USF Libraries
USF Digital Collections

Predicting the medical management requirements of large scale mass casualty events using computer simulation

MISSING IMAGE

Material Information

Title:
Predicting the medical management requirements of large scale mass casualty events using computer simulation
Physical Description:
Book
Language:
English
Creator:
Zuerlein, Scott A
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Computer modeling
Simulation
Blasts
Planning
Emergency care
Dissertations, Academic -- Health Policy and Management -- Doctoral -- USF   ( lcsh )
Genre:
non-fiction   ( marcgt )

Notes

Abstract:
ABSTRACT: Recent events throughout the world and in the US lend support to the belief that another terrorist attack on the US is likely, perhaps probable. Given the potential for large numbers of casualties to be produced by a blast using conventional exlosives, it is imperative that health systems across the nation consider the risks in their jurisdictions and take steps to better prepare for the possibility of an attack. Computer modeling and simulation offers a viable and useful methodology to better prepare an organization or system to respond to a large scale event. The real question, given the shortage, and in some areas absence, of experiential data, could computer modeling and simulation be used to predict the resource requirements generated by this type of event and thus prepare a health system in a defined geographic area for the possibility of an event of this nature? Research resulted in the identification of variables that surround a health system at risk, the development of a computer model to predict the injuries that would be seen in an injured survivor population and the medical resources required to care for this population. Finally, methodologies were developed to modify the existing model to match unique health system structures and processes in order to assess the preparedness of a specific geographic location or health system. As depicted in this research, computer modeling and simulation was found to offer a viable and usable methodology for a defined geographic region to better prepare for the potential of a large scale blast event and to care for the injured survivors that result from the blast. This can be done with relatively low cost and low tech approach using existing computer modeling and simulation software, making it affordable and viable for even the smallest geographic jurisdiction or health system.
Thesis:
Dissertation (Ph.D.)--University of South Florida, 2009.
Bibliography:
Includes bibliographical references.
System Details:
Mode of access: World Wide Web.
System Details:
System requirements: World Wide Web browser and PDF reader.
Statement of Responsibility:
by Scott A. Zuerlein.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 295 pages.
General Note:
Includes vita.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 002028887
oclc - 436688735
usfldc doi - E14-SFE0002836
usfldc handle - e14.2836
System ID:
SFS0027153:00001


This item is only available as the following downloads:


Full Text

PAGE 1

Predicting the Medical Management Requirement s of Large Scale Mass Casualty Events Using Computer Simulation by Scott A. Zuerlein A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Health Policy and Management College of Public Health University of South Florida Major Professor: Alan M. Sear, Ph.D. Barbara L. Orban, Ph.D. James Studnicki, Sc.D. Yiliang Zhu, Ph.D. Date of Approval: February 27, 2009 Keywords: computer modeling, simulation, blasts, planning, emergency care, care processes, health system prepar edness, health system resources Copyright 2009, Scott A. Zuerlein

PAGE 2

i Table of Contents List of Tables iv List of Figures vi Abstract vii Chapter 1: Introduction/Statement of the Rese arch Problem 1 Chapter 2: Review of the Lite rature 8 Systems Theory, Computer Modeling and Simulation 9 Research Methods 16 Arena specific computer modeling structures/methods 20 Probability Distributions 25 Verification and Validation 28 Blast Events and Terrorist Actions 30 Capacity (Structure) of the Health Care System and Mass Casualty 37 Events Public Policy 43 Studies Supporting the Development of a Computer Simulation Model 46 Chapter 3: Objectives of the Research 64 Chapter 4: Research Methods 70 Methods for Objective 1 71 Methods for Objective 2 75 Methods for Objective 3 81 Methods for Objectives 4 and 5 86 Methods for Objective 6 94 Chapter 5: Results 96 Results for Objective 1 97 Facility Characteristics 97 Frequency of Use 99 Structural Characteristics 100 Explosive Delivery Mechanisms 106 Location of Blast 110 Strength of Blast 111 Results for Objective 2 113

PAGE 3

ii Results for Objective 3 121 Results for Objectives 4 ad 5 147 Results for Objective 6 167 Chapter 6: Summary and Conclusions 183 Chapter 7: Limitations of the Study 197 References 202 Appendices 208 Appendix 1: Discussion of Add itional Probability Distributions Available In Arena Software 209 Appendix 2: Bombings of Fixed Structures Using Conventional Explosives: 1988 to 2005 214 Appendix 3: Distribution of Out door Stadiums within the United States (includes domed stadiums) 217 Appendix 4: Distribution of Indoor Arenas within the United States 218 Appendix 5: Example of a Resource Dens e Geographic Location 219 Appendix 6: Example of a Resource Spar se Geographic Location 220 Appendix 7: The Barell Injury Diag nosis Matrix 221 Appendix 8: The Barell Injury Di agnosis Matrix, Cl assification by Body Region and Nature of the Injury 222 Appendix 9: Matrix of Studies Identifyi ng Injury by Body Region 228 Appendix 10: Matrix of Studies Identifying Injury by Injury Type 229 Appendix 11: In Hospital Emergency Response Models 230 Appendix 12: Process and Decisi on Parameters as Defined in Hirshberg Study 233 Appendix 13: Distribution of Arena St adiums by State 234 Appendix 14: Arenas and Stadiums by State 236 Appendix 15: Distribution of Indoor Arenas by Location Population 247 Appendix 16: Distribution of Outdoor Stadiu ms by Location Population 247 Appendix 17: List of Explosive Ma terials 248 Appendix 18: Injury Prediction Model Ba sed on Frykberg Study 254 Appendix 19: Injury Prediction Model Based on Barell Injury Matrix 257 Appendix 20: ICD-9-CM Codes and Inju ry Description to Match Injury Prediction Model Based on Frykberg Study (Appendix 18) 261 Appendix 21: Resource Prediction Model 274 Appendix 22: Simulation Decision Module, Process Module, and Termination Module Parameters 280 Appendix 23: Simulation Results 7,000 Injured Survivors (no constraints) 288

PAGE 4

iii Appendix 24: Sim ulation Results 45,000 Injured Survivors (no constraints) 289 Appendix 25: Simulation Results 7,000 Injured Survivors (12 hours, resource sparse) 290 Appendix 26: Simulation Results 7,000 Injured Survivors (12 hours, resource dense) 291 Appendix 27: Simulation Results 45,000 Injured Survivors (12 hours, resource sparse) 292 Appendix 28: Simulation Results 45,000 Injured Survivors (12 hours, resource dense) 293 Appendix 29: Simulation Results 7,000 Injured Survivors (12 hour resource constraints) 294 Appendix 30: Simulation Results45,000 Injured Survivors (12 hour resource constraints) 295 About the Author End Page

PAGE 5

iv List of Tables Table 1. P robability Distributions Available in Arena 26 Table 2. Published studies providing data/inf ormation for this research 47 Table 3. CDC Overview of Explosive -related Injuries 50 Table 4. Terrorist Bombing Injury Analysis (220 events, 2,934 immediate survivors) 52 Table 5. Oklahoma City Injury Analysis ( 592 Injured Survivors) 53 Table 6. Oklahoma City Location of Soft Tissue Injuries (506 injuries) 54 Table 7. Oklahoma City Location of Musculoskeletal Injuries (Fractures And Dislocations (60 injuries) and Spra ins (152 injuries)) 55 Table 8. Peleg Study Injury Categoriz ation (623 patients) 57 Table 9. World Trade Center Attack Surviv ors (790 injured survivors) 58 Table 10. Khobar Towers Injury Analysis ( 401 injured survivors) 60 Table 11. Abbreviated Injury Scale 62 Table 12. Injury Severity Scores among Published Studies 63 Table 13. Terrorist related blast events causing more th an 500 injured survivors 76 Table 14. Simulation Run Matrix 87 Table 15. Resource Constraints 90 Table 13. Injury Prediction Model (Fr ykberg Data, 1988) 116 Table 14. Injury Prediction Model (Nature of Injury categories) 119 Table 15. Injury Predictions (7,000 in jured survivors, no constraints, 10 replications) 148 Table 16. Injury Predictions (45,000 injured survivors, no constraints, 10 replications) 150 Table 17. Injury Predictions (7,000 in jured survivors, 12 hours/resource dense, 10 replications) 152 Table 18. Injury Predictions (45,000 inju red survivors, 12 hours/resource dense, 10 replications) 153 Table 19. Comparison (7,000 injured survivors, 10 replication means) 154 Table 20. Comparison (45,000 injured survivors, 10 replication means) 156 Table 21. Resource Requirement Prediction (7,000 injured surv ivors, no constraints, 10 replications) 157 Table 22. Resource Requirement Prediction (45,000 injured survivors, no constraint s, 10 replications) 159 Table 23. Resource Requirement Predic tion (7,000 injured survivors, 12 hours/resource dense, 10 replications) 160

PAGE 6

v Table 24. Resource Requirem ent Predic tion (7,000 injured survivors, 12 hours/resource sparse, 10 replications) 161 Table 25. Scenario Comparison of Predicted Resource Requirements (7,000 injured survivors, 10 replication means) 163 Table 26. Scenario Comparison of Predicted Resource Requirements (45,000 injured survivors, 10 replication means) 164 Table 27. 7,000 Injured Survivors/Resour ce Dense Geographic Area (8 hour window) 168 Table 28. 7,000 Injured Survivors/Resour ce Sparse Geographic Area (8 hour window) 172 Table 29. 45,000 Injured Survivors/Res ource Dense Geographic Area (8 hour window) 173 Table 30. 45,000 Injured Survivors/Resource Sparse Geographic Area (8 hour window 174 Table 31. 7,000 Injured Survivors/Res ource Dense Geographic Area (12 hour window) 175 Table 35. 7,000 Injured Survivors/Resour ce Sparse Geographic Area (12 hour window) 176 Table 36. 45,000 Injured Survivors/Res ource Dense Geographic Area (12 hour window) 177 Table 37. 45,000 Injured Survivors/Resource Sparse Geographic Area (12 hour window) 178 Table 38. 7,000 Injured Survivors/Resour ce Dense Geographic Area 179 Table 36. 45,000 Injured Survivors/Resource Sparse Geographic Area 181

PAGE 7

vi List of Figures Figure 1. Sim ulation Model Example 21 Figure 2. Casualty Flow in a Disaster Situation 83 Figure 3. Basic Structure of the Simu lation Model Used for the Present Research 84 Figure 4. Indoor Facility above Ground Structure (Conseco Fieldhouse, Indianapolis, IN) 103 Figure 5. Indoor Facility below Ground St ructure (University Arena, Albuquerque, NM) 104 Figure 6. Outdoor Fac ility above Ground Structure (Raymond James Stadium Tampa, Florida) 107 Figure 7. Outdoor Facility below Gr ound Structure (Michigan Stadium Ann Arbor, MI) 108 Figure 8. Injury Prediction Al gorithm 115 Figure 9. Resource Prediction Algorithm (Res cue/Transportation Component) 123 Figure 10. Simulation Module Descripti ons 125 Figure 11. Resource Prediction Algorithm (CCP Component) 130 Figure 12. Resource Prediction Algorithm (Hos pital Component) 135 Figure 13. Resource Prediction Algorithm (Surgery Component) 139 Figure 14. Resource Prediction Algorithm (Sur gery Component) 141 Figure 15. Resource Prediction Algorithm (Hosp ital (2) Component) 144

PAGE 8

vii Predicting the Medical Managem ent Require ments of Large Scale Mass Casualty Events Using Computer Simulation Scott A. Zuerlein ABSTRACT Recent events throughout the world and in the US lend support to the belief that another terrorist attack on th e US is likely, perhaps probabl e. Given the potential for large numbers of casualties to be produced by a blast using conventi onal explosives, it is imperative that health systems across the nati on consider the risks in their jurisdictions and take steps to better prepare for the possibility of an attack. Computer modeling and simulation offers a viable and useful met hodology to better prepare an organization or system for the occurrence of a one time catastrophic event. The objective of this research was to determine if computer modeling a nd simulation offered a viable methodology to prepare a health system to respond to a large scale event. The real question; given the shortage, and in some areas ab sence, of experiential data, could computer modeling and simulation be used to predict the resource re quirements generated by this type of event and thus prepare a health system in a define d geographic area for the possibility of an event of this nature? Research resulted in the id entification of variables that surround a health system at risk, the development of a computer simulation model to predict the injuries that would be seen in an injured survivor populat ion and the medical resources required to care for this population. Fina lly, methodologies were developed to modify the existing model to match unique health syst em structures and processes in order to

PAGE 9

viii assess the preparedness of a sp ecific geographic location or he alth system. As depicted in this research, computer modeling and si mulation was found to offer a viable and usable methodology for a defined geographic regi on to better prepare fo r the potential of a large scale blast event and to care for the in jured survivors that result from the blast. This can be done with a relatively low co st and low tech approach using existing computer modeling and simulation software, making it affordable and viable for even the smallest geographic jurisdiction or health system.

PAGE 10

1 Chapter 1 Introductio n/Statement of the Research Problem Historically, blasts and pu rposeful bombings using c onventional explosives have resulted in thousands of deaths and injuri es. In December of 1917, an ammunition ship exploded in the harbor of Halifax, Nova Sco tia and left an area two miles around the site completely destroyed (Kernaghan, 2004). Terror ist bombings in the United States date as early as 1920 when a TNT bomb plante d in a horse-drawn wagon exploded on Wall Street killing 35 and injuring hundreds more (T errorist Attacks, 2004). The last 25 years have witnessed multiple acts of terrorism wh ich have garnered headlines and produced mass casualties. In 1983, suicide bombers exploded a truck near the U.S. military barracks in Beirut, Lebanon, k illing 234 immediately and leav ing another 112 injured of which 7 subsequently died (Frykberg, 1989). In 1993, a bomb exploded in the basement of the World Trade Center killing six and inju ring another 1,040 (Terro rist Attacks, 2004). In 1995, a truck bomb exploded outside the fe deral building in Oklahoma City which killed 168 and injured an additional 591 (T eague, 2004). In 1996, a truck bomb exploded outside the Khobar Towers military complex in Dhahran, Saudi Arabia, killing 19 and directly injuring another 401 (Thompson, 2004). In 1998, truck bombs exploded near the U.S. embassies in Nairobi, Kenya and Dar es Salaam, Tanzania, killing 213 in Kenya and 11 in Tanzania and injuring 4,500 (Terrorist Attacks, 2004). In 2001, the attack which destroyed the World Trade Ce nter, though not a bombing per se, killed 2,749 and injured 1,103 (Terrorist Attacks, 2004 and MMWR 2004). In 2004, 10 bombs exploded

PAGE 11

2 sim ultaneously on trains in Madrid, Spai n, killing 202 and injuring more than 1,400 (Terrorist Attacks, 2004). In 2007, five bombing s, four at a crowded bus station, killed at least 500 and injured at least 320 in two Yazidi villages in Northern Iraq (Iraqi Officials, 2007). This list does not address the ma ny comparatively small scale incidents experienced by Israel, Iraq, and other nations where individual suicide bombers kill and injure tens and twenties at a time. These events and many others throughout the world over the last 25 years lead one to conclude that there is a reasonable likeli hood of a conventional explosive event occurring again in the United States in the next five to ten years. Potential targets with large numbers of civilians throughout the United Stat es offer easy access to terrorists. On any given day large numbers of fans gather to cheer on their local s ports team. Outdoor sports venues often contain 30,000-100,000 sp ectators, while indoor venues contain 10,000-30,000. Many other sites (such as a conven tion center) host larg e gatherings of people. The potential for large numbers of deaths and injuries if a terrorist were to strike one of these venues is very high. A conventional explosive placed at an outdoor or indoor venue of this size c ould kill thousands of people out right and injure tens of thousands at one time. There is no litera ture to guide the pub lic health/medical community in how to deal with conventi onal explosive events on this scale. If an attack occurs, the medical infrastruc ture of the region becomes the first line of response to sustain and restore life. Any city (or region) that has not at least considered the possibility of an attack may find that it is unprepared in the event of a large scale mass casualty event. The medical infrastruc ture must be prepar ed to respond to, and

PAGE 12

3 m anage, the aftermath of an attack that kills thousands outright and injures multiple thousands more, for whom medical care must be provided. The problem presented here is not one of prevention. Prevention efforts are the concern of other entities. The concern for the public health/medical community is the resulting injuries if preventi on efforts are unsuccessful and a large scale blast event using conventional explosives does occur. If an attack does occur which results in a large number of casualties, the im mediate problem presented to the medical community is how best to respond and, how best to care for the injured. Effectively and accurately assessing the preparedness of a medical system to respond to a mass casualty event is difficult. Polic ies and procedures are often established, training occurs and exercises are conducted, how ever the ability to effectively test the adequacy of these systems is limited at best, especially in response to a large scale mass casualty event. As a result, an accurate assessment of the adequacy of these preparation efforts cannot be made using conventional methods. This research focuses on this belief; that an accurate assessment of the readiness of a given jurisdiction to manage a large s cale mass casualty event cannot be made using conventional methods. Multiple factors combine to make this problem worthy of research and demanding of a solution. These factors each shape the problem and set the context for further research. The first of these factors becomes the driver for all the others. The existence of a threat in the form of a terrorist attack is not questioned. Historical evidence, recent trends with respect to targets and the magnitude of the blasts (attacks), and the ongoing efforts of the United States to eradic ate terrorists and terrorism place the citizens of this country

PAGE 13

4 at risk of further attacks. Conceding that th er e is a threat or risk of a terrorist attack, identifying the most likely targets, or where the threat is the str ongest, is less certain. Part of the problem facing the public health/medical community is that one cannot accurately answer the questions of what are the most likely targets and where are those targets located? The problem then becomes one of how best to focus resources wherever a blast may occur. These unknowns hinder the ab ility of a given jurisdiction to prepare in an effective manner. Poor preparation is not b ecause of lack of effort or lack of concern, but rather incomplete information as to th e risk and requirements which may be placed on the system. Another factor contributing to the prep aredness problem for those in the United States is the lack of experience in dealing with terrorist attacks. Most health systems in the United States do not have experience in dealing with terrorist attacks, and in particular, do not have experience dealing with the medical management requirements that may result from a large scale terroris t attack using conven tional explosives. Thankfully, this lack of experience is prim arily a result of not having opportunities to respond to terrorist attacks. Th is is contrasted with Israel where terrorist attacks have been a common occurrence. The medical commun ity in Israel deals with the effects of terrorist attacks on a regular basis. They have experience as a result of their day-to-day activity. Trauma systems throughout the Un ited States have experienced trauma providers and support staffthey deal with trauma on a daily basis. What providers in the United States do not have is experience in dealing with mass casualty events, in general, and large scale mass cas ualty events in particular. This is a problem not easily addressed. It is only in responding to a mass casualty event that medical (and other

PAGE 14

5 responders) gain m ass casualty experience. The ability to gain wisdom from those with experience and to learn from past actions is critical to ensure readiness to handle a mass casualty event, in general, or a large s cale mass casualty event, in particular. Another contributing factor is the perspect ive or belief concerning readiness held by the medical community and policy makers. Confidence in the system in general may thwart any encouragement toward additional preparedness. Conversely, the mindset of those involved may also lend encouragement to not addressing this problem. If they perceive that there is no problem, then offe ring them a tool to help the planning and preparedness process will be worth very little Additionally, if resources are focused at prevention efforts and detection of weapons of mass destruction, attention will be drawn away from the more likely conventiona l explosive blast events/attacks. Pertinent to the problem of preparedness is an analysis or predic tion of the type of blast event that is likely to occur. Identif ying the likely (or possible) type of event (or blast) presents the first problem. Bombs a nd other explosive devices are composed and delivered in a variety of different ways. Th e way the bomb is constructed, its size, and the means by which it is delivered are factor s that directly influence the casualties produced. Characteristics of the location of a blast; the facility structure and the number of people involved combine with the specific characteristics of the blas t itself to produce a unique mix of casualties. The structure of the location is a str ong determinant of the impact in terms of casualties. Differences in casualties will occur between indoor and outdoor facilities and between differences in th e placement of the blast device in relation to the facility. An additional aspect of the problem is to identify the specific mix of

PAGE 15

6 casualtie s. This includes identifying the num bers that may be killed immediately by the blast or who die before medical care can be provided. For those surv iving the blast, the types and number of injuries requiring care will be the driver to determine the medical resources required to manage the care. Another component of the medical manageme nt challenge is where and how best to care for the casualties. Some will require tr auma care and thus rapid transportation to a designated trauma center. Some will require hospitalization. Others will require very minimal care, but in their haste to access care, they may prevent the more seriously injured from receiving the care they need. Among this mix of multiple casualties is the difficulty of determining who is seriously inju red, and in need of immediate care, versus those who are not seriously in jured and can wait for care. Given the risk, inherent problems with pr eparing for a large scale mass casualty event, and the shortage or absence of experiential data, can computer modeling and simulation (as a methodology) be used to pr edict the medical resource requirements generated by this type of event and thus be tter prepare a health system in a defined geographic are for the possibility of an event of this nature? The focus of this research problem is the methodology to predict the medi cal management requirements associated with a large scale mass casual ty event. This methodology in cludes the identification of the characteristics of the blast event; envi ronment, location, number of people at risk, nature of the blast and magnitude. It incl udes the immediate impact of the blast, the number of casualties produced (dead, injured, and types of injuries) and the impact to the structure in terms of damage and subse quent accessibility to the injured by the rescuers/first responders. Finally, this re search identifies the number and types of

PAGE 16

7 m edical resources which will be required to manage the casualties. This includes identifying the composition of emergency re sponse assets, the predicted needs for definitive/acute care resources, the transportation resources that will be required to transport the injured for care, and the additi onal support in all categories of medical care that will be required as the loca l resources become overwhelmed. Given the constraints to tes ting the ability of a health system to respond to a large scale blast event, the development of a co mputer model and subsequent simulation can bridge the gap between planni ng and testing. Computer mo deling and simulation provide the capacity to test theories and ideas quickly and economically, to visualize and understand complex situations, to prioritize labor and investment opportunities, and to reduce the risk inherent in business/system decisions (Virginia Modeling, Analysis and Simulation Center (VMASC), 2004). A mode l developed for a unique geographic region (or jurisdiction) could be easily modified to test the resources and capabilities of a different region. The use of computer mode ling and simulation could prove to be an effective tool to assess system capacity and to aid in policy development and planning to ensure an effective response to a mass casualty event. This research offers a first step toward using computer modeling and simula tion as a prediction methodology to predict the medical management requirements of large scale mass casualty event.

PAGE 17

8 Chapter 2 Review of the Literature The literature to support this research is varied and plentiful in some areas, but very sparse in others. From a methodological pers pective this review includes an assessment of the current state of activity and methods within the field of computer simulation to include research methods and a review of the simulation software used for this research. An historical overview of terrorist actions to include ta rgets, means of attack, and resulting casualties sets the stage for the li kelihood of an attack and also the potential consequences of an attack. The structural co mponents of the health care system and their disaster response capacity are assessed to set the foundation fo r a large scale mass casualty response. Public policy in relation to disaster (large scale mass casualty events) preparedness activities is a ssessed as an integral component to overall levels of preparedness and ability to manage a large scale mass casualty event resulting from the use of conventional explosives. Finally, the literatu re utilized to s upport the development of the computer simulation model used for this research is discussed.

PAGE 18

9 Systems Theory, Computer Modeling and Simulation Systems, models, and simulation have mu ch relevance for health care delivery. Systems thinking helps to understand and desc ribe many of the entitie s that are seen and dealt with on a daily basis. Models aid in the understanding and visualization of processes, structures, organizations and system s. Often these models represent an ideal state for the organization or system in que stion. Simulation then explains how the components of a model interact with each ot her to produce an output. Simulation is a tool that can improve the act ual interaction with the environment, and can make the achievement of organizational tasks mo re efficient and more effective. Systems and their accompanying models and simulation methodologies come in a variety of sizes. For our purposes, a system can be as small as a local emergency management system (EMS) or community hospi tal or as large as a regional trauma system or network of hospitals covering a la rge geographic area. Some researchers refer to global and large scale systems as those having multi-level organizations and multi-goal criteria for success (Balakrishnan, 1979). These global/large scale systems include electrical power supply and national and international ec onomic systems (Balakrishnan, 1979). Given this context of systems, models, and simulation, a regional health system is neither too large, nor too complex to model a nd simulate. To clarif y, in the context of this research the health system does not refe r to a set of component s affiliated through ownership or other formal means. Health syst em refers to the health care resources in a given geographic area which must work coopera tively in response to a large scale mass casualty event.

PAGE 19

10 System s have three basic components: input, output, and sy stem description (Severance, 2001). The system represents the multiple components that make up a process. The system description describe s the components, or resources, which will process the inputs and produce outputs. For researchers, if an y two of the basic components are specified, the remaining component follows. The point of view of the researcher will often dictate the interaction between the researcher and the basic components of the system. From a scientific pe rspective the system is part of nature. An analysis is done to determine what the syst em actually is. By analyzing the inputs, and outputs it is hoped that the system descripti on will reveal itself. For an engineer, the basic inputs and outputs are known. It is the engineers task to design a system that produces the desired output when a given input is presented. A manager takes a third view, usually the system is already in place a nd the desired output is known. It is the responsibility of the manager to manipulate (control) the inputs to produce the desired output. If the desired output cannot be achieve d, the system must be adjusted. Finally, system optimization assumes that the mathematical form of the system is known, but strives to find the parameter values to optimize the objective function. Fishwick (2003) states that models are fundamentally interfac es between humans and phenomena. Models are built to mimic something else and thus modeling is an important and integral component of the simulation process. The McGraw-Hill Concise Encyclopedia of Engineering (2002) defines si mulation as a broad collection of methods used to study and analyze the behavior and pe rformance of actual or theoretical systems. Simulation studies are performed, not on th e real-world system, but on a (usually computer-based) model of the system created for the purpose of studying certain system

PAGE 20

11 dynam ics and characteristics. The purpose of any model is to enable its users to draw conclusions about the real system by st udying and analyzing the model. The major reasons for developing a model, as opposed to analyzing the real system, include economics, unavailability of a real syst em, and the goal of achieving a deeper understanding of the relationships between the elements of the system. Arsham (2003) describes system simulation as the mimicki ng of the operation of a real system. The simulation model is a what-if tool that allows a manager (or planner) to experiment with alternatives to see the impact those decisions have on the remainder of the system. Rubenstein and Malamed (1998) provide an introduction to the modeling and simulation of discrete-event systems. Th e origins of simulation lie in experiments involving chance; simulation intr oduces random elements into the traditional model. In comparison to an analytic model which uses mathematical tools to compute quantities of interest, simulation generate s possible histories (sample data) and then calculates statistics from the data that have been produced. Traditional analytic models use mathematical tools while disc rete-event simulation uses statistical tools. Analytic methods usually yield exact solutions but can only handle simple models. Simulation can be applied to more complex models but w ill yield statistical estimates subject to experimental error. Real-life discrete-eve nt systems are often too complex to model analytically, which is where the statistical computer simulation model is used to approximate numerically the desired characteristics. Discrete-event simulation views both time and state as discrete, rather than continuous. Within discrete-event simulation, a particular state is considered to be constant over a certain time interval. A transi tion takes place at a disc rete point in time as

PAGE 21

12 som ething happens and the system progresses fr om one discrete stat e to another. The notion of state transitions and holding time are fundamental to discrete-event simulation. Theoretically, reality is simplified in that so mething new only happens at discrete points in time (transition) and nothing new takes place between the transitions. These transitions are triggered by events. Randomness is usually incorporated into the simulation and takes the form of either/bot h the transition from state-to-state being random and/or the holding time (time between transitions) being random. A significant component of discrete-event simulation is anal ysis of the output of the simulation runs. Output analysis consists of the collection of various statistics or data elements from each simulation run. This collection of statistics becomes the sample data set. These data are used to formulate statistical estimates and c onfidence intervals for elements of interest and the performance of statisti cal tests in order to support decisions based on statistical inference. Rubenstein and Malamed (1998) define a system as a set of related entities (components or elements). These entities in teract in time and cause changes to the systems state. When a change occurs (s omething happens), the system jumps to a new state. This would be the case in a health system when a disa ster occurs, the usual state of emergency response and care will be disrupted by a large scale even t that changes the state of the system. Since this change in st atus does not occur continuously, but rather at a discrete point in time, it is called a discrete-event system. Developing a model is the first step in studyi ng a system. In this case, the model is defined as an abstraction of some real system that can be used to obtain predictions and formulate control strategies. Specifically, m odels are used to analyze changes in the

PAGE 22

13 system that may affect other aspects of th at system. Realism and simplicity are two conflicting elements that are necessary for an effective model. The model needs to reasonably approximate the system being mode led but it can not be so complex as to preclude understanding and manipulation. It is not necessary for the model to capture all the system components but rather to pr ovide a high correlation between model predictions and real life performance (model validity). Once the model is developed for the problem under consideration, the next step is to derive a solution from the model. Both analytical and numeric solutions may be used. An analytical solution is usually obtained directly from its mathematical form ulas. A numeric solution is generally an approximation using a suitable approximation procedure. Computer simulation is a means of obtaining the numeric solution/approximation. Simulation has been used in a variety of applications in the health sector although not specifically to address or predict the medical requirements of a large scale mass casualty event. Historically the focus in hea lthcare has been on the as pects of facility and clinic design which closely align with modeling and simulation applications in the manufacturing and service industries; attempts to fine-tune the manu facturing or service processes while reducing costs and wasted ti me, space, or materials. Eliminating or minimizing patient waiting times and maximizi ng resource utilization are the areas that have received modeling and simulation a ttention although some in the operations research arena see the health care field as ripe for further applicati on of these tools. Recent publications have exposed an in creased application of modeling and simulation to the emergency and disaster res ponse areas of health care. In separate articles Levi and Bregman discuss the use of simulation as a drilling technique in Israeli

PAGE 23

14 hospitals to prepare staff for m ass casualty situations (Levi, 1998 and Bregman, date unknown). Weil and his partners used simulation as a method of preparing the public health response to a bioterrorism respons e. They simulated the triage and drug dispensing activities associated with a bioterrorism respons e (Weil, 2002). Researchers have used modeling and simulation as a means of addressing the multiple threats, tools, and components associated with an emergency response. Brady used simulation to help local responders more quickly develop, test, a nd refine plans prepared in response to an expanding list of threats (Brady, 2003). Jain and McLean proposed a modeling framework to integrate the various compone nts/or tools developed for the multiple independent aspects of an emergenc y response (Jain & McLean, 2003). Hershberg focused on surgical resource utiliz ation during and after an urban terrorist bombing, presenting the advantages and lim itations of using simulation (Hershberg, 1999). He modeled the in-hospital respons e to individual multiple (mass) casualty incidents as experienced regularly by Israeli hospitals. The model was used to predict capacity, identify bottlenecks in service areas, and to identify requirements for additional staff. The purpose of the Hershberg st udy and the methodology employed most closely represents the intended purpose and method of this study. Two other pertinent references to s imulation are ongoing research projects. Sokolowski explains that the policies and pr ocedures for managing a mass casualty event are often not tested for adequacy (Sokol owski, 2004). He proposes a computer simulation capable of providing training, anal ysis and decision support to health care decision makers faced with mass casualty care decisions. This research is ongoing at the Virginia Modeling, Analysis and Simulation Center of Old Dominion University.

PAGE 24

15 Hallber t used the label high consequence events for those events that result in challenges to the safety, health, or security of facilities, orga nizations and society (Hallbert, 2004). The objective of Hallberts ongoing research is to develop models of information management for high consequence events. This research is being conducted within the Civil and Environm ental Engineering Department at Vanderbilt University, with funding from the Idaho National Engineering and Environmental Laboratory. Neither study has yet produced anything more definitive than the initial intent.

PAGE 25

16 Research Methods Kelton, Sadowski and Sturrock (2004) describe sim ulation from a practical viewpoint; it is the process of designing and creating a computerized model of a real or proposed system. The purpose of the mode l is to conduct numerical experiments to provide a better understanding of the behavior of that system for a given set of conditions. The basis of the simulation is a logical (or mathematical) model comprised of a set of approximations and assumptions, both structural and quantitative, about the way the system does, or will work (Kelton, 2004). Structurally the mode l depicts the system or process being studied. The model is usua lly represented in the form of a computer program that is run to address questions about the models behavior and then the systems behavior if the model is valid (Kelton, 2004). This exercising of the model or computer program is where the structure of the model actively simu lates the operation of the real system. The rationale behind the use of a com puter simulation model is based on the benefits it affords. The computer simulati on model represents what is usually an easy, cheap, and fast way to get answers to questions about the model and system it represents. This is done by manipulating bo th the inputs to the simulatio n and the structure of the model (Kelton, 2004). The model becomes a su rrogate that can be manipulated more cheaply, safely, and conveniently than the system being simulated (Pidd, 2004). The easy, cheap, and fast labels applied by Kelton and his associates references the multiple simulation runs accomplished once the simulation model is constructed. The literature is full of references to the bene fits of computer simulation. But the caveat given is that creating a valid model is far from quick and easy. Once the model is created, the ability

PAGE 26

17 to easily m anipulate the inputs and the structur e of the model makes it very easy to ask the hypothetical what-if questi ons. From a financial and se rvice provider perspective, the cheap label is a reference to the comput er simulations ability to assess the system and then assess the impact of changes to th e system without impacti ng daily operations. With the delivery of health care being at th e center of this research, manipulating the model and testing the results can be done wit hout diminishing the day-to-day ability of a health system to care for patients. Fina lly, once constructed, simulation models lend themselves to modification to represent an entirely different system, thus making it easy for other systems to benefit from the work already done. Sensitivity analysis is the proce ss of analyzing and understanding how the outputs of a model change in relation to changes in the inputs to the model or to changes in the structure of the model. An important criterion for establishi ng the usefulness of the model and methodology used is the ability to manipulate both the inputs and structure of the model and then the subsequent analysis of the outputs. This process of sensitivity analysis is critical for medical planners seek ing to prepare their jurisdictions in the best manner possible for a large scale mass casualty event. Procedurally, each model identifies when and how an entity is created. For example, the entities in this research are de fined as injured survivor s of a blast event. Models then identify all the processes and decision points each individual entity (injured survivors in this research) must go through to exit the system. Ex amples of processes used for this research include extraction from th e blast site, initial triage, care in the field, transportation to a medical faci lity, and care provided within the medical facility. Each of these processes is required for the inju red survivor to proceed through the response

PAGE 27

18 system the health care system, and to ultimately exit the system. Additionally, decisions are made at different points to direct the inju red survivor(s) to the a ppropriate care path. As examples, decisions are made concerning the severity of th e injury, and thus the level of care required. Later in the care process a decision is made to admit the injured survivor to the hospital or pr ovide only outpatient care. E ach of these decision points are modeled with the use of a decision module containing the assigned decision parameters. Models identify the resources required/utilized within ea ch process to move the injured survivor through the sy stem. Typically th e resources are limited in terms of quantities of individual resources and will in clude personnel, equipment, transportation assets, and other items required to car ry out the different processes. Models identify the decisi on points and decision methodologies required for the injured survivors to progress through the process or system The concluding structural component of the model is the point at whic h the injured survivor(s) exit the system. Depending on the intent and boundaries of the model, the exit point could occur at a variety of different locations: for this researc h, departure from the blas t site, arrival at the hospital and the completion of emergency depa rtment care are examples. For a disaster response scenario, this would be the point wh en the injured survivor completes definitive treatment. The survivor is then released and no longer considered within the acute care part of the health care system. Quantitatively a model identifies the paramete rs associated with each structural component of the model. These parameters ar e either in the form of specified quantities (deterministic variables) or random quantities (stochastic variables). The parameters are quantities that represent the characteristics of the distribution in terms of arithmetic mean

PAGE 28

19 and standard deviation (Kelton, 2004). As an exam ple, the simulation model for this research begins at the point where the injure d survivors are generated by the model. One method of creating or inputting the injured su rvivors into the model is to specify time zero as the point where all injured survivor s are created. Once created, the injured survivors will begin to be processed through the model. A second method is to allow some variability in the creation of the injure d survivors. The parameter associated with the creation of the injured survivors consists of a probability distribution that the model uses to randomly generate the injured survivors. This random generation would be similar to that which confronts the workers at an emergency room. Rather than arriving all at once, the casualties might arrive singly or a few at a time, and with different time intervals between arrivals. This random patte rn continues until there are no more injured survivors (the total number of injured surv ivors being a parameter determined by the researcher). Similarly, processes included in the model have either a defined time required to process each injured survivor or a probability distribution of the likely processing times. Based on the probability distribution specified for the process, processing times are randomly assigned as each person (injured survivor) is processed. Finally, resources are typically deterministic in nature in that there are a set number of personnel, ambulances, CT scanners, and hospita l beds available for use. The lack of resource availability will generate waiting lin es or queues and incr ease total processing times. Including uncertainty (random or stoc hastic components) in the model through the use of specific probability distributions make s the model more realistic and allows the researcher to simulate events outside th e bounds of the observed data, and to explore situations where no data are av ailable (Kelton, 2004).

PAGE 29

20 The Arena software program (utilized for this research) is one of several computer modeling and simulation programs designed to support systems researchers. Arena is a computer modeling and simulation program based on the SIMAN simulation language (Kelton, 2004). Arena has proven to be fairly easy to use and offers the flexibility of a hierarchical structure, allowi ng users to build models using established Arena templates on one end of the hierarchy and user written code on the other end. Arena also provides the functionality to animate the computer simulation model. Arena specific computer modeling structures/methods : The structures of the computer simulation models for this research ar e fairly basic. A very general flow chart example depicting a model is presented in Fi gure 1. The simulation process starts with a creation module (discussed in prev ious section) where an entity is created, in this case an injured survivor. Each inju red survivor then proceeds th rough a series of decision modules. The decision modules are represented by the diamond shapes in the flow chart. Decision modules are characterized by one entr y point and at least two exit points (Arena allows up to eight exit points associated with a true path plus one associated with the false path). Within each decision module, the injured survivor is either placed on the true path (or one of the true paths) or the false path. The placement on one or the other path is based on the decision parameters built into th e decision module. In Figure 1 the decision determines if the injured survivor has a head injury. The statement might be, the injured survivor has a head injury. If this is true the injured survivor exits the decision module on the true path. If false, the injured surviv or exits on the false path. This decision is made by chance (probability) of a true outcome The chance (probability) is assigned by the researcher and is typically supported by data analysis but may be selected purely for

PAGE 30

Figure 1 Simulation Model Example Creation of Injured Survivo r Head Injury Assign head injury attribute Record head injury False Finish Each simulation in Arena begins with a creation module. A decision module uses chance or a condition to make decisions that determine the subsequent path of the entit y An assignment module assigns attributes to each entity. The attribute is common to all. The presence or not of the attribute is entity unique. The record module records or counts the number of entities with the particular attribute. Once the attribute processes through the entire model it finishes in the dispose module. True Start testing a des ired effect within the model. In the case of more than one true path being available, the cumulative tota ls of the chance assigned to each exit point equal 100% (including the false path). Following the decisi on, the true path will lead to the head injury attribute being assigned to the injure d survivor. In Arena, an attribute, once 21

PAGE 31

22 estab lished becomes a common characteristic possessed by all inju red survivors; the value assigned to the attribute is however, unique to each individual survivor (or entity) (Kelton, 2004). Each survivor either has the attribute (positive valu e assignment) or it does not (negative value assignment). The determination of whether the injured survivor takes the attribute (or injury) is based on the decision made in the decision module. As an example, the decision point presented in Figure 1 determines which injured survivors will be assigned the head injury attribute. The model may be built so the chance the survivor has a head injury is 13%. Each injured survivor who move s completely through the decision process will have a 13% chance of emerging with the head injury attribute. To clarify, the 13% chance of head injury is determined for each individual injured survivor. The entire population of injure d survivors is not a ssigned/found with 13% having a head injury. There is the potential for variability between re plications as to the overall percentage of the population with the head injury attribute. For data collection and analysis pusposes, following the assignment of an attribute, the injured survivor is counted as having the attribute in the record module, and then the process comes to an end in the dispose module. A second method for the decision to be made in a decision module is for the decision to be made based on an attribute the injured survivor car ries with them (one that has already been assigned). If the decision module is determining the need for surgery, rather than assigning a numerical probability of surgery, the decision can be made on the previous assignment of an attribut e. If the injured survivor has previously been assigned the severe injury attribute, the decision on the need for surgery can be based on the presence of that attribute. Ther efore, if the injured survivor has a severe

PAGE 32

23 injury, the d ecision module will direct the inju red survivor down the path to surgery. If the injured survivor does not have a severe inju ry, they will be directed out the false exit down the path and will not enter the surgery process module. The process module is the second Aren a module utilized for this study. The process modules are represented by rectangles in the flow charts. Each process module has a single entry point and a single exit point. The process modules represent the variety of processes requiring resources that an injured survivor wi ll procede through to reach the point of definitive care and ultimately exit the system/model. Each process module requires at least one resource to complete its process but ca n be constructed to require multiple resources for a single process. For th e purposes of this research, each process is constructed to determine the minutes of se rvice required from that resource/process by the injured survivor. Additionally, each proc ess is performed by a single resource entity or unit. Once the injured survivor (entity) en ters the process module, the system searches for a resource to match to the injured survi vor. If a resource is available, it will be matched to the injured survivor or seized. If no resource is available, the injured survivor is placed in a waiting queue until a resource is available. Once the injured survivor seizes a resource it holds the resource (known as delay in Arena) for a designated period of time. This designated period of time repres ents the time required for the resources to complete the process. Once the process is co mplete for that particular injured survivor the resource is released and becomes availabl e to serve another injured survivor. The availability of resources is a limitation that can be built into the model. If the time between arrival (interarrival ti me) of injured survivors is sh orter then the resource process time, then a queue or waiting lin e is created. If the interarrival time of the injured

PAGE 33

24 survivors is longer than the process ing time, the queue will get shorter or be nonexistent. If there is an infinite number of a resource th ere will never be a queue. As an example, if the injured survivor needs an x -ray, the injured survivor will be sent to the x-ray process. Once the injured survivor enters the x-ray process module, an x-ray resource is searched for and, if available, seized. If the x-ray process takes 19 minutes the resource will be held (or delayed) for 19 minutes and then rel eased. The injured survivor then moves on to the next process and the x-ray resource beco mes available for the next injured survivor who requires an x-ray. The time the resource is held can be assi gned in two ways. First, a set time can be assigned for each process. If all x-rays re quire 19 minutes and there is no variance then 19 minutes is assigned as the processing time. All injured survivors requiring an x-ray will then delay or utilize the x-ray resource for 19 minutes, there will be no variation to this standard. The alterna tive to assigning a constant tim e is to assign a probability distribution with a range of possible processing times. This method is appropriate where there is variance in the processing times. As each injured survivor seizes the resource, the simulation methodology selects a processi ng time from the range of possibilities based on the assigned distribution parameters. The processing time utilized by each individual survivor is then that which ha s been selected by the simulation methodology. Upon completion of processing, the resource is released and becomes available for another injured survivor. Once the injured survivor progresses th rough the entire model, they exit the model or system via a dispose module. Th e dispose module represents the point where all care or service to the injured survivor is complete or terminated. Based on the

PAGE 34

25 decisions made at each decision point for that particular injured survivor, either no more care op tions are available or no more care is needed. Probability Distributions As has been previously explained, th e methodology used within the simulation model to create the injured survivors and to assign times to each process as the injured survivors proceed through the system, is based on a probability distribution. Whether these distributions are based on an analysis of available data, or on an estimate, the probability distributions provide the basis for the simulation run. To assist this process, the Arena software contains a set of built -in functions for generating random variates based on the probability distributions (Kelton, 20 04). As an example, the travel times for an ambulance may be assigned a triangular distribution with a minimum value of 10 minutes, a maximum value of 60 minutes, and a mode or most likely value of 30 minutes. During the simulation, each time an ambulance tr ansports an injured survivor, Arena will randomly choose a travel time between 10 a nd 60 minuteswith the most likely being 30 minutes. The distributions avai lable in Arena and their associated parameter values are listed in Table 1. The distributions used in this resear ch include Exponential, Triangular, and Uniform. A description of each, as pres ented by Kelton, Sadowski, and Sturrock, are included below (Kelton, 2004). A discussion of the other distributions available in Arena is found in Appendix 6. Arena also contains an Input Analyzer th at will assist in analyzing a set of data and matching a distribution for us e in the simulation model.

