Wildfires and the COVID-19 pandemic: a systematized literature review of converging health crises

The emergence and global spread of the COVID-19 pandemic in 2020 converged with wildfire seasons of unprecedented extent. These co-occurring crises brought the potential for amplified health impacts. A systematized literature review was conducted to identify the health impacts from co-exposure to wildfires and the COVID-19 pandemic. A search of PubMed and Scopus identified 373 distinct references which were screened according to predetermined criteria. A total of 22 peer-reviewed publications were included in the final analysis. Studies were located in Australia and the western United States, with a single study in the Amazonian region of Brazil. The studies identified focused primarily on the impact of wildfire smoke exposure on COVID-19 infection and mortality, and the impact of exposure to both crises on mental health. The collective evidence shows that wildfire exposure within the context of the pandemic exacerbated COVID-19 infection and mortality as well as various adverse mental health effects. Additional research is needed in more diverse contexts and with individual-level data. Findings highlight the need for public health preparedness to anticipate overlapping, related crises and to advance climate change mitigation to protect public health.


Introduction and background
As the COVID-19 pandemic enters its fourth year, the Southern hemisphere reenters wildfire season after years of unusually intense wildfires in both hemispheres. Unprecedented wildfires in Australia in late 2019, coined the Black Summer, captured global attention in the months before detection of COVID-19. In fact, the spread of COVID-19 and the Black Summer were nearly concurrent crises in both their inception and escalation [1].
Before the COVID-19 pandemic was declared in March 2020, the year had already been extraordinary for extreme wildfires raging across Australia, followed by additional record-setting fires in Siberian tundra, Brazilian wetlands, and the western United States [2]. Further record-breaking wildfire seasons in areas such as the western United States, Siberia, and Turkey ensued in 2021; and in Europe, Mongolia, Argentina, and the western United States among other locations in 2022, all while the pandemic continued in parallel.
These events have been potentiated by anthropogenic greenhouse gas emissions. The United Nations' Intergovernmental Panel on Climate Change (IPCC) reports with high confidence that climate change has already increased the frequency and intensity of fire weather [3]. This has led, along with other factors, to up to twice the burned area compared to natural levels across global tropical, temperate, and boreal ecosystems [4]. Not only has global warming led to higher temperatures, aridity, and drought which promote the emergence wildfires [4], but climate change-induced lightning strikes may also increase ignition risk, and climate change-induced winds may accelerate spread [5]. Additionally, higher atmospheric carbon dioxide (CO 2 ) levels have created favorable conditions for invasive grasses in semi-arid lands which further increase fire risk [6].

Screening
Results were screened by a single reviewer according to pre-determined Population, Exposure, Outcome, and study Type (PEOT) exclusion and inclusion criteria (table S1). PEOT criteria were drafted a priori and criteria were refined during a pilot screening. Specific comparison group criteria were not established due to the lack of individual-level exposure measurements, the universal nature of the COVID-19 pandemic, and to allow for qualitative and mixed-methods studies. No studies were excluded based on language.
COVID-19 exposures were defined as any health effect of the COVID-19 pandemic, including the impacts of response policies or related impacts on social, cultural, and economic factors. Based on pilot screening, exposure criteria were set to allow a concurrent exposure span of up to six months. This decision was made based on the observation of a six month lag period of increased respiratory disease in those with high wildfire exposure [29]. Previous evidence suggests that mental health effects from wildfires are also persistent, remaining elevated five years after exposure [30,31]. This six month period was used for all health outcomes for consistency. Wildfire exposure was defined as any exposure to wildfires including to associated smoke, fire, displacement, loss of property, injury, fear, worry, or air pollution.
Studies were excluded if any criterion was not met. Studies labelled as Include, Unclear, or No Abstract at the title and abstract stage proceeded to the full text stage. Studies in the Unclear category were screened by a second reviewer. Reviews that met all other study criteria were used for hand-searching. Full articles were retrieved and screened according to the PEOT criteria. Studies meeting the PEOT criteria at the full text stage were included in the final study list and were hand-searched to identify additional candidates for inclusion.

Data extraction
For studies included in the final synthesis, characteristics and data were extracted by a single reviewer into Excel according to predefined categories. These included study location, study method and design, study sample (population) and demographics, wildfire exposure agent, COVID-19 exposure agent, whether these exposures were concurrent, risk factors and vulnerabilities identified, physical or mental health or well-being outcomes, effect and certainty measures for relevant findings, whether climate change context was given, and any stated research or policy implications. Key results of the updated search were added and discussed in the context of their additions to original findings.

