Nemesio Rodrigues Capocci
Faculdade de Tecnologia de Guarulhos, Brazil
E-mail: nemesio941@hotmail.com
Bárbara Soares Nascimento
Faculdade de Tecnologia de Guarulhos, Brazil
E-mail: barbara_elite@hotmail.com
Filipe Brito Lopes
Faculdade de Tecnologia de Guarulhos, Brazil
E-mail: jdfilipe.bl@gmail.com
Enio Fernandes Rodrigues
Instituto Federal de São Paulo, Brazil
E-mail: eniofr@uol.com.br
João Roberto Maiellaro
Faculdade de Tecnologia de Guarulhos, Brazil
E-mail: joao.maiellaro@fatec.sp.gov.br
Submission: 03/01/2017
Accept: 15/01/2017
ABSTRACT
This
study aims to demonstrate the use of the discrete event simulation technique as
a hospital management support tool, as well as all complex processes existing
in a health unit. There must be an analysis of the system as a whole from the
perspective of service level provided to patients regarding waiting times. The
role of this technique is to show the behavior of a given system. Data were
collected from employees of a public Polyclinic, located in a city of the
greater São Paulo, by means of interviews which questions were prepared to
determine the time spent in the processes of the service system. Such data were
inserted in the software Arena in flowchart format for analysis and
identification of the problem. Since the person responsible for the screening
process was overloaded, thus causing longer waiting times for patients
submitted for screening, some changes were made in the model in order to
propose an improvement, to balance the occupancy levels of the health unit’s
staff and, at the same time, reach a shorter withdrawal period of patients
throughout the system. Results showed a significant improvement in the
performance of the Polyclinic’s system, as well as a subsequent improvement in
the level of service provided to patients. Based on this study, one can note
that simulation allows for evaluating scenarios and projecting changes that
will impact the behavior of a certain system with no physical changes, thus
preventing the lack of scientific basis when making management decisions and
allowing for improvements.
Keywords: Operations Research; Simulation; Software Arena;
Hospital Management.
1. INTRODUCTION
In every organization, it is fundamental to improve processes,
service levels, and operation efficiency on a continuous basis. However, it may
be difficult to do so when the manager lacks a clear and holistic view of the
organization’s operations. At this point, Operations Research (OP) may play an
important supportive role through quantitative techniques and tools for the
analysis of processes implemented by organizations, allowing for improvements
that may not be very clear at first sight.
In the last few years, there has been increased
interest in OP-based techniques, especially because of recent software
developments that help with the calculation process of many optimization
methods. This is not enough though, since one of the biggest challenges is lack
of knowledge in the field, mainly when it comes to simulation techniques. This
study shows the implementation of simulation in a public polyclinic. In
addition to showing its implementation, this study aims at showing how
significant this technique is for the analysis of scenarios and improvements in
the service process of a basic healthcare system selected due to its importance
for the society in general.
The first part of this article addresses the
theoretical grounding of Operations Research, as well as the basis for this
simulation, which was selected for this study from a number of OP techniques.
Next, there will be an approach to some of the many simulation software
applications created in the past few years, with special emphasis on Arena.
Following, there will be a brief approach to hospital management by virtue of
the methodology used, i.e., the case study of a health unit. Following the
study of the current situation of the unit’s system and the identification of
the bottleneck, i.e., the most overloaded employee (including which queue has
longer waiting times), system improvements will be proposed in order to remove
such bottleneck and balance the occupancy levels of resources available at the
Polyclinic, while a greater service level is provided to patients.
2. THEORETICAL GROUNDING
2.1.
Operations
Research
Looking for improvements in the processes of a system
is no easy task for managers. The greater the complexity of a problem, the more
responsibility the decision requires. Operations Research is an applied
quantitative science that can be useful in such a difficult task. As the name
suggests, it studies operations, providing optimization tools and techniques
based on mathematical models for a clearer and more objective view of the
scenario at matter, thus providing a more solid basis for making decisions that
require great responsibility.