PAGE 35

Table 1. Probability Distribut ions Available in Arena Distribu tion Parameter Values Beta Beta, Alpha Continuous CumP1, Val1, CumPn, Valn Discrete CumP1, Val1, CumPn, Valn Erlang ExpoMean, k Exponential Mean Gamma Beta, Alpha Johnson Gamma, Delta, lambda, Xi Lognormal LogMean, LogStd Normal Mean, StdDev Poisson Mean Triangular Min, Mode, Max Uniform Min, Max Weibull Beta, Alpha Exponential Probability Distribution The exponential probability distribution is used to model inter-event times in random arrival and breakdown processes. The probability density function, f (x) = 1 ex/ for x > 0 and 0 if otherwise. Parameters: The mean ( ) specified as a positive real number. Range: [0, + ) 26

PAGE 36

27 Mean: Variance: 2 Triangular Probability Distribution The triangular probability distribution is commonly used in situations in which the exact form of the distribution is no t known, but estimates (or guesses) for the minimum, maximum, and most likely values ar e available. The tria ngular distribution is easier to use and explain than other distributions that may be used in this situation (e.g., the beta distribution). The probability density function, f (x) = __ 2( x a) __ for a x m and __2( bx )__ for m x b and 0 if otherwise. ( m a)( ba) ( bm )(b a) Parameters: The minimum ( a), mode ( m ), and maximum ( b) values for the distribution specified as real numbers with a < m < b. Range: [ a, b] Mean: (a + m + b)/3 Variance: ( a2 + m2 + b2 ma ab mb )/18 Uniform Probability Distribution The uniform distribution is used when all values over a finite range are considered to be equally likely. It is sometimes used when no information other than the range is available. The uniform distribution has a larg er variance than other distributions that are

PAGE 37

28 used when infor mation is lacking (e.g., the triangular distribution). The probability density function, f (x) = 1_ for a x b and 0 if otherwise. ba Parameters: The minimum ( a) and maximum ( b) values for the distribution specified as real numbers with a < b. Range: [ a, b] Mean: (a + b)/2 Variance: ( b a)2/12 Verification and Validation Verification and validation are two steps requ ired to ensure that the model accurately depicts the system under study and that the right problem is being addressed in an accurate manner. Verification is the task of ensuring that the model behaves as intended (Kelton, 2004). Validation is the task of en suring that the model behaves in the same manner as the real system (Kelton, 2004). Verification of the model o ccurs throughout the process of building the model. In its most basic form, verification is the debuggi ng of the code. In Arena this is done by responding to error messages when the model is run. Animation is a second method which is used for verification. Animation provides the means to visua lize the flow of an entity through the system. This allows the re searcher to ensure that the model appears to operate as designed. Finally, verificati on is accomplished by running the model and verifying the accuracy of the output. For this final verification, a set of conditions will be defined and an estimate of the results will be calculated. The conditions will then be

PAGE 38

29 applied to the m odel and the simulation run co mpleted. The results of the simulation run will then be compared to the estimates. If the results are the same, the model will be considered verified. Additional verification steps can include stressing the model with parameters and inputs more extreme than the sample data would support and then assessing the output. The results in each case should be pred ictable. If the simulation produces results outside the predicted range, additional steps will be required to identify the discrepancy and fix the problem. Validation of the model is a more diffic ult proposition. Valida tion is the process of ensuring that the computer model accurately mimics the real world situation. This is typically done by having those working with, or within, the system review the model and assess how well it matches the real system. Secondly, output from the simulation runs are compared with output from the real system to ensure that the model is behaving as the real system behaves. Since there have been no large scale blast events of the type being modeled here, and thus no systems that have responded to an event, validation is doubly problematic. If the model is validated and ba sed on sound logic and examples, it must be assumed that the simulation model presents an adequate representation of a system responding to a large scale blast event.

PAGE 39

30 Blast Events and Terrorist Actions Blast events and the damage caused by these events have been experienced by mankind since the discovery/creation of gunpowder. Some references place this discovery as early as 850 A.D. in China by al chemists seeking to create an elixir of immortality (Silkroad Foundation, 1997). An early account stated, some have heated together the saltpeter, sulfur and carbon charcoal with honey; smoke and flames result, so that their hands and faces have been burnt, and even the whole house burnt down (Silkroad Foundation, 1997). Datin g from this time there is evidence of blasts and the subsequent injuries and destruction resulting from accidents, acts associated with warfare, and terrorist actions. The introduction to this paper outlined mu ltiple events, but not all were terrorist related nor purely blast associated events; th e accidental explosion of the munitions ship Mont Blanc in Halifax harbor in 1917 and the use of airliners in bringing down the World Trade Center. Although not of terrori st origin, the explosion of the Mont Blanc provides an example of the destructive power of a bl ast. The destruction and devastation caused by a blast is a stark reminder of the potential danger that exists throughout the world. A BBC correspondent in Taba, Egypt, the site of the bombing of a Hilton Hotel, labeled the devastation as astonishing (BBC News, 2004) after a blast estimated to be caused by 441 pounds of explosives ripped through the hot el bringing down 11 stories on one side of the building (Spiller, 2004) and causing 34 deaths and 105 injuries (BBC News, 2004). Two of the largest incidents of chemical bl ast-caused devastation are associated with cargo ships while in port. Th e sheer tonnage of the cargo in volved make most terrorist bombings pale in comparison, but brings into question the safety, and preparedness, of

PAGE 40

31 port cities. As previously m entioned, the explosion of the munitions carrier Mont Blanc in Halifax harbor left 1,600 dead, 9,000 injured, 1,600 hundred buildings destroyed, 12,000 buildings damaged, 6,000 people homeless and 20,000 without adequate shelter, all in a town of 50,000 (Kernaghan, 2004). Th e ship that exploded carried benzol on deck and 2,766 tons of picric acid, TNT a nd guncotton in the hold (Kernaghan, 2004). A similar explosion at the docks in Texas C ity, Texas in 1947 left 576 persons known dead and injured thousands (Texas City Disaster, 2002). The SS Grandcamp carried 2,300 tons of ammonium nitrate fertilizer (The Te xas City Disaster, 2004). Assistance to Texas City from the Red Cross and other vol unteers amounted to 4,000 workers operating temporary hospitals, morgues, and shelters (T exas City Disaster, 2004). Of note with each of these blasts is the lo ss of life of the initial responde rs (mostly fire fighters) and curious onlookers. Frykberg labeled this the second-hit principle (Frykberg, 2002). Each event started with a fire which was responded to by fire fighters and curious onlookers. While the fire fighters fought the fires, the ships exploded, thus decimating the fire fighters and on-lookers. This demons trates the importance of protecting medical and other rescue resources in the initial response to a disast er/attack site (Frykberg, 2002). A definitive list of terrorist attacks has not proven to be readily available while multiple sources provide a piece of the ove rall picture. Frykberg cited multiple references in stating that there were ove r 12,000 bombings in the United States alone between 1980 and 1990 (Frykberg, 2002). He felt the trend continued into the 1990s with 1,582 bombings resulting in 222 injuri es and 27 deaths in 1990 alone (Frykberg, 2002). Slater and Trunkey point out that mo st of the 12,000 bombings were pipe bombs and that 53% of those in 1990 were pipe bombs, there was however an increasing trend in

PAGE 41

32 the num ber of bombings per year through the ten year period (Slater, 1997). Clearly volume does not necessarily lead to large scal e mass casualty events. It was not until the bombing of the federal building in Oklahom a City in 1995 and the attack on the World Trade Center in 2001 that Americans felt sus ceptible to events th at may produce a large number of casualties. A search of the International Terrorism data base maintained by the International Policy Institute for Counter-Terrorism f ound 33 bombings, 41 suicide bombings, and 23 car bomb attacks which caused more than 50 killed and injured between 1986 and 2005. The table found in Appendix 2 lists the re sults of the search for blast related injuries/attacks. A further search of the In ternational Terrorism data base for events resulting in 1,000 or more killed and injured identified 6 attacks. These six include: World Trade Center, New York, New York (11 Sep 01) U.S. Embassy in Nairobi, Kenya (7 Aug 98) Bank attack in Colombo, Sri Lanka (31 Jan 96) Tokyo subway attack (20 Mar 95) World Trade Center attack (26 Feb 93) Ammunition dump explosion in Pakistan (4 Oct 88) In order to look at the specific results of some of these blasts it is necessary to define additional terms commonly used when analyzing the impact of a blast. In 1997 Slater and Trunkey differentiated between conventional and nonconve ntional weapons. They define conventional weapons as prim arily kinetic energy delivery systems like firearms and explosive or thermal devices (Slater, 1997). Nonconventional weapons systems include nuclear, biological, and chemi cal elements (Slater, 1997). Slater and

PAGE 42

33 Trunkey ref erence the Textbook of Military Me dicine to define le thality and casualty generation. Lethality refers to the fraction of the total number injured who die (Slater, 1997). If a total of 100 are injured and 78 of those 100 die, the lethal ity of the weapon or blast is 78%. Casualty generation refers to the number of indivi duals in the target population who are injured by a single use of th e weapon or a single bl ast (Slater, 1997). The calculation of TNT (trinitr otoluene) equivalents is used as a method to compare the explosive energy from different materi als to a standard (Thompson, 2004). Injuries attributable directly to the ef fects of a blast are ca tegorized as primary, secondary, tertiary, and miscellaneous. Pr imary injuries are those caused by the blast wave as it travels through air or water with injuries be ing categorized as pulmonary contusions, hollow viscous perforations, and perforated eardrums (Slater, 1997). Secondary injuries are those caused by prim ary and secondary missiles as they are propelled outward by the explosion and are ca tegorized as penetrating missile injuries and orthopedic injuries (Slater, 1997). Tertiary injuries are caused when the casualty is propelled (displaced) through the air and then impacts onto a relatively fixed object, these include closed head injuries, cervical spine in juries, and orthopedic inju ries (Slater, 1997). Miscellaneous injuries are burns, inhalation injuries, and other injuries related to structural collapse that are categorized as burns, inhalation injuries, crush syndrome, and compartment syndrome (Slater, 1997). Slater and Trunkey cite a 1988 article by Frykberg and Tepas which provides an overview of injuries sustained from 220 sepa rate blasts (explosions). Injuries were sustained by a group of 3,357 (an average of 15 per blast) with 2,934 immediate survivors (Slater, 1997). Of the immediate survivors 881 (3 0%) were hosp italized and 40

PAGE 43

34 (1.4%) subsequently died (Slater, 1997). Th e common injuries were soft tissue (55.4%), head (31.4%), and bony extrem ity (10.9%) with most surgical operations being required for soft tissue (67%) and skeletal (17.5%) inju ries (Slater, 1997). Mortality rates were highest for those patients with abdominal (19% ) and chest (15.1%) injuries (Slater, 1997). As is referenced in other literature, Slat er and Trunkey point out that the challenge associated with the high percentage of relati vely minor injuries is to avoid overwhelming the medical resources with these minor injuries and to rapidly identify those with major injuries. Researchers have analyzed specific even ts to, in part, identify the mix of casualties produced by the events. Frykberg did this with a group of bombings that caused a relatively large number of casualties dating back to 1969. More recently the Oklahoma City, Khobar Towers, and World Trade Center events have been analyzed. Teague assessed the mass casualty situation that confronted the health system in Oklahoma City following the bombing of the federal building. On April 19, 1995 a truck holding 4,000 pounds of ammonium nitrate soaked in fuel o il was detonated outside the federal building (Teague, 2004). The blast le ft 168 dead, 591 injured and the federal building severely damaged (Teague, 2004). The st atistics tell the story of the blast. Of the 361 in the federal building, 319 (88%) were injured. One hundred sixty eight of the 319 (47%) died. The other injuries were either to those in the four adjacent buildings or those who were passing by. Five major hospita ls are within 1.5 miles of the blast which resulted in very little field care and an ear ly patient surge which overwhelmed the closest hospitals. Thirteen area hospita ls received patients. In se ven, the treatment provided was solely confined to the emergency department thus indicating the minor nature of the

PAGE 44

35 injuries. The first patient arrival occurred w ithin 15 m inutes of the blast with the peak arrival time being between 60 and 120 minutes after the blast. The means of transportation was known for 272 patients, 90 (33%) arrived by am bulance and 152 (57%) by private vehicle. The others walked or we re carried. At the lo cal hospitals 265 (68%) of 388 patients were treated in the emergenc y department and 114 (29%) were triaged to minor treatment areas. For those who died th e primary cause of death was determined to be multiple injuries (122), head trauma (24), chest trauma (13), and various causes (9). The death rate was higher for those in the collapsed portion of the building, 153 (87%) of 175 who were injured died. Only 10 (5%) of 186 injured died in the portion of the building that remained standing. Thirty-eight victims needed to be rescued from the building and most rescues occurred within 45 mi nutes of the blast, with only three being rescued more than 3 hours after the blast. The low volume of seve rely injured victims spared the health care system from a true test of its mass casualty (trauma) preparedness. One can only speculate about the outcome if 168 patients requir ing trauma care and emergency resuscitation had been extracted fr om the rubble and transported to the local hospitals. Similar to Oklahoma City, the blast which damaged a portion of the Khobar Towers military compound in Dhahran, Saudi Arabia was caused by a truck bomb parked outside the complex. Two events served to mitigate the effects of the blast, first a security fence had been constructed which di d not allow the truck closer than 80 feet from the building. Secondly, security pe rsonnel had noticed the truck and began evacuation procedures for those closest to the blast site. Even with these measures, the explosives, estimated at 20,000 pounds of TN T equivalents, produced considerable

PAGE 45

36 dam age and injury. The blast caused 19 d eaths and left 555 injured survivors (Thompson, 2004). Of the total number in jured, 288 (52%) were injured directly by the blast, 113 (20%) were injured both by the blast and during evacuation, 60 (11%) were injured during the evacuation only, and 24 (4%) were injured during the search and rescue or clean up (Thompson, 2004). Sixty-one (11%) re ported only auditory, and/or smoke and dust inhalation (Thompson, 2004). Of the 420 who were injured directly by the blast, 19 (5%) died at the scene, 66 (16%) were hospitalized, 171 (41%) were treated on an outpatient basis, and 164 (39%) self treate d (Thompson, 2004). A total of 2,065 injuries were documented for the 401 injured surv ivors with a range of 1-25 per person (Thompson, 2004). The average number of inju ries per hospitalized patient was nine, five per outpatient, and three for those who self treated (Thompson, 2004). Autopsy reports documented 381 injuries for those who died with a range of 10-50 and an average of 20 (Thompson, 2004). Most of the injuries were characterized as blunt, crushing, and penetrating injuries (Thompson, 2004).

PAGE 46

37 Capacity (S tructure) of the Health Ca re System and Mass Casualty Events Trauma systems in many areas of the US form the structure available to care for the casualties resulting from a large scale blas t event. Trauma Systems are designed to provide an organized and coordinated respons e to injury (MacKenzie, 2003). The true success of these systems is dependent upon th e transition between each phase of medical care and the level of integrati on between resources, with the goal being improved patient outcomes. In further defining trauma syst ems, the National Highway Traffic Safety Administration (NHTSA) says trauma systems are regionalized, based on the unique requirements of the population (rural, inner-city, urban, etc. ), and must emphasize the prevention of injuries in the context of comm unity health. The vision of the NHTSA is a more proactive approach where the trauma system seeks to maximize the integrated delivery of care to the injured, identifies threat s to community health and also identifies the intervention required to alleviate the threat. The system should not only address the daily demands for trauma care but also provi de the foundation for a response to a mass casualty event. The NHTSA defines a trauma patient as an injured person who requires timely diagnosis and treatment of actual or po tential injuries by a multidisciplinary team of healthcare professionals, supported by appropriate resources, to diminish or eliminate the risk of death or permanent disability. The Centers for Disease Control (C DC), in its Injury Fact Book-2002 states that as many as 35% of trauma patients who die do so because optimal acute care is not available (Injury Fact Book, 2003). The CDC defines a trauma care system as an organized effort, coordinated by a state or local agency, to de liver the full spectrum of care to injured persons in a defined geogra phic area. This syst em requires specially

PAGE 47

38 tra ined practitioners, and adequate resources and equipment. Despite existing evidence supporting the development of trauma systems, only 25% of the U.S. population is served by a trauma system. Trauma Systems and Public Policy, a 1991 article by Mendeloff and Cayten outlines support for trauma systems, threats to the existence of trauma centers/systems and lists what is needed to further develop tr auma systems as viable health care resources in the United States. They support their push for development of trauma systems by asserting that some 20% of deaths among inju ry victims who were treated at community hospitals could have been prevented had t hose patients been taken to a trauma center (Mendelhoff, 1991). This assertion is ba sed on their review of published studies and motor vehicle accident data. Mendelhoff and Cayten define the concept of a trauma system as the integration of prehospital emergency medical services (EMS), hospital care, and posthospitalization rehabilitation pr ograms. Within this context, trauma centers are not appropriate ve nues of care for all injured pe ople. A Los Angeles County study (using a fairly restric tive measure of need for a tr auma center), published in 1985, found that only 12% of injury victims transported by ambulance should have been sent to a trauma center. Thus it is important for trau ma systems to incorporate multiple levels of care to ensure the most effective a nd efficient use of resources. In its Statement on disaster and mass casualty management, the American College of Surgeons characterizes mass causalities following a disaster as having such numbers, severity, and diversity of inju ries that they overwhelm the ability of the local medical resources to deliver comprehensive and definitive medical care to all victims (ACS, 2003). Directed at surgeons, the statement points out that disaster management poses

PAGE 48

39 challenges that are distinct from normal surg ical practice. Disaster management requires a paradigm change, from the application of unlimited resources for the greatest good for each individual, to the allo cation of limited resources for the greatest good for the greatest number of casualties. This statemen t reinforces the need for a smooth integration of multidisciplinary local, state, and federal assets. Bowen and Pretto sought to determine the extent to which state emergency health and medical plans were designed to manage la rge numbers of critically-injured casualties following a catastrophic event (Bowen, 1999). They attempted to survey the 50 State Emergency Medical Services Directors asking if the state had a catastrophic casualty management plan. If told yes, they requested a copy of the plan which was then assessed on five criteria: (1) Is the plan based on a hazard-risk analysis? (2) Is the plan based on vulnerability analysis studies? (3) Has the plan been integrated into the la rger context of the Emergency Operations plan? (4) Have mutual aid agreements been established? (5) Have contacts for material and personnel resources, specifically for a disaster response, been identified? Twenty-eight states participated in the st udy, 21 indicated that th ey had a catastrophic casualty management plan. Twelve states submitted their plans and none of the 12 met all five of the review criteria. The efficiency of medical regulators (t hose who determine where each patient is transported) in a mass casualty situation may play a significant role in the effectiveness of

PAGE 49

40 the health care response. Based on obser vations during a Red Cross m ass casualty disaster exercise, the authors approached the regulators task from a decision sciences and management approach (Muller, 2001). It appears that the authors built a simulation model to compare the methodology/efficiency of the model with that of a benchmark best practice produced by medi cal specialists. The authors found limitations in the model and those who attempted to use it based on the following factors: (1) first time confrontation by the partic ipants with such an assignment, (2) differences in the perceived policy criteria, (3) difficulties in agreeing on concrete measurement means of commonly accepted policy criteria, (4) the existence of a variety of theoretically well-known dispatch ing heuristics, and (5) the lack of a trainees capability to transfer and use knowledge from well-known fields of applications for unexpected or even apparently unrelated problems. The authors concluded that the regulators task in a mass casualty event must be clearly defined and enforced. Training in the use of appropriate dispatching heuristics seems essential, and an expert-like computer si mulation is a necessity and a pedagogically valuable tool. Walter Green presents several fundamental questions concerning the definitions of a mass casualty incident (Green, 2000). When answered, these questions will help determine how a response will be made. (1) Is a mass casualty incident a set number of patients, or is it more flexibly defined as more casualties than you have ambulances or hospital beds? (2) Are there gradations of mass casualty incidents with some being worse than others?

PAGE 50

41 (3) Does a m ass casualty incident start when people are injured or does it start well before that? The State of Virginia defines a mass casualt y incident as one which generates more patients than available resources can manage using routine res ources. Green states this definition has the advantage of being linked to two critical, interrelated components: system capacity and operational procedures. System capacity is not static and will vary based on a number of components: number of ambulances ava ilable, number of available and qualified personnel, pers onnel efficiency, number of appropriate and available hospital beds, and communications systems cap acity. Operational procedures determine the efficiency of treatment of patients in terms of speed, resource commitment, and outcome. A mass casualty incident requires the use of emergency procedures for successful management as normal procedures will not be adequate for the task. Given this definition, a system for response can be outlined to address both capacity and procedural issues through all phases of an incident. These phases may include: (1) Preparedness, (2) Mass casualty event, (3) Response and characterization, (4) Patient clearance, (5) Transition to mass fatality incident if required, (6) Short distance transport to definitive care, (7) Long distance transport to definitive care, and (8) Patient discharge and return.

PAGE 51

42 A successful approach to a mass casualty in cident must address all of these phases through direct inclusion of new programs or i ndirectly through trai ning and the inclusion of existing programs. Virginia has devel oped a building block approach, first through training (awareness, operations, supervisor, emergency operations center), then hospital system, and then a course of action if th e number of casualties exceeds the statewide system capacity.

PAGE 52

43 Public Polic y The Federal government has not directly addressed the issue of large scale mass casualty events caused by conventional explos ives, but has encouraged the development of emergency medical systems or trauma systems since 1973. Public Law 93-154, signed on 16 November 1973 and entitled the Emerge ncy Medical Services Systems Act, amended the Public Health Services Act to provide assistance and encouragement for the development of comprehensive area emerge ncy medical services systems (S.2410, 1973)). This law authorized the Secret ary of Health, Education and Welfare to make grants and to enter into contracts for projects which we re to include both (1) studying the feasibility of establishing and operating an emergency medi cal services system, and (2) planning the establishment and operation of such a system. Additionally the secretary was authorized to make grants for (1) the establishment and initial operation of emergency services systems, (2) projects for the expansion and improvement of emerge ncy medical services systems, and (3) support of research in em ergency medical techniques, methods, devices and delivery. This law also provided criteria as to what an emergency medical services system would include: (1) An adequate number of personnel (hea lth and allied health professions and others) with appropriate training and experience. (2) Provision of appropria te training and continuing e ducation for its personnel. (3) A central communication system to join the components of the system. (4) An adequate number of surface and air transportation resources to meet the characteristics (needs) of the service area. (5) Other enumerated criteria.

PAGE 53

44 A second bill was signed on 16 November 1990. This bill was titled the T rauma Care Systems Planning Act of 1990 (H.R. 1602, 1990) As originally introduced, this bill was to amend the Public Health Service Act to improve emergency medical services and trauma care (and for other purposes). This act authorized and requi red the Secretary of Health and Human Service to act on multiple initiatives in relation to creating and sustaining trauma systems. The Secretary was authorized to make gr ants and enter into contracts (with respect to trauma care) to: (1) conduct and suppor t research, training, evaluations, and demonstration projects, (2) fo ster development of trauma care systems, (3) provide technical assistan ce to state and local agencies and (4) sponsor workshops and conferences. The Secretary was also aut horized to make grants for research and demonstration projects to improve emergency medical services in rural areas. In addition, the Secretary was directed to establish an Advisory Council on Trauma Care Systems, to establish and operate a National Clearinghous e on Trauma Care and Emergency Medical Services and to allot funds to each state for ea ch fiscal year for the purpose of developing, implementing, and monitoring modification to the trauma-care component of the state plan for the provision of emergency medical services. The state plans were to be modified with respect to: (1) Trauma care regions, centers, and systems, (2) Triage and transport of children, (3) Evaluation, (4) Data reporting and analyses systems, (5) Procedures for paramedical personne l to assess the severity of injuries, (6) Transportation a nd transfer policies,

PAGE 54

45 (7) Public education, (8) Coordination and cooperation, and (9) Other m atters Finally, the states were require d to adopt standards to desi gnate trauma centers and for triage, transfer, and transporta tion policies and were mandated to submit to the Secretary: (1) the trauma care component of the state plan for the prov ision of emergency services, (2) at least annually, the information received via its data reporting and analysis system, and (3) identify and submit to the Secretary a list of rural areas lacking certain emergency medical services. An assessment of legislation and progr ess in individual states has not been included as a component of this review. Th e author previously reviewed activity and systems within the state of Florida and found support for the wide disparities in preparedness that are likely to be found between states and within a given state.

PAGE 55

46 Studies Supporting the Development of a Computer Simulation Model The literature reviewed in th is section is that which supported the developm ent of the computer simulation model that is the focus of this research. The methodology of achieving objectives 2 and 3 is directly linked to the information presen ted in this section. Portions of this review were also included w ithin paragraph 2.3. Ta ble 2 lists the studies utilized to develop the compter simulati on model for this research. Injuries associated with a blast event are generally categorized into four major categories or types: primary, secondary, tertiary, and misce llaneous. These categories are defined in terms of the mechanism leading to th e injury and the types of injuries likely to be seen. The following definitions and characteristics are adapted from articles by Slater and Trunkey (1997), Riley, Cl ark, and Wong (2002), and a CDC Injury Prevention document titled Explosion and Blast Injuries, A Primer for Clinicians (CDC, 2005). Primary blast injuries are those injuries that are caused by the blast wave as it travels through air or water. Air containing (or gas filled) structures within the body like the lungs, gastrointestinal tr act, and the middle ear are the most susceptible to primary blast injuries. Riley, Clark, and Wong labele d pulmonary barotraumas (blast lung), acute arterial or venous gas embolism, and intestinal barotraumas as the three killers of primary blast injury. Barotrauma is an in jury caused by pressure (Dorlands, 1988). An embolism is a sudden blocking of an artery by foreign material that has been brought to the site by the flow of blood (Dorlands, 1988). Other common injuries include pulmonary contusions, hollow viscous perforations, perforated ear drums, abdominal hemorrhage and perforation, globe (e ye) rupture, and concussion.

PAGE 56

Table 2 Published studies providing da ta/information for this rese arch Author (yea r) CDC (2005) Frykberg and Tepas (1988) Mallonee, et al (1996) Morbidity and Mortality W eekly Report (2002) Peleg, et al (2004) Riley, Clark, and Wong (2002) Slater and Trunkey (1997) Thompson, et al (2004) Secondary blast injuries are caused by prim ary and secondary missiles (flying objects) as they are propelle d outward by the explosion. Any body part may be affected with a secondary blast injury. Primary missiles come from the bomb container; secondary missiles come from the environment surrounding the specific blast location. Common injuries include pe netrating missile wounds, blunt missile wounds, eye penetrations, and orthoped ic injuries. Tertiary blast injuries are caused when the casualty is propelled through the air and hits a relatively fixed object. Common tertiary injuries include fractures and traumatic amputations, both closed and open he ad injuries, cervical spine injuries, and orthopedic injuries. As is the case with the secondary blast injuries, any kind of trauma/injury is likely for those in close proxi mity to the blast. Children are especially 47

PAGE 57

48 prone to the ef fects of tertiary blast injuries, this being a result of their small size and ease of being thrown around. Miscellaneous (or quaternary as labeled by the CDC) blast injuries are comprised of burn injuries, inhalation injuries, and crush injuries resulting from the collapse of a structure. Basically, any inju ry that is not included in one of the other categories is included here, to include the exacerbation or complication of existing conditions. Common injuries in this cat egory are burns, closed and open brain injuries, asthma, angina, hyperglycemia, hypertension, injuries as sociated with the inhalation of dust or gases, crush syndrome, and compartment syndrome. Crush syndrome is found when victims are trapped under a collapsed building for four hours or longer. All crush injuries will have some degree of rhabdomyolysis which is a syndrome characterized by the release of intracellu lar contents from injured skeletal muscle (Riley, 2002). The mechanism of the crush injury presents th e perfect situation for compartment syndrome which is defined as an elevated pressure within a closed tissue space which impairs neurovascular function, leads to tissue ischemia (deficienc y of blood in a part due to constriction), and death (Riley, 2002). Of note when considering both crush and compartment syndrome is the ease with whic h an untrained eye will miss the diagnosis and thus the subsequent a ppropriate treatment. Injuries falling within each of these blast injury categories are included in the model. Injured survivors of any size blast are very likely to experience injuries from each category, the type of injury being dependent on the proximity of the injured survivor to the blast and the characteristics of the facility or location where the blast occurs. In an overview of explosive-related injuries, the CDC categorizes eight body systems affected

PAGE 58

49 by injuries or conditions that are blast related (CD C, 2005). The eight systems along with their associated injury or condition are presented in Table 3. Additionally, the CDCs National Center for Health Statis tics (NCHS) sponsors the International Collaborative Effort (ICE) on injury statisti cs which has produced the Barell Injury Diagnosis Matrix, Classificati on by Body Region and Nature of the Injury (CDC, 2007). The entire matrix, as presented on the CDC web page, is displayed in Appendix 7. The description of each ICD-9-CM code by Nature of the Injury category is in Appendix 8. This matrix is based on ICD-9-CM coded da ta and provides a framework for categorizing trauma injuries. The aim of the matrix is to provide a common format for reports from trauma registries, hospital discharge data sy stems, emergency department data systems, and other non-fatal injury data (Barell, 2002). The matrix iden tifies the body region or site of injury on the vertical axis and the nature of the injury on the horizontal axis. The cells of the matrix contain the pertinent ICD-9-CM code s. Although having received some criticism for the divisions between body region categories and between the nature of the injury categories, multiple published studies focusing on trauma care have used this matrix as the framework for their analysis. The major categories of the vertical (body region) axis of the matrix include ; head and neck, spine and back, torso, extremities, and unclassifiable by site. The ma jor categories of the horizontal (nature of the injury) axis include; fracture, disloca tion, sprains and strains, internal, open wound, amputations, injuries to the blood vessels, cont usion/superficial, crus h, burns, injuries to the nerves, and unspecified as a category to ca tch all injuries that do not fit one of the other categories. The intersec tion of the body region/nature of injury categories is further defined by the ICD-9-CDM codes.

PAGE 59

Table 3 CDC Overview of Ex plosive-related Injuries System Injury or Condition Auditory TM Rupture, ossicular disruption, cochlear damage, foreign body Eye, Orbit, Face Perforated globe, foreign body, air embolism, fractures Respiratory Blast lung, hemothorax, pneumothorax, pulmonary contusion and hemorrhage, A-V fistulas (source of air embolism, airway epithelial damage, aspiration pneumonitis, sepsis Digestive Bowel perforation, hemorrhage, ruptured liver or spleen, sepsis, mesenteric ischemia from air embolism Circulatory Cardiac contusion, myocardi al infarction from air embolism shock, vasovagal hypotension, peripheral va scular injury, air embolisminduced injury CNS Injury Concussion, closed and ope n brain injury, stroke, spinal cord injury, air embolism-induced injury Renal Injury Renal contusion, lacer ation, acute renal failure due to rhabdomyolysis, hypotension, and hypovolemia Extremity Injury Traumatic amputation, fractures, crush injuries, compartment syndrome, burns, cuts, lacerations acute arterial occlusion, air embolism-induced injury Source: CDC (2005) 50

PAGE 60

51 When categorizing blast injuries, pu blished studies typically present a com bination of injury categories and anatomic location. The injury categories included in the model and the parameters assigned to t hose categories are based on an analysis and consolidation of the published experiences from the blast events that have occurred. Several studies present data and analyses from single blast events, e.g. the bombing of the World Trade Center in 2001, the 1996 Khoba r Towers bombing, and the 1995 Oklahoma City bombing. In others, the injury assessment is an aggregation of multiple blast events. In either case, it must be assumed that a blas t leading to a large scale mass casualty event will follow similar injury patterns. One difficulty presented in utilizing multiple studies was the lack of consistency between studies as to the injury categories used. Similar categories are often found, however it is diffi cult to make a direct comparison between studies and/or events. The primary study used for this research was conducted by Frykberg and published in 1988. In this study Frykberg revi ewed 14 studies of blast events published between 1969 and 1983. These studies considered 220 separate blast events accounting for 3,357 casualties. Table 4 displays the in jury categories, number of injuries, and percentage of injuries in each category as de termined by Frykberg. The applicable Barell Matrix categories have been added to the ca tegories presented in the article for present research purposes. Frykberg used a comb ination of injury categories and anatomic location to present his findings. The events in cluded in this study were relatively small. The mean number of casualties per blast event was 15.3 with a range of four to 346 per blast event.

PAGE 61

Table 4 T errorist Bombing Injury Analys is (220 events, 2,934 immediate survivors) Injury Category/Location Number % of Survivors Barell Category Head 920 31.4 Head & Neck Chest 53 2.0 Torso (Chest) Blast Lung 18 0.6 Torso (Chest/Internal) Abdomen 42 1.4 Torso (Abdomen) Burns 146 5.0 Burns Traumatic Amputation 36 1.2 Amputations Bony Extremity 320 10.9 Extremities Soft Tissue 1,624 55.4 Open Wound/ Blood Vessels/ Contusion/ Superficial/ Crush/ Burns/Nerves Totals 3,159* 107.9* Source: Frykberg and T epas, 1988. The Barell Categories (CDC, 2005) were added by Zuerlein, 2007. Injury category totals and the cumulative percents are greater than the total injured survivors and 100% because an injured surviv or may have more than one injury. Mallonee, et al., (1996) i nvestigated the injuries among the survivors of the Oklahoma City Federal Building bombing. Tabl e 5 depicts the injury categories, number of injuries of that type, percentage of injured survivors within that category, and applicable Barell Matrix categories. This study also provided a further analysis of the anatomic location of the injuries within specif ic injury categories. Table 6 depicts the 52

PAGE 62

Table 5 Oklahoma City Injury Analysis (592 Injured Survivors) Category Number % of Survivors Barell Category Soft Tissue 506 85 Open Wound/Blood Vessels/Contusion/ Superficial/Crush/Burns/ Nerves Fracture/Dislocation 60 1 Fracture/Dislocation Sprains 152 25 Sprains & Strains Head 80 14 Head and Neck Ocular 59 1 Head and Neck Burns 9 2 Burns Totals 866* 128* Source: Mallonee, et al., 1996. The Ba rell Categories (CDC 2005) were added by Zuerlein, 2007. Total m ay add to more than 100% because some injured survivors had more than one injury. anatomic location of only those injuries categor ized as soft tissue injuries. Additional categories included in the study were musculoskeleta l injuries, head injuries, ocular injuries, burns, and auditory damage. Ta ble 7 depicts the anatomic location of the musculoskeletal injuries. Other specifics c oncerning the injured survivors include eighty head injuries with eight considered severe and 72 categorized as mild or moderate. Thirty-five (44%) were hospitalized. Fift y-nine ocular injuri es resulted in 23 hospitalizations, nine ruptured globes, 15 co rneal or scleral lacer ations and six with lacerations, contusions, and or glass. Nine received burns generating seven admissions. The face and neck accounted for the areas most frequently burned (67%). Finally, 78 53

PAGE 63

Table 6 Oklahoma City Location of Soft Tissue Injuries (506 injuries) Location Number % of Injuries ICD Code Extrem ities 374 74 880-884, 890-894, 904, 912-915, 917918, 923-924, 927928, 943, 945, 953, 955 Head and Neck 243 48 873-874, 900, 910, 920, 925, 940-941, 947, 950-951, 953954, 957 Face 228 45 872-873, 941 Chest 177 35 875, 879, 901, 922, 926, 942, 953 Total 1022* 202* Source: Mallonee, et al., 1996. The ICD code s (as listed for the spec ified location in the Barell Matrix (CDC, 2005)) were added by Zuerlein, 2005. Total adds to m ore than 100% because some injured survivors may possess more than one injury. 54

PAGE 64

Table 7 Oklahoma City Location of Mu sculoskeleta l Injuries (Fractures and Dislocations (60 injuries) and Sprains ( 152 injuries)) Location of Fracture/Dislo cations (60) % of injuries ICD Code Legs 40% 820-827, 835-838 Arms 38% 810, 831-834 Face and Neck 37% 802, 807, 830 Back, Chest, or Pelvis 25% 805-809, 839 Location of Sprains (152) % of injuries ICD Code Chest and Back 53% 843, 846, 847 Neck 29% 848 Extremities (legs) 27% 843-846 Extremities (arms) 9% 804-842 Source: Mallonee, et al., 1996. The ICD code s (as listed for the spec ified location in the Barell Matrix (CDC, 2005)) were added by Zuerlein, 2007. 55

PAGE 65

56 m edical diagnoses of auditory damage were recorded. Thirtyone for hearing loss, 22 with bilateral or unilateral tympanic membra ne perforation, 13 with acoustic trauma, and 12 with tinnitus, vestibular injury, or otalgia. Peleg, et al. (2004) stud ied 623 patient recorded in the Israeli National Trauma Registry who experienced expl osion related trauma injuries. Table 8 identifies the distribution of injuries, by body region categories, for the trau ma patients. The injuries described in this study were not the result of a single large scale mass casualty event but rather multiple small events causing trauma that occurred over a 21 month period (1 Oct 2000 Jun 2002). The January 11, 2002 Morbidity and Mortal ity Weekly Report (Rapid Assessment of Injuries Among Survivors of the Terrorist Attack on the World trade Center New York City, September 2001) focused on the inju ries resulting from the 2001 attack on the World Trade Center. This study identified 790 injured survivors treat ed within 48 hours of the attack. The mix and distribution of in juries among the injured survivors is listed in Table 9. Thompson, et al. (2004) assessed the injuries to 401 survivors injured directly by the blast at the Khobar Towers military complex. This study was conducted through a review of medical records and a survey of th e survivors. A total of 574 casualties were identified. Nineteen of thes e died immediately. Those injured directly by the blast numbered 401. The other injuries occurred during evacuation operations. Thompson identified the overall mix of injuries and the mix of injuries by anatomic region. Table 10 lists the overall mix of injuries. The study of injury severity was bolstered in 1969 with the development of the

PAGE 66

Table 8 Peleg Study Injury Categoriz ation (623 patients) Injury Ca tegory Number % of Patients Barell Categories Brain 115 18.5 Traumatic Brain Injury Other Head 335 53.8 Other Head Spinal Cord & Column 30 4.8 Spine & Back Chest 135 12 Torso (Chest) Abdomen 75 21.7 Torso (Abdomen) Pelvis/Trunk 104 12 Torso (Pelvis & Trunk) Upper Extremities 242 38.8 Extremities (Upper) Lower Extremities 234 37.6 Extremities (Lower) Total 1270* 199.2* Source: Peleg, 2004. The Barell Categories (CDC, 2005) were added by Zuerlein, 2007. Total adds to m ore than 100% because some injured survivors may possess more than one injury. 57

PAGE 67

Table 9 World Trade Center Attack Survivors (790 injured survivors) Injury Ca tegory Number % of survivors Barell Category Inhalation 386 49 Chest/Internal Ocular 204 26 Other Head, Face, Neck Laceration 110 14 Open Wound Sprain or Strain 108 14 Sprains & Strains Contusion 98 12 Contusion/Superficial Fracture 46 6 Fracture Burn 39 5 Burns Closed Head 14 2 Head and Neck Crush 8 1 Crush MTotal 1013* 129* Source: MWR Report, 2002. The Barell Categories (CDC, 2005) were added by Zuerlein, 2007. Total adds to more than 100% because some injured survivors may possess more than one injury. 58

PAGE 68

59 Abbreviated Injury Scale (AIS). It w as intended to support the measurement of severity of injury (emedicine, 2005). The AIS was developed by a joint committee comprised of members of the American Medical Associa tion, the Society of Automotive Engineers, and the American Association for Automotive Medicine (now the Association for the Advancement of Automotive Medicine (AAAM)) (Stevenson, 2001). The AAAM claims the AIS is the standard of choice for severi ty of injury assessment (AAAM, 2005) and the AIS is believed to be the most widely used severity scale in North America, Europe, Japan, New Zealand, and Australia (Stevenson, 2005). The Injury Severity Score (ISS) was introduced in 1974 by Baker et. al. (Trauma.Org, 2005). The ISS also receives at tention as the gold standard and most widely applied instrument for assessing sever ity of trauma, no matte r what the source (Husum, 2002, emedicine, 2005, and Aharonson-Daniel, 2004). The ISS is a commonly used reference in the literature discussing blast related injuries and is used as a means of identifying severity of trauma among the injured survivors. The ISS score is derived from the abbreviated injury scale (AIS) and provides an anatomical scoring system that provides an overall score for patients with multiple injuries (Trauma Scoring, 2005). Each injury is assigned an AIS score and is allocated to one of nine body regions. These body regions are the head, face, neck, thorax, abdomen, spine, upper extremity, lower extremity, a nd external (burns, and other skin, and subcutaneous tissue injuries). The ISS u tilizes six body regions which are formed from the AIS categories. The six categories are head/neck, face, thorax, abdomen, extremities, and external. To calculate the ISS, only th e highest AIS score in each body region is

PAGE 69

Table 10 Khobar Tow ers Injury Analysis (401 injured survivors) Injury Ca tegory Number % of survivors Barell Categories Soft Tissue 380 94.8 Open Wound/ Contusion/ Superficial/Crush/ Burns/Nerves Foreign Bodies 195 48.6 Unspecified by Site Severe Lacerations 36 9 Open Wound Eye 27 6.7 Head and Neck (Eye) Brain 25 6.2 Traumatic Brain Injury Fractures & Dislocations 17 4.2 Fracture/Dislocation Strains or Sprains 54 13.5 Strains & Sprains 734* 183* Source: Thom pson, et al., 2004. The Bare ll Categories (CDC, 2005) were added by Zuerlein, 2007. Total adds to more than 100% because some injured survivors may possess more than one injury. 60

PAGE 70

61 used, the three m ost severely injured body re gions have their score squared and added together to produce the ISS. The AIS ranks in juries on a scale of one to six. Table 11 lists the score and injury ch aracterization. The AIS repr esents the threat to life associated with the injury rather than a comprehensive measure of severity (AIS, 2005). The ISS takes values from 0 to 75. If an injury is assigned an AIS of 6, the ISS score is automatically 75. Identified wea knesses of the ISS are that any error in AIS scoring increases the ISS error, different inju ry patterns can yield th e same ISS score and injuries to different body regi ons are not weighted. Hirshberg, et al., (2001) id entified those with an I SS of greater than 15 as critically injured. With a few exceptions, this was common place of division throughout the studies reviewed for this research. Table 12 lists the ISS data available within the published studies used for this research. In formation on all categor ies was not available within each study. The number s in parentheses following the percent indicate a different ISS range than the column heading. This is included to indicate the variance between studies with respect to grouping the se verity of injury scores. The International Classification of Diseases, Ninth Edition (ICD-9) Injury Severity Score (ICISS) is an injury se verity methodology based on the ICD-9 codes (Rutledge). More accurate ly, it is a prediction methodol ogy based on the ICD-9 codes and can be used as a predictor of survival and other outcomes for an injured individual (Rutledge, 1998). Although shown to be a better predictor of survival than the ISS (Rutledge, 1998 and Osler 1996), the ICISS was not used for this research because of its lack of utilization in the st udies utilized for this res earch. The ISS was the common means of assigning severity in the studies review ed/utilized for this research and was thus

PAGE 71

Table 11 Abbreviated Injury Scale AIS Score Injury 1 Minor 2 Moderate 3 Serious 4 Severe 5 Critical 6 Unsurvivable Source: Traum a Scoring, 2005. used as the basis for differentiating between injured survivors with minor injuries and injured survivors with se vere injuries. 62

PAGE 72

Table 12 Injury Severity Scores among Published Studies Author (year) ISS of 0-8 ISS of 9-15 ISS of >15 Frykberg (1998) 19% (8-34%)* Shamir (2004) 9% (>16) Adler (1983) 87% (1-6) 3% (7-12) 10% (>13) Frykberg (1989) 62% (0-10) 15% (11-15) 22% Brismar (1982) 67% (0-10) 9% (11-15) 23% Frykberg (2004) 20% Frykberg (2004) 22% Frykberg (2004) 23% Hirshberg (1999) 78% 7% (9-16) 15% (>16) Hirshberg (2001) 10-15% *8-34% represents the range of severe injuries in the studies review ed be Frykberg for his research. 63

PAGE 73

64 Chapter 3 Objectives of the Research In order to answer the research questi on defined on page six, multiple objectives were identified. The research question being, can computer modeling and simulation (as a methodology) be used to predict the medi cal resource requirements generated by this type of event and thus better prepare a health system in a defined geographic are for the possibility of an event of this nature? Th e completion of these objectives, as presented and subsequently discussed below resulted in the findings and conclusions discussed in later chapters of the document. Objective 1 : To describe the characteristic s of structures (sports arenas, convention centers, etc.) where large scale ma ss casualty events might occur, including the frequency of events held in those struct ures. Also to be de scribed here are the structural characteristics of the facilities, the mechanisms for delivering the conventional explosives, the location of the bl ast relative to the facility and the strength of the blast. Objective 2 : To describe the mix of specifi c injuries for those who survive a blast from conventional explosives, resulting in a large scale mass casualty event, as the injury prediction component of the simulation model. Objective 3 : To identify the types of medical assets needed to respond to a blast event, to include fixed facilities and equipm ent, transportation, temporary facilities and equipment, personnel and other assets associ ated with an emergency response to a mass casualty event, as components of the simulation model.