Results
A total of 373 unique search results were identified (figure 1). Of these, 283 articles failed to meet inclusion criteria at the title and abstract stage and were excluded from further analysis. Of the remaining eligible 90 articles, 21 met all inclusion criteria at the full text stage. The most common reason for exclusion at the full-text stage (n = 34) was study type. One study [32] was identified through hand-searching of another included article [33]. A total of 22 studies were included in the final synthesis.

Study characteristics
Summaries for the 22 studies identified in the main search are found in table 1. Most included articles studied the United States (n = 13) or Australia (n = 8), with one study conducted in the Brazilian state of Pará in the Amazon region. The United States studies were concentrated on the West Coast, primarily focused on California (n = 9 of 13). Other Western states studied include Washington, Colorado, Oregon, and Nevada. Most Australian studies focused on New South Wales or the broader East Coast, while the remainder examined the nation overall.
Most study periods (n = 15 of 22) fell entirely within the 2020 calendar year. Study periods ranged from 2004 to 2021; however, only one study extended slightly into 2021. Two studies spanned several years to establish pre-COVID-19 pandemic baseline. Most studies used quantitative methods only (n = 20 of 22), though two Australian studies used qualitative or mixed methods.

Wildfire and COVID-19 pandemic exposures studied
Most studies (n = 12 of 22) measured smoke-related exposures to wildfire, either in general or in terms of specific air quality components of wildfire smoke. The most-studied smoke component was PM 2.5 (n = 8 of 12 studies measuring smoke exposure), with other smoke components less studied. Other than direct air monitors, studies measured smoke by self-report [34,35] or satellite [36][37][38]. Many other studies (n = 8 of 22) considered wildfire exposure in an open-ended framing within a given time period (typically qualitative studies, or time series analyses of periods associated with prominent wildfires). The remaining studies (n = 2 of 22) approximated exposure through local fire clusters [29] or percentage of land area burned [39]; the latter study also measured evacuation and sheltering-in-place behaviors. Most studies (n = 14 of 22) measured COVID-19 infection or mortality as a health outcome (i.e. implied exposures to the SARS-CoV-2 virus). Another contingent of studies (n = 6 of 22) considered any COVID-19 impact, often through qualitative study or by measuring the health outcome of interest during the COVID-19 pandemic in comparison to earlier time periods. The remaining (n = 2 of 22) studies measured actual exposures to SARS-CoV-2 (self-reported possible or actual exposures to the virus, close social impact), or to various COVID-19 pandemic related impacts including job loss, working from home, financial distress, and impairment in work and social domains [34,35].
Concurrency of exposure to wildfire and the COVID-19 pandemic was determined by time and place. All study populations in the United States and Brazil were exposed to wildfires during the COVID-19 pandemic (n = 14 of 22). Australian study populations were primarily exposed to wildfire in the months prior to the COVID-19 pandemic (n = 8 of 22).

Health outcomes studied
The included studies focused primarily on respiratory and mental health outcomes. Most studies measured the combined effect of wildfire and the COVID-19 pandemic on COVID-19 infection (n = 14 of 22). Another set of studies analyzed the impact of these co-exposures on mental health outcomes (n = 5 of 22). The three remaining studies measured the combined impact of the Australian Black Summer bushfires and COVID-19 pandemic; health effects studied include pregnancy and birth outcomes [40], health behaviors of those with multiple sclerosis [41], and unintentional fatal coastal drownings [42].
Of the studies measuring the impact of wildfire exposure on COVID-19 infection outcomes, the most commonly-measured was COVID-19 case counts or incidence [38] (n = 11 of 14). Half of these studies also measured death counts attributed to COVID-19 [43][44][45][46] (n = 7). Schwarz et al studied the effect on COVID-19 case-fatality ratios. Schroeder et al studied COVID-19 mortality rate ratios of high fire cities compared to low fire cities. Kiser et al studied SARS-CoV-2 test positivity rates in a Reno hospital during wildfires. Altogether, over half of studies on the COVID-19 impact of wildfires considered disease mortality (n = 9 of 14).
The mental health outcomes most commonly studied for association with wildfire and COVID-19 pandemic exposures were depression symptoms and anxiety symptoms (n = 4 of 5), followed by isolation, social connection, or loneliness (n = 3 of 5). Dawel et al explored cross-sectional symptoms of anxiety and depression and overall psychological wellbeing measured by self-report questionnaires [35]; Batterham et al studied the longitudinal trajectory of anxiety and depression symptoms in the same Australian National COVID-19 Mental Health, Behavior and Risk Communication (COVID-MHBRC) cohort [34]. Van Beek and Patulny explored loneliness and social cohesion outcomes through semi-structured in-depth interviews [47]. Arjmand et al surveyed participants on depressive and anxiety symptoms and social connectedness [48]; Sugg et al used crisis text line activity to measure overall mental health crisis but also stress and anxiety, depression, isolation, suicidal thoughts, self-harm, abuse, and substance abuse [49].

