Mass Casualty Incident Commander Decision-Making Models: Novice vs. Expert Decision Making

Abstract

Introduction: Managing a mass-casualty incident (MCI) challenges commanders in the pre-hospital phase. Exploring the differences between the decision-making processes of novice and expert commanders can reveal how to improve training. This study aims to compare the decision-making processes between paramedic students and veteran paramedics. Methods: Data from eight MCI field simulations of paramedic students with no experience, and one simulation of a veteran paramedic were collected. Results: In the first phase of the simulation, both the students and the veteran paramedic followed the MCI protocol. From the second phase, the paramedic students mostly responded to received cues, while the veteran paramedic initiated actions without preliminary cues. Discussion: The veteran paramedic initiates actions and utilizes "considered" decision-making process earlier than the students, that frequently acted in response to cues. Although these results are based on a small sample, this study indicates a difference between novice and expert commanders’ decision-making processes.

Keywords

Mass-Casualty Incident Decision Making Dual Process Model Incident Commander

Introduction

A mass-casualty incident (MCI) is an emergency in which the number of casualties exceeds the available medical resources (Einav et al., 2004; Rimstad & Braut, 2015; Rimstad & Sollid, 2015). Managing the pre-hospital response of emergency medical services (EMS) is a challenge for MCI commanders and requires them to cope with a lack (Rimstad & Sollid, 2015) or excessive amount of information and limited resources (Perry, Jaffe, & Bitan, 2019). Thus, improving the MCI commander’s decision-making process is vital, and exploring the differences between novice and expert commanders’ decision-making processes can potentially improve novice commanders’ training.

The differences between novice commanders and experienced commanders were explored (Berndt & Herczeg, 2019) previously based on the Situational Awareness theory (Endsley, 1995, 2021). The main differences were in the ability of incident commanders to perceive changes in the MCI, comprehend their effects, and project their consequences. In contrast, preliminary leaders could only perceive changes. Perry, Jaffe, & Bitan (2022a) used quantitative data from MCI field simulations to explore the decision-making process of paramedic students at the end of their MCI management qualification harnessing the “dual-process” model (Kahneman, 2003; Norman, 2009). Because field simulations are known to improve communication skills (Gordon et al., 2016; Jorm et al., 2016), knowledge (Zinan, Puia, & Kinsley, 2015), clinical skills (Currie, Kourouche, Gordon, Jorm, & West, 2018), and performance (Bajow, Alassaf, & Cluntun, 2018), they provide the most realistic exposure to gain experience toward managing a real MCI (Gillett et al., 2008).

The "dual-process model" represents two decision-making processes known as System 1, which is an “instantaneous” process, and System 2, which is a "considered" process and it is widely used to model, analyze, and explore medical decision-making processes (Croskerry, 2009; Djulbegovic, Hozo, Beckstead, Tsalatsanis, & Pauker, 2012; Marcum, 2012; Tsalatsanis, Hozo, Kumar, & Djulbegovic, 2015). Perry, Jaffe & Bitan (2022a) found that the decision-making process of paramedic students changes over the MCI timeline. At first, commanders rely on the MCI protocol, which guides them during the chaos and uncertainty that characterizes the first moment after arrival to a MCI (i.e., System 1). As the event unfolds, a commander responds immediately (i.e., System 1) to new information and, toward the end, initiates actions without any new information (i.e., System 2).

The goal of this study was to explore the conditions under which veteran paramedic and paramedic students utilized System 1 and System 2 responses, and to elucidate the differences between experts and novice commanders’ decision-making processes.

Methods

As this study followed previous paramedic students’ decision-making while managing an MCI (Perry, Jaffe, & Bitan, 2022b), it utilized a data collection method dedicated for MCI simulations that was previously harnessed (Perry, Jaffe, & Bitan, 2022a).

Study Design

Two data sets of MCI simulations were used, one of paramedic students and a second of a veteran paramedic. The simulation was divided into interactions, defined as "information exchanges between two or more persons without an interruption”. Each interaction was first tagged as a cue or an action. A cue was defined as “input of verbal, visual, and/or auditory information”. An action was defined as “an order given by the commander” For example, an interaction initiated by the commander to upload a casualty to an ambulance was tagged as an action.

