Simulation model to assist the planning and management of Health Services

Silvia Brand Silva^a, Heitor Mansur Caulliraux^a, Thais Spiegel^a

Abstract

The real challenge for managers has been satisfying the following dilemma: provide health care with a quality minimally satisfactory, provide access to this system with the least possible restriction and with an affordable cost. In this context, the aim of this paper is to propose a discrete event simulation model that represents a subsystem of a hospital and demonstrate how it can be used to reconfigure the portfolio of services offered by the Medical Unit. Aiming thereby facilitate user access, organize flows and improve quality which translates into a reduction in the size of the queues and decrease the waiting time for care.

Keywords: Simulation, Health Management, Health Services.

1. INTRODUCTION

According to Escrivão Junior (2012), Brazil is among the countries with the highest level of inequality, both in the health status of the population and access to health care of good quality. The difficulty to access has emerged as a critical factor even in the private sector. Patients are often subjected to long waiting times to perform an examination or consultation.

For Oliveira et al. (2011), the key point of this question is the balancing and the balance between increasing demand and an inadequate supply of hospital services. In this way, the capacity management plays a strategic role to monitor the allocation of resources, scarce and costly, in an organization.

However, the study's ability of service expressed particular characteristics, because its production has considerable differences related to the production of products. Unlike goods, services are:

• Intangibles - can not be touched or stored. A wasted time of a medical resource, for example, can not be recovered in the future;

• Co-produced - require the simultaneous presence of the client/user (patient) and the processing resources (doctors, nurses);

• Heterogeneous - two services are never exactly alike;

• Difficult to measure - quality, for example, is an indicator that depends on the customer/user perception. Two patients can have fairly different perceptions to services done in a similar way;

Faced with these difficulties, it grows the need for tools that enable the preliminary analysis and quantification of the impact of possible changes. To Gonçalves (2004), the discrete event simulation has great potential to assist managers in strategic and operational planning. According to Harrell et al. (2012), the power of simulation lies in the fact that it provides a method of analysis that is able to accurately predict the performance of even the most complex systems.

The study object of this research is a medical unit located in the western area of the Municipality of Rio de Janeiro. The hospital provides emergency care and outpatient consultations in 15 specialties. From the data provided by management, there were long queues for appointments while high levels of absenteeism. In addition, over 80% of patients who sought emergency were characterized as "not urgent" or "non-urgent". The analysis of these facts indicates a need for reconfiguration of the unit service portfolio.

In this context, this research aims to bring together elements that contribute to spread the use of simulation as an instrument to improve the analysis of processes and decision-making in health services. This study lies in presenting the construction experience of a simulation model to evaluate alternative scenarios and scale the resources needed in reshaping the portfolio of services from a health provider.

To achieve the proposed objectives, this document is structured as follows explained. Section 2 presents the concepts and definitions about the simulation, the formulation of the problem, the method of construction of the model and characterize the software used. Section 3 describes in detail the model built for adult emergency introducing the adequacy of simulation technique to this case. Section 4 is an analysis of the current situation and from there scenarios are constructed. And, finally, Section 5 presents the conclusions of the study.

2. SIMULATION E MODELING

The simulation is the reproduction of the conditions of a situation through a computational model in order to evaluate and improve system performance Harrell et al. (2012). For the operation of the simulation, was used the construction of logical-mathematical models that represent the dynamics of the system under study.

White et al. (1999) features a model as an explicit and external representation of part of the reality seen by people who want to understand it, change it, manage it and control it in some way. For the modeled reality has validity, its construction should be carefully and with sufficient detail so that their outputs are not different than actually occurs.

Observed the care in building the model, the simulation is a valuable technique. It allows the decision maker to evaluate alternative scenarios without the need to implement and quickly at low cost. This characteristic is especially relevant in healthcare where the experimentation could incur adverse effects on patients, if changes to the system harmed it’s performance (Pessôa et al., 2009).