PAGE 74

65 Objective 4 : Use com puter simulation to predict the mix of injuries (defined by ICD codes) experienced by those who surviv e a large scale mass casualty event caused by conventional explosives using the total num ber of injured survivors as input to the simulation model. Objective 5 : Use computer simulation to predict the medical resources required to care for the predicted mix of injuries found for Objective 4. The injuries predicted in the simulation for Objective 4 served as an input to the simulation for Objective 5. Objective 6 : To compare the predicted resource requirements of a given scenario to the resources available in an area consid ered resource dense and another considered resource sparse as an exampl e of how planners in a specifi c location can use the tool to assess their own requirements fo r facilities, equipment, and personnel. This comparison between the predicted and available resources will identify the differential that would require additional planning and coordination to adequately respond to a large scale mass casualty event. The overall objective of this research wa s to provide a tool to assist health systems in a defined geographic location to pr epare for the possibility of a large scale mass casualty event. In this context, heal th systems does not refer to an affiliated group of health care facilities or resources but rather to the aggregation of health care resources available in a define d geographic area. In particul ar, this research focused on the medical resources required to care fo r the physical injuries resulting from a conventional explosive blast in or near an indoor arena seating 10,000 or more or an outdoor stadium seating 30,000 or more, though a ny location in which thousands, or tens of thousands of people gather, could be the locus of such a study, e.g. an auto speedway,

PAGE 75

66 train station, airport, concer t venue, office building, hospital, festival site, or convention center. The arena or stadium thus becomes th e geographic center of the defined location. The defined location encompasses an area around the arena or stadium that contains health care resources available to treat the in jured. The defined area does not necessarily coincide with the boundaries of a city, county or state. The defined location could be part of, or a combination of, these units. Within the United States there are 130 indoor arenas and 182 outdoor or domed stadiums fitting this descri ption (Appendices 3 and 4 present a picture of the size and distribution of these stadiums ). Although the facility may seat more than 10,000 (or more than 30,000 if outdoors), seating capacity must be differentiated from the total number of casualties and from the total number of injured survivors. Seating cap acity represents an estimate of the number at risk of death or inju ry when the blast occurs In reality, if the event is sold out, the number at risk will be something more than the maximum seating capacity when those who are working at the event are taken into consideration. The injured survivors are those present when the bl ast occurs, receive inju ries as a result of the blast, and require medical car e. It is the injured survivors who were of interest for this study. A blast of the sort considered for this research could be expected to generate casualty numbers in the thousands for the indoo r arena and in the tens of thousands for the outdoor setting. The predic tions generated by this rese arch were based on a minimum of 5,000 injured survivors which is larger than most mass casualty situations experienced to this point in time and larger than any si ngle previously occurring terrorist-related blast event.

PAGE 76

67 Multip le subordinate objectives were met to reach the overall objective; first was the definition of a large scal e mass casualty scenario, sec ond, the use of a prediction model to identify the possible mix of casualtie s generated by a blast in this setting, and third, the use of a prediction model to identify the medical resources required to respond to, and care for, the injured. Finally, the comparison of the resu lts of the prediction model with the resource capacity in a given area or system provides an example of the application of this methodology in a real wo rld setting. This co mparison provided the means to help identify the size of the geogra phic area that resources must be drawn from to meet the requirements of treating the injured su rvivors from a large scale blast event. The intent of the initial objectives wa s to develop a pred iction methodology, a computer simulation model, and subsequent simulations, which describes large scale blast events that produce mass casualties, the li kely injuries resulting from the blasts, and the medical resources required to treat the victims (to include transportation). This methodology can be used to help prepare a given jurisdiction fo r a large scale mass casualty producing event and to help assess the capacity of th e system to respond to such an event. The centerpiece of the methodol ogy is the analytic description (computer model) of the structure of the system. Th e model can be modified to simulate the characteristics of a specific health sy stem in a specific geographic location. The application of this to ol and the associated assessment of a unique geographic areas capacity to care for the injured is th e end result of this research. Clearly the capacity of local resources to respond to, and manage, a large scale mass casualty event varies within the United States. It is assu med that most large metropolitan areas in the U.S. possess extensive resources (i.e., they are resource dense) and thus have a greater

PAGE 77

68 capacity to m anage a large scale mass cas ualty event with limited assistance from contiguous geographic areas. In contrast, many cities with small populations would likely have fewer health care resources (i.e., they are resource spar se) and would require assistance from a very large geographic area to meet the resource requirements to care for the injured should a blast resu lting in a large scale mass casualty event occur. The contrast between these two examples indi cates the level of cooperation and planning required to prepare for the possi bility of a large scale mass casualty event, both in areas considered resource dense and those consid ered resource sparse. A resource dense geographic location would be characterized by mu ltiple large health care facilities with trauma centers within close proximity to the bl ast site (stadium or arena). The resource dense area would be supported by a robust em ergency response system to include extensive transportation resources. A resource dense area would also be supported by the full spectrum of surgical subspecialties and the associated support staff that would be required to deal with the mix of injuries produced by the blast. These would include general surgeons, thoracic surgeons, neurologists, ophthalmol ogists, oral and maxillofacial surgeons, orthopaedic surgeons, otorhinolaryngologists, and plastic surgeons. In a resource dense area multiple surgeons in each specialty area would be available. The diagram in Appendix 5 depi cts a hypothetical res ource dense geographic location. A resource sparse area may be characteri zed by a single commun ity hospital, with or without a trauma center, w ithin the city where the blast occurs. A distance of 50 miles or more may need to be traveled to find medical resources capable of supporting/responding to a mass casualty event. The resource sparse area will be

PAGE 78

69 charac terized by limited resources in all ar eas to include emergency response and transportation assets, and surgical specialist s. As an example, medical specialists available in a resource sparse area could be limited to a single general surgeon, an orthopedic surgeon and an ophthalmologist. Th e point being that a resource sparse area is going to have severe limitations both in te rms of the range of specialties available in the number of physicians practicing within th at specialty. The diagram in Appendix 6 depicts a hypothetical resource sp arse geographic location. The final product will help the respon se system on two fronts. First, the methodology can be used by local health planne rs to better-prepare for a response to a large scale mass casualty producing event. Second, the methodology can be used by policy makers to help identify the best use of resources to ensure maximum preparedness for a large scale mass casua lty event caused by the use of conventional explosives.

PAGE 79

70 Chapter 4 Research Methods The methodology for this research utilized existing data to build a computer simulation model to generate the mix of inju ries that may be produced by a large scale mass casualty event caused by conventional expl osives and to model a health system responding to the medical care needs of the injured. This model was then used to simulate the impact of a large scale mass casua lty event on a health system and to predict the medical resources required to care for the in jured. The intent of this research was to identify the resources required to manage a large scale mass casualty event and establish a methodology to answer questions concerning the readiness of a health system to respond to such an event.

PAGE 80

71 Objective 1 : To describe the characteristics of structures (sports arenas, convention centers, etc.) where large scale m ass casualty events might occur, including the frequency of events held in those structures. Also to be described here are the structural characteristics of the facilities, the mechanisms for delivering the conventional explosives, the location of the blast relative to the f acility and the strengt h of the blast. Methods for Objective 1 : Conceptually, the numb er and mix of casualties produced by a blast is a functi on of the facility type, the se ating capacity, the construction and structural configuration of the facilit y, the blast magnitude, and the blast location relative to the structure. A review of the literature, including re fereed and non-refereed publications, was used to develop a conceptu al model of the vulnerability of a local health system to a large scale mass casualty event using the variable s delineated above. The methodology presented here leads to a de scription of the sett ings/locations that would be at risk if this type of attack were carried out a nd a discussion of the remaining variables of the concep tual model. The first step of this process identified communities and thus health systems which are at risk of experiencing a large scale mass casualty producing event. This step addressed the variables of facility type and seating capacity. Those communities and thus health systems that are defined as at risk were those which possess an outdoor stadium with a seating capacity of 30,000 or more (foot ball stadiums, baseball stadiums, and auto racing venues) or those with an indoor arena with a seating capacity of 10,000 or more (basketball arenas, hockey arenas, and mu lti-purpose events centers). Although not assessed for this research, and often not po ssessing a specified s eating/holding capacity, convention centers, office buildings, airpor ts, shopping centers, manufacturing plants,

PAGE 81

72 large churches, and large train or subways stat ions offer other indoor facility types with the potential of holding more than 10,000 pe ople at one time. Thirty thousand was selected as the m inimum capacity for an outdoor stadium to limit the number of stadiums, and thus communities and/or health systems included in this analysis. The basis of the selection of 30,000 as a minimu m for outdoor stadiums was th at this is the minimum seating capacity for a facility to be used by a National Collegiate Athletic Association (NCAA) Division I Football program. The 30,000 figure would also include all professional baseball and football stadiums plus a variety of other venues such as auto racing tracks, horse racing track s, and outdoor amphitheaters. This figure precludes the inclusion of smaller college football stadiums as well as all high school venues across the country. Although not based on an NCAA st andard, targeting indoor arenas with a seating capacity of greater than 10,000 provide s a similar number of arenas as does the 30,000 standard for the outdoor stadiums. It was also assumed that to produce injured survivors in excess of the 5,000, that is the fo cus of this study, a relatively large venue would be required. The selection of minimum seating capacities for inclusion in this study does not negate the risk to locations w ith smaller facilities. It, however, does not consider these facilities and locations as at risk for a large scale mass casualty event (defined as one with 5,000 or more injured survivors). An inventory of outdoor stadiums and indoor arenas to include location, population, and seating capacity was assembled through an extensive search of websites containing information about outdoor stad iums and indoor arenas. These websites included both individual we bsites (i.e. a university, prof essional team, or specific facilities website) and websites containing information about stadiums or arenas in

PAGE 82

73 general (i.e. www.collegegridirons.co m). Co mmunity census data were added to the inventory through a search of the U.S. Census Bureau website. The U.S. Census Bureau identifies both metropolitan and micropolitan st atistical areas. The U.S. Census Bureau describes the general concept of a metropolitan or micropolitan statistical area as a core area containing a substantial population nucle us, together with adjacent communities having a high degree of social and economic integration with the core (U.S. Census Bureau, 2005). Each metropolitan area must have at least one ur banized area of 50,000 or more inhabitants. Each micropolitan area mu st have at least one urban cluster of at least 10,000 but less than 50,000 inha bitants. If the community where the stadium or arena is located is part of a Metropolitan (o r Micropolitan) Statistical Area (MSA), that population figure was used. If the community is not part of an MSA, the census figure for that community was used. The population of the location serves as a rough surrogate measure of the density of health resources av ailable in that area. Areas with a larger population will presumably have a greater density of medical resources available. The construction/structural configuration element of the facility subject to a blast was addressed through a search of the literatu re, web sites, and through observation of a non-probability sample of facilities and blast ev ents. This analysis served to identify structural characteristics that would make a f acility more or less susceptible to the effects of a large scale blast event. These characteri stics are presented in the results section of this document. An assessment of the importance of blas t magnitude and blast location was also accomplished via a search of the literature and web sites which address blast events of all types, both purposeful and accidental. The types of devices consid ered in relation to

PAGE 83

74 m agnitude and location was limited to those which have been used by terrorists using conventional explosives and result in a blast (thus producing fractures, wounds, and burns). Consideration of the effects of biol ogical, chemical, and nuclear devices was not part of this research. The magnitude of the blast was based on pounds of TNT (trinitrotoluene) equivalents. A TNT equivalent is a method commonly used to compare explosive energy to a standard (Thompson, 2004).

PAGE 84

75 Objective 2 : To describ e the mix of specific injuries for those who survive a blast resulting in a large scale mass casualty event as the injury prediction component of the simulation model. Methods for Objectives 2: Objective two was accomplished by building a computer simulation model to describe the mi x of injuries resulting from a large scale blast producing a mass casualty event. The in jury categories and associated parameters included in the model are based on an assessm ent of research and published studies of historical blast events. The simulation model was built using the Arena software program. Of importance to note here is that the me thodology used to develop this model did not rely upon a data set describing past la rge scale mass casualty events. An event producing the number of cas ualties that is the focus for th is research has not occurred. The examples of events that have killed or injured thousands of people at one time are few (the 1993 bombing of the World Trade Center, the 1996 bombing of a bank in Colombo, Sri Lanka, the 1996 bombing of the Federal Building in Oklahoma City, the 1998 bombing of the U.S. Embassy in Nairob i, Kenya, the September 2001 attack which destroyed the World Trade Center [not a purel y blast event like the others], the multiple bombings in 2004 of trains in Madrid, Spain, the multiple bombings in 2005 in London, and the multiple bombing targeting the Yazidis in Northern Iraq in 2007). As shown in Table 13, each of these events resulted in gr eater than 500 casualties (deaths and injured survivors). Although the numbe r of casualties produced by th ese events approaches the intent for the large scale mass casualty event pr oposed for this research, none represents an example or model for an attack on an outdoor stadium seating 30,000 or more or indoor arena seating 10,000 or more. Additionall y, only the events in New York City and

PAGE 85

Table 13 Terrorist related blast events causing more th an 500 injured survivors Year Location Facility Killed Injured 1988 Islamabad, Pakistan Ammunition Dump 93 1,000 1993 New York City World Trade Center 6 1,040 1996 Colombo, Sri Lanka Bank 90 1,400 1996 Oklahoma City Federal Building 168 591 1998 Nairobi, Kenya US Embassy 254 5,000 2001 New York City World Trade Center 2,700 1,103 2004 Madrid, Spain Train 202 1,400 2005 London, England Subway 56 700 2007 Yazidi Villages Multiple Bus Stations 500+ 320+ Oklahom a City have sufficient data publishe d to support the development of an injury prediction model. The computer model built describing the mix of injuries for those who survive a large scale blast event addresses two components; the type and/or anatomic location of the injuries and the severity of the injuries. The availabl e literature/studies typically address both components; however they are not addressed togeth er. The studies that have been published present a picture of the injury categor ies along with a separate picture of injury severity. Injury severity has not typi cally been reported within a specific injury category. These components are thus discussed as two separate variables and are included in the model as separate component s, both potentially applying to every injured survivor. 76

PAGE 86

77 The injury categories were established through a process of first identifying an injury category baseline, the Barell Injury Diag nosis Matrix in this case (see appendices 7 and 8), and then refining the baseline to matc h available studies in terms of identified injury categories. Parameters for use in the simulation were then assigned to the injury categories based on findings in the various st udies. The parameters are quantities that represent the characteristics of the distribution in terms of arithmetic mean and standard deviation. In this case, the parameters are values that represent the number of injured survivors in the given injury category as a percentage of the total number of injured survivors. When applied w ithin the simulation, the parame ter value is the chance of having that type of injury applied to each in dividual survivor. The development of the injury categories and prediction parameters was similar to a meta-analysis in that multiple studies were used to develop the final list of injury categories and parameters. Since no single source provided all the categories and parameters that are included in the matrix, and no two sources provided the same data, this compilation and comparison of different studies was required to refine the injury categories. Appendi ces 9 and 10 display the data used for this study. Within each injury category, a range of severity exists. For the prediction purposes of this research, it was important to include injury severity to aid in the identification of the health care resources that will be required to care for the injured. Injury severity is identified in the published studies by very basic methods, such as differentiating between hospital inpatient and ou tpatient care, and by more specific means such a rating based on the Abbreviated Injury Scale (AAAM, 2005) or an Injury Severity Score (ISS) (Trauma.Org, 2005) fo r a given injured survivor.

PAGE 87

78 The injury s everity levels were assigned to each individual survivor at two levels, critical and non-critical. The injury severity levels were assigned to the individual survivors rather than in asso ciation with a particular in jury because published studies only provide a global level of injury severit y. Published studies do not describe injury severity within an injury category. The assi gned parameters (percent of injured survivors with critical injuries and the percent of injured survivors wi th non-critical injuries) were established by using data from the published studies. As with the other parameters, severity can be easily modified and the si mulation run again to assess the impact of a change in severity. Once the injury categories and severity leve ls were identified, the structure of the computer simulation model was developed using the Arena Simulation software. The construction of the model follows the proce ss discussed in the research methods portion of the literature review. The Basic Module Template was used to develop the structure of the model. The example in Figure 1 (page 20) presented the Arena modules used for the injury mix component of the model. The model begins with the creation module where the injured survivor enters th e system. Each injured survivor then proceeds through a series of decision modules, one for each of the injury categories, and subsequent assignment modules. The parameter assigned to each injury category serves as the chance of a true finding assigned to the deci sion module. A true finding would indicate the injured survivor would have the particul ar injury. The assignment of the specified injury as an attribute of th e injured survivor follows the decision module and occurs in the assignment module. The model and associat ed prediction parameters are presented in the Results section below.

PAGE 88

79 Verification (ensuring the m odel behaves as intended) was accomplished through two steps. The first step was the process of correcting any structural or functional errors in the model as identified by the software. The software prompted this check of the model whenever it was run, or when prompted by the researcher. This process ensured that, from a structural/functional standpoint the model was built co rrectly. Corrections and modifications were made to the model as pr oblems were identified by the software. The second step of the verification c onsisted of comparing the results of a simulation run using the model with what w ould be expected for outcomes based on the total number of injured survivors and the para meters applied at each decision point. By this method it is determined if the simulation model processes the in jured survivors as it is intended/expected to do. Expected outcomes were calculated by the researcher using the parameters assigned to the model. The expected outcomes were the standard against which the results of the simulation runs were compared. This comparison was made at three different injured survivor levels 100, 1,000, and 10,000. Ten replications at each level were run and the means of the 10 re plications as well as the maximum and minimum values for each prediction component were compared with the expected values as calculated by the researcher using correlati on coefficients. When the variance between the simulated and expected number was too great (correlation coefficients =<.80), changes to the model were considered. As wa s expected (and thus further verifying the model), as the number of injured survivors (N) processed through the models got larger, the percent deviation from expected got smaller. Thus, as is expected from the Central Limit Theorem, the large N resulted in a normal distribution centered on the mean. In this case, the larger N results in an allocation of injured survivors to the injury categories

PAGE 89

80 in num bers that are closer to the expected va lues. Conversely, the smaller N resulted in a greater deviation from the expected. When 100 injured survivors alone were processed through the model, a one or two survivor devi ation from expected had a large statistical impact. This same impact was not found with 1,000 or 10,000 injured survivors. Validation (ensuring the computer model accurately mimics the real world situation) was not accomplished because no real world situation exists to model. Health systems in the U.S. have never faced a situa tion like that modeled for this situation. Where available, applicable published studies have been used to construct the model and to apply decision and process time parameters.

PAGE 90

81 Objective 3 : To identify the types of m edical asse ts needed to respond to a blast event, to include fixed facilities and equipment, transportation, temporary facilities and equipment, personnel and other assets associ ated with an emergency response to a mass casualty event, as components of the simulation model. Methods for Objective 3 : Objective three was accomplished by building a computer simulation model that describes the components and a ssociated processes necessary for a health care system to res pond to a large scale mass casualty event. The structural components, processes, and associat ed parameters were based on an assessment of research and published st udies of responses to ma ss casualty events, planning documents, and researcher identifie d components and parameters. As with the methodology for Objective 2, an analysis of the available literature addressing responses to trauma and mass casualty events was used to construct a model of the components of a health system that are required to respond to a mass casualty event and the time parameters, which were meas ured in minutes, for providing the care or service. In areas where li terature was unavailable, rese archer-defined components and parameters were included. These parameters are the time distribu tions assigned to the different process modules. The model then uses the defined pa rameters during the simulation runs to assign times to the different processes. The parameters assigned to the different processes were deri ved from a combination of published studies and researcher assigned estimates and/or expect ations of process times. The intent of this model is not to mimic a specific system but rather to provide a general model that is representative of a health system in its response to a large scale mass casualty event.

PAGE 91

82 The process modules used within th e m odel are built using resource units rather than identifying each and every resource that would be required to complete the process. The Arena software does afford the ability to use multiple and varied resources to support each process. Resource units were used to ensure the model remained as simple as possible while demonstrating the flexibility of the model and its value as a planning tool. A single resource unit can easily be translated into total resource requirements. For example, if transporting an injured surv ivor via ambulance requires one ambulance and two emergency medical technicians (EMT), a requirement for one transportation resource unit translates into one ambulance and two EMTs The use of resource units also allows the local planner or health system to defi ne their resource units and establish policy concerning their response to a large scale ma ss casualty event to include utilization of available resources. In the absence of published studies a ddressing a particular resource unit or process, researcher (or user) defined parameters or expectations were used. Given that there are no reference points for a health system in the United States to respond to a mass casualty event of the magnitude modeled here, planners must establish expectations or best estimates as to different processing times. The parameters become assumptions as to how the model/system will work under conditions created by a large scale event. These assumptions could be modified based on user (planner) defined expectations. The final structure of the model and assigned parameters ar e presented in the resu lts section below. Frykberg (2002) provided the basic struct ure for the flow of injured survivors (casualties in Frykbergs figure) following a disa ster. This flow is presented in Figure 4. This figure presents the structure of the ma jor components of the model built for this

PAGE 92

Rescue Decontamination Sorting & Life Support Evacuation Definitive Care Figure 2 Casualty Flow in a Di saster Situation (Frykberg, 2002) research with the exception of decontaminati on. The present research is focused on blast events and not chemical, biological, or nucle ar, therefore the decontamination process was not included. Each of the components repr esents ongoing processes that are required to provide the care needed by the injured survivors. Models focused on the care provided to the injured survivors on ce they reached the hospital were presented by Hirshberg (1999), Shamir (2004), and Almogy (2004) and each contribute to or reinforce the model. These models are displayed in Appendix 11. The model displaying the health care system constructed for this research follows Frykbergs basic casualty flow model. The structure of the mode l to include injury assignment is displayed in Figure 3. Th e field activity submodel includes the 83

PAGE 93

Injury Identification Submodel (Objective 2) Injury Categories Severity Field Activity Submodel Rescue/Retrieval Process Field Care Process -Sorting and Life Support Transportation Process Start: Creation of Injured Survivors Facility Activity Submodel Sorting Emergency Department Care Trauma/Specialty Care Hospitalization End: Release of Injured Survivors from the hospital Figure 3 Basic Structure of the Simulation Model Used for the Present Research 84 rescue/retrieval process, th e field care process, and th e transportation (evacuation) process. W ithin the field care process are el ements of the initial sorting and life support process. The facility activity submodel incl udes a second element of the sorting and life support process, and then the definitive care process within the hospital, either in an emergency department/urgent care track, or a trauma/surgery track. Both the emergency

PAGE 94

85 departm ent/urgent care track and the trauma/surgery track may result in hospitalization and then release. In addition to presenting a model, Hirshberg (1999) conducted a computer simulation study designed to identify the point, in terms of patient load, where the quality of care began to decline in a mass casualty situation. This represents the only study found which offers data addressing the respons e to a mass casualty s ituation. Hirshbergs study used the Arena Software program. Para meters from Hirshbergs simulation model that were applicable to, and used in, this research are listed in Appendix 12. Once the structure and associated paramete rs of the model were identified, the model was built using the Arena software program. This portion of the model was a continuation of what was built to accomplis h objective two. The components of this portion of the model consist of decision m odules and process modules as described within the literature review. Once completed, the same verification pr ocess of the model that was described for Objective 3 was followed.

PAGE 95

86 Objective 4 : Use com puter simulation to predict the mix of injuries (defined by ICD codes) experienced by those who survive a larg e scale mass casualty event using the total number of injured survivors as input to the simulation model; Objective 5 : Use computer simulation to predict the medical resources required to care for the predicted mix of injuries found for Ob jective 4. The injuri es predicted in the simulation for Objective 4 served as an i nput to the simulation for Objective 5. Methods for Objectives 4 and 5: The fulfillment of objectives 4 and 5 provide examples of how the computer simulation model will be used to predict medical resource requirements for a large scale mass casualty event and to descri be the impact of constrained resources on caring for the in jured survivors. The methodology discussed here explains the selection of the criteria for conducting the multiple simulations and the steps required to set up the different simulations to run as desired. A total of six simulation runs were conduc ted to provide the framework for future use of the simulation model. Table 14 di splays a matrix which describes the six simulation runs used for this research. Two le vels of injured survivor s were selected with three separate runs of 10 iterations each c onducted for each level. The two levels of injured survivors were set at 7,000 and 45,000. This represents both a relatively small, large scale mass casualty event, representative of what might happen in an indoor arena, and a larger event that could o ccur in an outdoor stadium. The three separate runs at each level of injured survi vors serves to exhibit the flexib ility and effectiveness of using computer simulation to predict both the injuri es and the medical resource requirements in a large scale mass casualty event and then to assess the impact constrained resources have on caring for the injured survivors.

PAGE 96

Table 14 Simulation Run Matrix Injured Survivors Time Cons traint Resource Constraints 7,000 None None 7,000 12 Hours Sparse 7,000 12 Hours Dense 45,000 None None 45,000 12 Hours Sparse 45,000 12 Hours Dense The first of the three runs was conducte d with no constrai nts (unconstrained) placed on th e model. No constraints on the resources meant that it was assumed that there were no queues and no waiting times when an injured survivor sought to access the resources necessary to complete a care proce ss. Whenever an inju red survivor (entity) required a resource, a resource was available. This occurred because each resource category had an unlimited number of resources available. There were no constraints on the number of ambulances, hospital beds, surgeo ns, etc. The result of this first, unconstrained simulation run was to identify the total number of units of service required for each resource category. Each unit of se rvice represents a new resource unit that provides service to only one injured survi vor. The unconstrained model also included no time constraints; the simulation was allowe d to run until all injured survivors fully processed through the system, thus identifying the total units of service required to care for all injured survivors. A unit of service represents a single care process required by a 87

PAGE 97

88 single inju red survivor. A single injured survivor may require multiple units of service, each unit being comprised of a different care process. The second and third of the three s imulation runs were each conducted with constraints on both time and resources. A tim e constraint of 12 hours was placed within the computer simulation model. This time c onstraint was selected by the researcher as one which would allow the event to fully de velop in terms of engaging all injury and resource categories but not runni ng so long as to allow all in jured survivors to be fully processed through the model. This would thus contribute to dem onstrating the results produced by placing constraints on the m odel. After 12 hours of processing, the computer simulation run was terminated. Da ta produced by the simulation run indicated the number of injured survivors which were processed by each process module and the number of injured survivors who processed th rough the entire model, and then exited the model, within the 12 hour time constraint. Note that this 12 hour c onstraint is applied within the simulation model and is simulation time. Because the speed at which the simulation is run/processed can be increas ed, 12 hours of simulation time requires less time than this in actual processing time on the computer. Additionally, had the simulation been allowed to run for additional time (24 hours or longer), no change in the nature of the outcome would have been reali zed, only a change in the volume of outcome as a greater number of injure d survivors would have entere d the simulation. Constraining the resources meant that there was not an unlimited number of resources available for processing injured su rvivors within each process module. The constrained resources meant that the injured survivors processing through the computer simulation model will now experience queues and waits for the various processes

PAGE 98

89 included in the m odel. Two levels of resour ce constraints were placed on the resources included in the model. In the second si mulation run, resources representative of a resource sparse area were available. The res ources available in a re source sparse location were loosely modeled after a small geogra phic location with an MSA of less than 200,000. In the third simulation run, resources re presentative of a resource dense area were available. The resources available in a resource dense setting were loosely modeled after a large geographic location with a MSA greater than 2 million. In each case the researcher sought to use an internet search of city and county government websites, the websites of medical facilities, and the websites of other medical resources within the MSA or city to develop an assessment of the resource availability in the respective areas (resource sparse and resource dense). Where data were not readily accessible, estimates were developed to ensure that all resource categories were represented. In reality, definitive data are not readily available on the internet. Research often resulted in as many unanswered questions as answered que stions. All resour ce constraints are researcher defined and based loosely on available data for a relatively small and large MSA. Resource parameters are included in Table 15. The resource categories established for the representative areas incl ude all that were included in the computer simulation model and are described in the list below. In each case, the available resources were manually altered within the computer model to mimic what is found in either the resource sparse or resource dense area. Rescue Resource: For the purposes of this study, rescue resources as characterized as formally designated search and rescue units within a community. These units have primary responsibility for finding and rescuing those who may be trapped in

PAGE 99

Table 15 Resource Constraints Resource Category Sparse Dense Rescue 1 40 Informal Transportation Infinite Infinite Formal Transportation 4 40 Field Triage 1 10 Field Care 4 40 Hospital Triage 2 20 Emergency Care 4 40 Trauma Care 1 10 Imaging 4 40 Hospital 100 1,000 ICU 8 80 Thoracic Surgery 2 20 Orthopedic Surgery 6 60 General Surgery 8 80 Neurosurgery 2 20 Researcher Defined Cate gories and Parameters the structu re of the blast site. Formal Transportation Resource: Th e formal transportation resource is comprised of the number of ambulances and other dedicated emergency response vehicles readily available to transport patients. 90

PAGE 100

91 Inform al Transportation Resource: Inform al transportation cons ists of any means of transportation other than the formal system of ambulances and other emergency response vehicles. The primary means of info rmal transportation will be privately owned vehicles. A key aspect to the informal tr ansportation resources is that there is no predesigned control/dispatch mechanism. Field Triage Resource: Th e person or persons located in the field at a central casualty collection point to perf orm triage; both prior to transport for definitive care or prior to care administered in the field. Field Care Resource: A common theme in the literature addressing the medical response to mass casualty events is the need to identify those in crit ical need of care and those who can wait. The ability to deploy resour ces to the field, to triage and to care for the injured survivors is one means of keepi ng the less injured out of the emergency rooms and reducing the initial burden on the hospitals and emerge ncy rooms. This category is not intended to include first responders and t hose who would be requi red to extricate the casualties. This research begins with the process of caring for th e injured survivors post extrication. Hospital Triage Resource: This is the person or persons performing triage as injured survivors arrive at the hospital. Emergency Care Resource: A substantia l amount of care is expected to be provided on an outpatient basis. The capacity of the ER to care for the injured in the initial surge of patients will contribute to the determination of waiting times and additional capacity that will be required. A Le vel I, II, or III trauma designation will also

PAGE 101

92 serve as an indicato r of capacity across the medical facilities to care for the injured survivors. Trauma Care Resource: This resource is in addition to the emergency care resource. The trauma care resource is dedi cated to providing trauma and shock care to those with injuries requiring this level of care. Imaging Resource: Support from the anci llary services (imaging in particular) must be adequate to support the requirements of the injured survivors. If adequate services are not available, po licies may be required in term s of the use of imaging and other ancillary services to ensure care is prompt and expedient. Most literature recommends limiting studies as much as possi ble, therefore no delineation of specific types of studies was included for this research. Hospital Resource: The bed capacity of the facilities will provide the preliminary estimate of capacity of the local resources to care for those injured survivors who will require hospitalization. A s ubsequent differentiation of specialty care beds further delineates capacity and identifies need in t hose specialty areas. The hospital resource implies not only the space for the injured survivor but also the staffing to provide care. Intensive Care Unit (ICU) Resource: Similar to the hospital resource, the ICU resource identifies both the space and equipment to provide intensive care and the associated staffing to administer the care. Surgery Resource (Thoracic, Orthopedic, General, and Neuro): The OR capacity will provide the means of determining how quickly patients requiring surgery can receive surgical care. Surgical resources in the different specialty areas include the space, surgeon, and support staff to perform the surgery.

PAGE 102

93 Upon completion of the 10 iterations of each of the six simulation runs, the arithmetic means, standard deviations, a nd minimum and maximum values (arithmetic means) for the data elements for each of the simulation runs was used to display the results of the six simulation runs. The results of the runs are listed in the results section below. Comparisons between the different simulation runs provide examples of the flexibility of a computer simulation model to meet the needs of local planners in all areas.

PAGE 103

94 Objective 6 : To com pare the predicted resource re quirements of a given scenario to the resources available in an area considered resource dense and another considered resource sparse as an example of how planners in a sp ecific location can use the tool to assess their own requirements for facilities equipment, and personnel. This comparison between the predicted and available resources will id entify the differential that would require additional planning and coordina tion to adequately respond to a large scale mass casualty event. Methods for Objective 6: Objective six was fulfilled by presenting an example of how a medical planner or public health professional woul d use this tool in both a resource dense area and a resource sparse ar ea. This was done in order to provide an example of how the computer simulation model can be used at a local level. In this role, the researcher compared the medical requi rements generated by the model with the resources available in the local area in an e ffort to assess the preparedness of the given area and to establish a baseline for planning a response to an ev ent of this nature. This assessment was made at two levels, one with 7,000 injured survivors and the other with 45,000 injured survivors. The results of the assessment provide the basis for planning how best to use the existing resources and for planning how to access supporting resources to meet the medical car e requirements of the injured. The data produced by the unconstrained simulation runs of 7,000 and 45,000 injured survivors for Objectives 4 and 5 were used to start this assessment process. The data produced by the simulation runs identi fied the total number of resource units required to provide care for all injured su rvivors with no queues or waiting times to access the resource units. Again, a resource unit represents the aggregation of resources

PAGE 104

95 required to perform a certain task, for example, a surgical resource unit is made up of an OR, Surgeon, Nurse, Anesthetist, etc., all of the personnel, equipment, and facilities it takes to provide the service or perform that task. The arithmetic means for the total resource requirements were entered into data tables designed to display the comparison between the required resources and the available resources. These tables are displayed and discussed in the results section below. The second step of the assessment was to identify the resources available in the planners (or public health professionals) geographic location. The data obtained for objective five were used to establish the re presentative resource dense and representative resource sparse geographic area. Once collected these data were entered into the previously mentioned data tables to conduct the comparison/analysis. The tabl es incorporate columns that identify the resource requirements as predicted by the computer simulation model, resources available in the local area, resource required to care for patient in an 8, 12, and 20 hour window, and the differential between resource availability and the predicted resource requirements. As a final step, the resources required to provide care within a 12 hour window were placed in the model as the resource c onstraints. The simula tion was then run to completion to assess the impact of this level of resource availability. The results of this simulation are included in the results section below.

PAGE 105

96 Chapter 5 Results The sections that follow discuss the results related to each objectives identified to fulfill the intent of this research. The end point of the research being a tool that can be used by local area health pl anners to prepare their jurisdictions in the best manner possible for a large scale blast event.

PAGE 106

97 Results for Objective 1 : This objective describes the charac teristics of facilities (sports arenas, convention centers, etc.) where larg e scale mass casualty events might occur, including the frequency of events held in those facilities. Also described are the structural characteristics of the facilities, the mechanisms for delivering the explosives, the location of the blast relative to the f acility and the strengt h of the blast. Facility Characteristics The characteristics of faci lities located across the United States are highly varied in terms of size, appearance, and locati on within the community. The description included in this section focuses on indoor arenas and outdoor stadiums with limited attention to convention centers and other types of facilities. Research substantively based on internet searches revealed 130 indoor arenas with seating capacities of 10,000 or greater and 189 outdoor stadiums (and domed football fields ) with seati ng capacities of 30,000 or greater. If arenas and stadiums of all sizes were added to the list these totals would grow exponentially. Texas alone po ssesses 988 facilities designated as High School Football Stadiums (www.texasbob.com, 2005). In addition to these structures, more than 600 Convention Centers exist across the United States (www.convention centers.us/, 2005). The Convention Centers rang e in size from relatively small exhibition halls of several thousand square feet to large complexes of several hundred thousand square feet spread across multiple facilities and multiple meeting rooms. Convention Centers were not included in the analysis of indoor arenas and outdoor stadiums presented here. The point of identifying the facilities that exist acr oss the country is not to label each as being at risk of a blast event, but rather to develop th e idea that the health care systems in the communities where these fa cilities are located will be overwhelmed if

PAGE 107

98 a larg e scale mass casualty event occurs. As will be described further, some of these communities are very large and have extensive health care resources that may be able to adequately respond to a large scale event, othe rs are small and have limited resources and a limited ability to respond to a large scale event using local resources alone. In each case, without extensive planning, the health care systems will not be able to mount effective responses to blast events resulting in mass casualties. Establishing seating capacities of 10 ,000 for an indoor arena and 30,000 for an outdoor stadium as a minimum, 45 states and the District of Colu mbia possess at least one facility that meets one of the minima. Alaska, Connecticut, Maine, Montana, and Vermont do not have a facility that m eets the 10,000 or 30,000 seat capacity for the respective facilities. Delaware, Hawaii, New Hampshire, and Rhode Island each possess only one facility which meets the criteria. On the other extreme, Texas (25), California (23), and Florida (19) lead the way with the most facilities meeting the minimum capacity requirements. Appendix 13 lists the distribution of arenas and stadiums across the states. Appendix 14 further defines th e state-by-state situ ation by listing the individual arena or stadium, its location, the population of the location, and the seating capacity of the facility. Within the states, arenas and stadiums are located in a broad spectrum of locations, ranging from inner city urban locations, to th e suburbs and rural settings. Appendices 15 and 16 list the distribution of arenas and stadiums based on the population of their location (i.e. the 2000 census of the city li sted as the location or the MSA for that location). The mean population of a city wher e an indoor arena with a seating capacity of 10,000 or more is located is 2,186,620. Fifty per cent of these arenas are in cities with a

PAGE 108

99 population less than 1,000,000 and 30% in cities of less than 350,000. The m ean population of a city with an outdoor stadiu m is 2,133,204. Fifty percent of the stadiums are in cities of less than 1,000,000, 40% in c ities of less than 500,000, and 25% in cities of less than 200,000. A point of importance he re is that there are many stadiums which, when full, have a population rivaling the populat ion of the city in which they are located, and in some cases a population larger than th at city. Whichever the case, the stadium may become one of the largest concentrations of population in the state for that period of time. The surge loads for the health systems in these locations would appear to be much greater than in a location where 50,000 or 60,000 fans in attendance at a football game represent only a small fraction of the overall population of the loca tion. For example the University of Iowas Kinnick Stadium hol ds 70,000 people and is located in the Iowa City, Iowa MSA with a population of 131,676. The University of Wyomings War Memorial Stadium has a seating capacity of 33,500 while its host city, Laramie, Wyoming, has a population of 32,014. In contrast, Shea Stadium holds approximately 50,000 people and is located in the New York City MSA with a population of 16 million. Pasadena Californias Rose Bowl holds 91,136 and is located within the Los Angeles MSA with a population of 12 million. Frequency of Use The frequency that the facility hosts even ts varies between locations and between arenas and stadiums. As a rule indoor arenas are used more frequently than are outdoor stadiums. The vast majority of the outdoor and domed stadiums are associated with a specific professional or college football or ba seball team. It would appear that these stadiums are primarily used for the intended pur pose, either football or baseball. The use

PAGE 109

100 of these stadium s for other events is not common. Stad iums which host a professional football team will typically offer 10 events per year. Those hosting a college football team will likely offer no more the 7 events per year. Facilities hosting a professional baseball team have significantly more events hosting at least 80 games per year. Teams that reach the playoffs host additional games. In contrast, many of the indoor arenas, although being the home of a professional or college basketball or hockey team, appear to be used more often for other events, such as other forms of entertainment (concerts, show s, exhibitions). A review of the websites of several indoor arenas gi ves listing for events in 2004 numbering 94 (American Airlines Arena in Miami), 95 (Taco Bell Aren a in Boise, Idaho and the Sun Dome in Tampa, Florida), 161 (Fleet Center in Bost on, Massachusetts), and the Pepsi Center in Denver, Colorado claims more than 200 spor ting events, concerts, and special events each year. The events held in these facili ties cover a broad spectrum of entertainment and special events; basketba ll, hockey, arena football, l acrosse, soccer, commencement ceremonies, concerts, and home and sports product exhibitions. This frequency of events supports the idea that that there are many opportunities fo r a large scale mass casualty event to occur. Structural Characteristics When analyzing the structural characterist ics of arenas and stadiums, there are as many appearance and styles as there are lo cations. Although a basic design may be followed, each facility is typically distinctive of that location. What does not vary from facility to facility are the primary constr uction materials; steel and concrete. These materials were present in the c onstruction of the oldest facili ties in the inventory and are

PAGE 110

101 present in the newest. One addition to these m a terials is a greater use of glass in many of the modern designs. Outdoor stadiums present a mix of very old facilities with multiple modifications and improvements resulting in their current cap acity and configuration, to new, state of the art facilities. Georgia Techs Bobby D odd Stadium is the oldest stadium used for college Division I football (C ollege Gridirons, 2005). Th e stadium was originally constructed in 1913 with a capac ity of 5,600. The current cap acity is 55,000. Ohio State Universitys football stadium was built in 1922 with an original capacity of 66,100. It now holds 101,568 (College Gridirons, 2005). The University of Michigans football stadium was built in 1927 with a seating capac ity of 72,000. It has grown to its present capacity of 107,501 (College Gridirons, 2005). The University of Floridas Ben Hill Griffin Stadium has gone through six separate construction phases to reach its current capacity of 88,548. Not all facilities are old, new construction is common. As examples, Raymond James Stadium in Tampa, Florida was completed in 1998 and Heinz Field in Pittsburgh was completed in 2001. These examples are representative of what the research revealed with respect to stadiums and arenasthe last 10-20 years has seen an abundance of new construction. Whether it is modifications to an existing structure or the construction of an entirely new facility, enhancing the facility amenities, increasing capacity, and providing a better viewing experience to the cust omers/fans is an active business. Indoor arenas present a contra st in that as a whole they are more modern than their outdoor counterparts and less likely to ha ve been modified. Many old field houses

PAGE 111

102 have been replaced by m odern facilities provid ing more seats, more comfort, and a better view of the playing surface. Indoor arenas are typically stand alone fac ilities constructed pr imarily of steel and concrete with a steel supported roof. In most instances floor level (i.e. the playing floor) is at ground level, therefore th e entire structure is built from the ground up and may reach 10 stories into the air. Figure 6 displays this type of facility structure. The lowest level of seating is at the ground or floor level and seating is then built at increasingly higher levels, thus support for the diffe rent levels of seating is pa rt of the construction. The entrance to the facility is typically at ground level with the walkways/hallways around the facility then being underneath much of the s eating. The outside perimeter of the building is then associated with the highest level of seating which is also the farthest from the playing area (court). The concessions and ve nding areas are located in the walkways or hallways which circle the build ing. An indoor arena typically has at least one loading dock where large trucks delivering supplies a nd/or equipment can, at a minimum, reach the perimeter of the facility. In some inst ances these loading docks may be configured within the structure of the facility. An alternative to the above ground structur e is one where the court is sunk into the ground. Figure 7 presents an outside picture of this type of structure. In this case some or all of the seating is at or below ground level. Support for the seating is the excavated ground or a portion of the facility whic h is not exposed to the open air. In this setting the entrance is on the ground level a nd the walkways at ground level are equal to the highest level of seating. The extremes have been described he re; either 100% above

PAGE 112

ground or 100% below ground, facil ities and the level of the playing surface and seating with respect to ground level c over the range be tween the two. Figure 4 Indoor Facility above Ground Structure (Conseco Fieldhouse, Indianapolis, IN) (source: http://basketball.ballparks.com/NBA/IndianaPacers/newsindex.htm) 103

PAGE 113

Figure 5 Indoor Facility below Ground Structure (University Arena, Albuquerque, NM) Outside view of University Arena Inside view of University Arena taken from ground level. (source: http://www.virtualalbuquerque.com ) 104

PAGE 114

105 One additional construction material that is very prevalent in many new indoor facilities is glass. Many mode rn facilities use a great de al of glass in the outside construction. The danger to those on the inside exposed to flying glass would seemingly be much increased over those in a facility with a more solid exterior construction. Conversely, a blast on the inside of a glass structure would seemingly be less contained within the structure but would present an increased danger to anyone on the outside of the building. Outdoor stadiums, including large domed stadiums, present a similar picture in regard to basic construction. Most facilities are those where the field level is at ground level with the seating built up from that point. As with the indoor arenas, the seating in this configuration is entirely supported by the structure of the facili ty (i.e. by steel and concrete). Three alternatives are seen with the structures that are 100% above ground. In some facilities the underside of the stands is en tirely open. In this case there is direct access to the underside of the structure from below. In the context of a blast event, nothing stands between the blast location and the structure which holds the seating. Another alternative is where the space under the stands is enclosed because of various uses of the space; offices, meeting rooms, athl etic training and weight rooms. Similar to the indoor arena, the internal st ructure is not visible and is pr otected to a certain extent. In this case the supporting structure of the st ands would be protected by the structure of the facilities taking up the space under the sta nds. A third alternat ive would represent a modification to one of the previ ous alternatives. In this case cantilevered decks are added to the structure and are placed above (or overh anging) a portion of the existing stadium. Websters dictionary defines cantilevered as a bracket or block supporting a balcony or

PAGE 115

106 cornice. In the case of th e stadiums, these brackets or blocks support an additional balcony for spectator seating. Figure 8 displays the above ground structure of an outdoor facility as described here. Alternatively, some stadiums are sunk into the ground which provides an extra measure of protection (and support) against any type of blast fr om outside the facility. In this instance the bulk of the seating is supported by the ground itself. There is no access to the structure of the facility. The only acce ss is to the components of the structure that are at ground level. Figure 9 displays the in-ground or below ground structure for an outdoor facility. Again, as with the indoor ar enas, this is not an either/or proposition, there is a range of in-ground vs. out-of-ground configurations. Critical to the consideration of the casua lties that may be generated by a blast is the accessibility and susceptibility of a structure to a blast. Those structures that are entirely above ground are seemingly more sus ceptible to damage from a blast than are those that are protected by the ground. Those stadiums with cantilevered balconies also appear to be at greater risk of collapse if a blast were to occur. A planner might question if older facilities are more susceptible to damage and collapse than is a newer facility. An additional consideration may also be the chance for collateral damage. It is not unusual for facilities to be located in a downtown area where close proximity to other buildings may lead to additional damage and in juries. In contrast, with a facility that stands by itself a good distance from other f acilities, damage will be limited to that facility alone. Explosive Delivery Mechanisms Blasts leading to mass casualty situations have been delivered to the intended

PAGE 116

Figure 6 Outdoor Facility above Gr ound Structure (Raymond James Stadium Tampa, Florida) (source: http://www.buccaneers.com/news/newsdetail.aspx ) 107

PAGE 117

Figure 7 Outdoor Facility below Ground Structure (Michigan Stadium Ann Arbor, MI) (source: http://bentley.umich.edu/athdept/stadium/stadtext/scrapp0.htm ) 108

PAGE 118

109 target in a variety of ways. Past events would dictate, whatever the m aterial used, a large scale blast is likely to require a large capaci ty mechanism for delivery. The large blasts experienced thus far have required a truck to deliver the explosives. It must be assumed that any future large scale blast event woul d likely require the use of a truck as the delivery mechanism. The blast material in th e 1983 attack on the Marines in Beirut was carried by a Mercedes truck. The material used in the1993 bombing of the World Trade Center was carried by a Ryder truck. The ma terial used in the Oklahoma City bombing was carried by a U-Haul truck. The material used in the attack on the Khobar Towers military complex was carried by a tanker truck. With the explosives that have been used, the size of the blast will be dependent on the size of the vehicle available/used for transportation. A reference to a Bureau of Alcohol, Tobacco and Fi rearms endeavor to identify/predict the lethal range of a blast in the late 1990s stated a compact sedan carrying 500 pounds of dynamite would be expect ed to produce an airblast that can kill up to 100 feet away and has a shrapnel ra nge of up to 1,500 feet (Kitfield, 1998). A semi-trailer carrying 60,000 pounds of explosives could be e xpected to produce a lethal shrapnel cloud that extended n early 1.5 miles (Kitfield, 1998). When considering the large scale nature of the blast event to be modeled, the use of a truck as the delivery m echanism must be viewed as the most likely method. Cars, ambulances, preset explosives (as in the case in Beslan, Russia), aircraft, and individual or multiple suicide bombers strapping explosives to their bodies, as well as any other innovative means could all be cons idered possibilities. The real ity of the situation is that the amount of explosives required to destroy a steel and concrete st ructure would require a large truck or airplane. Individual or multiple suicide bombers could cause widespread

PAGE 119

110 injur ies but it is not likely that they would be able to pack the explosives required to severely damage or destroy a large building or portion of a stadium. Two of the largest man made (and accidental) blasts recorded are those associated with ships. This places most arenas and stadiums in a safe pos ition although port cities if there is close proximity to a stadium, arena, or convention center, must take note of the magnitude of the explosion and the potential for damage simply based on the tonnage a ship can carry. Location of the Blast The location of the explosive (i.e. wher e the explosive is located when it is detonated), relative to the f acility, is a strong determinant of the damage that will be caused to the structure, and thus determines the number and types of injuries. The likelihood of a structure collapsing increases if the blast comes from within the structure. As an example, the truck carry ing the explosives that hit th e marine barracks in Beirut made it into the building and destroyed much of the building. Th e blast that hit the Khobar Towers military complex was adjacent to the building but not with in the structure. The death and injury totals would likely have been much higher if the blast had caused a portion of the building to collapse. A blast in the open air rapi dly dissipates, thus reducing the damage that is caused. The ability or likelihood of a terro rist to be able to get within the structure of the facility to det onate a blast could act as an indicator of the destruction/injuries that might be caused. Th e assumption is that a blast from within the structure of the facility is going to cause mo re damage and generate more casualties than a blast from the outside of th e facility. Improved security measures since September 11, 2001, especially at big events, provide hope that creating a blast within a structure is less

PAGE 120

111 likely now than before that date. The locati on of the blast relative to the structure is, however, a key component for the pred iction of casualties. Strength of the Blast In the theoretical model the strength of the blast is the last major component that contributes to the damage inflicted and the casualties generated from a blast event. The strength of the blast will be dependent on the material used and, as previously discussed, the delivery mechanism (i.e. how much material can be carried/deliver ed to the target). Identifying the material that may be used is a difficult proposition. Most of the published literature limits the specifics concerning the explosives used and means of transportation, to estimated strengths of the explosives in TNT equivalents and a tr uck as the delivery mechanism. With a few notable exceptions, published literature does not identify the specific types of explosives used in many of the blas t events. Those that have been identified have not been caused by exotic, high tech ex plosives. The blasts that destroyed the barracks in Beirut and the federal building in Oklahoma City were comprised of ammonium nitrate (fertilizer). In the case of Beirut, th e equivalent of 12,000 pounds of TNT and in Oklahoma City, 1,800 kilograms (equivalent to 4,000 pounds of TNT) of ammonium nitrate soaked in fuel oil (Slate r, 1997). A US Army Corps of Engineers report listed the material used in the Khobar Towers attack as 20,000 pounds of TNT equivalents in a tanker truck filled with raw sewage (Thompson, 2004). A subsequent government report stated the tanker truck was loaded with at least 5,000 pounds of plastic explosives (Department of Justice, 2001). The resulting blast was then comparable to 20,000 pounds of TNT (Department of Justice, 2001). The bomb in the 1993 attack on

PAGE 121

112 the W orld Trade Centers consisted of 1,300 pounds of urea pellets (fertilizer additive), nitroglycerin, sulfuric acid, aluminum azi de, magnesium azide, and bottled nitrogen. Finally, the 2001 attack on the World Trade Centers utilized jet fuel and its combined effect with the impact of the jet ai rliners on the building. Although the literature does not readily id entify the explosives used in many of the blasts, it does indicate that many comm only (easily) found ingredients can be combined to produce an explosive. Presumab ly enough of any of these ingredients could generate a large scale blast event. Appe ndix 17 lists those materials identified by the Department of Justice, Bureau of Alc ohol, Tobacco, Firearms and Explosives as explosives. This list is labeled as comprehensiv e but not all-inclusive. This list is very extensive and points out the wi de range of possible ingredie nts that could be used to cause a blast. Although the list is very large, experience thus far i ndicates a large scale blast would require a large volume of explos ives and the means to carry them. The experience also indicates some form of plastic explosives or ammonium nitrate bomb are the most likely explosives to be used to generate a large scale blast.