Two categories of cues were defined: cues with a following response, and cues without a response. This distinction was based on the assumption that a response indicates that the commander managed to complete a decision-making process (e.g., managed to find a solution to a certain problem). Following the definitions of cues, three action categories were defined: a response following a cue, an action without a cue, and an action following the MCI protocol. The three categories were differentiated according to the following assumptions. An action following the protocol was defined “as an action that was included in the EMS MCI protocol and was performed without any cue”. The other two categories of initiated action, following a cue and without a cue, referred to interactions whose content was not included in the MCI protocol. These two categories differed by the presence of the stimuli that triggered the action following a cue and by the stimuli that were missing from the action without a cue.

Normalization Method

To compare different simulation timelines, each timeline was normalized into time frames. Let $S$ be the set of $η$ interactions in all simulations, $s_{i}$ the subset of interactions occurring in simulation $i$ , and $x_{i j}$ the $j$ ’th interaction in the $i$ simulation. Each interaction is tagged as one of four different subsets: (1) cue with a following response – $σ$ , (2) action following cue – $ρ$ , (3) action without cue – $α$ , (4) action following protocol – $δ$ , and (5) cues without a response – $μ$ , such as $\cup_{x_{i j} \in {σ, ρ, α, δ, μ}} = S$ . Let $s t_{i j}$ be the starting time of interaction $x_{i j}$ , $e t_{i j}$ the ending time of the last interaction of simulation K ≥ 1, and the number of overall time frames. Interaction $x_{i j} \in k \geq 1$ will be tagged in time frame if $\frac{s t_{i j} * k}{e t_{i}} + 1 = k$ .

MCI Simulations

Data from eight MCI simulations of paramedic student were collected. These simulations were part of a paramedic MCI training course, in which most of the students were medics in the EMS, with 2 or 3 years of experience as military medics, but no experience as MCI commanders. The exercised scenario was a simulated missile attack on an urban area and included approximately 20–30 actors as casualties. The local EMS participated in the simulation and provided all the necessary resources (e.g., ambulances, drivers, medics, stretchers, and communications equipment). In addition, a dedicated dispatch was assembled in the EMS dispatch center to communicate with the MCI commander.

The data of the veteran paramedic is a part of the yearly operational EMS training. The exercised scenario was a terror incident at a live concert and included 40–50 actors as casualties and 100 actors as a crowd. In these events, a dedicated EMS team is deployed to provide medical and evacuation services in case of a medical event or an MCI. The setup included several medical units scattered at the concert area, a pre-made treatment and evacuation plan, and an experienced commander qualified to manage an MCI. The data of all simulations were tagged interdependently by two observers (O.P. and T.K.) using OBSERVER XT software. A third observer (Y.B.) resolved any disagreements.

Statistical Analysis

Based on the normalization method, cues, action following a cue, action without a cue, and action following protocol were divided into time frames and are reported by means and standard deviations (standard deviations were not calculated for the veteran paramedic since the data include a single simulation). Time frames were divided into the three MCI management phases: (1) evaluation and triage, (2) treatment and preparedness for evacuation, and (3) evacuation according to casualty priority. Matplotlib and Statistics Python packages were used for statistical analysis and graphing.

Results

The data for this study included two data-sets, the first of paramedic students, which was presented in Perry, Jaffe & Bitan (2022a) and included eight simulations with an $η_{N} =$ 1,068 interactions, and the second of an experienced MCI commander, which included one simulation with $η_{E} = 201$ interactions. The categorization of interactions into sub-categories is presented in Table 1. The data of each simulation were divided into ten timeframes (i.e., $K =$ 10). For example, interactions in the third timeframe were tagged in the 20%–30% timeframe.

Table 1.

Interaction Sub-Category Division.

Interaction Sub-Categories	Paramedic Students	Veteran Paramedic
Action Following Protocol - $δ$	67 (6.27%)	29 (14.43%)
Cues with a Following Response - $σ$	318 (29.78%)	36 (17.91%)
Action Following a Cue - $ρ$	386 (36.14%)	36 (17.91%)
Action Without a Cue - $α$	158 (14.79%)	97 (48.26%)
Cues without a Response - $μ$	139 (13.01%)	3 (1.49%)

Paramedic Students’ Decision-Making Analysis

Both the overall number of cues and the overall number of actions increased towards the treatment phase, reaching mean peaks of 4.98%±1.36% and 6.14%±2.04%, respectively, and decreased as the second phase unfolded. As the third phase began, overall actions and cues peaked at mean values of 4.73%±2.05% and 6.95%±2.32%, respectively. The actions following the protocol were at their peaks at initiation of the simulations, at 2.65%±1.33%, and decreased as the simulations proceeded. In contrast, actions without a cue increased as the simulations unfolded and peaked in the 80%–90% time frame, at 2.57%±1.60%. The number of responses following cue maintained similar trends as actions and cues, at 5.24%±1.79% in the 30%–40% time frame, and at 4.38%±3.09% in the 80%–90% time frame. Cues without a response increased toward the 40%–50% (1.39%±1.40%) time frame and maintained similar trends to the end of the simulations. Figure 2 represents the average percentage (black bars) across the different MCI phases.

