Improving Traffic Incident Management Using Team Cognitive Work Analysis

Abstract

Effective traffic incident management requires separate responder agencies, with different and sometimes competing priorities and purposes, to come together as a team. Their priorities include optimizing casualty outcomes, minimizing the disruption to the flow of traffic, and maintaining responder team safety. In this study, team Cognitive Work Analysis was used in a desktop exercise setting to analyze a complex traffic incident management exercise. The study investigated decisions made at the scene of an incident to determine system issues and system support solutions. Participants were all senior officers and decision makers in traffic incident management environments. Results indicated that team Cognitive Work Analysis was highly beneficial in determining gaps in team coordination, communication, and structures. Information regarding shared and not shared work elements between agencies highlighted novel coordination and education requirements within and between agencies, such as disparate priorities at the scene creating the risk of interoperability issues. Analyses of operational, coordination, and structural strategies offered new insights into the traffic incident management work domain and recommendations for improvements to the safety and performance of the overall traffic incident management system.

Keywords

cognitive work analysis critical decision method teamwork traffic incident management

Introduction

Traffic incident management involves the coordinated response from emergency services, traffic agencies, and local government agencies to remove incidents and restore traffic capacity safely and efficiently (Charles, 2007b; Farradyne, 2000). Aside from improved road safety, effective traffic incident management reduces congestion costs, improves the reliability of all forms of transport, and reduces vehicle emissions. As an example, congestion costs in the United States in 2005 were estimated at 78.2 billion dollars, with 52% to 58% of congestion attributed to traffic incidents (Carson, 2010).

However, traffic incident management involves inherent risks and the eliminating these risks is difficult due to the complexity of traffic incident environments. The traffic incident management work environment is dynamic, and the characteristics at each incident vary. Examples of issues related to incident specifics include the resources available at the scene (including availability of equipment like lighting, variable message signs, vests, traffic cones and so on, but also including personnel—quantity and levels of expertise/specialization for requirements at the scene), communication at the scene (intra- and interagency), the physical characteristics of the scene (scene size, topography, weather conditions, time of day), and incident specifics (stability of vehicles, presence of hazardous materials [HAZMATs], number, and types of casualties; Charles, 2007).

Although the incident specifics can seem unique and random, experts report similarities across incidents and recognition of incident types (Klein, 1998). Therefore, despite the complexity of traffic incident work environments, one way to determine improvement opportunities for traffic incident management is through an operator-centered focus. In a study by Cattermole, Horberry, Wallis, and Cloete (2014), emergency responders from police, fire, and traffic control agencies were asked to rate and describe their experiences of the greatest safety issues at traffic incident scenes. The highest rated issues related to interoperability between the emergency responder agencies. This finding is supported by Fiore and Salas (2004), who cited team coordination as a major issue for teams by researchers and a major aim of work in this area is to reduce the “process loss” of poor team coordination.

Teamwork involves the adaption of coordination strategies through closed-loop communication and a sense of collective orientation (Salas & Fiore, 2004). A good team in a temporally challenged, high stakes and dynamic environment such as in the traffic incident management environment requires a shared awareness of team goals, congruence between individual and team goals, and good coordination between team members conducting their separate tasks as part of the whole output (Charles, 2007). Training within agencies is strict, and professional development ongoing, so intra-agency teams display the characteristics described (Cattermole, Horberry, & Hassall, 2016). However, aside from occasional joint exercises, there is no training to better understand interagency roles and responsibilities at incident scenes. Also, due to the distributed and hectic nature of traffic incident management systems, team awareness is reduced, and team awareness is an important factor for successful collaboration (Gutwin & Greenberg, 2002).

The issue of interoperability at the scene of a traffic incident extends wider than the operational context. The traffic incident management environment can be thought of as a single system (Cattermole-Terzic, 2017). However, it is supported by policies and directives from separate agencies, departments, and industry, each developed with a focus on one aspect or group of the incident management system rather than the system as a whole. It is likely that some policies and practices will not be compatible. These incompatibilities may not have an immediate visible effect but may contribute to the potential for secondary incidents. Secondary incidents are incidents that occur after the initial incident and as a direct result of the changed conditions caused by the initial incident. The impact on secondary incidents includes loss of life, serious injury, and community, economic, and environmental costs. Secondary incidents also have a direct impact on emergency responder safety. For example, in the United States, an average of one police officer per month is killed in a roadside crash (Fischer, Krzmarzick, Menon, & Shankwitz, 2012).

The unique nature of each incident scene means that although training of standard operating procedures plays an important role in effecting optimal safety and output at traffic incident scenes, a large part of the work for teams at incidents requires problem solving, building knowledge, dynamic risk assessments, and flexibility. Agency training to build these skills at an intra- and interagency level may be one potential tool required to improve interoperability at incidents.

Frameworks from human factors have previously been used to map complex environments to improve safety and interoperability (e.g., Cooke, Gorman, Myers, & Duran, 2013; Flavell & Wellman, 1975; Kenny & La Voie, 1984; Klein & Wright, 2016; Langan-Fox, Code, & Langfield-Smith, 2000; McNeese, Rentsch, & Perusich, 2000; Rasmussen, 1983; Rentsch, Mello, & Delise, 2010; Scherer & Petrick, 2001; Shiffrin & Schneider, 1977). One such framework is Cognitive Work Analysis.

Cognitive Work Analysis

Cognitive Work Analysis is a framework that was originally developed by Jens Rasmussen (Rasmussen, 1983; Rasmussen, Pejtersen, & Schmidt, 1990). It is a theoretical framework that analyzes how people work in complex environments, with the aim of providing recommendations to improve system design (Vicente, 1999). Whereas other work analyses are descriptive (describing how work is currently done) or normative (describing how work should be done), Cognitive Work Analysis is a formative model (describing how work can be done; Naikar & Elix, 2016). It does this by identifying system controls and constraints over five phases of analysis. Each of the phases of Cognitive Work Analysis focuses on different types of constraints. The initial phases focus on ecological elements and there is a gradual shift to more cognitive issues.