PAGE 122

113 Results for Objective 2 : This objective was to develop the injury prediction com ponent of the computer simulation model. This is the portion of the computer simulation model that predicts the mix of specific injuries (amongst a defined injure d survivor population) that would result from a large scale blast event. Note that the prediction of the mix of specific injuries is done via the simulation methodology/model. The simulation methodolo gy provides for the assignment of the specific injuries to each individual injured su rvivor as they proceed through this portion of the model. Each individual injured surv ivor has the same chance (probability) of being assigned each specific injury category. Each injured survivor therefore has the chance of being assigned multiple injuries. There is also the chance that the injured survivor could be assigned no injuries. By definition (i.e. injured survivor) this should not happen, but the nature of the simula tion methodology and the assigned prediction parameters leave it as an option. The inju ry assignments are made for each individual survivor rather than in aggregate for the en tire population. Each individual survivor is assigned a unique set of injuries. Although the prediction parameters are the same for each individual survivor, the nature of the simulation methodology is that each individual survivor is treated as an indi vidual experiencing injuries and requiring care, rather than as part of a larger, aggregate population. The benefit being that the individual injured survivor carries the assigned injury categor ies through the entire si mulation model, thus driving care processes based on the specific injuries. The analysis of available data/studies resulted in the development of two models in an attempt to address the shortcomings of applying the data to a truly large scale mass casualty event. The two models are displayed in flowchart format in Appendices 18 and

PAGE 123

114 19. An abbreviated representation of the m odel found in Appendix 18 is found in Figure 10. Each of these models behaves and is cons tructed exactly alike. The difference is in the injury categories and decision parameters used in the simulation model. Each decision or determination is based on the ch ance or probability of the decision statement being true, e.g. the injured survivor has a seve r injury. If true, th at particular injured survivor is assigned the injury category or characteristic. Tables 19 and 20 display the sequential decision points and decision parame ters (probabilities) for each model. The output of both models feed into the resource prediction model and as such, all injury characteristics for each injured survivor are assigned at the beginning of the overall simulation process. These characteristics provide the information that the resource prediction portion of the model requires to make resource utilization decisions with respect to each individual injured survivor. Each injury prediction model assigns three categories or characteristics to the injured surv ivor; injury severity, hospitalization status (inpatient care is required or not ), and injury location and/or the nature of the injury. The first model (Figure 10 and Appendix 18) takes an expedient (practical) route and applies injury categories and associated parameters from a single study, Frykbergs 1988 publication. This model presents greater limita tions in terms of in jury categories, but better quality of data matching selected categories. This model utilizes the injury categories and parameters presented by Frykberg in his 1988 study. This study was assessed to be the best single representation of blast injuries because of its inclusion of 220 blast events and nearly 3,000 injuries. Th e resulting model, presented in Figure 10 and Appendix 18, sequentially steps through nine decision points. These decision points and associated decision parameters are found in Table 16. In order, the decision points

PAGE 124

User Defined Input Minor injury severity (p=0.813) Hospitalization (p=1 for severe, p=.14 for minor injury) Presence of head injury (p=0.314) Presence of lung injury or blast lung (p=0.006) Presence of abdomen injury (p=0.014) Presence of burn injury (p=0.05) Presence of traumatic amputation (p=0.012) Presence of bony extrem ity injury (p=0.109) Presence of soft tissu e injury (p=0.554) En d Figure 8 Injury Prediction Algorithm Notes: 1. User Defined Input re presents the user input to the model. In this case, this input is the users determ inati on of the total number of injured survivors to be considered for the study/simulation. 2. Each box where a determination is made in Figure 1 represents a decision module in the comput er simulation model. Therefore, each determination is a decision, either the statem ent is true, e.g. the injured survivor has a severe injury, or the statement is false. The decision is based on the assigned decision parameter or probability that the decision statem ent is true (the injury is present/will be assigned). The decision parame ter is the probability that the statement is true and is randomly assigned to each injured survivor as applied by the simulation program. 115

PAGE 125

Table 16 Injury Prediction Model (Frykberg Data, 1988) Decision Module Probability of experiencing the injury Minor Injury Severity 81.3% Hospitalization (severe) 100% Hospitalization (minor) 13.93% Head Injury 31.4% Blast Lung 0.6% Abdomen Injury 1.4% Burned 5% Traumatic Amputation 1.2% Bony Extremity 10.9% Soft Tissue Injury 55.4% determ ine injury severity, hospitalization status, the presence of a head injury, blast lung, abdomen injury, burns, traumatic amputati on, bony extremity injury, and soft tissue injury. Subsequent to the a ssignment of injury severity and hospitalization status, the model uses the injury categor ies and parameters as presented by Frykberg. Appendix 20 matches the injury categories presented by Frykberg to the corresponding Barell Matrix categories and there subordinate ICD-9-CM code s to describe the range of injuries that may be experienced within each category. It is important to note that the range of potential injuries is an extr apolation from the categories used by Frykberg. No direct correlation exists between the Frykberg study an d the specific injuries included in the 116

PAGE 126

117 Barell Matrix. The injury severity and hospitalization decision param eters were extrapolated from other available studies (included in the Studies Supporting Computer Model Development section of the literature review). Each deci sion point assigns the specific injury based on a random assignment co rrelating with the value of the parameter assigned to the decision module. For example the chance of an injured survivor having a head injury is 31.4%. All injured survi vors proceed through th e decision module and face a 31.4% chance of being assigned the head injury characteristic. This parameter does not represent a cumulative total, but ra ther the chance of having the injury faced by each individual injured survivor. Each injured survivor also proceeds through each subsequent decision module and faces corr esponding chance of having a new injury category assigned. Therefore, each injured survivor has the chance of being assigned multiple injury categories and conversely, al though categorized as an injured survivor, the chance exists that the injured survivor will be assigned no inju ries in the assignment process. To paraphrase using the head injury example, the simulation algorithm randomly assigns the head injury category with a probability of 0.314. Conversely, the probability of an injured survivor not experiencing a head injury is 0.686. The second model reaches for a more op timal solution in terms of injury categories by utilizing the nature of injury categories from the Barell Injury Diagnosis matrix and assigns parameters from existing data /studies as it best fits the categories. This model offers a more robust presentation of injury categories but is limited by data to support the assignment of the decision parame ters to each category. This model is presented in standard flowchart form in Appe ndix 19. This model util izes the nature of injury categories as presented in the Barell Matrix to form the sequential decision points

PAGE 127

118 f or the assignment of injuries to the injured survivors. These decision points and range of associated decision parameters as presented in published studies are included in Table 17. The assignment of parameters to this mode l was not as clean as for the first model because no single data source addresses each of the injury categories, and no two data source address the same categories. A mix of parameters from the available studies was used to assign the parameters to the differe nt injury categories. Where multiple studies addressed the same category, a range has been provided. Appendix 8 provides a full list of ICD-9-CM codes/categories, matched to ea ch of the Barell Matr ix injury categories that may result from this type of injury. In the absence of definitive data sets addressing the injuries resulting from a blast produced large scale mass casualty event, thes e two models represent initial steps toward a more comprehensive simulation/prediction model. A more comprehensive model for predicting the injury mix would address each of the categories and sub-categories (body region and nature of injury) in the Barell Matrix. Appendices 9 and 10 display the primary categories by body regi on and nature of the inju ry correlated with the corresponding categories utilized in the studies available for this research. Although seemingly well populated, a close review reveals many shortfalls in assigning parameters to the primary and secondary categories listed in the matrix. The output of this portion of the pred iction model is twofold. First, each individual injured su rvivor is assigned injury characteristics such as severity, hospitalization status, and an injury category or categories. These assignments are simulated according to the probability assigned to each injury category or characteristic.

PAGE 128

Table 17 Injury Prediction Model (Nature of Injury categories) Injury Ca tegories Pr obability of Occurrence Minor Injury Severity 81.3% Hospitalization (if severe) 100% Hospitalization (if minor) 13.93% Fracture 1 6% Dislocation 1 6% Sprains & Strains 13.5 25% Internal 0.6 49% (yes) Open Wound 55.4 94.8% (yes) Amputation 1.2% Blood Vessels NS Contusion/Superficial 12 94.8% Crush 1% Burns 2 5% Nerves NS* Unspecified NS *NS indicates no studies reviewed for th is research included this category. W ithin the construct of the computer model, each injured survivor has the potential to be assigned from 0 to 7 injury categories. The second output of this por tion of the model is a cumulative total of the number of each injury category to include severity and hospitalization status assi gned through the duration of the simulation run. 119

PAGE 129

120 The user has the opportunity to provide i nput or modify this model in two ways. First, the user identifies the total number of injured survivors to be studied. The computer model and simulation can accept any number of injured survivors. Secondly, the user has the ability and/or opportunity to modify the injury assignment parameters (probabilities), either to matc h another study or data source or because of assumptions in terms of the expected outcomes of a blast. In both cases the user is able to modify the simulation to better represent the scenar io they desire to prepare for.

PAGE 130

121 Results for Objective 3 : This objec tive was to devel op a computer simulation model composed of the medical assets and proce sses needed to respond to a blast producing a large scale mass casualty event. Research identified the va rious processes and resource units that this injured population would re quire, and where available, the process parameters (processing times) for each process. Additionally, the associated decision points were identified. Th e decision points determine how the injured survivor progresses through the model (i.e. the path the injured survivor follows through the simulation model) and when and/or if a certa in process is required or accessed. Where available, decision parameters (probabilities ) were identified in pub lished studies. These processes, resource units, and decision point s were used to construct the simulation model and are displayed as se quential algorithms in Figures 11-17. While figures 11-17 provide the basic structure and flow of the simulation model, appendix 21 provides a detailed flow chart depicting the entire simulation model. Appendix 22 compliments appendix 21 with a brief description of each component of the simulation model to include assigned parameters. The remaining narrative in this chapter describes the algorithms found in Figures 11-17 and provides a detailed description of each of the structural components of the simulation m odel; process modules, resource units, decision modules, and termination modules. Figures 11 and 13-17 are not displayed in fo rmal flow chart form (the formal form is found in Appendix 21) but rather seek to depict the structure and components of the simulation model in simplified form. In all cases the elements found in Figures 11 and 13-17 can be linked to the formal flowchar ts in Appendix 21, the descriptions in Appendix 22, and the narrative descripti ons found throughout th is chapter by the

PAGE 131

122 nom enclature found in parenthesis at the beginning of each component description. (D.) indicates a decision point (dec ision module), (P.) indicates a process (process module), and (T.) indicates a termination point (ter mination module). Additionally, descriptions have been provided for each resource unit. As has been previously discussed, each decision is based on a probability that the deci sion statement is true. Each process is supported by one or more resource units and has assigned parameters and distribution to determine the processing time required fo r the individual injured survivor. The distribution and designated parameters are presented in the following format: [distribution (time unit, minimum, most likely, maximum, mean)]. Finally, each termination module represents the point at which an injured survivor exits the simulation because they no longer require care with in the construct of the model. Figure 9 displays the porti on of the model that addresses the initial contact by the response system with the injured survivor. Th is portion of the simulation model receives input in the form of injured survivors from the Injury Pred iction portion of the model. Decision modules are used to determine if the injured survivor is trapped (D.1), if they are mobile (D.2), the means of transportation (D.3), and the destination for initial care; a casualty collection point (CCP) in the field or a hospital (D.4 D.7). Processes are included to provide for the rescue of an injure d survivor (P.1) if th ey are trapped and to provide transportation to the de signated point of care; either the CCP or a hospital (P.2 P.5). As has been previously discussed, decisions made in the decision modules are based on assigned probabilities of an outcome o ccurrence; this probabili ty is either based on research results found in published studies or simply researcher defined parameters. The actual process time for the rescue or the actual transportation time experienced by an

PAGE 132

(D.1) The Injured Survivor is trapped. (p=0.2)* *Interpretation; 20% of injured survivors will be trapped. (D.2) The Injured Survivor is mobile. (p=0.8) (P.1) Rescue process initiated [triangular (5, 10, 60, 25)] N o Yes (D.3) The injured survivor is transported by informal means. (p=0.8) Rescue process completed. Travel will be by formal means. (D.4 D.5, D.6, & D.7) Travel destination is the hospital. If no, travel destination is the casualty collection point (CCP). Probabilities: D.4 (p(hospital)=0 .2), D.5 (p(hospital)=0.8), D.6 (if injury sever, then hospital), D.7 (p(hospital)=0.8) (P.2, P.3, P.4, & P.5) Transportation processes to the hospital or CCP. Distributions and processing times for each means of travel and destination: P.2=triangular (3, 5, 20, 9.33), P.3=triangular (10, 20, 60, 30), P.4=triangular (3, 5, 30, 12.67), P.5=triangular (5, 20, 90, 38.33) Injured survivor(s) arrive at CCP. Injured survivor(s) arrive at hospital. CCP N o Yes to D.7 N o to D.4 To D.6 Yes to D.5 P.2, formal to CCP P.3, formal to hos p ital P.4, informal to CCP P.5, informal to hos p ital Hospital Start; injured survivors received from injury assi gnment model. Figure 9. Resource Prediction Alg orithm (Rescue/Transportation Component) 123

PAGE 133

124 individual injured surv ivor is established by the random assignment of a processing time based on the parameters and distribution a ssigned to the process. Again, assigned parameters are based on research results f ound in published studies or are researcher defined when no appropriate information existed in the literature. Th e end result of this portion of the model is for the in jured survivor to be directed to one of the care locations; CCP or hospital, and for data to be collect ed in terms of number s entering each process (or total number requiring the specific servic e) and total (or cumulative) time of care or service provided to the injured survivors requiring that particular process. Figure 9 Component Descriptions; decision modules, process modules, and resource units. See Figure 10 for descriptions of each type of module. D.1: This decision module determines if th e injured survivor is trapped as a result of the blast and thus requires the utilization of resources to be extracted from their trapped position and access a pa th to receive health care. If not trapped, the injured survivor can directly access a path leading to health ca re. (Decision Statement = The Injured Survivor is trapped.) D.2: Decision module D2 determines if th e injured survivor is mobile. If mobile, the injured survivor can proceed to access an ap propriate level of care. If not mobile, the injured survivor requires assistance to access car e, thus a transportation resource will be required for movement. (Decision Statement = The injured survivor is mobile.) D.3: Decision point D.3 determines if the injured survivor will be transported by informal or formal means. If informal, an informal transportation resource will transport the injured survivor to the location where ca re will be accessed. If formal, a formal transportation resource unit will be required to transport the injured su rvivor for medical

PAGE 134

Termination Points : The termination points represent the location at which an injured survivor exits the model. This is the point where the injured survivor either no longer needs medical care, or the care that is required is not included as a component of the simulation model. Resource Units : The resource units represent the cumulative resources in terms of personnel, equipment, and supplies that provide the care or perform the work identified in each process module. The exact composition of each resource unit is not identified. Process Modules : The process modules represent the types of work or activities that are performed to find, retrieve, transport, and care fo r the injured survivors. Each process is part of the overall system needed to effectively care for the injured surviv ors of a large scale mass casualty event. Decision Modules : The decision modules utilize the assigned decision parameters to make decisions concerning the path followed and char acteristics possessed by each injured survivor. Each decision module is necessary to process th e injured survivor through the model to a point of termination for the simulation. (The statement in parenthesis at the end of each description is the actual true/false statement included in the model that the decision parameters address.) Figure 10. Simulation Module Descriptions care. (Decis ion Statement = The IS travels by informal means.) D.4: Decision point D.4 determines if the injured survivor being transported by formal means (identified at decision point D.3) is taken directly to the hospital to access medical care or if they are taken to a triage and medical care site in the field for initial assessment and care. If true, the injured survi vor is transported direc tly to the hospital. (Decision Statement = The injured survivor travels by formal means to the hospital.) D.5: Decision point D.5 determines if the injured survivor being transported by informal means (identified at decision point D .3) is taken directly to the hospital or if 125

PAGE 135

126 they are taken to a triage and m edical care site in the field for initial assessment and care. (Decision Statement = The injured survivor tr avels by informal means to the hospital.) D.6: Decision point D.6 determines if th e trapped injured survivor (identified at decision point D.1), who is now rescued, w ill be transported by fo rmal means to the hospital. An assumption is made that all trapped injured survivors who experience the rescue process will then be transported via a formal transportation resource. The alternative to being transporte d to a hospital is that the fo rmal transportation resources will transport the rescued injured survivor to a field triage and car e location. (Decision Statement = The rescued IS travels by formal means to the hospital.) D.7: Decision point D.7 determines if th e mobile injured survivor (identified at decision point D.2) is transported by informal means (identified at de cision point D.3) to the hospital or the field assessment and tr eatment location. (Decision Statement = The mobile IS is transported by informal means to the hospital.) P.1: Rescue. The rescue process accounts for the effort expended to retrieve injured survivors who have been trapped in a collapsed structure. This process is not intended to represent the extended effort to find unaccounted for persons after a disaster event. Rather, it is likely that a certain number of injured survi vors will not be easily accessible and will thus require the utilization of special or dedicated resources to remove them from the damaged structure. These in jured survivors will be easily noticeable and accessible and will thus not require an extended effort to find them, but will require effort to be extracted from their location. Unique skills and equipment may be required to carry out this rescue operation.

PAGE 136

127 P.2: Formal transportation process to the casualty collection point (CCP) in the field. This process represents the effort to transport an inju red survivor to the CCP in the field. Time, distance, and severity of inju ry considerations will influence travel and destination options. At minimum the establis hment of a CCP in the field will provide a central location to gather injured survivors prio r to being transported to a hospital. In its most robust form the CCP will provide care and treatment in the field as appropriate and necessary. This process represents the use of formal transportation assets like ambulances and other emergency response vehicl es to transport inju red survivors from the blast site and point of injury to the CCP The local emergency response system is expected to provide oversight a nd control of the emergency respons e resources. P.3: Formal transportation to hospital from the field. This process represents the effort to transport an injured survivor from a location in the field near the blast site or point of injury to a hospital. This process represents the use of the formal transportation assets described in P.2 to transport injured survivors directly from the point they meet the formal transportation asset to the hospital. This process also utilizes the Emergency Response Resource. P.4: Informal transportation to CCP in the field. This pr ocess represents the informal process of transporti ng injured survivors from a point of injury to a CCP in the field. The large number of injured survivors de signated for this research is expected to mimic the experience of other large scale disasters. A portion of all injured survivors will find their way to health care resources by informal means. This informal means will consist of traveling by foot, being carried by those who are not injured, and being transported by vehicles; cars, truck, and/or buses. The Info rmal Transportation Resource

PAGE 137

128 in this p rocess carries the injured survivor to the CCP in the field. The informal transportation resources act on their own with limited or no oversight and control. The inclusion of an informal transportation res ource/process was done to account for the nonformal means many injured survivors will use to access a source of medical care. P.5: Informal transportation to ho spital. Similar to P.4 the Informal Transportation Resource will carry the injured su rvivor from the blast location directly to a hospital. Limited or no cont rol is exerted over these resources and the destinations they choose. Rescue: The Rescue Resource Unit is co mprised of the professionally trained and equipped personnel who carry out search and rescue operations. A single Rescue Resource Unit consists of the required pers onnel and equipment to carry out rescue operations for one injured survivor. Emergency Response: The Emergency Response Resource unit consists of professional trained personnel with appropria te vehicles and equipment to transport critically injured survivors. This resource unit typically represents an ambulance and the crew that operates the ambulance but coul d also include other types of emergency response vehicles and crew that could be used to transport an injure d survivor or injured survivors. The Emergency Response Res ource unit consists of the personnel and equipment required to transport a single injured survivor. Informal Transportation: The Informal Transportation Resource consists of untrained personnel and transport vehicles th at are available at the time and at the location of the blast. Transportation could be by foot, in the hands or on the shoulders of other injured or uninjured survivors, or in a personal vehicle availabl e at the blast site.

PAGE 138

129 Figure 11 displays the casua lty collection point (CCP) por tion of the model. This portion describes the variety of health care ac tivities that occur in the field prior to transportation for definitive care (hospital). The injured survivor first enters the triage process where they are triaged (P.6). The triage process will culminate in a decision determining whether the injuries are severe or minor (D.8). This particular decision is based on the severe or minor injury seve rity assignment made during the injury assignment portion of the simulation model. Based on the severity assignment, the injured survivor enters a path for either those with severe injuries or those with minor injuries. If severe, the next decision module determines if the injured survivor requires stabilization before being transported to a hosp ital (D.9). If stabilization is required, the injured survivor enters the st abilization process (P.7) and, when complete, enters a formal transportation process with the destination be ing a hospital (P.9). If stabilization is not required, the injured survivor proceeds directly to the formal transportation process (P.9) where they are transported to a hospital. As has been previously discussed, each of the decisions made in the decision module in this portion of the simulation module are based on the probability of a true out come or the identification and segregation of a previously assigned characteristic. The process times for each required process to be executed are based on randomly generated process times selected from the parameters and distribution assigned in each process module. For those in jured survivors determined to have minor injuries, the next step is the determination (decision module) if field treatment is adequate or if transportation to the hosp ital is required (D.10). If fiel d treatment is deemed to be adequate they next enter the fi eld care process module (P.8). If field care is not adequate,

PAGE 139

(P.6) Triage process [triangular (seconds, 5, 20, 90, 38.33] CC; from Figure 2 (D.8) Injured survivor has minor injury. (If minor injury, then minor injury path) (D.10) Field treatment is appropriate/adequate. (p=0.8) (D.9) The injured survivor requires stabilization before transport. (p=0.2) (P.7) Field Stabilization Process [triangular (minutes, 11, 21, 38, 23.33)] (P.8) Field care process. [triangular (minutes, 11, 21, 38, 23.33)] (P.9) Formal transportation to hospital. [triangular (minutes, 10, 20, 60, 30) (D.11) No hospital required. (p=0.86) (D.12) No additional care required. (p=0.8) (T.1) Termination; field care adequate, injured survivor released from care process. Hospital N o Yes Yes N o Yes Yes N o Yes N o N oFigure 11. Resource Prediction Algorithm (CCP Component) 130

PAGE 140

131 the injured survivor will next enter the for m al transportation process module (P.9). Once field care is complete, the injured survivor wi ll enter a sequence of decision modules to determine if additional stabilization and then hospitalization is require d (D.11), or if field care was adequate (D.12). In each case, if hos pitalization is required the injured survivor will move into the formal transportation process (P.9) to be transported to the hospital. If field care was determined to be adequate, th e injured survivor will enter the termination module where they will exit the system (T.1). As was previously described, data is collected to account for the total demand for each type of care or service as well as the cumulative process (service) time provi ded within each process module. Figure 11 Component Descriptions; d ecision modules, process modules, resource units, and termination module. See Figure 10 for descriptions of each type of module. D.8: Decision point D.8 addresses only those injured survivors who were not trapped, were not mobile, and were transpor ted by formal means to the field casualty collection point. This decision point separates those who were identified in the injury assignment subcomponent of the model as havi ng non-critical injuri es from those with other than non-critical injuri es (critical injuries). (T he IS has a minor injury.) D.9: Decision point D.9 considers only those identified at de cision point D.8 as having critical injuries and determines if they require field stabilization prior to being transported by formal means to the hospital. The alternative is that the injured survivor does not require stabilization prior to tran sport to the hospital. (The IS requires stabilization in the field.)

PAGE 141

132 D.10: Decision point D.10 considers only those identified at decision point D.8 as having non-critical injuries and determines if field treatment is adequate to meet their needs. The alternative is that field treatment is not adequate and care in a hospital setting is required. (Field treatment is ade quate for the injured survivor.) D.11: Decision point D.11 determines if hospitalization is required for those injured survivors who received fi eld treatment. The alternative is that hospitalization is not required. (The injured survivor do es not require hospita lization.) D.12: Decision point D.12 determines if field treatment is adequate to meet the needs of those who had non-critical injuries a nd received care in the field. If field treatment is adequate, the inju red survivor proceeds to term ination point T.1. If field treatment is not adequate, the injured survivor will be transported to the hospital. (The injured survivor requires no additional care. Field treatment was adequate.) P.6: Field CCP Triage. Triage is the process of assessing and sorting injured survivors as they arrive at a central access point to a health care resource. This assessing and sorting is utilized to make the best use of limited health care resources. In this case the injured survivors are triaged as they arrive at the CCP. This triage process utilizes the triage resource. Once the triage process is complete, the injured survivors will either receive care in the field (if field care resources are availabl e) or be transported to a hospital. In both cases the priority for trea tment or transportation in relation to other injured survivors will be based on the assessment made during the triage process. P.7: Field Stabilization. Many injure d survivors will require care beyond the capability available in the field. In more se vere cases these injured survivors will require some form of stabilization prior to transporta tion to a hospital. This process represents

PAGE 142

133 the stabilization that takes place p rior to transportation for definitive care. The stabilization process is accomplis hed by a Field Care Resource. P.8: Field Care. When local resources and time allow, field care assets will be utilized to care for the less cr itically injured survivors. Th e primary purpose of field care is to keep the less critically injured out of the hospital and thus allow the hospital resources to focus on the more severely injured. Field care resources can be used to care for those with less severe injuries. This proc ess utilizes the same Fi eld Care Resource as does the field stabilization process. P.9: Formal transportation to hospital. This is the same process as is found in P.3 except that the injured survivor has received some form of fi eld care or stabilization prior to being transported to the hos pital. Once again, the formal transportation resource is used to provide transportation. Triage: The Triage Resource consists of a triage officer and support personnel to document and assists with the triage process as the injured survivors arrive at the triage location; field or hospital. No unique equipmen t or supplies are requir ed as part of this resource unit. The triage resource provides triage fo r one injured survivor at a time. Field Care: The Field Care Resource unit represents the personnel, equipment, and supplies present in the field to care for injured survivors. This resource unit does not imply there is a field structure where care is provided. A structure could be present, or the care could be provided in a structure of opportunity or in the open air. Each resource unit represents that which is required to care for one injured survivor. Emergency Response: The Emergency Response Resource unit consists of professional trained personnel with appropria te vehicles and equipment to transport

PAGE 143

134 critica lly injured survivors. This resource unit typically represents an ambulance and the crew that operates the ambulance but coul d also include other types of emergency response vehicles and crew that could be used to transport an injure d survivor or injured survivors. The Emergency Response Res ource unit consists of the personnel and equipment required to transport a single injured survivor. T.1: T.1 represents the first termination point. Injured survivors who exit the simulation model at this point received care via the Field Ca re module (P.8) utilizing the Field Care Resource unit. The subsequent decision module (D.12) determined that no more care was required; therefore the injured su rvivor exits the health system (simulation model) at this termination point. Figure 12 describes the hospital component of the resource prediction model. This component of the model receives the inju red survivors that were determined (Figure 4) to require more definitive care in a hospital, and those who bypassed the CCP and were transported directly to the hospital (F igure 11). Upon arrival at a hospital, the injured survivor first enters a triage process (P.10). Upon ex iting the triage process, the injured survivor enters a decision module that determines if they have a severe or minor injury (D.13), and then directs them as appr opriate. For those with severe injuries, a determination is made if the injured survivor is stable (D.14). Separate trauma care processes are available for the stable (P.12) a nd unstable (P.13) injured survivors. Based on the determination made in the decision module (D.14), the injured survivor would enter one or the other of the trauma care pro cess modules. Subsequent to the trauma care process is the imaging study process (P.15) a nd then a decision module to determine if the injured survivor requires su rgery (D.16). If surgery is required, the injured survivor

PAGE 144

( P.10 ) Hos p ital tria g e p rocess [ trian g ular ( seconds 5 20 60 28.33 )] (D.13) Minor injury(ies) (If minor injury category assigned, then yes) (P.11) Emergency Department Care Process. [triangular (minutes, 22, 41, 116, 59.67)] (D.14) The injured survivor is stable. (p=0.73) (P.12) Trauma Care Process (stable). [triangular (minutes, 14, 47, 89, 50)] (P.13) Trauma Care Process (unstable). [triangular (minutes, 3, 22, 52, 25.67)] (D.15) Imaging study is required. (p=0.69) (P.14) Initiate imaging study process. [triangular (minutes, 11, 19, 58, 29.33)] Hospital (2) (P.15) Imaging Study process. [triangular (minutes, 11, 19, 58, 29.33)] (D.16) Surgery is required. (p=0.17) N o Yes Yes N o N o N o Yes Yes Surgery Hospital; from Figure 11 Figure 12. Resource Prediction Al gorithm (Hospital Co mponent) 135

PAGE 145

136 continues to the surgery com ponent of the simulation model (Surgery). If surgery is not required, the injured survivor continues to th e hospitalization portion ( hospital (2)) of the simulation model. Injured survivors with minor injuries (D.12) progress to the emergency department care process (P.11). This care process is followed by a decision module that determines if an imaging study is required (D.15). If im aging is required the injured survivor enters the imaging study pr ocess (P.14). The injured survivors then proceed to the second hospital component of the simulation model for additional disposition (Hospital (2)). Figure 12 Component Descriptions; decision modules, process modules, and resource units. See Figure 10 for descriptions of each type of module. D.13: Decision point D.13 immediately fo llows the initial tria ge of all injured survivors who have arrived at the hospital. This decision point separates those injured survivors who have non-critical injuries from those who have critical injuries. (Decision Statement = The injured survivor has only minor injuries). D.14: Decision point D.14 considers inju red survivors who have arrived at the hospital and have been identified as having critical injuries. This decision point determines if the injured survivor is stable. The alternative is that the injured survivor is not stable. (Decision Statement = The injured survivor has critical injuries and is stable). D.15: Decision point D.15 follows the emergency department care process for those injured survivors who a rrived at the hospital with non-critical injuries. This decision point determines if an imaging st udy is required. The alternative is that no imaging study is required. (Decision Stat ement = The injured survivor with minor injuries requires an imaging study).

PAGE 146

137 D.16: Decision point D.16 considers all pa tients identified with critical injuries. After initial trauma care has been provided, th is decision point determines if surgery is required. The alternative is that surgical car e is not required. (Decision Statement = The injured survivor re quires surgery.) P.10: Hospital Triage. Similar to P.6 (Field CCP Triage) this process represents the triage performed as injured survivors arrive at a hospital. As is described in P.6, this triage process assesses and sorts the injured survivors as they arrive at the hospital. The intent is to prioritize and match the injured survivors with the res ources that best meet their needs. Additionally, injured survi vors who require imme diate attention are separated from those who can wait to receive ca re. The triage process is performed by a Triage Resource unit. P.11: Emergency Department Care. This process represents the care provided in the emergency department to those injured surv ivors with minor injuri es. These injuries are likely to be made up of the minor soft tissue injuries and bony extremity injuries that do not require surgery or long term hospitaliz ation. This process is performed by the Emergency Department Resource unit. P.12: Initial trauma care for the stable injured survivor. This process represents the trauma care provided to th e injured survivor who arrives in stable condition. This process is performed by a hospital based Trauma Care Resource unit. P.13: Initial trauma care for the uns table injured survivor. This process represents the trauma care provided to the injured survivor who arrives in unstable condition. This process is performed by the same hospital based Trauma Care Resource unit as is discussed in P.12.

PAGE 147

138 P.14: Imaging for the injured survivor with minor injuries. This process represents the imaging or x-ray procedure requ ired for the injured survivor arriving in the hospital with minor injuries. Th is process is accomplished by an Imaging Resource unit. P.15: Imaging for the injured surviv or who was not stable. This process represents the imaging or x-ray procedure fo r the injured survivor who arrived at the hospital and required initial tr auma care to stabilize their condition. This process is carried out by the same Imaging Resource unit as is required for P.14. Emergency Department Care: The Emergency Department Resource unit represents the personnel, equipment, and suppl ies necessary to care for injured survivors in an emergency department setting. The Emergency Department Resource unit includes the space in an Emergency Department se tting. Each resource unit represents the resources required to care for one injured survivor. Trauma Care: The Trauma Care Re source unit represents the personnel, equipment, and supplies required to care for th e injured survivors requiring trauma care in a formal trauma setting. This will include sp ace in a designated trauma facility. Each resource unit represents the resources required to care for one injured survivor. Imaging: The Imaging Resource unit re presents the personnel, equipment, and supplies necessary to provide imaging suppor t to those injured survivors requiring an imaging study. This resource unit includes the space to provide the service. Each resource unit represents the resources requi red to care for one injured survivor. Figures 13 describes the surgery components of the resource prediction portion of the simulation model. Only those injured surv ivors determined to need surgery (Figure 5) enter this portion of the mode l. The first module in this portion of the model is a

PAGE 148

(D.17) Determination of which surgical specialty is required? (decision based on assessment of assigned injury categories) (P.16-23) Surgical care [triangular (minutes, 47, 97, 218, 120.67)] Surgical care is provided by the surgical specialist and team as determined in the previous decision module. (D.19) Intensive care is required. (p=0.20) (P.26) Intensive care [triangular (1440, 4320, 9640, 4800)] (P.27) Hospital care [triangular (minutes, 1440, 4320, 8640, 4800)] No Yes A surgical specialty is selected based on injury categories assigned to the injured survivor. Once selected the injured survivor enters the care process for the assigned surgical specialty. (T.3) Terminate; hospital care no longer required. Surgery; from figure 11 Figure 13. Resource Prediction Al gorithm (Surgery Component) 139

PAGE 149

140 decision m odule that determines the surgical specialty that is required (D.17). This decision is based on the type of injuries assi gned to the injured survivor in the injury assignment portion of the simulation model. The injured survivor then enters the appropriate surgical care process for the required surgical specia lty (P.16-23). Figure 16 provides greater detail for this portion of the model. Once the surgical care process is complete, a decision module is used to determ ine if intensive care is required (D.19). This decision is based on the assigned severity of injury. If intensive care is required, the injured survivor next enters the inpatient intensiv e care process (P.26). If intensive care is not required, the injured survivor enters the hospital care process (P .27). In either case, the injured survivor will complete their co urse of treatment/care and then exit the simulation model via a termination module (T.3) If no further hospital care is required, the patient enters the termination module and exits the simulation model. Figure 14 Component Descriptions; d ecision modules, process modules, resource units, and termination module. See Figure 12 for descriptions of each type of module. D.17: Decision point D.17 considers onl y those patients iden tified at decision point D.16 as requiring surgical care. This decision point determines the surgical specialist required to pr ovide care. This decision is de pendent on the type of injuries assigned to the patient in the injury assi gnment submodel. (Decision statement = The injured survivor requires a surgical specialist.) D.19: Decision point D.19 considers all pa tients identified with critical injuries. This decision point determines if intensive ca re is required in an intensive care unit. The alternative is that intensive care is no t required. (ICU care is required.)

PAGE 150

(P.18, 20 & 22) General Surgery Process [triangular (minutes, 47, 97, 218, 120.67)] (P.16 & 19) Cardioth oracic Surgery Process [triangular (minutes, 47, 97, 218, 120.67)] (P.17 & 23) Orthopedic Surgery Process [triangular (minutes, 47, 97, 218, 120.67)] (P.21) Neurosurgery Process [triangular (minutes, 47, 97, 218, 120.67)] (D.19) Intensive care is required (p=0.20) Yes No (P.26) (P.27) (D.17) Determination of which surgical specialty is required? (decision based on assessment of assigned injury categories) If head injury, then neurosurgery process. If chest injury, then cardiothoracic surgery process. If blast lung, then cardiothoracic surgery process. If abdomen injury, then general surgery process. If burn injury, then general surgery process. If traumatic amputation, then orthopedic surgery process. If bony extremity injury, then orthopedic surgery process. Otherwise (soft tissues injury), general surgery. Figure 14. Resource Prediction Al gorithm (Surgery Component) 141

PAGE 151

142 P.16-23: Surgical care. Depending on the type of injury experienced by the injured survivor, processes 16-23 represent surgical procedures carried out by the identified surgical specialties. This se quence includes duplicate specialties. These duplicate processes occurred because of lim itations in the process of building the structure of the model in the software. It is not significant for any reason. In each case, the duplicate process is supported by the same resource unit. The su rgical processes are performed by a Surgical Resource unit unique to each surgical specialty. P.16: Cardiothoracic Surgery Process P.17: Orthopedi c Surgery Process P.18: General Surgery Process P.19: Cardiothoracic Surgery Process P.20: General Surgery Process P.21: Neurosurgery Process P.22: General Surgery Process P.26: Intensive Care Unit (ICU) care. Th is process represents the care provided in an intensive care unit setting following an injured survivor rece iving trauma and/or surgical care in a hospital. This proce ss is performed by an ICU Resource Unit P.27: Hospital care following ICU care or su rgical care. This process represents the hospital care provided to th e injured survivors directly following either ICU care or surgical care. This care is provided by the same Inpatient Care Resource unit that provides care in P.25. Surgical: The Surgical Resource unit repr esents the personnel, equipment, and

PAGE 152

143 supplies n ecessary to provide surgical care to one injured survivor. This resource unit includes the operating suite and r ecovery space necessary to care for the injured survivor. A separate resource unit is used for e ach surgical specia lty; general surgery, cardiothoracic surgery, neuros urgery, and orthopedic surgery. ICU Care: The ICU Resource unit cons ists of the people, equipment, and supplies needed to care for a single injured survivor who requires extended care in an ICU setting. This resource unit will include physician and nursing time, an ICU bed, monitoring equipment, and supplies necessary to care for a single injured survivor requiring ICU care. Inpatient Care: The Inpatient Care Resource unit consists of the people, equipment, and supplies needed to care for a single injured survivor requiring extended care in a hospital set ting. This resource unit includes physician and nursing time, a hospital bed, monitoring equipment, and supplies necessary to care for a single hospitalized patient. Figure 15 describes the final components of the simulation model. This portion of the model determines the final disposition for the injured survivors with minor injuries. The first decision point determines if hospitaliz ation is required for injured survivors who received emergency department care for minor injuries (D.18). If hospitalization is not required, one final round of emergency departme nt care is received (t his care follows any required imaging study) (P.24). Once all em ergency department care is provided, the injured survivor enters the termination module and exits th e simulation model (T.2). If hospitalization is requir ed, the injured survivor enters th e hospitalization pr ocess (P.25).

PAGE 153

(D.18) Hospital Care is requir ed (if previously designated as requiring hospital care, then yes) Hospital (2); from figure 11 (P.24) Post-imaging emergency department care [triangular (minutes, 29, 53, 137, 73)] (P.25) Hospital care [triangular (minutes, 1440, 1440, 4320, 2400)] (T.2) Terminate; care no longer required. (T.3) Terminate; hospital care no longer required. No Yes Figure 15. Resource Prediction Algor ithm (Hospital (2) Component) When care is no longer required (the hospital process is com p lete), the injured survivor enters the termination module a nd exists the simulati on model (T.3). If unconstrained by time, the simulation run will be complete when all injured survivors have reached one of the three termination modules. Figure 15 Component Descriptions; d ecision modules, process modules, resource units, and termination module. See Figure 12 for descriptions of each type of module. D.18: Decision point D.18 considers only t hose patients with noncritical injuries who received care in an emergency department This decision point determines if hospitalization is required. The alternativ e is that hospitalizat ion is not required. 144

PAGE 154

145 (Decision S tatement = Hospitalization is requir ed for the injured survivor with a minor injury). P.24: Post-imaging emergency department car e. This process represents the care provided in the emergency department for in jured survivors with minor injuries who require an imaging (x-ray) study. This proc ess represents the a dditional care provided after the imaging study. This process is performed by the Emergency Department Resource unit. P.25: Hospital care following emergency department care. The process represents the hospital care (inpatient services) provided to the injured survivors who received emergency department care and then required at least one night of care in the hospital. This process is performed by an Inpatien t Care Resource unit. Imaging: The Imaging Resource unit re presents the personnel, equipment, and supplies necessary to provide imaging suppor t to those injured survivors requiring an imaging study. This resource unit includes the space to provide the service. Each resource unit represents the resources requi red to care for one injured survivor. Inpatient Care: The Inpatient Care Resource unit consists of the people, equipment, and supplies needed to care for a single injured survivor requiring extended care in a hospital set ting. This resource unit includes physician and nursing time, a hospital bed, monitoring equipment, and s upplies necessary to care for a hospitalized patient. T.2: Termination point T .2 represents the second termination point where an injured survivor may exit the simulation mode l. An injured survivor who exits the simulation model at this point has received all required medical care in the emergency

PAGE 155

146 departm ent setting. Additional care and/or hospitalization is not required (decision module D.18 in the injury assignment model). T.3: Termination point T.3 represents the final termination point for the simulation model. At this point, all required medical care has been provided to include ICU care and inpatient hospital care. At this point the patient either requires no additional care or the care requ ired is beyond the scope of this simulation model.