Using this beacon, a model that represents the subsystem adult emergency was built. We conducted a comprehensive study of diagnosis in the medical unit in question. Therefore, it was divided into subsystems that could be analyzed separately for better understanding of the problems considering its connections. They are: Adult Emergency, Pediatric Emergency, Ambulatory, Pediatrics, Laboratory, Image (radiology), Examinations (cardiology) and Pharmacy / Sterilization.

The model building method follows the following steps: field visit to understanding the functioning of the clinic, the patient flow construction, the flow validation, patient flow transformation into a logical model, model programming in computer language and analysis of output data. Developed the ProModel software in package simulation designed specifically to model health systems called MedModel. The package has a library of characteristic objects of the hospital environment, as patients, beds, medical equipment, ambulances and others.

The structure of the model began with the layout design, made respecting the constructed area in clinic: 1130 m² per floor. All rooms have been represented in scale. Adopted was a value of 38 m/min for the transit speed which is the speed at which resources and entities move within the model. The simulation period, ie time in which the model performs processes, was five weeks. The defaults followed the reference MedModel User Guide. The following section explains in detail the model built.

3. THE ADULT EMERGENCY SUBSYSTEM

Emergency services are offered in the specialty Orthopedics and Internal Medicine. Are available in five offices, four ordinary and one specially dedicated to orthopedics. The Adult Emergency also has a plaster room and two risk stratification and a lounge. There are present a medication room, an inhalation, a nursing station, eight beds (one isolated), a trauma room and a dressing room.

The record of the arrival of the emergency patients to the unit is done through the totems located at the reception. By registering, the patient specifies whether looking for medical or orthopedic care. In the second case, already goes to the waiting room consultations, which will wait for the service. In the case of Internal Medicine, the user is directed to the waiting room for layering. The medical unit chose to standardize care by importing and implementation of screening protocol of Manchester, England.

According Escrivão Junior (2012), the Manchester Triage System (MTS) establishes a risk rating in five categories. After identifying the main complaint of the patient by the nurse, a specific flowchart, guided by discriminating and presented in the form of questions, it is selected. On the clinical history and presented signs and symptoms, the patient is classified into one of five categories: Emerging (red) very urgent (orange), urgent (yellow), low urgent (green) and not urgent (blue). The purpose of the adoption of the protocol is to organize the service. Thus, there is the improvement of user access, changing the traditional way of entry by rows in order of arrival. Patients with higher severity signs have priority over the other.

After risk stratification, patients await the consultation in the waiting room. On that first visit, the Internal Medicine patient could be referred to the Procedure Room (medication room or boxes of rest), to do some examination (laboratory, radiological and cardiological) or be released. The same patient can be referred to more than one type of exam and also for medication room. Followed the proper doctor's recommendations, the patient returns to the clinic for the second service and is then released.

For patients from orthopedic consultations, which have not been released are mostly referred for radiological examinations. Then, the patient goes to the plaster room (Figure 1). The following explains the construction of the model and then a description of the processes for each local entity is given. According Pessôa et al. (2009) this description enables the conversion to the model and details its operation.

Entities: the patient.

Resources: the totem, the nurse responsible for risk stratification, doctors of internal medicine, nurses responsible for collecting and medication, x-ray technician, the echocardiography technician, the orthopedist and the plaster room technician. The types and amounts of resources represented in the model are shown in Table 1.

Table 1. Human resources and equipments

Source: Own elaboration

Figure 1. Patient flow in adult emergency

Source: Own elaboration

The local for emergency structure representation are also specified. It includes admission of patients, the registration on the totem pole, the offices, the treatment rooms and the exit. The following table sets out the local, unity, capacity and admission rule for the entities. Capacity is the number of entities that the local involves a time, since the unit is defined as the amount of independent operation station. Admission rule defines the priority in which the entities are met.