Veteran Paramedic Decision-Making Analysis

Although this part of the data is based on a small sample, these preliminary results present different patterns. Both actions and cues show a similar trend over the simulation timeline. Both increased toward the end of the first phase peaking at the 40%–50% time frame with 10.52% and 3.15, respectively. From the 50%–60% time frame, both gradually increased until the end of the simulation. Actions without a cue maintain a similar trend as the actions, peaking at the end of the second phase with 6.31% and the 40%–50% time frame with 7.89%, and gradually increasing from the 50%–60% time frame until the end of the simulation. Responses to cues peaked at 40%–50% with 2.63% and maintained a similar trend until the end of the simulation. Actions following the protocol are mostly dominant in the first phase, reaching a peak at the 10%-20% time frame with 6.84%. Cues without responses maintain a similar trend over the simulation timeline, ranging from 0% to 0.52%. Figure 2 represents the average percentage (grey bars) across the different MCI phases.

Discussion

This study’s results reveal the difference between the decision-making process of paramedic students and veteran paramedics (Figure 1). Two main findings may shed light on these differences. First, for paramedic students, the triggers for most actions are cues received from team leaders and EMS team members. Thus, the commander utilized the cues to manage the MCI that fits System 1. For the veteran paramedic, most actions were initiated without a preliminary cue, indicating that fewer cues were required to manage the MCI, which fits System 2. In addition, the veteran paramedic was able to shift between MCI phases more quickly than the paramedic students.

Figure 1.

Paramedic Students’ (Top Graph) and Veteran Paramedic’s (Bottom Graph) Decision Making.

Figure 2.

Paramedic Students’ and Veteran Paramedic’s Average Percentage Actions Following Protocol, Action Following a Cue, Action Without a Cue, and Cues Without Response.

These findings align with previous results indicating that expert paramedics operate more strategically than novice paramedics (Smith et al., 2013). In terms of Situational Awareness (Endsley, 1995, 2021) and the Recognition Primed Decision-Making framework (G. Klein, 1999; G. A. Klein, Calderwood, & Macgregor, 1989), it can be assumed that the experience of the commander allows projection generation (i.e., Situational Awareness Level-3) or to integrate and follow only the required information on the time to shift between phases.

The second finding is that the veteran commander utilized the MCI protocol and then shifted to initiated actions versus the paramedic students who utilized the protocol, but then waited for cues to arrive before making a response. As mentioned above, it can be assumed that the expertise of the expert allowed him to recognize the situation easily and to recall what possible actions can be performed to manage the MCI. Similar findings were reported by Berndt & Herczeg (2019), in which experienced MCI commanders demonstrated better situational awareness than preliminary leaders with no experience.

An additional finding is the number of cues without a response. For paramedic students, there is a consistent trend of receiving cues without a response in contrast to the veteran paramedic, in which there were almost no cues without a response. If a cue did not yield a response, it might indicate that it was ignored or did not perceive. The ability to ignore a cue was observed as an advantage of experts to ignore cues that are not relevant to the problem at hand (G. Klein, Pliske, Crandall, & Woods, 2005). A cue that was not perceived could be due to working memory limitations - number of stored items (Wickens, Helton, Hollands, & Banbury, 2021) or items decay (Brown, 1958; Peterson & Peterson, 1959), lack of required encoded knowledge in the long-term memory (Barsalou, 2008, 2010), or high mental workload (Wickens, 2002, 2008).

Lastly, there is a difference in the quantities of the actions and cues. While the action and cue trends are similar, the quantities differ. The veteran paramedic simulation included more actions than the paramedic students’ simulations. This could be explained by the complexity of the MCI (i.e., the number of casualties in the MCI) (Busby & Witucki-Brown, 2011). Since there were more casualties, it is safe to assume that more information must be controlled, resulting in more procedures and actions that needed to be performed.