PAGE 156

147 Results for Objectives 4 and 5 : Objectives 4 and 5 were to run the sim ulation model to first predict the mix of injuries (defined by ICD codes) experienced by those who survive a large scale mass casualty event (objective 4) and to then predict the resources required to care for this injured popul ation (objective 5). Injury predictions are provided for a large scale blast producing 7,000 injured surviv ors and for a large scale blast producing 45,000 injured survivors. The simulations we re run with no constr aints (resources or time) to predict the total resources required to care for the injured survivors. Additionally, the impact of resource constraints is de scribed through two scenarios within each specified large scale blast event; a 12 hour time limit in a medical ly sparse geographic area, and a 12 hour time limit in a medically dense geographic area. Table 18 displays the data produced by the Arena software after 10 replications of the injury prediction simulation for 7,000 inju red survivors with no constraints. The injury categories are those that were previous ly described in the Results for Objective 2 section; abdomen, bony extremity, burned, ches t, head, lung, soft ti ssue, and traumatic amputation. The two severity categories are al so listed in the tabl e; minor injury, and severe injury. Appendix 20 li nks these injury categories to the corresponding Nature of the Injury Category and associated ICD-9-CM codes as presented in the Barell Matrix. For each injury category Table 18 lists the mean, half width, minimum, and maximum values as calculated by the Arena Software pr ogram. The mean represents the arithmetic mean calculated from the 10 replications for the particular injury category. The half width represents the 95 % confidence interval value, plus or minus, for the particular injury category based on the 10 replications; in 95% of the simulation replications, the prediction for that particular injury categorie s falls within the confidence interval. Min

PAGE 157

Table 18 Injury Predictions (7,000 inju red su rvivors, no constraints, 10 replications) Injury Category Mean* Half Width** Min*** Max**** Abdom en 102 5.9 93 119 Bony Extremity 761.3 20.3 699 788 Burned 349.9 8.2 334 371 Chest 137.4 6.7 123 152 Head 2,192.7 36 2,119 2,278 Lung 44.9 5.2 37 59 Soft Tissue 3,890.9 39.5 3,775 3,978 Traumatic Amputation 84.4 7.5 64 103 Minor Injury 5,704.9 25 5,642 5,764 Severe Injury 1,295.1 25 1,236 1,358 Mean: Arithm etic mean calculated from the 10 replications. **Half Width: in 95% of repeated tria ls, the sample mean would be reported as within the interval sample mean +half width. ***Min: The smallest average across all replications. ****Max: The largest average ac ross all replications. 148

PAGE 158

149 repres ents the single minimum predicted value fo r that particular injury category from the 10 replications. Max represents the single ma ximum predicted value for that particular injury category from the 10 replications. The results presented in Table 18 align closely with the decision parameters presented in Ta ble 16. This alignment is expected due to the nature of the decision process; the inju ry decisions and/or a ssignment for a specific injured survivor are assigned randomly one injured survivor at a time. Thus the percentage of injured survivors predicted to have a specific category of injury follows closely, but not exactly, th e assigned prediction parameters because on the random assignment. As an example to help interpret th e data, in Table 18 the Abdomen Injury Category has a corresponding Mean injury pred iction of 102 with a Half Width of 5.9. The Min is 93 and the Max is 119. The Mean of 102 indicates the arithmetic mean of the number of predicted abdomen injuries for th e 10 replications of the simulation was 102; 102 abdomen injuries are predicted to be found among the 7,000 injured survivors. There is 95% confidence that another simulation run will produce a predicted number of abdomen injuries between 96.1 and 107.9 (102 +/ 5.9). Within the 10 replications, the minimum predicted number of abdomen injuries was 93 and the maximum prediction was 119. Table 19 displays the data produced by the Arena software after 10 replications of the injury prediction simulation for 45,000 inju red survivors with no constraints. The injury categories and column heading duplic ate that presented in Table 18, the only difference being the total number of injured surv ivors utilized for the simulation. As with the prediction data for 7,000 injured survivor s, the injury predicti ons for 45,000 injured

PAGE 159

Table 19 Injury Predictions (45,000 inju red survivors, n o constraints, 10 replications) Injury Category Mean* Half Width** Min*** Max**** Abdom en 632.1 25 574 684 Bony Extremity 4,901.5 49.6 4,799 4,988 Burned 2,250 17.9 2,208 2,289 Chest 900.9 15.3 866 932 Head 14,084.2 69 13,964 14,248 Lung 266.5 35.5 230 285 Soft Tissue 24,943.2 12.5 24,840 25,082 Traumatic Amputation 543.6 11.9 511 570 Minor Injury 36,597.8 64 36,470 36,740 Major Injury 8,402.2 64 8,260 8,530 Mean: Arithm etic mean calculated from the 10 replications. **Half Width: in 95% of repeated tria ls, the sample mean would be reported as within the interval sample mean +half width. ***Min: The smallest average across all replications. ****Max: The largest average acro ss all replications. 150

PAGE 160

151 survivors closely resemble the decision param eters from Table 16. Tables 23 and 24 present the same categorie s of data as is found in Tables 21 and 22. Although labeled as Injury Predictions, in reality this tables represent the mix of injuries experienced by the number of injured survivors who enter the simulation within the constraints placed on the simulation. Tabl es 23 presents the resu lts of 10 replications of the simulation of 7,000 injured survivors with a 12 hour time constraint. Tables 24 presents the results of 10 repl ications of the simulation of 45,000 injured survivors with a 12 hour time constraint. In each case, 7,000 and 45,000 injured survivors, simulation runs were conducted with resour ce constraints; first with resource sparse constraints and then with resource dense constraints. Only the results of the simulation runs with resource dense constraints are displayed here. In each case, resource sparse and resource dense, the time constraint combined with th e speed in which the injured survivors entered the simulation, and not the resource constr aint, influenced the number of injured survivors to enter the system and be assi gned injury categories. Thus the results displayed in Tables 22 and 23 are similar and represent the number of injured survivors that entered the model (constrained by th e speed they entered the model, or, for simulation purposes, were created). The speed with which the injured survivors entered the simulation was not fast enough and the dur ation of the simulation not long enough to allow the difference in resources to impact the output/outcome. Because of this, the resource constraints did not impact the assi gnment of injuries nor the total number of injured survivors to enter the system. Had greater time been allowed and/or had the injured survivors entered the model more quickly, difference in the results would be seen. Tables 25 displays a comparison of the in jury prediction results from the three

PAGE 161

Table 20 Injury Predictions (7,00 0 injured survivors, 12 hours/resource dense, 10 replications) Injury Category Mean* Half Width** Min*** Max**** Abdom en 30.9 3.8 26 43 Bony Extremity 235.8 8.3 223 260 Burned 111.5 9.4 92 129 Chest 44.3 2.7 39 49 Head 673.6 17.6 625 703 Lung 14.1 1.9 11 20 Soft Tissue 1,212.9 25.2 1,147 1,266 Traumatic Amputation 26.6 1.9 22 31 Minor Injury 1,770.1 30.6 1,708 1,823 Major Injury 411.3 10.5 388 434 Mean: Arithm etic mean calculated from the 10 replications. **Half Width: in 95% of repeated tria ls, the sample mean would be reported as within the interval sample mean +half width. ***Min: The smallest average across all replications. ****Max: The largest average acro ss all replications. 152

PAGE 162

Table 21 Injury Predictions (45,0 00 injured survivors, 12 hours/resource dense, 10 replications) Injury Category Mean* Half Width** Min*** Max**** Abdom en 30.9 3.8 26 43 Bony Extremity 235.8 8.3 223 260 Burned 111.5 9.4 92 129 Chest 44.3 2.7 39 49 Head 673.6 17.6 625 703 Lung 14.1 1.9 11 20 Soft Tissue 1,212.9 25.2 1,147 1,266 Traumatic Amputation 25.6 1.9 18 33 Minor Injury 1,770.1 30.6 1,708 1,823 Major Injury 411.3 10.5 388 434 Mean: Arithmetic mean calculated from the 10 replications. **Half W idth: in 95% of repeated tria ls, the sample mean would be reported as within the interval sample mean +half width. ***Min: The smallest average across all replications. ****Max: The largest average acro ss all replications. 153

PAGE 163

Table 22 Comparison (7,000 injured survivors, 10 replication means) Injury Category Unconstrained 12 hour/dense 12 hour/sparse Abdom en 102 30.9 31.5 Bony Extremity 761.3 235.8 231.2 Burned 349.9 111.5 111.2 Chest 137.4 44.3 44.1 Head 2,192.7 673.6 688.6 Lung 44.9 14.1 15.1 Soft Tissue 3,890.9 1,212.9 1,206.3 Minor Injury 5,704.9 1,770.1 1,768.2 Major Injury 1,295.1 411.3 407.6 154

PAGE 164

155 sim ulations for 7,000 injured survivors. Tabl e 23 present a compar ison of the results from the three simulations for 45,000 injured su rvivors. As previously described, the comparisons display the impact of constraini ng the total time allowed for the simulation (12 hours) but display no impact of the associated resource constraints. Table 24 displays the data produced by the Arena software after 10 replications of the resource requirements prediction simu lation for 7,000 injured survivors with no constraints. This table displays the total number of the particular resource required to provide care to all injured su rvivors with no constraints in terms of time or resources. The resource categories are the same as thos e described in the Results for Objective 3 section; Thoracic Surgery, Emergency Depart ment, Field Care, Field Triage, Formal Transportation, General Surgery, Hospital, Hospital Triage, ICU, Imaging, Informal Transportation, Neurosurgery, Orthopedic Surgery, Rescue, and Shock/Trauma. As in Tables 21-24, the mean, half width, minimum, and maximum values as calculated by the Arena Software program are displayed for each resource categories. The mean represents the arithmetic mean calculated from the 10 replications for the particular resource category. The half width represents the 95% co nfidence interval value, plus or minus, for that particular resource category based on the re sults of the 10 replications; in 95% of the simulation replications, the prediction for that particular resource cate gory falls within the confidence interval. Min represents the single minimum predicted value for that particular injury category fr om the 10 replications. Max represents the single maximum predicted value for that particular inju ry category from the 10 replications. As an example to help interpret the data, in Table 24 the Thoracic Surgery resource category has a corresponding Mean re source prediction of 7.7 with a Half Width

PAGE 165

Table 23 Comparison (45,000 injured survivors, 10 rep lication means) Injury Category Unconstrained 12 hour/dense 12 hour/sparse Abdom en 632.1 30.9 31.5 Bony Extremity 4,901.5 235.8 231.2 Burned 2,250 111.5 111.2 Chest 900.9 44.3 44.1 Head 14.084.2 673.6 688.6 Lung 266.5 14.1 15.1 Soft Tissue 24,943.2 1,212.9 1,206.3 Minor Injury 36,697.8 1,770.1 1,768.2 Major Injury 8,402.2 411.3 407.6 156

PAGE 166

Table 24 Resource Requirement Prediction (7,000 injured survivors, no constraints, 10 replications) Resource Category Mean* Half Width** Min*** Max**** Thoracic Surgery 7.7 2 3 12 Emergency Department 8,308.1 59.1 8,150 8,412 Field Care 1,807.3 24.4 1,759 1,874 Field Triage 2,314 24.4 2,268 2,364 Formal Transportation 2,780.1 31.7 2,701 2,835 General Surgery 302.4 10.1 285 325 Hospital 2,090.4 30.3 2,003 2,138 Hospital Triage 5,846.8 19.7 5,794 5,893 ICU 257.7 8.9 235 281 Imaging 4,089.4 27.1 4,006 4,145 Informal Transportation 5,380.7 24.5 5,342 5,434 Neurosurgery 164.5 9.5 139 190 Orthopedic Surgery 38.3 3.8 27 45 Rescue 1,396.4 22.1 1,340 1,429 Shock/Trauma 1,295.1 24.7 1,236 1,358 Mean: Arithm etic mean calculated from the 10 replications. **Half Width: in 95% of repeated tria ls, the sample mean would be reported as within the interval sample mean +half width. ***Min: The smallest average across all replications. ****Max: The largest average acro ss all replications. 157

PAGE 167

158 of 2. The Min is 3 and the Max is 12. The Mean of 7.7 indicates the arithm etic mean of the 10 replications of the simulation was 7.7; rounding up, 8 thoracic surgery resource units are required to care for the predic ted thoracic surgery needs of 7,000 injured survivors. There is 95% confidence that another simulation run will produce a thoracic resource requirement between 5.7 and 9.7 (7.7 +/2). Within the 10 replications, the minimum predicted thoracic surgery resour ce requirement was 3 and the maximum prediction was 12. Table 25 displays the data produced by the Arena software after 10 replications of the resource requirements prediction simu lation for 45,000 injured survivors with no constraints. The resource cat egories and column headings duplicate those described in Table 24, the only difference in the tables is related to the total number of injured survivors processed within each simulation. Tables 29 and 30 display the results of the simulations when constraints are placed on the model. Note that although tit led as Resource Requirement Predictions, each of the constrained scenarios produces results that predict the number of injured survivors who required the the different care processes and who were able to access the process within the constraints placed on the simulation. Thus producing resource utilization estimates rather than true resource requirement predictions. Tables 29 displays the results after 10 replicati ons with 7,000 injured survivors with constraints of 12 hours and resource dense resource constraints. Table 27 displays the resu lts of 10 replications with 7,000 injured survivors, 12 hour time c onstraint and resource sparse resource constraints. Additional simulations were run with 45,000 injured survivors and the new constraint scenarios; 12 hours/ resource sparse and 12 hours/res ource dense. As with the

PAGE 168

Table 25 Resource Requirement Prediction (45,000 inju red survivors, no constraints, 10 replications) Resource Category Mean* Half Width** Min*** Max**** Thoracic Surgery 58.9 5.2 47 69 Emergency Department 53,357.5 144 53,040 53,632 Field Care 11,543 53.1 11,431 11,645 Field Triage 14,797 73.9 14,663 14,966 Formal Transportation 17,868.3 75.1 17,743 18,027 General Surgery 1,978.7 31.8 1,900 2,019 Hospital 13,498.9 76.3 13,294 13,650 Hospital Triage 37,629.3 67.1 37,468 37,763 ICU 1,653.5 33.5 1,576 1,721 Imaging 26,294.8 102.5 26,085 26,469 Informal Transportation 34,558.8 44.1 34,446 34,666 Neurosurgery 1,039.8 25.6 958 1,078 Orthopedic Surgery 236.9 10.8 213 261 Rescue 9,034.8 48.2 8,953 9,156 Shock/Trauma 8,402.2 64 8,260 8,530 Mean: Arithm etic mean calculated from the 10 replications. **Half Width: in 95% of repeated tria ls, the sample mean would be reported as within the interval sample mean +half width. ***Min: The smallest average across all replications. ****Max: The largest average acro ss all replications. 159

PAGE 169

Table 26 Resource Requirement Prediction (7,000 injured survivors, 12 hours/resou rce dense, 10 replications) Resource Category Mean* Half Width** Min***Max**** Thoracic Surgery 0.7 0.5 0 2 Emergency Department 460.4 7 446 475 Field Care 451.8 14.6 415 476 Field Triage 580.8 19.4 539 621 Formal Transportation 636 20.7 582 662 General Surgery 33.4 2.1 30 38 Hospital 164 5.4 147 174 Hospital Triage 1,636 29.3 1,556 1,695 ICU 26.6 2.4 22 33 Imaging 348.5 7.7 334 366 Informal Transportation 1,677.4 24.9 1,608 1,723 Neurosurgery 19 3.4 14 28 Orthopedic Surgery 2.9 1.1 1 6 Rescue 290.3 6.4 280 307 Shock/Trauma 161.4 3.6 154 167 Mean: Arithm etic mean calculated from the 10 replications. **Half Width: in 95% of repeated tria ls, the sample mean would be reported as within the interval sample mean +half width. ***Min: The smallest average across all replications. ****Max: The largest average acro ss all replications. 160

PAGE 170

Table 27 Resource Requirement Prediction (7,000 injured survivors, 12 hours/resou rce sparse, 10 replications) Resource Category Mean* Half Width** Min*** Max**** Thoracic Surgery 0 0 0 0 Em ergency Department 49.3 1.3 46 52 Field Care 121.8 2.2 117 125 Field Triage 340.7 9.6 326 370 Formal Transportation 122.9 4.4 114 133 General Surgery 4.1 1.3 2 7 Hospital 18.5 1.9 14 22 Hospital Triage 1,374.2 35.9 1,288 1,467 ICU 3 1.4 0 6 Imaging 41.5 1.6 37 44 Informal Transportation 1,669.5 35.4 1,596 1,756 Neurosurgery 1.7 1.1 0 4 Orthopedic Surgery 0.5 0.5 0 2 Rescue 30.1 2 24 33 Shock/Trauma 16.9 2 14 23 Mean: Arithm etic mean calculated from the 10 replications. **Half Width: in 95% of repeated tria ls, the sample mean would be reported as within the interval sample mean +half width. ***Min: The smallest average across all replications. ****Max: The largest average acro ss all replications. 161

PAGE 171

162 sim ulation to predict the mix of injury categ ories, the speed that the injured survivors enter the simulation impacted the results. In each case the output for each scenario with 45,000 injured survivors matched that produced w ith 7,000 injured survivors. Data tables were not included for these two simulation runs The that that have been included (29 and 30) display the impact in terms of the num ber of injured survivors that are cared for when the simulation model is constrained in different ways, similar to what might be found in a unique geographic area. Tables 31 displays the comparisons be tween the simulations for 7,000 injured survivors. The tables list each resource cat egory and the predicted resource requirement or utilization for the three scenarios; no cons traints, 12 hour time cons traint and resource dense resource constraints, and 12 hour time constraint and resource sparse resource constraints. Table 25 displays data categories identical to Table 24 but for simulations with 45,000 injured survivors. Appendices 21 through 27 describe additional data produced during the six simulation scenarios and related to the proces ses which provide the ca re and the access to those processes. This data affords the opportunity for additional analysis and, thus, more detailed planning on the part of a local planner. Data for all process categories as labeled and defined in the results section for Objec tive 3 and in Appendix 22 are included in the appendices. The additional da ta categories included in th e appendices are presented as they are labeled in the Aren a output and include the follo wing; number in, value added time, total time, accumulated time, queue waiting, and queue number waiting. Definitions for these categories are as follows: Number In (# in): The number of injured survivors who require service from the

PAGE 172

Table 28 Scenario Comparison of Predicted Resource Requirements (7,000 injured survivors, 10 replication means) Resource Category No Constr aints 12 hour/dense 12 hour/sparse Thoracic Surgery 7.7 0.7 0.0 Emergency Department 8,308.1 460.4 49.3 Field Care 1,807.3 451.8 121.8 Field Triage 2,314 580.8 340.7 Formal Transportation 2,780.1 636 122.9 General Surgery 302.4 33.4 4.1 Hospital 2,090.4 164 18.5 Hospital Triage 5,846.8 1,636 1,374.2 ICU 257.7 26.6 3 Imaging 4,089.4 348.5 41.5 Informal Transportation 5,380.7 1,677.4 1,669.5 Neurosurgery 164.5 19 1.7 Orthopedic Surgery 38.3 2.9 0.5 Rescue 1,396.4 290.3 30.1 Shock/Trauma 1,295.1 161.4 16.9 Note: Results of constrained scenarios produce predictions of resource utilization rather than resource requirements. Given the cons traints placed on the simulation, the listed value indicates the number of injured surv ivors requiring each process/service and who were able to access that service. 163

PAGE 173

Table 29 Scenario Comparison of Predicted Resource Requirements (45,000 injured survivors, 10 replication means) Resource Category No Constr aints 12 hour/dense 12 hour/sparse Thoracic Surgery 58.9 0.7 0 Emergency Department 53,357.5 460.4 49.3 Field Care 11,543 451.8 121.8 Field Triage 14,797.8 580.8 340.7 Formal Transportation 17,868.3 636 122.9 General Surgery 1,978.7 33.4 4.1 Hospital 13,498.9 164 18.5 Hospital Triage 37,629.3 1,636 1,374.2 ICU 1,653.5 26.6 3 Imaging 26,294.8 348.5 41.5 Informal Transportation 34,558.8 1,677.4 1,669.5 Neurosurgery 1,039.8 19 1.7 Orthopedic Surgery 236.9 2.9 0.5 Rescue 9,034.8 290.3 30.1 Shock/Trauma 8,402.2 161.4 16.9 Note: Results of constrained scenarios produce predictions of resource utilization rather than resource requirements. Given the cons traints placed on the simulation, the listed value indicates the number of injured surv ivors requiring each process/service and who were able to access that service 164

PAGE 174

165 particular process category and have entered or in itiated the process. As an example, in Appendix 23 Process Category P.11 (Emergency Department Care) had 4,552 injured survivors who required and accessed the emerge ncy department care process/resource. Value Added Time (VA Time): This is the arithmetic mean of the time each injured survivor accessing the partic ular process received care from the resource/process (presented in minutes). Utilizing the exam ple of Process Category P.11 in Appendix 23, the arithmetic mean of the time each injured su rvivor received care in the Emergency Department care process, or from the Emer gency Care resource was 59.6 minutes. Total Time (T Time): The total time an injured survivor spends in a specific process. This is the sum of the value adde d time and any time spent waiting in a queue. The Emergency Department process describe d in the previous paragraphs had no time listed for the queue waiting time, therefore the value added time and the total time are equivalent. There is no waiting time found in the data presented in Appendices 23 and 24 because there are no resource constraint s (i.e. available resources are unlimited). Whenever a resource is required a resource is available. Accumulated Time (A Time): The total tim e of service provided in/by a process. This is the cumulative sum of the value added time provided to each injured survivor that enters and exits the process. In Ap pendix 23 Process Category P.11 and its accompanying Emergency Care resource provid ed a total of 271,379.9 minutes of service to the 4,552 injured survivors that required care in the Emergency Department process. Queue Waiting (QW Time): The arithmetic mean of the time an injured survivor spends waiting for the specific service. Wa iting time is a function of the number of injured survivors requiring a se rvice, the time required to provide the service, and the

PAGE 175

166 num ber of resource units availa ble to provide the service. As was previously mentioned, Appendices 23 and 24 display data from simu lations with no resource constraints therefore there were no waiting lines or queues. In Appendix 25 the arithmetic mean of the time the injured survivor waited for emergency Department Care was 310.9 minutes. This indicates all Emergency Care resour ces were providing service/care when the injured survivor entered the Emergency Department process. They therefore had to wait on average 310.9 minutes to receiv e the care they required. Queue Number Waiting (Q#W): The arithm etic mean of the number of injured survivors waiting for a specific service at any point in time. Utilizing the data for Process Category P.11 in Appendix 25, at any point dur ing the simulation, an injured survivor requiring Emergency Department Care and entering the Emergency Department Care process could expect to find 494 other injured survivors waiting for the same service/resource.

PAGE 176

167 Results for Objective 6 : This objective was to provide an example of how planners in a specific area may use the results of the simu lation model to better prepare their location to respond to a large scal e mass casualty event. The tables that are included below repres ent a portion of the data generated by the Arena Software that are available to a pla nner and/or leader preparing their geographic location for a large scale mass casualty event. Critical to any pl anning effort is an accurate assessment of the availa bility of health resources wi thin the specified geographic area and the capacity of that system to de liver health care. The comparison of the predicted resource requirements and the availa ble resources allow the planner/leader to focus on the differential between the two. Deve loping processes and policies to care for the injured survivors with locally ava ilable resources and methodologies to access additional resource and cover the resource differential will be the task of the planner. Armed with both an assessment of the availabl e health resources/capacity of the area and a prediction of the required res ources to care for a truly larg e scale mass casualty event, the planner/leader is in a pos ition to prepare their geographic area for the possibility of a large scale mass casualty event. Table 30 displays the results of an unconstrained simulation of 7,000 injured survivors in relation to a goal of providing care to all injure d survivors within an 8-hour window in a resource dense geographic area. This table provides data for each of the previously described resource categories; Emergency Care, Field Care, Field Triage, Formal Transportation, Hospital Care, Hospital Triage, ICU Care, Imaging, Shock/Trauma Care, General Surgery, Neurosur gery, Orthopedic Surgery, and Thoracic Surgery. For each resource category the followi ng data elements are displayed: Total

PAGE 177

Table 30 7,000 Injured Survivors/Resource Dens e Geog raphic Area (8 hour window) Total Resource Resource Requirements Reso urces Available Differential Category Number Time Number Time Capacity Number Time Emergency Care 8,309 545,316 40 19,200 292 (8,017) (526,116) Field Care 1,808 42,167 40 19,200 804 (1,004) (22,967) Field Triage 2,314 1,094 10 4,800 10,212 7,898 3,706 Formal Trans 2,781 55,918 40 19,200 954 (1,827) (36,718) Hospital 2,091 8,146,777 1,000 480,000 123 (1,968) (7,666,777) Hospital Triage 5,847 2,760 20 9,600 20,425 14,578 6,840 ICU 258 1,236,261 80 38,400 80 (178) (1,197,861) Imaging 4,090 120,012 40 19,200 654 (3,436) (104,248) Shock/Trauma 1,296 55,882 10 4,800 111 (1,185) (51,082) General Surgery 303 36,458 80 38,400 319 1,942 16 Neurosurgery 165 20,072 20 9,600 78 (87) (10,472) Orthopedic Surgery 39 4,600 60 28,800 244 205 24,200 Thoracic Surgery 8 930 20 9,600 82 76 8,670 168

PAGE 178

Resource Requirem ents (Number, Time), Resour ces Available (Number, Time, Capacity), and Differential (Number, Time). The Total Resource Requirements (Number, Time) are those predicted by the simulation model to be required to provide care for 7,000 injured survivors with the mix of injuries previously desc ribed in the results section for Objective 4. Described differently, Total Resource Requirements repres ents the demand for a particular resource category without regard to the amount (number) of that resource that is available. Number represents the total amount of that particular resource that is utilized (demanded) during the simulation. Each tim e an injured survivor requires the care provided by a particular resource category it c ounts as one. If a si ngle injured survivor required emergency care at two different tim es, it would be counted as two. Time represents the cumulative time of service provided by the resource category or the total time that particular resource category is providing care or service. In Table 30 the Emergency Care resource responded to a to tal of 8,309 care requireme nts that required a cumulative total of 545,316 minut es of service/care. The Resources Available (Number, Time, Capacity) are the researcher defined resource quantities representing a resource de nse geographic area. Number represents the total amount (number) of the resource av ailable in the define d geographic area. Time represents the minutes of service available in an 8-hour time period (# of resources available x [8 x 60 minutes]). In Table 30 a total of 40 Emergency Care resource units are available in the geographic area. In an 8-hour time period, these resource units can provide a maximum of 19,200 minutes of care. Capacity represents the total number of injured su rvivors that could be cared for in the amount of time the 169

PAGE 179

particular resource is available. This is ba sed on the arithmetic mean from the simulation for the amount of time each injured survivor re quired care from the particular resource (available time/arithmetic mean process time) In Table 30, using the arithmetic mean from the simulation for emergency care provided to each injured survivor requiring emergency care, 40 emergency care resource units can be expected to care for 292 injured survivors in an 8-hour time period. Differential (Number, Time) represents the differential in terms of total resources required/available and in time (minutes of se rvice) required/available between that which is predicted by the simulation and what is available in the specified geographic area. In the case of Emergency Care, the differential between the total number of emergency care resource units required and th e total number available in an 8-hour time period is 8,017. Stated differently, the predicted demand for emergency care exceeds the capacity of the local resources by 8,017 units if the goal is to provide care to all injured survivors within 8 hours of the first arrival of a patient. Th e differential between the predicted demand for emergency care resource minutes of service a nd that available in the local area over an 8 hour time period is 526,116 minutes. In each case a deficit in resource capability is indicated by numbers in (parentheses). There is no parenthesis when there is a surplus. As examples, Emergency care resources are clearly incapable of meeting the demand, Triage resources have excess capacity. The data presented for each additional re source category can be interpreted in the same manner as has been explained in the previous paragraphs. Clearly changes in resource demand and/or resource capacity will produce changes in the differential. 170

PAGE 180

171 Table 31 displays the predicted resour ce requirements to care for 7,000 injured survivors in comparison with the capacity of a resource sparse geographic area to provide care in an 8 hour time period. Resource categorie s, data categories, and data calculation and interpretation are exactly as described above for Table 30. Tables 35 and 36 display the predicted re source requirements for a simulation of 45,000 injured survivors in a resource dense ge ographic area (Table35 ) and in a resource sparse geographic area (Table 36). Both Ta bles 35 and 36 display results for an 8-hour time period. Tables 37 and 38 display th e predicted resource requirements for a simulation of 7,000 injured survivors in a reso urce dense geographic area (Table 37) and in a resource sparse geographic area (Table 38), but this time utilizing a 12-hour time period. Tables 39 and 40 revisit the scenar io of 45,000 injured survivors utilizing the 12hour time period. In each case the resource categories, data headings, and data calculation and interpre tation are the same as was describe d for Table 30. Table 38 presents a comparison between the number of available resources required to meet the predicted resource requirements in eight, 12, and 20 hour time periods. Table 38 utilizes the 7,000 injured su rvivor scenario in a resource dense geographic area. As with the other tables in this section, data are presented for each of the previously described resource categories. The data presented in the tables is the Total Resource Requirements (Number, Time) and th e Resources Required to Meet Total Time Requirements (8-Hour, 12-Hour, and 20-Hour). The data in these categories were calculated by dividing the total service time requirement for each resource category by the 8 (480 minutes), 12 (720 minutes), and 20 (1,200 minutes) hour time windows. The number of resource units displayed under the 8, 12, and 20 hour columns indicate the

PAGE 181

Table 31 7,000 Injured Survivors/Resource Sparse Geo graphic Area (8 hour window) Resource Resource Requirements Reso urces Available Differential Category Number Time Number Time Capacity Number Time Em ergency Care 8,309 545,316 4 1,920 29 (8,280) (543,396) Field Care 1,808 42,167 4 1,920 80 (1,728) (40,247) Field Triage 2,314 1,094 1 480 1,021 (1,293) (614) Formal Trans 2,781 55,918 4 1,920 95 (2,686) (53,998) Hospital 2,091 8,146,777 100 4,800 100 (1,991) (8,141,977) Hospital Triage 5,847 2,760 2 960 2,042 (3,805) (718) ICU 258 1,236,261 8 3,840 8 (250) (1,232,421) Imaging 4,090 120,012 4 1,920 65 (4,025) (118,092) Shock/Trauma 1,296 55,882 1 480 11 (1,285) (55,402) General Surgery 303 36,458 8 3,840 31 (272) (32,618) Neurosurgery 165 20,072 2 960 7 (158) (19,112) Orthopedic Surgery 39 4,600 6 2,880 24 (15) (1,720) Thoracic Surgery 8 930 2 960 8 0 30 172

PAGE 182

Table 32 45,000 Injured Survivors/Resource Dens e Geographic Area (8 hour w indow) Resource Resource Requirements Reso urces Available Differential Category Number Time Number Time* Capacity Number Time Em ergency Care 53,358 3,505,095 40 19,200 292 (53,066) (3,485,895) Field Care 11,543 269,004 40 19,200 823 (10,720) (249,804) Field Triage 14,798 6,989 10 4,800 10,163 (4,635) (2,189) Formal Trans 17,869 360,571 40 19,200 951 (16,918) (341,371) Hospital 13,499 52,496,564 1,000 480,000 1,000 (12,499) (52,016,564) Hospital Triage 37,630 17,762 20 9,600 20,338 (17,292) (8,162) ICU 1,654 7,935,553 80 38,400 80 (1,574) (7,897,153) Imaging 26,295 771,001 40 19,200 654 (25,641) (751,801) Shock/Trauma 8,403 364,760 10 4,800 110 (8,293) (359,960) General Surgery 1,979 238,138 80 38,400 319 (1,660) (199,738) Neurosurgery 1,040 125,550 20 9,600 79 (961) (115,950) Orthopedic Surgery 237 28,506 60 28,800 239 2 294 Thoracic Surgery 59 7,130 20 9,600 79 20 2,470 173

PAGE 183

Table 33 45,000 Injured Survivors/Resource Sparse Geographic Area (8 hour w indow) Resource Resource Requirements Reso urces Available Differential Category Number Time Number Time* Capacity Number Time Em ergency Care 53,358 3,505,095 4 1,920 29 (53,329) (3,503,175) Field Care 11,543 269,004 4 1,920 82 (11,461) (267,084) Field Triage 14,798 6,989 1 480 1,016 (13,782) (5,973) Formal Trans 17,869 360,571 4 1,920 95 (17,774) (358,651) Hospital 13,499 52,496,564 100 4,800 100 (13,399) (52,491,764) Hospital Triage 37,630 17,762 2 960 2,033 (35,597) (16,802) ICU 1,654 7,935,553 8 3,840 8 (1,646) (7,931,713) Imaging 26,295 771,001 4 1,920 65 (26,230) (769,081) Shock/Trauma 8,403 364,760 1 480 11 (8,392) (364,280) General Surgery 1,979 238,138 8 3,840 31 (1,948) (234,298) Neurosurgery 1,040 125,550 2 960 7 (1,033) (124,590) Orthopedic Surgery 237 28,506 6 2,880 23 (214) (25,626) Thoracic Surgery 59 7,130 2 960 7 (52) (6,170) 174

PAGE 184

Table 34 7,000 Injured Survivors/Resource Dens e Geog raphic Area (12 hour window) To tal Resource Resource Requirements Reso urces Available Differential Category Number Time Number Time Capacity Number Time Em ergency Care 8,309 545,316 40 28,800 438 (7,871) (516,516) Field Care 1,808 42,167 40 28,800 1,206 (602) (13,367) Field Triage 2,314 1,094 10 7,200 15,318 13,004 6,106 Formal Trans 2,781 55,918 40 28,800 1,431 (1,350) (27,118) Hospital 2,091 8,146,777 1,000 720,000 1,000 (1,091) (7,426,777) Hospital Triage 5,847 2,760 20 14,400 30,637 8,553 27,877 ICU 258 1,236,261 80 57,600 80 (178) (1,178,661) Imaging 4,090 120,012 40 28,800 981 (3,109) (91,212) Shock/Trauma 1,296 55,882 10 7,200 166 (1,130) (48,682) General Surgery 303 36,458 80 57,600 478 175 21,142 Neurosurgery 165 20,072 20 14,400 117 (48) (5,672) Orthopedic Surgery 39 4,600 60 43,200 366 327 38,600 Thoracic Surgery 8 930 20 14,400 123 115 13,470 175

PAGE 185

Table 35 7,000 Injured Survivors/Resource Sparse Geo graphic Area (12 hour window) Resource Resource Requirements Reso urces Available Differential Category Number Time Number Time Capacity Number Time Em ergency Care 8,309 545,316 4 2,880 43 (8,266) (542,436) Field Care 1,808 42,167 4 2,880 120 (1,688) (39,287) Field Triage 2,314 1,094 1 720 1,531 (783) (437) Formal Trans 2,781 55,918 4 2,880 143 (2,638) (53,038) Hospital 2,091 8,146,777 100 7,200 100 (1,991) (8,139,577) Hospital Triage 5,847 2,760 2 1,440 3,063 (2,784) (1,360) ICU 258 1,236,261 8 5,760 8 (250) (1,230,501) Imaging 4,090 120,012 4 2,880 97 (3,993) (117,132) Shock/Trauma 1,296 55,882 1 720 16 (1,280) (55,162) General Surgery 303 36,458 8 5,760 46 (257) (30,698) Neurosurgery 165 20,072 2 1,440 10 (155) (18,632) Orthopedic Surgery 39 4,600 6 4,320 32 (7) (280) Thoracic Surgery 8 930 2 1,440 12 4 510 176

PAGE 186

Table 36 45,000 Injured Survivors/Resource Dens e Geographic Area (12 hour w indow) Resource Resource Requirements Reso urces Available Differential Category Number Time Number Time* Capacity Number Time Em ergency Care 53,358 3,505,095 40 28,800 438 (52,920) (3,476,295) Field Care 11,543 269,004 40 28,800 1,206 (10,337) (240,204) Field Triage 14,798 6,989 10 7,200 15,318 520 211 Formal Trans 17,869 360,571 40 28,800 1,431 (16,438) (331,771) Hospital 13,499 52,496,564 1,000 720,000 1,000 (12,499) (51,776,564) Hospital Triage 37,630 17,762 20 14,400 30,637 (6,993) (3,362) ICU 1,654 7,935,553 80 57,600 80 (1,574) (7,877,953) Imaging 26,295 771,001 40 28,800 981 (25,314) (742,201) Shock/Trauma 8,403 364,760 10 7,200 166 (8,237) (357,560) General Surgery 1,979 238,138 80 57,600 478 (1,501) (180,538) Neurosurgery 1,040 125,550 20 14,400 117 (923) (111,150) Orthopedic Surgery 237 28,506 60 43,200 366 129 14,694 Thoracic Surgery 59 7,130 20 14,400 123 64 7,270 177

PAGE 187

Table 37 45,000 Injured Survivors/Resource Spa rse Geographic Area (12 hour w indow) Resource Resource Requirements Reso urces Available Differential Category Number Time Number Time* Capacity Number Time Em ergency Care 53,358 3,505,095 4 2,880 43 (53,315) (3,502,215) Field Care 11,543 269,004 4 2,880 120 (11,423) (266,124) Field Triage 14,798 6,989 1 720 1,531 (13,267) (6,269) Formal Trans 17,869 360,571 4 2,880 143 (17,726) (357,691) Hospital 13,499 52,496,564 100 7,200 100 (13,399) (52,489,364) Hospital Triage 37,630 17,762 2 1,440 3,063 (34,567) (16,322) ICU 1,654 7,935,553 8 5,760 8 (1,646) (7,929,793) Imaging 26,295 771,001 4 2,880 97 (26,198) (768,121) Shock/Trauma 8,403 364,760 1 720 16 (7,683) (364,040) General Surgery 1,979 238,138 8 5,760 46 (1,933) (232,378) Neurosurgery 1,040 125,550 2 1,440 10 (1,030) (124,110) Orthopedic Surgery 237 28,506 6 4,320 32 (205) (24,186) Thoracic Surgery 59 7,130 2 1,440 12 (47) (5,690) 178

PAGE 188

179 Table 38 7,000 Injured Survivors/Resource Dense Geographic Area Total Resource Resource Requirements Resources Required to Meet Total Time Requirements Category Number Time 8-Hour 12-Hour 20-Hour Imaging 4,090 120,012 250 167 100 Hospital Triage 5,847 2,760 6 4 3 Hospital 2,091 8,146,777 2,091 2,091 2,091 Emergency Care 8,309 545,316 1,137 758 455 ICU 258 1,236,261 258 258 258 General Surgery 303 36,458 76 51 31 Shock/Trauma 1,296 55,882 117 78 47 Formal Trans 2,781 55,918 117 78 47 Neurosurgery 165 20,072 42 28 17 Field Care 1,808 42,167 88 59 36 Field Triage 2,314 1,094 3 2 1 Thoracic Surgery 8 930 2 2 1 Orthopedic Surgery 39 4,600 10 7 4

PAGE 189

total num ber of that specific resource unit required to provide the total amount of service time predicted to be required. In Table 38 the predicted demand for the Emergency Care resource unit is 8,309 units of care and 545,316 minutes of care/service. One thousand one hundred thirty seven emergency care res ource units would be required to be available to provide all required emergency ca re minutes of service in an 8-houor time period. Seven hundred fifty eight emergency care resource un its would be required to be available to provide all required emergency care in a 12-hour time period. Four hundred fifty five emergency care resource units woul d be required to be available to provide all required emergency care in a 20-hour time peri od. The interpretation of the remaining resource categories in Table 38 and the reso urce categories in Table 39 are the same as described in this paragraph. Appendices 29 and 30 provide two additional examples of how a planner/leader in a specified geographic area could use the simu lation model to better prepare the location to respond to a large scale mass casualty even t. Appendices 29 and 30 display the same Process Categories and corresponding data as are described for Appendices 23-28 in the results section for Objectives 4 and 5. The difference between Appendices 23-28 and Appendices 29 and 30 is that all the injured survivors receive care in Appendices 29 and 30. In the scenarios for Appendices 23-28, onl y those who enter the simulation within the specified time period access the care processes. Appendix 29 displays the simulation resu lts of the interaction of 7,000 injured survivors with a health system with adequa te resource capacity to provide care to the injured survivors within a 12-hour time pe riod. Appendix 30 display the simulation results of the interaction of 45,000 injured survivors with a hea lth system with adequate 180

PAGE 190

Table 39 45,000 Injured Survivors/Resource Sparse Geographic Area Resource Resource Requirements Resources Required to Meet Total Time Requirements Category Number Time 8-Hour 12-Hour 20-Hour Em ergency Care 53,358 3,505,095 7,303 4,869 2,921 Field Care 11,543 269,004 561 374 225 Field Triage 14,798 6,989 15 10 6 Formal Trans 17,869 360,571 752 501 301 Hospital 13,499 52,496,564 13,499 13,499 13,499 Hospital Triage 37,630 17,762 37 25 15 ICU 1,654 7,935,553 1,654 1,654 1,654 Imaging 26,295 771,001 1,607 1,071 643 Shock/Trauma 8,403 364,760 760 507 304 General Surgery 1,979 238,138 497 331 199 Neurosurgery 1,040 125,550 262 175 105 Orthopedic Surgery 237 28,506 60 40 24 Thoracic Surgery 59 7,130 15 10 6 181

PAGE 191

resource capacity to provide care to the in jured survivors within a 12-hour time period. As an example of how to interpret the data in these Appendices, the Emergency Department process (P.11) in Appendix 29 re ceived and provided care to 4,448 injured survivors who required emergency department care. The arithmetic mean for the minutes of service/care provided to each of these injured survivors (VA Time) was 59.16 minutes. The arithmetic mean of the Total Time (T Time) spent by each injured survivor in the Emergency Department process was 64.39 minutes This indicates each injured survivor waited approximately five minutes to receive emergency care after entering the emergency department. Emergency care res ources provided a total of 235,609 minutes of service (AVA Time) to injured survivors in the Emergency Department. This accounts for 100% of the care/service provided in the Emergency Department to the injured survivors. The arithmetic mean of th e waiting times (QW Time) experience by injured survivors in the Emergency Department was 5.98 minutes meaning each injured survivor who entered the Emergency Department wa ited approximately 5.98 minutes to receive care/service. Finally, upon entering the Em ergency Department each injured survivor found approximately 37 other injured survivor s waiting to receive care. This same interpretation applies to the other Proce ss Categories in Appendix 29 and all of the Process Categories in Appendix 30.