Table 2. Local characteristics represented in the model

Source: Own elaboration

A variable called "adult_color" was created. Every time the entity passes through the patient risk stratification, this variable is assigned the value of a random number of 0 to 1. Through this random number is identified the color classification of the patient in accordance with the percentages shown in Table 3. The "adult_color " variable specifies the attribute "Patient_Type ". For the red color attribute "Patient_Type" takes the value 1, to orange value of 2, and so on. With this, the attribute values gets smaller the more severe is the disease, so the admission rule is the minimum value of the attribute "Patient_Type".

The classified patient percentage in each color were calculated from sample given by the unit management and are explained in table 3.

Table 3. Percentage of patients classified into certain color in risk stratification

Source: Own elaboration

Finished the first consultation, patients can be referred to one of the rooms previously mentioned. The forwarding rates were also calculated from the sample provided by management and are shown in Table 4. Note that the same patient may be referred to more than one type of procedure, which explains why the percentage sum of 113.05%.

Table 4. Percentage of patients seen in the emergency are directed to the treatment rooms.

Source: Own elaboration

The model built in ProModel software can be seen in Figure 2.

Figure 2. Simulation model of subsystem adult emergency

Source: Own elaboration

Because of the difficulty in obtaining formal and reliable data to power model, a team remained in place clocking times. The collected data were entered in the Stat Fit ProModel tool. The Kolmogorov-Smirnov test of Anderson Darling examined the adherence to probability distributions through d. The parameters used are shown in table 5.

Tabela 5. Model Parameters

Source: Own elaboration

There are three broad sets of indicators that are relevant for planning the unit and can be viewed on simulation results: about patients, about resources and about locations. In the case of the first indicator during the simulation period for measuring the user lead-time in the system, 8.123 patients were "processed" within the unit, with an average time of crossing of 1h7m (67.03 minutes). Of this total time on average, patients were 29 minutes "in operation", ie performing some activity that adds value to it.

The second and third sets of indicators relate to the use of various resources and places accessible to the adult emergency patients. The graphics generated by the tool allow us to understand the percentage of time that such resources and places were in operation and percentage of time that were idle. Indicators of resources and places allow you to map the bottlenecks and idle points system.

4. EXPERIMENTS

In the clinic, in addition to regular consultations, there is System Emergency Care (SPA). The SPA serves patients in order of arrival through passwords. There are also some places for patients who are redirected from emergency room. These are patients classified as blue in the emergency and because there is no urgency are referred to the clinic. Passwords are distributed estimating a duration of 15 minutes for consultations.

In short, the Internal Medicine has three gateways to the patient: the elective enviroment (outpatient), the SPA and the emergency. In the current situation, patients classified as "green" and "blue" are the 87, 84% of total emergency. They are treated at emergency rooms, despite not being.

The patient finds it very difficult to be able to schedule an appointment at the Internal Medicine. From data supplied by the unit management, it has been estimated that the size of the queue is of 75 days for an appointment. On the other hand, the long wait of more than two months translates into high frequency of absences of scheduled patients. The calculated index of absenteeism is 50%.

What is suggested in this paper is that patients seeking emergency and are classified as "blue" or "green" in the risk stratification to be forwarded to the SPA. The "green" patients are classified as "low urgent", but often require medication or service in the unit and therefore have priority in the SPA. As stated earlier, the direction for the SPA has been going on for patients classified in the "blue" color. However, they do not gain priority over SPA calls.

This proposal was based on the Ministry of Health document, which proposes the construction of flows separated by degree of risk in order to promote the organization of space and clarity and control over the process. The design proposal builds at least two axes: the critically ill patients, called the red axis, and the patient not serious but needs or seeks emergency care, called blue axis. The new patient flow is represented in Figure 3.

The simulated scenarios, so check the interface of emergency adult subsystem with adult outpatient clinic subsystem. The experiments considered the medical resource from the SPA and the demand that already existed for this service. SPA care will now be called the blue axis and the emergency service red axis.

