Articles

Smoke health costs and the calculus for wildfires fuel

management: a modelling study
Nicolas Borchers-Arriagada, David M J S Bowman, Owen Price, Andrew J Palmer, Stephanie Samson, Hamish Clarke, Gonzalo Sepulveda,
Fay H Johnston

Summary Lancet Planet Health 2021;

Background Smoke from uncontrolled wildfires and deliberately set prescribed burns has the potential to produce 5: e608–19

substantial population exposure to fine particulate matter (PM2·5). We aimed to estimate historical health costs Menzies Institute for Medical
Research
attributable to smoke-related PM₂·₅ from all landscape fires combined, and the relative contributions from wildfires
(N Borchers-Arriagada MEnv,
and prescribed burns, in New South Wales, Australia. Prof A J Palmer MBBS,
Prof F H Johnston PhD),
Methods We quantified PM2·5 from all landscape fire smoke (LFS) and estimated the attributable health burden and New South Wales Bushfire Risk
Management Research Hub
daily health costs between July 1, 2000, and June 30, 2020, for all of New South Wales and by smaller geographical
(N Borchers-Arriagada),
regions. We combined these results with a spatial database of landscape fires to estimate the relative total and per and School of Natural Sciences
hectare health costs attributable to PM2·5 from wildfire smoke (WFS) and prescribed burning smoke (PBS). (Prof D M J S Bowman PhD),
University of Tasmania,
Hobart, TAS, Australia;
Findings We estimated health costs of AU$ 2013 million (95% CI 718–3354; calculated with the 2018 value of the New South Wales Bushfire Risk
AU$). $1653 million (82·1%) of costs were attributable to WFS and $361 million (17·9%) to PBS. The per hectare Management Research Hub
health cost was of $105 for all LFS days ($104 for WFS and $477 for PBS). In sensitivity analyses, the per hectare costs (O Price PhD, S Samson MSc,
associated with PBS was consistently higher than for WFS under a range of different scenarios. H Clarke PhD), and Centre for
Environmental Risk
Management of Bushfires
Interpretation WFS and PBS produce substantial health costs. Total health costs are higher for WFS, but per hectare (O Price, H Clarke), University of
costs are higher for PBS. This should be considered when assessing the trade-offs between prescribed burns and Wollongong, Wollongong,
wildfires. NSW, Australia; Centre for
Health Policy, School of
Population and Global Health,
Funding None. The University of Melbourne,
Melbourne, VIC, Australia
Copyright © 2021 The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY-NC-ND (Prof A J Palmer); Faculty of
Information Technology,
4.0 license. Monash University, Melbourne,
VIC, Australia (G Sepulveda MSc)
Introduction wider range of compounds and more severe pollution Correspondence to:
Since 2017 different regions in the world, including than more intense flaming combustion. PM2·5 is accepted Prof Fay H Johnston, Menzies
Australia, Canada, the USA, Brazil, and parts of Europe, as the most appropriate pollutant for the assessment of Institute for Medical Research,
University of Tasmania, Hobart,
have been affected by unprecedented wildfires (also the health effects of fire smoke, but it is acknowledged TAS 7000, Australia
called bushfires, forest fires, or wildland fires).1,2 Wildfires that analysis of PM2·5 alone will not capture the full range [email protected]
can have large effects on society, including injuries and of health harms caused by the entire smoke mixture.6
loss of life, destruction of homes and other infrastructure, Prescribed burning, also referred to as planned
disrupted productivity and water supplies, and pollution burning or hazard reduction burning, is the planned
of the air by smoke.3 Wildfire smoke is a mixture of application of fire to a landscape. The generic term,
airborne particles and gases, many of which are harmful landscape fire, encompasses all fire types including
to health. At a population level, the most important and wildfires and prescribed burns. Prescribed burning is
best studied pollutant is fine particulate matter (PM2·5), commonly used to modify fuel loads to reduce the risks
which has a similar spectrum of health effects to the well associated with future wildfires by making them more
characterised PM2·5 from sources such as industrial and controllable.7,8 Prescribed burning is also done for
traffic emissions.4 PM2·5 from wildfire smoke is asso­ cultural, ecological, and silvicultural purposes. Non-
ciated with an increase in a myriad of adverse health out­ burning approaches for fuel management include the
comes, including exacerbations of asthma and chronic manual thinning of woody fuels, and the mowing or
obstructive pulmonary disease and increases in cardio­ grazing of grass fuels; however, non-burning approaches
vascular and respiratory hospital admissions and all- are typically applied at local scales, whereas prescribed
cause mortality.4,5 Other pollutants in smoke include burning is applied at the broad landscape scale.
carbon monoxide, nitrogen oxides, ozone, and volatile Like wildfires, smoke produced by prescribed burning
organic compounds; the specific composition of smoke has a negative effect on air quality and health.9 Prescribed
constantly changing due to ongoing chemical reactions fires, although generally smaller in scale than wildfires,
and the accumulation and deposition of particles.4 are frequently done at interfaces between urban and
Smouldering and incomplete combustion produce a rural environments to protect the built environment;

