Proceedings of the 2016 Winter Simulation Conference

T. M. K. Roeder, P. I. Frazier, R. Szechtman, E. Zhou, T. Huschka, and S. E. Chick, eds.

MODELING HEALTHCARE DEMAND USING A HYBRID SIMULATION APPROACH

Bożena Mielczarek
Jacek Zabawa

Department of Operational Research

Wrocław University of Science and Technology
Wyb. Wyspiańskiego 27
PL50-370 Wrocław, POLAND

ABSTRACT

This paper describes a hybrid simulation model that uses a system dynamics and discrete event simulation
to study the influence of long-term population changes on the demand for healthcare services. A dynamic
simulation model implements an aging chain approach to forecast the number of individuals who belong
to their respective age-sex cohorts. The demographic parameters that were calculated from a Central
Statistical Office Local Data Base were applied to the Wrocław Region population from 2002 to 2014,
and the basic scenario for the projected trends was adopted for a time horizon from 2015 to 2035. The
historical data on hospital admissions were obtained from the Regional Health Fund. A discrete event
model generates batches of patients with cardiac diseases and modifies the demand according to the
demographic changes that were forecasted by a population model. The results offer a well-defined
starting point for future research in the health policy field.

1 INTRODUCTION
The effectiveness of the performance of a healthcare system depends largely on an appropriate level of
financial and material resources. Although different countries are at different stages of economic
development and can afford different packages of care, in the case of any public healthcare system, the
number of provided services is permanently much lower than the growing demand, and the supply of
services is usually insufficient to meet all of the associated needs. A population’s health is therefore
determined not only by the quantity of available resources but also by their appropriate distribution. The
credible knowledge of a population’s needs is the key determinant that can enable the optimal allocation
of resources and provide the information that policy makers can use to improve people’s access to health
facilities.
In the application of quantitative modeling to support managerial decisions, healthcare needs are
usually converted into indicators of demand. The terms that are used in this process include descriptors of
the delivery of healthcare services such as volume, structure and dynamics. The most common
methodological approach assumes that the future demand for healthcare services may be properly
approximated by the past supply of the services that have been delivered to the patients; see, for example,
(Bowers et al. 2009; Steins et al. 2010; Matta and Patterson 2007). However, this is not always the best
solution that can be applied. According to Roberfroid et al. (2009), this approach is justified only when
the current supply of healthcare services is appropriate and the population’s demography changes
according to currently observed trends. It is widely recognized that the population’s needs for healthcare
services are highly influenced by age-gender demographic profiles such as the proportion of elderly

978-1-5090-4486-3/16/$31.00 ©2016 IEEE 1535

 Mielczarek and Zabawa

people, average expected length of life, birth and death rates, and the number of people of working age
(Ansah et al. 2014). For example, Barber and Lopez-Valcarcel (2010) simulated a demographic pyramid
to analyze the demand for medical specialties in Spain; Masnick and McDonnel (2010) modelled
population evolution to link individuals with health conditions to clinical workload; Lagergren (2005)
projected the population of older persons to estimate future needs for publicly financed long-term care.
Furthermore, the supply projection approach does not consider the important uncertainties that could
influence future demand, such as the uncertainty of future incidence rates of specific diseases. The
random and uncertain factors that play a significant role in healthcare management issues and influence
these types of external and internal factors on the overall demand for healthcare services should be
considered (Cardoso et al. 2012).
The factors that are discussed above have an impact on the undesirable accumulation of demand,
despite the fact that there is a satisfactory average supply level (calculated for the specific period). This in
turn may lengthen the waiting time that is required for a particular service and generate additional costs.
The overall goal of our project is to develop the general approach for predicting the demand for
healthcare services. In an attempt to extend our previous studies (Mielczarek et al. 2014; Mielczarek
2013), we searched for new solutions that would enable the preservation of the unique and valuable
features of the discrete simulation with the possibility of applying a holistic analysis of the problem. We
decided to build a hybrid simulation model to capture different aspects of demand for healthcare services,
in particular the effects of ongoing demographic changes, the uncertainty that surrounds the key
determinants of future incidence rates and the healthcare indicators that vary geographically. The
submodel, which was developed according to a continuous system dynamics (SD) paradigm, describes
population evolution. The discrete event simulation (DES) submodel generates data that can facilitate an
assessment of the demand for healthcare services on the regional level.
This paper presents the results of an experiment that was conducted to analyze the effects of long-
term demographic changes on future demand for healthcare services. We outline the overall idea of the
hybrid model and discuss the initial findings of the simulation of population projections. The remainder
of the paper is organized as follows: section 2 provides background information about healthcare demand
modeling and outlines the conceptual framework for using hybrid simulation. The methodology and the
details of the model are presented in section 3. The features of the experiment and the analysis of the
results are presented in section 4. Finally, section 5 ends the paper with discussion and conclusions.

