Conference PaperPDF Available

Pre-hospital simulation model for medical disaster management

Authors:

Abstract and Figures

Medical disaster management research aims at identifying methodologies and rules of best practice and evaluates performance and outcome indicators for medical disaster management. However, the conduct of experimental studies is either impossible or ethically inappropriate. We generate realistic victim profiles for medical disaster simulations based on medical expertise. These profiles are used in a medical disaster model where victim entities evolve in parallel through a medical response model and a victim pathway model. The medical response model focuses on the pre-hospital phase which includes triage procedures, evacuation processes and medical processes. Medical decisions such as whether to evacuate or to treat the current victim are based on the RPM (respiratory rate, pulse rate, motor response) parameters of the victim. We present results for a simulated major road accident and show how the level of resource can influence outcome indicators.
Content may be subject to copyright.
Proceedings of the 2013 Winter Simulation Conference
R. Pasupathy, S.-H. Kim, A. Tolk, R. Hill, and M. E. Kuhl, eds.
PRE-HOSPITAL SIMULATION MODEL FOR MEDICAL DISASTER MANAGEMENT
Christophe Ullrich
Filip Van Utterbeeck
Emilie Dejardin
Department of Mathematics
Royal Military Academy
Brussels, BELGIUM
Michel Debacker
Research Group on Emergency
and Disaster Medicine
Vrije Universiteit Brussel
Brussels, BELGIUM
Erwin Dhondt
Queen Astrid Military Hospital
Neder-Over-Heembeek, BELGIUM
ABSTRACT
Medical disaster management research aims at identifying methodologies and rules of best practice and
evaluates performance and outcome indicators for medical disaster management. However, the conduct of
experimental studies is either impossible or ethically inappropriate.
We generate realistic victim profiles for medical disaster simulations based on medical expertise. These
profiles are used in a medical disaster model where victim entities evolve in parallel through a medical
response model and a victim pathway model. The medical response model focuses on the pre-hospital
phase which includes triage procedures, evacuation processes and medical processes. Medical decisions
such as whether to evacuate or to treat the current victim are based on the RPM (respiratory rate, pulse
rate, motor response) parameters of the victim. We present results for a simulated major road accident and
show how the level of resource can influence outcome indicators.
1 INTRODUCTION
Until recently, reports of disaster responses primarily have been anecdotal and descriptive. There are
currently no defined (validated) performance outcome measures as to what constitutes a “good” disaster
response or not. Operational research in medical disaster management is limited by the fact that the
conduct of prospective and randomized controlled studies under real world conditions is impossible or
ethically inappropriate. Computer simulation has been used to overcome these methodological problems.
For example, Escudero-Marin and Pidd (2011) use an agent based modeling system to simulate a hospital
emergency department. Duguay and Chetouane (2007) describe a discrete event simulation study of another
emergency department. G¨
unal and Pidd (2010) present a literature review about discrete event simulation for
performance modeling in health care. Su (2003) uses an object oriented simulation software to improve the
emergency medical service of Taiwan. Brailsford and Hilton (2000) compare system dynamics and discrete
event simulation to see which method should be applied in specific circumstances. Mes and Bruens (2012)
develop a discrete-event simulation model for an integrated emergency post. They presented a generalized
and flexible simulation model, which can be adapted to several emergency departments. McGuire (1998)
dedicates a chapter to the application of simulation tools in health care. Tomasini and Van Wassenhove
2432978-1-4799-2076-1/13/$31.00 ©2013 IEEE
Ullrich, Van Utterbeeck, Dejardin, Dhondt, and Debacker
(2009) discuss the evolution of supply chain management in disaster relief and the role of new players
like the private sector. Van Wassenhove and Pedraza Martinez (2012) adapt supply chain best practices to
humanitarian logistics.
In contrast to real world disaster exercises, computer simulations of medical disaster response allow
the consecutive execution of a particular scenario with changes to the occurrences and timing of particular
medical interventions or modifications to the utilization of human and material resources. This enables the
evaluation of (medical) operational interventions in multiple plausible disaster situations and the development
of a resource-efficient medical response without the costs and time constraints associated with full scale
exercises.
The research presented in this paper is part of the SIMEDIS (Simulation for the assessment and
optimization of medical disaster management in disaster scenarios for the Queen Astrid Military Hospital)
project. The objective of SIMEDIS is the development of a stochastic discrete event simulation model
which will be used to evaluate applicable methodologies and identify rules of best practice for medical
disaster and military battlefield management in different large-scale event scenarios for the military hospital
of the Belgian Defense. The four initial scenarios under investigation are an airplane crash and airport
disaster, a CBRNE (Chemical, Biological, Radiological, Nuclear and Explosives) incident, mass gatherings
and a hospital disaster.
A stochastic discrete event simulation model was constructed using Arena (a commercially available,
SIMAN programming language-based simulation software). References for Arena include Altiok and
Melamed (2007), Kelton et al. (2010) and Rossetti (2010). This simulation model is shown in figure 1
and consists of 3 interacting components: the medical response model (where the victim interacts with
the environment and with the resources at the disposal of the disaster manager), the victim creation model
and the victim pathway model (where the current clinical condition of every victim is monitored). The
specificity of our simulation model is the fact that the victim entities will evolve through both the medical
response model and the victim pathway model in parallel, while the interaction between both models is
ensured through triggers. This paper focuses on the improvement of the medical response model presented
in Van Utterbeeck et al. (2011). A pilot scenario depicting a major road traffic accident has been studied
and has allowed the validation and verification of the simulation model and its outcome indicators (victim
flow, morbidity, mortality).