www.thelancet.com/planetary-health Vol 5 September 2021 e608

 Articles

Research in context
Evidence before this study and 2) to estimate daily wood heater smoke, to exclude it from
Current evidence shows that air pollution, and specifically fine the analysis. We then estimated the total and per hectare costs
particulate matter (PM2·5), is associated with a wide range of for WFS and PBS. Our results establish new estimates for the
adverse health effects, including but not restricted to, increased smoke health costs of these two major types of fire. Total costs
mortality and admissions to hospital for cardiovascular and are higher for WFS than for PBS, but on a per hectare basis they
respiratory conditions. Multiple studies have estimated the are higher for PBS. Our sensitivity analyses show that these
health burden attributable to short-term and long-term results remain consistent across different scenarios.
exposure to PM2·5, and its associated health costs, by
Implications of all the available evidence
implementing health impact assessments at local, national,
Our study found similar results compared with other studies
and global scales. Since 2012, several studies have applied these
with respect to the level of landscape fire smoke PM2·5
methods to systematically assess the impacts associated with
exposure: exposure was relatively larger for WFS compared
elevated smoke-related PM2·5 attributable to landscape fire
with PBS. Our results also suggest that smoke-related health
activity, including wildfires and prescribed burns. Some studies
impacts are coupled with area burned in both wildfires and
have shown that prescribed burns, which tend to be smaller in
prescribed burns, with estimated health costs per area burned
duration and extension, might produce large increases in PM2·5
higher for prescribed burns than for wildfires. There is
and public health impacts, when large, populated areas are
considerable variation in health impacts at a regional level,
affected. However, only one study has assessed the relative
and important inter-regional effects of PM2·5 exposure and
health impacts and costs of wildfire smoke (WFS) and
public health outcomes. Current evidence for
prescribed burning smoke (PBS), and this study was restricted
New South Wales shows that to reduce the risk of burning of
to days when the average daily PM2·5 was above 25 µg/m³.
1 hectare due to a wildfire, 3–6 hectares need to be burned
Added value of this study through prescribed fires. Given that our results always yield
This study assessed the historical smoke-related health costs higher per hectare health costs for PBS than for WFS (between
attributable to WFS and PBS, over a 20 year period, in 1·2-times and 4·6-times higher), it would be expected that if
New South Wales, Australia. We combined a spatial fires prescribed fires impacts follow the historical patterns
database with publicly available air quality, health, described in our analysis, then the smoke from these practices
demographic, and economic data to attribute smoke-related might produce health costs at least 3·6 times higher than the
health costs to wildfires and prescribed burns. We included the avoided smoke impacts from wildfires. These results should
use of machine learning methods, particularly random forest, be investigated elsewhere in Australia and other fire prone
for two purposes: 1) to classify days as being affected by WFS or regions in the world, and incorporated into the population
PBS, when more than one fire type was active on the same day risk calculus for wildfire management strategies.