Ashoori and Burns (2013) modified the traditional five-phase approach of Cognitive Work Analysis into two sets of four. Work Domain Analysis, Control Task Analysis, Strategies Analysis, and Worker Competencies Analysis were paired with a parallel set of social or team models called team Work Domain Analysis, team Control Task Analysis, team Strategies Analysis, and team Worker Competencies Analysis. Using this modification, Ashoori and Burns successfully mapped teamwork and shared tasks, strategies to accomplish tasks, and the required qualifications of operators in effective medical teams. Therefore, aside from the ability of Cognitive Work Analysis to analyze complex systems where unanticipated events occur, the modification of the framework to focus on team interactions makes it ideal for analyzing traffic incident management. This study will adopt the Ashoori and Burns’s (2013) adaptation, given its explicit focus on teamwork.

The four phases of team cognitive work analysis

A Work Domain Analysis provides a description of the constraints governing the functions and purpose of a particular work environment (Vicente, 1999). Team Work Domain Analysis investigates which team members have shared processes, components, and objectives within a work environment and also which elements only influence individuals. The goal of team Work Domain Analysis is to create a set of models that describe shared values, purpose, and priorities in the work environment.

Team Control Task Analysis investigates team activity and collaboration. One technique used in this process links the decision ladders of individual team members to determine team collaboration points in “decision wheels.” Because this section of team Cognitive Work Analysis investigates how decisions are made, by whom, and when, it is particularly useful in determining team decision support requirements.

Team Strategies Analysis looks at how teams coordinate, form, and regroup to handle different tasks. Analysis of traffic incident management using team Strategies Analysis could yield valuable information for scenario testing. The work environment in traffic incident management is dynamic, but establishing lines of communication and determining team make up and procedures for different possibilities at incidents could improve synchronous collaboration and the overall effectiveness of traffic incident management.

Team Competency Analysis aims to determine a series of desirable competencies operators must possess to effectively work in a team. Team Competency Analysis extends the analysis of traditional Cognitive Work Analysis and includes a study of social competency, investigating the interpersonal skills required for effective teamwork. The analysis requires the identification of skills/rules/knowledge-based requirements in the work situations.

This study will use three of the four phases of team Cognitive Work Analysis—team Work Domain Analysis, team Control Task Analysis, and team Strategies Analysis—to investigate the collaboration requirements between responder agencies at traffic incidents. The framework was chosen primarily due to its focus on constraints and possibilities for behavior rather than describing how activities actually occur. The dynamic complexity of the traffic incident management environment would be impossible to encapsulate using descriptive or normative models.

In this Australian study, decision makers from Queensland Police Service, Queensland Fire and Emergency Services, and Royal Automotive Club Queensland’s Traffic Response Unit participated in a complex desktop exercise. The results were analyzed using modified team Cognitive Work Analysis tools. The aim of this study was to establish collaboration points and to develop recommendations to improve intra- and interagency coordination, collaboration, and interoperability at traffic incident scenes.

Method

Participants

Participants for this study were required to be decision makers at traffic incidents and therefore needed to be senior officers. The researcher contacted the agencies to request for appropriate representation. The group Critical Decision Method (Klein, Calderwood, & Macgregor, 1989) for the desktop exercise included five participants—a Senior Traffic Response Officer, the Officer in Charge of the Forensic Crash Unit for the Queensland Police Service, two Queensland Fire and Emergency Services inspectors, and one Queensland Fire and Emergency Services Assistant Commissioner. All participants were male. The level of experience of the officers ranged from 15 to 32 years with an average experience of just over 25 years (M = 25.8 years, SD = 5.9) in the area of traffic incident management/emergency response.

Desktop Exercise and Photograph of the Incident Scene

The incident scenario was based on a major Australian incident on the Sydney F3 Sydney to Newcastle Freeway on April 12, 2010, when a 16-ton flatbed truck collided with the rear of a fully laden fuel carrier. The road was closed for a significant period, stranding motorists in some cases for more than 8 hr, resulting in significant media and political attention. For the exercise, the scenario was shifted to a South East Queensland location on the Pacific Motorway, just before the Logan River Bridge (Figure 1).

Figure 1.

Map of the incident area provided to participants at the exercise.

Procedure

The Critical Decision Method interview procedure was altered to suit a group environment to conduct a group desktop exercise. Participants were sent an incident scenario and a map of the incident location 2 days prior to the desktop exercise. They were asked to consider their agency’s response to the given incident.

On the day, participants sat in a closed meeting room with a whiteboard at the University of Queensland. The desktop exercise was conducted using one interviewer. The interviewer had considerable experience working with the interviewing technique and also working in transport, police, and traffic incident environments. All participants gave informed consent to be part of the desktop exercise and to be audio-recorded throughout the process. The audio recording was taken to assist with later analyses. The desktop exercise process conformed to preapproved University of Queensland ethics procedures. In total, the interview process for the desktop exercise took 3.5 hr.

Interview Process

A modification of the classic Critical Decision Method approach was used, applying four “sweeps” of the incident. In Critical Decision Method, sweeps are described as follows:

Sweep 1: Incident familiarization: Participants were sent a scenario of an incident and asked to consider their agency response prior to attending the desktop exercise. At the exercise, the facilitator requested that the officers remain in the mind-set of their own agency to prevent “group think” throughout the process.

At the beginning, the facilitator read through the incident details from beginning to end and then prompted each agency to begin with a description of what would happen from their agency’s point of view, from beginning to end of the incident. Participants were encouraged to interject when they considered their agency would take over functions or collaborate. It was also the facilitator’s role to keep the dialogue flowing and, as much as possible, in order of likely events/tasks.

Once the group finished, the interviewer retold the incident to the participants and wrote the details onto a whiteboard. The participants were encouraged to clarify and provide additional information where required so that a more complete description of the likely occurrences at the incident were represented on the whiteboard. At the end of this sweep, the interviewer and participants had a “shared view” of the incident. For this study, a shared view was considered to be reached when the interviewer retold the story according to what was written on the whiteboard and the participants agreed that all the details of likely occurrences of the incident were accurately described. This was considered to be the likely workflow for the scenario. At this point, the interviewer took photographs of the whiteboard to assist in the analysis stage of the study.

