Article Text
Abstract
Objectives Hospital attendance related to fire, flame or smoke exposure is commonly associated with work. The aim of this study was to examine time trends and risk factors for work-related fire/flame/smoke injuries in Victoria, Australia.
Methods This study was based on emergency department (ED) presentation records from the Victorian Emergency Minimum Dataset, 2003–2021. Cases were people aged 15–74 years with injury-related ED presentations, if cause of injury was recorded as fire/flame/smoke, based on coded data and/or narratives. Work-related rates were calculated per employed persons; non-work rates were calculated per population. Work-related and non-work-related cases were compared using logistic regression modelling.
Results There were 11 838 ED presentations related to fire/flame/smoke: 1864 (15.7%) were work-related. Non-work-related rates were 12.3 ED presentations per 100 000 population, and work-related rates were 3.43 per 100 000 employed persons annually. Over the study period, work-related rates decreased annually by 2.0% (p<0.0001), while non-work rates increased by 1.1% (p<0.0001). Work-related cases (vs non-work) were associated with summer (vs winter), but the association with extreme bushfire periods (Victorian ‘Black Saturday’ and ‘Black Summer’) was not statistically significant. Work-related cases were less severe than non-work-related cases, evidenced by triage status and subsequent admission.
Conclusions Rates of occupational fire/flame/smoke-related injury presentations decreased over the past two decades in Victoria, while non-work-related rates increased. This could reflect improved safety in the workplace. Hospital data, however, cannot be used to distinguish occupation or industry therefore, employment data linkage studies are recommended to further inform workplace preventive measures.
- Wounds and Injuries
- Materials, exposures or occupational groups
- Occupational Health
- Firefighters
- Epidemiology
Data availability statement
Data may be obtained from a third party and are not publicly available.
Statistics from Altmetric.com
- Wounds and Injuries
- Materials, exposures or occupational groups
- Occupational Health
- Firefighters
- Epidemiology
WHAT IS ALREADY KNOWN ON THIS TOPIC
Australia has experienced extreme bushfires in recent years and the occurrence of these is expected to increase. Fire, flame and smoke-related injuries are common and many of these occur in a work setting.
WHAT THIS STUDY ADDS
In Victoria (Australia), 2003–2021, emergency department (ED) presentations for injuries related to fire, flame and smoke exposure at work were common: rates were 3.4 ED presentations per 100 000 employed persons annually. Work-related rates showed a decrease over the study period while non-work-related cases increased. Work-related cases were associated with summer in Victoria, which is generally bushfire season.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
Although the overall trend in fire/flame/smoke-related occupational injury showed improvements over time, suggesting improvements in workplace safety, further targeted preventive measures should be developed to prepare for bushfires. Studies linking hospital data with employment data would be beneficial to inform prevention specific to relevant occupational groups and industries.
Introduction
Morbidity and mortality related to fire, flame and smoke constitute a significant public health burden in Australia. In terms of morbidity, in Australia 2021/2022, there were 2192 hospital admissions related to exposure to fire, flames and smoke: this equates to a rate of 8.5 per 100 000 population.1 In terms of mortality, in 2020/2021, 79 deaths were attributed to fire/flame/smoke exposure. Extreme bushfires are a contributor to fire and smoke-related morbidity and mortality in Australia.2 Following on from the recent Black Summer bushfires of 2019/2020, which ranked among the worst bushfires on record, extreme fires such as these are expected to increase in incidence in the coming decades.3
Exposure to fire/flame/smoke exposure can occur at work: occupational exposure is not limited to first responders (firefighters, defence personnel, police, ambulance workers) but can also occur in other occupations commonly exposed to intentional fire (eg, chefs) or among workers in any occupation or industry exposed to unintentional fires, for example, house-fires or bush-fires. Therefore, occupational exposure constitutes a subset of fire, flame and smoke-related morbidity and mortality in Australia: in the Burns Registry of Australia and New Zealand, 12% of burns due to fire were work-related in people of working age.4 Work-related settings are more amenable to intervention through occupational health and safety measures: workplaces are regulated environments, and workers have the right to a safe workplace. Work-related fires in the past have triggered significant workplace safety reforms. Historically, a workplace fire that broke out over a century ago in the USA, resulting in 146 fatalities, sparked outrage and protests, which eventually led to workplace reform through improved workplace safety regulations.5 This landmark fire occurred in Manhattan, New York City, in 1911 and is known as the Triangle Shirtwaist Factory fire. The majority of the victims were recent immigrant girls and women aged 14–23 years.