Figure 1: top-level view of the SIMEDIS simulation model
In this paper we propose an improved medical disaster model. In Van Utterbeeck et al. (2011), the
medical response was limited to a single medical process on the disaster scene. We extended the model to
a pre-hospital simulation model with 3 areas of interest. Concerning the victim pathway the limitation was
that we could have only one medical intervention trigger per clinical condition and the implementation of
the victim pathway was specific to the victim profile. We remove this limitation on the pathway definition
and propose a logic to handle all the different pathways. We focus on improving the implementation of the
two models and their interaction, and we show an implementation using the Arena software. In section 2
we present a way to manage multiple intervention triggers, a new implementation of the victim pathway
2433
Ullrich, Van Utterbeeck, Dejardin, Dhondt, and Debacker
model and new time factors for medical intervention. In section 3 we propose a more realistic medical
response model. In Section 4 we describe the pilot model and discuss the results. Finally, in section 5 we
conclude and discuss future work.
2 VICTIM PATHWAY
The victim creation model generates all the disaster victims needed in the simulation and maps these victims
to victim profiles corresponding to the scenario. Each victim profile consists of general victim data, a set of
possible clinical conditions (specifying primary survey, triage and diagnostic test data and injury severity
scores) and a set of potential transitions in between.
The victim pathway model represents the clinical evolution of each victim in the disaster scenario and
manages the transition of one clinical condition to another. These are triggered either by elapsed time or
by medical treatment interventions (according to procedures, available equipment and supplies as well as
skill levels of the on-site medical care providers).
2.1 Victim Profile
Victim profiles are created according to the hazard type and injury mechanism and their severity. The
transition of one clinical condition to another clinical condition depends on time intervals (time interval of
clinical deterioration if no treatment is provided, time interval to deliver the treatment and time interval of
treatment procedures to be effective), treatment procedures and resources including the health care providers
with their respective skill levels, medical equipment and supplies. The “no treatment” time interval and
effect time interval of treatment are determined by medical experts. The treatment delivery time interval
is based on experimental studies (see for examples: Cwinn et al. (1987)).
Figure 2 shows an example of a victim pathway with 24 clinical conditions (CC) and 41 transitions.
CC0 is the clinical condition immediately after the impact of the disaster on the victim’s health.
Figure 2: The different pathways of the clinical condition of a victim.
2.2 Implementation
In Van Utterbeeck et al. (2011), three possible transitions from one clinical condition to another were
considered, within the victim pathway:
only time trigger (TT);
only one medical intervention trigger (MIT);
one time trigger and one medical intervention trigger.
2434
Ullrich, Van Utterbeeck, Dejardin, Dhondt, and Debacker
The first extension of the pathway model is the possibility to have several medical interventions per
clinical condition. This modification is used to represent the incremental nature of treatment capabilities
(i.e. partial treatment) and to avoid the situation where a medical intervention trigger is not performed
because of the unavailability of one or more assets required for the intervention. Pathways with multiple
medical intervention triggers allows medical experts to define several possible treatments in function of
the available assets (with better treatment generally requiring more assets).
Each clinical condition is represented by a VBA class module. The transitions between the current
clinical condition and the next one in the pathway are represented by a linked ordered list. The criterion for
ordering the list is the number of medical interventions. The first item of the list is the medical intervention
trigger with the highest number of interventions. A specific pathway is defined by all the clinical conditions
and transitions and is implemented as a collection of instances of the class Clinical Condition.
The second extension is this “Effect Time”: for a medical intervention trigger we now define two time
delays. The first one is for the delivery time of the intervention, during which medical assets are assigned
to the victim. The second one is the effect time which is the time required for the intervention to have its
effect on the clinical parameters of the victim. Medical assets are not seized by the victim during this time.
It is however possible that a medical intervention is initiated in the medical response model before
the time trigger delay elapses or while the entity is being held until the next intervention. The entity is
then sent to the “Being Treated” logic branch. The victim in question will be held there until the Medical
Process logic sends the victim entity onward to the “Effect Time” logic. The victim is held during the
effect time of the medical intervention and is sent to the entry station corresponding to the type of the new
CC.
The third extension concerns the implementation of the victim pathway model. We implemented a new
version in such way that it can handle all possible kinds of pathway with the same logic module. Each
clinical condition can evolve by transitions, which are : time trigger and medical intervention triggers, only
TT, only MITs and end CC. the duplicated victim enters the victim pathway model through a station (Victim
Pathway IN), passes the clinical condition test. The time trigger module consists of a delay module where
it will be held until the time trigger delay for this CC elapses. The value of the time trigger delay is stored
in an attribute of the victim entity. The medical intervention trigger module consist of the “Being Treated”
logic branch and “Effect Time” logic. When the delay elapses (time trigger or delivery and effect time), the
victim entity enters the clinical adaptation module which updates the parameters to those corresponding
to the next CC. The entity enters a second check module which evaluates if the new clinical condition is
an end clinical condition or not. If it is not an end clinical condition, the victim entity returns to the first
check else the victim enters the end clinical condition module. This logic is represented in Figure 3 below.
Figure 3: Clinical condition logic
3 MEDICAL RESPONSE MODEL
The medical response model represents the environment (areas of interest, time), the available human and
material resources, a rule-set of medical/operational decisions and the localization of the victims as they are
evacuated from one area to another. Typically, the three areas of interest are the disaster site, the forward
2435
Ullrich, Van Utterbeeck, Dejardin, Dhondt, and Debacker
medical post and the health care facilities of destination, these areas are depicted in the Figure 4. The
non-urgent care area represents the family doctors.
Figure 4: Medical response model
For each area of interest we can define three main processes, these processes and their interactions are
illustrated in Figure 5. These processes, interactions and relations between areas of interest are function
of the rescue policy. Two rescue policies have been implemented in our pre-hospital simulation model. In
the “scoop & run” rescue policy, victims are directly evacuated to the health care facilities. In the “stay &
play” victims are transferred to the forward medical post, where they are treated, stabilized or evacuated
to health care facilities according to decision rules and availability of resources.