2 BACKGROUND
Simulation plays a vital role in healthcare decision making, and healthcare systems have benefited greatly
from the use of simulation (Gul and Guneri 2015, Katsaliaki and Mustafee 2011). Simulation methods
that are applied to the healthcare sector are categorized in various ways, but they are most commonly
classified (Brailsford et al. 2009; Sobolev et al. 2011; Mielczarek and Uziałko-Mydlikowska 2012;
Marshall et al. 2015) into four categories: Monte Carlo (MC), discrete-event simulation (DES), system
dynamics (SD) and agent-based simulation (ABS). In the selection of a simulation method, the most
central consideration is the type of the problem that is being investigated. For example, when modelling
emergency departments, discrete-event simulation is definitely the preferred technique (Gul and Guneri
2015). In contrast, models of epidemics and disease prevention are usually built using the system
dynamics approach (Homer and Hirsch 2006). ABS is used to study systems for which the consequences
on the collective level are not predictable although the modeler is able to describe the behavior of
individuals (Kasaie et al. 2010).
The estimation of the future demand for healthcare services is crucial in addressing the majority of
the decision support management problems in healthcare systems. Mielczarek (2014) demonstrated that
the issue of modeling and forecasting healthcare demand is present: in diagnosing and improving the
performance of a healthcare system; in studying the cost-effectiveness and/or the clinical effectiveness of
medical procedures, in connection with medical treatments that are associated with clinical pathways, and

1536
 Mielczarek and Zabawa

in the development of prevention strategies or contemporary health trends; and as support for decision
makers who are engaged in the capacity planning process at the regional or national level.
In the projection of the demand for healthcare services, population demography and epidemiological
estimates of prevalence are usually considered. For example, long-term care (Ansah et al. 2014) or highly
age-related conditions such as dementia (Jagger et al. 2009) require that population projections with
demand modeling are linked. So-called cohort modeling enables the representation of chronological
ageing, i.e., the process that describes the dynamic movement of people from one population group to
another over time. However, there is also a need to include more individual-specific components in the
model, such as a region’s socio-economic profiles, temporal factors (i.e., time of day, day of the week,
season, and calendar year) or geographical characteristics (i.e., the location of healthcare providers,
patients’ place of residence) and other factors. The influence of different types of uncertainty and
randomness should also be considered (Cardoso et al. 2012). The neglect of this type information can lead
to forecasts that are constructed without links to the specifics of the observed changes in demographics, as
well as the epidemiological, geographical and health related structures of the population. An incorrectly
estimated demand could lead to the erroneous assumptions and could significantly lower the explanatory
power of demand projections. Consequently, this may result in accepting a solution for which the planned
supply of services far from meets the current demand.
Demand modeling is performed using system dynamics (Desai et al. 2008), discrete-event (Vissers et
al. 2007) or agent-based simulation (Taboada et al. 2011). We assumed that the quantitative and
qualitative factors as well as the deterministic and stochastic variability should all be present in the model.
Such a possibility is offered by the hybrid structure that has already proven to be successful in many
fields and has obtained benefits that were unattainable with the use of only one of the components alone.
In hybrid modeling, different elements of a system are modelled by different simulation and/or analytical
paradigms (Viana 2014). Balaban (2014) used a multi-method simulation approach to model the return-
to-work behavior of people with disabilities. Gao et al. (2014) developed a hybrid simulation model to
project the cost and health impacts of diabetic end stage renal disease. Crowe et al. (2015) used
simulation and analytical modeling to study certain aspects of a pediatric heart transplantation program.
According to Djanatliev and German (2015), the separate use of simulation techniques is not sufficiently
powerful to solve large-scaled problems. The authors also suggest the use of a hybrid simulation term if a
clear distinction between continuous and discrete methods is of particular importance.
Healthcare demand modeling requires the integration of two opposite perspectives in the simulation
model. The projections of long-term population evolutions are performed with the aggregated data and
focus on pre-specified age-sex cohorts. These groups are described using such demographic parameters as
birth and death rates, life expectancy, and migration descriptors. The common approach to the study of
demographic trends is system dynamics (SD), which is a method that focuses on the dynamic analysis of
complex phenomena. SD uses a holistic perspective to describe a system by means of a set of stocks and
flows, unique theoretical constructions that are capable of accumulating objects (i.e., individuals from a
specific age-sex group) and pushing them from one reservoir to another during a specified time period (as
individuals age during the natural passage of time). The opposite perspective, which is based on discrete
modeling, concentrates on individuals, i.e., patients entering the healthcare system. The model keeps track
of the patients’ movements, tries to capture their individual choices, and attempts to incorporate the
dynamic and uncertain factors that influence the time and place of admission and the type of services that
are provided to patients.