This study’s contribution in evaluating the differences between novice and expert commander decision-making can serve as a paradigm for researchers, trainees, and instructors to assess current decision-making skills and the required level of decision-making skills (i.e., how experts make decisions while managing an MCI). Instructors should utilize this study’s results as a benchmark for their trainees’ decision-making and to strive for expert decision-making at the end of a qualification or operational training.

Like any study, the analysis method used in this study has limitations. In our simulation, information exchanges were mostly verbal, aside from a few cases where interactions were tagged as responses to visual input (e.g., the commander pointed to specific casualties and asked the team to evacuate them). The assumption that there could have been cases where verbal interactions, combined with visual inputs, led to a response should be considered. Thus, the assumption that commanders acquire visual data should be considered and reviewed in post-simulation interviews. An additional limitation is the sample size of the veteran paramedic simulation, which due to logistics and administrative challenges included only one simulation. Lastly, the method made it difficult to differentiate between a cue that was not perceived or was ignored. Post-simulation interviews or questionnaires could shed light on the reason for cues without a response. Future studies should be based on more experienced MCI simulations to further explore the differences between novice and expert MCI commanders.

Footnotes

Acknowledgements

This study was conducted as part of SimReC - The research center for simulation in healthcare at Ben-Gurion University of the Negev. The contribution of Heli Bergman and Michal Hergas to the manuscript is acknowledged with gratitude.

References

Bajow

N. A.

Alassaf

W. I.

Cluntun

A. A.

(2018). Course in Prehospital Major Incidents Management for Health Care Providers in Saudi Arabia. Prehospital and Disaster Medicine, 1–9. https://doi.org/10.1017/S1049023X18000791

Barsalou

L. W.

(2008). Grounded cognition. Annual Review of Psychology, 59, 617–645. https://doi.org/10.1146/annurev.psych.59.103006.093639

Barsalou

L. W.

(2010). Grounded Cognition: Past, Present, and Future. Topics in Cognitive Science, 2(4), 716–724. https://doi.org/10.1111/j.1756-8765.2010.01115.x

Berndt

Herczeg

(2019). The role of mental models and situation awareness for computer system support in mass casualty incident management. ECCE 2019 - Proceedings of the 31st European Conference on Cognitive Ergonomics: “‘Design for Cognition’,” 220–227. https://doi.org/10.1145/3335082.3335086

Brown

(1958). Some tests of the decay theory of immediate memory. Quarterly Journal of Experimental Psychology, 10(1), 12–21.

Busby

Witucki-Brown

(2011). Theory development for situational awareness in multi-casualty incidents. Journal of Emergency Nursing, 37(5), 444–452. https://doi.org/10.1016/j.jen.2010.07.023

Croskerry

(2009). Clinical cognition and diagnostic error: Applications of a dual process model of reasoning. Advances in Health Sciences Education, 14(1 SUPPL), 27–35. https://doi.org/10.1007/s10459-009-9182-2

Currie

Kourouche

Gordon

Jorm

West

(2018). Mass casualty education for undergraduate nursing students in Australia. Nurse Education in Practice, 28(October 2017), 156–162. https://doi.org/10.1016/j.nepr.2017.10.006

Djulbegovic

Hozo

Beckstead

Tsalatsanis

Pauker

S. G.

(2012). Dual processing model of medical decision-making. BMC Medical Informatics and Decision Making, 12(1). https://doi.org/10.1186/1472-6947-12-94

10.

Einav

Feigenberg

Weissman

Zaichik

Caspi

Kotler

Freund

H. R.

(2004). Evacuation Priorities in Mass Casualty Terror-Related Events: Implications for Contingency Planning. Annals of Surgery, 239(3), 304–310. https://doi.org/10.1097/01.sla.0000114013.19114.57

11.

Endsley

M. R.

(1995). Toward a theory of situation awareness in dynamic systems. Human Error in Aviation, 37(1), 32–64. https://doi.org/10.4324/9781315092898-13

12.

Endsley

M. R.

(2021). A Systematic Review and Meta-Analysis of Direct Objective Measures of Situation Awareness: A Comparison of SAGAT and SPAM. Human Factors, 63(1), 124–150. https://doi.org/10.1177/0018720819875376

13.

Gillett

Peckler

Sinert

Onkst

Nabors

Issley

… Arquilla

(2008). Simulation in a disaster drill: Comparison of high-fidelity simulators versus trained actors. Academic Emergency Medicine, 15(11), 1144–1151. https://doi.org/10.1111/j.1553-2712.2008.00198.x

14.