Triage and
Decision
process
Medical
process
Evacuation
process
zone entry
Figure 5: Area logic
In the following subsection we present the three areas of interest and detail which processes are involved
as a function of the rescue policy chosen and the relation between the areas of interest.
3.1 Disaster Site
The on-site sub-model represents the disaster site, a part of the simulation model in which all the victims
involved in the scenario are created in. The two rescue policies share a common initial process which
consists of the primary triage. This process allows the separation of the urgent victims from the non urgent
victims, the same rules are applied for both rescue policies. Victims are separated in function of their
NATO triage category (defined in their profiles, NATO STANAG (1998)), T1, T4 and T2 victims are urgent
and T3 victims are non urgent. The remaining processes and relation between the areas are specific to the
rescue policy.
In the “scoop & run” policy, victim entities enter a second decision process called “disposition” after
the primary triage. The process helps to decide whether to evacuate the victim to health care facilities or
treat and/or stabilize the victim. The victims enter either the medical process or the evacuation process in
function of the decision process. The evacuation to health care facilities depends on:
2436
Ullrich, Van Utterbeeck, Dejardin, Dhondt, and Debacker
the severity of the injuries, the associated survival probability and the deterioration rate or change
of survival probability over time;
the availability of ambulances or other transport vehicles;
the transport time from the scene to health care facilities;
the treatment capacity of the health care facilities.
In the “stay & play” policy, victims enter an evacuation process where all the urgent victims are
evacuated to the forward medical post in function of their priority and the non-urgent victims are evacuated
to the non-urgent care area. No medical process is modeled at the on-site sub-model as the victims are
treated and/or stabilized at the forward medical post.
The aim of assigning priorities to victims when transferring them from the disaster scene or advanced
medical post to health care facilities is to maximize the expected number of survivors. Priority is evaluated
with the triage category and the RPM severity score. The injury severity is defined by the so-called RPM
score which consist of respiratory rate, pulse rate and best motor response. The RPM score is the sum of
coded values for respiratory rate, pulse rate and best motor response and takes integer values from 0 to 12
(Table1).
Table 1: RPM Score
Coded value Respiratory rate Pulse rate Best motor response
(per minute) (per minute)
0 0 0 None
1 1-9 1-40 Extends/flexes from pain
2 36+ 41 - 60 Withdraws from pain
3 25 - 35 121+ Localizes pain
4 10-24 61 - 120 Obeys commands
Survival probability estimates were determined for each RPM value using a logistic regression on data
obtained from a retrospective analysis of data from a trauma registry (Sacco et al. (2005)).
3.2 Forward Medical Post
Victims arrive at the Forward Medical Post (FMP) only if the “stay & play” policy of rescue is used as
for the “scoop & run” the victims are directly evacuated to the health care facilities. The three processes
and their interaction shown in the Figure 5 are used to model the FMP. First the victims enter the decision
process, the rules used by this process are the same as the disposition described in the on-site area for
the “scoop & run” policy. If the disposition process decides to evacuate the victim and this victim enters
the evacuation process. If no means of transport are available the decision process checks the possibility
of treating and/or stabilizing the victim, this victim enters the medical process. At the FMP the medical
process is also composed of palliative care (if the severity of injury is too great to be treated or stabilized)
and temporary mortuary. If no treatment and/or evacuation is possible the victim waits until resources are
available.
3.3 Health Care Facilities
Actually the health care facilities model is limited to the arrival of the victims at the correct hospital,
writing the log file, and updating the performance indicators (performance indicators are described in the
subsection 4.2). A simple model of the emergency department will be implemented in the future using
only the three processes presented and their interaction presented by the Figure 5. Victims will enter the
decision process which will consist of the triage procedure, after this the victims will be routed according
2437
Ullrich, Van Utterbeeck, Dejardin, Dhondt, and Debacker
to the triage procedure either to medical processes (diagnostic test, treatment) or transferred to another
department of the hospital.
4 MODEL AND RESULTS
4.1 Model
The model presented in this paper is the pilot model of the SIMEDIS project which simulates a major road
traffic accident. The vehicles involved in this accident are a truck and a bus, the number of victims is 62
with the following distribution of the triage category:
T1: 10
T2: 15
T3: 36
T4: 1
We wanted to study the influence of several parameters on performance indicators. We vary the number
of hospitals, Medical Mobile Teams (MMT), Advanced Life Support (ALS) ambulances, Basic Life Support
(BLS) ambulances, distances and the speed of the vehicles. The time of alerting and sending medical
mobile teams and ambulance is 5 minutes. If the “stay & play” policy is used, the victims are transferred
from the disaster scene to the forward medical post with a travel time of 2 minutes with stretchers. The
primary triage takes 30 seconds for urgent victims and 5 seconds for non-urgent victims. The distance
column (in the tables 2, 3, 4 and 5) correspond to the high level of resource for the distance. The medium
level is twice the distance for the high level (distanceMedium =2distanceH igh ) and the Low level is four
times the distance of the high level (distanceLow =4distanceHigh ).
Table 2 presents the three levels of health care facility resources.
Table 2: Health care facility
High Medium Low
Trauma HTC Trauma HTC Trauma HTC
ID distance Level T1 T2 T3 Level T1 T2 T3 Level T1 T2 T3
1 0.3 1 5 7 18 2 3 6 15 4 1 2 9
2 2 3 2 2 9 3 2 2 9 3 1 2 9
3 4 3 2 3 14 1 5 7 18 2 2 2 10
4 7 3 2 2 9 3 2 2 9 3 1 2 9
5 9 3 2 3 9 4 1 2 9 1 4 7 17
6 10 2 3 4 14 2 3 4 14
7 12 4 1 2 9 4 1 2 9
8 12 2 3 6 16 2 3 6 16
9 14 4 1 2 9
10 15 2 3 6 15
11 16 4 1 3 9
12 19 4 1 3 8
The Hospital Treatment Capacity (HTC) is the number of victims that the hospital can handle in one
hour for each triage category. The HTC for each level of resource is presented in Table 2.