This science is widely used by governments,
industries, service companies, and business companies. With systemic focus on
study issues, it attempts to find the optimal solution using a teamwork
methodology that comprises Engineering, Statistics, Mathematics, Computing, and
Economics (MARINS, 2011).
Operations Research as we know it came into being
during World War II, as a result of studies carried out by interdisciplinary
teams of scientists hired to solve military issues of strategic and tactical
nature (SILVA et al., 1998, p.11)
As mentioned before, OP uses models that enable the
analysis of decisions, meaning that a decision can be tested and evaluated
before being implemented, thus reducing the risks of possible losses because of
decisions made by the manager. It is worth to emphasize that the progress of
Operations Research in recent years is because the evolution of computers made
it possible to design software applications that significantly help with the
calculations of many techniques provided by this science due to its data
processing and storage abilities (REZENDE FILHO, 2006).
Among the many OP-based techniques, one can mention
linear programming in issues relating to production line optimization,
production scale, transportation, maximum flow, shorter path, etc. All these
problems have direct impact on calculations (maximizing or minimizing), and the
result provides managers with the optimal solution, i.e., the best distribution
of resources available based on system demands and characteristics, in order to
meet the study goal (LACHTERMARCHER, 2009).
Accordingly, this work will show exactly the
significance of implementing one of the OP-based techniques with the help of
one of the many software applications created in the last few years.
2.2.
Discrete
Events Simulation
Discrete events simulation is a technique not as
objective as linear programming. It is originated from an old theory within
Operations Research: the queueing theory.
“This theory addresses system queueing issues, which
main characteristic is the presence of ‘customers’ requesting ‘services’
somehow” (ANDRADE, 2009).
“In the context of the queueing theory, when one uses
the word ‘queue’, one is referring to the set of customers that will be waiting
for the service [...], in addition to that one who is being served”
(CAIXETA-FILHO, 2004).
Queueing theory is the study of waiting in various guises. It uses queueing models to represent the various types of queueing systems (systems that involve
queues of some kind) that arise in practice. Formulas for each model indicate
how the corresponding queueing system should perform, including the average
amount of waiting that will occur, under a variety of circumstances (HILLIER; LIEBERMAN, 2010, p. 1)
Queues can be observed in a series of situations. Not
only in people’s everyday lives, clinical offices, banks, dental offices, etc.,
but also in the industry, such as productive systems, facilities planning,
inventories, work force, etc. Discrete events simulation helps one to consider
these scenarios by showing in a clear, objective, and quantitative manner the
study system behavior, the occupancy percentage of resources used, the size of
queues, waiting times, number attended, average and maximum travel time within
the system, and many additional information that is important for managers.
Simulation is used mainly when system changes are too
expensive or difficult to be implemented. This technique allows one to test
many changes in the study system model, as well as to analyze which
combinations render the best results. Subsequently, one can observe all the
effects of these changes over the system without taking any unwanted
cost-related risks, which would mean a loss (SILVA et al., 1998).
2.3.
Simulation
Software
According to Prado (2009), the
study of system simulation has two stages. First, the analyst should build a
model, feed data into it, and collect other data that are identical to the ones
of the study system. The second stage consists of changing the model so that, based
on the results obtained, one can perform analyses that will yield
recommendations and a conclusion.
Arena is a graphic simulation and animation software
created by Rockwell and distributed by Paragon.
According
to Prado (1999), SIMAN is a simulation language, which, in 1983, gave name to
the first simulation program for personal computers (PCs). Created in 1984,
CINEMA was the first PC simulation animation program. This set was continually
improved. In 1993, the programs were combined into a single software application: Arena.
Arena is an integrated graphics
simulation environment with a number of modeling, animation, statistical
analysis, and result analysis tools (LAW; KELTON, 2000).
One of Arena’s differentials is the ability of
creating templates, i.e., a collection of modeling objects/tools, through which
users can describe the behavior of the process in analysis based on answers to
questions prepared in advance, with no programming, in a visual and interactive
manner. Through the use of templates (customization), Arena can easily
transform into a specific simulator to be used for reengineering, natural gas
transportation, manufacturing, mining purposes, etc (PARAGON, 2016).
2.4.