PAGE 192

Chapter 6 Summary and Conclusions Recent even ts throughout the world and in the United States lend support to the belief that another attack on the United States is likely, perhaps even probable. Given the potential for large numbers of casualties to be produced by a blast using conventional explosives, it is imperative that health system s across the nation consider the risks in their jurisdictions and take steps to better prepare for the possibil ity of an attack. Computer modeling and simulation offers a viable and useful methodology to better prepare an organization or system for both day-to-day operations, and for the occurrence of a one time catastrophic event. This research presents computer modeling and simulation as a tool capable of assisting lo cal health care leaders and health and emergency response planners to better prepare thei r jurisdictions for a large scal e mass casualty event. In the case considered for this research, the larg e scale mass casualty event is caused by a purposeful blast using conven tional explosives. Although this research focused on a single scenario, the methodology presented he re could be used to model and assess preparedness for a variety of disaster scenario s. Published literature presents examples of computer modeling and simulation being us ed to assess and improve individual service units in a health care setting. These simulati on models typically recr eate the structure of an individual clinic or service, then seek to analyze the processes performed in the setting and the movement of patients receiving care or service in that clinic. The objective of 183

PAGE 193

this kind of research and analysis (com puter m odeling and simulation) is improved performance in terms of resour ce utilization and efficiency in patient care and patient movement. The overall objective of this particular research was to determine if computer modeling and simulation offered a viable me thodology to prepare a health system to respond to a large scale event. This particular large scale scenario is outside the norm of day-to-day preparedness activit ies and processes. The r eal question was, given the shortage, and in some areas ab sence, of experiential data, could computer modeling and simulation (as a methodology) be used to pred ict the resource requirements generated by this type of event and thus be tter prepare a health system in a defined geographic area for the possibility of an event of this nature? It is difficult if not impossible for health systems in the United States to fully test their readiness to respond to a truly large scale event. H ealth system being defined as the aggregation of health re sources in a defined geographi c location; this aggregation being anything from a single community hospital and supporting emergency response capability in a defined, but geographically remo te location, to the aggregation of hospitals and other health resources in a large metropo litan area. The reason is twofold. First, health systems cannot cease all daily operations to concentrate on an exercise of this magnitude. Health systems cannot completely halt the daily need to care for patients to fully test or exercise their ability to respond to a disaster. Subseque ntly, they cannot test their ability to provide the variety and volum e of care that would be required by even 7,000 injured survivors in a manner that would be needed in response to a disaster of this magnitude. Not only are these health systems not able to fully test or assess their readiness for a large scale mass casualty event, but they have not fully considered the 184

PAGE 194

ram ifications of casualties numbering in the multiple thousands; 7,000, 20,000, or even 45,000. Health systems across the United States have no effective and efficient way to fully assess the resource requirements of an injured population of this size or how their system would respond to such a need. It is essential that pr eparedness activities go beyond consideration of how best to ensure survival of the physical plant in case of a natural disaster and the development and training of internal/hospita l emergency response teams to assemble and respond when a disaster strikes. In response to the shortcomings in term s of health system preparedness, this research first identified the characteristics or parameters that might indicate a health system at risk of experiencing a large scale mass casualty situation as a result of a purposeful blast. The charac teristics of the scenario or variables that would likely surround a large scale blast even t were then identified. Un fortunately, published studies and existing data did not allo w the development of a computer model which would utilize the location, facility, and blast characteristics to predict the total number of casualties to be expected from a population at risk. As a proxy, this re search developed a computer model and subsequent simulation methodology to assist in identifying the likely mix of injuries that would be experi enced by a defined injured survivor population. This mix of injuries was identified in terms of injury types or categories and volume of each type of injury that would result from this type of bl ast. A computer model was also developed to identify the resources that would be required (defined by specific categories of resource units) to care for this populati on of injured survivors. Finall y, the process of tailoring the computer model to a specific geographic location in terms of available resources was presented and various scenarios were simulated to facilitate an assessment of how ready a 185

PAGE 195

health system in a specific geograph ic location is for this type of event and to prompt consideration and preparation wi thin the health system for the possibility of responding to a large scale event of this nature. As a point of clarification, the intent of the first objective was not to ascertain the risk or probability that a large scale blast event using conventional explosives will occur, but rather to describe the pa rameters and characteristics th at would surround a large scale blast that produced a large numbe r of casualties; with injured survivors being the focus. The findings reveal an endle ss supply of opportunities where a large number of people gather in a single location. The venues for athletics, entertainment, worship, shopping, travel, and conferences in the United States afford weekly if not daily gatherings of large numbers of people. Health planners in thes e locales must consider the possibility of a purposeful or man-made blast producing a mass casualty situation. This research identified hundreds of venues, by name, across the United States where a large scale blast could be expected to result in thousands of casualties and 5,000 or more injured survivors. The venues identified for this research are primarily venues built specifically for sporting teams or events; basketball, hockey, baseball football, and auto racing. Additionally, only indoor facilities with seati ng capacities of 10,000 or more and outdoor facilities with seating capacities of 30,000 or more were includ ed. This research did not specifically identify smaller sports facilities or other venues that could also house large numbers of people: churches, malls, conference centers, trai n stations, and airports to name a few. Each of these smaller venues should not be discounted in that each offers a population at risk of a large scale mass casualty even t in its own right. 186

PAGE 196

The interaction of m ultiple variables will produce the population of injured survivors: the characteristics of the blast itself (blast ma terial, strength, location, and delivery mechanism), the structural characte ristics of the venue, and the number of people in the venue or immediate area who are at risk. Based on the experience documented in published studies and analyses, a blast producing the number of casualties considered for this research would be logistically challenging in terms of these characteristics. A single blast capable of producing the number of casualties considered for this research must, by definition, exceed the magnitude of that found in any currently published studies. Based on existing anal yses, a single blast using conventional explosives would also require an extremely large delivery mechanism. As an alternative, multiple smaller scale blasts in a single location/venue must be considered. Even though a blast of the magnitude necessary to produce a truly large scale mass casualty situation appears to be logistically challenging, the mere fact that the United States offers a large number of venues or targets housing larg e and concentrated populations of people mandates health system leaders and planners consider the risks. When considering the computer simulation models developed fo r this research, it is important to reinforce th at both portions of the mode l are based solely on publicly available data. Furthermore, the injury pred iction portion of the si mulation model is not designed to predict the number of injuries ex pected within a populati on at risk, but rather, given a pre-defined number of injured survi vors, the prediction model was designed to predict the mix of injuries that would be expected with in the population. The injury prediction model was based on experience gained as a result of historic al blast events and the published analyses of these events. This experience was then extrapolated to the 187

PAGE 197

larger population of injured survivors to identify the m ix of injuries that would be expected. No published studies have iden tified a population at risk from which a calculation can be made for the mix of dead, injured survivors, and uninjured survivors for this type of event. The assumption supporting this methodology is that the application of prediction parameters based on sm all scale events is applicable to a larger injured survivor population. The effort applied for objectives two and three resulted in the development of the computer simulation model for this resear ch. The first portion of the computer simulation model, as discussed in the prev ious paragraph, was de signed to predict the types of injuries by ICD-9 codes that would result from a large scale blast event. This portion of the computer model took the i nput, a predetermined number of injured survivors (excluding those killed by the blast and uninjured survivors) and predicted the injuries that would be expected in this population. The pred iction parameters utilized for this research were based on findings in published research studies. To clarify, this model was not designed to predict the tota l number of casualties, both killed and wounded, from a population at risk. It was designed to take a pre-defined number of injured survivors and predict the mix of injuries that woul d be found based on past experience. Two computer simulation models (prediction met hodologies) were developed. The first (and the one utilized to accomplish the subsequent objectives of this research) was based on a single study of multiple blast events. It was subjectively judged by the researcher as the best (most representative) study for the purposes of this research. Injury categories and prediction parameters as presented in the study were applied dire ctly as the injury categories and prediction para meter for the injury predic tion methodology. The injury 188

PAGE 198

categories and prediction param eters were thus extrapolated directly to the larger injured survivor population that was the focus of this research. The methodology for the second injury prediction model was based on a published matrix of injury categories designed to facilitate research and analyses of trauma injuries. The Barell Injury Diagnosis Matrix (see Appendix 7) presents a very detailed categorization of trauma associated injury categories. Th e shortcoming of this methodology is that no studies of the injuries resulting from blast events use this matrix as the basis for categorizing injuries and thus no published prediction parameters exist which can be applied directly to the injury prediction portion of the model. Although this methodology provided a much more expansive presentation of injury categories, without published studies to develop the injury pred iction parameters, utilization of this methodology would require application of resear cher defined prediction parameters to all injury categories. The second portion of the computer simulation model was designed to predict the medical resources that would be required to care for the injured survivors (with the injuries identified in the injury prediction po rtion of the model), up to the point that the injured survivors completed their hospital stay. This portion of the model identified the series of processes or services and supporting resources that an injured survivor would experience or require following a blast related injury. This portion of the simulation did not consider the long term car e requirements and rehabilitation requirements that some injured survivors would require. As with the injury prediction portion of the model, this portion of the model, in terms of processes, resources, and process parameters, was based on published studies when available and user defined parameters when published data 189

PAGE 199

were not available. The processes included in th e model represent those health care services that would be re quired to care for the injure d population. The resources represent the resources (people, space, equi pment, supplies) requi red to complete or accomplish the processes, to provide the service, or to perform the task. Finally, the process parameters were the expected times or range of expected times to perform each service or accomplish each task (or process). The processes described in this model provided for the initial contact of an injured su rvivor with the health care system in the field, the triage process and either care in the field or transportation to a definitive care facility (hospital). The model then incorporated the em ergency care processes of triage, trauma or shock care for those in need, im aging studies, surgery, and hospital care, to include both intensive care a nd regular hospitalization. The m odel terminated at the point the injured survivor completed their c ourse of care in the hospital. Each identified process was supported by a resource unit. The resource unit was comprised of all the resources required to comp lete the process and/or provide the service provided in that process. For example, a single surgery resource unit was comprised of an operating room, a surgeon, the operating room nurse, the anesthesiologist, and all other supporting personnel, suppl ies, and equipment. The ut ilization of resource units provided for the identification of the multiple ty pes of resources that would be required to care for the injured survivors while allowing the flexibility for a local planner to determine how best to define the resource units and how best to utilize existing resources to respond to the health care needs. Each resource unit could be further divided into multiple sub-components that when combined, result in a complete resource unit able to accomplish the required task or provide the requi red service. The result of the simulation 190

PAGE 200

using this portion of the model was the identi fication of the total num ber of resource units required to meet the needs of the injured population and the cumulative time of service required to be provided by each category of resource units. These predictions provide a starting point, they provide the basis for a planning effort to determine the total requirements in terms of resources, and th en consider the available resources, the differential, and how the requirements might best be met. The results of objectives four, five, and six displayed the output of the simulation methodology when used to predict the types a nd quantities of the in juries experienced by the injured survivors, the resources require d to care for the inju red population based on the predicted mix of injuries, and potential resource differen tial that would be faced by a health system with a finite set of resources. The results of the simulation runs for each of the objectives were included in the fi ndings section of this document. First, the injury prediction portion of the model was run and results provided for several different scenarios. Injury pred ictions, using ICD-9 codes, were made for populations of both 7,000 and 45,000 injured surv ivors with three different scenarios each. The three scenarios included no constr aints in terms of resources and time, a resource sparse scenario with a 12-hour time limit, and a re source dense scenario with a 12-hour time limit. The starting point for each simulation process was the scenario with no constraints in terms of time allowed and ava ilable resources. This scenario/simulation served to identify the total requirements in terms of resource volume and total time of service or care required. The subsequent simulations presented a subset of the overall prediction for resources and required time of se rvice. Limitations in terms of time and resources served to limit the number of injure d survivors who entered the health system 191

PAGE 201

and who consequently were iden tif ied within the model. With no change in the rate or frequency that injured survi vors entered the system or model (also known as the speed with which injured survivors were created in the simulation model), the time constrained scenarios simply reduced the number of predicte d injuries to a proporti on of the total. The speed with which the injured survivors en tered the system, when combined with time constraints and resource constr aints, restricted the total num ber of injured survivors who entered the system and thus the total number of injuries predic ted for the population. Basically, if the speed at which injured su rvivors entering the system did not change, restricting time and resources only served to restrict the number of injured survivors who gained access to the health system. So a portion of the injured survivor population was not considered in the prediction data. Second, the simulations were run to pred ict the medical resour ces that would be needed to care for the population of injured su rvivors, with the injuries predicted in the first portion of the model. The same six scen arios as described above were utilized to predict the resources required to care for the population of injured survivors. As in the injury prediction portion of the model, the first simulation scenario predicted the total requirements in terms of health care resource s required to care for the injured survivor population. The limitations in the subsequent simulation runs combined with the consistent speed of the injured survivors entering the simulation resulted in only a portion of the injured survivors entering the system and thus identifying only a portion of the resources required to provide the necessary care for the total population. Finally, multiple scenarios were run to show how a defined geographic location would utilize the model and simulation methodol ogy to better prepare themselves for the 192

PAGE 202

possibility of a large scale blast and resulting m ass casualt y situation. Both the 7,000 and 45,000 injured survivor scenarios were utilized with time and resource constraints similar to what was described for the injury prediction portion of the model. Four scenarios were tested for each injured survivor scenario: re source sparse and resource dense constraints were matched with 8-hour and 12-hour time constraints. The multiple scenarios allowed for comparison in terms of total requirem ents, both in number and cumulative time of service required to care for the injured survivor s. The primary result of this portion of the research was to identify the resource differential a health system with a finite set of resources would face if confr onted with a large scale mass casualty situation. In each case the health care leader or planner can us e the identified differential between available and required resources to prepare alternate solu tions to meet the care requirements of the injured population. Conclusions As a bottom line up front statement, the utili ty of this research and the computer simulation model in its current form is twofold. First, it o ffers a prediction methodology to identify the types of injuries that are like ly to be experienced in a large scale blast event. Secondly, it serves to identify the reso urces that would be required to care for an injured population of this size, and thus provides the means to identify the differential between resource requirements and resource cap acity in the defined geographic area. With this information in hand, local health leaders and health planners can begin to explore methods for responding to this type of event with the resources that are available in the local geographic area. Furthermore, whether at the local, regional, state, or national level, health leaders and health pla nners can pursue the necessary affiliations and 193

PAGE 203

support agreem ents to effectivel y respond to an event of this nature in a collaborative manner. Computer simulation and m odeling affords a viable and usable option for a defined geographic region to bett er prepare for the potential of a large scale blast event and caring for the injured survivor s who are injured as a result of a blast. As depicted in this research, this can be done with a rela tively low cost and low tech approach using existing computer modeling and simulation software, thus making it affordable and viable for even the smallest geographi c jurisdiction or health system. With respect to the risk of a blast producing a large scale mass casualty event, the volume of venues which house large concentratio ns of people indicate health leaders and planners must take notice. Increased safety measures at these venues and the fact that no events of the magnitude descri bed in this research have o ccurred does not negate the need for action and preparation. Recent catastr ophic events, both man-made and natural, should be an adequate incentiv e for health systems across th e Unites States to consider the possibilities. Computer modeling and simulation afford the opportunity for any health system to start asking th e what if questions. A standard or common methodology is n eeded for researchers studying injury patterns as a result of blast injuries. One lim itation of this research was the inconsistency between studies in terms of injury categorie s and the differentiation between those with severe injuries and those with other-than-severe injuries. The identification of a common analysis methodology will ensure comparabil ity between studies and would allow for multiple complimentary studies to be used when identifying prediction parameters in a model like that developed for this research. A tool like the Barrel Injury Diagnosis 194

PAGE 204

Matrix affords an available too l for sta ndardization purposes. At this point, any disagreement in injury categories and di vision between categories is overshadowed by the benefit of utilizing consistent injury cat egories and research methodology. Note that the literature discussing the Barrel Matrix ha s cited the injury categories and divisions used in the matrix as one source of criticism. The methodology utilized for predicting the mix of injuries amongst a defined injured survivor population is the only viable option given the current limitations in data. Current data and published studies do not pr ovide the means to predict the total number of casualties that would be experienced among an identified populati on at risk, therefore predicting injury patterns and the resulting health resource requirements based on a specified number of injured survivors is the best option. The magnitude of the injured population as depicted in this re search requires that jurisdictions consider field re sponse/care capabilities as a m eans of caring for the injured while attempting to avoid the immediate influx of all patients, both severe and other-thansevere, to the local facilities. The sheer volum e of need dictates that local facilities will be overwhelmed quickly. The triage process is designed to match an injured survivor with the appropriate level of care/service. Providing additional out lets where care is provided will be essential to ensure severely injured survivors have the best chance of receiving appropriate care in a hospital setting. When considering how best to respond to a disaster and resul ting injured survivor population of this magnitude, health system l eaders/planners must consider an expanded geographic region when pursuing regional s upport relationships. The overwhelming nature of an event like this will require an overwhelming response. As regional alliances 195

PAGE 205

are cons idered, communication and transportati on considerations become critical; both the process of notifying regi onal partners of a need, th e transportation of medical personnel to the location of the injured, and a system of transporting the injured to regional facilities for care must be considered. Although viable in its current form as a prediction and assessment methodology, the next logical step is to a pply specific resources and respons e structures (i.e. identify a real world location and match the model to this location) to the model. This combined with the application of local response e xpectations/parameters would provide the opportunity for the model to be used to a ssess current preparedne ss levels and improve the overall response capacity of a specified location. Additionally, the application of alternative response methodologies, process parameters, and resource capacity would allow the comparison between methodologies in the search for an optimal or best response plan or capability. The fact that the United States has not experienced an attack since September 11 of 2001 cannot be used as justification for not fully considering the ramifications of a truly large scale mass casualty event. Health systems across the Unites States owe it to their beneficiary populations to consider and be ready for any contingency. The combination of resource constraints and da y-to-day operating pressures dictate using alternative means to assess and prepare. Computer modeling and simulation is one of those alternative means that offers the opportu nity to better prepare a jurisdiction with minimal cost and impact on currently operating health care systems. 196

PAGE 206

Chapter 7 Limita tions of the Study As discussed in the previous chapters c onclusions section, this study and associated computer simulation model offers a foundation for health leaders and health planners to consider how the health system in their jurisdiction could/will respond to a large scale mass casualty scenario resulting from a conventional blast. Although offering a viable methodology for planning purposes, there are limita tions to this research that must be considered when utilizing the model/met hodology. When understood, these limitations can serve to shape subsequent research to further refine the model/methodology in pursuit of a level of optimum preparedness for a large scale mass casualty scenario. Case Simulation: As presented in th is research, the model and methodology represents what can be described as a cas e simulation. Since pub lished studies/data are limited (reference the next section of this chapter), the injury prediction portion of this research is based on a single study (Frykbe rg, 1988) or case. Although this study compiled data and analyzed multiple blast act ivities, it did not present a range of experience to apply as injury prediction parameters. Thus th e injury predictions resulting from this research are limited to a specific case rather than presenting a range of possible outcomes and covering a range of blast scenarios. Published Studies/Data Supporting the Research : As discussed in previous sections, the data used to develop the computer simulation model for this study, to include the 197

PAGE 207

decision and process param eters, was based on p ublished studies. Current literature was limited in several ways. First, no large scale mass casualty events of the nature described in this study have occurred. The result being that data from small scale blast events was extrapolated to the larger injured population. Second, alt hough a number of studies have analyzed the injury patterns experienced by survivors of blast events, no two studies presented the same set of injury categories. Th is resulted in the use of one representative study as the source of injury prediction paramete rs. Finally, limited literature exists that describe the response of a health system to a mass casualty situation. Where published studies provided parameters for care process times and distribution, they were used. In most instances however, resear cher defined parameters were utilized to establish the process times. Note that this last point al igns with one of the benefits of computer simulation; it affords the opportunity for the re searcher or user to alter the structure and process parameters to test the system under difference circumstance, thus allowing for the pursuit of an optimal resource/methodology mix. Correlation between types of injuries: Re lated to the previous section, published studies did not allow or provide for the identification of correla tions between injury categories. It is likely that certain injury categories have a high corre lation, therefore, as an example; an injured survivor who experiences a head injury might also be expected to experience a bony extremity injury, or one who experience blast lung might also be expected to experience an abdominal injury. As the current model is constructed, every injured survivor has the chance of experien cing each injury category. As previously described, the assignment of an injury categor y is a random event w ith the chance being the probability of the true response assigned to the particular deci sion module. Better 198

PAGE 208

data to supp ort the injury assignment/predic tion process would clearly result in a more comprehensive injury prediction model. Simulation Model Design/Construction: The flexibility of the computer simulation software used for this research allows a vari ety of methods to be utilized to develop and construct the model. The limitations discusse d here are not limitati ons associated with the capabilities of the software, but are lim itations associated with how the researcher chose to construct the simulation model. Future research efforts could easily alter the simulation model to address these limitations The ease of alteri ng the simulation model is one of the previously discussed bene fits of using computer simulation. One limitation is the speed with which the entities are created in the simulation model. For this study, this is the speed the injured survivors enter (or are created in) the simulation model. The research scenario s where time and resource constraints were placed on the model were influenced less by these constraint and were limited more by the speed the injured survivors entered the model. The speed th e injured survivors entered the model and the overall time constraint placed on the simulation run served to limit the overall number of injured survivors who entered the model, and thus impacted the resources and time of service required to care for the injured survivors. Speeding up the rate the injured survivor enter the model or placing a large number of injured survivor in the model at time zero would serve to eliminate this limitation. The specificity of model components is a second limitation. The simulation software used for this research allows as much specifi city in the design of the simulation model as the researcher or user desire s to establish. The researcher purposefully chose a more basic or simple simulation model design. One of the guiding principles behind computer 199

PAGE 209

modeling an d simulation is to ensure the simulation model is not to complex to ensure it is understandable but to also ensure it can be easily modified. In many instance, the processes included in the current model c ould be broken down into more precise subprocesses. Additionally, the resource units used to support the process modules could be subdivided into individual reso urce requirements. As an ex ample, a surgical resource could be subdivided into a surgeon re source, operating room nurse resource, anesthesiology resource, the operating room itself, and each piece of equipment or supply item category required to perform the surgical procedure. Future research could easily include this type of resource specificity if it were required to meet the needs of the researcher or user. Model Utilization: The final limitation has to do with the utilization of the computer model and simulation as presented in this research. As stated in the previous chapter, the utility of this simulation model in its current form is to first predict the total resource requirements to care for an inju red population like that describe d in this research. Second is to identify the differentials between the resources required to care for a large injured population and that available in a defined geographic area. The step this research did not take was to utilize the simulation methodol ogy to make comparisons between resource allocation alternatives and/or different methodologies for provid ing care to a large injured population. This type of resear ch/analysis is a logical next step as a user (researcher, health leader, and health planner) search es for an optimal or best allocation and utilization of available resource to respond to the defined scenario. One of the benefits of the simulation methodology is as a means of produc ing data that can be analyzed with the results being used to modify an existing system. This opportunity applies to both the 200

PAGE 210

injury prediction portion of the m odel and the resource requirements prediction portion of the model. 201

PAGE 211

References Adler, J., Golan, E., Golan, J., Yitz haki, M., and Ben-Hur, N. (1983). Terrorist Bom bing Experience During 1975-79 Casualties Admitted to the Shaare Zedek Medical Center. Israel Journal of Medical Sciences, 19(2), 189-193. Aharonson-Daniel, L., Kluger, Y., Giveon, A., and Peleg, K. (2004). Terror Related Explosion Injuries The Need fo r Revised Injury Severity Score. Public Health and the Environment Annual Meeting Disaster and Terrorism Poster Session. Retrieved July 26, 2005 from http://apha.confex.com/apha/132am/techprogram/paper_89173.htm Almogy, G., Belzberg, H., Mints, Y., Pikarsky, A.K., Zamir, G. and Rivkind, A.I. (2004). Suicide Bombing Attacks Update and Modifications to the Protocol. Annals of Surgery, 239(3), 295-303. American College of Surgeons (2003). Statement on disaster and mass casualty management. Retrieved February 17, 2004, from www.facs.org/fellows_info/statements/st-42.html. Arsham, H. (2003). Simulation: In troduction and Summary. Retrieved June 15, 2003, from http://home.ubalt.edu/ntsbarsh/simulation/sim.htm Association for the Advancement of Automotive Medicine (AAAM) (2005). Retrieved July 27, 2005 from http://www.aaam.org Balakrishnan, A.V. & Thomas, M. (E ditors). (1979). Lecture Notes in Control and Information Sciences: Global and Large Scale System Models New York: SpringerVerlag. BBC News (2004). Al-Qaeda behind Egypt bombings. Retrieved October 19, 2004 from http://newsvote.bbc.co.uk BBC News (2004). Death toll rises in Egypt blasts. Retrieved October 19, 2004 from http://newsvote.bbc.co.uk Bowen, N.J., & Pretto, E.A. (1999). Su rvey of State Level Catastrophic Casualty Management Plans in the United States of Am erica. Prehospital and Disaster Medicine, 14(supplement 1), S28. 202

PAGE 212

Brady, T. (2003). Emergency Managem ent: Capability Analysis of Critical Incident Response. Proceedings of the 2003 Winter Simulation Conference. 1863-1867. Brismar, B. and Bergenwald, L. (1982) The Terrorist Bomb Explosion in Bologna, Italy, 1980: An Analysis of the Effects and In juries Sustained. The Journal of Trauma, 22(3), 216-220. Bregman, D., Gilat, D., and Levi, L. ( no date listed). The Use of an Arena system in Developing an Implementing Readiness of Hospita l Management to Disastrous Situations. Retrieved February 17, 2004 from www.satgmbh.de/download/pdf/Arena system inhospitalmanagement.pdf Centers for Disease Control (2003). Trauma Care Systems. In Injury Fact Book 2001-Retrieved from www.cdc.gov Centers for Disease Control (2005). Explosions and Blast Injuries, A Primer for Clinicians. Retrieved 5 Jul 05 from www.bt.cdc.gov Centers for Disease Control (2007). International Collaborativ e Effort (ICE on Injury Statistics, The Barell Injury Diagnosis Ma trix, Classification by body Region and Nature of the Injury. Retrieved 5 Jul 05 from www.bt.cdc.gov CNN.com (2007). Iraqi officials: Truck bombings killed at le ast 500. Retrieved 20 Oct 07 from www.cnn.com/2007/WORLD/meast/08/15/iraq.main Dorland Illustra ted Medical Dictionary, 27th Edition (1988). Philadelphia: W.B. Saunders Company, Harcourt Brace Jovanovich, Inc. Einav, E., Feige nberg, Z., Weissman, C., Zaichik, D., Caspi, G., Kotler D., and Freund, D. (2004). Evacuation Priorities in Mass Casu alty Terror-Related Ev ents, Implications for Contingency Planning. Annals of Surgery, 239(3), 304-310. Emedicine (2005). Excerpt from Trau ma Scoring Systems. Retrieved July 27, 2005 from http://www.emedicine.com/med/byname/trauma-scoring-systems.htm Frykberg, E.R. (2004). Principles of Mass Casualty management Following Terrorist Disasters. Annals of Surgery, 239(3), 319-321. Frykberg, E.R. (2002). Medical Management of Disasters and Mass Casualties From Terrorist Bombings: How Can We Cope? Th e Journal of TRAUMA Injury, Infection, and Critical Care, 53, 201-212. Frykberg, E.R., Tepas, J.J., and Al exander R.H. (1998). The 1983 Beirut Airport Terrorist Bombing Injury Patterns and Implications for Disaster Management. The American Surgeon, 55(3), 134-141. 203

PAGE 213

Frykberg, E.R. and Tepas, J.J. (1988). Terrorist Bom bings Lessons Learned From Belfast to Beirut. Annals of Surgery, November, 569-576. Green, W.G. (2000). Mass Casualty In cident Management: The Virginia Model. Retrieved February 17, 2004, from www.urich.edu/~wgreen/conf7.pdf Gun and Gunpowder. Retrieved 21 July 2007 from http://www.silk-road.com/artl/gun.shtml Hallbert, B. (2004). Engineering Capability Brief: Emergency Response Reliability in Mass Casualty Events. Retrieved September 30, 2004 from www.cee.vanderbuilt.edu/ Hillsborough County Trauma Agency (2004). Hillsborough County Trauma Agency. Retrieved February 17, 2004, from www.hillsboroughcounty.org Hirshberg, A., Holcomb, J.B., and Mattox, K. (2001). Hospital Trauma Care in MultipleCasualty Incidents: A Critical View. Annals of Emergency Medicine, 37(6), 647-652. Hirshberg, A., Stein, M., and Walden, R. (1999). Surgical Resource Utilization in Urban Terrorist Bombing: A Computer Simulation. The Journal of Trauma Injury, Infection and Cr itical Care, 47(3), 545-550. Husum, H. and Strada, G. (2002). Injury Severity Score Versus New Injury Severity Score for Penetrating Injuries. Prehosp ital Disaster Medicine, 17(1), 27-32. Jain, S. and McLean, C. (2003). A Framework for Modeling and Simulation for Emergency Response. Proceedings of the 2003 Winter Simulation Conference, 10681076. Kelton, W.D., Sadowski, R.P., & Sturroc k, D.T. (2004). Simulation with Arena. Boston:McGraw Hill. Kernaghan, L. (2004) Halifax Explosion. In The Canadian Encyclopedia. Retrieved from http://tceplus.com. Levi, L., Bregman, D., Geva, H., and Revach, M. (1998). Hospital Disaster Management Simulation System. Prehospita l and Disaster Mana gement, 13(1), 29-34. Mallonee, S., Shariat, S., Stennies, G., Waxweiler, R., Hogan, D., & Jordan, F. (1996). Physical Injuries and Fatal ities Resulting From the Oklahoma City Bombing. The Journal of the American Medical Association, 276(5), 382-387. MacKenzie, E.J., Hoyt, D.B., Sacra, J.C., Jurkovich, G.J., Car lini, A.R., Teitelbaum, S.D., & Teter, H. (2003). National Inventory of Hospital Trauma Centers. Journal of the American Medical Association, 289(12), 1515-1522). 204

PAGE 214

McGraw-Hill Concis e Encyclopedia of Engineering (2002). Simulation. Retrieved on January 4, 2009 from http://encyclopedia2.thefreedictionary.com/simulation Mendeloff, J.M. & Cayten, C.G. (1991) Trauma Systems and Public Policy. Annual Review of Public Health, 12, 401-424. Muller, N., Muller, H. & Martens, P. (2001). Do Computer Programs Perform Better Than Human Regulators in Mass Casualty Disa sters? Prehospital Disaster Management, 16(2), S51. National Highway Traffi c Safety Administration (2002). Trauma System Agenda For the Future. Retrieved January 20, 2004, from www.nhtsa.dot.gov Osler, T., Rutledge, R., Deis, J., a nd Bedrick, E. (1996). ICISS: An International Classification of Disease-9 Based Injury Se verity Score. Trauma, 41(3), 380-386. Peleg, K., Aharanson-Daniel, L., St ein, M., Michaelson, M., Kluger, Y., Simon, D., Israeli Trauma Group, & Noji, E.K. ( 2004). Gunshot and Explosion Injuries, Characteristics, Outcomes, and Implications for Ca re of Terror-Related Injureis in Israel. Annals of Surgery, 239(3), 311-318. Pidd, M. (2004). Systems Modelling Th eory and Practice. West Sussex, England: John Wiley & Sons, Ltd. Rapid Assessment of Injuries Among Su rvivors of the Terrorist Attack on the World Trade Center New York City, September 2001. (2002, January 11). Morbidity and Mortality Weekly, pp. 1-5. Riley, D., Clark, M., and Wong, T. ( 2002). World Trade Cent er Terror: Explosion Trauma Blast, Burns, and Crush Injury. Topi cs in Emergency Medici ne, 24(2), 47-59. Rubenstein, R.Y. & Malamed, B. (1998). Modern Simulation and Modeling. New York: John Wiley & Sons. Rutledge, R., Osler, T., Emery, S., a nd Kromhout-Schiro, S. (1998). The End of the Injury Severity Score (ISS) and the Trauma and Injury Severity Score (TRISS). Trauma, 44(1), 41-49. Severance, F.L. (2001) System Mode ling and Simulation: An Introduction. New York: John Wiley & Sons, LTD. Shamir, M.Y., Yoram, G.W., Willner, D, Mintz, Y., Bloom, A.I., Weiss, Y., Sprung, C.L., and Weissman, C. (2004). Multip le Casualty Terrorist Events: The Anesthesiologists Perspec tive. Anesth Analg, 98, 1746-1752. 205

PAGE 215

Slater, M.S. & Trunkey, D.D. (1997). Te rrorism in Am erica: An Evolving Threat. Archives of Surgery, 132(10), 1059-1066. Sokolowski, J. (2004). Employing Simulation for Mass Casualty Management. Retrieved on November 3, 2004 from http://www.odu/engr/vmasc/jsokolowski.shtml Spiller, P. (2004). A scene of devastation. BBC News, retrieved October 19, 2004 from http://newsvote.bbc.co.uk State of Florida. Trauma Chapter 395.40 Legislative Findings and Intent. Retrieved February 17, 2004, from www.hillsboroughcounty.org Stevenson M., Segui-Gomez M., Lescohier, I., Di Scala, C ., and McDonal-Sm ith, G. (2001) An Overview of the Injury S e verity Score and the New Injury Severity Score. Injury Prevention, 7, 10-13. Teague, David C. (2004) Mass Casualties in the Oklahom a City Bombing. Clinical Orthopaedics and Related Research, 422, 77-81. Thom pson, D., Brown, S., Mallonee, S., and Sunshine, D. (2004). Fatal and NonFatal Inju ries am ong U.S. Air Force Personnel Resulting from the Terrorist Bom bing of the Khobar Towers. The Journal of TRAUM A Injury, Inf ection, and Critical Care, 57 (2), 208-215. Terrorist Attacks (within the United States or against Am e r ican abroad) (2004). infoplease. Retrieved Septem ber 24, 2004, from http://print.infoplease.com Texas City Disaster (2002). In The Handbook of Texas Online. Retrieved from http://www.tsha.utexas.edu/handbook/online The Texas City Disaster Apr il 16, 1947 (2004). Retrieved October 20, 2004 from http://www.local1259iaff.org/disaster.htm l TRAUMA.ORG (2005). Retrieved July 6, 2005 from http://www.traum a .org/scores/ U.S. Census Bureau (2005). Metropolitan and Micropolita n Statistical Areas. Retrieved 28 June 2005 from www.c e nsus.gov U.S. Department of Justice (2001). Press Release. Retrieved 13 Jul 2005 from http://www.fbi.gov/pressrel/pressrel01/khobar.htm Vemuri, V. (1978). Modeling of Co mplex Systems: An Introduction. New York: Academic Press. 206

PAGE 216

207 Weil, J. (2002). Weill-Cornell: Comput er simulation model could fill preparedness gap. Cornell Chronicle. Retrieved September 30, 2004 from http://ww.news.cornell.edu/Chroni cle/02/12.5.02/Weill-sim_model.html. Zemke, R. (2001). System s Thinking. Training, 38(2), 40-46. 93rd Congress (1973). S.2410 Bill Summary and Status for the 93rd Congress. Retrieved February 24, 2004, from www.nscot.org 101st Congress (1990). H.R. 1602 Bill Summary and Status for the 101st Congress. Retrieved February 24, 2004, from www.nscot.org

PAGE 217

Appendices

PAGE 218

Appendix 1: Discussion of Additional Proba bility Distributions Available in Arena Software Beta Probability Distribution This distribution is often us ed as a rough model in the absence of data. It has the ability to take on a wide variety of shapes. The probability density function, f (x)= X -1(1-x) -1 for 0
PAGE 219

Appendix 1 (Continued) type, the v isitation sequence, or the batch size for an arriving entity. The probability mass function, p( xj) = cjcj-1 where c0 = 0. Parameters: The discrete function in Arena returns a sample from a user-defined discrete probability distribution. The dist ribution is defined by the set of n possible discrete values (denoted by x1, x2, xn) that can be returned by the function and the cumulative probabilities (denoted by c1, c2, cn) associated with thes e discrete values. The cumulative probability ( cj) for xj is defined as the probabi lity of obtaining a value that is less than or equal to xj. Hence, cj is equal to the sum of p( xk) for k going from 1 to j. By definition, cn = 1. Range: { x1, x2, xn} Erlang Probability Distribution The Erlang distribution is us ed in situations where an activity occurs in successive phases and each phase has an e xponential distribution. For large k the Erlang approaches the normal distribution. The Er lang distribution is often us ed to represent the time required to complete a task. The Erlang di stribution is a special case of the gamma distribution in which the shape parameter, a, is an integer ( k ). The probability density function, f (x)= kxk-1ex/ for x > 0 and 0 if otherwise. ( k -1)! Parameters: If X1, X2, Xk are independent and identically distributed exponential random variables, then the sum of these k samples has an Erlang-k 210

PAGE 220

Appendix 1 (Continued) distribution. The m ean ( ) of each of the component exponential random variables ( k ) are the parameters of the di stribution. The exponential mean is specified as a positive real number, and k is specified as a positive integer. Range: [0, + ] Mean: k Variance: k 2 Gamma Probability Distribution The gamma probability distribution is th e same as the Erlang distribution for integer shape parameters. The gamma is of ten used to represent the time required to complete some task (for example a treatme nt time or equipment repair time). The probability density function, f (x) = x -1ex/ for x > 0 and 0 if otherwise. ( ) Where is the complete gamma function given by ( ) = 0t -1etdt Parameters: The shape parameters ( ) and scale parameter ( ) are specified as positive real values. Range: [0, + ) Mean: Variance: 2 211

PAGE 221

Appendix 1 (Continued) Lognormal Probability Distribut ion The lognormal distribution is used in situ ations where the quantity is the product of a large number of random quantities. It is also frequently used to represent task times that have a distribution skewed to the right. This distribution is related to the normal distributionif X has a Lognormal (l, l) distribution, then ln (X) has a normal (, ) distribution. The probably density function, f (x) = __1 ___ e-(ln( x))(ln( x))/(2 ) for x > 0 and 0 if otherwise. x 2 Parameters: Mean LogMean (l > 0) and standard deviation LogStd (l > 0) is the lognormal random variable. Both LogMean and LogStd must be specified as strictly positive real numbers. Range: [0, + ) Mean: LogMean = l = e+ /2 Variance: (LogStd)2 = l 2 = e2+ (e 1) Poisson Probability Distribution The Poisson distribution is a discrete distribution that is often used to model the number of random events occurr ing in affixed interval of time. If the time between successive events is exponentially distributed, then the number of events that occur in a fixed time interval has a Poisson distribution. The Poisson distributi on is also used to model random batch size. Th e probability mass function, p( x ) = ex for x {0, 1, } and 0 if otherwise. x 212

PAGE 222

Appendix 1 (Continued) Param eters: the mean () specified as a positive real number. Range: {0, 1, } Mean: Variance: 213

PAGE 223

Appendix 2: Bom bings of Fixed Structures Using Conventional Explosives: 1988 to 2005 Note: The table below includes several ex ceptions to the Fixed Structure label. Year Location Target Killed Injured 2005 Hillah, Iraq Military Recruit Center 122 170 2004 Madrid, Spain Trains 202 1400 2004 Taba, Egypt Hotel 34 160 2004 Multan, Pakistan Religious Gathering 40 100 2004 Karachi, Pakistan Mosque 14 38 2004 Moscow, Russia Subway 39 130 2003 Aceh Province, Indonesia Concert 10 45 2003 Khaldiya, Iraq Police Station 17 33 2003 Turkey Synagogue 20 300 2003 Baghdad, Iraq Jordanian Embassy 11 65 2003 Jakarta, Indonesia Hotal 15 150 2003 Mozdok, Russia Military Hospital 35 24+ 2003 Riyadh, Saudi Arabia Residential Compound 20 200 2003 Davao City, Phillipines Ferry Terminal 16 50 2003 Bombay, India Train 10 75 2003 Davao City, Phillipnes Airport Terminal 21 170 2002 Jerusalem, Israel Bus 11 50 2002 Kfar Sava, Israel Shopping Mall 2 69 2002 Tel-Aviv, Israel Bus 6 59 2002 Meron Junction, Israel Bus 10 40 2002 Jerusalem, Israel Bus Stop 7 50 2002 Jerusalem, Israel Bus 20 52 2002 Megiddo Junction, Israel Bus 17 50 2002 Petah Tikvah, Israel Restaurant 2 57 2002 Rishon, LeTzion, Israel Civilians 2 50 2002 Netanya, Israel Marketplace 3 59 2002 Rishon Le Zion, Israel Pool hall 16 60 2002 Jerusalem, Israel Marketplace 6 100 2002 Halfa, Israel Restaurant 35 15 2002 Netania, Israel Hotel 29 150 2002 Jerusalem, Israel Civilians 5 86 2002 Jerusalem, Israel Restaurant 11 50 2002 Jerusalem, Israel Place of Worship 10 40 2002 Jerusalem, Israel Civilians 1 150 2002 Mombassa, Kenya Hotel 13 80 2002 Bali, Indonesia Tourists/Nightclubs 182 250 2002 Karachi, Pakistan U.S. Consulate 11 40 2002 Vallavicencio, Columbia Civilians 12 60 2002 Ambon, Indonesia Civilians 4 58 2002 Zamboanga, Phillipines Shopping Center 5 100 2002 Satkhira, Bangladesh Civilians 10 200 2002 Jeerusalem, Israel School 9 86 2002 Larba, Algeria Civilians 38 86 2002 Dagestan, Russia Civilians 42 150 2002 General Santos, PI Civilians 15 70 214

PAGE 224

Appendix 2: (Continued) Year Location Target Killed Injured 2001 Jerusalem, Israel Civilians 11 188 2001 New York & Washington D.C. Buildings 3600 250 2001 Nahariya, Israel Civilians 3 90 2001 Jerusalem, Israel Restaurant 15 130 2001 Tel Aviv, Israel Nightclub 20 120 2001 Netanya, Israel Shopping Center 6 100 2001 Kfar Sava, Israel Bus Stop 1 60 2001 Netanya, Israel Marketplace 3 50 2001 Hadera, Israel Bus Stop 0 65 2001 Medellin, Columbia Entertainment District 7 138 2001 Jakarta, Indonesia Place of Worship 0 64 2001 Narayangang, Bangladesh Civilians 22 100 2001 Dhaka, Bangladesh Civilians 9 51 2001 Dhaka, Bangladesh Civilians 6 50 2000 Aden Harbor, Yemen U.S. Warship 15 33 2000 Muttur, Sri Lanka Election Rally 24 50 2000 Colombo, Sri Lanka Government Personnel 20 46 2000 Hadera, Israel Bus 2 55 2000 Several Churches in Indonesia Places of Worship 14 100 2000 Islamabad, Pakistan Marketplace 16 50 2000 Moscow, Russia Civilian 7 93 2000 Ozamis, Phillipines Bus 45 30 1999 Volgodonsk, Russia Apartment Bldg. 17 100 1999 Buinaksk, Dagestan, Russia Apartment Bldg. 64 66 1999 Moscow, Russia Civilians 118 150 1999 Moscow, Russia Civilians 94 150 1999 Jalpaiguri Railway Station, India Train 10 59 1999 London, United Kingdom Restaurant 3 65 1999 Vladikavkaz, Russia Marketplace 3 65 1998 Colombo, Sri Lanka Civilians 36 257 1998 Sri Lanka Military Personnel 79 0 1998 Tandy, Sri Lanka Civilians 19 34 1998 Omagh, Northern Ireland Civilians 55 530 1998 Dar es Sala'am, Tanzania U.S. Embassy 10 77 1998 Nairabi, Kenya U.S. Embassy 254 5000 1998 Coimbatore, India Civilians 50 0 1998 Antioquia Dept., Columbia Pipeline/Powerline 71 100 1998 Algiers, Algeria Marketplace 17 60 1998 Coimbatore, India Other/Unknown 43 200 1997 Jerusalem, Israel Shopping Center 8 200 1997 Jerusalem, Israel Marketplace 16 178 1997 Colombo, Sri Lanka Hotel 18 110 1996 Atlanta, Georgia Olympic Park 2 110 215

PAGE 225

Appendix 2: (Continued) Year Location Target Killed Injured 1996 Tel Aviv, Israel Shopping Center 20 75 1996 Jerusalem, Israel Bus 26 80 1996 Manchester, United Kingdom Shopping Center 0 206 1996 Colombo, Sri Lanka Bank 90 1400 1996 Paris, France Subway/Train 4 86 1996 London, United Kingdom Garage 2 100 1995 Oklahoma City Federal Building 168 591 1995 Islamabad, Pakistan Egyptian Embassy 16 60 1995 Jerusalem, Israel Bus 4 100 1995 Bet Lid Junction, Israel Bus Stop 18 69 1995 Paris, France Train 7 86 1994 Tel Aviv, Israel Bus 22 46 1994 Buenos Aires, Argentina Building 100 200 1994 Aula, Israel Bus Stop 8 51 1993 New York Building 6 1040 1992 Buenos Aires, Argentina Israeli Embassy 29 242 1992 Algiers, Algeria Airport 12 128 1991 Barcelona, Spain Military Personnel 9 50 1988 near Islamabad, Pakistan Other/Unknown 93 1000 1988 Chaman, Pakistan Other/Unknown 6 50 1988 Islamabad, Pakistan Civilian 18 60 Source: This table is comprised of data from the International Institute for CounterTerrorism and the New York Times Archives. 216

PAGE 226

Appendix 3: Distribution of Outdoor Stadium s within th e United States (includes domed stadiums) Category Outdoor Number 182 Unknown Capacity 7 Minimum Capacity 30,000 Maximum Capacity 157,000 W/out auto racing minimum 30,000 W/out auto racing maximum 107,501 Stadiums Cum % Cum % 150,000 plus 3 1.60% 100.00% 140,000 149,999 0 1.60% 98.40% 130,000 139,999 1 2.20% 98.40% 120,000 129,999 2 3.30% 96.70% 110,000 119,999 1 3.90% 96.20% 100,000 109,999 6 7.10% 92.90% 90,000 99,999 6 10.00% 89.60% 80,000 89,999 16 21.00% 90.00% 70,000 79,999 24 32.00% 81.00% 60,000 69,999 30 49.00% 68.00% 50,000 59,999 28 64.00% 50.00% 40,000 49,999 30 81.00% 36.00% 30,000 39,999 35 100.00% 19.00% Arithmetic Mean Capacity 61,266 Arithmetic Mean w/out auto racing 57,138 217

PAGE 227

Appendix 4: Distribution of Indoor Arenas within the United States Category Indoor Number 130 Minimum Capacity 10,000 Maximum Capacity 24,535 Arenas Cum % Cum % 24,000 plus 1 0.80% 100.00% 23,000 23,999 3 3.10% 99.20% 22,000 22,999 2 4.60% 96.90% 21,000 21,999 2 6.20% 95.40% 20,000 20,999 9 13.10% 93.90% 19,000 19,999 16 25.40% 86.90% 18,000 18,999 13 35.40% 74.60% 17,000 17,999 12 44.60% 64.60% 16,000 16,999 7 50.00% 55.40% 15,000 15,999 11 58.50% 50.00% 14,000 14,999 8 64.60% 41.50% 13,000 13,999 11 73.10% 35.40% 12,000 12,999 11 81.50% 18.50% 11,000 11,999 10 89.20% 18.50% 10,000 10,999 14 100.00% 10.80% Arithmetic Mean Capacity 15,830 218