Relevant findings
Full summaries of relevant findings from each included study are presented in the supplemental file (table S2).

COVID-19 infection
Ten ecological studies and two case studies reported on the impact of wildfire on COVID-19 infection. Of these, most (n = 9 of 12) examined wildfire smoke or its components as the main exposure, most commonly measuring PM 2.5 components (n = 8). Each study found a significant positive association of COVID-19 infection with smoke overall [38,45] or smoke components. Studies of smoke-associated PM 2.5 consistently found significantly positive associations with COVID-19 infections with the exception of two studies which did not test for significance [32,37].  [38]; however, Cortes-Ramirez et al suggest their use of a spatial interpolation model to estimate PM 10 exposure may have been insufficiently accurate. Ademu et al found that a 1 µg m −3 increase in CO was associated with 36% more daily COVID-19 cases [50]; Meo et al also identified this significant positive correlation in an initial cross-sectional study [44] and in a longitudinal study later in the pandemic [45]. Results for NO 2 are less clear-cut. Ademu et al found a significant negative relationship between NO 2 and COVID-19 cases in most models [50]; Sannigrahi et al found a positive significant relationship between NO 2 and COVID-19 cases, albeit weaker and less consistent across time periods than PM [46]. Finally, Meo et al found that ozone also exhibited a positive, significant association with COVID-19 infections [45]. Lag-time effects between exposure and COVID-19 infection effects following exposure were reported by four studies and included 2-6 d [33]; and the following one [32,50], two [50], three [50], and four weeks [43].
Page-Tan and Fraser studied the impact of general wildfire intensity (percentage of land area burned) and evacuation or sheltering-in-place behaviors on COVID-19 cases [39]. In otherwise similar cities, wildfire intensity was associated with higher COVID-19 infection. Increased wildfire-related mobility was associated with lower COVID-19 infection. Sheltering-in-place was associated with lower COVID-19 infection, especially within cities.
Two case studies of COVID-19 in wildfire responders in the western U.S. were identified. In Backer et al's study of 1631 frontline firefighters and conservation support personnel seen for medical care at base camps, 7.23% of patient care records were for suspected COVID-19 infection during deployment [51]. Of 74 emergency department transfers, 8 were for pulmonary problems including asthma and potential COVID-19 infection. No large-scale COVID-19 outbreaks at these base camps were reported, but longer-term outcomes were not explored. In Metz et al's study of 6123 responders to a separate wildfire, 79 cases were identified of which 78 were confirmed [52]. Most cases (n = 73 of 79) were firefighters for an attack rate of 5.8% among 1260 full-time firefighters. Of the 79 cases, 13 visited an emergency department, 3 were hospitalized, and no deaths were reported. Phylogenetic and social network analysis showed spread between and within distinct firefighting crews. Contact tracing identified at least 273 close contacts.
Although some studies identified regional variability in the relationship between wildfire exposure and COVID-19 infection, none identified specific risk factors or vulnerabilities at the individual or community level.

