Risk management of emergency service vehicle crashes in the United States fire service: process, outputs, and recommendations

Firefighting remains a dangerous occupation, resulting in over 63,000 injuries and an average of 81 deaths per year over the last 10 years [1‐3]. Emergency service vehicle crashes (ESVCs) contribute to a significant portion of firefighter morbidity and mortality [4‐6]. In 2014, there were over 14,000 ESVCs reported to the National Fire Protection Association (NFPA), resulting in approximately 600 injuries to firefighters [3]. In 2015, there were 13 vehicle-related fatalities including 8 ESVCs and 5 roadside struck by incidents, making ESVCs the 2nd leading cause of fatal injury in the United States (US) fire service for that year [2]. Furthermore, emergency service vehicles, with their large sizes and high speed operations, pose a significant hazard to civilians and public roadway motorists [5]. The cost of repair, property damage, injury and litigation as a result of ESVCs can result in considerable financial costs that are passed on to municipalities and the public [7, 8].

Risk management (RM) is a formalized proactive process that has been used in a range of occupational settings, including firefighting and mining, to reduce workplace hazards and injuries [9‐11]. Organizations use formal RM to manage risks and hazards through identification, assessment, and prioritization of risks for mitigation and intervention [11]. RM follows a cyclical set of three phases, including 1) hazard scoping, 2) risk assessment, and 3) implementation of controls within a feedback loop (Fig. 1). While commonly required in other countries to protect the health and safety of industrial workers, formal RM is not as widely practiced in the US [12, 13], despite growing evidence suggesting it may effectively reduce a range of occupational hazards and injuries [13‐15]. Widespread implementation of RM in Australia was associated with up to 78% reductions in lost-time injury rates in the Australian coal mining industry, compared to only a 20% reduction in the US coal mining industry which relied on compliance-based safety programs during the same timeframe [12]. An international comparison of fire service injury rates found that fire departments in the UK, with RM-based occupational safety regulations, had significantly lower fire service injury rates than US fire departments. RM has been previously used to reduce injuries and their economic consequences in the fire service. In the US, RM has been applied in the fire service to address injuries stemming from patient transport, fireground activities, and physical exercise [14]. Implementation of a risk management program aimed at improving the structure and management of physical exercise among recruit firefighters was associated with a statistically significant reduction in injury frequency and worker’s compensation claims cost [15]. These prior studies have shown that RM is was perceived to be useful to and valued by firefighters [13].

The goal of this study was to apply RM processes in three distinct US fire department settings to identify risks and hazards associated with ESVCs, determine potentially effective control measures, and evaluate the acceptability and utility of RM as a tool for mitigating ESVCs in the US fire service.

Three fire departments (A, B, and C), representing a range of urban-rural geographic classifications, population served sizes, and career/volunteer firefighter mix, participated in this study. Department A originally requested the research to assist in reducing ESVCs. Departments B and C were already fire service partners with the researchers at the time of study initiation. Department A was a career fire department, serving a large major metropolitan area. Department B was a combination career/volunteer fire department serving a mostly suburban population. Department C was the only volunteer department in the study, serving largely rural and wildland areas. Department C is staffed by <100 volunteer firefighters, and serve a population of less than 100,000. The department covers approximately 100 mile² that is 97% farmland, forestlands and recreational areas.

Teams within each department participated in a facilitated RM process delivered through focus groups and Delphi method panels. Three to six 2-h meetings were planned for each department to cover the scoping, risk assessment, and control implementation phases within the RM framework. Meetings were conducted in a combination of in-person and web-conference forums. All study procedures were approved by the University of Arizona Institutional Review Board.

RM has been used in the fire service to reduce occupational injury and associated economic costs [14, 15]. Previous research shows that RM is well-accepted by firefighters and that specific components of the RM process, especially the participatory approach and analysis of risks associated with specific tasks, are useful to and valued by firefighters [14]. In this study, we applied formalized RM to address ESVCs in our partner fire departments, which was a new application of the approach.