Today, insights into the injury burden associated with work-related fire/flame/smoke exposure can be gained from workers’ compensation data, through analysis of injury claims. Studies using workers’ compensation data, however, tend to focus on occupational groups such as firefighters,6–9 rather than specific exposure to fire. In these studies, the most commonly reported injury types are generally musculoskeletal, caused by overexertion and falls, reflecting occupational hazards overall rather than fire-exposure specifically. Furthermore, workers’ compensation data does not capture all work-related injuries: in a study of work-related hospital cases and workers’ compensation claims, work-related emergency department (ED) presentation rates for open wounds and burns were almost three times greater than workers’ compensation claim rates for these injury types.10 Notably, hospital admission rates in this category were lower than compensation claim rates.
Firefighters are not the only workers potentially exposed to fire, flame and smoke: in a US study of occupational fatalities due to fire and flames, one-fourth affected manufacturing workers and another one-fourth affected construction or service workers, whereas the highest rates were observed among workers in the mining industry.11 Although we have some knowledge of the broader context through research on occupational burn injuries,12–14 burn injuries can result from causes other than fire and flames. In a recent Victorian study of ED presentations, admissions and deaths due to burns, only 37% of unintentional burn injury-related hospital admissions and less than 10% of unintentional burn injury-related ED presentations were recorded as caused by fire, flames or smoke.15 A US study of the National Burns Registry reported 32% of work-related burns to be caused by fire/flame.14
In light of the recent extreme bushfires in Australia, specifically the Black Saturday and the Black Summer bushfires, alongside predictions that the occurrence of extreme bushfire events is on the increase, the purpose of this study is to determine the occupational injury burden associated with fire and smoke exposure in Victoria, Australia. Hospital data are the primary data source for estimating population-level morbidity: this study draws on this resource but also demonstrates its limitations for use in fire-related occupational injury research. Specifically, the study aims are to: (1) determine 19-year time trends in work-related and non-work-related fire/flame/smoke injuries; (2) describe work-related injuries due to fire/flame/smoke in terms of patient demographics, setting, time of occurrence in relation to fire seasons and ED presentation outcomes; and (3) compare work-related versus non-work-related fire/smoke/flame injuries in Victoria, to better understand the differences between the occupational versus community patient profile.
Methods
This study entails a retrospective analysis of population-based administrative hospital data from Victoria, Australia.
Data sources
This study used data from the Victorian Emergency Minimum Dataset (VEMD), provided to the Victorian Injury Surveillance Unit by the Victorian Government Department of Health. The VEMD is a collection of clinical data from all patients presenting to Victorian public hospitals with a designated, 24-hour ED. To determine work-related rates per employed persons, labour force statistics were sourced from the Australian Bureau of Statistics.16 The numbers of employed persons in Victoria in 2003–2022 were used as the denominator for calculating work-related injury rates. To determine non-work-related rates per population, population statistics were sourced from the Australian Bureau of Statistics.17 The numbers of Victorian residents aged 15–74 years were used as the denominator for calculating non-work-related injury rates.
Sample
The study sample was selected as ED presentations between 1 January 2003 and 31 December 2021 (by arrival date), limiting the analyses to patients aged between 15 and 74 years. The sample was further limited to ED presentations with an injury caused by fire, flames and smoke (coded as 14 in the injury cause variable of the VEMD injury surveillance suite). ED presentations were also included based on analysis of the narrative (free text) for mention of fire, flames and smoke through a search using keywords and spelling variations, followed by a manual review of the resulting case selection. ED presentations that were classified as planned return visits and prearranged admissions were excluded.
Variables
Work-relatedness was determined from the activity when the injured variable in the VEMD (where the activity was coded as working for income) and/or the compensable status (where status was recorded as WorkSafe Victoria, indicating workers’ compensation).
Ten-year age groups were created based on the 5-year age bands provided in the source data. Season was coded based on the presentation month. Extreme bushfire season was coded as ED presentations that occurred between 7 February and 14 March 2009 (Black Saturday bushfires) or between 1 November 2019 and 29 February 2020 (Black Summer bushfires). Place of occurrence was determined from the injury surveillance place variable in the VEMD. Geographical region was determined from the area of residence of the patient and coded as Melbourne metropolitan area, regional/rural Victoria or interstate/overseas/unknown. As measures of urgency and injury severity, the triage category was considered, alongside departure status (returning to usual residence outcomes were considered a non-severe outcome and transfer or admission outcomes were denoted as of greater severity).