therefore, they usually occur closer to populated areas.10 contribution of wildfires and prescribed burns to smoke-
Although PM2·5 emissions tend to be lower following related health impacts.A study published in 2020
prescribed burning, population exposure to air pollution estimated the health impacts attributable to smoke from
can be substantial because of the closer proximity of the wildfires and prescribed burns for Western Australia,
fires, the more frequent exposure, and weather conditions Australia, but the analysis was restricted to days with
that generally favour accumulation rather than dispersal daily PM2·5 concentrations exceeding the national daily
of smoke.11–13 To better under­ stand the existing trade- standard (25 µg/m³), which excluded a large number of
offs and effectiveness of fire management tools, such days with moderate pollution levels.19
as prescribed burning, the relative public health Between October, 2019, and February, 2020, the eastern
burden attributable to wildfire and prescribed fire smoke seaboard of Australia was affected by an unprecedented
exposure needs to be considered.11,14 An improved under­ fire season, with more than 7 million hectares of
standing of the public health burden is especially Eucalyptus forests burned causing harm to biodiversity
important in the context of a changing climate in which and substantial carbon emissions.24,25 The fires caused
extreme wildfires might increase both in magnitude and harm to human populations through the destruction of
frequency15 and the availability of suitable days for more than 3000 homes, 33 direct deaths, widespread
prescribed burning might be restricted or changed in social trauma, and chronic smoke pollution.24 The state
seasonality.16,17 of New South Wales, Australia, was the most affected,
Health impact assessments have been extensively used and estimates show that the PM2·5 that covered the state
to estimate the health impacts associated with air during this period probably translated into an additional
pollution;18 and in recent years these methods have been 200 deaths, 1600 hospital admissions, and 700 asthma
increasingly applied to assess the health impacts of PM2·5 emergency department visits, with health costs of more
associated with smoke from wildfires and prescribed than AU$1 billion.2,26 This catastrophic event triggered a
burns.2,19–23 Nevertheless, little is known about the relative national inquiry: the Royal Commission into National

e609 www.thelancet.com/planetary-health Vol 5 September 2021

 Articles

Natural Disaster Arrangements.27 The inquiry noted the or PBS; (2) we did a health impact assessment to quantify
importance of fuel management in mitigating fire risk the LFS-attributable health impacts and estimate the daily
through the application of Aboriginal traditional smoke-related health costs; and (3) we combined results
knowledge and use of deliberately set fires (prescribed from the first two steps to estimate the total and per hectare
burns) to reduce fuel loads. Yet, the inquiry highlighted health costs attributable to WFS and PBS. For the health
that smoke generated through these activities might pose impact assessment, we estimated the burden for all-cause
significant harm to the population, particularly for mortality, hospitalisations for circulatory and respiratory
vulnerable groups (eg, children, adults aged 65 years and diseases, and emergency department presen­ tations for
older, and people with pre-existing health conditions).27 asthma-like illnesses, using previously tested methods.2,19–21
Resolving the trade-offs of the benefits of bushfire risk Our main analysis considered the entire state of
reduction through intentional burning against health New South Wales as a single spatial unit, but we also did a
costs remains a major science and policy challenge, regional analysis, which considered grouping fires and
which accordingly demands economic analyses of costs health impacts according to their location using the eight
and benefits across different sectors of the economy. geographical administration branches defined by the
New South Wales is an ideal setting for such an analysis NPWS (appendix p 3). See Online for appendix
because of the flammable native Eucalyptus forests close
to many large population centres and a longstanding Procedures
programme of prescribed burning to manage the risks We used the NPWS Fire History database,29 and
associated with severe wildfires. In the Sydney region considered fires within New South Wales that were active
about 1% of forests get treated with prescribed burns during the study period (July 1, 2000, to June 30, 2020).
each year, often in close proximity to large populations;10 Information in this database included fire season
additionally an average of 4% of forests are burnt by (12 consecutive months between July and June), fire type
wildfire per year.28 Previous studies in New South Wales (prescribed burn or wildfire), start date, end date, and
have found that to effectively reduce the area burned by area burned (appendix p 35). Every fire is given a unique
wildfires an area about three-times as large needs to identifier in the NPWS database. We identified duplicate
be burned through prescribed fires.28 Furthermore, fires, based on the NPWS identifiers, and wrong or
New South Wales is also an ideal setting because it has a missing dates, and used the following procedure to
reliable fire history, well developed air quality monitoring estimate the daily area burned for each fire. First, we
network, and regularly updated health statistical infor­ identified fires that had missing start and end dates and
mation that provide the necessary data for such an those with complete start and end dates. Second, for fires
analysis. In this study, we aimed to estimate the health with complete start and end dates, we estimated the daily
costs of PM2·5 attributable to wildfire smoke (WFS) and area burned for each fire. Daily burned areas were
prescribed burning smoke (PBS) in New South Wales, calculated as the total area burned divided by the number
and to estimate a per hectare health cost indicator to help of days the fire was active (from start day to end day).
incorporate the potential fire smoke-related health Third, for each fire season and fire type, we aggregated
impacts of fire management practices. the daily area burned for fires with complete start and
end dates (identified in step two) and estimated the daily
Methods area burned distribution. Finally, for fires with missing
In this modelling study, we applied a quantitative health start and end dates, we assumed that their total area
impact assessment framework to estimate the health costs burned had the same temporal distribution as those in
of PM2·5 attributable to WFS and PBS in New South Wales, step three (ie, we assumed that fires with missing dates
since July, 2000. Additionally, we estimated a per hectare had the same temporal distribution as those with
health cost indicator, to help incorporate the potential complete dates).
fire smoke-related health impacts associated with fire Once we had an estimation of the daily area burned for
management practices. each fire, we used these results to identify if each day was
We estimated the health burden attributable to landscape dominated by PBS or WFS. For each day, we calculated
fire smoke (LFS) in New South Wales between July 1, 2000, the proportion of area burned by a wildfire (perc ha wf),
and June 30, 2020. We combined a spatial landscape fires and then classified each day according to the likely
database containing detailed character­ isation of all dominant LFS category.
registered fires within the state, with estimated daily health We used a random forest model machine learning
costs attributable to LFS-related PM2·5 (figure 1). The algorithm to predict the dominant LFS category for each
analysis approach used the following three steps: (1) we day. A random forest is a supervised learning algorithm
analysed the National Parks and Wildlife Service (NPWS) that can be trained with observations having known
fires database, cleaned the data, estimated the daily area categories (eg, wildfire or prescribed burn) to predict
burned by each fire type (wildfire or prescribed burn), and these categories on a new set of observations.19,30 First,
applied a machine learning algorithm (random forest) to we identified days that were dominated by wildfires
classify each day across the state as being affected by WFS (perc ha wf ≥0·9) or prescribed burns (perc ha wf ≤0·1).