Each mobile medical team is composed of a doctor and a nurse. The first MMT assumes the role
of medical director, the medical director ensures the logistic part for medical rescue teams. The second
MMT does the primary triage on the scene and when the primary triage is finished they help in medical
2438
Ullrich, Van Utterbeeck, Dejardin, Dhondt, and Debacker
interventions at the FMP or on site (in function of the rescue policy). The third and following MMT
ensure directly medical interventions at FMP or on-site. T1 victims need a doctor and a nurse to treat or
stabilize them, T2 victims need one doctor or nurse depending on available manpower. Table 3 shows the
availability of medical mobile teams as a function of the three level of resources. Tables 4 and 5 present
Table 3: MMT
ID Distance Number
High
Medium
Low
1 0,3 1
2 2 0/1/1
3 4 1
4 9 1
5 10 1
6 12 1
7 15 1
the three levels of evacuation means available. T1 victims can only be evacuated by ALS ambulances,
T2 victims are evacuated by BLS ambulances or ALS ambulances if there are no more T1 victims at the
forward medical post and/or on site area. ALS and BLS ambulances can transport only one victim per
journey.
Table 4: Ambulance BLS
ID Distance Number
High
Medium
Low
1 1 2
2 3 1
3 5 1
4 9 2
5 11 2
6 11 2
7 12 1
8 13 2
9 14 1
10 15 2
11 16 1
12 17 1
13 17 1
14 18 2
15 19 1
16 20 1
The last input parameter we studied is the velocity of rescue vehicle : MMT, ALS ambulances and
BLS ambulances.We considered two values for the velocity : High = 60 km/h and Low = 30 km/h.
4.2 Outcome Indicators
In the previous section, input parameters have been described. In this section we present the outcome
indicators used for this paper. We chose 4 outcome indicators which are: mortality, morbidity, time of
clearance, and last arrival at the health care facilities.
2439
Ullrich, Van Utterbeeck, Dejardin, Dhondt, and Debacker
Table 5: Ambulance ALS
ID Distance Number
High
Medium
Low
1 1 1
2 3 0
3 4 1
4 8 1
5 10 0
6 12 1
7 14 0
8 16 1
9 18 0
10 20 1
The first indicator is the mortality, which consists of two sub-indicators: the “immediate death” and
the “pre-hospital death”. The immediate death corresponds to a victim who dies before being seen by any
medical rescue team. The pre-hospital death is when a victim dies after the primary triage and before the
arrival at the health care facilities.
The second indicator is the morbidity, which corresponds to a deterioration of the RPM score or Glasgow
coma scale score between the primary triage and the arrival of the victims at HCF. This deterioration can
lead to permanent injury or incapacity in the future life of the victims.
The third outcome indicator is the time of clearance defined as the time necessary for the number of
victims at the disaster scene to reach 0. The number of victims decreases if the victim is evacuated or
died. We split the time of clearance for each triage category, so we define the time of clearance for T1,
T2 and T3.
The last outcome is the time needed by the last victim to arrive at the health care facilities. We split
the time of arrival for the last victim for each triage category, so we define last T1, T2 and T3 arrival at
health care facilities.
4.3 Results
For each possible couples of parameters, we ran 30 replications. The total number of couples is 2 ×3×
3×3×2=108. Therefore 30 ×108 =3240 replications where needed to obtain all the results presented
below. Once all the results were available for each outcome indicators, we tested them with an ANOVA
test or an Independent Samples Student test in function of each parameter in order to see their influences.
Table 6 shows which input parameters have a significant influence on which outcome indicators.
Table 6: Results
Death Immediate Time Clearance Time Clearance Last T1 Last T2
Death T1 T2 HCF HCF
Rescue policy X X X X
Hospitals X X X X X
Ambulances X X X X X
Distance
Speed X X X X X X
We can see that the policy of rescue has a significant influence on the total number of dead, the time
for the victims to be evacuated from the disaster site and to the hospital. The time to evacuate the victims
2440
Ullrich, Van Utterbeeck, Dejardin, Dhondt, and Debacker
T1 and T2 from the disaster site is shorter in the “stay & play” policy of rescue. In this scenario, the
victims are directly evacuated from the disaster site via the primary evacuation to a place near the site, the
Forward Medical Post, where the victims are sorted, treated and prepared for evacuation. In the “scoop
& run” policy there is no primary evacuation. The victims are evacuated from the disaster site directly to
the hospital. However, the evacuation can sometimes be delayed if a treatment is needed or if the hospital
capacities already are busy. On the contrary, the time for the last T1 to arrive at the hospital is shorter with
the “scoop & run” policy of rescue, where the victims are directly evacuated to the hospital, than with the
“stay & play” policy of rescue, where they first go to the Forward Medical Post for triage, treatment and
preparation for evacuation. There is no influence of the policy of rescue in the immediate dead because the
time for the ambulances to arrive on the disaster site is the same in the two policies. But, as said earlier,
the policy of rescue has an influence on the total number of dead, which is lower in the “stay & play”
policy. We can see in the Figure 6 that the median in the “stay & play” policy is lower. The quartile 1 and
3 are quite the same for the two policies, but the interquartile interval between quartile 3 and 4 is bigger
in the “scoop & run” policy. It can be explained by the fact that the victims receive a treatment earlier in
“stay & play” policy than in the “scoop & run” policy, giving the more critical victims more chance to
stay alive.
Figure 6: Death as function of rescue policy
There is an influence of the type of hospitals on nearly all the outputs: only the number of immediate
dead is not influenced by it, which is normal because that output depends only on the medical staff on the
site. All the other outputs are lower when the capacities of the hospitals are high: in this case, there are
more places available to evacuate the victims, which results in a shorter time to evacuate all the victims
and consequently in fewer dead.