Statistical methods
Work-related injury rates were calculated as the number of work-related ED presentations per 100 000 employed persons in Victoria, for each year. Non-work-related injury rates were calculated as the number of non-work-related ED presentations per 100 000 population in Victoria for each year. The time trend for rates during the study period was determined using Poisson regression modelling of cases as a function of time (in years), with the log of the number of employed people (work-related model) or general population (non-work-related model) as a model offset. The Poisson models scaled deviance was forced to equal 1. The time trend for the proportion of cases that were work-related during the study period was determined using logistic regression modelling of work-relatedness as a function of time (in years).
Descriptive statistics were computed for presenting fire/flame/smoke-related injury details and patient profile in work-related and non-work-related settings. To determine the differences in patient profile, setting and severity in work-related versus non-work-related settings, logistic regression modelling was used, where work-relatedness was the dependent variable, and each variable was modelled separately, with and without adjustment for age group and sex. Statistical analyses were carried out using SAS V.9.4.
Results
Between 2003 and 2021, there were 11 838 ED presentations for injuries due to fire/flame/smoke exposure, 1864 (15.7%) of which were recorded as work-related and 9974 (84.3%) as non-work-related. This corresponds with an average annual work-related fire/flame/smoke injury rate of 3.43 ED presentations per 100 000 employed persons in Victoria, Australia. The average annual non-work-related fire/flame/smoke injury rate was 12.3 ED presentations per 100 000 population (aged 15–74 years). Non-work-related fire/flame/smoke injury trends over time are shown in figure 1 and work-related rates are shown in figure 2. The rates in non-work-related ED presentations for fire/flame/smoke injuries increased by 1.05% (95% CI 0.53 to 1.58) (p<0.0001) annually during the study period. In contrast, the rates in work-related ED presentations decreased at a rate of −2.01% (95% CI −2.84 to −1.17) (p<0.0001) annually.
Patient demographics, setting and severity of fire/flame/smoke related injury ED presentations are presented in tables 1 and 2. Fire/flame/smoke related injuries in work-related settings (vs non-work-related settings) were less common in people aged 55 years and above, in females, and those living in regional Victoria: these groups were relatively under-represented in the work-related cases. Work-relatedness was relatively less common in more recent years compared with the 2003–2007 time period. Additional crude modelling was conducted with time (years) as a continuous variable to statistically test time trends in work-relatedness as proportion of all cases. This demonstrated a statistically significant decrease over time in the proportion of fire/flame/smoke injury ED presentations that were work-related: ‘year’ had an OR of 0.97 (0.96 to 0.98), p<0.0001 in ED presentation modelling.
The results in tables 1 and 2 show that work-related cases were associated with summer (vs winter). Work-related cases were not statistically significantly related to extreme bushfire periods. Work-relatedness was associated with the place of occurrence recorded as a trade or service area or industrial or construction area. There was an association between work-relatedness and lower severity of the injury, as indicated by the negative association between work-relatedness and ‘resuscitation’ and ‘emergency’ (vs ‘semi-urgent’) triage status, as well as by the negative association with admission to ward and transfer (vs returning to usual residence) in the departure status. Departure before treatment was completed, which could signify less severe injury, was negatively associated with work-relatedness: this contrasts the overall pattern of lower injury severity associated with work-related injuries. It should be noted that in the modelling, triage and departure status are included as proxies for severity of injury, whereas all other model variables can be interpreted as predictors of work-relatedness of injury.
Discussion
The findings in this study demonstrated that hospital-treated occupational injuries related to fire, flame and/or smoke decreased in Victoria during the 19-year study period, while non-work-related rates increased. Work-relatedness was associated with males; ages 25–54 years; metropolitan residence; occurrence in summer months, and relatively less severe injuries.