Sweep 2: Timeline verification and decision point identification: The interviewer again retold the incident, encouraging the participants to organize the incident around a likely timeline. The interviewer then identified points where decisions would be made, and actions taken. These “decision points” formed the focus of the final two sweeps. Decisions were defined as points where the participant was required to act when more than one option was available to act upon. Decision points were identified by the interviewer. The interviewer clarified decision points with participants by confirming that the action was taken and then further understood through probe questions identifying that other options were available (e.g., questioning the participant about what a novice would have done in the situation, or what he would have done if some situational cues were altered).

It should be noted that all participants in the study considered the decision points as merely actions. Further probing and questions were needed at this point to determine information about how the participants determined what potential actions to take.

Sweep 3: Deep probes: Probe questions were used to focus the participants on particular aspects of the cognitive processes and context behind the hypothetical decisions made by the experts at the incident. The questions included cue usage, prior knowledge, goals, expectations, and options.

Sweep 4: Hypotheticals—What if . . .? In the final sweep, the participants were asked to shift their perspective to alternative views and outcomes. What if you were a novice in this situation? What if some particular aspect of the incident scene was different? This section examined the possibilities and consequences of other options and errors, and also extended the interviewer’s understanding of the activation points for expert decisions.

Postinterview Activities

The desktop exercise was transcribed prior to analysis. To gain a richer understanding of the traffic incident management environment, the interviewer went on shift with participants from the Traffic Response Unit and Queensland Fire and Emergency Services. The data from the desktop exercise were then analyzed using three of the phases of team Cognitive Work Analysis and later validated by two senior emergency responders—one of the officers was a participant at the session, and one was external to the process.

Results

Hypothesized Workflow for Agencies at the Incident

In Sweep 1 of the desktop exercise, participants identified the likely workflow for the scenario, depicted in Figure 2. In the workflow diagram, different roles and responsibilities are shown along the vertical axis, and the horizontal axis shows the work progress over time. From this point, it is possible to analyze team structure, inter-team and intra-team interactions, shared work domain elements, and expertise using team Cognitive Work Analysis tools.

Figure 2.

Workflow diagram of the hypothetical traffic incident.

Team Work Domain Analysis

Team Work Domain Analysis investigates which team members have shared processes, components, and objectives within a work environment and also which elements only influence individuals. The first step in team Work Domain Analysis is to construct a regular Work Domain Model. A Work Domain Model is a diagram depicting the entire work domain, using five levels of abstraction for analysis. The top three levels consider the overall purpose of the work domain and what it is required to do. The bottom two levels consider capability and resource components. The five levels of abstraction are functional purpose, abstract function, generalized function, physical function, and physical form.

The different levels of the Work Domain Model are connected through a means-ends relationship. As an example, in the diagram below, at the physical form level, computer screens and traffic management center screens are required to visualize the CCTV footage (physical function level). The CCTV footage is required for the Brisbane Metropolitan Traffic Management Centre (BMTMC) to establish a visual assessment of the reported crash (generalized function). The visual assessment enables the BMTMC and traffic response unit to fulfill its priority of establishing an accurate incident assessment (abstract function), and the accurate assessment enables the traffic response to fulfill its higher level purpose for all personnel to arrive at the scene safely and efficiently. Figure 3 depicts the work domain model for the traffic incident scenario being analyzed.

Figure 3.

Work domain model for the incident.

At the functional purpose level, the abstraction hierarchy corresponds to work domain purposes. Four overall purposes were identified for traffic incident management at this incident at the functional purpose level: arrive at scene safely and efficiently, optimize casualty survival, investigate incident to provide accurate information about causal factors, and promote continued operation of the road transport system.

The abstract function level relates to the values, priorities, and principles of the teams at the incident. Eight processes were identified at this level: accurate incident assessment; timely response by appropriate groups; safe, effective, and quick accident assessment; appropriate resources for timely rescue; safe and appropriate diversions minimizing congestion; specialist teams arrive at appropriate times; effective investigations; quick clearance and road operational. The links between the abstract function level and the functional purpose level establish the why–how relation between the purposes of the work and the values and priorities.

Ten processes describe the generalized function level of the work domain model. These are visual assessment, required information obtained, required crews at the site, consulting, incident command post managing operation, rescue conducted, traffic diverted, specialist teams operating, investigations conducted, and road cleared and assessed. This level identifies the main work processes at the incident and the links between generalized function level and the abstract function level identify the work domain processes that meet the values and priorities of the organizations at the scene.

The physical function level identifies the physical work domain resources. The eight resources identified were CCTV, witnesses, communication teams, policy teams, general responders, specialist teams, casualties, and motorists. Links between this level and the generalized function level identify the work domain resources.

At the physical form level, the physical characteristics of work domain resources are listed. The eight identified areas were computer screens and Traffic Management Center screens; incident records (kept by each agency); procedural handbook (for each agency); policies, regulations, and legislation (for each agency and overarching); responder placement (where, when, availability); responder equipment (time to use, type, location, availability); investigation records (Scenes of Crime Unit, Forensic Crash Unit, coroner’s report, Environmental Protection Agency report, road assessment); and other equipment (for groups such as environmental teams, local government, road assessors, tow truck operators). The links between physical form and physical function levels identify the attributes of work domain resources.

The regular Work Domain Analysis identifies work domain purpose, values, work processes, and the physical elements of the work domain. However, it does not identify shared values or work processes. To better understand the responsibilities for each agency, a responsibility map was developed (Figure 4).

Figure 4.

Responsibility map for the incident.

A significant level of shared responsibility at the intra- and interagency levels is evident at the incident.

To better identify shared/not shared elements across the work domain, the responsibility map was separated according to the four functional purposes identified for the incident. These were “arrive at the scene safely and efficiently,” “optimize casualty survival,” “investigate incident to provide accurate information about causal factors,” and “promote continued operation of the transport system.” As an example, the responsibility map for “optimize casualty survival” is described below.

Queensland Fire and Emergency Services and Queensland Ambulance Service are directly responsible for casualty outcomes at the scene (Figure 5). Differences in priorities for this purpose relate to the need for specialist teams from Queensland Fire and Emergency Services to be at the scene due to the nature of this particular incident. Conflicts can occur due to differing priorities at this incident, indicating a point at which Queensland Fire and Emergency Services and Queensland Ambulance Service need to establish shared understanding to prevent conflicts or counterproductive actions. The specialist team operation at the generalized function level is a clear area of Queensland Fire and Emergency Services activity that does not require coordination with Queensland Ambulance Service. All boundary objects at the physical form level are shared between agencies indicating a requirement to ensure they are compatible with the needs and purposes of both agencies.