Emergency response driving was cited as a major concern across all departments and is a frequently cited risk factor in prior studies. Donoghue et al. reported that 66% of firetruck crashes occurred during emergency response [5]. Custalow and Gravitz reported that 91% of emergency vehicle collisions in Denver occurred under lights and sirens [17]. While the exact mechanisms that increase a driver’s risk of collision during emergency response is not completely clear, the association does seems to be driven by generally higher speeds and reduced reaction times during emergency responses [16]. Savolainen et al. reported emergency response crashes to be common in Michigan and cited speeding, overtaking, and passing to be common factors contributing to the risk of these collisions [18]. High speeds reduce driver’s ability to react to hazards, safely control the vehicle, and navigate traffic. Policy changes reclassifying certain categories of calls as not requiring an emergency response have reportedly proven effective in reducing ESVCs in other fire departments like in St. Louis, Virginia Beach and Phoenix [16] .

Poor visibility due to weather, road environment, and apparatus design were medium to high priorities for all fire departments. Visibility concerns were often raised in conjunction with backing incidents since driver’s rearward field of vision is practically obscured in all fire apparatuses. Fortunately, back up cameras and rear facing cameras have been found to be generally effective in reducing backing related incidents in civilian drivers, though their use has been rarely if ever tested in the emergency services [19, 20]. Clarke et al. found that emergency responders in the UK had generally low “blame worthiness” levels in emergency response collisions; however, failure to check blind spots was the most common factor in at-fault collisions [21]. Furthermore, Custalow and Gravitz have noted that visual clearance of intersections is necessary in reducing intersection collisions which have historically been the crash type associated with the highest risk of injury [17]. Department C reported that driver-vehicle incompatibility to be common concern since a driver’s field of vision can be diminished in fire apparatuses. This problem of driver-vehicle incompatibility is particularly exacerbated when drivers are not assigned to permanent vehicles which they may customize to fit their driving style [16].

Insufficient driver training and education was also cited as a common concern across all partner fire departments. Driver error has been cited as a primary cause in up to 93% of ambulance-involved collisions; thus, training remains a first-line priority for all departments despite being inconsistently applied and minimally required [22]. While NFPA 1002: Standard for Fire Apparatus Driver/Operator Professional Qualifications provides minimum job performance standards for emergency vehicle operators, the standards are not legally required nor enforceable. The only legal requirement is that drivers are trained to meet minimum state and federal laws for operating large heavy trucks. Moreover, since 1986, virtually all states have exempted emergency responders from Commercial Driver’s License (CDL) requirements which was designed to discourage unsafe drivers from operating large commercial vehicles [16]. Training space is also another concern, as Department A reported the lack of dedicated training spaces as a critical limitation to their hands-on driver training program. Dedicated driving space is an important feature in driver training, since training conducted on closed ranges have been associated with a reduction in crash risk over traditional on-road driver training [23].

A key advantage of RM—in contrast to compliance based safety programs—is the emphasis on establishing context and tailoring interventions to site-specific needs. The advantage of considering context and needs in selecting controls is that limited resources may be concentrated towards targeted interventions to address high priority risks instead of broad target interventions that may have only secondary or tertiary effects on a specific high priority need. The needs between urban and rural fire departments are particularly prescient given they operate in very different conditions [22]. This important advantage of RM was evident in the notable differences in prioritized risks and incidents in our dense urban departments and rural fire departments. The fire departments with more urban environments tended to prioritize risks and hazards associated with driving in traffic congested areas, while rural departments were more concerned with animal collisions and environmental hazards such as inclement weather on less developed roads. This is consistent with prior work by Ray and Kupas where they found that compared to urban settings, rural ambulance crashes tend to be more likely on snowy roads (13% vs 5%) and at night without street lighting (25% vs. 4%) [22]. Moreover, in comparison to rural departments, urban departments tended to experience more angled collisions with other vehicles (54% vs 19%), intersection collisions (57% vs 36%) and incidents at traffic signals or stop signs (53% vs. 14%), which is consistent with the risks experienced by drivers in dense urban areas [22]. Nearly all volunteer firefighters (95%) work in suburban and rural departments serving populations under 25,000, facing different training and working conditions than urban-based career firefighters [24]. RM scoping and risk assessments should account for volunteer firefighters responding from home, and factor in differences in apparatus training programs and driving experience.