The number of dead, immediate dead, the time to evacuate T1 and T2 from the site and to the hospital
are significantly influenced by the number of ambulances in the two policies of rescue. There are fewer
dead when there are more ambulances, which is normal: the more ambulances there are, the faster the
victims are seen by medical staff. With the same logic, the victims are evacuated more quickly when there
are more ambulances.
The speed influences significantly all the outputs. When the ambulances can go faster, they arrive
earlier on the disaster site and the victims are seen faster. That results in less dead and immediate dead
and the site is more rapidly evacuated.
2441
Ullrich, Van Utterbeeck, Dejardin, Dhondt, and Debacker
5 CONCLUSIONS AND FUTURE WORK
We presented a simulation model of pre-hospital medical disaster response using realistic victims. The
simulation model has been developed specifically to allow medical researchers to use it without changing
the model in ARENA. Improvements have been made to the main components of our previous model
(Van Utterbeeck et al. (2011)). The first one consists of the victim modeling and aims at representing
more realistic victims and at defining pathways with multiple intervention triggers which enables the use
of scarce resources for treatment. The victims profiles used in our model are highly realistic and their
validity is ensured by experienced professionals in disaster medical management. The second one concerns
the medical response model, where we define three main zones: disaster scene, forward medical post
and hospital, which are the essential parts of the pre-hospital medical response system. The decision
process based on the RPM (respiratory rate, pulse rate, motor response) score, the evolution of the survival
probability over time and evacuation information (ambulance availability and travel time) allows choosing
between treatment or evacuation of the victim. Threshold values used in the decision process have been
validated by medical experts.
A pilot case study describing a major road traffic accident has been studied for verification and
validation of the implemented medical response and victim pathway models and for performance and
outcome measures evaluation. We presented the impact of several input parameters on four outcomes
measures. This validation will also be useful to analyze the flexibility of our model and to improve the user
interface for medical researchers. The medical response model will be extended to the emergency room
processes. After the validation phase, future works will focus on continuing the development of the victim
profiles database and four scenarios will be investigated: an aeronautical catastrophe, a CBRNE (Chemical,
Biological, Radiological, Nuclear and Explosives) incident, mass gatherings and hospital catastrophes.
The expected outcomes of the SIMEDIS project are evidence based recommendations and rules of best
practices for optimal disaster management and medical battlefield management in different large-scale
event scenarios, as well as evidence based recommendations for teaching, training and research in medical
disaster management.
REFERENCES
Altiok, T., and B. Melamed. 2007. Simulation Modeling and Analysis with Arena. Academic press.
Brailsford, S. C., and N. A. Hilton. 2000. “A Comparison of Discrete Event Simulation and System Dynamics
for Modeling Healthcare Systems”. In Proceedings of ORAHS, 18–39. School of Management University
of Southampton, UK.
Cwinn, A. A., P. T. Pons, E. E. Moore, J. A. Marx, B. Honigman, and N. Dinerman. 1987. “Prehospital
Advanced Trauma Life Support for Critical Blunt Trauma Victims”. Annals of Emergency Medicine:399–
403.
Duguay, C., and F. Chetouane. 2007, April. “Modeling and Improving Emergency Department Systems
using Discrete Event Simulation”. SIMULATION 83 (4): 311–320.
Escudero-Marin, P., and M. Pidd. 2011. “Using ABMS to simulate emergency departments”. In Proceedings
of the 2011 Winter Simulation Conference, edited by S. Jain, R. R. Creasey, J. Himmelspach, K. P. White,
and M. Fu, 1239–1250. Piscataway, New Jersey: Institute of Electrical and Electronics Engineers, Inc.
G¨
unal, M. M., and M. Pidd. 2010, March. “Discrete event simulation for performance modelling in health
care: a review of the literature”. Journal of Simulation 4 (1): 42–51.
Kelton, W. D., R. P. Sadowski, and D. A. Sadowski. 2010. Simulation with Arena. 5 ed. McGraw-Hill.
McGuire, F. 1998. “Simulation in healthcare”. In Handbook of Simulation : Principles, Methodology,
Advances, Applications, and Practice, edited by J. Banks, Chapter 17, 605–627. Wiley-Interscience.
Mes, M., and M. Bruens. 2012. “A generalized simulation model of an integrated emergency post”.
In Proceedings of the 2012 Winter Simulation Conference, edited by C. Laroque, J. Himmelspach,
2442
Ullrich, Van Utterbeeck, Dejardin, Dhondt, and Debacker
R. Pasupathy, O. Rose, , and A. M. Uhrmacher, 872–882. Piscataway, New Jersey: Institute of Electrical
and Electronics Engineers, Inc.
NATO STANAG, 2879 1998. “Principles of Medical Policy in the Management of a Mass Casualty
Situation”.
Rossetti, M. D. 2010. Simulation Modeling and Arena. Wiley.
Sacco, W. J., D. M. Navin, K. E. Fiedler, R. L. Waddell II, W. B. Long, and R. F. Buckman Jr.. 2005.
“Precise formulation and evidence-based application of ressource-constrained triage”. Acad. Emerg.
Med. 12 (8): 759–770.
Su, S. 2003, December. “Modeling an emergency medical services system using computer simulation”.
International Journal of Medical Informatics 72 (1-3): 57–72.
Tomasini, R. M., and L. N. Van Wassenhove. 2009. “From preparedness to partnerships: case study research
on humanitarian logistics”. International Transactions in Operational Research 16:549–559.
Van Utterbeeck, F., C. Ullrich, E. Dhondt, M. Debacker, J. L. Murray, and S. Van Campen. 2011. “Generating
and Managing Realistic Victims for Medical Disaster Simulations”. In Proceedings of the 2011 Winter
Simulation Conference, edited by S. Jain, R. R. Creasey, J. Himmelspach, K. P. White, and M. Fu,
2679–2689. Piscataway, New Jersey: Institute of Electrical and Electronics Engineers, Inc.