Figure 5.

Responsibility map for “optimize casualty survival.”

To understand and analyze shared and individual purpose, values, processes, boundary objects, team structures and interactions in intra- and interagency context, the next step for team Work Domain Analysis is to build collaboration and abstraction tables.

Collaboration tables

The collaboration tables are divided into four levels—functional purpose, abstract function, generalized function, and physical function. As an example, this paper will outline the collaboration table for the functional purpose level. At this level, the Collaboration Table depicts roles and responsibilities that contribute to the work domain purpose. They are also useful to identify what collaboration should occur at scenarios versus the actuality of collaboration and coordination at incident scenes.

In Table 1, all communication teams and operational teams from the different agencies share the functional purpose of ensuring that the appropriate operational teams arrive at the scene efficiently and safely. This suggests that coordination among the agencies should be apparent for activities that occur for this purpose. The other agency purposes at the scene are divided. Optimizing casualty outcomes is the shared purpose of Queensland Fire and Emergency Services and the Queensland Ambulance Service. Investigating the scene is the responsibility of the Queensland Police Service, the Environmental Protection Agency, and the Coroner. Ensuring that the road system is operational as quickly as possible is the responsibility of the Traffic Response Unit, tow truck operators, the Queensland Police Service, and the Department of Transport and Main Roads. This disparity of purpose in a safety critical environment sets the scene for possible interoperability issues. In the first row of Table 1, QPS refers to the Queensland Police Service, QFES is the Queensland Fire and Emergency Services, QAS is the Queensland Ambulance Service, BMTMC is the Brisbane Metropolitan Traffic Management Centre, TRU is the Traffic Response Unit, and DTMR is the Department of Transport and Main Roads.

Table 1

Traffic Incident Management Collaboration Table at the Functional Purpose Level

Functional Purpose	Communication Team QPS	Communication Team QFES	Communication Team QAS	BMTMC	QPS	QFES	QAS	TRU	Specialist Teams QFES	Specialist Teams QPS	Tow Trucks	DTMR/Road Assessors
Arrive at scene safely and efficiently	X	X	X	X	X	X	X	X	X	X	X
Optimize casualty survival						X	X		X
Investigate incident to provide accurate information about causal factors					X					X
Promote continued operation of the road system					X				X		X	X

Team Work Domain Analysis Findings

The purpose of this section of the analysis was to identify shared/not shared purposes, values, priorities, and principles. Ashoori, Burns, d’Entremont, and Momtahan (2014) found that team Work Domain Analysis provided information about shared elements of the information space in surgical teams. In support of the findings of Ashoori et al. (2014), this study also found that team Work Domain Analysis identified shared and not shared elements across the emergency responder agencies, therefore identifying likely areas of weakness for interoperability.

More specifically, for the functional purpose of arriving at the scene safely and efficiently, all agencies have this purpose. At the abstract function level, however, whereas Queensland Fire and Emergency Services and the Queensland Ambulance Service prioritize optimal casualty outcomes at the scene, the Queensland Police Service prioritizes incident investigation and traffic management, and the Traffic Response Unit prioritizes traffic management, optimizing traffic flow through the area and road clearance and assessment. The different agency priorities in a high-stakes, complex and dynamic environment become a risk point for interoperability unless the responders from each agency have a shared awareness of the overall priorities of the Traffic Incident Management work domain, and the role each agency plays within it. However, in a study by Cattermole et al. (2014) investigating decision making and perspectives of emergency responders at traffic incidents, a shared awareness was not evident. For example, Queensland Fire and Emergency Services participants indicated that Queensland Police Service officers arrive too late to manage incident scenes, putting emergency responders from Queensland Ambulance Service and Queensland Fire and Emergency Services at greater risk and reducing the effectiveness of the incident response. Many did not seem aware that this is due to a different level of priority given to attending road crashes between the agencies.

Optimizing casualty outcomes at traffic incidents is a focus for Queensland Fire and Emergency Services and Queensland Ambulance Service only. This indicates a need to ensure tight coordination between these agencies for this function, but little requirement for Queensland Police Service or Traffic Response Unit to understand the processes involved.

Incident investigation to determine causal factors at the incident is the purpose of Queensland Police Service and other specialty teams at the scene. Although it is not necessary for other teams to understand the investigation process, in the exercise, the impact of these teams taking several hours to attend the incident became evident. Although it is important for the site to remain untouched for criminal investigations, if chemicals are on the road for more than 3 hr, it is likely that the bitumen will need to be removed and therefore the road closed for several days. Although this aligns with the priorities of the investigation teams, it is in direct opposition to the priorities of the Traffic Response Unit and the Department of Transport and Main Roads.

Promoting continued operation of the road transport system is a functional purpose for the Queensland Police Service, Traffic Response Unit, Department of Transport, and tow truck operators. Aspects of this function require specialist activities for each agency individually, but joint functions should be coordinated so that policies, processes, and practices align and the agencies have a shared understanding of their joint roles in the function.

Team Control Task Analysis

The team Control Task Analysis tools used to examine traffic incident management team structures, interactions, shared workflows, and boundary objects were decision wheels and the Contextual Activity Template for teams.

Decision wheels

Ashoori and Burns (2013) linked the decision ladders of individual team members to determine team collaboration points in “decision wheels.” The maps become quite complex, so links are numbered for simplification. Using this technique, it is possible to create a decision wheel table and to determine whether the collaboration points were “synchronous” or “asynchronous.” A synchronous collaboration point is one that is observed to occur efficiently and effectively. An asynchronous collaboration is one that is observed to deviate from optimal effectiveness, efficiency, or safety. The decision wheels for the traffic incident management incident were complex and focused around Queensland Fire and Emergency Services due to the HAZMAT issues at the scene. Figure 6 depicts decision wheels for the incident. Each wheel represents an organization and each piece of “pie” within the wheel represents an actor/area/crew from the organization. The arrow pointing to a piece of pie represents the beginning point of a decision. For example, a line from witness call-ins goes to each of the agency communication teams, with the arrow pointing to the “activation” circle. Due to the hypothetical nature of the exercise, the flow through the decision ladder template cannot be depicted; however, the exit point gives information about when the communication teams make decisions about communicating with other units, and who they contact. The general complexity of traffic incident management is evident from the diagram also.