Looking for the ideal configuration for blue and red axes in order to balance the demand and supply of services, providing a more efficient service to all patients. There is an expectation – shared by clinic management – that the increase in the supply of SPA trigger a positive impact on the emergency and elective environment by reducing absenteeism. This expectation is a consequence of the fact that demand in the emergency by low urgent patients is due to the difficulty in booking appointments for the short term.

Figure 3. New patient flow

Source: Own elaboration

Scenario 1 is to separate the care of emergency patients classified as "red", "orange" and "yellow" from the rest. An emergency doctor performs the care of these patients and the other four (three doctors in the emergency plus one doctor who was already allocated in the SPA) doctors perform the service of those classified as "green", "blue" and SPA calls. The objective of this scenario is to see if really only one doctor in the red axis could meet demand.

As suggested by Oliveira et al. (2011), established the maximum rate of 80% usage in the analysis of the results of the resources provided by the simulation. This maximum value was established because was considered that the professional does not work overly. As a utilization rate was observed for the emergency physician below this value, it appears that only one doctor can meet the demand of the red axis.

The average patient time in the system increased from 67 minutes to 117 minutes. However, the operation time remains practically the same, about 29 minutes. This means that the time does not add value, ie, the queue time increased to 88 minutes.

Scenario 2 tests the one-unit increase in the number of doctors in blue axis. It aims to verify that the patient's time in the system decreases significantly with the introduction of another doctor in blue axis. In it, the average patient's time in the system became 63.54 minutes. This time has been halved compared to scenario 1.

Scenario 3 increases by another unit the number of doctors in the blue axis, totaling 5 doctors, in it, the average time of the pacient in the system became 57.96, and the average of operation to 26.20.

It was found that the average time of the patient in the system decreases little in the scenario that includes five doctors of SPA. This means that the addition of one doctor does little to reduce the patient's time in the system. Because of this, the scenario 5 is tested with five doctors to SPA. The risk stratification nurse presents a percentage of utilization of 77%, very close to the limit. Scenario 4 tests the impact of the increase of a risk stratification nurse in the average patient's time in the system, and the results point to an average time in the system and operation of 58.12 and 28.25, respectively.

It was found that the average time of the patient in the system decrease a little in the scenario that includes two risck stratification nurses. So in the scenario 5 was considered two nurses.

The results provided by the simulation showed that collection nurses presents 78% as a percentage of use. Thus, the scenario 5 tests the increase by another collection nurse in the average queue time. In this scenario, the time in the system was 59.63; the operation was 28.40.

In scenario 1 the number of doctors in blue axis is configured as a restriction on the system because the introduction of another doctor halves the patient average time. However, a comparison of scenario 2 to 3, 4 and 5 it can be seen that there are no significant changes in queue time. Table 9 shows these scenarios reached a maximum reduction of 5 minutes in the queue time. The hiring of other professional could be quite costly to the system without a significant change in queue times. Thus, it is considered positive performance presented in scenario 2 especially for a short-term horizon.

Tabela 6. Parameter summary of each scenario and the patient's time in the system

Source: Own elaboration

5. FINAL CONSIDERATIONS

During this work, a simulation model was presented to discrete events able to describe about the operation of an important subsystem of a hospital: the adult emergency. From the analysis of the classification percentage of risk stratification and absenteeism, it was found that most users of the health system sought Medical Unit because wanted quick responsiveness. However, they did not require emergency treatment.

It is suggested the separation of the flow of patients in the blue axis, which receive non-urgent and urgent calls of low and medium complexity, and the red axis directed to urgent patients of high complexity and emergencies. The division of flows and the redesign of processes is fundamental in reshaping the services offered by the unit and the pursuit of efficiency. According to the Ministry of Health, the realization of the risk classification alone does not guarantee an improvement in the quality of care. It is necessary to build internal pacts for the viability of the process, with the construction of light flows by degree of risk.

In addition to the division of the flow, the simulation model was used in the search for a better design and configuration of human resources, improving service capacity. As a result, access to health care was facilitated by reducing the waiting time for consultation at the Internal Medicine specialty.