3 METHODS

3.1 Overview of the Model

The initial population for the model was the 2002 Wrocław Region (WR), and the revised population
estimates were extracted from the data that were published by the Polish Central Statistical Office (GUS,

1537
 Mielczarek and Zabawa

2015). The WR belongs to Lower Silesia, which is the fourth largest region in Poland. From among the
five subregions that make up the Lower Silesia voivodship, the central area, which consists of the capital
(i.e., the city of Wrocław) and another subregion that encompasses the administrative districts and which
are located near the capital, were selected for our study.
The general concept of the hybrid SD–DES model is shown in Figure 1. The main objective is to
estimate the level and the structure of the exposed and unexposed demand for healthcare services. The
key element is the inclusion of the factors for which the incidence, intensity and consequences are
uncertain in relation to demography, human behavior, time and prevalence. The SD model uses an aging
chain approach to forecast the demographic changes that will be observed within the WR population over
next 20 years. The DES model generates batches of patients with a certain disease (e.g., patients with
suspected cardiovascular disease) who will arrive at the WR hospitals. The model takes into account
temporal (month, year), spatial (place of residence, location of the healthcare unit) and epidemiological
(incidence rates) factors. The final volume of the demand is determined by integrating the outputs of the
two submodels.

SD population model

0_4 5_19 20_39 40_59 60+

CV patients: historic arrivals Patients belonging

to different CV patients: projected
from 9 WR districts demand from 9 WR districts
age-sex cohorts

DES model

Figure 1: The general concept of the hybrid simulation model.

3.2 Population Submodel

The SD population submodel consists of two aging chains, with five age-sex cohorts in every chain. A
detailed description may be found in (Mielczarek and Zabawa 2016). The stocks represent individuals
who belong to a certain age-sex group, while the flows are responsible for the dynamic behavior of the
population. Input flows, i.e., births and immigrations, increase the number of individuals inside the
cohort; and output flows, such as deaths and emigrations, decrease this number. There are five cohorts
inside each aging chain: F0_4; F5_19; F20_39; F40_59; F60+ and M0_4; M5_19; M20_39; M40_59;
M60+. For example, the cohort F20_39 describes women at the ages between 20 and 39 years old. The
cohorts are connected by the maturation input-output flows (see Figure 2). The movement between
cohorts may be described by the time in cohort, i.e., the time that each individual needs to “live” from the
moment at which she enters the younger cohort until the moment at which she moves to the older one.
The demographic variables are assumed to be exogenous. There are four main demographic variables:
fertility, mortality, net migration and life expectancy. The last variable applies only to cohorts F60+ and

1538
 Mielczarek and Zabawa

M60+, and it measures the average time that an individual at age 60 is expected to live based on sex and
current age.
The members of each age cohort are updated on every computational interval so that the changes
inside the cohorts are registered not once a year but on a continuous basis within a year. The time-step
mechanism, which operates at discrete moments, aggregates all of the input and output flows that move to
and from a particular cohort into one dynamic object. The singular resultant flow instantly increases or
decreases the number of individuals in the cohort.

Figure 2: Female aging chain: five stocks (cohorts) with input (births), output (deaths) and input-output
flows (migrations).