Figure 6.

Decision wheel of traffic incident management incident.

The numbers near each line identify the sequence of decisions and communications at the intra- and interagency level. Colors of the arrows and numbers identify agencies—red for Queensland Fire and Emergency Services, blue for Queensland Police Service, purple for the Queensland Ambulance Service, and green for the Traffic Response Unit, BMTMC, and Department of Transport and Main Roads. Other groups represented in the diagram are the Environmental Protection Agency and tow truck operators.

The connections within and across decision wheels give information about communication and coordination requirements and the types of decisions and tasks required at each of the decision points. Due to the hypothetical nature of the scenario, the reported decision processes from the participants were primarily at the outer sections of the decision wheel. If this was actually the case, a focus on coordination would be less important. Decision processes that require knowledge-based decisions are more likely to require consideration of the impacts of other agencies and advice from agencies. As an example, in a previous study by Cattermole et al. (2016), a Queensland Fire and Emergency Services Officer required input from paramedics to determine the appropriate rescue operation. A series of changing circumstances communicated by paramedics altered the Station Officer’s decisions about rescue operation requirements. It is highly likely that a complex incident like in this exercise would also contain several points where issues required the decision-making process to go to the higher levels of the decision ladder. Analysis of those points at the incident would identify crucial points for interagency communication and coordination.

Table 2 is an excerpt of a decision wheel table for the incident, using examples from Queensland Fire and Emergency Response to showcase the table’s function. In the table, each of the decision points from each agency is listed by color-coded number. The table identifies the team making the decision, the tasks or information they were distributing, and the boundary object for the decision. In a real scenario, the table would also identify whether the decision was synchronous or asynchronous; however, in hypothetical situations, all decisions are synchronous, so the section of the table was obsolete for this exercise. As an example, in the table, the first decision for Queensland Fire and Emergency Services comes from their communication team. They deployed crews to the scene. The boundary objects they used to gain information to make the decision were witnesses, and the decision point was synchronous because, at least in this hypothetical scenario, the crews were successfully contacted, available, and acted on the information to go to the scene as requested. The distribution of the decision point was at an intra-agency level for Queensland Fire and Emergency Services.

Table 2

Decision Wheel Table Excerpt for the Incident—Queensland Fire and Emergency Services

	Teams	Task	Boundary Object	Distribution
1	Firecom–crews	Send to scene	Witnesses	Queensland Fire and Emergency Services
2	Firecom–QAScom	Inform/request at scene	Witnesses	Queensland Fire and Emergency Services-Queensland Ambulance Service
3	Crews–Firecom	Visual assessment/updates	Visual assessment	Queensland Fire and Emergency Services
4	Firecom–Scientific	Send to scene	Visual assessment	Queensland Fire and Emergency Services
5	Firecom–Road crash rescue	Send to scene	Visual assessment	Queensland Fire and Emergency Services
6	Firecom–Environmental Protection Agency	Send to scene	Visual assessment	Queensland Fire and Emergency Services
7	Crews–Queensland Police Service	Advise re HAZMAT	Visual assessment	Queensland Fire and Emergency Services-Queensland Police Service
8	Crews–Queensland Ambulance Service	Advise re HAZMAT	Visual assessment	Queensland Fire and Emergency Services-Queensland Ambulance Service
9	Crews–Traffic Response Unit	Advise re HAZMAT	Visual assessment	Queensland Fire and Emergency Services-Traffic Response Unit
10	Scientific–Firecom	Advise re HAZMAT	Results of testing	Queensland Fire and Emergency Services
11	Scientific–Crews	Advise re HAZMAT	Results of testing	Queensland Fire and Emergency Services
12	RCR–Firecom	Inform re rescue	Casualty	Queensland Fire and Emergency Services

Note. HAZMAT = hazardous material; RCR = road crash rescue.

This type of analysis would be beneficial to establish teams requiring coordination, linked with the process, tasks, and resources required. Establishing optimal tables would enable comparisons against actual events at incidents. This would prove informative for postincident investigation, as well as for resource allocation and training teams in responder organizations.

Team contextual activity template

The contextual activity template is a representation to show how teams are involved in multiple functions over the totality of the incident. In Table 3, the Contextual Activity Template (adapted from Naikar, Moylan, & Pearce, 2006) for the incident is represented by four situations—evaluation of the incident, rescue and casualty treatment, incident management and investigation, and road clearance and assessment. The different functions required to complete these areas of work at the scene are the initial assessment, emergency response, visual assessment, incident assessment, consultation, road crash rescue, casualty treatment, incident management, road clearance operation, and road assessment. In Table 3, functions are listed in the left column. The occurrence of each function is represented by the box. For example, the initial assessment only occurs during the evaluation phase of the incident; however, incident assessment continues beyond the evaluation phase and across rescue and treatment and incident management and investigation phases.

Table 3

Contextual Activity Template of the Incident

SituationFunction	Evaluation	Rescue and Treatment	Incident Management and Investigation	Road Clearance and Road Assessment
Initial assessment
Emergency response
Visual assessment
Incident assessment
Consult
Road crash rescue
Casualty treatment
Incident management
Investigation
Road clearance operation
Road assessment

Source. Adapted from Naikar, Moylan, and Pearce (2006).

Ashoori and Burns (2013) extended the Contextual Activity Template to identify team requirements at the scene. In team Contextual Activity Template, team workflow is mapped to represent the distribution of work problems to the different teams. Roles and responsibilities of teams are in the rows and the work problems allocated to teams are illustrated with circles. Communication and coordination associated with reallocation of work problems are shown with arrows. In Figure 7, the work functions of the agency communication teams, general responders, and specialist teams are identified. Due to the complexity of the workflow, the work functions have been coded:

IA = Initial assessment

ER = Emergency response

VA = Visual assessment

In E = Incident evaluation

C = Consulting

ER2 = Second emergency response (representing the sending of specialist teams)

CT = Casualty treatment

RCR = Road crash rescue

IM = Incident management

Inv = Investigation

RCl = Road clearance

RA = Road assessment

Figure 7.

