Flex Serv Manuf J DOI 10.1007/s10696-014-9205-z
A policy management game for mass casualty incidents: an experimental study Marion S. Rauner • Helmut Niessner • Ulrike Leopold-Wildburger • Natasa Peric Teresa Herdlicka
•
Springer Science+Business Media New York 2014
Abstract The number of complex and unique mass casualty incidents has increased due to natural and technological disasters as well as man-made disasters such as political instabilities, economic recession, and terrorism. Thus, health care policy-makers such as the Austrian Samaritan Organization have been continuously improving the training of emergency staff to enable them to quickly evacuate an emergency site, to minimize the number of fatalities at the incident site, and to decrease the patients’ waiting time for treatment. We developed a policy management game to provide a training tool for emergency staff to support such policymakers. In addition, with this game students can be educated on scheduling and planning techniques such as simulation, queuing theory, and resource allocation. To investigate the potential of our policy management game, we conducted an experimental study with 96 participants including students, practitioners from health care services, and researchers. They acted as incident commanders to decide on sending medical staff to triage, to different treatment rooms for care and to on-site transportation, as well as to transportation to hospitals during three game runs. The participants rated the general structure and organization of the experiment as high. The performance was also improved by many participants during the experiment. We found differences in performance among the different participant groups.
M. S. Rauner (&) Department of Innovation and Technology Management, University of Vienna, Oskar-Morgenstern-Platz 1, 1090 Vienna, Austria e-mail:
[email protected] H. Niessner N. Peric T. Herdlicka School of Business and Economics, University of Vienna, Vienna, Austria U. Leopold-Wildburger Department of Statistics and Operations Research, Karl-Franzens-University Graz, Graz, Austria
123
M. S. Rauner et al.
Keywords Policy simulator Management game Health care Mass casuality incidents Experimental study
1 Introduction Since the Industrial Revolution, urbanization has accelerated tremendously (Batty 2008). Nowadays, more than 50 % of the global population lives in huge, densely populated urban areas which produce more than 90 % of the world economy (Solecki et al. 2013). For the functioning of such urban societies, a large-scale and most critical infrastructure has to be provided, such as water supply, food supply, power supply, transportation, public health, telecommunication, and security (Rinaldi et al. 2001). Disruptions in these critical infrastructures represent the Achilles’ heel of urbanized societies and can cause a lot of damage both to the economy and humans (Rinaldi et al. 2001). The World Health Organization Collaborating Centre for Research on the Epidemiology of Disasters has been maintaining an Emergency Events Database and has reported a high increase in natural disasters (e.g., geophysical disasters, hydrometeorological disasters, climatological disasters, or biological disasters) and technological disasters (e.g. industrial disasters or transportation disasters) that threaten urban societies (Center for Research on the Epidemiology of Disasters 2013). In addition, political instabilities, economic recession, and terrorism cause man-made disasters with similar consequences (Pidgeon and O’Leary 2000). If such disasters happen at crowded places such as event locations (e.g., sport stadiums, concert halls), transportation hubs (e.g., airports, train stations, metro stations), or huge gathering places (e.g., squares, parks), then policy-makers will have to handle mass casualty incidents (MCIs) with high impact. For example, the mass panic of the hundreds of thousands of visitors at the Love Parade in Duisburg, Germany, in 2010 caused 575 injured people and 21 deaths (Marx et al. 2013). Due to a conductor’s mistake, a large train accident near Santiago de Compostela in Spain in 2013 injured and killed 140 and 79 passengers, respectively (Marot 2013). Terrorists bombed the Boston marathon in USA in 2013, resulting in 264 injured people and three fatalities (Kotz 2013; Lown 2013). A standardized concept to handle such MCIs is used by emergency medical services in many central European countries such as Germany and Austria (Hansak et al. 2003). The main focus of this concept is to put up a small field hospital which is called advanced medical post (AMP) next to the incident site. At the AMP, all injured people are categorized, treated, and stabilized prior to their transportation to hospitals. However, severely injured patients might die at the location of the incident or during the rescue and treatment at the incident site. To support emergency health policy-makers such as the Austrian Samaritan Organization at a strategic, a tactical, and an operational level at an AMP, we have developed a discrete-event-based policy simulator. This tool helps emergency policy-makers to analyze not only vehicle and patient scheduling but also staff and material planning on site (Rauner et al. 2012).
123
A policy management game for mass casualty incidents
Subsequently, we transformed the policy simulator into a policy management game for a single player who has to make several key decisions during a MCI in discrete time intervals. This new tool can be used to train medical staff for rare MCIs. In addition, students can be educated on scheduling and planning techniques such as simulation, queuing theory, and resource allocation (Ozcan 2009; Pidd 2004; Pinedo 2008; Vissers and Beech 2005). We conducted an experimental study with 96 participants including students, practitioners from health care services, and researchers to investigate the potential of this AMP-Management Game. In the next section, we illustrate management games as experimental learning behavior environments. Section 3 describes the AMP-Management Game in detail, while Sect. 4 presents the questionnaire of our experimental study. Section 5 investigates the results of our experimental study concerning the evaluation of the general structure and organization as well as the process of decision-making and the results achieved by the players. Next, we statistically analyze the performance of the participants on selected outcome measures among runs depending on participant groups, gender, practical experience in emergency medical service, and experience with management games. The final section concludes the paper by drawing key policy implications for decision-makers.
2 Management games as experimental learning environments Providing health services comprises complex resource allocation decisions at a strategic, tactical, and operation management level (Brailsford et al. 2012). Thus, health care policy-makers need quantitative decision-making tools to cope with such situations. For this purpose, health care management games have gained importance for different stakeholders due to rising health care costs and scarce resources in the last decades. Health care management-oriented games are applied for educating health care students, medical staff and professionals, as well as health care policy-makers (Kraus et al. 2010) and focus on improving the decision-making process in complex medical environments (Lane et al. 2001). In addition to such games, medical treatment-oriented games and simulators utilize ‘‘standardized patients’’ and mannequins to prepare medical staff and students for real patient treatment (Streufert et al. 2001). In this paper, we focus on the application of a health care management-oriented game (cf. Kraus et al. 2010), the AMP-Management Game for MCIs. Another example of a health care management-oriented game is the ‘‘Operating Room Manager Game’’ in which players run an operating room department of a large hospital. The game involves a wide range of decisions from strategic choices on capacities to a daily online planning of surgeries for each operating room (Hans and Nieberg 2007). The internet-based ‘‘COREmain Hospital Game’’ offers a broad variety of decisions on resource, process, and financial result management in different departments to illustrate independencies among the different managerial goals internally, externally, and within the entire region (Rauner et al. 2008). Pfaff