3.3 Arrivals Submodel

We used dynamic random Poisson processes to model patient arrivals to the WR healthcare facilities.
There were nine input patient flows (see Figure 3) describing cardiovascular (CV) patients incoming from
nine WR districts.

no 5
CV patient in
population cohorts
no 1

no 6
WR districts

no 2
CV 0_4 CV 20_39 CV 60+ CV patients: projected
no 7 demand from 9 WR
no 3
districts
CV 5_19 CV 40_59
no 8

no 4
no 9

Figure 3: Patients with CV health problems arriving at the WR healthcare facilities.

The parameters for the exponential distributions with time-varying parameters were defined
according to the results of the historical data analysis. All input parameters were based on the data sample
from the Lower Silesian Health Fund branch registry. The flows of CV patients fluctuate considerably

1539
 Mielczarek and Zabawa

depending on the calendar month, and their intensity is highly dependent on the number of older people
who live in particular districts. The output of the model predicted the number of CV patients who live in
nine WR districts and will create the demand for healthcare services in the next few years.

3.4 Data
The simulation begins in 2002 and runs through 2014 according to parameters that were estimated on the
basis of CSO (GUS, 2015) data. Beyond 2015, the input values were estimated based on the different
scenarios according to (Waligórska et al. 2014). The goal of the simulation was to explore the past and
future structure of the WR population and to generate numerical forecasts that can be used by the DES
model. Table 1 presents the basic set of parameters calculated based on the historical data. The incidence
rates that describe the number of WR patients with CV disease are based on the statistics from the Lower
Silesia Health Fund (NFZ) branch registry from 2010 and 2011. The arriving patients were defined by
two descriptive characteristics: age-sex group and place of residence (district code).

Table 1: Historical population parameters calculated based on (GUS, 2015) for the WR.
Parameter 2002 2008 2014
Total female population 606158 613093 633074
Population of children F0_4 23974 26924 30527
The oldest population F60+ 124988 136798 165694
Fertility rates (F) 2.31% 2.99% 2.85%
Death rates (F0_4; F40_59) 0.23%; 0.32% 0.32%; 0.38% 0.07%; 0.29%
Migration rates (F20_39) 0.21% 0.28% 0.43%
Life expectancy (F60+) 22.15 23.18 24.40

Total male population 558302 559861 579707

Population of children M0_4 25452 28093 32560
The oldest population M60+ 78835 87899 113164
Fertility rates (M) 2.55% 3.09% 3.08%
Death rates (M0_4; M40_59) 0.22%; 0.84% 0.17%; 1.00% 0.11%; 0.73%
Migration rates (M20_29) 0.14% 0.20% 0.39%
Life expectancy (M60+) 17.19 18.05 19.49

3.5 SD Model Testing

The model was calibrated to determine the optimal values of times in cohorts (so-called maturation
lengths). The values of the maturation lengths were adjusted using the ExtendSim optimizer. The primary
optimization function was to minimize the total differences between the number of females (and,
respectively, males) from historical and simulation data in cohorts 0_4, 5_19, 20_39 and 40_59. The
second optimization function was to minimize the total differences between the total number of women
(and men, respectively) from the historical and simulation data in 60+ cohorts (Figure 4). The results of
the calibration were acceptable. The total number of the WR population was simulated with a high level
of accuracy. The Mean Percentage Error (MPE) differs from -1.49% to 1.96%. The discrepancies in the
MPE values for particular cohorts were at an acceptable level: from -4.87% to 9.88% for the female
cohorts and from -5.72% to 10.93% for the male cohorts.
It should be emphasized that the optimization of maturation lengths was necessary because the model
must consider the diverse ages of the individuals within the same cohort. In addition, the time ranges that
were defined for the particular cohorts differed from 5 to 20 years and this differentiation in the sizes of
the cohorts adversely affects the correctness of the chronological aging. Another problem is caused by the
intermediate inflows and outflows that dynamically increase or decrease the quantity within the stocks

1540
 Mielczarek and Zabawa

and disrupt the flow between the age cohorts. This blending problem lead to the differences between the
theoretical values of the parameters time in cohort and the adopted values of maturation length. The time
that is needed to move from one cohort to another takes on average longer than the time in cohort variable
would indicate (see Table 2). The life expectancy parameter was also adjusted. The last two cohorts (F60+
and M60+) describe individuals who are older than 60 years. Life expectancy, as defined for a certain
point of time, differs for a woman at age 61 and a woman at age 81. Therefore, during the optimization
process, the values of numerical multipliers were found and applied.