Modified contextual activity template representing the distribution of work functions at the incident.

As shown by the arrow directions, the horizontal axis in Figure 7 does not specifically represent time.

Team Control Task Analysis Findings

In this study, team Control Task Analysis identified areas that shared intra- and interagency elements across the work domain.

The Decision Wheels and accompanying Decision Wheel table clearly outlined the complexity of the traffic incident scene. The hypothetical incident could not be accurately mapped by the wheels, but the exercise identified an optimal scenario between agencies. The use of this tool to map actual incidents, especially when issues occur, is evident.

The basic Contextual Activity Template of the incident identified “consulting” as the only activity required to be conducted across all stages of the incident. The importance of consulting within and between agencies at the scene supports the findings of the current research. In previous research conducted by the authors (Cattermole, Horberry, & Hassall, 2016; Cattermole et al., 2015), a major focus of participants when asked about inefficiencies and safety concerns was the lack of shared understanding and coordination between agencies.

The modified Contextual Activity Template enabled work functions to be grouped according to activities. In accordance with the Work Domain Analysis, this analysis highlighted that all teams have an intra- and interagency requirement for effective incident evaluation (initial response, visual assessment, accident assessment). It is likely that of all functions at the incident, incident evaluation requires the highest level of communication, and shared and complimentary policies and practices. In the study, participants discussed that an interoperability channel is currently being trialed by the Queensland Police Service, Queensland Fire and Emergency Services, and the Queensland Ambulance Service, which may be a reflection of agency understanding of the importance of interagency communication for successful interoperability. However, currently, the channel will only be used between the Queensland Police Service, Queensland Fire and Emergency Services, and Queensland Ambulance Service. This analysis suggests that consideration should perhaps be given to extending the functions so that communication is tailored to the requirements of specific scenes. For example, a communication channel can accept other groups such as the BMTMC, Traffic Response Unit, and specialist groups if the incident requires them to be working in the environment. Also, incident channels should be available so that everyone working at a particular incident can be in contact.

Overall, the Work Domain Analysis and Control Task Analysis sections of the analysis identified shared aspects of the incident and highlighted aspects of the scene that required coordination as well as identifying scene tasks that were individual functions of agencies. This information is useful in determining focus of effort for interagency coordination.

Team Strategies Analysis

In team Strategies Analysis, how tasks are executed at incidents is investigated across four categories. The first is operational—strategies that explain how to carry out control tasks. The second category is coordination—strategies for analyzing coordination structures and the process of coordinating structures. The third is team development—strategies that use Tuckman’s team development model to understand how behaviors change during the team life cycle (Tuckman & Jensen, 1977). The final category is structural—the strategies that build on work domain constraints (Ashoori et al., 2014). For this exercise, a discussion between responders about the different strategies for incidents involving HAZMAT versus “normal” (non-HAZMAT) incidents enabled an analysis of likely operational strategies and team coordination strategies. Team development strategies analysis and structural strategies require an analysis of actual teams working together, and as this analysis was conducted on a hypothetical incident, it was impossible for these analyses to be conducted.

Operational strategies

Using a modification of Information Flow Maps and the Contextual Activity Template, operational strategies at incident scenes can be identified. In Queensland, regulations dictate that the Queensland Police Service manages the scene of standard traffic incidents. However, in the case of incidents involving HAZMAT, Queensland Fire and Emergency Services become the controlling agency at the scene. In Figure 8, the operational strategy for standard visual assessments by responder agencies is compared with the operational strategy for incidents where there is HAZMAT. This information is useful for traffic incident management process development and also training as it identifies team structures and interaction patterns under different circumstances.

Figure 8.

Operational strategies table for visual assessment at traffic incidents.

Coordination strategies

Using the HAZMAT example from operational strategies, an example coordination strategy can be investigated (Figure 9). Although a HAZMAT incident has a more streamlined operational strategy, communication between agencies and their communication teams remains separate. According to the coordination strategy for communication of HAZMAT requirements following the Queensland Fire and Emergency Services visual assessment, a diplomatic coordination structure is evident. Rasmussen, Pejtersen, & Schmidt (1990) identified that in diplomatic coordination structures, the individual decision makers can only coordinate with their neighbor decision makers and the information traffic is locally planned. In Figure 9, when the station officer of the crew sent to visually assess the incident saw the HAZMAT situation, he or she consults with guidelines and then communicate them to Firecom, the Queensland Police Service officer in charge, the senior Traffic Response Unit responder, and the senior Queensland Ambulance Service officer at the scene. Those officers then contact their own communication teams as well as the lower level officers at the scene. As this is a hypothetical incident, it is impossible to investigate actual coordination processes, but team coordination strategy analysis for traffic incident management could be used to understand the processes underlying effective coordination and to identify poor coordination structures. It is likely, given the identified nonoptimal coordination strategy for this desktop exercise, that the analysis would yield useful results.

Figure 9.

Team coordination strategy for communication of HAZMAT requirements at the incident.

Team Strategies Analysis findings

Strategies Analysis has been successfully used to map the complexity of the road system previously. Cornelissen, Salmon, McClure, and Stanton (2012) used a Strategies Analysis diagram to map different road user types at complex intersections. Ashoori et al. (2014) used team Strategies Analysis to identify different options for actors in medical teams in different workplace situations. For this hypothetical incident, an example was used to test the applicability of the analysis to traffic incident management. Operational strategies for a normal incident versus a HAZMAT incident were quite different. The HAZMAT incident was far more streamlined. This type of analysis would be useful in the traffic incident management environment to map real versus ideal operational strategies of incidents. For example, the participants in this exercise identified that the rules for HAZMAT coordination are strictly followed in the urban environment but less so in the regional/rural environment, largely due to the reduced resources and officer experience in regional/rural areas.

Using the same example, an analysis of coordination strategies for a HAZMAT incident identified that a diplomatic coordination strategy is used in the traffic incident management environment. This strategy is not optimal for HAZMAT incidents due to the urgency of the information flow requirements. Consideration may need to be given to developing alternative coordination strategies for emergency scenarios.