123
M. S. Rauner et al.
et al. (2010, 2013) focused on decision-space visualization to support emergency responders to identify the most robust options for complex emergency situations. 2.1 General experimental learning structure Players learn to solve problems by identifying and evaluating relevant information and by trying out different decisions (Geier 2006). Being confronted with different situations awakens the interest and curiosity of the players (Schulmeister 1996). In addition, the players are motivated to act. By concentrating on the principles of problem-solving and self-monitoring of interest as well as motivation, the discovering learning coincides with the constructivist guidelines for the creation of learning environments (Geier 2006). The same principles can be discovered for management games. Depending on a current position, the players have to analyze the changed environment and make decisions several times (Geier 2006). They can try different hypotheses for decision-making. The impact of those game decisions on the simulated situation in turn has to be analyzed and evaluated too. The approach of learning through experience (‘‘experimental learning’’) by Kolb (1984) also contains the cyclic execution of the learning process. The building of experience is seen as a central criterion of the learning process. Learning is therefore the process of generating knowledge through the transformation of experience (Geier 2006). Experience acts as a starting point of the cyclic learning process. Games are the best way to facilitate ‘‘learning by doing’’ and ‘‘learning by failing’’ without negative consequences for the real world (Dieleman and Huisingh 2006). Most studies on educational games propose a learning model as illustrated in Fig. 1. As an input, instructional content has to be specified and all game characteristics have to be explained. The game cycle consists of the judgments or reactions of the user including entertainment and/or interest, the user behavior, as well as the system feedback. Depending on the successful combination of content PROCESS
INPUT
OUTCOME
Game Cycle Instructional Content
User Judgments
Learning Outcomes Game Characteristics
System Feedback
User Behavior
Fig. 1 Input-process-outcome game model (Garris et al. 2002)
123
A policy management game for mass casualty incidents
with game functions, the cycle results in recurring and self-motivated playing. Finally, the engagement with the game leads to an achievement of training goals and learning results (Garris et al. 2002). Each game run involves repeated judgment-behavior-feedback loops leading to increased interest, enjoyment, or involvement resulting in a greater persistence or intensity of effort. These behaviors of the player are rewarded by direct system feedback on performance in the game context (Garris et al. 2002). The instrument of management games is a learning and teaching method in which problem statements are systematically created to be solved by the participants (Blo¨tz 2005; Becker et al. 2007; Brandl et al. 2010). Management games are used to simulate operation or incidence situations and present them in a more comprehensible and easily assessable way. This is especially favorable for undetermined, nontransparent, or complex situations that need quick realization, evaluation, and solution. Usually, management games deal with situations that can hardly be demonstrated in reality or only with high (financial) expenditures. In particular, the gaming aspect together with the focus on the learning perspective distinguishes such management games from pure simulations. Management games can therefore be conducted especially as a preparation for complex situations in reality (Rempe and Klo¨sters 2006). They contain the possibility to experiment and to even make risky decisions. The outcomes are simulated without direct consequences for the involved parties. In this respect, experimenting with different possibilities does not harm anything or anyone, but can lead to a modified evaluation and perception of procedures and situations. The player should learn to act tactically and logically in different situations, to make suitable decisions, and to draw conclusions from different situations and decisions made. In the following section, we describe how these theories were implemented in our AMP-Management Game. 2.2 Experimental learning structure of the AMP-Management Game As MCIs are very complex and rare situations dealing with human lives, a management game is a very suitable instrument to educate managerial staff (e.g., incident commanders) in particular for acting at different positions within the organization of an AMP. The educated players can repeatedly learn in very different settings how a MCI should be handled, how the processes at an AMP should function, and where the bottlenecks at an AMP are. Additionally, our AMPManagement Game focuses on personnel placement planning (e.g., triage, treatment, and transport of patients) as a key factor of success (Sefrin 2005). To ensure optimal learning behavior in the AMP-Management Game (Peric and Herdlicka 2012), we developed a clear structure of six parts for each experiment consisting of three games runs for evaluation (cf. Fig. 2). We provided the participants with the basic knowledge needed to deal with MCIs and simultaneously explained the handling of the AMP-Management Game. We also ensured that all participants of our experiment had the same game conditions. We took into account that lapses in attention and memory resulted in some dysfunctions, including boredom after a certain time (cf. Carriere et al. 2008). These effects could diminish
123
M. S. Rauner et al.
Part 1
Introducon: AMP-Management Game (30-45 min)
Part 2
Short Break (5 min)
Part 3
Demonstraon: AMP-Management Game (15 min)
Part 4
Audion: AMP-Management Game (15 min)
Part 5
Rounds: Three runs of AMP-Management Game (60 min)
Part 6
Quesonnaire (15 min)
Fig. 2 Structure of the AMP-Management Game sessions
the results and the performance of the participants could vary over several runs. In a beta test of the AMP-Management Game, we tested the framework and found that players had paid attention during up to three runs. Thus, we investigated the learning behavior of the participants only for three evaluation runs. We started each experiment session with a presentation to introduce the concept of an AMP and the AMP-Management Game in general (part 1). After a short break (part 2), the handling of the AMP-Management Game and the three main decisions that could be taken were demonstrated (part 3; cf. Sect. 3). Then, a trial game could be played by the participants (part 4). During part 5, the participants had 60 min to play three runs for evaluation of the AMP-Management Game. Immediately afterwards, the players were asked to fill in a questionnaire (part 6; cf. Sect. 4). Two moderators set-up the AMP-Management Game and supervised the players. They prevented any interaction among the participants during game. There was only one break. Refreshments and a small buffet helped to keep up the motivation of the participants during the game sessions. Students were mainly extrinsically motivated to participate in the study by gaining additional credit points for their courses. In contrast, practitioners had an intrinsic motivation to play the game due to their general interest in health care management games. Although extrinsic factors are less effective than intrinsic ones, both factors play an important role in learning behavior (Garris et al. 2002).
3 The AMP-Management Game Our AMP-Management Game for MCIs deals with the complex process of an AMP to be built up on an incident site (cf. Sect. 1). It contains triage, treatment, and transportation of the patients (to, within, and from the AMP). The concrete
123
A policy management game for mass casualty incidents
Fig. 3 Structure of the AMP-Management Game
procedures of rescue at the incident site as well as the patients’ transportation to the hospitals are not part of the game (cf. Fig. 3). The situation settings at the beginning of the game allow the moderator of the game to set up various different scenarios from simple urban to rather complex remote incidents with many affected people and a high rate of severely injured ones. The AMPManagement Game was implemented by the software ‘‘Anylogic’’ (XJ Technologies Company 2007). As it can be exported as a JAVA Applet, this game can be played by a standard web browser on PCs or tablets, either locally or via the internet. Based on the simulation of MCIs (cf. Rauner et al. 2012), the user interface (cf. Fig. 4) is orientated towards the schematic representation of the Austrian MCI concept (cf. Sect. 1). It consists of the following main elements (cf. blue heading boxes): • • • • • • • • • •
Ambulance Vehicle Loading Site, Ambulance Vehicle Parking Space, Care Units (Treatment Rooms I, IIa, IIb, III, and IV), Command Center (EL), Fatalities Collection Point (t), Incident Site, Pre-Triage (BT), Registration Office, Triage (T), and Uninjured Collection Point (U).
Beside the standard components of an AMP, the AMP-Management Game’s user interface is characterized by several special features: •
The speed of the visualization of the simulated processes can be changed (As a standard value, one real minute is simulated in one second. For example, an incident that needs a total rescue time of 3 h, can be simulated within 3 min).
123
M. S. Rauner et al.
Fig. 4 User interface of the AMP-Management Game
•
The simulated real time is displayed by a clock in the upper left corner and the total rescue time is indicated in the lower left corner. The health status (triage group) of a patient is indicated by a colored dot beside the patient’s head after the pre-triage was done at the incident site. The health status of a patient changes over time and is a function of both waiting and
123
A policy management game for mass casualty incidents
treatment during the AMP-Management Game using scoring systems (cf. Rauner et al. 2012). The patients with the worst health status are fatal patients who either die when the incident occurs or during rescue and treatment at the incident site. To keep the AMP-Management Game simple, over-triage and its consequences are not considered yet. •
•
•
•
•
•
•
We assigned a single value for an individual’s health status based on standardized scoring systems (Horsman et al. 2003; Rabin and de Charro 2001) adapted to the Austrian situation. The values for the health status range between 0 and 100 based on Koehler et al. (1992) and Smith et al. (2009). A score of 0 represents ‘dead,’ while a score of 100 stands for ‘‘uninjured’’. At the beginning of the AMP simulation, each casuality is assigned a random initial health status between 0 and 100 using uniform distributions depending on the rate of the severely injured persons and the emergency scenario itself (cf. Rauner et al. 2012). Dead persons have 0 points, heavily injured persons obtain 1–40 points, slightly injured persons get 41–99 points, and not injured persons receive 100 points. After triage, the casualties are assigned based on their health status to treatment rooms (room IV: 1–10 points, room I: 11–20 points, room IIa: 21–40 points, room IIb: 41–65 points; room III: 66–99 points) or to collection points (dead persons: 0 points; uninjured persons: 100 points) as explained by Rauner et al. (2012). During waiting for treatment, the health status of a heavily injured person might deteriorate and he/she might even die. In addition, the health status of heavily injured persons might also worsen after treatment during waiting for transportation (cf. Rauner et al. 2012). We have modeled a linear decay approach based on expert interviews as described by Rauner et al. (2012). Treatment might improve the health status using a linear improvement approach (cf. Rauner et al. 2012).