A potential limitation of the risk assessment process that we used was the reliance on the perceived likelihood and severity of incidents to rank and prioritize vehicle incident types. While experiences and perceptions are valid, data driven assessment may provide more objective and substantive rankings of hazards to be controlled. However, given that data are often not available or poorly collected, opinion-based risk assessments are generally the norm, particularly in cases where the involved team members clearly understand the situations, operations and risks being reviewed [9, 14]. Risk matrices are rarely validated in their ability to improve risk management decisions, and are prone to poor resolution (i.e., very different risks can have the same ranking) as well as errors where a qualitatively ranked risk may be high with very low quantitative risk [25]. Another key limitation to our study is that the departments self-selected to participate in the RM process, resulting in a potential for selection bias. Though the effects of selection bias may be difficult to quantify, it is expected that the personnel that elected to participate in our study may differ from the general emergency service population (for instance in education or experience), thus the identified priorities and hazards during the risk assessment may not be representative of true priorities and hazards in the emergency services (i.e., external validity is limited). Regarding Department B, the crash data reviewed during scoping were limited to only career firefighters, thus inferences about risks and hazards in this combination department does not reflect the volunteer firefighter force.

Education and training featured prominently as a control adopted by our partner fire departments. Given the paramilitary nature of firefighting organizations and skill-based operations, it was not surprising that drivers training was a commonly advocated strategy to improve driver safety [16, 17]. Although the evidence base is mixed, prior studies suggest that licensing programs for young novice drivers may reduce the risk of fatal crash involvement, particularly in graduated driver licensing programs where the learning period and amount of practice is extended [26‐28]. The effectiveness of graduated licensing programs in civilian populations suggests graduated driver training programs where independent driving responsibilities are gradually increased may be effective in reducing the risk of crash involvement in the fire service.

It should be noted that published evaluations of driver training programs have predominantly been conducted on young novice drivers with little to no driving experience—not emergency services populations. We have found no published studies or reports evaluating the effect of emergency service vehicle operations training and recommend these programs be evaluated. In our collective experience working with the fire service, we have found initial fire service driver training programs to be variable in duration, availability, quality and curriculum, rendering the evaluation of fire service training programs difficult.

Though initial drivers training and licensure is important in developing safe driving skills, there have been no published studies to indicate post-licensure training or remedial training to be effective in reducing crashes in drivers that have already been trained [29, 30]. In a recent systematic review, Masten and Peck report that driver remediation is only slightly effective in reducing road crashes among civilian drivers, but most of the benefits came from driver suspension programs and only then could be due to more careful driving in the post suspension intervals evaluated [30]. In a recent meta-analysis of 21 randomized controlled trials studying the effects of post-licensure driver education, Ker et al. found small non-significant reductions in road traffic crashes among drivers receiving post-licensure training [31]. Custalow and Gravitz have found 71% of drivers involved in emergency response collisions had a history of multiple collisions, suggesting that remedial training may not be effective since drivers sent to remedial training remain at risk for repeated collisions [17]. These findings suggest that reliance on remedial training should be tempered and used in tandem with other approaches to reduce the risk of collision. Furthermore, there is evidence to suggest that training programs which focus on changing risk perceptions and reducing risk tolerance in additional to driving competency may be more effective than traditional skill-based training programs in reducing the future risk of crash involvement [28].

The bulk of challenges in the RM process occurred during the implementation of controls. The potential for using telematics driving data in a punitive fashion was a great concern for department union leaders. Engagement of union representatives and members from the outset is critical to a successful RM program, especially one targeting driving and crashes where fault and liability may prompt punitive action as opposed to education and proactive prevention. Simulator training in Departments A and B was hampered by reports of trainee nausea caused by ‘simulator sickness,’ which is a commonly reported negative side effect of stimulator use [32]. As previously stated, the format of the telematics driving data was provided directly from the company of the fire department partners. The data provided were too detailed for the safety managers and required substantial time to review and analyze. Based on feedback from our fire department partners, we are developing weekly visual ‘dashboard’ summary reports for each department which provides fleet-level as well as vehicle-level statistics of driving performance. The dashboard reports summarize miles driven, trips driven, driving rule exceptions, and ranks each vehicle against each other based on the number of driving rule exceptions per mile. The report is emailed to the safety department on a weekly basis. Each dashboard is customized to meet the needs of the department and what they deem important to safety and operations. We strongly recommend that departments planning to incorporate VDR or telematics data into a comprehensive safety program engage a vendor that can customize data reports and visualizations based on specific department goals and needs.

Risks and hazards for ESVCs are site dependent. Proactive RM introduced in our three partner fire departments was considered, by the RM teams, a beneficial and acceptable process, allowing each fire department to implement controls tailored to their own priority hazards. The findings of this study support the broader use of RM to increase the knowledge and understanding of ESVCs in the US fire service.