These findings are novel, as previous studies have not focused on all injuries caused by fire, flame and smoke (but instead, focused on burn injuries, for example), either general or work-related. The observed increase in rates of non-work-related fire/flame/smoke injuries appears to be inconsistent with the slight decrease in burn injury rates reported by Cleland et al, based on ED presentations and hospital admissions in Victoria 2008–2017.15 The burn injuries in their study, however, were caused by fire/flame/smoke in less than 10% of the unintentional burn-related ED cases, while the remainder were caused by contact with heat/hot substances.15 It can, therefore, be surmised that fire/flame/smoke-related injuries are distinct in patterns and trends and cannot be equated to burns injuries: while fire/flame/smoke exposure commonly results in burns, the majority of burns injuries are not caused by fire/flame/smoke.
The reported patient demographics associated with occupational (vs other) fire/flame/smoke injuries were: working age (as expected), male sex and residence in metropolitan Victoria. The over-representation of males in unintentional burn injuries overall is aligned with other reports of burn injury demographics in Victoria,15 and male over-representation in occupational burn injuries has also been reported previously.4 12 14 18 19 The association between work-related fire/flame/smoke-related injury in males could be partly attributed to the over-representation of males in occupations that are likely to be exposed to fire/flame/smoke, including firefighters.
Under-representation of work-related cases in fire/flame/smoke injury among those residing in regional Victoria is unexpected, as occupational (vs other) exposure to fire is likely to be relatively high in regional Victoria due to planned burning and other controlled (as well as uncontrolled) outdoor fires. This suggests that exposure to fire and smoke in regional settings is not necessarily in the line of work: people residing in regional areas may be more likely to experience non-work-related fire/flame/smoke exposure. Alternatively, work-relatedness could be inadequately reported (which can happen, eg, in family-run businesses) or inadequately recorded. In a US study comparing self-reported injury with workers’ compensation claim filing, agriculture/forestry/fishing ranked high in self-reported work-related injury/illness but relatively low in workers’ compensation claim filing.20 Work-relatedness could be under-reported in certain industries and occupations in our study, as work-relatedness was in part determined from workers’ compensation payer status in the ED data. Further research on fire/flame/smoke injury in farm settings is warranted, as farms are a setting where work-relatedness of injuries can be difficult to determine; this will require project-specific data collection beyond administrative claims and health services records.
The findings of this study suggest that injuries were more severe in non-work-related cases, evidenced from triage status and subsequent admission/transfer rates. This pattern could, in part, be due to a lower threshold for healthcare seeking in work-related versus other settings: plausible drivers for this are OHS protocols in the workplace, organisational culture, and/or the need to seek workers’ compensation for workplace injuries. This could explain the observed negative association between departure before treatment was completed and work-relatedness, which could be an indicator of a stronger motivation to receive medical advice. An alternative explanation of this observation could be that work-related injuries were less severe: however, this is unlikely as the other injury severity proxies do not support this. A similar pattern was observed in an Australian and New Zealand burn centre study comparing work-related versus non-work-related burns: the latter had double the proportion of burn total body surface area ≥50%, indicating more severe injuries.4 Potentially, higher rates of cooling interventions applied at the scene in workplace settings, which was also reported in their study, may contribute to better outcomes in this group. Increased availability of personal protective equipment for tasks involving fire risks may also have contributed. In general, workplace injury prevention protocols and practices are not only designed to prevent fire/flame/smoke injuries but also to lower the severity of burn injuries when they occur, for example, through the use of sprinkler systems and fire blankets (to contain fire); safety showers and eye-wash stations (to limit burn injuries); and immediate access to first aid supplies (for immediate burn treatment). These factors all point to the workplace as a better-regulated environment in which fire/flame/smoke-related injuries can be prevented and controlled: ongoing monitoring and feedback play a key role in this.
To understand the observed decrease in occupational fire/flame/smoke injuries, the study findings should be considered in the context of the heterogeneous circumstances of these injuries, which are not limited to occupational exposure through firefighting or other emergency response work. Although bushfires and, in particular, extreme bushfires in Australia have increased in frequency over the last two decades,3 the direct impact of this on fire/flame/smoke-related injuries cannot be precisely measured in hospital data, which also includes other causes of fire/flame/smoke injuries, such as house fires and controlled fires. Even if we were to limit case selection to injuries caused by uncontrolled fires only, the sample would not be directly reflective of the impact of bushfires. The majority of house fires are due to cooking, improper use of appliances and heating, and incorrect use of powerboards21 and not bushfires. The overall decrease in work-related fire/flame/smoke injury rate observed in this study is therefore unlikely to be related to bushfire trends but could be attributed to improved occupational health and safety regulation and compliance; for example, through mandatory use of smoke detectors and routine audits of electrical appliance safety in the workplace. Improved workplace safety would also explain why occupational injuries decreased while non-work-related rates in the general population increased. Smoking bans in the workplace may have also contributed positively: smoking was banned in enclosed workplaces in Victoria in March 2006 (within the study period).