For each care unit (treatment rooms I, IIa, IIb, III, and IV), the AMPManagement Game separately shows how many patients are waiting for care (WC), waiting for transfer (WTr) or transportation (WT), and how many treatment squads (TSq) are allocated. A treatment squad contains a certain number of doctors and medics depending on the treatment room: one doctor and two medics for treatment room I, one doctor and one medic for treatment rooms IIa, IIb, and IV, and one medic for treatment room III. The players are informed that each treatment room requires a certain squad-mix and also that the duration of treatment differs among treatment rooms. This information is essential for resource allocation of squads to treatment rooms (cf. Table 1). Furthermore, it can be seen how many physicians and medics are available at the registration office and could be sent to the care units for treatment of patients or for transportation of patients to the hospital.
123
M. S. Rauner et al. Table 1 General guidelines for decision-making in the AMP-Management Game Current situation
Potential decision
Long queue before triage (T)
Increase triage capacity
A
No queue before triage (T)
Decrease triage capacity
A
Waiting for transfer to another care unit (WTr) [ 0
Increase number of medics in AMP per 100 individuals affected (decrease number of transportation staff)
B
Waiting for transportation to hospital (WT) [ 0
Decrease number of medics in AMP per 100 individuals affected (increase number of transportation staff)
B
Waiting for medical care in treatment units I–IV (WC) [ 0
Increase number of medics in AMP per 100 individuals affected (decrease number of transportation staff)
B
Waiting for medical care in treatment units I–IV (WC) [ 0
Delegate additional treatment squads to treatment rooms I–IV
C
Idle treatment squads (TSq)
Withdraw treatment squads from treatment rooms I–IV to registration office
C
The aim of the AMP-Management Game is to improve the above-mentioned processes. Players act as commanders and allocate scarce personnel to triage, treatment, and transportation of patients. In order to rescue as many patients as possible and to quickly clear the incident site, the priority at the beginning of a game should be given to the triage. Afterwards the treatment of the patients, especially of the severely injured ones, should be the focus of the player’s attention. Towards the end of the game, transportation of the patients to the hospitals becomes more and more important. The handling of an intermediate emergency scenario from the incident until the entire clearing of the incident site might take about 180 min in reality, while such a scenario is played in about 20 min by the participants. Every 15 simulated emergency scenario’s minutes (which is less than a real minute for a player), the AMP-Management Game automatically stops and the players have to take the following three decisions A to C (cf. Fig. 5). Depending on the duration of a scenario played, this decision-making process happens about twelve times per game run for an intermediate emergency scenario:
123
A policy management game for mass casualty incidents
Fig. 5 Three decisions made during the AMP-Management Game
Decision A) Triage capacity Decision B) Medics in AMP
Decision C) Sending treatment squads
How many physicians and medics should be used to categorize the patients? How many medics are necessary in the AMP for treatment and transport of patients within the AMP and how many medics should wait in the ambulance vehicles to transport the patients to the hospitals? How many treatment squads (medical doctors and medics) are needed in which care units (treatment room I to IV) for patient treatment?
For players without previous knowledge about the AMP concept, these decisions are rather complex. For this reason, a short four-page guidance summary on all essential information and decisions to be taken are given to the players in the introduction session of the AMP-Management Game (cf. Fig. 2). During the demonstration and audition sessions, the players get familiar with the game to finally play the three evaluation rounds. This process ensures that the players who are not familiar with the management of MCI at an AMP can develop and improve strategies to cope with such decision problems by using the AMP-Management Game. For example, we prepared general decision guidelines for them as illustrated in Table 1. We briefly explained for which state of the AMP which potential decisions should be made based on main system parameters such as length of queues or number of patients waiting. These guidelines helped the players to better understand the main activities of the AMP-Management Game.
123
M. S. Rauner et al.
Fig. 6 Main statistics of a game run
The players learn how to improve the treatment of the patients and their transportation to hospitals by allocating scarce medical staff to the corresponding activities at the AMP (decisions A, B, and C). The main decision-making goal focuses on saving human lives as the highest priority. A lower priority is generally given to the clearing of the incident site, except if the incident happened at a critical infrastructural point (e.g. airport, train station). Depending on the selected scenario and the explained goal of the rescue mission, the players focus on improving different performance measures such as average waiting time for treatment, number of fatalities, and overall duration (cf. Fig. 6). In the introduction session (cf. Fig. 2), the instructors of the AMP-Management Game explain which performance measures or weighted performance measures are used to benchmark the success of the players. At the end of each game, the player can evaluate his/her success regarding quickly treating patients and afterwards transporting them to hospitals. For this purpose, we provide main statistics for each game run including patient-related and staff-related outcome measures (cf. Fig. 6). The main patient-related outcome measures include: (1) patients’ waiting times for treatment in each treatment room I–IV (e.g., treatment room I: 118 min (minimum); treatment room IIb: 141 min (maximum)), (2) number of fatalities (e.g., 14 fatalities until triage ? 2 fatalities in treatment rooms = 16 fatalities), and (3) overall duration (total rescue time, e.g., 320.17 min). The staff-related outcome measures comprise the utilization rates of physicians and medics during the entire game run as well as their average overall utilization
123
A policy management game for mass casualty incidents
rates. Especially, the number of medics directly after the MCI is a general bottleneck. Their allocation planning plays a key role for decision-making (decisions A, B, and C). With this structured system feedback and the possibility to replay a certain scenario, the players can judge their achievement, adapt their behavior, and improve outcomes in the subsequent run. This judgment-behavior-feedback loop leads to high learning behavior outcomes as explained in Sect. 2. We found that the concentration of the players lasts for up to three game runs in one experimental session. As explained above, the AMP-Management Game both displays the results of each run and also saves all essential patient-related and staff-related outcome measures as well as all decisions made by the player in an output file. These data are essential for the statistical analyses of the experimental economic study (cf. Sect. 6). In the AMP-Management Game, players learn the key concepts behind efficiency (‘‘doing things right’’) and effectiveness (‘‘doing the right things’’) applied in the performance literature of management/economics (Drucker 2006; Ozcan 2008). Both efficiency and effectiveness influence performance/outcome measures. During the game, the players aim at improving efficiency (cf. Fig. 5 and Table 1) in order to optimize the performance/outcome measures of the AMP (cf. Fig. 6) for a given number of inputs (i.e., medical staff available). In addition, players understand that enhanced effectiveness is necessary to achieve best possible outcome measures (cf. Fig. 6) by improved decision making (cf. Fig. 5 and Table 1). The players apply basic management concepts such as improved planning and scheduling to reach this aim. The better the outcome measures of a certain player, the better his/her performance in managing the AMP. For example, short waiting times for treatments reduce the overall duration. Consequently, the incident site will be cleared sooner. The medical staff will become quicker available for further duties. The incidents site will become productive earlier. In addition, fewer fatalities and/or less severely injured individuals due to quicker/better treatment denote a gain in productivity for the society.