Hospital
Management
The healthcare area is
fundamental to society by nature. At a certain point, everyone needs medical
attention, and this service must be performed with a high level of efficiency,
unlike any other segment. The smallest mistake can mean losing a life, even if
it is not an emergency. Any delay or procedure error can worsen the patient’s
condition, subsequently impairing the unit’s service level, which is not good
for the population. The manager is responsible for managing in a way that
efficiently and effectively provides balance to the whole service process.
By virtue of their high
service level, many times, issues that are incumbent on the unit’s manager,
which issues, if not resolved, can result in low service quality, challenge the
mission of health units (DA SILVA; SILVA, 2008).
3. METHODOLOGY
A
case study approach was used with the aim of studying a Polyclinic located in a
city of the state of São Paulo, in order to show the significance of
implementing the technique in hospital management. The study was selected due
to its important role in management decisions. The first step of this process
was to collect data in a regular workday of the health unit, including
available resources (equipment and employees), service time in the different
stages of the unit’s system, patient arrival interval, and average percentage
of patients rated in four priority levels.
Modeling
came next, i.e., the data from the health unit’s system were handled with the
help of Arena in order to evaluate and identify the process with the longer
waiting time and the most overloaded/idlest employees, so that service and
operational level improvements could be proposed. It is worth emphasizing that
discrete events simulation provides a number of possibilities of changes in the
study system, which is made evident throughout the case study.
4. CASE STUDY
As previously mentioned, the object of the study was a
Polyclinic where the exponential average patient arrival interval is six
minutes. These patients have to fill out a form, which time in minutes follows
a triangular distribution (TRIA) of (2, 3, 5), meaning minimum, average, and
maximum times, respectively. The form is applied by 1 of 2 attendants of the
health unit.
Based on these forms, 20% of the patients arriving are
returning patients with no need to go through the screening process, but are still
rated as “priority”, while usually 1% is “urgent” and 9% “slightly urgent”. The
screening process is performed by a nurse in a TRIA of (5, 7, 9) minutes. Graph
1 represents the average percentage of patients rated in four different
priority levels.
Graph 1:
Priority Percentage
Source:
Prepared by the Authors (2016).
From 5% of the patients rated as “emergency”, 80% are
transferred to another hospital, due to the Polyclinic’s structure and medical
service capacity. For the medical service, the unit relies on 2 professionals
that see new patients in a TRIA of (5, 11, 13) minutes and returning patients
in another TRIA of (4, 7, 10) minutes.
Typically, 24% of the new patients require X-ray
examination, that is performed by a technician in a TRIA of (10, 15, 23)
minutes and takes 30 minutes, other 37% are submitted for laboratory
examination, performed by 2 nurses designated exclusively for this procedure,
with a TRIA of (6, 8, 13) minutes to obtain the samples, plus 240 minutes to
complete the examination.
The other 8% of the new patients are submitted for
ECG, which takes a TRIA of (30, 45, 60) minutes to be completed by another
specialized nurse. The remaining 31% of the new patients require no
examinations. From all examined patients, 53% return on the same day to see the
doctor, while the other 47% return on another date.
Most of the times, only 1% of the patients require no
drug at the unit. However, 19% require inhalation procedure, which is prepared
by a nurse from the drug room in a TRIA of (0.5, 1.5, 2.5) minutes. Next, these
patients are transferred to a room with six places for them to receive the drug
in another TRIA of (8, 10, 13) minutes.
30% of the new patients are submitted for
intramuscular drug, given by the nurse in a TRIA of (3, 3.5, 5) minutes, the
other 50% receive intravenous drug prepared by the same nurse in a TRIA of
(0.5, 1.5, 2.5) minutes, and then they are addressed to another room with ten
places, where they receive their drugs in a TRIA of (40, 70, 120) minutes.
In general, 60% of the new patients have to return.
The other 40% are discharged. 2% of the returning patients require new
examinations, 20% receive the drug, and 78% are discharged.
4.1.
System
Simulation
Based on the data collected, the health unit’s system
was simulated in Arena. After the simulation, the software generated a number
of reports, including occupancy level of resources, waiting time of each
process, size of queues, among others.