Figure 4: Mean Percentage Errors (MPE) calculated between historical and simulation data for the
particular age-sex cohorts of the WR population.

Table 2: The comparison of the parameters describing the delays in the movement of individuals between
cohorts. Values extracted from the data and calculated through the optimization process.
Cohort data [years] model [years]
F M F M
Time in cohort Maturation length
0_4 5 5 6.1 6
5_19 15 15 13.5 12
20_39 20 20 37 30
40_59 20 20 32 27
Life expectancy
60+ n m n x 1.9 m x 1.3

3.6 Simulation Scenario

We used the population projections for the period from 2015 to 2035 as published by the Polish
Government. According to (Waligórska et al. 2014) four alternative scenarios forecast the demographic
changes for Poland over the next 20 years. The scenarios assume different values of four basic
parameters: births, deaths, migrations and life expectancy. We formulated the demographic assumptions
for the WR according to the most likely scenario that was predicted for the whole country. It should be
emphasized, however, that the governmental forecasts are published for the larger administrative regions
(voivodships), such as, for example, the Lower Silesia. Our simulation study focused on two small
regions (out of five) that belong to the Lower Silesia voivodship. These two regions are classified as an
urban area, and the global population parameters are highly differentiated for rural and urban regions.
There are four parameters that are used in the simulation models that reflect possible future
population changes: fertility rates, which are calculated as the percentage of the number of births in
relation to the F20_39 cohort; death rates, which are calculated as the number of deaths in relation to a

1541
 Mielczarek and Zabawa

particular cohort; migration balance, which is calculated as the resulting balance of immigrations over
emigrations in relation to a particular cohort, and average life expectancy, which is linked to the last two
cohorts. The basic scenario makes the following assumptions for the period from 2014 to 2035:

The fertility rates will first slightly decrease and then gradually increase by approximately
15.55% (females) and 13.50% (males).
Death rates will increase gradually; however, the number of deaths in the middle-aged cohorts
(F40-59 and M40-59) will slightly decline. The growth rates of deaths will be higher in urban
areas compared to rural regions.
The difference between the international and internal net migrations will decrease to almost zero;
however, the total number of immigrating and emigrating populations will decrease by
approximately 20%.
Women and men will live longer; however, the differences in life expectancy between Poland and
European countries will remain at the same levels. In 2035, a woman age 60 will live on average
for 27.75 years, and a man at the same age will live on average for 24.27 more years.

Under the basic scenario, it was assumed that the age-specific prevalence of CV disease remains the
same during the forecasted time horizon. However, we plan to explore other trends during the next phase
of the study.

4 RESULTS
The simulation begins in 2002 and continues through to 2014 and runs according to the parameters that
were calculated based on the historical data. Next, from 2015 on, the values of the parameters are
projected according to the population scenario that is described in section 3.6. The simulation reveals the
important demographic trends for the WR population for the next 20 years. The total number of
individuals from the 60+ cohort, both women and men, will systematically rise (Figure 5); however, the
difference between the volume of the oldest men’s population in relation to the oldest women’s
population will slowly decrease. For example, in 2002 the M60+ cohort consisted of 63.34% of the F60+
age-sex group, and in 2035, this ratio will rise to 69.08%.

Figure 5: Forecasted trend of the WR population within two age groups: the middle-aged and the oldest,
separately for women and men (simulation data).

The simulation shows that population aging will result in a 23.2% (women) and 25.6% (men) increase
in the number of people aged 60+ over the next 20 years. The youngest cohorts, i.e., small children and
adolescents from 5 to 19 years old, will systematically decrease from 2002 to 2020; however, in 2020,
this drop will stop, and a small but continuous increase is observed (Figure 6). The number of the

1542
 Mielczarek and Zabawa

population between 40 and 59 years old (Figure 5) will fluctuate: a clear upward trend is observed for the
female cohort F40_59; however, in the case of the male cohort M40_59, the falling trend will start in
2029. The forecasted changes in the structure of the WR population, both for women and men age-sex
groups, are presented in Figure 6.

Figure 6: Changes of structure of the WR population by age cohorts, separately for women and men
(simulation data).