4 The questionnaires of the experimental economic study To gain deeper insight into the players’ decision-making process and the quality of our AMP-Management Game, we developed two questionnaires (Peric and Herdlicka 2012): (1) result questionnaire (sub-Sect. 4.1) and (2) evaluation questionnaire (sub-Sect. 4.2). The result questionnaire focuses on economic outcome measures to both increase efficiency as well as effectiveness of rescue and treatment at the AMP as discussed at the end of Sect. 3. 4.1 Result questionnaire A result questionnaire for manual completion was designed. The participants had to fill in outcome measures of the AMP-Management Game such as minimum and maximum average waiting times for treatment, number of fatalities, overall duration (total rescue time), and the overall utilization rates of physicians and medics. In
123
M. S. Rauner et al.
addition, the participants had to prioritize the five main outcome measures for each evaluation run: (1) overall utilization rate of physicians, (2) overall utilization rate of medics, (3) overall duration (total rescue time), (4) average waiting time for treatment, and (5) number of fatalities. This questionnaire supported the participants in judging and benchmarking their results of the three game runs. In addition, this information was a useful indicator for an increasing or a decreasing performance of the participants among the runs. All decisions made by the participants during each evaluation run and the outcome measures at the end of each evaluation run were electronically reported for further evaluation. 4.2 Evaluation questionnaire An evaluation questionnaire with three sections was developed to obtain additional information on: (1) the participants’ experiences, (2) the participants’ evaluation of the game session, and (3) the participants’ evaluation of the AMP-Management Game. 1st section: contained general questions about previous knowledge in the fields of project management, emergency medical service, and/or management games. 2nd section: surveyed the ratings of the participants on the quality of the AMPManagement Game sessions with a four-stage scale: (1) very good, (2) good, (3) satisfactory, and 4) unsatisfactory. 3rd section: dealt with the participants’ evaluation of the AMP-Management Game and was divided into three parts: (1) usability, (2) decision-making, and (3) demographics of the participants. The first part dealt with the entertainment factor of the game, the challenge for the players, traceability of the processes and results, the applicability of the AMP-Management Game in practice as well as improvement of the participants’ performance during game runs. All these questions had the following four-stage answer scale to avoid average rating (Brandt 2008): (1) very, (2) quite, (3) not very, and (4) not at all. The second part investigated the prioritization of the five outcome measures: (1) mean utilization of medics, (2) mean utilization of physicians, (3) total rescue time, (4) mean waiting time, and (5) number of fatalities. In the third part, the demographic data (age, gender, education, and current profession) of the participants were collected. Furthermore, the participants were assigned to one of the following participant groups: current Bachelor students, former master students, and PhD students (mainly of International Business Administration and Business Administration); practitioners from health care services including emergency medical services; and researchers. For these participant groups, we analyzed their performance differences (cf. Sect. 6).
5 General results of the experimental study The first experimental economic study of our AMP-Management Game took place at the University of Vienna, Austria, in May and June 2011 and was supported by the Experimental Economical Center of the University of Graz, Austria (Peric and
123
A policy management game for mass casualty incidents
Herdlicka 2012). The main groups of participants consisted of Bachelor students of International Business Administration and Business Administration of the University of Vienna with no experience in the health care sector (called ‘‘greenhorns’’) and practitioners from health care and emergency medical services (called ‘‘professionals’’). The 96 participants were split into eight groups with twelve participants each. For all eight groups, we prepared an identical setting by using our mobile management game laboratory consisting of laptops and additional equipment. The two game organizers supervised the twelve participants who were located on distant tables in a lecture room to ensure that participants could not interact with their neighbors. The organization of the game session is shown in Fig. 2 and each participant played a trial run and three evaluated game runs. Furthermore, the participants filled in questionnaires for further evaluation (cf. Sect. 4). As a realistic scenario for an MCI for the AMP-Management Game, we chose a gas explosion at a small town’s farmers’ market. The sudden, shocking explosion caused a mass panic. In total, 80 people were affected and 40 % of them were severely injured. All participants were given this identical scenario in all three evaluated game runs. Prior to this experimental economic study, the AMP-Management Game software was tested and optimized for user-friendliness. A large group of Bachelor students of International Business Administration and Business Administration acted as beta testers for the software by obtaining credit points for their university courses. They reported errors and suggested improvements. Furthermore, the structure of the AMP-Management Game sessions and the accompanying presentation of the instructors, the handouts given to the players, and the questionnaires were evaluated and improved too. After this trial phase, the eight game sessions were held as described above. The feedback of the participants on the general structure and organization of the game sessions is analyzed in sub-Sect. 5.1. Sub-Sect. 5.2 provides insight into the decision-making process of the participants, while sub-Sect. 5.3 analyzes the results achieved by the participants. 5.1 Evaluation of the general structure and organization by the players As mentioned in Sect. 4.2, we used a numeric scale from 1 to 4 to obtain participants’ evaluation of the game sessions (1 = very good, 2 = good, 3 = satisfactory, and 4 = unsatisfactory). The participants of the AMP-Management Game reported a high satisfaction rate with the organization of the sessions (mean value = 1.14; standard deviation = 0.37). We received a high rating for the introductory session of the AMP-Management Game (mean value = 1.32; standard deviation = 0.49) and for the expertise of the AMP-Management Game instructors (mean value = 1.30; standard deviation = 0.46). These results confirmed that the AMP-Management Game sessions were professionally organized. The participants did not report any barriers for experimental learning behavior and success during the exercise. Furthermore, the participants enjoyed playing the AMP-Management Game because 61 % of the participants evaluated the entertainment factor as high and
123
M. S. Rauner et al.
35 % as rather high (1 = high, 2 = rather high, 3 = rather not, and 4 = certainly not). In total, 56 % of the participants felt rather challenged by the AMPManagement Game (mean value = 2.13; standard deviation = 0.70). Only 17 % were highly challenged, 25 % did not feel rather challenged, and only 2 % reported no challenge. According to the theory on experimental learning environments (cf. Geier 2006), the AMP-Management Game enormously motivated the participants to rescue human lives and thus enhanced their learning curves. A key issue for the application of management games encompasses the quality, scope, and traceability of the underlying processes for which we used the following scale (Rempe and Klo¨sters 2006): 1 = high, 2 = rather high, 3 = rather not, and 4 = certainly not. Most of the participants clearly understood the processes (mean value = 1.65; standard deviation = 0.62). It was also obvious for the participants how the results of the AMP-Management Game were calculated (mean value = 1.90; standard deviation = 0.62). These results significantly demonstrate the importance of a high quality introductory session to an unknown topic. Such a session supports the qualified decision-making of participants in a training situation. In addition, the participants evaluated the applicability of the AMP-Management Game in practice with a mean value of 2.10 (standard deviation = 0.72). About 18 % of the students, 10 % of the practitioners from medical services, 25 % of the health care practitioners, and 50 % of the researchers highly rated the practical relevance of the AMP-Management Game. Nobody of the 96 participants found the AMP-Management Game absolutely unsuitable for practical application. 5.2 Evaluation of the process on decision-making by the players Several questions in the questionnaire dealt with the self-evaluation of the participants’ performance during the runs. A significant and a rather strong performance increase in the second run were reported by 21 and 49 % of the participants, respectively. While 70 % of the players stated an increase in their performance from the first to the second run, only 56 % of the players could reach an improvement in performance in the third run. This self-evaluation of the players is statistically tested in sub-Sect. 6.6. We observed that 78 % of the participants (mean value = 1.86; standard deviation = 0.88) were motivated to play additional runs of the AMP-Management Game to improve their performance and to try further strategies. We disclosed that despite the seriousness of the topic, an entertainment factor existed. This demonstrates that the complex processes of a rescue operation after a MCI can be learnt by playing an artificial game situation. The participants had to prioritize the following five outcome measures: (1) mean utilization of medics, (2) mean utilization of physicians, (3) total rescue time, (4) average waiting time for treatment of patients, and (5) number of fatalities. A value of one, for example, indicated the highest priority for a certain outcome measure. As anticipated, the number of fatalities had the highest impact on the decision-making for 40 % of the players. The utilization of physicians scored 2.18 (standard deviation of 1.18), the number of fatalities reached 2.43 (standard deviation of 1.42), while the utilization of medics gained 3.23 (standard deviation of 1.09). This
123
A policy management game for mass casualty incidents Table 2 Results of the key outcome measures of the AMP-Management Game for all 96 participants from run 1 to run 3 Number of fatalities
Total rescue time
Run 1
Run 2
Run 3
Run 1
17.28
16.63
17.97
224.51
227.63
3.24
2.81
3.11
31.16
44.06
44.71
Median
17.00
17.00
18.00
221.00
216.00
220.00
Variance
10.52
7.88
9.69
971.24
1941.67
1999.14
Skewness
0.25
0.38
0.25
1.21
3.69
5.59
Minimum
10.00
9.00
11.00
174.00
178.00
153.00
Maximum
27.00
25.00
26.00
335.00
524.00
585.00
Mean Standard deviation
Run 2
Run 3 224.29
prioritization of the outcome measures by the players did not significantly change among runs. A worrying result was that the participants did not focus on total rescue time during the entire AMP-Management Game session. The participants primarily focused on the number of fatalities and efficient human resource management. The main goal of emergency medicine to optimally and quickly treat many patients and transport them to hospitals was obviously not followed by the greenhorns (Hansak et al. 2003). Only the practitioners in emergency medical services imposed a higher priority on the parameter of total rescue time and on parameters of waiting times for treatment of patients. 5.3 Evaluation of the results achieved by the players The rescue principle to optimally and quickly treat many patients and transport them to hospitals is mainly reflected by the outcome measure of minimizing the number of fatalities as well as minimizing the total rescue time (Hansak et al. 2003). Thus, we investigated these two parameters for all 96 participants in detail (cf. Table 2), while the differences among participant groups, genders, practical experience in emergency medical services, and experience with management games are statistically analyzed in Sect. 6. We identified a U-shaped distribution of the number of fatalities. While the mean number of fatalities decreased from run 1 to run 2, it increased from run 2 to run 3. However, the median number nearly remained the same in all three runs. Thus, we used a number of 17 fatalities as a threshold value between ‘‘good’’ and ‘‘poor’’ players. The minimum and maximum values for the number of fatalities reflected the same trend as the mean value. The outcome measure of the total rescue time did not show a specific distribution pattern. For example, the median total rescue time ranged from 216 to 221 min. During an experimental study, researchers observed that the longer the experiments took the higher the learning behavior success might be, but the lower the attention level became and the more likely an end effect would