Van Wassenhove, L. N., and A. J. Pedraza Martinez. 2012. “Using OR to adapt supply chain managemman
best practices to humatirian logistics”. International Transactions in Operational Research 19:307–322.
AUTHOR BIOGRAPHIES
CHRISTOPHE ULLRICH graduated from the University of Li`
ege as a chemical civil engineer with
an orientation in chemical process design and simulation in 2005. He obtained a PhD in engineering
science in 2010 from the same university. The main topic of his PhD research was chemical process
simulation and data reconciliation for dynamic chemical processes. He currently works as researcher at the
Department of Mathematics of the Royal Military Academy (RMA) in Brussels. His main area of research
is currently simulation optimization and its applications in medical disaster management. His e-mail is
christophe.ullrich@rma.ac.be.
FILIP VAN UTTERBEECK graduated from the Royal Military Academy (RMA) in Brussels as a poly-
technical engineer in 1995. He worked as a technical officer and material manager in the Air Defense branch
of the Belgian Air Force until 2002, when he returned to the RMA to teach at the Department of Mathemat-
ics. He obtained a PhD in engineering sciences from the Katholieke Universiteit Leuven and the RMA in
2011. He currently lectures several courses in the fields of Operations Research and Computer Simulation.
His main area of research is simulation optimization and its applications in human resource management,
supply chain management and medical disaster management. His e-mail is filip.van.utterbeeck@rma.ac.be.
EMILIE DEJARDIN graduated from the Royal Military Academy in Brussels as a in Social and Military
Sciences. She worked as Operation & Training Officer in the medical component of the Belgian Army
until when she returned to the RMA to teach at the Department of Mathematics and Statistics. Whith her
experience in Main Disaster Management and in Statistics, she’s helpful in the SIMEDIS project. Her
e-mail is emilie.dejardin@rma.ac.be.
MICHEL DEBACKER graduated from the Vrije Universiteit Brussel as medical doctor in 1970. Subse-
quently, he specialized in internal medicine, intensive care and emergency medicine. As military physician
he was chairman of the Joint Medical Committee at NATO from 1995 to 1997. He was the co-founder of
the European Master in Disaster Medicine (EMDM) in 1999 and the Emergency Management and Disaster
Medicine Academy in 2008. He is currently working as professor in disaster medicine at the Research
Group on Emergency and Disaster Medicine of the Vrije Universiteit Brussel. His main area of research is
the triage process and modeling of the disaster medical response. His e-mail is Michel.debacker@vub.ac.be.
2443
... Several studies modeled on-scene treatment using DES. For instance, de Oliveira & Toscano [17] developed an entire emergency care system for treatment on the scene or Debacker et al. [33] and Ullrich et al. included this step in their pre-hospital emergency response studies [34]. Niessner et al. [1] and Rauner et al. [35] also developed DES models considering on-field treatment in some treatment rooms. ...
... Therefore, the treatment can be delayed for up to two hours, while the red urgency level signifies the highest severity of the injury; that is, the life of the injured person is in danger, and they must receive treatment immediately. The distribution of the casualties on arrival at the hospital usually comprises 20% red level, 30% yellow level, and 50% green level patients [34]. This distribution pattern, which is derived from previous work, is used in designing these scenarios. ...
Article
Crisis occurrence in the healthcare context is, for different reasons, a phenomenon that happens abundantly. The priority of the healthcare system during a crisis is to provide quality care and superior services to the injured people. However, given the usually extreme severity of the crisis that results in a significant number of injured people, proper and timely responsiveness of healthcare systems is a challenging issue This study proposes a novel framework using a hybrid simulation–optimization approach to measure the healthcare responsiveness in crisis to address this real-world problem. This paper closely connects operations research techniques to critical systems thinking notions to evaluate the behavior of a system in the face of crisis. Since all arriving casualties to the hospital are first taken to the emergency department (ED), the ED in a case study is used to illustrate the performance of the presented approach. We designed seven crisis scenarios and one scenario of the ED system in a normal situation and modeled them using discrete-event simulation (DES). Patients’ interarrival times act as the driver of workload experienced in ED during crisis scenarios of varying severity. For crisis simulation scenarios that are unable to cope with the severity of the crisis, we developed an optimization model in an optimization tool to determine the optimal configuration of resources. The optimal configuration can improve healthcare resilience. The results show that an interarrival time of 13.8 min is the maximum threshold, below which feasible solutions could not be found, and the ED system is likely to collapse.
... To validate the correctness of the model's logic, the movement of patients and vehicles and the occurrence of every event in the model were tracked in the simulation log. The model behaviour was evaluated in a pilot study in which the response to normal as well as extreme situations was examined in order to determine if the model output behaves as expected [53]. Face-validity of the model was tested by users and subject matter experts who assessed the accuracy and consistency of the simulation output and outcome data compared to the real-world system. ...
... However, looking at the frequency distribution of the subtraction of the number of deaths in de stayand-play policy and the number of deaths in the scoop-andrun policy [Δdeath = Deaths (Policy = S&P) -Deaths (Policy = S&R)], we found that in 11 % of the 1152 scenarios, the stay-and-play policy is the best option to minimize the number of deaths. Moreover, this result may not be generalized to other MCIs or DMRS, since for instance in the pilot study to validate the model the stay-and play policy had a lower mortality in a pileup scenario [53]. ...