Discussion

This study investigated intra- and interagency team requirements at traffic incidents using a modification of team Cognitive Work Analysis. The aim was to establish collaboration points and to develop recommendations to improve intra- and interagency coordination, collaboration, and interoperability at traffic incident scenes. In line with this aim, the analysis uncovered themes for effective traffic incident management team coordination and generated new knowledge, especially regarding agency interactions, by analyzing traffic incident management holistically.

Improving Traffic Incident Management

Overall, a number of key themes were identified in the study that could be used to focus efforts for emergency responder agencies looking to improve traffic incident management.

The first theme relates to priorities and shared awareness. The lack of shared awareness at the incident scene creates the risk of interoperability issues. As an example, when a fatality occurs at a HAZMAT incident, the police require the scene to remain untouched until an investigation can be carried out. However, the road authority requires the HAZMAT material to be cleared from the road within a time period to prevent damage to the road. The responders from each agency do not have visibility of the competing priorities while at the scene. Therefore, it is likely that one agency will hamper the requirements of the other. If the road authority clears the HAZMAT material, the investigation scene will be corrupted. If the HAZMAT material remains on the road, the road will need to remain closed for days to repair, causing wider community issues. These types of issues indicate the requirement for an overarching governance arrangement between agencies, as well as risk matrices/decision-making processes for complex incidents. In a future study, it would be useful to conduct workshops using human factors and human-centered design techniques to co-design traffic incident management training and processes to better support shared awareness at incidents.

Communication and consultation was another emerging theme that requires consideration. Agency communication teams are not aligned, despite the fact that all agencies share the same communication functions and priorities. Consultation is also a shared requirement by agencies throughout the entirety of an incident. It is likely that aligning communication teams and creating a high-level interagency governance structure for traffic incident management would accommodate this shared requirement.

The desktop exercise was of a complex incident. A requirement to review the current coordination strategy for HAZMAT incidents was evident in the findings of the analysis, but it is likely that all complex incident types need review. Policies and processes written by agencies in isolation despite the fact that the functions at the scene require coordination and collaboration are more likely to become a safety issue in complex environments.

Limitations and Further Research

A limitation of this study is that the scenario was a desktop exercise rather than a real incident. It is possible that processes identified in an office would not occur in real incident situations. A further limitation was that only one incident was analyzed so the team coordination and collaboration requirements identified for the incident might be specific to the incident rather than generalized requirements. A third limitation was that the group were all officers from an urban center. Differences between urban and regional/rural responses were discussed in the exercise and it is likely the analysis could not be generalized to regional/rural incidents. The study could also have benefited from representation from other agencies at the scene—the Queensland Ambulance Service, the Environmental Protection Agency, the Coroner and tow truck operators. Due to the level of experience of the participants, however, they were able to report on likely perspectives and actions of the agencies that were not represented.

Despite the limitations of the study, the potential of team Cognitive Work Analysis as a tool to analyze the traffic incident management environment is evident. Future analyses could conduct studies interviewing decision makers from attending agencies at real incidents, in urban, regional, and rural environments. A study conducted at real incidents would also include asynchronous decisions, which were a noted limitation in team Control Task Analysis. An interesting future study could also include a wider array of officer types from the participating agencies. For example, communication teams, training staff, policy makers, and specialist teams aside from the Queensland Police’s Forensic Crash Unit (who were represented) may add extra perspective about the work domain and work flow at traffic incidents. Information from this type of analysis would inform optimal processes and practices in the traffic incident management environment. This paper contributes to the Cognitive Work Analysis (CWA) field by providing an example of applying team CWA to a new domain. The desktop exercise approach used, and the obtained findings about improving interagency coordination and communication, may be relevant to other emergency management fields. Any parallels with other complex environments, for example medical, military, and mining environments, might offer valuable insights to improve traffic incident management. Finally, given the effectiveness of applying team Cognitive Work Analysis to the traffic incident management domain, it would be beneficial for future studies to attempt a similar application across other domains, perhaps requiring some minor adjustments to the methodology as was the case for this study.

Conclusion

The traffic incident management system requires intra- and interagency team coordination and collaboration. The current study investigated traffic incident management using an adaptation of team Cognitive Work Analysis. Previously, analyses using team Cognitive Work Analysis were mainly restricted to medical teams. The successful application of team Cognitive Work Analysis into a new domain indicates the framework may be useful more broadly. This analysis was also helpful in raising new issues that may lead to better interagency coordination and communication. Where the individual analyses of previous studies by the authors (Cattermole, Horberry, Burgess-Limerick, Wallis, & Cloete, 2015; Cattermole et al., 2016; Cattermole et al., 2014) identified system issues and solutions related to individuals—for example, the need to create stronger policies to better support novice responders at their first fatality crashes—the group process identified system issues relating to gaps in team coordination, communication and structures as well as the policies and processes related to intra- and interagency teams. The framework enabled an examination of the traffic incident management system as a whole—establishing the purpose, priorities, tasks, and resource requirements at incidents over time. Key weaknesses in the traffic incident management system identified by the analysis were

disparate priorities at the scene creating the risk of interoperability issues,

nonalignment of agency communication teams despite the fact that all agencies share the same functions and priorities regarding the tasks related to the communication teams,

policies written in isolation by separate agencies despite the fact that the functions at the scene require coordination and collaboration between agencies,

a requirement to improve consultation across agencies and at the incident scene, and

a requirement to review the current coordination strategy for HAZMAT traffic incidents.

The application of team Cognitive Work Analysis for traffic incident management has produced novel insights that could potentially improve the safety of responders working at traffic incidents, and also the effectiveness of the traffic incident management system, leading to improvements to casualty outcomes and reduced risks for motorists and responders in and around traffic incidents.

Footnotes

Acknowledgments

The authors thank the participants who took place in this research. In addition, they thank Mr. Brendan Lawrence for his research assistance in creating and revising the figures in this paper.

ORCID iD

Tim Horberry

Vanessa Cattermole-Terzic is the Director, Research and Insights for Queensland’s Department of Transport and Main Roads, with 20 years’ experience in government roles across transport, police and education agencies.