COVID-19 mortality
The identified studies reported COVID-19 severity outcomes exclusively in relation to mortality; hospitalization, long-term sequalae, and other severity measures were not available. Seven quantitative ecological studies examined the impact of wildfire smoke [36,45], wildfire smoke components [43][44][45][46], or number of local wildfires [29] on COVID-19 mortality. In Pará, Brazil, cities with a higher number of fires had 80% higher COVID-19 mortality rates [29], and COVID-19 deaths increased after wildfire exposure in all ten California counties studied [45]. Heavy wildfire smoke coverage was associated with heightened COVID-19 case-fatality-ratios in two of five California counties (Alameda and San Francisco) by up to 58 and 42 deaths per 1000 cases, respectively [36].
Wildfire smoke-associated PM 2.5 showed positive, significant associations with COVID-19 death counts in multiple studies, although these associations are less common and smaller in effect than PM 2.5 associations with COVID-19 infection. Across 92 western American counties, a 10 µg m −3 PM 2.5 increase was associated with an 8.4% (95% CI, 2.1-15.3) increase in COVID-19 deaths over the next four weeks, and 17 of the 92 individual counties showed this same relationship when analyzed independently [43]. Meo et al found that higher PM 2.5 was associated with higher cumulative, but not daily, COVID-19 deaths (r = 0.562, p < 0.001) [44], and in a later study, with daily COVID-19 death counts [45]. Sannigrahi et al found that maximum monthly PM 2.5 levels, but not mean PM 2.5 levels, were positively associated with COVID-19 death counts [46]. These authors also found that monthly maximum and mean PM 10 concentrations were significantly associated with COVID-19 deaths for all months analyzed, while NO 2 showed a similar relationship only in some months; overall, single unit increases in mean PM 10 and maximum NO 2 were associated with COVID-19 death increases of β = 0.085 and β = 0.260 [46]. For Meo et al, CO was associated with higher cumulative deaths (r = 0.315, p < 0.001) but not number of daily cases [44]. In a subsequent study, Meo et al found CO had an overall positive significant relationship with COVID-19 death counts which was reflected in some but not all fire zones analyzed [45]. These authors did not find an overall significant relationship between ozone and COVID-19 deaths in an analysis of five California fire zones, although a positive significant relationship was found for one fire zone [45].
Four studies comparing the strength of association between wildfire exposure and COVID-19 mortality between regions (e.g. counties, states) all found regional variability in this relationship [36,43,45,46]. No study examined risk factors or vulnerabilities related to wildfire impact on COVID-19 mortality.

Mental health outcomes 3.6.1. Depression symptoms
At the outset of the COVID-19 pandemic in Australia, financial distress, and work and social impairment due to COVID-19, were each associated with higher depression symptoms in a cohort of 1296 adults [35]. Although these associations were not significant in final adjusted models [35], subsequent longitudinal studies of the same COVID-MHBRC cohort found that those exposed to the bushfires were more likely to grow increasingly depressed during the pandemic [odds ratio (OR): 2.95; 95% CI: 1.24, 7.04] [34].
An Australian cohort of mood-monitoring app (MoodPrism) users followed from before the bushfires throughout the Black Summer and COVID-19 lockdown periods showed a significant increase in depression symptoms during the bushfire period compared to baseline [48]. Depression symptoms during the COVID-19 period were significantly greater than the baseline, but not greater than the bushfire period; this suggests that depression symptoms increased at the outset of the bushfire crisis and were sustained through the COVID-19 period [48].

Anxiety symptoms
At the start of the COVID-19 lockdown, the COVID-MHBRC cohort showed positive though non-significant associations between anxiety symptoms and both financial distress, and work and social impairment due to COVID-19 [35]. Continued study of this cohort showed that higher exposure to bushfire (OR: 5.78; 95% CI: 1.25, 26.9) or its smoke (OR: 2.79; 95% CI: 1.44, 5.38) prior to the pandemic, COVID-related job loss (OR: 3.16; 95% CI: 1.07, 9.35), or working from home (OR: 2.57; 95% CI: 1.02, 6.48) were each associated with significantly higher odds of a moderate, increasing trajectory of anxiety symptoms [34]. In the MoodPrism user cohort, anxiety increased significantly during COVID-19 but not during the Black Summer [48].
Although non-significant after controlling for demographic differences, crisis text lines received from areas affected by wildfires during the COVID-19 pandemic more frequently related to stress and anxiety than during the COVID-19-only periods [49].
A qualitative study in Australia found that the bushfires and COVID-19 pandemic exacerbated pre-existing anxiety conditions [47], with a cumulative effect of the two crises. As one participant explained, 'My anxiety since the pandemic has just been so high. It is also compounded by the bushfires. That whole time was such intense anxiety and then I feel like we barely had time to catch our breath [47].'