Before proposing any kind of improvement, some report
results were compared to the information provided by the Polyclinic’s
management in order to check if the simulation was actually representative of
the study system. The first report shows the number of patients seen in
average. This study resulted in 194 patients, while according to the unit’s
management, the average number is 190, i.e., a minimum difference.
Another analysis was carried out to find the most
occupied and the idlest employees. According to the management, the screening
nurse is usually the most occupied professional, while the idlest employees
were the attendants and the drug room nurse. Figure 1 shows the report of
occupancy levels of the health unit’s attendants and resources.
Figure 1:
Occupancy Level
Source:
Prepared by the Authors (2016).
One can see from Figure 1 that the employee with the
highest occupancy level is exactly the screening nurse, as pointed out by the
management, and that the idlest ones are the attendants and the nurse. It is
worth emphasizing that Arena allows no special characters, such as accents in
words.
Another important detail is the unit of value of
results in Figure 1, which are expressed as decimals. Therefore, the nurse has
an occupancy level of 85.30%, while the nurse has only 5.8% approximately; in
the case of the 2 attendants, the number 26.38% accounts for the average
occupancy between them both. Figure 2 shows the report of waiting times in the
health unit’s different processes.
Figure 2: Waiting Times in Queues
Source:
Prepared by the Authors (2016).
The values in this report are expressed as minutes.
Again, the simulation matches the information provided by the management, since
the process with the longest waiting time is the screening one, with
approximately 16 minutes in average.
4.2.
Proposed
Improvements
Based on the analysis of the results reported in
Arena, the proposed improvements aim at balancing the occupancy levels of the
screening nurse and the drug room nurse, in order to make better use of their
qualifications without overloading anyone and to test a potential change in the
number of attendants.
Subsequently, a new simulation was carried out, where
the nurse had to administer the drugs and also support the screening nurse.
There was a change in the number of attendants throughout the day. In the first
12 hours, the simulation took into account 1 attendant, while in the other 12
hours, 2 employees were taken into account. The inverse was tested as well,
i.e., 2 in the first 12 hours, and 1 in the other 12. Figure 3 shows the
occupancy levels with the proposed improvements.
Figure 3:
Occupancy Level with the Proposed Improvements
Source:
Prepared by the Authors (2016).
In both the tests, changing the number of attendants
during the day caused no change in the occupancy levels. Therefore, one can
consider the report of Figure 3 as a result of the proposal. One can see an
increase in the occupancy levels of the attendants and the drug room nurse, as
well as a significant decrease in the occupancy level of the screening nurse.
Figure 4 shows the impacts of the proposed improvement compared to waiting
times in queues at the Polyclinic.
In this report, one can see a significant decrease in
the average waiting time of patients in screening process, as well as a maximum
waiting time of approximately 15 minutes. Based on these results, the
Polyclinic is suggested to use one of the receptionists, during half of the
workday, in administrative services or another activity, in order to make
better use of their workload.
Figure 4:
Waiting Times in Queues with the Proposed Improvements
Source:
Prepared by the Authors (2016).
Based on the changes suggested, a last change is
proposed. While there is an improvement in the system performance, according to
Figure 3, the drug room nurse remains as the idlest employee, with an average
occupancy level of only 9.47%. This resource accounts for 2 employees.
Accordingly, another system simulation was carried out taking into account only
1 nurse on duty. Figure 5 shows the impact of this change on the occupancy
levels of resources.
Figure 5:
Occupancy Levels with Change in the Number of Nurses
Source:
Prepared by the Authors (2016).
One can see from this report that there is no
significant change in the occupancy levels, except for the nurse, whose
occupancy level now is 17.75%. Figure 6 shows the impact of this last change on
the waiting time of each process.
Figure 6:
Waiting Times with Change in the Number of Nurses
Source:
Prepared by the Authors (2016).
What is interesting from this last report is that
there is a greater decrease in the waiting time of each process of the health
unit with only 1 nurse less, showing that it is important to carry out several
tests in the same study model, until the best distribution of the resources
available is found.