Tim Horberry works at both Monash University Accident Research Centre in Australia and Coventry University in the UK. He has conducted applied human factors research in the road transport, mining, rail and medical domains for over 25 years.

References

Ashoori

Burns

(2013). Team cognitive work analysis structure and control tasks. Journal of Cognitive Engineering and Decision Making, 7, 123–140.

Ashoori

Burns

C. M.

d’Entremont

Momtahan

(2014). Using team cognitive work analysis to reveal healthcare team interactions in a birthing unit. Ergonomics, 57, 973–986.

Carson

J.L.

(2010). Best Practices in traffic incident management. Federal Highway Administration Report # FHWA-HOP-10-050.

Cattermole

V. T.

Horberry

Burgess-Limerick

Wallis

Cloete

(2015, October). Using the critical decision method and decision ladders to analyse traffic incident management system issues. Paper presented at the 1st Australasian Road Safety Conference, Gold Coast, Australia.

Cattermole

V. T.

Horberry

Hassall

(2016). Using naturalistic decision making to identify support requirements in the traffic incident management work environment. Journal of Cognitive Engineering and Decision Making, 10, 309–324.

Cattermole

V. T.

Horberry

Wallis

Cloete

(2014, November 12–14). An operator-centred investigation of safety issues for emergency responders at traffic incidents. Paper presented at the Australasian Road Safety Research, Policing and Education Conference, Melbourne, Australia.

Cattermole-Terzic

(2017). A human factors investigation into the effectiveness of traffic incident management systems (Unpublished PhD thesis). The University of Queensland, Brisbane, Australia.

Charles

(2007a). Review of current traffic incident management practices (Research Report No. R297/07). Sydney, Australia: Austroads.

Charles

(2007b). Traffic incident management guide to best practices (Research Report No. AP-R298/07). Sydney, Australia: Austroads.

10.

Cooke

N. J.

Gorman

J. C.

Myers

C. W.

Duran

J. L.

(2013). Interactive team cognition. Cognitive Science, 37, 255–285.

11.

Cornelissen

Salmon

P. M.

McClure

Stanton

N. A.

(2013). Using cognitive work analysis and the strategies analysis diagram to understand variability in road user behaviour at intersections. Ergonomics, 56(5), 764–780.

12.

Farradyne

P. B.

(2000). Traffic incident management handbook. Prepared for Federal Highway Administration, Office of Travel Management. Retrieved from https://hcm2010.org/system/datas/65/original/Traffic%20Incident%20Management%20Handbook.pdf

13.

Fiore

S. M.

Salas

(2004). Why we need team cognition. In Salas

Fiore

S. M.

(Eds.), Team cognition: Understanding the factors that drive process and performance (pp. 235–248). Washington, DC: American Psychological Association.

14.

Fischer

Krzmarzick

Menon

Shankwitz

(2012). Performance analysis of squad car lighting, retro-reflective markings, and paint treatments to improve safety at roadside traffic stops (Report CTS No. 12–13). Minneapolis: University of Minnesota.

15.

Flavell

J. H.

Wellman

H. M.

(1975, August). Metamemory. Paper presented at the 83rd Annual Meeting of the American Psychological Association, Chicago, IL.

16.

Gutwin

Greenberg

(2002). A descriptive framework of workspace awareness for real-time groupware. Computer Supported Cooperative Work, 11, 411–446.

17.

Kenny

D. A.

La Voie

(1984). The social relations model. Advances in Experimental Social Psychology, 18, 141–182.

18.

Klein

(1998). Sources of power: How people make decisions. Cambridge, UK: The MIT Press.

19.

Klein

Calderwood

Macgregor

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

20.

Klein

Wright

(2016). Macrocognition: From theory to toolbox. Frontiers in Psychology, 7, Article 54.

21.

Langan-Fox

Code

Langfield-Smith

(2000). Team mental models: Techniques, methods, and analytic approaches. Human Factors: The Journal of the Human Factors and Ergonomics Society, 42, 242–271.

22.

McNeese

M. D.

Rentsch

J. R.

Perusich

(2000). Modeling, measuring, and mediating teamwork: The use of fuzzy cognitive maps and team member schema similarity to enhance BMC/sup 3/I decision making. Proceedings of the IEEE International Conference on Systems, Man, and Cybernetics, 2, 1081–1086.

23.

Naikar

Elix

(2016). Reflections on cognitive work analysis and its capacity to support designing for adaptation. Journal of Cognitive Engineering and Decision Making, 10, 123–125.

24.

Naikar

Moylan

Pearce

(2006). Analysing activity in complex systems with cognitive work analysis: Concepts, guidelines and case study for control task analysis. Theoretical Issues in Ergonomics Science, 7, 371–394.

25.

Rasmussen

(1983). Skills, rules, and knowledge; signals, signs, and symbols, and other distinctions in human performance models. IEEE Transactions on Systems, Man, and Cybernetics, 13, 257–266.

26.

Rasmussen

Pejtersen

A. M.

Schmidt

(1990). Taxonomy for cognitive work analysis. Roskilde, Denmark: Risø National Laboratory.

27.

Rentsch

J. R.

Mello

A. L.

Delise

L. A.

(2010). Collaboration and meaning analysis process in intense problem solving teams. Theoretical Issues in Ergonomics Science, 11, 287–303.

28.

Salas

E. E.

Fiore

S. M.

(2004). Team cognition: Understanding the factors that drive process and performance. Washington, DC: American Psychological Association.

29.

Scherer

R. F.

Petrick

J. A.

(2001). The effects of gender role orientation on team schema: A multivariate analysis of indicators in a US federal health care organization. The Journal of Social Psychology, 141, 7–22.

30.

Shiffrin

R. M.

Schneider

(1977). Controlled and automatic human information processing: II. Perceptual learning, automatic attending and a general theory. Psychological Review, 84, 127–190.

31.

Tuckman

B. W.

Jensen

M. A. C.

(1977). Stages of small-group development revisited. Group & Organization Management, 2(4), 419–427.

32.

Vicente

K. J.

(1999). Cognitive work analysis: Toward safe, productive, and healthy computer-based work. Mahwah, NJ. Lawrence Erlbaum.