Under the assumption that the proportion of CV patients per age group remains unchanged during the
forecasted time horizon, the global trends for the population are well reflected in the decreases and
increases of the prevalence of CV patients. It can be seen from Figure 7 that the trend in the prevalence of
CV patients is upwards except for the M40_59 cohort. The older population (F60+ and M60+) will
generate increasing demands, and the growth of CV patients between 2014 and 2035 will be more intense
among the male population 60+ than the female 60+ (34.4% and 30.15%, respectively). The middle aged
cohorts (F40_59 and M40_59) will produce a slightly increased demand; however, this growth will not be
extensive, and the number of CV patients will remain at a stable level.

Figure 7: Forecasted trend of CV patients in the WR population within two age groups: the middle-aged
and the oldest, separately for women and men (simulation data).

5 DISCUSSION AND CONCLUSIONS

The population model that is presented in this paper represents a step towards a hybrid simulation with
the overall goal to forecast future volume of healthcare demands within different groups of patients. The
valid prediction of the demand for healthcare services may contribute to the improvement of the quality of

1543
 Mielczarek and Zabawa

the decisions that are made to properly distribute the available resources and implement policies that aim
to improve people’s health. However, in Poland, no data have been collected on the regional level that
show how services are provided in relation to certain measures of needs-for-service. There are also no
published studies that show how the care is allocated within the region and the grounds for long-term
plans that have been prepared to cover the forecasted population demand.
The basic concept of a computer simulation model assumes that it is possible to estimate future
demand by taking into account the past supply of services and the projected trends of population
demography. We believe that both types of healthcare needs, i.e. the demands exposed and recorded in
healthcare system, as well as the demands that was not revealed and therefore cannot be found in NHF
registries, may be estimated with the acceptable accuracy.
This study is an attempt to link the demographic trends with the regional volume of the demand for
healthcare services. The major strength of our approach lies in the ability to integrate different simulation
approaches and data from different sources. A dynamic simulation model implements an aging chain
approach to forecast the number of individuals who belong to age-sex cohorts. A discrete event model
generates batches of patients with cardiac diseases based on the historical data from the Regional Health
Fund and modifies the demand according to demographic changes that have been forecasted by a
population model. To the best of our knowledge, only a few studies have been conducted to quantify the
relationship between expected patient volume and demographic, environmental, socio-economic and
geographical variables. The issues of random and uncertain factors, such as the needs that originate in
different geographical regions or forecasted changes in the future morbidity rates, have not been
extensively studied in relation to long-term care planning.
The model described in the paper has its limitations. For example, it assumes a given and stable level
of the proportion of CV patients per age group and the absence of technological progress effect. Further
research is needed to more deeply examine the relationship between WR demographic parameters and CV
prevalence indicators. It is clear that these trend extrapolations should be modified according to different
scenarios.

ACKNOWLEDGMENTS

This work, Simulation modeling of the demand for healthcare services, was financed by the grant from
the National Science Centre, Poland, which was awarded based on the decision 2015/17/B/HS4/00306.

REFERENCES

Ansah, J. P., R. L. Eberlein, S. R. Love, M. A. Bautista, J. P. Thompson, R. Malhotra, and D. B. Matchar.

2014. "Implications of Long-term Care Capacity Response Policies for an Aging Population: A
Simulation Analysis." Health Policy 116(1):105–113.
Balaban, M. 2014. "Return to Work Behavior of People with Disabilities: A Multi-method Approach." In
Proceedings of the 2014 Winter Simulation Conference, edited by A. Tolk, S. D. Diallo, I. O.
Ryzhov, L. Yilmaz, S. Buckley, and J. A. Miller, 1561-1572. Piscataway, New Jersey: Institute of
Electrical and Electronics Engineers, Inc.
Barber, P., and B. G. Lopez-Valcarcel. 2010. "Forecasting the Need for Medical Specialists in Spain:
Application of a System Dynamics Model." Human Resources for Health 8(24):1–9.
Bowers, J., M. Ghattas, and G. Mould. 2009. "Success and Failure in the Simulation of an Accident and
Emergency Department." Journal of Simulation 3(3):171–178.
Brailsford, S. C., P. R. Harper, B. Patel, and M. Pitt. 2009. "An Analysis of the Academic Literature on
Simulation and Modelling in Health Care." Journal of Simulation 3(3):130–140.
Cardoso, T., M. Oliveira, A. Barbosa-Póvoa, and S. Nickel. 2012. "Modeling the Demand for Long-term
Care Services under Uncertain Information." Health Care Management Science 15(4):385–412.

