Modeling the systemic risks of COVID-19 on the wildland firefighting workforce

Abstract

Wildfire management in the US relies on a complex nationwide network of shared resources that are allocated based on regional need. While this network bolsters firefighting capacity, it may also provide pathways for transmission of infectious diseases between fire sites. In this manuscript, we review a first attempt at building an epidemiological model adapted to the interconnected fire system, with the aims of supporting prevention and mitigation efforts along with understanding potential impacts to workforce capacity. Specifically, we developed an agent-based model of COVID-19 built on historical wildland fire assignments using detailed dispatch data from 2016-2018, which form a network of firefighters dispersed spatially and temporally across the US. We used this model to simulate SARS-CoV-2 transmission under several intervention scenarios including vaccination and social distancing. We found vaccination and social distancing are effective at reducing transmission at fire incidents. Under a scenario assuming High Compliance with recommended mitigations (including vaccination), infection rates, number of outbreaks, and worker days missed are effectively negligible, suggesting the recommended interventions could successfully mitigate the risk of cascading infections between fires. Under a contrasting Low Compliance scenario, it is possible for cascading outbreaks to emerge leading to relatively high numbers of worker days missed. As the model was built in 2021 before the emergence of the Delta and Omicron variants, the modeled viral parameters and isolation/quarantine policies may have less relevance to 2022, but nevertheless underscore the importance of following basic prevention and mitigation guidance. This work could set the foundation for future modeling efforts focused on mitigating spread of infectious disease at wildland fire incidents to manage both the health of fire personnel and system capacity.

Figure 1

Historical assignment/reassignment data for a single fire in Montana. The map of incoming assignments shows the range of origins for personnel assigned to a fire that started on July 15, 2017. The outbound reassignments shown include all incidents to which personnel went, given nine or fewer days between demobilization at the first fire and mobilization at the second fire.

Figure 2

(a) The possible viral states which individuals may travel through in simulations. The arrows indicate possible paths that individuals may take through the viral states. An individual may move directly from susceptible to recovered only if vaccinated. (b) Interactions between personnel on a single fire. Crew module members (individuals of the same color) interact only with other members of the same module, with the exception of module leaders, who interact both with their module members and with other module leaders. Management personnel cannot effectively form modules and thus interact with all other management personnel as well as a proportion who interact with the crew module leaders.

Figure 3

Daily cumulative infections by compliance scenario on and off fire (a) and annual cumulative infections by personnel type (b). In (a), each line is associated with a single scenario run while the bolded lines show a smoothed number of cumulative infections incurred. The total cumulative infections across the 2017 season by scenario and personnel role are shown in (b), with cases attributed to assignment status at time of exposure.

Figure 4

(a) Percentage of runs for which each fire had an outbreak by scenario and maximum number of personnel assigned to the fire on a single day. Two fires are singled out: the points associated with a “many outbreaks” fire are circled in blue and the points associated with a “fewer outbreaks” fire have a pink square around them. (b) The number of personnel over time for the “many outbreaks fire” and the “fewer outbreaks fire” that are indicated in (a).

Figure 5

(a) The number of infectious assignments and reassignments by scenario and personnel type for the 2017 fire assignment data. (b) A map of the infectious reassignments that occurred during the Low Compliance run using 2017 data that had the highest number of infectious reassignments (i.e., the worst case scenario observed). (c) A map of the infectious reassignments that occurred during the High Compliance run using 2017 data that had the highest number of infectious reassignments. All large fires included in the analysis are mapped as points, with the point size corresponding to the maximum number of personnel assigned to the fire on a single day. Lines connecting fires indicate infectious reassignments.

Figure 6

The distribution of worker days missed by scenario. The red denotes all workdays missed by vaccinated and unvaccinated firefighters while the blue denotes workdays missed by only unvaccinated firefighters. The Only Unvaccinated indication captures guidance current at the time. Brackets indicate the interquartile range and plus signs indicate the median value for each distribution.