Loneliness, social cohesion, and social connectedness
During the COVID-19 pandemic, crisis texts from wildfire-affected area codes relating to isolation, relationship concerns, and abuse increased, while bullying texts decreased; although these changes were explained by demographic differences in texts compared to non-wildfire periods [49]. On the other hand, the MoodPrism user cohort showed a decrease in social connectedness during COVID-19 relative to both the bushfire and baseline periods, suggesting that decreases in social cohesion may be particular to COVID-19 rather than wildfire [48].
Qualitative evidence from semi-structured in-depth interviews of Australian adults showed that the bushfire crisis brought communities closer together, while COVID-19 contributed to feelings of isolation and distrust in rural communities and cohesiveness in urban communities [47].

Other mental health outcomes
Directly after the Australian bushfires and at the beginning of the COVID-19 lockdown, recent bushfire smoke exposure, financial distress related to COVID-19, and work and social impairment due to COVID-19 were each independently associated with lower general psychological wellbeing in the COVID-MHBRC cohort; however, none of these associations remained significant in final adjusted models [35].
The MoodPrism user cohort showed evidence of a cumulative effect of the Black Summer bushfires and the subsequent COVID-19 pandemic on positive psychological functioning; personal control, motivation, meaning and purpose, self-esteem, and sense of achievement: each declined significantly first during the bushfires, and then further declined significantly during the COVID-19 pandemic [48]. In the same cohort, emotional valence (pleasantness or unpleasantness of emotions [53]) and arousal did not significantly change during the bushfire period but did decline during COVID-19 relative to both the bushfire and baseline periods [48].

Mental health risk factors
Of the four studies reporting risk factors or demographic characteristics associated with increased vulnerability to mental health impacts during converging wildfire and COVID-19 crises, consistent themes emerge. Non-male gender identity (female, non-binary, other) [34,35,49], younger age [34,35,49], and pre-existing mental health conditions [34,35,47] were each commonly identified as risk factors. Batterham et al also identified recent adversity experiences-possibly highlighting the importance of converging crises-and fewer years of education as risk factors [34]. Rural or regional residence [47] and certain racial and ethnic identities (Asian-American, Hispanic, mixed-race) [49] were also identified as at greater risk for adverse mental health outcomes.

Other health outcomes
Brew et al found that Australian pregnancy and birth outcomes varied significantly according to the combination and timing of exposure to bushfires and COVID-19 lockdown and restrictions [40]. The nature of the health associations of exposure to one or both health crises during pregnancy varied by trimester of exposure. Birth cohorts were defined according to perinatal exposure to the bushfires only, COVID-19 lockdown only, or exposure to both crises (divided into two cohorts according to trimester). Outcomes significantly associated with perinatal exposure to both bushfires and COVID-19 lockdown included birth weight, pre-labor membrane rupture, (un)planned caesarean sections, and gestational diabetes [40]. Compared to baseline controls, the cohort exposed to both crises earlier in pregnancy experienced higher risk of prelabor membrane rupture [adjusted odds ratio (aOR) 1.21, 95% CI 1.07, 1.27], unplanned caesarean sections (aOR 1.09, 95%CI 1.00, 1.20), planned caesarean sections (aOR 1.10, 95%CI 1.00, 1.20), and low birth weight (aOR 1.18, 95%CI 1.03, 1.37). The cohort exposed to both crises later in pregnancy likewise experienced higher risk of unplanned caesarean sections (aOR 1.15, 95%CI 1.04, 1.27) but also high birth weight (aOR 1.16, 95%CI 1.02, 1.31) and lowered risk of gestational diabetes (aOR 0.87, 95% CI 0.79, 0.96).
Analysis of unintentional fatal coastal drowning in Australia during 2019-2020 showed that the risk of drowning more than doubled [relative risk (RR) = 2.08; 95% CI = 1.61-2.7; p < 0.001)] when adjusted for outdoor activity levels [42]. This risk during the bushfire portion of this period (RR = 2.32; 95% CI = 1.67-3.24; p < 0.001) was higher than during the COVID-19 portion (RR = 1.75; 95% CI = 1.11-2.77; p = 0.02) [42]. The authors propose that this increased lethality may be related to an influx of less experienced boaters and other coastal recreators as more people headed onto the water to pursue socially-distanced activities or to avoid smoke, fires, and beachgoing restrictions. Further analysis is needed to better characterize this relationship.
Both the Black Summer bushfires and COVID-19 pandemics led to worsening health behaviors for most affected Australians with multiple sclerosis [41]. Participants affected by both crises reported decreased physical activity due to facility closures or air quality concerns, and many experienced increases in unhealthy eating and alcohol consumption (more so during the pandemic than bushfires) and disrupted sleep due to boredom or stress [41]. Participants and their healthcare providers were particularly concerned that changes in these health behaviors would be sustained post-pandemic due to their associated increase in physical disability.