PAGE 228

Appendix 5: Exam ple of Resource Dense Geographic Locatio n where multiple/duplicate resources are available in close proximity to the locati on of the large scale mass casualty event. Legend: TC B-450 ED-25 OR-20 N-338 TC B-500 ED-45 OR-32 N-375 H B-220 ED-10 OR-8 N-165 A-3 A-5 A-7 GS-6 OS-2 O-3 X H B-180 ED-6 OR-4 N-135 H B-260 ED-20 OR-10 N-195 TC B-460 ED-30 OR-24 N-345 H B-220 H B-310 ED-25 OR-8 N-233 TC H B-260 ED-15 OR-6 N-195 A6 A-8 20 mile radius 10 mile radius A-2 GS-10 OS-6 O-5 GS-8 OS-6 O-2 GS-12 OS-8 O-4 GS-3 OS-1 O-1 GS-7 OS-5 O-2 B-380 E D-20 BU-10 OR-22 N-300 E D-20 OR-12 N-165X = Locatio n of mass casualty event A = Ambulances TC = Trauma Center GS = General Surgeons B = Inpatient Beds OS = Orthopedic Surgeons ED = Emergency Department Beds O = Ophthalmologists BU = Burn Unit Beds N = Nurses OR = Operating Rooms 219

PAGE 229

Appendix 6: Exam ple of a Resource Sparse Geographic Loca tion where resources are very limited in the area surrounding the site of the large scale mass casualty event. Legend: 10 mile radius A-3 H B-310 ED-25 OR-8 N-233 A-2 GS-10 OS-6 O-5 GS-7 OS-5 O-2 50 mile radius H B-260 ED-15 OR-6 N-195 H X GS-6 OS-2 O-3 B-220 E D-10 OR-8 N-165 A-3 X = Locatio n of mass casualty event A = Ambulances TC = Trauma Center GS = General Surgeons B = Inpatient Beds OS = Orthopedic Surgeons ED = Emergency Department Beds O = Ophthalmologists BU = Burn Unit Beds N = Nurses OR = Operating Rooms 220

PAGE 230

Appendix 7: Barrell Injury Diagnosis Matrix 221

PAGE 231

Appendix 8: The Barell Injury Diagnosis Matrix, Classi fication by Body Region and Nature of the Injury Nature of the Injury ICD-9-CM Code Description Body Region Fracture 800 Fracture of vault of skull Head and Neck 801 Freacture of base of skull Head and Neck 803 Other and unqualified skull fractures Head and Neck 804 Multiple fractures in volving skull or face with other bones Head and Neck 802 Fracture of face bones Head and Neck 807 Fracture of rib(s), sternum, larynx, and trachea Head and Neck/Torso 806 Fracture of vertebral column with spinal cord injury Spine and Back 805 Fracture of vertebral column without mention of spinal cord injury Spine and Back 808 Fracture of pelvis Torso 809 Ill-defined fractures of bones of trunk Torso 810 Fracture of clavical Extremities 811 Fracture of scapula Extremities 812 Fractcure of humerus Extremities 813 Fracture of radius and ulna Extremities 814 Fracture of carpal bone(s) Extremities 815 Fracture of metacarpal bone(s) Extremities 816 Fracture of one or more phalanges of hand Extremities 817 Multiple fractures of hand bones Extremities 818 Ill-defined fractures of upper limb Extremities 820 Fracture of neck of femur Extremities 821 Fracture of other and unspecified parts of femur Extremities 822 Fracture of patella Extremities 823 Fracture of tibia and fibula Extremities 824 Fracture of ankle Extremities 825 Fracture of one or more tarsal or metatarsal bones Extremities 826 Fracture of one or more phalanges of foot Extremities 827 Other, multiple, and ill-defined fractures of lower limb Extremities 819 Multiple fractures involving both upper limbs, and upper imb with rib(s) and sternum Unclassifiable by Site 828 Multiple fractures involving both lower limbs, lower with upper limb, and lower limb(s) with rib(s) and sternum Unclassified by site 222

PAGE 232

Appendix 8: (Continued) Fracture (continued) 829 Fracture of unspecified bones Unclassified by site 839 Other, multiple, and ill-defined dislocation Spine and Back/Unclassifiable by Site 831 Dislocation of shoulder Extremities 832 Dislocation of elbow Extremities 833 Dislocation of wrist Extremities 834 Dislocation of finger Extremities 835 Dislocation of hip Extremities 836 Dislocation of knee Extremities 837 Dislocation of ankle Extremities 838 Dislocation of foot Extremities Sprains and Strains 848 Other and ill-defined sprain s and strains Head and Neck/Torso 847 Sprains and strains of other and unspecified parts of back Spine and Back/Torso/ Unclassifiable by Site 846 Sprains and strains of sacroiliac region Torso 840 Sprains and strains of shoulder and upper arm Extremities 841 Sprains and strains of elbow and forearm Extremities 842 Sprains and strains of wrist and hand Extremities 843 Sprains and strains of hip and thigh Extremities 844 Sprains and strains of knee and leg Extremities 845 Sprains and strains of ankle and foot Extremities Internal 850 Concussion Head and Neck 851 Cerebral laceration and contusion Head and Neck 852 Subarachnoid, subdural, and extradural hemorrhage, following injury Head and Neck 853 Other and unspecified intracranial hemorrhage following injury Head and Neck 854 Intracranial injury of other and unspecified nature Head and Neck 952 Spinal cord injury without evidence of spinal bone injury Spine and Back 860 Traumatic pneumothorax and hemothorax Torso 861 Injury to heart and lung Torso 862 Injury to other and unspecified intrathoracic organs Torso 863 Injury to gastrointestinal track Torso 864 Injury to liver Torso 865 Injury to spleen Torso 223

PAGE 233

Appendix 8: (Continued) Internal 866 Injury to kidney Torso 868 Injury to other intra-abdominal organs Torso 872 Open wound of ear Head and Neck 870 Open wound of ocular adnexa Head and Neck 871 Open wound of eyeball Head and Neck 874 Open wound of neck Head and Neck 875 Open wound of chest (wall) Torso 879 Open wound of other and unspecified sites, except limbs Torso/Unclassifiable by Site 877 Open wound of buttock Torso 878 Open wound of other and unspecified sites, except limbs Torso 876 Open wound of back Torso 880 Open wound of shoulder and upper arm Extremities 881 Open wound of elbow, forearm, and wrist Extremities 882 Open wound of hand except finger(s) alone Extremities 883 Open wound of finger(s) Extremities 884 Multiple and unspecified open wound of upper limb Extremities 892 Open wound of foot except toe(s) alone Extremities 893 Open wound of toe(s) Extremities 890 Open wound of hip and thigh Extremities 891 Open wound of knee, leg (except thigh), and ankle Extremities 894 Multiple and unspecified open wound of lower limb Extremities Amputations 887 Traumatic amputation of arm and hand (complete)(partial) Extremities 885 Traumatic amputation of thumb (complete)(partial) Extremities 886 Traumatic amputation of other finger(s) (complete)(partial) Extremities 897 Traumatic amputation of leg(s) (complete)(partial) Extremities 895 Traumatic amputation of toe(s) (complete)(partial) Extremities 896 Traumatic amputation of foot (complete)(partial) Extremities 224

PAGE 234

Appendix 8: (Continued) Blood Vessels 900 Injury to blood vessels of head and neck Head and Neck 901 Injury to blood vessels of thorax Torso 902 Injury to blood vessels of abdomen and pelvis Torso/Unclassifiable by Site 903 Injury to blood vessels of upper extremity Extremities 904 Injury to blood vessels of lower extrimity and unspecified sites Extremities /Unclassifiable by Site Contusion/ Superficial 918 Superficial inju ry of eye and adnexa Head and Neck 921 Contusion ofeye and adnexa Head and Neck 910 Superficial injury of face, neck, and scalp except eye Head and Neck 920 Contusion of face, scalp, and neck except eye(s) Head and Neck 922 Contusion of trunk Torso 911 Superficial injury of trunk Torso 912 Superficial injury of shoulder and upper arm Extremities 923 Contusion of upper limb Extremities 914 Superficial injury of hand(s) except finger(s) alone Extremities 915 Superficial injury offinger(s) Extremities 913 Superficial injury of elbow, forearm, and wrist Extremities 924 Contusion of lower limb and of other and unspecified sites Extremities/ Unclassifiable by Site 917 Superficial injury of foot and toe(s) Extremities 916 Superficial injury of hip, thigh, leg, and ankle Extremities 919 Superficial injury of other Unclassifiable by Site 225

PAGE 235

Appendix 8: (Continued) Crush 925 Crushing injury of face, scalp, and neck Head and Neck 926 Crushing injury of trunk Torso 927 Crushing injury of upper limb Extremities 928 Crushing injury of lower limb Extremities 929 Crushing injury of multiple and unspecified sites Unclassifiable by Site Burns 941 Burn of face, head, and neck Head and Neck 940 Burn confiend to eye and adnexa Head and Neck 947 Burn of internal organs Head and Neck/Torso/ Unclassifiable by Site 942 Burn of trunk Torso 943 Burn of upper limb, except wrist and hand Extremities 944 Burn of wrist(s) and hand(s) Extremities 945 Burn of lower limb(s) Extremities 946 Burn of multiple specified sites Unclassifiable by Site 948 Burns classified according to extent of body surface involved Unclassifiable by Site 949 Burn, enspecified Unclassifiable by Site 226

PAGE 236

Appendix 8: (Continued) Nerves 950 Injury to optic nerve and pathways Head and Neck 951 Injury to other cranial nerve(s) Head and Neck 953 Injury to nerve roots and spinal plexus Head and Neck/Torso/ Extremities/ Unclassifiabale by Site 954 Injury to other nerve(s) of trunk, excluding shoulder and pelvic girdles Head and Neck/Torso 957 Injury to other and unspecified nerves Head and Neck/ Unclassifiable by Site 955 Injury to peripheral nerve(s) of pelvic girdle and lower limb Extremities 956 Injury to peripheral nerve(s) of pelvic girdle and lower limb Unclassifiable by Site Unspecified 959 Injury, other and unspecified Head and Neck Undefinable by Site -System Wide 930-939 Effects of foreign body entering through orifice 930 Foreign body on external eye 931 Foreign body in ear 932 Foreign body in nose 933 Foreign body in pharynx 934 Foreign body in trachea, bronchus, and lung 935 Foreign body in mouth, esophagus, and stomach 936 Foreign body in intestine and colon 937 Foreign body in anus and rectum 938 Foreign body in digestive system, unspecified 939 Foreign body in genitourinary tract 958 Certain early complication of trauma 960-979 Poisoning by drugs, medicinal and biologic substances 980-989 Toxic effects of substances chiefly nonmedical as to source 990-995 Other and unspecified effects of external causes 905-909 Late effects of injuries, poisoning, toxic effects, and other external causes 227

PAGE 237

Appendix 9: Matrix of Studies Identifying Injury by Body Re gion Study Author/Date Head & Neck Spine & Back Torso Extremities Unclassified by Site CDC/2005 (no %'s, only categories) Eye, Orbit, Face', CNS, Auditory (3 categories) CNS Respiratory, Digestive, Circulatory, Renal (4 categories) Extremity Frykberg/1988 Head (31.4%) Chest (2%), Abdomen (1.4%) Bony Extremities (10.9%) Mallonee/1996 Head (14%), Ocular (1%) Mallonee/1996 (location of soft tissue injuries) Head & Neck (48%), Face (45%) Chest (35%) Extremities (74%) Mallonee/1996 (location of Fractures/ Dislocations) Face & Neck (37%) Back, Chest, or Pelvis (25%) Back, Chest, or Pelvis (25%) Legs (40%), Arms (38%) Mallonee/1996 (Location of Sprains) Neck (29%) Chest & Back (53%) Chest & Back (53%) Extremities (legs) (27%), Extremities (arms) (9%) Peleg/2004 Brain (18.5%), Other Head (53.8%) Spinal Cord & Column (4.8%) Chest (12%), Abdomen (21.7%), Pelvis/Trunk (12%) Upper Extremities (38.8%), Lower Extremities (37.6%) MMWR/2002 Ocular (26%), Closed Head (2%) Thompson/2004 Eye (6.7%), Brain (6.2%) (xx%) = % of total injuries. 228

PAGE 238

Appendix 10: Matrix of Studies Identifying Injury by Injury Type Study Author/Date Injury Type CDC/2005 (no %'s, categories only) Frykberg/1988 Malonee/1996 MMWR/2002 Thompson/2004 Fracture Extremity, CNS (2 categories) Fractures/ Dislocations (1%) Fracture (6%) Fractures & Dislocations (4.2%) Dislocation Extremity, CNS (2 categories) Fractures/ Dislocations (1%) Fractures & Dislocations (4.2%) Sprains & Strains Extremity, CNS (2 categories) Sprains (25%) Sprain or Strain (14%) Strains or Sprains (13.5%) Internal Auditory, Respiratory, Renal, Digestive, Circulatory (5 categories) Blast Lung (0.6%) Inhalation (49%), Closed Head (2%) Brain (6.2%) Open Wound Extremity Soft Tissue (55.4%) Soft Tissue (85%) Laceration (14%) Soft Tissue (94.8%), Severe Lacerations (9%) Amputations Extremity Traumatic Amputation (1.2%) Blood Vessels Extremity Contusion/ Superficial Extremity Soft Tissue (55.4%) Soft Tissue (85%) Laceration (14%), Contusion (12%) Soft Tissue (94.8%) Crush Extremity, CNS (2 categories) Crush (1%) Burns Extremity Burns (5%) Burns (2%) Burn (5%) Nerves Extremity, CNS (2 categories) Unspecified (xx%) = % of total injuries 229

PAGE 239

Appendix 11: In Hospital Em ergency Response Models Hirshberg Model (adapted from Hirshberg, 1999) Start Triage in ER ER Care Process Shock Room Care X-Ray Process Additional Care Process Exit CT Scanner Surgery Process ICU ICU Exit 230

PAGE 240

Appendix 11: (Continued) Sha mir Model (adapted from Shamir, 2004) Ambulance Bay Triage ED Trauma Uni t ED Care CT Scanners Operating Rooms Angiography PACU Surgical ICUs Beds 231

PAGE 241

Appendix 11: (Continued) Al mogy Model (adapted from Almogy, 2004) Hospital Transfer EMS from Scene Surgeon in Charge Admitting Area Trauma Room OR Surgeon in Charge Imaging OR ICU Subspecialties Surgion in Charge Admitting Floor Non-EMS Evacuation 232

PAGE 242

Appendix 12: Process and Decision Para m eters as Defined in Hirshberg Study Process/Decision Parameter Triage Mean = 20 Seconds Care in Shock Room (Stable Patients) Mean = 47 minutes (range 14-89) Care in Shock Room (Unstable Patien ts) Mean = 22 minutes (range 3-52) Stable/Unstable Patients 73%/27% ER Time for Initial Care Mean = 41 minutes (range 22-116) ER Time for Post-imaging care Mean = 53 minutes (range 29-137) Routing of New Arrivals 69% to x-ray/31% to exit X-Ray Time Mean = 19 minutes (range 11-158) Operating Room Time Mean = 97 minutes (range 47-218) Source: Hirshberg, 1999 233

PAGE 243

Appendix 13: Distribution of Arenas and Stadiu ms by State State # of Arenas # of Stadiums Alabama 3 4 Alaska 0 0 Arizona 3 4 Arkansas 2 2 California 10 13 Colorado 1 6 Connecticut 0 0 Delaware 0 1 District of Columbia 1 1 Florida 8 11 Georgia 2 5 Hawaii 0 1 Idaho 3 1 Illinois 2 7 Indiana 4 5 Iowa 3 2 Kansas 3 2 Kentucky 2 2 Louisiana 4 5 Maine 0 0 Maryland 1 5 Massachusetts 2 4 Michigan 3 8 Minnesota 2 1 Mississippi 1 3 Missouri 3 5 Montana 0 0 Nebraska 1 1 Nevada 3 3 New Hampshire 0 1 New Jersey 1 2 New Mexico 2 2 New York 4 6 North Carolina 6 8 North Dakota 2 0 Ohio 5 8 Oklahoma 2 3 Oregon 2 3 Pennsylvania 6 9 234

PAGE 244

Appendix 13: (Continued) State # of Arenas # of Stadiums Rhode Island 1 0 South Carolina 2 4 South Dakota 2 0 Tennessee 5 7 Texas 9 15 Utah 4 2 Vermont 0 0 Virginia 2 4 Washington 2 3 West Virginia 2 2 Wisconsin 2 5 Wyoming 1 1 235

PAGE 245

Appendix 14: Arenas and Stadium s by State S eating State/Name Location Population Capacity Alabama Jordan Hare Stadium Auburn 115,092 86,063 Beard-Eaves Memorial Coliseum Auburn 115,092 10,500 Bryant-Denny Stadium Tuscaloosa 192,034 83,818 Coleman Coliseum Tuscaloosa 192,034 15,043 Legion Field Birmingham 1,052,238 80,601 Mobile Civic Center Mobile 399,843 10,112 Talladega Superspeedway Talladega 15,143 108,000 Arizona Arizona Stadium Tucson 843,746 57,803 McHale Center Tucson 843,746 14,545 Sun Devil Stadium Tempe 3,251,876 73,273 Wells Fargo Arena Tempe 3,251,876 14,198 America West Arena Phoenix 3,251,876 19,023 Bank One Ballpark Phoenix 3,251,876 49,075 Phoenix International Raceway Phoenix 3,251,876 78,000 Arkansas Convocation Center Jonesboro 107,312 10,529 Bud Walton Arena Fayetteville 347,045 20,000 War Memorial Stadium Little Rock 610,518 53,727 Razorback Stadium Fayetteville 347,045 72,000 California Spartan Stadium San Jose 1,735,819 30,000 Bulldog Stadium Fresno 799,407 41,031 Save Mart Center Fresno 799,407 13,800 Stanford Stadium Palo Alto 4,123,740 85,500 Memorial Stadium Berkeley 4,123,740 75,662 Walter A Hass Jr. Pavilion Berkeley 4,123,740 11,877 Los Angeles Coliseum Los Angeles 12,365,627 92,000 The Rose Bowl Pasadena 12,365,627 91,136 Pauley Pavilion Los Angeles 12,365,627 12,819 Memorial Sports Arena Los Angeles 12,365,627 15,000 Pacific Bell Park San Francisco 4,123,740 40,800 Edison International Field Anaheim 12,365,627 45,050 Qualcomm Stadium San Diego 2,813,833 71,000 Cox Arena San Diego 2,813,833 12,000 Staples Center Los Angeles 12,365,627 16,021 Dodger Stadium Los Angeles 12,365,627 56,000 Arrowhead Pond Anaheim 12,365,627 17,174 Arco Arena Sacramento 1,796,857 17,317 236

PAGE 246

Appendix 14: (Continued) S eating State/Name Location Population Capacity California (continued) Network Associates Coliseum Oakland 4,123,740 63,146 The Compaq Center San Jose 1,735,819 17,496 The Arena in Oakland Oakland 4,123,740 19,596 3 Com Park San Francisco 4,123,740 69,734 California Speedway Fontana 3,254,821 118,000 Colorado Hughes Stadium Ft Collins 251,494 35,000 Folsom Field Boulder 291,288 53,245 Falcon Field Colorado Springs 537,484 52,123 Pepsi Center Denver 2,196,028 18,129 Mile High Stadium Denver 2,196,028 76,098 Coors Field Denver 2,196,028 50,381 Pikes Peak International R aceway Fountain 537,484 42,000 Delaware Dover International Speedway Dover 126,697 107,000 District of Columbia MCI Center Washington 4,796,183 20,600 RFK Stadium Washington 4,796,183 56,000 Florida Citrus Bowl Orlando 1,644,561 70,300 Miami Arena Miami 5,007,564 15,200 Doak S. Campbell Stadium Tallahassee 320,304 80,000 Gator Bowl Jacksonville 1,122,750 82,000 Leon County Civic Center Tallahassee 320,304 13,500 Orange Bowl Miami 5,007,564 72,319 The Sundome Tampa 2,395,997 11,300 Ben Hill Grifin Stadium Gainesville 232,392 85,000 Stephen P. OConnell Center Gainesville 232,392 12,000 Raymond James Stadium Tampa 2,395,997 66,321 Tropicana Stadium St. Petersburg 2,395,997 45,000 Pro Player Stadium Miami 5,007,564 74,916 American Airlines Arena Miami 5,007,564 19,600 Alltell Stadium Jacksonville 1,122,750 73,000 Ice Palace Tampa 2,395,997 19,764 TD Waterhouse Centre Orlando 1,644,561 17,248 National Care Rental Center Sunrise 85,779 19,088 Daytona International Speedway Daytona Beech 64,112 150,000 Homestead-Miami Speedway Homestead 31,909 60,000 237

PAGE 247

Appendix 14: (Continued) S eating State/Name Location Population Capacity Georgia Bobby Dodd Stadium Atlanta 4,247,981 55,000 Sanford Stadium Athens 166,079 92,746 Stegeman Coliseum Athens 166,079 10,523 Turner Field Atlanta 4,247,981 49,831 Georgia Dome Atlanta 4,247,981 71,228 Philips Arena Atlanta 4,247,981 19,455 Atlanta Motor Wpeedway Hampton 4,247,981 Hawaii Aloha Stadium Honolulu 876,156 50,000 Idaho Holt Arena Pocatello 83,103 12,000 Bronco Stadium Boise 464,840 30,000 Taco Bell Arena Boise 464,840 12,380 Kibbie Dome Moscow 34,935 16,000 Illinois Huskie Stadium DeKalb 8,453,960 30,076 Memorial Stadium Champaign 210,275 70,000 Assembly Hall Champaign 210,275 16,000 Ryan Field Evanston 8,453,960 49,256 Soldier Field Chicago 8,453,960 66,944 United Center Chicago 8,453,960 20,500 Wrigley Field Chicago 8,453,960 38,765 Comiskey Park Chicago 8,453,960 44,321 Gateway International Raceway Madison 2,721,491 45,000 Indiana Memorial Stadium Bloomington 175,506 52,000 Allen County Memorial Coliseum Fort Wayne 390,156 10,000 Assembly Hall Bloomington 175,506 17,484 Notre Dame Stadium South Bend 316,663 80,795 Joyce Center South Bend 316,663 11,418 Ross Ade Stadium West Lafayette 178,541 68,000 RCA Dome Indianapolis 1,525,104 60,272 Conseco Fieldhouse Indianapolis 1,525,104 18,340 Indianapolis Motor Speedway Indianapolis 1,525,104 250,000 Iowa Kinnick Stadium Iowa City 131,676 70,000 Carver-Hawkeye Stadium Iowa City 131,676 15,500 Jack Trice Stadium Ames 79,981 50,000 238

PAGE 248

Appendix 14: (Continued) S eating State/Name Location Population Capacity Iow a (continued) James H. Hilton Coliseum Ames 79,981 14,023 Veterans Memorial Auditorium Des Moines 481,394 11,277 Kansas KSU Stadium Manhattan 108,999 50,300 Ahearn Fieldhouse Manhattan 108,999 11,700 Bramlage Coliseum Manhattan 108,999 13,500 Kansas Memorial Stadium Lawrence 99,962 50,250 Allen Fieldhouse Lawrence 99,962 16,300 Kentucky Commonwealth Stadium Lexington-Fayette 408,326 67,600 Adolph Rupp Arena Lexington-Fayette 408,326 23,000 Freedom Hall Louisville 1,161,975 18,865 Papa Johns Cardinal Stadium Louisville 1,161,975 42,000 Louisiana Malone Stadium Monroe 170,053 30,427 Joe Aillet Stadium Ruston 227,959 30,600 New Orleans Arena New Orleans 1,316,510 18,000 The Cajun Dome Lafayette 239,086 11,500 Cajun Field Lafayette 239,086 31,000 Maravich Assembly Center Baton Rouge 705,973 14,500 UNO Lakefront Arena New Orleans 1,316,510 10,000 Tiger Stadium Baton Rouge 705,973 91,644 Louisiana Superdome New Orleans 1,316,510 69,082 Maryland Navy-Maine Corps Stadium Annapolis 38,838 35,000 Byrd Stadium College Park 4,796,183 62,000 Comcast Center College Park 4,796,183 17,950 FedEx Field Landover 4,796,183 80,116 Oriole Park at Camden Yards Baltimore 2,552,994 48,262 PSInet Stadium Baltimore 2,552,994 69,084 Massachusetts Alumni Stadium Boston 4,391,344 44,500 Harvard Stadium Cambridge 4,391,344 36,739 Worcester Centrum Worcester 750,963 14,800 Fleet Center Boston 4,391,344 19,600 Foxboro Stadium Foxboro 4,391,344 60,292 Fenway Park Boston 4,391,344 33,871 239

PAGE 249

Appendix 14: (Continued) S eating State/Name Location Population Capacity Michigan Waldo Stadium Kalamazoo 314,866 30,200 Rynearson Stadium Ypsilanti 22,362 30,200 Kelly-Shorts Stadium Mount Pleasant 25,946 30,199 Crysler Stadium Ann Arbor 322,895 13,751 Michigan Stadium Ann Arbor 322,895 107,501 Spartan Stadium East Lansing 447,728 72,027 Pontiac Silverdome Pontiac 4,452,557 80,311 Palace of Auburn Hills Auburn Hills 4,452,557 22,076 Joe Louis Arena Detroit 4,452,557 19,275 Comereica Park Detroit 4,452,557 40,000 Cobb Arena Detroit 4,452,557 12,191 Michigan Speedway Brooklyn 158,422 126,000 Minnesota Target Center Arena Minneapolis 2,968,806 19,006 Hubert H. Humphrey Metrodome Minneapolis 2,968,806 64,121 Williams Arena Minneapolis 2,968,806 14,625 Mississippi Vaught Hemingway Stadium Oxford 37,744 60,850 Roberts Stadium Hattiesburg 123,812 33,000 Scott Field Starkville 42,902 55,082 Humphrey Coliseum Starkville 42,902 10,500 Missouri Kemper Arena Kansas City 1,836,038 17,500 Faurot Field Columbia 145,666 62,000 Mizzou Arena Columbia 145,666 15,000 Savvis Center St. Louis 2,721,491 21,000 Busch Stadium St. Louis 2,721,491 57,673 Arrowhead Stadium Kansas City 1,836,038 79,409 Trans World Dome St. Louis 2,721,491 66,000 Kauffman Stadium Kansas City 1,836,038 40,625 Nebraska Memorial Stadium Lincoln 266,787 73,198 Bob Devaney Sprots Center Lincoln 266,787 13,500 Nevada Sam Boyd Stadium Las Vegas 1,375,765 36,800 Mackay Stadium Reno 342,885 31,545 MGM Grand Arena Las Vegas 1,375,765 15,200 Thomas and Mack Center Las Vegas 1,375,765 18,000 240

PAGE 250

Appendix 14: (Continued) S eating State/Name Location Population Capacity Nevada (continued) Lawlor Events Center Reno 342,885 11,600 Las Vegas Motor Speedway Las Vegas 1,375,765 120,000 New Hampshire NH International Speedway Loudon 4,481 90,000 New Jersey Rutgers Stadium Piscataway 50,482 42,000 Continental Arena East Rutherford 16,669,062 19,040 Giants Stadium East Rutherford 16,669,062 79,469 New Mexico University Arena Albuquerque 729,649 18,018 University Stadium Albuquerque 729,649 37,370 Aggie Memorial Stadium Las Cruces 174,682 40,000 Pan American Center Las Cruces 174,682 13,000 New York UB Stadium Buffalo 1,170,111 31,000 Pepsi Arena Albany 825,875 17,500 Carrier Dome Syracuse 650,154 49,250 Michie Stadium West Point 7,138 39,929 Ralph Wilson Stadium Orchard Park 1,170,111 75,339 Nassau Veterans Coliseum Uniondale 16,669,062 16,297 Madison Square Garden New York City 16,669,062 19,763 Shea Stadium Flushing/Queens 16,669,062 55,777 HSBC Arena Buffalo 1,170,111 18,595 Yankee Stadium Bronx 16,669,062 57,546 North Carolina Lawrence Joel Coliseum Winston-Salem 421,961 14,407 Carter Finley Stadium Raleigh 797,071 51,500 RBC Center Raleigh 797,071 19,722 Reynolds Coliseum Raleigh 797,071 12,400 Dowdy Ficklen Stadium Greenville 152,772 43,000 Keenan Stadium Chapel Hill 426,493 60,000 Dean E. Smith Center Chapel Hill 426,493 20,000 Wallace Wade Stadium Durham 426,493 33,941 Groves Stadium Winston-Salem 421,961 31,500 Charlotte Coliseum Charlotte 1,330,448 23,698 Entertainment and Sports Arena Raleigh 797,071 19,000 Ericsson Stadium Charlotte 1,330,448 73,250 Lowes Motor Speedway Concord 1,330,448 157,000 241

PAGE 251

Appendix 14: (Continued) S eating State/Name Location Population Capacity North Carolina (contin ued) North Carolina Speedway Rockingham 9,672 North Dakota The Fargodome Fargo 174,367 18,700 Ralph Engelstad Arena Grand Forks 97,478 11,406 Ohio Rubber Bowl Stadium Akron 694,960 35,202 Fred C. Yager Stadium Oxford 21,943 30,012 Dix Stadium Kent 27,906 30,520 Nippert Stadium Cincinnati 2,009,632 38,000 Ohio Stadium Columbus 1,612,694 101,568 St. John Arena Columbus 1,612,694 13,276 Value City Arena Columbus 1,612,694 19,500 Shoemaker Center Cincinnati 2,009,632 13,176 Paul Brown Stadium Cincinnati 2,009,632 65,600 Cinergy Field Cincinnati 2,009,632 39,000 Cleveland Browns Stadium Cleveland 2,148,143 73,200 The Gund Arena at Gateway Cleveland 2,148,143 20,750 Jacobs Field Cleveland 2,148,143 42,865 Nationwide Field Columbus 1,612,694 18,500 Oklahoma Skelly Stadium Tulsa 859,532 40,385 Boone Pickens Stadium Stillwater 68,190 53,000 Gallagher-Iba Arena Stillwater 68,190 13,611 Ownen Field at Memorial Stadium Norman 1,095,421 81,207 Lloyd Noble Center Norman 1,095,421 12,000 Oregon Reser Stadium Corvallis 78,153 35,362 Gill Coliseum Corvallis 78,153 10,400 Autzen Stadium Eugene 322,959 54,000 The Rose Garden Portland 1,927,881 21,300 Portland International Raceway Portland 1,927,881 86,000 Pennsylvania Wachovia Center Philadelphia 5,687,147 20,000 Lincoln Financial Field Philadelphia 5,687,147 66,000 Heinz Field Pittsburgh 2,431,288 64,450 Peterson Events Center Pittsburgh 2,431,288 12,500 Franklin Field Philadelphia 5,687,147 52,000 Beaver Stadium State College 135,758 107,282 242

PAGE 252

Appendix 14: (Continued) S eating State/Name Location Population Capacity Pennsylvania (continued) Bryce Jordan Center State College 135,758 15,261 The Apollo of Temple Philadelphia 5,687,147 10,201 Veterans Stadium Philadelphia 5,687,147 65,352 First Union Center Philadelphia 5,687,147 19,500 Civic Arena Pittsburgh 2,431,288 17,181 Three Rivers Stadium Pittsburgh 2,431,288 59,600 PNC Park Pittsburgh 2,431,288 38,127 Nazareth Speedway Nazareth 6,023 Pocono Raceway Pocono 9,607 Rhode Island Providence Civic Center Providence 1,582,997 23,150 South Carolina Memorial Stadium Clemson 11,939 86,400 Littlejohn Coliseum Clemson 11,939 10,980 Williams Brice Stadium Columbia 647,158 80,250 The Colonial Center Columbia 647,158 18,000 Darlington Raceway Darlington 193,155 Myrtle Beech Speedway Myrtle Beech 196,629 South Dakota Dakota Dome Vermillion 9,765 10,000 Rushmore Plaza Civic Center Rapid City 112,818 10,000 Tennessee Floyd Stadium Murfreesboro 68,816 31,000 Mid-South Coliseum Memphis 1,205,204 11,667 Vanderbilt Stadium Nashville 1,311,789 41,203 Liberty Bowl Memphis 1,205,204 62,380 The Pyramid Memphis 1,205,204 20,142 Neyland Stadium Knoxville 616,079 104,079 Thompson-Boling Arena Knoxville 616,079 24,535 The McKenzie Arena Chattanooga 476,531 11,218 Gaylord Entertainment Cent er Nashville 1,311,789 20,000 Adelphia Coliseum Nashville 1,311,789 67,000 Bristol Motor Speedway Bristol 24,821 135,000 Nashville Speedway USA Nashville 1,311,789 50,000 Texas SBC Center San Antonio 1,711,703 18,500 Toyota Center Houston 4,715,407 18,300 American Airlines Center Dallas 5,161,544 20,000 243

PAGE 253

Appendix 14: (Continued) S eating State/Name Location Population Capacity Texas (continued) Fouts Field Wichita Falls 151,524 30,000 Sun Bowl El Paso 679,622 50,426 Rice Stadium Houston 4,715,407 70,000 Gerald J. Ford Stadium Dallas 5,161,544 32,000 Amon Carter Stadium Fort Worth 5,161,544 46,000 Jones SBC Stadium Lubbock 249,700 50,000 United Spirit Arena Lubbock 249,700 15,000 Floyd Casey Field Waco 213,517 50,000 The Ferrell Center Waco 213,517 15,000 Kyle Field College Station 184,885 82,600 Reed Arena College Station 184,885 12,500 Robertson Stadium Houston 4,715,407 32,000 Darrell K. Royal Stadium Austin 1,249,763 80,082 Frank Erwin Special Events Center Austin 1,249,763 16,755 Enron Field Houston 4,715,407 42,000 Astrodome Houston 4,715,407 54,350 Compaq Center Houston 4,715,407 16,279 Texas Stadium Irving 5,161,544 65,675 Reunion Arena Dallas 5,161,544 17,007 Arlington Stadium Arlington 5,161,544 49,166 Alamodome San Antonio 1,711,703 65,000 Texas Motor Speedway Justin 5,161,544 154,861 Utah Dee Events Center Ogden 442,656 12,000 Rice-Eccles Stadium Salt Lake City 9,8,858 45,634 Jon M. Huntsman Center Salt Lake City 968,858 15,000 Cougar Stadium Provo 376,774 65,000 Marriott Center Provo 376,774 22,700 Delta Center Arena Salt Lake City 968,858 19,911 Virginia Scott Stadium Charlottsville 174,021 60,000 Hampton Coliseum Hampton 1,564,804 13,800 Lane Stadium/Worsham Field Blacksburg 151,272 65,115 Cassell Coliseum Blacksburg 151,272 10,052 Martinsville Speedway Martinsville 15,416 70,000 Richmond International Speedway Richmond 1,096,957 95,000 Washington Martin Stadium Pullman 40,740 40,000 244

PAGE 254

Appendix 14: (Continued) S eating State/Name Location Population Capacity Washington (continued) Safeco Field Seattle 3,043,878 47,116 Husky Stadium Seattle 3,043,878 72,500 Bank of America Arena Seattle 3,043,878 10,000 Key Arena Seattle 3,043,878 17,000 West Virginia Charleston Civic Center Charleston 549,033 13,600 Marshall University Stadium Huntington 288,649 38,016 Mountaineer Field Morgantown 26,809 63,500 West Virginia University Coliseum Morgantown 26,809 14,000 Wisconsin Kohl Center Madison 501,774 17,142 Camp Randall Stadium Madison 501,774 79,500 Milwaukee County Stadium Milwaukee 1,500,561 53,192 Bradley Center Arena Milwaukee 1,500,561 18,633 Lambeau Field Green Bay 282,599 60,890 Miller Park Milwaukee 1,500,561 42,500 Wisconsin (continued) The Milwaukee Mile West Allis 1,500,561 45,000 Wyoming War Memorial Stadium Laramie 32,014 33,500 Arena-Auditorium Laramie 32,014 15,000 245

PAGE 255

Appendix 15: Distribution of I ndoor Arenas by Location Population Cum % Cum % Population Interval Frequency % of Total Ascending Descending < 50,000 6 4.62 4.62 100.00 60,000 99,999 7 5.38 10.00 95.38 100,000 149,999 8 6.15 16.15 90.00 50,000 199,999 7 5.38 21.54 83.85 200,000 249,999 5 3.85 25.38 78.46 250,000 299,999 1 0.77 26.15 74.62 300,000 349,999 5 3.85 30.00 73.85 350,000 399,999 3 2.31 32.31 70.00 400,000 449,999 4 3.08 35.38 67.69 450,000 499,999 3 2.31 37.69 64.62 500,000 549,999 2 1.54 39.23 62.31 550,000 599,999 0 0.00 39.23 60.77 600,000 649,999 2 1.54 40.77 60.77 650,000 699,999 0 0.00 40.77 59.23 700,000 749,999 2 1.54 42.31 59.23 750,000 799,999 5 3.85 46.15 57.69 800,000 849,999 2 1.54 47.69 53.85 850,000 899,999 0 0.00 47.69 52.31 900,000 949,999 0 0.00 47.69 52.31 950,000 999,999 2 1.54 49.23 52.31 1.0M 1.49M 12 9.23 58.46 50.77 1.5M 1.99M 13 10.00 68.46 41.54 2.0M 2.49M 7 5.38 73.85 31.54 2.5M 2.99M 4 3.08 76.92 26.15 3.0M 3.49M 4 3.08 80.00 23.08 3.5M 3.99M 0 0.00 80.00 20.00 4.0M 4.49M 4 3.08 83.08 20.00 4.5M 4.99M 7 5.38 88.46 16.92 5.0M 5.49M 4 3.08 91.54 11.54 5.5M 5.99M 3 2.31 93.85 8.46 6.0M 7.99M 0 0.00 93.85 6.15 8.0M 8.49M 2 1.54 95.38 6.15 8.5M 11.99M 0 0.00 95.38 4.62 12.0 12.49M 4 3.08 98.46 4.62 12.5M 16.49M 0 0.00 98.46 1.54 >16M 2 1.54 100.00 1.54 Total 130 Maximum Population 16,669,062 Mean Population 2,186,620 Minimum Population 9,765 Mean Seating Capacity 15,830 246

PAGE 256

Appendix 16: Distribution of Outd oor Stadium s by Location Population Cum % Cum % Population Interval Frequency % of Total Ascending Descending < 50,000 20 10.58 10.58 100.00 60,000 99,999 7 3.70 14.29 89.42 100,000 149,999 7 3.70 17.99 85.71 150,000 199,999 14 7.41 25.40 82.01 200,000 249,999 6 3.17 28.57 74.60 250,000 299,999 6 3.17 31.75 71.43 300,000 349,999 7 3.70 35.45 68.25 350,000 399,999 1 0.53 35.98 64.55 400,000 449,999 5 2.65 38.62 64.02 450,000 499,999 1 0.53 39.15 61.38 500,000 549,999 3 1.59 40.74 60.85 550,000 599,999 0 0.00 40.74 59.26 600,000 649,999 3 1.59 42.33 59.26 650,000 699,999 3 1.59 43.92 57.67 700,000 749,999 2 1.06 44.97 56.08 750,000 799,999 2 1.06 46.03 55.03 800,000 849,999 1 0.53 46.56 53.97 850,000 899,999 2 1.06 47.62 53.44 900,000 949,999 0 0.00 47.62 52.38 950,000 999,999 1 0.53 48.15 52.38 1.0M 1.49M 18 9.52 57.67 51.85 1.5M 1.99M 12 6.35 64.02 42.33 2.0M 2.49M 13 6.88 70.90 35.98 2.5M 2.99M 7 3.70 74.60 29.10 3.0M 3.49M 6 3.17 77.78 25.4 3.5M 3.99M 0 0.00 77.78 22.22 4.0M 4.49M 15 7.94 85.71 22.22 4.5M 4.99M 7 3.7 89.42 14.29 5.0M 5.49M 7 3.7 93.12 10.58 5.5M 5.99M 2 1.06 94.18 6.88 6.0M 7.99M 0 0 94.18 5.82 8.0M 8.49M 5 2.65 96.83 5.82 8.5M 11.99M 0 0 96.83 3.17 12.0 12.49M 4 2.12 98.94 3.17 12.5M 16.49M 0 0 98.94 1.06 >16M 2 1.06 100.00 1.06 Total 189 Maximum Population 16,669,062 Mean Population 2,133,204 Minimum Population 4,481 Mean Seating Capacity 61,266 247

PAGE 257

Appendix 17: List of Explosive Material s (Bureau of Alcohol, Tobacco, Firearm s and Explosives, Department of Justice) Note: This list is stated to be comprehensive but not all-inclusive. Acetylides of heavy metals. Aluminum containing polymeric propellant. Aluminum ophorite explosive. Amatex. Ammonal. Ammonium nitrate explosive mixtures (cap sensitive). Ammonium nitrate explosive mi xtures (non-cap sensitive). Ammonium perchlorate having part icle size less than 15 microns. Ammonium perchlorate composite propellant. Ammonium perchlorate explosive mixtures. Ammonium picrate (picrate of ammonia, Explosive D). Ammonium salt lattice w ith isomorphously substituted inorganic salts. ANFO (ammonium nitrate-fuel oil). Aromatic nitro-compound explosive mixtures. Azide explosives. Baranol. Baratol. BEAF (1, 2-bis [2, 2-difluoro-2-nitroacetoxyethane]). Black powder. Black powder based explosive mixtures. Blasting agents, nitro-carbo-ni trates, including non-cap sens itive slurry and water gel explosives. Blasting caps. Blasting gelatine. Blasting powder. BTNEC (bis [trinitroethyl] carbonate). BTNEN (bis [trinitroethyl] nitramine). BTTN (1, 2, 4 butanetriol trinitrate). Bulk salutes. Butyl tetryl. Calcium nitrate explosive mixtures. Cellulose hexanitrate explosive mixture. Chlorate explosive mixtures. Composition A and variations. Composition B and variations. Composition C and variations. Copper acetylide. Cyanuric triazide. Cyclonite (RDX). 248

PAGE 258

Appendix 17: (Continued) Cyclotetramethylen etetranitramine (HMX). Cyclotol. Cyclotrimethylenetrinitrami-ne (RDX). DATB (diaminotrinitrobenzene). DDNP (diazodinitrophenol). DEGDN (diethyleneglycol Dinitrate). Detonating cord. Detonators. Dimethylol dimethyl methane dinitrate composition. Dinitroethyleneurea. Dinitroglycerine (glycerol dinitrate). Dinitrophenol. Dinitrophenolates. Dinitophenyl hydrazine. Dinitroresorcinol. Dinitrotoluene-sodium nitrate explosive mixtures. DIPAM (dipicramide; diaminohexanitrobiphenyl). Dipicryl sulfone. Dipicrylamine. Display fireworks. DNPA (2, 2-dinitrop ropyl acrylate). DNPD (dinitropentano nitrile). Dynamite. EDDN (ethylene diamine dinitrate). EDNA (ethylenedinitramine). Ednatol. EDNP (ethyl 4,4-dinitropentanoate). EGDN (ethylene glyc ol dinitrate). Erythritol tetranitrate explosives. Esters of nitro-substituted alcohols. Ethyl-tetryl. Explosive conitrates. Explosive gelatins. Explosive liquids. Explosive mixtures containing oxygen-releas ing inorganic salts and hydrocarbons. Explosive mixtures containing oxygen-releas ing inorganic salts and nitro bodies. Explosive mixtures containing oxygen-releasing inorganic salts and water insoluble fuels. Explosive mixtures containing oxygen-releasing inorganic salts and water soluble fuels. Explosive mixtures containing sensitized nitromethane. Explosive mixtures containing te tranitromethane (nitroform). Explosive nitro compounds of aromatic hydrocarbons. Explosive organic nitrate mixtures. Explosive powders. 249

PAGE 259

Appendix 17: (Continued) Flash powder. Fulm inate of mercury. Fulminate of silver. Fulminating gold. Fulminating mercury. Fulminating platinum. Fulminating silver. Gelatinized nitrocellulose. Gem-dinitro aliphatic explosive mixtures. Guanyl nitrosamino guanyl tetrazene. Guanyl nitrosamino guanylidene hydrazine. Guncotton. Heavy metal azides. Hexanite. Hexanitrodiphenylamine. Hexanitrostilbene. Hexogen (RDX). Hexogene or octogene and a nitrated N-methylaniline. Hexolites. HMTD (hexamethylenetriperoxidediamine). HMX (cyclo-1,3,5,7-tetramethylene 2,4,6,8-tetranitramine; Octogen). Hydrazinium nitrate/hydrazine /aluminum explosive system. Hydrazoic acid. Igniter cord. Igniters. Initiating tube systems. KDNBF (potassium dinitrobenzofuroxane). Lead azide. Lead mannite. Lead mononitroresorcinate. Lead picrate. Lead salts, explosive. Lead styphnate (styphnate of lea d, lead trinitroresorcinate). Liquid nitrated polyol and trimethylolethane. Liquid oxygen explosives. Magnesium ophorite explosives. Mannitol hexanitrate. MDNP (methyl 4,4-dinitropentanoate). MEAN (monoethanolamine nitrate). Mercuric fulminate. Mercury oxalate. Mercury tartrate. Metriol trinitrate. 250