123
M. S. Rauner et al.
occur (cf. Rapoport and Dale 1966; Carriere et al. 2008). These effects will be discussed below in detail. The participants were able to modify the decisions every 10 min of the real simulated time. We compared their results to benchmark results of an optimized run (1,800 repetitions) in which a perfect commander made decisions based on advanced knowledge whenever necessary. For such a perfect commander, we got a mean of 16.39 fatalities as a benchmark result which was very close to the mean number of fatalities of the second run achieved by the participants. The perfect commander’s benchmark of the mean total rescue time was 218.37 min compared to 227.63 min of the participants in the second run.
6 Statistical analyses of the experimental study In this section, we statistically investigate the performance of the participants among the three runs in detail (Peric and Herdlicka 2012). The experience gained in one game run represents the feedback to the cyclic process of experimental learning behavior (cf. sub-Sect. 2.1). The analyses were based on the outcome measure of the numbers of fatalities depending on participant group, gender, practical experience in emergency medical service, and experience with management games. We exemplarily tested six hypotheses H1 to H6 related to learning behavior tendencies in sub-Sects. 6.1, 6.2, 6.3, 6.4, 6.5, 6.6, respectively. 6.1 Performance increase by exercise First, we tested for the performance increase and learning behavior effects of participants during the experiment using the Mann–Whitney U test (also known as Wilcoxon rank-sum test): H1
There were no performance differences between run 1 and run 2.
Table 3 shows that the differences among the number of fatalities in run 2 compared to run 1 were significant (p = 0.090). The number of 54 negative differences/ranks (i.e., increase in performance) was considerably higher than the number of 31 positive ones (i.e., decrease in performance). Furthermore, 11
Table 3 Mann–Whitney U test: number of fatalities in run 2 versus run 1 Ranks
N
Mean rank
Sum of ranks
Number of fatalities in run 2–
Negative ranks
54
40.98
2213.00
Number of fatalities in run 1
Positive ranks
31
46.52
1442.00
Bindings
11
Total
96
Test statistics
Z Asymptotic significance (2-tailed)
123
-1.695 0.090
A policy management game for mass casualty incidents
participants achieved the same number of fatalities in run 1 and run 2 (i.e., bindings). Then, we further investigated the performance increase by exercise of our participants depending on the following split categories: (1) participant group, (2) gender, (3) practical experience in emergency medical service, and (4) experience with management games using the Mann–Whitney U test. Although negative ranks predominated in each split category, only two results were significant (cf. Table 4). Only practitioners from the emergency medical service and participants with experience in management games could significantly improve their performance. For example, while nine of ten practitioners from the emergency medical service improved their performance in run 2, only one of them stagnated. Furthermore, 21 of 29 participants with experience in management games could increase their performance in run 2, while five decreased their performance and three stagnated. To summarize and which was also intuitively assumed, we could statistically prove that the performance significantly increased by exercise and found related learning behavior success for some of our participants from run 1 to run 2. 6.2 Influence of experimentation The second hypothesis H2 analyzed the influence of experimentation on the performance in the last game run (i.e., run 3): H2
There were no performance differences between run 2 and run 3.
We again used the Mann–Whitney U test to investigate the performance differences of the participants regarding the number of fatalities in runs 2 and 3. With 48 positive ranks, exactly 50 % of the participants achieved a higher number of fatalities in run 3 than in run 2 and thus worsened their performance (p = 0.004). We found significant differences among the following split categories (cf. Table 5): (1) Bachelor students, (2) practitioners from health care, (3) male participants, (4) participants without practical experience in emergency medical service, and (5) participants without experience in management games. This result clearly illustrated the trend to experimentation of the players in the last run. The participants’ low performance in the third run can be explained by the phenomenon of the end effect. Among other factors, the distribution of the results is affected by the so-called ‘‘end effect’’, which has a measurable impact not only on the distribution in general but also on the performance on the last round(s) of a session. The end effect has been discussed in various versions as well as contexts and is most intensively studied for game theoretical situations (cf. Rapoport and Dale 1966). 6.3 Performance differences among the participant groups We further examined whether or not the performance differed between the practitioners (‘‘professionals’’) and students (‘‘greenhorns’’): H3
There were no performance differences between practitioners and students.
123
123
* p B 0.05; ** p B 0.01; *** p B 0.001
Experience with management games
Practical experience in emergency medical services
Gender
No
Yes p = 0.540
p = 0.138 p = 0.263
Yes
No
p = 0.214 p = 0.228
Female
Male
p = 0.306 p = 0.655
Practitioners from health care
Research
Practitioners from emergency medical services
p = 0.408 p = 0.715
Bachelor students
Former master students and PhD candidates
Participant group
Not significant
Parameter value
Split category (variable)
Table 4 Mann–Whitney U test: number of fatalities in run 2 versus run 1 by split category
p = 0.008**
p = 0.007**
Significant
32
21
43
11
28
26
1
5
9
2
37
Negative ranks
26
5
26
5
15
16
1
2
0
2
26
Positive ranks
8
3
9
2
6
5
0
1
1
0
9
Bindings
66
29
78
18
49
47
2
8
10
4
72
N
M. S. Rauner et al.
* p B 0.05; ** p B 0.01; *** p B 0.001
Experience with management games
Practical experience in emergency medical services
Gender
No
Yes
No
Yes
Male
Female
Research
p = 0.086
p = 0.102
p = 0.170
p = 0.655
p = 0.327
Practitioners from emergency medical services
Practitioners from health care
p = 0.461
Bachelor students
Participant group
Not significant
Former master students and PhD candidates
Parameter value
Split category (variable)
Table 5 Mann–Whitney U test: number of fatalities in run 3 versus run 2 split by category