Gordon

C. J.

Jorm

Shulruf

Weller

Currie

Lim

Osomanski

(2016). Development of a self-assessment teamwork tool for use by medical and nursing students. BMC Medical Education, 16(1), 1–7. https://doi.org/10.1186/s12909-016-0743-9

15.

Jorm

Roberts

Lim

Roper

Skinner

Robertson

… Osomanski

(2016). A large-scale mass casualty simulation to develop the non-technical skills medical students require for collaborative teamwork. BMC Medical Education, 16(1), 1–10. https://doi.org/10.1186/s12909-016-0588-2

16.

Kahneman

(2003). A Perspective on Judgment and Choice: Mapping Bounded Rationality. American Psychologist, 58(9), 697–720. https://doi.org/10.1037/0003-066X.58.9.697

17.

Klein

(1999). Sources of Power: How People Make Decisions. MIT Press Books, 1.

18.

Klein

G. A.

Calderwood

Macgregor

(1989). Critical decision method for eliciting knowledge. IEEE Transactions on Systems, Man, and Cybernetics, 19(3), 462–472.

19.

Klein

Pliske

Crandall

Woods

D. D.

(2005). Problem detection. Cognition, Technology and Work, 7(1), 14–28. https://doi.org/10.1007/s10111-004-0166-y

20.

Marcum

J. A.

(2012). An integrated model of clinical reasoning: Dual-process theory of cognition and metacognition. Journal of Evaluation in Clinical Practice, 18(5), 954–961. https://doi.org/10.1111/j.1365-2753.2012.01900.x

21.

Norman

(2009). Dual processing and diagnostic errors. Advances in Health Sciences Education, 14(1 SUPPL), 37–49. https://doi.org/10.1007/s10459-009-9179-x

22.

Perry

Jaffe

Bitan

(2019). Measuring the Amounts of Information During Mass Casualty Incidents. In Proceedings of the International Symposium on Human Factors and Ergonomics in Health Care (Vol. 8, pp. 284–285). SAGE Publications Sage CA: Los Angeles, CA.

23.

Perry

Jaffe

Bitan

(2022a). Dynamic Communication Quantification Model for Measuring Information Management During Mass-Casualty Incident Simulations. Human Factors, 64(1), 228–249. https://doi.org/10.1177/00187208211018880

24.

Perry

Jaffe

Bitan

(2022b). Mass Casualty Incident Commander Decision-Making Models. Proceedings of the Human Factors and Ergonomics Society Annual Meeting, 66(1), 427–428. https://doi.org/10.1177/1071181322661091

25.

Peterson

M. J.

(1959). Short-term retention of individual verbal items. Journal of Experimental Psychology, 58(3), 193–198. https://doi.org/10.1037/h0049234

26.

Rimstad

Braut

G. S.

(2015). Literature review on medical incident command. Prehospital and Disaster Medicine, 30(2), 205–215. https://doi.org/10.1017/S1049023X15000035

27.

Rimstad

Sollid

S. J. M.

(2015). A retrospective observational study of medical incident command and decision-making in the 2011 Oslo bombing. International Journal of Emergency Medicine, 8(1), 1–10. https://doi.org/10.1186/s12245-015-0052-9

28.

Smith

M. W.

Bentley

M. A.

Fernandez

A. R.

Gibson

Schweikhart

S. B.

Woods

D. D.

(2013). Performance of experienced versus less experienced paramedics in managing challenging scenarios: A cognitive task analysis study. Annals of Emergency Medicine, 62(4), 367–379. https://doi.org/10.1016/j.annemergmed.2013.04.026

29.

Tsalatsanis

Hozo

Kumar

Djulbegovic

(2015). Dual processing model for medical decision-making: An extension to diagnostic testing. PLoS ONE, 10(8), 1–16. https://doi.org/10.1371/journal.pone.0134800

30.

Wickens

C. D.

(2002). Multiple resources and performance prediction. Theoretical Issues in Ergonomics Science, 3(2), 159–177. https://doi.org/10.1080/14639220210123806

31.

Wickens

C. D.

(2008). Multiple resources and mental workload. Human Factors, 50(3), 449–455. https://doi.org/10.1518/001872008X288394

32.

Wickens

C. D.

Helton

W. S.

Hollands

J. G.

Banbury

(2021). Engineering psychology and human performance. Routledge.

33.

Zinan

Puia

Kinsley

(2015). Results of a mass casualty incident simulation in an undergraduate nursing program. Journal of Nursing Education and Practice, 5(12), 71–78. https://doi.org/10.5430/jnep.v5n12p71