Stated implications and context
Authors' statements on study implications, if given at all, ranged from research to policy recommendations. Some studies called for more research on the mechanism of action for the influence of wildfire smoke on COVID-19 while others called for more research to disentangle and better characterize the various health impacts of converging crises. Only Meo et al explicitly recommended strengthening wildfire prevention efforts [44]. Leifer et al recommended reduction of biomass smoke exposure in general [32]. Arjmand et al and Van Beek et al recommended community-based interventions to strengthen social support and resiliency to crises [47,48].
Several authors called for policy and preparedness measures to include converging health crises. Lawes et al recommended that drowning prevention efforts be strengthened but also consider the influence of other public health crises on drowning [42]. Other studies made recommendations specific to considering pandemic prevention within wildfire response; Metz et al recommended testing and screening responders on arrival, encouraging vaccination, improving physical distancing onsite, and educating responders on the 'overlapping symptoms of smoke inhalation and COVID-19' [52]. Backer et al found their prevention measures instituted at fire base camps were effective in preventing outbreaks [51]. Page-Tan and Fraser recommended sheltering-in-place when safe for wildfire regions, and for evacuees, providing socially-distanced accommodations along with health screening procedures [39]. Kiser et al noted that American federal recommendations on wildfire smoke safety in COVID-19 such as staying inside, using air cleaners, and wearing N95 masks 'may not be helpful during a pandemic that requires social distancing and limits the availability of personal protective equipment,' instead recommending 'lowering the recommended healthy limit for PM 2.5 in cities with a high prevalence of SARS-CoV-2, establishing 'clean air' shelters that maintain social distancing, and allocating sufficient quantities of appropriate respirators to areas at high risk for wildfires [33]. ' Less than half of studies (n = 10 of 22) mentioned climate change as a contextual factor. Some briefly stated that wildfire research will be of increasing importance due to climate change [32,37,[40][41][42]. Others referred to the role of climate change in potentiating the wildfires being studied [29,36,37,40,43]. Multiple authors situated climate change as a source of concurrent, cascading crises that exacerbate other health crises like infectious disease events [36,43]. Schroeder et al emphasized the reinforcing nature of wildfires and climate disasters, alluding to the contribution of wildfires to climate change through the destruction of important carbon sinks [29].
A COVID-19 outbreak investigation in Colorado wildland firefighters by Metz et al is the first to employ whole genome sequencing and social network analysis to this context and showed that wildfire response led to 79 cases among these responders from multiple introductions of COVID-19 that spread both between and within crews [52].