p = 0.009**
p = 0.016*
p = 0.014*
p = 0.046*
p = 0.038*
Significant
21
11
28
5
14
19
1
1
3
2
26
Negative ranks
34
14
39
9
26
22
1
5
5
2
35
Positive Ranks
11
4
11
4
9
6
0
2
2
0
11
Bindings
66
29
78
18
49
47
2
8
10
4
72
N
A policy management game for mass casualty incidents
123
M. S. Rauner et al.
One would have intuitively assumed a difference among these two groups, which has to be statistically tested. Using the Chi-squared test for independence, we tested hypothesis H3 by comparing the outcome measure ‘‘number of fatalities’’ of each run among participant groups (i.e. Bachelor students, former master students and PhD candidates, practitioners from emergency medical services, practitioners from health care, and researchers). For this outcome measure of run 2, the Chi-squared test showed a highly significant difference (Chi-squared value: 95.07; degrees of freedom = 15; significance level: p \ 0.000) among the results of participant groups. However, it could not be determined which participant group was responsible for the differences. As Table 4 from sub-Sect. 6.1 shows that practitioners from the emergency medical services significantly reached better results compared to other participant groups for run 2 (significance level: p = 0.039), hypothesis H3 can be clearly rejected for run 2. Due to the fact that hypothesis 3 could only be rejected for run 2, we conducted an additional Chi-squared test to analyze the results of run 3 by merging the participant groups into two aggregated participant groups: students (‘‘greenhorns’’) and practitioners (‘‘professionals’’: former Master students and PhD candidates, practitioners from emergency medical services and health care, researchers). By applying this grouping, we additionally determined a significant difference between the results of ‘‘professionals’’ and ‘‘students’’ for run 3 (Chi-squared value: 24.51; degrees of freedom = 15; significance level: p = 0.057). Thus, hypothesis H3 can be rejected for the two aggregated participant groups of run 3 too. As hypothesis H3 for run 1 could not be rejected, we assumed that most participant groups used run 1 as a basic learning run. It has to be mentioned that run 1 was the first real evaluation run for all participants in the AMP-Management Game. This effect is well known for complex management games (Dill and Doppelt 1963). An additional Chi-squared test disclosed that female participants felt more challenged by the AMP-Management Game compared to male participants (p = 0.018). The ‘‘Doing Gender’’ concept provides a possible explanation for this effect (Paseka 2008; Solga and Pfahl 2009). Technical knowledge, interest in computers, and playing computer games corresponds primarily with the male gender stereotype. This might be the reason why male participants felt less challenged by the AMP-Management Game than female participants. 6.4 No random effect of best performance Next, we tested whether or not the best performance of a single game run was a random effect. We compared the rank order of all participants by two performance indicators for the three runs (best and second-best performances) using hypothesis H4: H4
The best performance randomly arose.
The Spearman’s rank correlation coefficient, rs, is a statistical measure of the strength of a monotonic relationship between paired data. The closer rs is to ± 1, the stronger the monotonic relationship. For H4, rs amounted to 0.52 (significance level: p \ 0.01) and proved a positive correlation between the best and the second-
123
A policy management game for mass casualty incidents
best performances of all participants. Thus, the hypothesis H4 can be rejected as the best performances did not randomly arise. 6.5 Existence of ‘‘good’’ and ‘‘poor’’ players We then examined the existence of ‘‘good’’ and ‘‘poor’’ players among the participants by investigating hypothesis H5: H5 The performance of an arbitrary participant cannot be continuously classified as ‘‘good’’ or ‘‘poor’’. We used the median of the number of fatalities as a threshold value between ‘‘good’’ and ‘‘poor’’ players (cf. Table 2). We classified participants with a number of fatalities B17 as ‘‘good’’ players, while participants that reached a number of 18 C fatalities are assigned to the group of ‘‘poor’’ players. First, we analyzed how the degree of membership to the group of ‘‘good’’ and ‘‘poor’’ players changed from one run to another and the coincidence by the McNemar’s test. The McNemar’s test is applied to a matched pair of samples (in our case the runs) and analyzes by a 2 9 2 contingency table whether or not marginal frequencies of rows and columns are equal (Benesch 2008). We tested two different threshold values for our classification scheme: (1) ‘‘good’’ players with a number of fatalities B17 (cf. Table 6) and (2) ‘‘very good’’ players with a number of fatalities B16. Table 6 shows that nine more participants improved compared to worsened (25–16) from run 1 to run 2. This difference was not significant (p = 0.212). From run 1 to run 3, nine more participants worsened compared to improved (34–25) which was not significant (p = 0.298), while 18 more participants worsened compared to improved from run 2 to run 3 (32–14) which was significant (p = 0.012). Thus, ‘‘good’’ players used run 3 as an experimental run. For the ‘‘very good’’ players’ threshold value of 16 fatalities, we could not find any statistical significance among performance differences of the three runs (p = 0.12–0.658). The McNemar’s Test disregarded which participants remained in the same classification category between two runs. Thus, we conducted a sensitivity analysis of ‘‘good’’ and ‘‘poor’’ players among the three runs for ‘‘good’’ players (cf. Table 7) and ‘‘poor’’ players (cf. Table 8). The sensitivity analysis on the threshold value of the number of fatalities between 12 and 17 showed that the continuity of ‘‘good’’ and ‘‘poor’’ results of the players changed among the runs. For example, 18 participants could be classified as ‘‘good’’ players in all three runs for a threshold value of 17 fatalities, while five ‘‘very good’’ players could be identified in all three runs for a threshold value of 16 fatalities. Continuously ‘‘very good’’ results for low threshold values could be achieved by only a few players. For a threshold value of 17 fatalities, 18 participants could be classified continuously as ‘‘good’’ players, while only 7 participants could be assigned continuously as ‘‘poor’’ players (cf. Tables 7 and 8). The lower the threshold value for the number of fatalities, the more continuously ‘‘poor’’ players could be identified (cf. Table 8).
123
123
1.561
0.212
Chi-square
Exact significance (2-sided)
25
Poor
96
37
Good
Good
18
16
Poor
0.298
1.085
96
Poor
Good
Number of players in run 1
Number of players in run 1
Number of players in run 2
Run 1 versus run 3
Run 1 versus run 2
N
Analysis of ‘‘good’’ players
Table 6 McNemar’s test: classification of ‘‘good’’ players with a number of fatalities B17
25
19
Good
18
34
Poor
Number of players in run 3
0.012
6.283
96
Poor
Good
Number of players in run 2
Run 2 versus run 3
14
30
Good
20
32
Poor
Number of players in run 3
M. S. Rauner et al.
A policy management game for mass casualty incidents Table 7 Sensitivity analysis on the threshold value for the number of fatalities: Classification of ‘‘good’’ players Number of fatalities
Number of ‘‘good’’ players in
Number of ‘‘good’’ players in
Number of ‘‘poor’’ players in
Threshold value
Runs
Run
Runs
1 and 2
2 and 3
1 and 3
1, 2, and 3
1
2
3
1, 2, and 3
17
37
30
19
18
53
62
44
7
16
18
12
9
5
39
43
31
17
15
10
7
4
1
28
34
23
31
14
6
1
1
0
19
21
12
52
13
2
0
0
0
10
12
8
68
12
0
0
0
0
5
5
2
84
Table 8 Sensitivity analysis on the threshold value for the number of fatalities: Classification of ‘‘poor’’ players Number of fatalities
Number of ‘‘poor’’ players in
Number of ‘‘poor’’ players in
Number of ‘‘good’’ players in
Threshold value
Runs
Run
Runs
1 and 2
2 and 3
1 and 3
17
18
20
18
16
32
34
35
15
42
48
49
14
58
67
65
13
73
79
12
85
90
1, 2, and 3
1
2
3
1, 2, and 3
7
43
34
52
18
17
57
53
65
5
31
68
62
73
1
52
77
75
84
0
78
68
86
84
88
0
89
84
91
91
94
0
Next, we analyzed in detail the single player results for the 18 ‘‘good’’ and seven ‘‘poor’’ participants (threshold value of 17 fatalities). While the five worst players of the ‘‘good’’ players achieved a mean value of 16 fatalities, the five best players of the ‘‘poor’’ players reached a mean value of 19 fatalities. To summarize, only for the threshold value of 17 fatalities, we could show that ‘‘good’’ players achieved good results during the whole experiment. Once, the threshold value amounted to 16 fatalities, we disclosed the effect that ‘‘poor’’ players achieved poor results during the entire experiment. 6.6 Self-evaluation of participants Our last hypothesis dealt with the self-evaluation of the participants: H6 There exists a difference between the self-evaluation and the real performance of the participants.