Article
Full-text available
It is recognized that the study of the disaster medical response (DMR) is a relatively new field. To date, there is no evidence-based literature that clearly defines the best medical response principles, concepts, structures and processes in a disaster setting. Much of what is known about the DMR results from descriptive studies and expert opinion. No experimental studies regarding the effects of DMR interventions on the health outcomes of disaster survivors have been carried out. Traditional analytic methods cannot fully capture the flow of disaster victims through a complex disaster medical response system (DMRS). Computer modelling and simulation enable to study and test operational assumptions in a virtual but controlled experimental environment. The SIMEDIS (Simulation for the assessment and optimization of medical disaster management) simulation model consists of 3 interacting components: the victim creation model, the victim monitoring model where the health state of each victim is monitored and adapted to the evolving clinical conditions of the victims, and the medical response model, where the victims interact with the environment and the resources at the disposal of the healthcare responders. Since the main aim of the DMR is to minimize as much as possible the mortality and morbidity of the survivors, we designed a victim-centred model in which the casualties pass through the different components and processes of a DMRS. The specificity of the SIMEDIS simulation model is the fact that the victim entities evolve in parallel through both the victim monitoring model and the medical response model. The interaction between both models is ensured through a time or medical intervention trigger. At each service point, a triage is performed together with a decision on the disposition of the victims regarding treatment and/or evacuation based on a priority code assigned to the victim and on the availability of resources at the service point. The aim of the case study is to implement the SIMEDIS model to the DMRS of an international airport and to test the medical response plan to an airplane crash simulation at the airport. In order to identify good response options, the model then was used to study the effect of a number of interventional factors on the performance of the DMRS. Our study reflects the potential of SIMEDIS to model complex systems, to test different aspects of DMR, and to be used as a tool in experimental research that might make a substantial contribution to provide the evidence base for the effectiveness and efficiency of disaster medical management. Electronic supplementary material The online version of this article (doi:10.1007/s10916-016-0633-z) contains supplementary material, which is available to authorized users.
... However, an information system and changes in the health status of the casualties are not considered during the process. Ullrich et al. (2013) simulated the pre-hospital processes in the In the study, triage procedures and medical service processes were evaluated, but DSS were not mentioned. Gul and Guneri (2015) conducted a simulation study with different triage possibilities, according to Istanbul's earthquake magnitudes. ...
Article
Purpose – The purpose of this paper is to propose a data-driven casualty transportation system based on Radio Frequency Identification technology to improve the related process during the response stage. Besides, it aims to statistically assess the effects of the design factors on the casualty transportation system's performance in large-scale disasters. Design/methodology/approach – In this study, we applied the simulation and factor effect analysis to evaluate the proposed casualty transportation process based on the data-driven decision support tool. An experimental design is made, and sixteen scenarios are formulated. Simulation models are developed for all scenarios, and a factor effect analysis is applied to the simulation results. Finally, the factor effect analysis results are assessed, and the significance of the factors is discussed. Findings – The sixteen scenarios are compared according to the performance criteria that are the number of unrecoverable casualties and time-spent by casualties until arriving at the hospital. It is seen that the proposed tool is more suitable than the current system. The results demonstrate the effectiveness of the proposed system based on factors. The managerial implications were presented in order to make suitable and timely decisions. Originality/value – In this study, both the data-driven decision support tool, the system's design process, and the casualty transportation process were examined together. This research considers the whole systems’ components, contributes to developing the response stage operations by filling gaps between using the decision support tool and casualty transportation.
... It also tracks effect time that is very important in such situations when resources are limited and need to be tracked when they will be free again. For example, a doctor checks five patients in 25 minutes, rescue workers can send five more patients after 25 minutes when the doctor is free again [17]. ...
Article
Full-text available
Disaster is a sudden accident or a catastrophic calamity that causes incredible damage or loss of life. Disaster has different types like Tornadoes, Floods, Wildfires, and Earthquakes. When a disaster occurs, many people get injured and many people die due to delay in timely treatment. But in a traditional rescue system, rescue workers are unaware of suitable and nearest health centers (hospitals) per patient condition. Rescue teams need to be updated about the capacity of the hospital and to know the shortest route to bring disaster patients to the most suitable hospital in minimum possible time. If resources are not available or occupied once they have arrived, retransfer from one hospital to another will be required which takes longer time and in severe condition, the patient could die. Smart Resource Allocation and Information System increases the chances of life in disaster by providing timely treatment. With the help of Smart Resource Allocation and Information System, the rescue teams will be aware of the shortest path, availability of specialist per patient condition and capacity of the hospital where disaster patient is to be assigned. So, the application allows timely treatment and better resuscitation services for catastrophic victims. Our designed system provides the solution for patient load balancing and patient load migration, better utilization of available resources, especially in resource constraint scenarios.
... In this paper we describe a comprehensive MCI medical management simulator incorporating search and rescue, on-site triage and treatment of casualties , their evacuation and admission to the emergency department of healthcare facilities, based on previous work (Debacker et al. 2016;Van Utterbeeck et al. 2011;Ullrich et al. 2013) . The objective of this project is to analyze and test the effect of MCI medical management policies and procedures in order to identify the optimal response strategies for various MCI scenarios. ...
... Specialized Transport [215][216][217][218][219][220][221][222] DES (2); ABS (1); Misc. (5) Cross-institutional (8) Arena (2); ArcGIS (1); Google Cloud (1); Microsoft Excel (1); Misc. ...
