ, Möller J, Karnon J, Stahl J, Ishak J. Special Topics. Discrete Event
36. Pendharkar SR, Bischak DP, Rogers P, Flemons W, Noseworthy TW. Using Simulation for Health Technology Assessment. 1st ed. Boca Raton: CRC
patient flow simulation to improve access at a multidisciplinary sleep Press; 2016. p. 273–310.
Centre. J Sleep Res. 2015;24(3):320–7. 59. Southard PB, Chandra C, Kumar S. RFID in healthcare: a six sigma DMAIC
37. Fialho AS, Oliveira MD, Sa AB. Using discrete event simulation to compare and simulation case study. Int J Health Care Qual Assur. 2012;25(4):291–321.
the performance of family health unit and primary health care Centre 60. Langley I, Doulla B, Lin HH, Millington K, Squire B. Modelling the impacts of
organizational models in Portugal. BMC Health Serv Res. 2011;11:274. new diagnostic tools for tuberculosis in developing countries to enhance
38. Tako AA, Kotiadis K, Vasilakis C, Miras A, le Roux CW. Improving patient policy decisions. Health Care Manag Sci. 2012;15(3):239–53.
waiting times: a simulation study of an obesity care service. Bmj Quality & 61. Lee D, Wilson K, Akehurst R, Cowie MR, Zannad F, Krum H, et al. Cost-
Safety. 2014;23(5):373–81. effectiveness of eplerenone in patients with systolic heart failure and mild
39. Bard JF, Shu Z, Morrice DJ, Wang DE, Poursani R, Leykum L. Improving patient symptoms. Heart. 2014;100(21):1681–7.
flow at a family health clinic. Health Care Manag Sci. 2016;19(2):170–91. 62. Higashi H, Barendregt JJ. Cost-effectiveness of total hip and knee
40. Asamoah DA, Sharda R, Rude HN, Doran D. RFID-based information visibility replacements for the Australian population with osteoarthritis: discrete-
for hospital operations: exploring its positive effects using discrete event event simulation model. PLoS One. 2011;6(9):e25403.
simulation. Health Care Manag Sci. 2018;21(3):305–16. 63. Vataire AL, Aballea S, Antonanzas F, Roijen LH, Lam RW, McCrone P, et al.
41. Crawford EA, Parikh PJ, Kong N, Thakar CV. Analyzing discharge strategies Core discrete event simulation model for the evaluation of health care
during acute care: a discrete-event simulation study. Med Decis Mak. 2014; technologies in major depressive disorder. Value Health. 2014;17(2):183–95.
34(2):231–41. 64. Katsaliaki K, Mustafee N. Applications of simulation within the healthcare
42. Lebcir R, Demir E, Ahmad R, Vasilakis C, Southern D. A discrete event context. J Oper Res Soc. 2011;62(8):1431–51.
simulation model to evaluate the use of community services in the
treatment of patients with Parkinson’s disease in the United Kingdom. BMC
Health Serv Res. 2017;17(1):50.
43. Santos A, Gurling J, Dvorak MF, Noonan VK, Fehlings MG, Burns AS, et al.
Modeling the patient journey from injury to community reintegration for
persons with acute traumatic spinal cord injury in a Canadian Centre. PLoS
One. 2013;8(8):1–10.
44. Matta A, Li J, Sahin E, Lanzarone E, Fowler J. Editors. Proceedings of the
international conference on health care systems engineering. Int Conf
health care Syst Eng 2013; Milan, Italy. Bern: Springer; 2014.
45. Chemweno P, Brackenier L, Thijs V, Pintelon L, Van Horenbeek A, Michiels D.
Optimising the complete care pathway for cerebrovascular accident
patients. Comput Ind Eng. 2016;93:236–51.
46. Soto-Gordoa M, Arrospide A, Merino Hernandez M, Mora Amengual J,
Fullaondo Zabala A, Larranaga I, et al. Incorporating budget impact analysis
in the implementation of complex interventions: a case of an integrated
intervention for multimorbid patients within the CareWell study. Value
Health. 2017;20(1):100–6.
47. Monks T, Worthington D, Allen M, Pitt M, Stein K, James MA. A modelling
tool for capacity planning in acute and community stroke services. BMC
Health Serv Res. 2016;16(1):530.
48. Willoughby KA, Chan BTB, Marques S. Using simulation to test ideas for improving
speech language pathology services. Eur J Oper Res. 2016;252(2):657–64.
49. Bowers J, Mould G, Marshall C. Location of services and the impact on
healthcare quality: insights from a simulation of a musculoskeletal
physiotherapy service. J Oper Res Soc. 2015;66(7):1212–21.
50. Pilgrim H, Tappenden P, Chilcott J, Bending M, Trueman P, Shorthouse A,
et al. The costs and benefits of bowel cancer service developments using
discrete event simulation. J Oper Res Soc. 2009;60(10):1305–14.
51. Katsaliaki K, Brailsford S, Browning D, Knight P. Mapping care pathways for
the elderly. J Health Organ Manag. 2005;19(1):57–72.
52. Hartz S, Getsios D, Tao S, Blume S, Maclaine G. Evaluating the cost
effectiveness of donepezil in the treatment of Alzheimer's disease in
Germany using discrete event simulation. BMC Neurol. 2012;12:2.