Discussion
Twenty-two studies identified in this review demonstrate substantial negative health impacts associated with co-exposure to wildfire and COVID-19. This included a consistent association between wildfire exposure-primarily to smoke or its components such as PM-and higher rates of COVID-19 infection, and to a lesser extent, COVID-19 mortality. COVID-19 mortality outcomes were mainly studied in isolation from overall infection rates; therefore, higher COVID-19 mortality in wildfire-exposed areas may be explained in part by higher COVID-19 incidence. The most consistently studied and implicated smoke component was PM 2.5 .
Studies also suggest nuanced impacts of the convergent crises on mental health. This includes both evidence of increased anxiety and depression symptoms as well as evidence of differential impact by crisis type; e.g. bushfires may be associated with social cohesion and COVID-19 with social isolation. Multiple studies found that mental health impacts from the COVID-19 pandemic and wildfires disproportionately impacted individuals of non-male gender, younger age, and with pre-existing mental health conditions. These findings are reinforced by Bühler et al's evidence that the COVID-19 pandemic, wildfires, and other contemporary sociopolitical tensions inflicted cumulative mental health stressors that harmed young adults' psychosocial development [54]. Remarkably, no risk factors or vulnerable groups were identified in the impact of wildfire on COVID-19 infection outcomes within the existing evidence base, primarily because such risk factors were not considered in the study designs.
Of the three studies on outlying health outcomes, the most notable is Brew et al's large perinatal cohort study [40]. This natural experiment demonstrated that co-exposure to wildfires and the pandemic can be associated with major physical health impacts beyond COVID-19 itself. The study highlights the vulnerability of the perinatal life-stage broadly; but through its distinct exposure cohorts, it also shows the sensitivity of health outcomes to specific exposure windows.
The breadth of exposures associated with each health crises was underexplored apart from air pollutants and viral exposure. Wildfire-associated exposures other than air pollution or prevalence of local wildfires were understudied, with only one analysis of evacuation and sheltering in place behavior. Likewise, most COVID-19 pandemic exposures were somewhat poorly characterized; other than those studying general population exposure to the virus, only two studies measured pandemic-related social and financial effects. In order to inform policies and programs to limit the health impacts of these converging crises, additional research is needed to characterize the health impacts of specific social, economic, and psychological exposures associated with each crisis.
The most consistently studied relationship-wildfire smoke air pollutants and COVID-19 infection-would benefit from consistent measurement that could enable meta-analysis. A first step in this process may be an investigation that identifies a common metric for future studies of this kind. Such a study should allow for the still-evolving SARS-CoV-2 virus, symptoms, and pandemic context.

Limited geographies studied
The existing scientific literature on the health impacts of co-exposure to wildfires and the COVID-19 pandemic focuses almost exclusively on Australia and the western United States. This affects the extent to which findings are generalizable. Wildfires also occurred elsewhere during this period including Siberia [2], Ukraine [55], the Brazilian Pantanal [56], China [57,58], Indonesia [59], India [60], Colombia [61], and across the Mediterranean [62]. Further research is warranted into how these wildfires converged with the COVID-19 pandemic in these localities (i.e. incidence, response policies) and the resultant health impacts. Some cumulative health effects of wildfires and the COVID-19 pandemic may be culturally specific to the extent that they are socially-mediated. The United States and Australia are both wealthy, western nations. Little is known about the effects of these converging crises in low-or middle-income countries or non-western cultures.
Wildfires also expose populations thousands of miles from their source. Direct exposure to harmful wildfire smoke is much more far-ranging than indicated by the studies identified in this review. For example, radioactive smoke from the 2020 wildfires in the Chernobyl Exclusion Zone in Ukraine was detected in Western Europe [55]. Harmful smoke from the 2021 Bootleg fire in the western United States traveled across the continent [63,64]. Furthermore, the mental health impact of wildfires may extend beyond boundaries determined by physical smoke exposures; Byrne et al found in 2006 that vicarious exposure to wildfires, for instance via consumption of media coverage, has been associated with traumatization in the general Australian population [65].