Article
Full-text available
Background Resource allocation in patient care relies heavily on individual judgements of healthcare professionals. Such professionals perform coordinating functions by managing the timing and execution of a multitude of care processes for multiple patients. Based on advances in simulation, new technologies that could be used for establishing realistic representations have been developed. These simulations can be used to facilitate understanding of various situations, coordination training and education in logistics, decision-making processes, and design aspects of the healthcare system. However, no study in the literature has synthesized the types of simulations models available for non-technical skills training and coordination of care. Methods A systematic literature review, following the PRISMA guidelines, was performed to identify simulation models that could be used for training individuals in operative logistical coordination that occurs on a daily basis. This article reviewed papers of simulation in healthcare logistics presented in the Web of Science Core Collections, ACM digital library, and JSTOR databases. We conducted a screening process to gather relevant papers as the knowledge foundation of our literature study. The screening process involved a query-based identification of papers and an assessment of relevance and quality. Results Two hundred ninety-four papers met the inclusion criteria. The review showed that different types of simulation models can be used for constructing scenarios for addressing different types of problems, primarily for training and education sessions. The papers identified were classified according to their utilized paradigm and focus areas. (1) Discrete-event simulation in single-category and single-unit scenarios formed the most dominant approach to developing healthcare simulations and dominated all other categories by a large margin. (2) As we approached a systems perspective (cross-departmental and cross-institutional), discrete-event simulation became less popular and is complemented by system dynamics or hybrid modeling. (3) Agent-based simulations and participatory simulations have increased in absolute terms, but the share of these modeling techniques among all simulations in this field remains low. Conclusions An extensive study analyzing the literature on simulation in healthcare logistics indicates a growth in the number of examples demonstrating how simulation can be used in healthcare settings. Results show that the majority of studies create situations in which non-technical skills of managers, coordinators, and decision makers can be trained. However, more system-level and complex system-based approaches are limited and use methods other than discrete-event simulation.
... Ullrich, Debacker and Dhondt developed a simulation model focusing on the prehospital phase consisting of field triage, evacuation and medical processes. With the model, they studied the effects of several parameters, such as the number of hospitals, medical teams and ambulances on several performance indicators [4]. ...
Article
Full-text available
We present a detailed integrated simulation model for casualty transportation and emergency room management after a major disaster. Usually, the two important processes are handled separately in literature. However, to be able to correctly evaluate preparedness of disaster and emergency and minimize potential loss of lives, these two interconnected processes must be analysed together. We present a case study for the proposed integrated simulation model using real life data for Bornova, İzmir in Turkey. Results obtained by running the simulation model with a detailed experimental design are statistically analysed and findings are discussed.
Article
Full-text available
We propose a methodology to generate realistic victim profiles for medical disaster simulations based on victims from the VictimBase library. We apply these profiles in a medical disaster model where victim entities evolve in parallel through a medical response model and a victim pathway model. These models interact in correspondence with the time triggers and intervention triggers from VictimBase. We show how such a model can be used to assess the impact of asset availability and implemented victim prioritisation rule on the clinical condition of the victims.
Conference Paper
Full-text available
This paper discusses the development of a discrete-event simulation model for an integrated emergency post. This post is a collaboration between a general practitioners post and an emergency department within a hospital. We present a generalized and flexible simulation model, which can easily be adapted to several emergency departments as well as to other departments within the hospital, as we demonstrate with our application to the integrated emergency post. Here, generalization relates to the way we model patient flow, patient prioritization, resource allocation, and process handling. After presenting the modeling approach, we shortly describe the implemented and validated model of the integrated emergency post, and describe how it is currently being used by health care managers to analyze the effects of organizational interventions.
Article
The aim of this chapter is to discuss whether the choice of simulation methodology - discrete-event simulation or system dynamics - is purely the personal preference of the modeller, or whether there are identifiable features of certain problems that make one approach intrinsically superior to the other. Although from a methodological standpoint the overall comments are generic and applicable to any setting, the chapter has a bias towards health care applications as this is the area of domain expertise of the author. A case study in emergency care is presented, where both DES and SD were used to tackle different aspects of the overall problem. The chapter concludes with some general guidelines to assist the modeller in choosing the technique.
Conference Paper
Computer simulation methods have enjoyed widespread use in healthcare system investigation and improvement. Most reported applications use discrete event simulation, though there are also many reports of the use of system dynamics. There are few reports of the use of agent-based simulations (ABS). This is curious, because healthcare systems are based on human interactions and the ability of ABS to represent human intention and interaction makes it an appealing approach. Tools exist to support both conceptual modelling and model implementation in ABS and these are illustrated with a simple example from an emergency department.
Article
Computer simulation methods have enjoyed widespread use in healthcare system investigation and improvement. Most reported applications use discrete event simulation, though there are also many reports of the use of system dynamics. There are few reports of the use of agent-based simulations (ABS). This is curious, because healthcare systems are based on human interactions and the ability of ABS to represent human intention and interaction makes it an appealing approach. Tools exist to support both conceptual modelling and model implementation in ABS and these are illustrated with a simple example from an emergency department.
Book
The textbook which treats the essentials of the Monte Carlo discrete-event simulation methodology, and does so in the context of a popular Arena simulation environment. It treats simulation modeling as an in-vitro laboratory that facilitates the understanding of complex systems and experimentation with what-if scenarios in order to estimate their performance metrics. The book contains chapters on the simulation modeling methodology and the underpinnings of discrete-event systems, as well as the relevant underlying probability, statistics, stochastic processes, input analysis, model validation and output analysis. All simulation-related concepts are illustrated in numerous Arena examples, encompassing production lines, manufacturing and inventory systems, transportation systems, and computer information systems in networked settings. Chapter 13.3.3 is on coal loading operations on barges/tugboats.
Article
The demand for humanitarian aid is extraordinarily large and it is increasing. In contrast, the funding for humanitarian operations does not seem to be increasing at the same rate. Humanitarian logistics has the challenge of allocating scarce resources to complex operations in an efficient way. After acquiring sufficient contextual knowledge, academics can use operations research (OR) to adapt successful supply chain management best practices to humanitarian logistics. We present two cases of OR applications to field vehicle fleet management in humanitarian operations. Our research shows that by using OR to adapt supply chain best practices to humanitarian logistics, significant improvements can be achieved.
Chapter
Introduction Comparison of System Types Research: The First Step Steps in a Simulation Project Barriers to Implementation and How to Deal with Them Problems Unique to Healthcare During Implementation Planning Case Study Summary Appendix 17.1: Emergency Department Process Survey Appendix 17.2: Data Collection Criteria for an Operating Room Process Simulation Project Appendix 17.3: Data Collection Criteria for an Emergency Department Simulation Project References