PAGE 260

Appendix 17: (Continued) Minol-2 (40% TNT, 40% amm oni um nitrate, 20% aluminum). MMAN (monomethylamine nitrate); methylamine nitrate. Mononitrotoluene-nitroglycerin mixture. Monopropellants. NIBTN (nitroisobutametriol trinitrate). Nitrate explosive mixtures. Nitrate sensitized with gelled nitroparaffin. Nitrated carbohydrate explosive. Nitrated glucoside explosive. Nitrated polyhydric alcohol explosives. Nitric acid and a nitro aromatic compound explosive. Nitric acid and carboxylic fuel explosives. Nitric acid explosive mixtures. Nitro compounds of furane explosive mixtures. Nitrocellulose explosive. Nitroderivative of urea explosive mixture. Nitrogelatin explosive. Nitrogen trichloride. Nitrogen tri-iodide. Nitroglycerine (NG, RNG, nitro, glycer yl trinitrate, trinitroglycerine). Nitroglycide. Nitroglycol (ethylene gl ycol dinitrate, EGDN). Nitroguanidine explosives. Nitronium perchlorate propellant mixtures. Nitroparaffins Explosive Grade a nd ammonium nitrate mixtures. Nitrostarch. Nitor-substituted carboxylic acids. Nitrourea. Octogen (HMX). Octol (75 percent HMX, 25 percent TNT). Organic amine nitrates. Organic nitramines. PBX (plastic bonded explosives). Pellet powder. Penthrinite composition. Pentolite. Perchlorate explosive mixtures. Peroxide based explosive mixtures. PETN (nitropentaerythrite, pentaerythrite tetranitrate, pentaerythritol tetranitrate). Picramic acid and its salts. Picramide. Picrate explosives. Picrate of potassium explosive mixtures. 251

PAGE 261

Appendix 17: (Continued) Picratol. Picric acid (m anufactured as an explosive). Picryl chloride. Picryl fluoride. PLX (95% nitromethane, 5% ethylenediamine). Polynitro aliphatic compounds. Polyolpolynitrate-nitroce llulose explosive gels. Potassium chlorate andlead sulfocyanate exlosive. Potassium nitrate explosive mixtures. Potassium nitroaminotetrazole. Pyrotechnic compositions. PYX (2, 6-bis[picrylamino]) 3.5-dinitropyridine. RDX (cyclonite, hexogen, T4, cyclo-1,3,5-tr imethylene-2,4,6,-trinitramine; hexahydro1,3,5-trinitro-S-triazene). Safety fuse. Salts of organic amino sulf onic acid explosive mixture. Salutes (bulk). Silver acetylide. Silver azide. Silver fulminate. Silver oxalate explosive mixtures. Silver styphnate. Silver tartrate explosive mixtures. Silver tetrazine. Slurried explosive mixtures of water, inorga nic oxidizing salt, gelling agent, fuel, and sensitizer (cap sensitive). Smokeless powder. Sodatol. Sodium amatol. Sodium azide explosive mixture. Sodium dinitro-ortho-cresolate. Sodium nitrate explosive mixtures. Sodium nitrate-potassium n itrate explosive mixture. Sodium picramate. Special fireworks. Squibs. Styphnic acid explosives. Tacot (tetranitro-2,3,5,6-dibenzo1,3a,4,6a tetraz apentalene). TATB (triaminotrinitrobenzene). TATP (triacetonetriperoxide). TEGDN (triethylene glycol dinitrate). Tetranitrocarbazole. Tetrazene (tetracene, tetrazine, 1[-tet razolyl]4-guanyl tetrazene hydrate). 252

PAGE 262

Appendix 17: (Continued) Tetrazo le explosives. Tetryl (2,4,6 tetranitro-N-methylaniline). Tetrytol. Thickened inorganic oxidizer salt slurried explosive mixture. TMETN (trimethylolethane trinitrate). TNEF (trinitroethyl formal). TNEOC (trinitroethylorthocarbonate). TNEOF (trinitroethylorthoformate). TNT (trinitrotoluene, tr otyl, trilite, triton). Torpex. Tridite. Trimethylol ethyl methane trinitrate composition. TRimethylolthane trinitratenitrocellulose. Trimonite. Trinitroanisole. Trinitrobenzene. Trinitrobenzoic acid. Trinitrocresol. Trinitro-meta-cresol. Trinitronaphthalene. Trinitrophenetol. Trinitrophloroglucinol. Trinitroresorcinol. Tritonal. Urea nitrate. Water-bearing explosives having sa lts of oxidizing acids and n itrogen bases, sulfates, or sulfamates (cap sensitive). Water-in-oil emulsion explosive compositions. Xanthamonas hydrophilic colloid explosive mixture. Source: Federal Register/Volume 69, Nu mber 62/Wednesday, March 31, 2004/Notices 253

PAGE 263

Appendix 18: Injury Prediction Model Based on Frykberg Study Severity Level? Hospital ? Head Injury? Create Injured Survivor Hospital Yes 1 Yes Severe Not Severe Yes No No Start 254

PAGE 264

Appendix 18: (Continued) Ab Injury ? Burns? Traum. Amput? Yes Yes Yes Yes No Bla s t Lu ng? No No No 2 1 255

PAGE 265

Appendix 18: (Continued) Soft Tissue? Enter care delivery model Bony Extrem? No Yes Yes No E nd 2 256

PAGE 266

Appendix 19: Injury Prediction Model Based o n Barell Injury Matrix 257 Start Create Injured Survivor Severity Level? Hospital ? Fracture ? Hospital Yes Severe Not Severe No Yes Yes No 1

PAGE 267

Appendix 19: (Continued) Sprains/ Strains? Internal? Open Wound? No No No Di sl o cation? No Yes Yes Yes Yes 2 1 258

PAGE 268

Appendix 19: (Continued) Cont/ Super? Blood Vessels? Amputat ion? Crush No No No No Yes Yes Yes Yes 3 2 259

PAGE 269

Appendix 19: (Continued) 3 Nerves? Enter care delivery model Burns? No E nd Un sp eci f ied ? Yes No Yes Yes No 260

PAGE 270

Appendix 20: ICD-9-CM Codes and Injury Descriptions to Ma tch Injury Prediction Model Based on Frykberg Study (Appendix 18) Simulation Model Injury Category: Head Nature of the Injury Category ICD-9-CM Code Description Fracture 800 Fracture of vault of skull 801 Freacture of base of skull 803 Other and unqualified skull fractures 804 Multiple fractures in volving skull or face with other bones 802 Fracture of face bones 807 Fracture of rib(s), sternum, larynx, and trachea Dislocation 830 Dislocation of jaw Sprains and Strains 848 Other and ill-defined sprains and strains Internal 850 Concussion 851 Cerebral laceration and contusion 852 Subarachnoid, subdural, and extradural hemorrhage, following injury 853 Other and unspecified intracranial hemorrhage following injury 854 Intracranial injury of other and unspecified nature 995 Certain adverse effects not elsewhere classified Open Wound 873 Other open wound of head 872 Open wound of ear 870 Open wound of ocular adnexa 871 Open wound of eyeball 874 Open wound of neck Blood Vessels 900 Injury to blood vessels of head and neck Contusion/ Superficial 918 Superficial injury of eye and adnexa 921 contusion ofeye and adnexa 910 Superficial injury of face, neck, and scalp except eye 920 Contusion of face, scalp, and neck except eye(s) Crush 925 Crushing injury of face, scalp, and neck 261

PAGE 271

Appendix 20: (Continued) Sim ulation Model Injury Category: Head (continued) Nature of the Injury Category ICD-9-CM Code Description 940 Burn confined to eye and adnexa 947 Burn of internal organs Nerves 950 Injury to optic nerve and pathways 951 Injury to other cranial nerve(s) 953 Injury to nerve roots and spinal plexus 954 Injury to other nerve(s) of trunk, excluding shoulder and pelvic girdles 957 Injury to other and unspecified nerves Unspecified 959 Injury, other and unspecified Undefinable by Site -System Wide 930-939 Effects of foreign body entering through orifice 930 Foreign body on external eye 931 Foreign body in ear 932 Foreign body in nose 933 Foreign body in pharynx 934 Foreign body in trachea, bronchus, and lung 935 Foreign body in mouth, esophagus, and stomach 262

PAGE 272

Appendix 20: (Continued) Sim ulation Injury Category: Chest Nature of the Injury Category ICD-9-CM Code Description Fracture 807 Fracture of rib(s), sternum, larynx, and trachea Dislocation 839 Other, multiple, and ill-defined dislocation Sprains and Strains 848 Internal 860 Traumatic pneumothorax and hemothorax 861 Injury to heart and lung 862 Injury to other and unspecified intrathoracic organs Open Wound 875 Open wound of chest (wall) 879 Open wound of other and unspecified sites, except limbs Blood Vessels 901 Injury to blood vessels of thorax Contusion/ Superficial 922 Contusion of trunk Crush 926 Burns 942 Nerves 953 Undefinable by Site -System Wide 934 Foreign body in trachea, bronchus, and lung 263

PAGE 273

Appendix 20: (Continued) Simulation Injury Category: Blast Lung Nature of the Injury Category ICD-9-CM Code Description Internal 860 Traumatic pneumothorax and hemothorax 861 Injury to heart and lung 862 Injury to other and unspecified intrathoracic organs 264

PAGE 274

Appendix 20: (Continued) Sim ulation Injury Category: Abdomen Nature of the Injury Category ICD-9-CM Code Description Internal 863 Injury to gastrointestinal track 864 Injury to liver 865 Injury to spleen 866 Injury to kidney 868 Injury to other intra-abdominal organs 869 Internal injury to unspecified or illdefined organs Open Wound 879 Open wound of other and unspecified sites, except limbs Blood Vessels 902 Injury to blood vessels of abdomen and pelvis Contusion/ Superficial 922 Contusion of trunk Burns 942 Burn of trunk 947 Burns of internal organs Nerves 953 Injury to nerve roots and spinal plexus Undefinable by Site -System Wide 935 Foreign body in mouth, esophagus, and stomach 936 Foreign body in intestine and colon 265

PAGE 275

Appendix 20: (Continued) Sim ulation Injury Category: Burns Nature of the Injury Category ICD-9-CM Code Description Body Region Burns 941 Burn of face, head, and neck Head and Neck 940 Burn confined to eye and adnexa Head and Neck 947 Burn of internal organs Head and Neck/Torso/ Unclassifiable by Site 942 Burn of trunk Torso 943 Burn of upper limb, except wrist and hand Extremities 944 Burn of wrist(s) and hand(s) Extremities 945 Burn of lower limb(s) Extremities 946 Burn of multiple specified sites Unclassifiable by Site 948 Burns classified according to extent of body surface involv ed Unclassifiable by Site 949 Burn, enspecified Unclassifiable by Site 266

PAGE 276

Appendix 20: (Continued) Sim ulation Injury Category: Traumatic Amputation Nature of the Injury Category ICD-9CM Code Description Body Region Amputations 887 Traumatic amputation of arm and hand (complete)(partial) Extremities 885 Traumatic amputation of thumb (complete)(partial) Extremities 886 Traumatic amputation of other finger(s) (complete)(partial) Extremities 897 Traumatic amputation of leg(s) (complete)(partial) Extremities 895 Traumatic amputation of toe(s) (complete)(partial) Extremities 896 Traumatic amputation of foot (complete)(partial) Extremities 267

PAGE 277

Appendix 20: (Continued) Sim ulation Model Injury Category: Bony Extremity Nature of the Injury Category ICD-9-CM Code Description Fracture 810 Fracture of clavical 811 Fracture of scapula 812 Fractcure of humerus 813 Fracture of radius and ulna 814 Fracture of carpal bone(s) 815 Fracture of metacarpal bone(s) 816 Fracture of one or more phalanges of hand 817 Multiple fractures of hand bones 818 Ill-defined fractures of upper limb 820 Fracture of neck of femur 821 Fracture of other and unspecified parts of femur 822 Fracture of patella 823 Fracture of tibia and fibula 824 Fracture of ankle 825 Fracture of one or more tarsal or metatarsal bones 826 Fracture of one or more phalanges of foot 827 Other, multiple, and ill-defined fractures of lower limb 819 Multiple fractures involving both upper limbs, and upper imb with rib(s) and sternum 828 Multiple fractures involving both lower limbs, lower with upper limb, and lower limb(s) with rib(s) and sternum 829 Fracture of unspecified bones Dislocation 831 Dislocation of shoulder 832 Dislocation of elbow 833 Dislocation of wrist 834 Dislocation of finger 835 Dislocation of hip 836 Dislocation of knee 837 Dislocation of ankle 838 Dislocation of foot 268

PAGE 278

Appendix 20: (Continued) Sim ulation Model Injury Category: Bony Extremity (continued) Nature of the Injury Category ICD-9-CM Code Description 841 Sprains and strains of elbow and forearm 842 Sprains and strains of wrist and hand 843 Sprains and strains of hip and thigh 844 Sprains and strains of knee and leg 845 Sprains and strains of ankle and foot 848 Other and ill-defined sprains and strains Open Wound 880 Open wound of shoulder and upper arm 881 Open wound of elbow, forearm, and wrist 882 Open wound of hand except finger(s) alone 883 Open wound of finger(s) 884 Multiple and unspecified open wound of upper limb 892 Open wound of foot except toe(s) alone 893 Open wound of toe(s) 890 Open wound of hip and thigh 891 Open wound of knee, leg (except thigh), and ankle 894 Multiple and unspecified open wound of lower limb Amputations 887 Traumatic amputation of arm and hand (complete)(partial) 885 Traumatic amputation of thumb (complete)(partial) 886 Traumatic amputation of other finger(s) (complete)(partial) 897 Traumatic amputation of leg(s) (complete)(partial) 895 Traumatic amputation of toe(s) (complete)(partial) 896 Traumatic amputation of foot (complete)(partial) Blood Vessels 903 Injury to blood vessels of upper extremity 904 Injury to blood vessels of lower extrimity and unspecified sites 269

PAGE 279

Appendix 20: (Continued) Sim ulation Model Injury Category: Bony Extremity (continued) Nature of the Injury Category ICD-9-CM Code Description 923 Contusion of upper limb 914 Superficial injury of hand(s) except finger(s) alone 915 Superficial injury offinger(s) 913 Superficial injury of elbow, forearm, and wrist 924 Contusion of lower limb and of other and unspecified sites 917 Superficial injury of foot and toe(s) 916 Superficial injury of hip, thigh, leg, and ankle Crush 927 Crushing injury of upper limb 928 Crushing injury of lower limb Burns 943 Burn of upper limb, except wrist and hand 944 Burn of wrist(s) and hand(s) 945 Burn of lower limb(s) Nerves 953 Injury to nerve roots and spinal plexus 270

PAGE 280

Appendix 20: (Continued) Sim ulation Model Injury Category: Soft Tissue Nature of the Injury Category ICD-9-CM Code Description Body Region Open Wound 873 Other open wound of head Head and Neck 872 Open wound of ear Head and Neck 870 Open wound of ocular adnexa Head and Neck 871 Open wound of eyeball Head and Neck 874 Open wound of neck Head and Neck 875 Open wound of chest (wall) Torso 879 Open wound of other and unspecified sites, except limbs Torso/Unclassifiable by Site 877 Open wound of buttock Torso 878 Open wound of other and unspecified sites, except limbs Torso 876 Open wound of back Torso 880 Open wound of shoulder and upper arm Extremities 881 Open wound of elbow, forearm, and wrist Extremities 882 Open wound of hand except finger(s) alone Extremities 883 Open wound of finger(s) Extremities 884 Multiple and unspecified open wound of upper limb Extremities 892 Open wound of foot except toe(s) alone Extremities 893 Open wound of toe(s) Extremities 890 Open wound of hip and thigh Extremities 891 Open wound of knee, leg (except thigh), and ankle Extremities 894 Multiple and unspecified open wound of lower limb Extremities 271

PAGE 281

Appendix 20: (Continued) Sim ulation Model Injury Category: Soft Tissue (continued) Nature of the Injury Category ICD-9-CM Code Description Body Region 901 Injury to blood vessels of thorax Torso 902 Injury to blood vessels of abdomen and pelvis Torso/Unclassifiable by Site 903 Injury to blood vessels of upper extremity Extremities 904 Injury to blood vessels of lower extrimity and unspecified sites Extremities /Unclassifiable by Site Contusion/ Superficial 918 Superficial injury of eye and adnexa Head and Neck 921 contusion ofeye and adnexa Head and Neck 910 Superficial injury of face, neck, and scalp except eye Head and Neck 920 Contusion of face, scalp, and neck except eye(s) Head and Neck 922 Contusion of trunk Torso 911 Superficial injury of trunk Torso 912 Superficial injury of shoulder and upper arm Extremities 923 Contusion of upper limb Extremities 914 Superficial injury of hand(s) except finger(s) alone Extremities 915 Superficial injury offinger(s) Extremities 913 Superficial injury of elbow, forearm, and wrist Extremities 924 Contusion of lower limb and of other and unspecified sites Extremities/ Unclassifiable by Site 917 Superficial injury of foot and toe(s) Extremities 916 Superficial injury of hip, thigh, leg, and ankle Extremities 919 Superficial injury of other Unclassifiable by Site 272

PAGE 282

Appendix 20: (Continued) Sim ulation Model Injury Category: Soft Tissue (continued) Nature of the Injury Category ICD-9-CM Code Description Body Region 940 Burn confiend to eye and adnexa Head and Neck 947 Burn of internal organs Head and Neck/Torso/ Unclassifiable by Site 942 Burn of trunk Torso 943 Burn of upper limb, except wrist and hand Extremities 944 Burn of wrist(s) and hand(s) Extremities 945 Burn of lower limb(s) Extremities 946 Burn of multiple specified sites Unclassifiable by Site 948 Burns classified according to extent of body surface involved Unclassifiable by Site 949 Burn, enspecified Unclassifiable by Site Nerves 950 Injury to optic nerve and pathways Head and Neck 951 Injury to other cranial nerve(s) Head and Neck 953 Injury to nerve roots and spinal plexus Head and Neck/Torso/ Extremities/ Unclassifiabale by Site 954 Injury to other nerve(s) of trunk, excluding shoulder and pelvic girdles Head and Neck/Torso 956 Injury to peripheral nerve(s) of pelvic girdle and lower limb Unclassifiable by Site 957 Injury to other and unspecified nerves Head and Neck/ Unclassifiable by Site 955 Injury to peripheral nerve(s) of pelvic girdle and lower limb Extremities 273

PAGE 283

Appendix 21: Resource Prediction Model Rescue Transportation C omponent 274 Start D.1 P.1 D.2 D.3 D.4 P.2 P.3 P.4 P.5 D.5 D.6 D.7 c.1 c.1 c.2 c. 2 c. 3 2 1 2 1 True False True True True True True True False False False False False False c.3

PAGE 284

Appendix 21: (Continued) Resource Prediction Model Sorting and Life Support at th e CCP Component P.6 D.8 D.9 D.10 P.7 P.8 D.11 D.12 P.9 2 True False False False False False True True T 1 1 275

PAGE 285

Appendix 21: (Continued) Resource Prediction Model Hospital Com ponent 2 P.10 D.13 P.11 D.14 P.13 P.12 3 D.15 P.14 False False False True True True 4 276

PAGE 286

Appendix 21: (Continued) Resource Prediction Model Surgery Com ponent 3 P.15 D.16 D.17 D.16-23 True True False False 5 277

PAGE 287

Appendix 21: (Continued) Resource Prediction Model Hosp ital Com ponent (continued) D.18 P.24 P.25 T 2 True False T .3 4 278

PAGE 288

Appendix 21: (Continued) Resource Prediction Model Hosp ital Com ponent (continued) D.19 P.26 P.27 True False T.3 5 279

PAGE 289

Appendix 22: Sim ulation Decision Module, Process Modul e, and Termination Module Parameters Decision Module Parameters Page Module Label Title Basis of Decision True Parameter Alternative Source of Parameters 1 D.1 The IS is trapped. Chance 20% The IS is not trapped. Researcher Defined 1 D.2 The IS is mobile. Chance 80% The IS is not mobile. Researcher Defined 1 D.3 The IS travels by informal means. Chance 80% The IS travel by formal means. Researcher Defined 1 D.4 The IS travels by formal means to the hospital Chance 20% The IS travels by formal means to the CCP. Researcher Defined 1 D.5 The IS travels by informal means to the hospital. Chance 80% The IS travels by informal means to the CCP. Researcher Defined 1 D.6 The rescued IS travels by formal means to the hospital.Condition If the IS has a major injury, the IS will be transported to the hospital. The rescued IS travels by formal means to the CCP. Researcher Defined 1 D.7 The mobile IS is transported by informal means to the hospital. Chance 80% The mobile IS is transported by informal means to the CCP. Researcher Defined 2 D.8 The IS has a minor injury. Condition If the IS was previously assigned the minor injury attribute then true. The IS was assigned the major injury attribute Researcher Defined 2 D.9 The IS requires stabilization in the field. Chance 20% The IS does not require stabilization. Researcher Define d 2 D.10 Field treatment is adequate for the IS. Chance 80% The IS requries more care then is available in the field. Researcher Defined 280

PAGE 290

281 Appendix 22: (Continued) Decision Module Parame ters (continued) Page Module Label Title Basis of Decision True Parameter Alternative Source of Parameters 2 D.12 The IS requires no additional care. Field treatment was sufficient Chance 80% The IS requires additional care. Researcher Defined 3 D.13 The injured survivor has only minor injuries. Condition If the IS has previously been assigned the minor injury attribute. The IS has major injuries Original assignment adapted from frykberg (1988). 3 D.14 The IS has major injuries and is stable. Chance 73% The IS has major injuries and is unstable. Adapted from Frykberg (1988) 3 D.15 The IS with minor injuries requires and imaging study. Chance 69% The IS with minor injuries does not require an imaging study. Adapted from Hirschberg (1999). 4 D.16 The IS requires surgery. Chance 17% The IS does not require surgery. 4 D.17 The IS requires a surgical specialist Condition Selection is based on the injury previously assigned to the IS. The only alternatives is one of the surgical specialties Researcher Defined 5 D.18 Hospitalization is required for the IS with a minor injury Condition If the IS has previously been assigned the hospitalization attribute. The IS does not require hospitalization. Original assignment adapted from Frykb erg (1988). 6 D.19 ICU Care is Required Chance 20% The IS does not require ICU care. Researcher Defined

PAGE 291

Appendix 22: (Continued) Process Module Param eters Page Module Label Process Title Process Explanation Resource Unit Distribution Unit Mean Min Most Likely Max Source of Parameters 1 P.1 Rescue Rescue of trapped IS. Rescue Triangular Minutes 25 5 10 60 Researcher Defined 1 P.2 Formal transport to CCP. Formal transportatin of IS to CCP. Emergency Response Triangular Minutes 9.33 3 5 20 Researcher Defined 1 P.3 Formal transport to hospital. Formal transportation of IS to hospital. Emergency Response Triangular Minutes 30 10 20 60 Researcher Defined 1 P.4 Informal transport to CCP. Informal transportation of IS to CCP. Informal transportationTriangular Minutes 12.67 3 5 30 Researcher Defined 1 P.5 Informal transport to hospital. Informal transportation of IS to hospital. Informal transportation Triangular Minutes 38.33 5 20 90 Researcher Defined 2 P.6 Field CCP Triage Triage of IS as the arrive at the field CCP. Triage Triangular Seconds 28.33 5 20 60 Adapted from Hirshberg (1999) 282

PAGE 292

Appendix 22: (Continued) Process Module Param eters (continued) Page Module Label Process Title Process Explanation Resource Unit Distributio n Unit Mean Min Most Likely Max Source of Parameters 2 P.8 Field Care Care provided in the filed to IS. Field care Triangular Minute s 23.33 11 21 38 Adapted from Hirshberg (1999) 2 P.9 Formal transport to hospital. Formal transportation of IS to hospital from CCP. Emergency Response Triangular Minute s 30 10 20 60 Researcher Defined 3 P.10 Hospital Triage Triage for IS as the arrive at the hospital. Triage Triangular Second s 28.33 5 20 60 Adapted from Hirshberg (1999) 3 P.11 ED Care Care provided in the ED for those with minor injuries. Hospital Care Triangular Minute s 59.67 22 41 116 Adapted from Hirshberg (1999) 3 P.12 Initial Trauma Care for the Stable IS Initial trauma care provided in the hospital to the IS who is stable. Trauma Care Triangular Minute s 50 14 47 89 Adapted from Hirshberg (1999) 283

PAGE 293

Appendix 22: (Continued) Process Module Param eters (continued) Page Module Label Process Title Process Explanation Resource Unit Distribution Unit Mean Min Most Likely Max Source of Parameters 3 P.14 Imaging for the IS with Minor Injuries. Imaging study within the hospital for the IS with minor injuries Imaging Triangular Minute s 29.33 11 19 58 Adapted from Hirshberg (1999) 4 P.15 Imaging for the IS who was not stable. Imaging study within the hospital for the IS who was required stabilization. Imaging Triangular Minute s 29.33 11 19 58 Adapted from Hirshberg (1999) 4 P.16-23 The IS requires surgery. Surgical procedues with the hospital for those IS who required surgery. Surgical Triangular Minute s 120.67 47 97 218 Adapted from Hirshberg (1999) P.16 & 19 Cardiothoracic Surgery Surgical Triangular Minute s 120.67 47 97 218 Adapted from Hirshberg (1999) 284

PAGE 294

Appendix 22: (Continued) Process Module Param eters (continued) Page Module Label Process Title Process Explanation Resource Unit Distribution Unit Mean Min Most Likely Max Source of Parameters P.18, 20, & 22 Genereal Surgery Surgical Triangular Minutes 120.67 47 97 218 Adapted from Hirshberg (1999) P.21 Neurosurgery Surgery Surgical Triangular Minutes 120.67 47 97 218 Adapted from Hirshberg (1999) 5 P.24 PostImaging ED Care. Medical care provided in the ED for IS with minor injuries after imaging study. Hospital Care Triangular Minutes 73 29 53 137 Adapted from Hirshberg (1999) 285

PAGE 295

286 Appendix 22: (Continued) Process Module Parameters (continued) Page Module Label Process Title Process Explanation Resource Unit Distribution Unit Mean Min Most Likely Max Source of Parameters 6 P.26 ICU Care Care provided for the IS with major injuries who required intensive care. ICU Care Triangular Minute s 4800 1440 4320 9640 Adapted from Hirshberg (1999) 6 P.27 Hospital Care after ICU or Surgery Care provided for the IS follwing ICU Care or Surgical Care. Inpatient Care Triangular Minute s 4800 1440 4320 8640 Adapted from Hirshberg (1999)

PAGE 296

287 Appendix 22: (Continued) Term ination Module Parameters (Appendix 22 continued) Page Module Label Title Reason 1 T.1 Termination Field care was adequate for the IS. No additional care is required. 5 T.2 Termination The IS has received all required care in the ED and has been released. 5 T.3 Termination The hospitalized IS has completed care and has been discharged. 6 T.3 Termination The hospitalized IS has completed care and has been discharged.

PAGE 297

Appendix 23: Sim ulation Results 7,000 Injured Survivors (no constraints, 10 replication averages, time in minutes) Note: The process category descriptio ns are located in the Objective 3 Results Section. Process categories are presented in the order as provided in the Arena output for the simulation run. Process # in VA Time T Time A Time QW Time Q#W Category P.1 1 4,552 59.6 59.6 271,379.9 NA NA P.8 1,684 23.3 23.3 39,307.6 NA NA P.9 1,161 30 30 34,818.4 NA NA P.2 1,324 9.3 9.3 12,254.5 NA NA P.3 295 29.9 29.9 8,844.9 NA NA P.10 5,847 0.5 0.5 2,759.3 NA NA P.25 795 2,402.9 2,402.9 1,910,901.6 NA NA P.27 1,295 4,814.7 4,814.7 6,235,874.9 NA NA P.28 258 4,797.6 4,797.6 1,236,260 NA NA P.14 3,145 29.4 29.4 92,359.1 NA NA P.4 990 12.6 12.6 12,432.5 NA NA P.5 4,391 38.4 38.4 168,550.7 NA NA P.12 944 49.7 49.7 46,943.3 NA NA P.13 351 25.5 25.5 8,937.8 NA NA P.6 2,314 0.5 0.5 1,093.9 NA NA P.24 3,756 72.9 72.9 273,935.9 NA NA P.15 944 29.3 29.3 27,652.68 NA NA P.1 1,396 25 25 34,956.73 NA NA P.7 123 23.2 23.2 2,858.47 NA NA P.16 2 101.3 101.3 209.6 NA NA P.17 34 120.8 120.8 4,140.1 NA NA P.18 20 119.6 119.6 2,401.6 NA NA P.19 6 119.7 119.7 719.8 NA NA P.20 5 124.7 124.7 597.2 NA NA P.21 165 121.9 121.9 20,071.6 NA NA P.22 277 120.6 120.6 33,457.9 NA NA P.23 4 115.8 115.8 458.6 NA NA 288

PAGE 298

Appendix 24: Sim ulation Results 45,000 Injured Survivors (no constraints, 10 replication averages, time in minutes) Note: The process category descriptio ns are located in the Objective 3 Results Section. Process categories are presented in the order as provided in the Arena output for the simulation run. Process # in VA Time T Time A Time QW Time Q#W Category P.11 29,227 59.6 59.6 1,743,049.9 NA NA P.8 10,743 23.3 23.3 250,348.2 NA NA P.9 7,427 30 30 222,616.8 NA NA P.2 8,475 9.3 9.3 79,091.7 NA NA P.3 1,967 29.9 29.9 58,861.6 NA NA P.10 37,629 0.5 0.5 17,761.1 NA NA P.25 5,097 2,399.6 2,399.6 12,229,994 NA NA P.27 8,402 4,792.3 4,792.3 40,266,570 NA NA P.26 1,654 4,799.3 4,799.3 7,935,552.5 NA NA P.14 20,168 29.3 29.3 591,536.9 NA NA P.4 6,323 12.6 12.6 79,774.8 NA NA P.5 28,236 38.3 38.3 1,082,181.35 NA NA P.12 6,127 50 50 306,548.4 NA NA P.13 2,275 25.6 25.6 58,210.1 NA NA P.6 14,798 0.5 0.5 6,988.3 NA NA P.24 24,130 73 73 1,762,044.9 NA NA P.15 6,127 29.3 29.3 179,464 NA NA P.1 9,035 25 25 225,809 NA NA P.7 800 23.3 23.3 18,654.7 NA NA P.16 14 123.8 123.8 1,695.12 NA NA P.17 214 120.5 120.5 25,794.5 NA NA P.18 111 120.5 120.5 13,380.7 NA NA P.19 45 120.8 120.8 5,433.9 NA NA P.20 29 121.5 121.5 3,505.3 NA NA P.21 1,040 120.7 120.7 125,549.2 NA NA P.22 1,839 120.3 120.3 221,250.9 NA NA P.23 23 118 118 2,710.5 NA NA 289

PAGE 299

Appendix 25: Sim ulation Results 7,000 Injured Survivors (12 hours, resource sparse, 10 replication averages, time in minutes) Note: The process category descriptio ns are located in the Objective 3 Re sults Section. Process categories are presented in the order as provided in the Arena output for the simulation run. Process # in VA Time TP Time AP Time QW Time Q#W Category P.11 1,095 59.4 343.05 2679.7 310.9 494 P.8 226 23.3 182.1 2469.3 165 53.8 P.9 134 29.7 186 2,077.6 164 30.5 P.2 73 9.2 184.2 369.1 175.6 17.7 P.3 19 30.1 172 278.6 153.1 4.2 P.10 1,375 0.5 0.6 650.4 0.1 0.2 P.25 8 Run time not long enough to complete process, no data produced. P.27 11 Run time not long enough to complete process, no data produced. P.26 3 Run time not long enough to complete process, no data produced. P.14 30 29 29.2 819.1 0.1 0.01 P.4 306 12.6 12.6 3,803.3 0.0 0.0 P.5 1,364 38.1 38.1 49,399.4 0.0 0.0 P.12 205 50.3 325.4 576.6 299.8 89.4 P.13 73 22.7 350 99.5 327.4 32.3 P.6 341 0.5 0.5 159.9 0.1 0.03 P.24 36 5.5 31.2 5.5 85 15.7 P.15 12 28.9 29 327.6 0.1 0.0 P.1 443 24.4 332.7 702.3 319.8 203.5 P.7 23 24 201.6 275.2 183.4 5.8 P.16 0 NA NA NA NA NA P.17 1 32.4 32.4 32.4 0.0 0.0 P.18 1 39.3 39.3 54.1 0.0 0.0 P.19 0 NA NA NA NA NA P.20 0 NA NA NA NA NA P.21 2 62.4 64.2 133.3 1.8 0.01 P.22 4 103.1 103.1 304.1 0.0 0.0 P.23 0 NA NA NA NA NA 290

PAGE 300

Appendix 26 Sim ulation Results 7,000 Injured Survivors (12 hours, resource dense, 10 replication averages, time in minutes) Note: The process category descriptio ns are located in the Objective 3 Re sults Section. Process categories are presented in the order as provided in the Arena output for the simulation run. Process # in VA Time T Ti me AVA Time QW Time Q#W C ategory P.11 1,273 59 242.1 20,476.3 205 396.9 P.8 415 23.2 23.2 9,240 0.0 0.0 P.9 286 29.5 29.5 8,056.9 0.0 0.0 P.2 287 9.2 9.2 2,593.4 0.0 0.0 P.3 64 30.5 30.5 1,854.9 0.0 0.0 P.10 1,636 0.5 0.5 772.3 0.0 0.0 P. 25 55 Run time not long enough to complete process, no data produced. P.27 109 Run time not long enough to complete process, no data produced. P.26 27 Run time not long enough to complete process, no data produced. P.14 239 29.5 29.5 6,773.79 0.0 0.0 P.4 303 12.7 12.7 3,789.6 0.0 0.0 P.5 1,375 37.9 37.9 49,277.8 0.0 0.0 P.12 262 49.2 209.8 5,402.1 174.3 66.6 P.13 100 25.4 202 1,056.3 182 25.8 P.6 581 0.5 0.5 274.8 0.0 0.0 P.24 283 70.7 305.7 5,182 256.9 96.9 P.15 110 29.1 29.1 3,065.2 0.0 0.0 P.1 434 24.8 133.5 6,932.5 112.81 68.5 P.7 37 22.8 22.8 815.2 0.0 0.0 P.16 0 34 34 34 0.0 0.0 P.17 3 131.6 131.6 261 0.0 0.0 P.18 2 92 92 181.1 0.0 0.0 P.19 0 41.1 41.1 57.1 0.0 0.0 P.20 0 50.1 50.1 50.1 0.0 0.0 P.21 19 120.3 120.3 1,852.7 0.0 0.0 P.22 31 117.7 117.7 2,932.2 0.0 0.0 P.23 0 29.7 29.7 44.2 0.0 0.0 291

PAGE 301

Appendix 27 Sim ulation Results 45,000 Injured Survivors (12 hours, resource sparse, 10 replication averages) Note: The process category descriptio ns are located in the Objective 3 Re sults Section. Process categories are presented in the order as provided in the Arena output for the simulation run. Process # in VA Time T Time AVA Time QW Time Q#W Category P.1 1 1,095 59.4 343.1 2,679.7 310.9 494 P.8 226 23.3 182.1 2,469.3 165 53.8 P.9 134 29.7 186 2,077.6 164 30.5 P.2 73 9.2 184.2 369.2 175.6 17.7 P.3 19 30.1 172 278.6 153.1 4.2 P.10 1,375 0.5 0.6 650.41 0.1 0.2 P.25 8 Run time not long enough to complete process, no data produced. P.27 11 Run time not long enough to complete process, no data produced. P.26 3 Run time not long enough to complete process, no data produced. P.14 30 29.0 29.2 819.1 0.1 <1 P.4 306 12.6 12.6 3,803.3 0.0 0.0 P.5 1,364 38.1 38.1 49,399.4 0.0 0.0 P.12 205 50.3 325.4 576.6 299.8 89.4 P.13 73 23.7 350 99.5 327.4 32.3 P.6 341 0.5 0.5 159.9 0.1 0.0 P.24 36 5.5 31.2 5.5 85 15.7 P.15 12 28.9 29 327.6 0.1 0.0 P.1 443 24.4 332.7 702.3 319.8 203.5 P.7 23 24 201.6 275.2 183.4 5.8 P.16 0 NA NA NA NA NA P.17 1 32.4 32.4 32.4 0.0 0.0 P.18 1 39.3 39.3 54.1 0.0 0.0 P.19 0 NA NA NA NA NA P.20 0 NA NA NA NA NA P.21 2 62.4 64.2 133.3 1.8 0.0 P.22 4 103.1 103.1 304.1 0.0 0.0 P.23 0 NA NA NA NA NA 292

PAGE 302

Appendix 28 Sim ulation Results 45,000 Injured Surv ivors (12 hours, resource dense) Note: The process category descriptio ns are located in the Objective 3 Re sults Section. Process categories are presented in the order as provided in the Arena output for the simulation run. Process # in VA Time T Ti me AVA Time QW Time Q#W Category P.11 1,273 59 242.1 20,476.3 205 397.9 P.8 415 23.2 23.2 9,240 0.0 0.0 P.9 286 29.5 29.5 8,056.9 0.0 0.0 P.2 287 9.2 9.2 2,593.4 0.0 0.0 P.3 64 30.5 30.5 1,854.9 0.0 0.0 P.10 1,636 0.5 0.5 772.3 0.0 0.0 P.25 55 Run time not long enough to complete process, no data produced. P.27 109 Run time not long enough to complete process, no data produced. P.26 27 Run time not long enough to complete process, no data produced. P.14 239 29.5 29.5 6,773.8 0.0 0.0 P.4 303 12.7 12.7 3,789.6 0.0 0.0 P.5 1,375 37.9 37.9 49,277.8 0.0 0.0 P.12 262 49.2 209.8 5,402.1 174.3 66.6 P.13 100 25.4 202 1,056.3 182 25.8 P.6 581 0.5 0.5 274.8 0.0 0.0 P.24 283 70.7 305.7 5,182 256.9 96.9 P.15 110 29.1 29.1 3,065.2 0.0 0.0 P.1 434 24.8 133.5 6,932.5 112.8 68.5 P.7 37 22.8 22.8 815.2 0.0 0.0 P.15 0 34 34 34 0.0 0.0 Surgery 2 3 131.6 131.6 261 0.0 0.0 Surgery 3 2 92 92 181.1 0.0 0.0 Surgery 4 0 41.1 41.1 57.1 0.0 0.0 Surgery 5 0 50.1 50.1 50.1 0.0 0.0 Surgery 6 19 120.3 120.3 1,852.7 0.0 0.0 Surgery 7 31 117.7 117.7 2,932.2 0.0 0.0 Surgery 8 0 29.7 29.7 44.2 0.0 0.0 293

PAGE 303

Appendix 29 Sim ulation Results 7,000 Injured Survivors (12 hour resource constraints, 10 replication means) Note: The process category descriptio ns are located in the Objective 3 Re sults Section. Process categories are presented in the order as provided in the Arena output for the simulation run. Process # in VA Time T Time AVA Time QW Time Q#W Category P.11 4,448 59.16 64.39 35,609 5.98 37 P.8 1,706 23.33 57.04 36,948 34.72 83 P.9 1,125 29.81 61.42 30,296 32.69 51 P.2 1,348 9.33 40.92 12,535 31.67 59 P.3 307 29.97 60.08 9,056 30.70 13 P.10 5,860 0.47 32.93 2,720 32.46 264 P.25 629 Run time not long enough to complete process, no data produced. P.27 810 Run time not long enough to complete process, no data produced. P.26 202 Run time not long enough to complete process, no data produced. P.14 2,725 29.14 31.54 75,054 2.51 10 P.4 1,004 12.64 12.64 12,691 NA NA P.5 4,541 38.46 38.46 174,648 NA NA P.12 957 49.45 64.93 41,172 16.75 23 P.13 348 25.54 41.71 8,062 16.71 8 P.6 2,348 0.47 1.50 1,109 1.03 3 P.24 3,205 71.96 77.79 197,880 6.77 30 P.15 833 29.11 31.45 22,974 2.45 3 P.1 1,436 24.79 24.79 35,605 NA NA P.7 129 23.21 56.22 2,832 33.82 6 P.16 2 73.97 76.82 159 9.94 <1 P.17 32 117.79 127.96 2,821 13.51 1 P.18 15 119.13 122.31 1,538 3.44 <1 P.19 6 107.84 117.43 448 12.04 <1 P.20 4 113.28 116.64 340 3.75 <1 P.21 142 117.81 126.57 13,249 9.83 2 P.22 247 117.81 120.97 23,299 3.51 1 P.23 3 121.29 133.38 278 11.69 <1 294

PAGE 304

295 Appendix 30 Simulation Results 45,000 Injured Survivors (12 hour resource constraints, 10 replication means) Note: The process category descriptio ns are located in the Objective 3 Re sults Section. Process categories are presented in the order as provided in the Arena output for the simulation run. Process # in VA Time T Ti me AVA Time QW Time Q#W Cat4gory P.11 28,141 59.01 61.89 1,511,462 3.1 121 P.8 10,716 23.26 30.82 236,023 7.92 119 P.9 7,220 29.68 83.92 198,435 52.09 518 P.2 8,486 9.35 57.84 79,344 48.49 571 P.3 1,986 30.01 78.59 59,603 48.58 134 P.10 36,894 0.47 83.65 17,254 83.14 4,227 P.25 3,911 Run time not long enough to complete process, no data produced. P.27 5,342 Run time not long enough to complete process, no data produced. P.26 1,330 Run time not long enough to complete process, no data produced. P.14 17,638 29.13 30.05 489,396 0.93 23 P.4 6,306 12.61 12.61 79,509 0.00 0.00 P.5 28,222 38.32 38.32 1,081,552 0.00 0.00 P.12 6,113 49.58 53.38 269,332 4.89 46 P.13 2,270 25.52 29.67 53,498 4.84 17 P.6 14,792 0.47 87.72 6,994 87.25 1,793 P.24 20,863 72.08 75.27 1,308,723 3.52 102 P.15 5,432 29.07 29.97 150,313 0.92 7 P.1 9,036 24.97 24.97 225,641 0.00 0.00 P.7 828 23.24 30.51 18,566 7.51 9 P.16 15 120.97 147.84 1,319 34.34 1 P.17 199 117.21 128.16 18,206 14.11 4 P.18 94 117.23 119.78 9,009 2.85 <1 P.19 44 120.27 149.77 3,745 38.64 3 P.20 29 120.15 123.00 2,810 3.08 <1 P.21 943 118.57 123.14 89,511 5.48 7 P.22 1,642 118.35 121.34 156,557 3.3 8 P.23 23 119.94 130.76 2,076 13.52 <1

PAGE 305

About the Author Scott Zuerlein received a Bachelors Degree in General Busi ness Management from the University of Wyoming in 1988 and a Mast ers in Health Services Administration from Baylor University in 1996. He was commissioned into the US Air Force Medical Service upon graduation in 1988 and currently ho lds the rank of Lieutenant Colonel. He entered the Ph.D. program at the University of South Florida with Air Force sponsorship in 2002. While in the Ph.D. program at the Univ ersity of South Florida, Lt Col Zuerlein was active in the Student Chapter of the American College of Health Care Executives in in the summer of 2005 returned to an active duty assignment in the Command Surgeons office of the US Central Command. He cu rrently leads the co mmands international health engagement program.


xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 2200397Ka 4500
controlfield tag 001 002028887
005 20090915092558.0
007 cr mnu|||uuuuu
008 090915s2009 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0002836
035
(OCoLC)436688735
040
FHM
c FHM
049
FHMM
090
RA425 (Online)
1 100
Zuerlein, Scott A.
0 245
Predicting the medical management requirements of large scale mass casualty events using computer simulation
h [electronic resource] /
by Scott A. Zuerlein.
260
[Tampa, Fla] :
b University of South Florida,
2009.
500
Title from PDF of title page.
Document formatted into pages; contains 295 pages.
Includes vita.
502
Dissertation (Ph.D.)--University of South Florida, 2009.
504
Includes bibliographical references.
516
Text (Electronic dissertation) in PDF format.
3 520
ABSTRACT: Recent events throughout the world and in the US lend support to the belief that another terrorist attack on the US is likely, perhaps probable. Given the potential for large numbers of casualties to be produced by a blast using conventional exlosives, it is imperative that health systems across the nation consider the risks in their jurisdictions and take steps to better prepare for the possibility of an attack. Computer modeling and simulation offers a viable and useful methodology to better prepare an organization or system to respond to a large scale event. The real question, given the shortage, and in some areas absence, of experiential data, could computer modeling and simulation be used to predict the resource requirements generated by this type of event and thus prepare a health system in a defined geographic area for the possibility of an event of this nature? Research resulted in the identification of variables that surround a health system at risk, the development of a computer model to predict the injuries that would be seen in an injured survivor population and the medical resources required to care for this population. Finally, methodologies were developed to modify the existing model to match unique health system structures and processes in order to assess the preparedness of a specific geographic location or health system. As depicted in this research, computer modeling and simulation was found to offer a viable and usable methodology for a defined geographic region to better prepare for the potential of a large scale blast event and to care for the injured survivors that result from the blast. This can be done with relatively low cost and low tech approach using existing computer modeling and simulation software, making it affordable and viable for even the smallest geographic jurisdiction or health system.
538
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
590
Advisor: Alan M. Sear, Ph.D.
653
Computer modeling
Simulation
Blasts
Planning
Emergency care
690
Dissertations, Academic
z USF
x Health Policy and Management
Doctoral.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.2836