Limitations of included studies
Most studies in this review did not measure either the wildfire or COVID-19 pandemic exposures at the individual level. Rather, wildfire exposures were typically measured by regional smoke, air pollutant, or wildfire count or overall time period. On the other hand, COVID-19 exposures were often not explicitly measured at all (e.g. general population viral exposure) and otherwise were typically defined by the overall pandemic or lockdown time period. Ecological study designs are limited; although new quasi-experimental and synthetic control methodologies offer improved certainty of effects at the ecological level, this lack of individual-level exposure data limits the differentiation and nuance possible amongst individuals. More granular data is needed for analysis of subpopulation and risk factors, particularly for physical health outcomes.
Potentially vulnerable populations have been under-explored. Brew et al's perinatal cohort study emphasizes the need for deeper attention to pregnant populations. Two studies have examined COVID-19 in wildfire responders. Although Backer et al did not find evidence of large COVID-19 infection outbreaks at wildfire firefighting base camps, the authors did not have access to testing. In a study of a concurrent wildfire, Metz et al had access to widespread testing and found an outbreak among responders (particularly firefighters, attack rate: 5.8%). This outbreak spread within, between, and beyond response crews. There was an active response from the local health department which likely kept the outbreak from growing further. Neither study indicates whether occupational exposures may predispose firefighters to COVID-19 long-term. Indeed, long-term impacts of wildfire smoke on firefighters has been underexplored in general [66]. Navarro et al hypothesize that wildfire smoke exposure in firefighters is likely to increase the risk for COVID-19 infection and severity [67], and a longitudinal study in this occupational group is warranted.
Farmworkers represent another key occupational group disproportionately exposed not only to wildfire smoke [68] but also to COVID-19 [69]. A study of California farmworkers found that COVID-19 infections were elevated by a factor of four due to essential worker status and socioeconomic vulnerabilities [69]. Notably, no studies on farmworkers were identified for inclusion. Research on health burdens of the wildfire-pandemic nexus in this key population requires further attention, particularly where wildfire-prone and agriculture-intensive regions overlap (e.g. the western U.S.).
The open-ended study of wildfire and pandemic exposures highlights the need for more research into which particular aspects of wildfires and the COVID-19 pandemic may interact. Without naming and measuring the specific exposure agents associated with each crisis, the key exposures and their interactions cannot be characterized. These steps are necessary for characterizing the mechanisms of action which lead to the cumulative health effects of these converging crises.
Moreover, studies thus far have focused largely on smoke and viral exposure, reflecting a quantitative, physical-biomedical perspective. Additional analysis of social exposures and non-infectious outcomes is warranted; social-related exposure results were fairly preliminary and undetailed in precisely what aspects of the COVID-19 pandemic or wildfires most impacted mental health, and how. Particular non-biomedical exposures warranting additional analysis are displacement, eco-grief [70] or solastalgia [71], vicarious exposures [72,73], and social isolation. Another open question is what type of social support may be most protective against these effects. Furthermore, long-term mental health impacts from wildfires can last years [30]; future research might examine the duration of impacts on individuals exposed to wildfires during COVID-19.
More broadly, a conceptual model is needed to elucidate the particular aspects of wildfires and the COVID-19 pandemic that might interact to impact health. For example, both exposure to wildfire smoke [74][75][76] and COVID-19 [77][78][79][80] have been associated with cardiovascular manifestations, although they are poorly understood. Such a framework might also consider that the issues of wildfire and COVID-19 are intertwined upstream of health (e.g. reinforcing feedback loop between wildfires and climate change, impact of wildland destruction on zoonotic disease emergence).

Implications
The findings of this study carry implications not only for COVID-19 but also for other contagious respiratory diseases. For example, a recent case-crossover study shows that wildfire smoke exposure is associated with higher rates of active tuberculosis [81].
Prevention and preparedness for health crises must assume that any single disaster may overlap with others, particularly as climate change unfolds. Disaster management plans for crises including wildfires and infectious epidemics should anticipate converging crises rather than compartmentalizing crisis or disaster type. Public health officials must plan for key interactions between crises both in health effects and in possible interventions or response.
Health emergency preparedness efforts must mainstream not just all hazards preparedness but multiple hazards preparedness together with climate preparedness [15]. Climate change impacts share risk pathways with other health disasters [82]. For example, Banwell et al provide a useful framework for linking disaster risk reduction and climate change adaptation work, particularly through taking advantage of synergies in infrastructure and community resilience building, policy, early warning systems, risk assessment, and health systems strengthening [83].
Beneficially, interaction also exists in potential prevention efforts for crises such as wildfires and emergent diseases. Specifically, climate change mitigation, land management, and conservation represent key opportunity areas for synergistic, upstream prevention efforts. The IPCC provides future climate change adaptation options, noting with high confidence, 'Integrated, multisectoral solutions that address social inequities, differentiate responses based on climate risk and cut across systems, increase the feasibility and effectiveness of adaptation in multiple sectors' [3]. This recommendation aligns neatly with the intersectoral, equity-driven public health discipline.

Conclusion
The emergence and subsequent spread of the COVID-19 pandemic converged with unprecedented wildfires. Evidence suggests that exposure to wildfires in the months prior to or during the COVID-19 pandemic exacerbated COVID-19 infection, COVID-19 mortality, and adverse mental health effects. Current evidence exists primarily for Australia and the western United States. Additional research is needed to explore this relationship in non-Western cultures and low-and middle-income contexts and to determine key risk factors and mechanisms of action. Future research should include more extensive analyses of risk factors and vulnerable populations. As we emerge from the initial phase of the COVID-19 pandemic, recovery efforts must include collective agency building and meaningful climate change mitigation actions. Future health emergency preparedness efforts should focus on potential crises not in isolation but in the context of multiple, cascading climate disasters.