123
M. S. Rauner et al.
Using the Chi-squared test for independence, we only proved for run 2 that hypothesis H6 could be rejected (Chi-squared value: 59.53; degrees of freedom = 39; significance level: p = 0.019). Thus, the self-evaluation was different from the real performance of the participants in runs 1 and 3.
7 Conclusion and policy implications We developed an AMP-policy simulator for MCIs in Central European countries such as Germany and Austria (Rauner et al. 2012). This tool helps emergency policy-makers to analyze not only vehicle and patient scheduling but also staff and material planning on site. In this paper, we present the improved AMP-Management Game that can be used to train medical staff for rare MCI and educate students on scheduling and planning techniques/methods such as simulation, queuing theory, and resource allocation (Ozcan 2009; Pidd 2004; Pinedo 2008; Vissers and Beech 2005). Currently, the Austrian Samaritan Organization applies the AMP-Management Game for training medical staff. We conducted an experimental study with 96 participants, including students, practitioners from health care services, and researchers to investigate the potential of the AMP-Management Game. We investigated three evaluation runs to account for the effects resulting from lapses of attention and memory (including boredom) which occurred after a certain time (cf. Carriere et al. 2008). The participants primarily focused on the number of fatalities and efficient human resource management. We found that the complex processes of a rescue operation after an MCI could be learnt by playing an artificial game situation. To summarize, we could statistically prove the performance increase of participants by practicing during the three runs and the related learning behavior success for the AMP-Management Game. For example, practitioners from the emergency medical services significantly reached better results compared to other participant groups in runs 2 and 3. The participants’ low performance in the third run can be explained by the phenomenon of the end effect (cf. Rapoport and Dale 1966). We found that female participants felt more challenged, which could be explained by the ‘‘Doing Gender’’ concept (Paseka 2008; Solga and Pfahl 2009). We disclosed that the first run could be characterized as a basic learning run similar to the literature (Dill and Doppelt 1963). Generally, participants reached their best results in the second run, whereas the third and last run was mostly used to experiment with alternative rescue and resource allocation strategies. In addition, good results of participants were not single successes. We further confirmed the non-random effect of participants’ performance. In addition, the self-evaluation of the participants’ performance differed from their real performance in runs 1 and 3. The investigation of learning behavior by educational neuroscience is a significantly emerging field that brings together researchers from different fields, mainly cognitive neuroscience, educational psychology and technology, as well as further related disciplines to explore the interactions between biological processes and education in general and learning as a specific process. Neuroeconomics seeks
123
A policy management game for mass casualty incidents
to ground economic theory in detailed neural mechanisms which are expressed in a quantitative way and make behavioral predictions (cf. Camerer 2007; Fischer 2009). The science of learning and becoming advanced in specific fields is nowadays dominated by neuroscientists. It is work in progress and will become a new field of great importance. For practical application, learning behavior effects might even be higher for additional realistic scenarios (e.g., urban versus rural MCI sites, complex versus simple circumstances) and even more game runs. Regularly, the AMP-Management Game has to be adapted to the current rescue principle such as the strict transportation rule of patients from only two treatment rooms, IIa and IIb to ¨ sterreichisches Rotes Kreuz 2007). In addition, the user interface could hospitals (O be improved with regard to detailed outcome measures for pre-triage and triage. The differences among participant groups, gender, culture, and countries as well as the effect of experience with computer games offer high potential for further research. As a next step, our AMP-Management Game will be expanded by a heuristic optimization approach for generating benchmark solutions of different scenarios for players.
References Batty M (2008) The size, scale, and shape of cities. Science 319:769–777 Becker O, Feit T, Hofer V, Leopold-Wildburger U, Selten R (2007) Educational effects in an experiment with the management game SINTO-Market. Cent Eur J Oper Res Econ 15(4):301–308 Benesch T (2008) Der Schlu¨ssel zur Statistik: datenbeurteilung mithilfe von SPSS. Facultas Verlag Blo¨tz U (2005) Planspiele in der beruflichen Bildung. W. Bertelsmann Verlag, Bielefeld Brailsford S, Kozan E, Rauner MS (2012) Health care management. Flex Serv Manuf J 24(4):375–378 Brandl B, Leopold-Wildburger U, Mietek A, Pickl S (2010) How do commission rates influence a firm’s success? Statistical analysis of a corporate strategy simulation experiment. CEJOR 18(4):553–566 Brandt H (2008) Planung und entwicklung von psychologischen tests und fragebogen. Testtheorie und Fragebogenkonstruktion. Springer, Berlin, pp 27–72 Camerer CF (2007) Neuroeconomics: using neuroscience to make economic predictions. Econ J 117(519):C26–C42 Carriere JSA, Cheyne JA, Smilek D (2008) Everyday attention lapses and memory failures: the affective consequences of mindlessness. Conscious Cogn 17(3):835–847 Center for Research on the Epidemiology of Disasters - CRED (2013) EM-DAT: the international disaster database. Retrieved October 21, 2013, from http://www.emdat.be/ Dieleman H, Huisingh D (2006) Games by which to learn and teach about sustainable development: exploring the relevance of games and experiential learning for sustainability. J Clean Prod 14(9):837–847 Dill WR, Doppelt N (1963) The acquisition of experience in a complex management game. Manage Sci 10(1):30–46 Drucker PF (2006) Classic drucker: essential wisdom of peter drucker from the pages of harvard business review. Harvard Business Press, Harvard Fischer KW (2009) Mind, brain, and education: building a scientific groundwork for learning and teaching. Mind Brain Educ 3(1):3–16 Garris R, Ahlers R, Driskell J (2002) Games, motivation, and learning: a research and practice model. Simul Gaming 33(4):441–467 Geier B (2006) Evaluation eines netzbasierten unternehmensplanspiels—Eine problemorientierte lernumgebung fu¨r die kaufma¨nnische Aus- und Weiterbildung. Doktorarbeit, LMU, Mu¨nchen Hans E, Nieberg T (2007) Operating room manager game. Inf Trans Educ 8(1):25–36
123
M. S. Rauner et al. Hansak P, Petutschnigg B, Bo¨bel M, Hu¨ndorf H-P, Lipp R, Veith J (2003) LPN-San o¨sterreich—lehrbuch fu¨r rettungssanita¨ter, betriebssanita¨ter und bundesheersanita¨ter. Stumpf and Kossendey, Edewecht Horsman J, Furlong W, Feeny D, Torrance G (2003) The health utilities index (HUI): concepts, measurement properties and applications. Health Qual Life Outcomes 1:54 Koehler G, Foley D, Jones M (1992) A computer simulation of a California casualty collection point used to respond to a major earthquake. Prehosp Disaster Med 7(4):339–347 Kolb D (1984) Experiential learning: experience as the source of learning and development. Englewood Cliffs, New Jersey Kotz D (2013) Injury toll from Marathon bombs reduced to 264. The Boston Globe, Boston Kraus M, Rauner MS, Schwarz S (2010) Hospital management games: a taxonomy and extensive review. CEJOR 18(4):567–591 Lane J, Slavin S, Ziv A (2001) Simulation in medical education: a review. Simul Gaming 32(3):297–314 Lown BA (2013) Reflections on compassion in the midst of violence. J Gen Intern Med 28(12):1554–1555 Marot J (2013) Zugunglu¨ck in Spanien: Lokfu¨hrer viel zu schnell unterwegs. Der Standard Marx F, Binsfeld M, Franke T (2013) Medizinischer Einsatz bei der Loveparade 2010 in Duisburg. Der Anaesthesist 62(12):1010–1019 Ozcan YA (2008) Health care benchmarking and performance