Occupational fire/flame/smoke-related injuries were not statistically significantly associated with extreme bushfire periods (ie, the Victorian Black Saturday and Black Summer periods) in the ED presentations data. Exposure in occupational (vs non-work-related) settings is likely to be disproportionately affected by bushfires, and therefore, the lack of association in the ED presentations data is surprising. If we were to be able to limit case selection to fire/flame/smoke exposure related to bushfires, we would be better placed to determine the impact on the general population and the workforce; this again points to the lack of specificity in the data and the inherent limitations in the utility of hospital data for measuring the health impacts of bushfires. Notably, an association between summer months and occupational (vs general) fire/flame/smoke injuries was observed, but may not be reflective of bushfires alone as this also includes other causes such as scheduled burns on farmland (although these tend to take place in Autumn and Spring) and summer-specific work-related controlled fires.
Although this study has brought to light some important findings, there are limitations that need to be acknowledged. Work-related fire/flame/smoke exposure is heterogeneous and administrative hospital data is not able to differentiate between injuries sustained while firefighting, working with flames or smoke in a commercial kitchen, or while conducting controlled burning on farmland. Although useful for monitoring population trends, capturing patient profiles, calculating the burden of injury and tracking outcomes over time, the data is very limited in its utility for providing industry or occupation-specific information that can be used to guide and improve primary and secondary prevention in the workplace. Relevant information that is routinely captured in hospital data was in some cases of poor quality, such as place of occurrence, which was not specified in 22% of work-related cases. To understand the impact of (1) occupational fire/flame/smoke exposure on health and long-term outcomes, (2) the impact of bushfires specifically and (3) health impacts of fire/flame/smoke exposure in first responders, hospital/ED data will need to be linked with other data sources to capture occupational and environmental exposure. Some examples would be linkage with workers’ compensation claims; linkage with labour force surveys; and linkage (by geocoding and time) with meteorological data. To determine the factors driving non-work-related fire/flame/smoke injury rates, other data sources such as ambulance data coronial reports could be used to distinguish sources of exposure (such as bushfire, housefire, controlled burns, firepit, gas stovetops) and use this information to inform prevention. Various ambient temperatures from extreme heat, cold and rainfall could directly and indirectly impact occupational and non-occupational injury rates. This impact is modified by behaviour, which affects exposure: when extreme weather is predicted, people adjust their plans regarding leisure activities and outside work. Studying the impact of weather conditions on occupational injury rates will require a complex study design that takes (outdoor) exposure into account; this is outside the scope of the current study.
In conclusion, this study demonstrates that work-related fire/flame/smoke exposure resulting in hospital-treated injury decreased over the past two decades in Victoria, while non-work-related fire/flame/smoke-related injury rates increased. The decrease in occupational fire/flame/smoke injury rates could be reflective of improved safety, prevention and/or early intervention in the workplace. Hospital data, however, cannot be used to distinguish work-related exposure by occupation or industry and, therefore, is of limited utility in informing specific workplace safety policies and practices. Harm resulting from occupational exposure to bushfires specifically also cannot be established based on hospital data. Linkage of hospital data with other data sources to determine occupational setting and specific environmental exposure is recommended to inform preventive measures that need to be taken to protect workers from harm caused by bushfires and other fire/flame/smoke exposure in Australia.
Data availability statement
Data may be obtained from a third party and are not publicly available.
Ethics statements
Patient consent for publication
Ethics approval
The Victorian Injury Surveillance Unit (VISU) has obtained ethical approval from the Monash University Human Ethical Committee (MUHREC) (project ID 21427) for the custody, use and dissemination of VEMD data.
Footnotes
X @kwb_doc
Contributors JB-G: conceptualisation, data curation, analysis, interpretation, writing–original draft. WW: writing–review and editing. KW-B: writing–review and editing. JB-G, KW-B, WW: responsible for the overall content as the guarantor.
Funding This study was partly supported by a National Health and Medical Research Council (NHMRC) grant (application 1201353).
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.