As mentioned in the methodology, discrete events
simulation is a technique that lets one consider a number of scenarios where
there might be a greater or lesser performance of the resources available in
the study system. In this research, the proposal only comprises the screening
process, reception, and nurse. However, other scenarios could be explored, such
as increasing or decreasing the number of doctors and technicians on duty.
This change was implemented in the simulation of this
case study, but the proposed improvement was not addressed due to the high
level of idleness of the health unit’s resources and longer waiting times. When
the number of doctors was changed from 2 to 3, this resulted in zero queues,
but there was 50% less occupancy of each available resource, except for the ECG
nurse, with 52.31%, and the doctors, with average occupancy of 59.21%. When
changed downwards, considering 1 doctor, this doctor would have occupancy of
99.69% based on a new simulation through the software.
Since the Polyclinic has only 1 X-ray technician, the
test consisted of changing it to 2. Again, results were not found to be
satisfactory. The technician had 70.21% of idleness, the doctor’s occupancy
level went up to 90.95%, and there was an average waiting time of approximately
27 minutes. This would cause a new bottleneck in the system, so the change was
not included in the proposed improvements.
These last data confirm that the possibilities are
many. Accordingly, the aim of the proposed improvements was only to remove the
bottleneck and adjust the Polyclinic’s system in order to provide a certain
balance in the occupancy levels of resources, while increasing the service
level provided because of shorter waiting times throughout the health unit’s
service system.
5. FINAL CONSIDERATIONS
The objective of the study was to introduce the
discrete events simulation technique as a hospital management tool, objective
which was met based on the study of the real case of a health unit. The data
were collected in the field.
Another important finding was the need to carry out
several tests and simulations based on different characteristics, in order to
consider a scenario of a general greater system performance with a greater
service level at the same time. These results show how the technique is
implemented. This allowed for an analysis of the system behavior through
changes in simulations and avoided empiricism and, subsequently, potential
losses or unnecessary works in the reallocation of resources.
This research increases our knowledge of the study of
operations with the discrete events simulation technique, which, as mentioned
in the section “Theoretical Grounding”, is one of many OR-based techniques.
Finally, many other techniques can be tested in real cases, allowing one to
consider and evidence the significance of these tools in the study of
operations in order to avoid changes in operations without any scientific
basis.
REFERENCES
ANDRADE, E. L. (2009) Introdução à Pesquisa Operacional - Métodos
e Modelos para Análise de Decisões. 4ª
Ed. Rio de Janeiro: LTC - Livros
Técnicos e Científicos.
CAIXETA-FILHO, J. V. (2004) Pesquisa Operacional: Técnicas de
Otimização Aplicadas a Sistemas Agroindustriais. 2ª Ed. São Paulo: Atlas.
HILLIER, F. S.; LIEBERMAN,
G. J. (2010) Introdução à
pesquisa operacional. McGraw Hill.
LACHTERMARCHER, G. (2009) Pesquisa Operacional na tomada de decisões. 4ª
ed. Pearson Prentice.
LAW,
A. M.; KELTON, W. D. (2000) Simulation
modeling and analysis. New York: McGraw- Hill.
MARINS, F. A. S. (2011) Introdução à Pesquisa Operacional. São Paulo: Cultura Acadêmica: Universidade Estadual Paulista.
PARAGON. [S.l.]:
[c2016?]. Available at: <http://www.paragon.com.br/arena-academico-student/
>. Accessed on
26 Mar. 2016.
PRADO, D. (1999) Usando o Arena em Simulação.
Belo Horizonte:
Editora de Desenvolvimento Gerencial.
REZENDE FILHO, M. (2006) Programação Linear - Otimização de Recursos em Apoio à Tomada de
Decisão. Rio de Janeiro: Fábrica do livro, v.1.
SILVA, E. M. et al. (1998) Pesquisa Operacional na tomada de decisões. 3ª
ed. Atlas S.A..
SILVA, J. A.; SILVA, M. V.
(2008) O administrador hospitalar nas organizações públicas de saúde frente às
novas concepções de administração. Saber
Científico, v. 1, n. 2, p. 323-341.