evaluation, an assessment using data envelopment analysis (DEA). Springer, New York Ozcan YA (2009) Quantitative methods in health care management: techniques and applications. JosseyBass/Wiley, San Francisco ¨ sterreichisches Rotes Kreuz (2007) Vorschriften. Retrieved November 26, 2013, from http:// O vorschriften.roteskreuz.at Paseka A (2008) Wie Kinder zu Ma¨dchen und Buben werden. Einige Erkenntnisse aus der Sozialisationsund Geschlechterforschung. In Geschlecht lernen. Innsbruck: Studienverlag, pp 15–31 Peric N, Herdlicka T (2012) Das SanHiSt-Planspiel fu¨r den Aufbau einer Sanita¨tshilfsstelle bei Großschadenseinsa¨tzen—Eine experimentelle Analyse des simulationsbasierten Managementplanspiels zur Unterstu¨tzung des effektiven und effizienten Einsatzes von Rettungsdiensten. Vienna: Master Thesis, University of Vienna Pfaff MS, Drury JL, Klein GL, More LD, Moon SP, Liu Y (2010) Weighing decisions: aiding emergency response decision making via option awareness. Proceedings of the 2010 IEEE Technologies for Homeland Security Conference, Waltham, MA, pp 251–257 Pfaff MS, Klein GL, Drury JL, Moon SP, Liu Y, Entezari SO (2013) Supporting complex decision making through option awareness. J Cogn Eng Decis Mak 7(2):155–178 Pidd M (2004) Computer simulation in management science. Wiley, Chichester Pidgeon N, O’Leary M (2000) An-made disasters: why technology and organizations (sometimes) fail. Saf Sci 34(1):15–30 Pinedo M (2008) Scheduling: theory, algorithms, and systems. Springer, Newton Rabin R, de Charro F (2001) EQ-5D: a measure of health status from the EuroQol group. Annu Med 33(5):337–343 Rapoport A, Dale PS (1966) The ‘‘end’’ and ‘‘start’’ effects in iterated prisoner’s dilemma. J Confl Resolut 10(3):363–366 Rauner MS, Kraus M, Schwarz S (2008) Competition under different reimbursement systems: the concept of an internet-based hospital management game. Eur J Oper Res 185(3):948–963 Rauner MS, Schaffhauser-Linzatti M, Niessner H (2012) Resource planning for ambulance services in mass casualty incidents: a DES-based policy model. Health Care Manag Sci 15(3):254–269 Rempe A, Klo¨sters K (2006) Das Planspiel als Entscheidungstraining. W. Kohlhammer-Verlag, Stuttgart Rinaldi SM, Peerenboom JP, Kelly TK (2001) Identifying, understanding, and analyzing critical infrastructure interdependencies. Control Syst IEEE 21(6):11–25 Schulmeister R (1996) Grundlagen hypermedialer Lernsysteme. Theorie - Didaktik - Design. AddisonWesley, Bonn Sefrin P (2005) Die Rolle des Behandlungsplatzes bei Massenanfall von Verletzten. Der Notarzt 21(06):189–194 Smith SW, Portelli I, Narzisi G, Nelson LS, Menges F, Rekow ED, Mincer JS, Mishra B, Goldfrank LR (2009) A novel approach to multihazard modeling and simulation. Disaster Med Public Health Prep 3(2):75–87 Solecki W, Seto KC, Marcotullio PJ (2013) It’s time for an urbanization science. Environ: Sci Policy Sustain Dev 55(1):12–17
123
A policy management game for mass casualty incidents Solga H, Pfahl L (2009) Doing Gender Im Technisch-Naturwissen-Schaftlichen Bereich. Fo¨rderung des Nachwuchses In Technik und Naturwissenschaft. Springer, Berlin Heidelberg, pp 155–218 Streufert S, Satish U, Barach P (2001) Improving medical care: the use of simulation technology. Simul Gaming 32(2):164–174 Vissers J, Beech R (eds) (2005) Health operations management: patient flow logistics in health care (routledge health management). Routledge Chapman and Hall, New York XJ Technologies Company (2007) AnyLogic 6—user’s guide
Marion S. Rauner is Associate Professor, Department of Innovation and Technology Management, University of Vienna, Austria. She received an MBA in Business Informatics, a Ph.D. in Social and Economic Sciences, and her Habilitation in Business Administration, all from the University of Vienna, Austria, and an MBA in Business Administration from Vienna University of Economics and Business Administration, Austria. Professor Rauner’s research interests include international health care systems, disease policy modeling, operations research in public health, disaster management, and evaluation and management of healthcare technologies. Her refereed research has been published in a wide variety of journals, including Operations Research, European Journal of Operational Research, Journal of the Operational Research Society, Computers & Operations Research, OR Spectrum, Central European Journal of Operations Research, IMA Journal of Management Mathematics, Socio-Economic Planning Sciences, Health Policy, Health Care Management Science, International Journal of Healthcare Technology and Management, and International Journal of Technology Management. She has co-edited several special issues and serves in many editorial boards. Professor Rauner was awarded the Young, Talented Scientists Award of the Vienna Municipal Government in 2000, the Pharmig Prize for Health Economics in 2002, the Dr. Maria Schaumeyer Prize, as well as Kardinal-Innitzer-Prize, in 2003, and a best paper conference award in 2012. She has also supported the Austrian sick fund for occupational injuries in optimally allocating prevention budgets since 2001 and participates in the ftp7 EU-project S-Help (Securing Health.Emergency. Learning.Planning) from 2014-2016. She was President of the Austrian Operations Research Society from 2006-2012 for which she also coordinates the working group ‘‘Operations Research applied to Health Services.’’ Helmut Niessner received an MBA Business Administration (University of Vienna, Austria, 2010). His research focus is on simulations and management games for disaster situations in health care applied to projects in practice. He has teaching experience at different Universities and Institutions. He serves as a volunteer paramedic at Samaritan Vienna (specialized on organization and disaster management, depute emergency management system incident commander for five districts of Vienna, technical officer for the CBRNE special emergency group of Samaritan Vienna). Currently, he is in the UNIVIE-team of Associate Professor Marion Raunerin in the ftp7 EU-project S-Help (Securing Health.Emergency. Learning.Planning) from 2014-2016. Ulrike Leopold-Wildburger studied mathematics, philosophy, and logic at University of Graz (Austria) and University of Oslo, Norway. She was a scholar at the Institute for Advanced Studies, Vienna, Austria. She received an Master of Science and Ph.D. summa cum laude, University of Graz and an Habilitation/a Venia docendi for Operations Research and Mathematical Economics [Referees: Wolfgang EICHHORN, (Karlsruhe), Reinhard SELTEN, (Bonn, Nobel Laureate in Economics 1994), John HARSANYI; (Berkeley, Nobel Laureate 1994)]. During 1988/89, she was full Professor at the Department of Operations Research, University of Zu¨rich, CH and Guestprofessor at University of Minnesota at Minneapolis, Dept. of Economics with Lenonid HURWICZ (Nobel Laureate 2007) in 1991. Since 1992, she has been full professor at the University of Graz. From 2001 to 2003, she was Dean of Studies, Faculty of Economics and Social Sciences, University of Graz. Since 2010, she has been Head of the department of Statistics and OR, University of Graz. She serves as Editor-in-chief of the Central European Journal of Operations Research/Springer. She has received several Best Paper Awards and Research Awards and Science Awards. She is member of the European Academy of Sciences and High Council member at the European University Institute of Florence. Natasa Peric is Clinical Trial Coordinator in the pharmaceutical industry and researcher, Competence Center for Nonprofit Organizations and Social Entrepreneurship, Vienna University of Economics and
123
M. S. Rauner et al.
Business, Austria. She received her Master in International Business Administration, with focus on Health Care Management, from University of Vienna, Austria in 2012. Her research interests include social impact measurement and social return on investment analysis (SROI-analysis) in public health. Teresa Herdlicka works at the Chamber of Agriculture,Vienna, Austria. She focuses on green care. She received an MBA from University of Vienna, Austria, in 2012, where she specialized in Health Care Management.
123