Ferraz (2012) recommends that there is a great common challenge to many and varied world’s health systems. He sets up in how to reconcile expectations and needs of users of this system to the available resources in an extremely complex environment and constantly changing. In other words, how to define and reconcile health care, according to three variables: quality of health care, access to health care and cost of care or attention. Thus, this study sought to demonstrate how managers can use the simulation tool to assist decision-making on these three pillars.

REFERENCES

Brasil, Ministério da Saúde. (2013), Normas para projetos físicos de estabelecimentos assistenciais de saúde, Ministério da Saúde, Brasília, DF, disponível em: http://bvsms.saude.gov.br/bvs/publicacoes/normas_montar_centro_.pdf (Acesso em 28 de Fevereiro de 2013).

Brasil, Ministério da Saúde, Secretaria de Atenção à Saúde (2009), Política Nacional de Humanização da Atenção e Gestão do SUS, Acolhimento e classificação de risco nos serviços de urgência. Ministério da Saúde, Brasília, DF, disponível em http://bvsms.saude.gov.br/bvs/publicacoes/acolhimento_classificaao_risco_servico_urgencia.pdf (Acesso em: 28 de Fevereiro de 2013).

Escrivão Junior, A. (2012), “A epidemiologia e o processo de assistência à saúde”, em Vencina Neto, G et Malik, A. M., Gestão em Saúde, Guanabara Koogan, Rio de Janeiro, RJ, p. 15-31.

Ferraz, M. B. (2012), “Avaliação econômica em saúde”, em Vencina Neto, G et Malik, A. M., Gestão em Saúde, Guanabara Koogan, Rio de Janeiro, RJ, pp. 184-190.

Ferreira Filho, V. J. M. (2005), Processos estocásticos e teoria de filas, Programa de Engenharia de Produção, Universidade Federal do Rio de Janeiro, Rio de Janeiro, RJ.

Gonçalves, A. A. (2004), Gestão da capacidade de atendimento em hospitais de câncer, Tese de Doutorado em Ciências, Universidade Federal do Rio de Janeiro, Rio de Janeiro, RJ.

Harrell, C., Ghosh, B. K., Bowden, R. O. (2012), Simulation using promodel, McGraw-Hill, New York, NY.

Oliveira, M. J. F., Oliveira, D. G., Oliveira, F. B., Chaves, W. B. (2011), “Modelo integrado de simulação para avaliar o desempenho de sistema de admissão de pacientes na emergência hospitalar”, apresentado no XLIII Simpósio Brasileiro de Pesquisa Operacional, Ubatuba, SP, 15-18 de agosto de 2011.

Paes, L. (2011), Gestão de operações em saúde para hospitais, clínicas, consultórios e serviços de diagnóstico, Atheneu, São Paulo, SP.

Paiva, L. F. (2012), Instituto Nacional do Câncer – Estudo de Simulação da Radioterapia – HCI, Trabalho de conclusão de curso de Graduação em Engenharia de Produção, Universidade Federal do Rio de Janeiro, Rio de Janeiro, RJ.

Pessôa, L. A. M., Lins, M. P. E., Da Silva, A.C.M., Fizsman, R. (2010), “Apoio à gestão de centro cirúrgico por meio de mapas”, apresentado no XLII Simpósio Brasileiro de Pesquisa Operacional, Bento Gonçalves, RS, 30 de agosto-03 de setembro de 2010.

Pinto Junior, D., Salgado, P. O., Chianca, T. C. M. (2012), “Validade preditiva do protocolo de classificação de Risco de Manchester: avaliação da evolução dos pacientes admitidos em um pronto atendimento”, Revista Latino-Americano de Enfermagem, Vol.20, No.6.

White, R. E., Pearson, J. N., Wilson, J. R., Pidd, M. (1999), “Systems just modeling through: a rough guide to modeling”, Interfaces Archive, Vol.29, No.2.