1544
 Mielczarek and Zabawa

Crowe, S., S. Gallivan, and C. Vasilakis. 2015. "Informing the Management of Pediatric Heart Transplant
Waiting Lists: Complementary Use of Simulation and Analytic Modeling." In Proceedings of the
2015 Winter Simulation Conference, edited by L. Yilmaz, W. K. V. Chan, I. Moon, T. M. K. Roeder,
C. Macal, and M. D. Rossetti, 1654–1665. Piscataway, New Jersey: Institute of Electrical and
Electronics Engineers, Inc.
Desai, M. S., M. L. Penn, S. Brailsford, and M. Chipulu. 2008. "Modelling of Hampshire Adult Services--
gearing up for future demands." Health Care Management Science 11(2):167–76.
Djanatliev, A., and R. German. 2015. "Towards a Guide to Domain-Specific Hybrid Simulation." In
Proceedings of the 2015 Winter Simulation Conference, edited by L. Yilmaz, W. K. V. Chan, I.
Moon, T. M. K. Roeder, C. Macal, and M. D. Rossetti, 1609-1620. Piscataway, New Jersey: Institute
of Electrical and Electronics Engineers, Inc.
Gao, A., N. D. Osgood, W. An, and R. F. Dyck. 2014. "A Tripartite Hybrid Model Architecture for
Investigating Health and Cost Impacts and Intervention Tradeoffs for Diabetic End-stage Renal
Disease." In Proceedings of the 2014 Winter Simulation Conference, edited by A. Tolk, S. D. Diallo,
I. O. Ryzhov, L. Yilmaz, S. Buckley, and J. A. Miller, 1676–1687. Piscataway, New Jersey: Institute
of Electrical and Electronics Engineers, Inc.
Gul, M., and A. F. Guneri. 2015. "A comprehensive Review of Emergency Department Simulation
Applications for Normal and Disaster Conditions." Computers & Industrial Engineering 83:327–344.
GUS. 2015. Główny Urząd Statystyczny. Accessed December 26, 2015. www.stat.gov.pl.
Homer, J. B., and G. B. Hirsch. 2006. "System Dynamics Modeling for Public Health: Background and
Opportunities." American Journal of Public Health 96(3):452–458.
Jagger, C., R. Matthews, J. Lindesay, T. Robinson, P. Croft, and C. Brayne. 2009. "The Effect of
Dementia Trends and Treatments on Longevity and Disability: a Simulation Model Based on the
MRC Cognitive Function and Ageing Study (MRC CFAS)." Age and Ageing 38(3):319–325.
Kasaie, P., W. D. Kelton, A. Vaghefi, and S. G. R. J. Naini. 2010. "Toward Optimal Resource Allocation
for Control of Epidemics: An Agent-Based Simulation Approach." In Proceedings of the 2010 Winter
Simulation Conference, edited by B. Johansson, S. Jain, J. Montoya-Torres, J. Hugan, and E.
Yücesan, 2237–2248. Piscataway, New Jersey: Institute of Electrical and Electronics Engineers, Inc.
Katsaliaki, K., and N. Mustafee. 2011. "Applications of Simulation within the Healthcare Context", The
Journal of the Operational Research Society 62(8):1431–1451.
Lagergren, M. 2005. "Whither Care of Older Persons in Sweden?—a Prospective Analysis Based upon
Simulation Model Calculations, 2000–2030." Health Policy 74(3):325–334.
Marshall, D. A., L. Burgos-Liz, M. J. IJzerman, W. Crown, W. V. Padula, P. K. Wong, K. S. Pasupathy,
M. K. Higashi, and N. D. Osgood. 2015. "Selecting a Dynamic Simulation Modeling Method for
Health Care Delivery Research – Part 2: Report of the ISPOR Dynamic Simulation Modeling
Emerging Good Practices Task Force." Value in Health 18(2):147–160.
Masnick, K., and G. McDonnell. 2010. "A Model Linking Clinical Workforce Skill Mix Planning to
Health and Health Care Dynamics." Human Resources for Health 8(11):1–10.
Matta, M. E., and S. Patterson. 2007. "Evaluating Multiple Performance Measures Across Several
Dimensions at a Multi-Facility Outpatient Center." Health Care Management Science 10(2):173–194.
Mielczarek, B. 2013. "Estimating Future Demand for Hospital Emergency Services at the Regional
Level." In Proceedings of the 2013 Winter Simulation Conference, edited by R. Pasupathy, S.-H.
Kim, A. Tolk, R. Hill, and M. Kuhl, 2386–2397. Piscataway, New Jersey: Institute of Electrical and
Electronics Engineers, Inc.
Mielczarek, B., J. Zabawa, and M. Lubicz. 2014. "A System Dynamics Model to Study the Impact of an
Age Pyramid on Emergency Demand." In Proceedings of the 4th International Conference on
Simulation and Modeling Methodologies, Technologies and Applications SIMULTECH 2014, edited
by M. S. Obaidat, J. Kacprzyk, and T. Oren, SciTePress, Vienna, Austria, 879 –888.

1545
 Mielczarek and Zabawa

Mielczarek, B. 2014. "Simulation Modelling for Contracting Hospital Emergency Services at the
Regional Level." European Journal of Operational Research 235(1):287–299.
Mielczarek, B., and J. Uziałko-Mydlikowska. 2012. "Application of Computer Simulation Modeling in
the Health Care Sector: a Survey." Simulation - transactions of the society for modeling and
simulation international 88(2):197–216.
Mielczarek, B., and J. Zabawa. 2016. "Modelling Population Growth, Shrinkage and Aging Using a
Hybrid Simulation Approach: Application to Healthcare." In Proceedings of the 6th International
Conference on Simulation and Modeling Methodologies, Technologies and Applications
SIMULTECH 2016, Lisbon, Portugal, accepted paper.
Roberfroid, D., C. Leonard, and S. Stordeur. 2009. "Physician Supply Forecast: Better than Peering in a
Crystal Ball?" Human Resources for Health 7:1–13.
Sobolev, B. G., V. Sanchez, and C. Vasilakis. 2011. "Systematic Review of the Use of Computer
Simulation Modeling of Patient Flow in Surgical Care." Journal of Medical Systems 35(1):1–16.
Steins, K., F. Persson, and M. Holmer. 2010. "Increasing Utilization in a Hospital Operating Department
Using Simulation Modeling." Simulation 86(8–9):463–480.
Taboada, M., E. Cabrera, M. L. Iglesias, F. Epelde, and E. Luque. 2011. "An Agent-Based Decision
Support System for Hospitals Emergency Departments." Procedia Computer Science 4:1870–1879.
Viana, J. 2014. "Reflections on Two Approaches to Hybrid Simulation in Healthcare." In Proceedings of
the 2014 Winter Simulation Conference, edited by A. Tolk, S. D. Diallo, I. O. Ryzhov, L. Yilmaz, S.
Buckley, and J. A. Miller, 1585–1596. Piscataway, New Jersey: Institute of Electrical and Electronics
Engineers, Inc.
Vissers, J. M. H., I. J. B. F. Adan, and N. P. Dellaert, N.P. 2007. "Developing a Platform for Comparison
of Hospital Admission Systems: An Illustration." European Journal of Operational Research
180(3):1290–1301.
Waligórska, M., Z. Kostrzewa, M. Potyra, and L. Rutkowska. 2014. Population Projection 2014–2050.
CSO, Demographic Surveys and Labour Market Department.

AUTHOR BIOGRAPHIES

BOŻENA MIELCZAREK is an Associate Professor in the Department of Operational Research at

Wrocław University of Science and Technology, Poland. She received an MSc in Management Science
and a PhD in Economics from Wrocław University of Science and Technology. Her research interests
include simulation modeling, health-service research, decision support and financial risk analysis. Her
email address is [email protected].

JACEK ZABAWA is an Assistant Professor in the Department of Operational Research at Wrocław

University of Science and Technology, Poland. He received an MSc in Management Science and a PhD
in Economics from Wrocław University of Science and Technology. His research interests include
financial statement projections and application of simulation in management, with particular emphasis on
discrete rate modeling. His email address is [email protected].

1546