www.thelancet.com/planetary-health Vol 5 September 2021 e610

 Articles

Analysis of fires database Quantification of health costs

ABS data PM2.5 elemental

Gridded population
component
by year
studies
NPWS fire history:
wildfires and
prescribed burns
Spatial Air quality Random forest
interpolation Population exposure monitoring model
(IDW)

Database cleaning
ID LFS affected Exclusion of
days WHS PM2·5

Random
Classification of days
forest Health risk Baseline
by fire type LFS-attributable
model coefficients incidence
health impacts
rates

Unitary health
Area burned by fire type Days by fire type Daily health costs
costs

Costs per area burned

Costs by
fire type
• State-wide analysis
• Regional analysis (NPWS branches)
• Sensitivity analysis
Costs per
area burned

Figure 1: Summary diagram for analysis

ABS: Australian Bureau of Statistics. ID=identifier. IDW=inverse distance weighting. LFS=landscape fire smoke. NPWS=National Parks and Wildlife Service. WHS=wood
heater smoke.

We trained a random forest model with this data, and medium sized geographical area with average population
used this model to predict the LFS category on other days of 10 000 people) and year; and exposure response
(0·1< perc ha wf <0·9). This model used the following functions or health risk coefficients (β₀) for short-term
explanatory variables: year, month, week, weekday, fire exposure to PM2·5. For unit costs (UnitCosts0) we used the
season, area burned, mean temperature, minimum value of a statistical life (VSL) for mortality, and Australia-
temperature, maximum temper­ ature, rainfall, solar specific average costs for morbidity, adjusted to the value
exposure, vapour pressure, and relative humidity. With of the Australian dollar in 2018 (appendix pp 7–8).
this model, we achieved an accuracy score of over 96% LFS-related PM2·5 exposure (ΔCd) was calculated as the
(appendix pp 4–6). difference between the estimated daily PM2·5 for days
In the health impact assessment we estimated health identified as being affected by LFS and the long-term
costs for each day (d) using the following equation: mean daily PM2·5 by SA2 and month, excluding the
contribution of wood heater smoke. To estimate exposure
Costsd = ∑ [IR0 × Popd × (eβ0 × ∆Cd–1)] × UnitCost0 for each SA2, we first used station level PM2·5 data, and
0 applied an inverse distance weighting method to
= ∑ [Cases0 × UnitCost0 interpolate values at mesh block (smallest geographical
0 area defined by the Australian Bureau of Statistics)
centroids, and then calculated a population weighted
We used baseline incidence rates (IR₀) for all-cause PM2·5 for each SA2 (each SA2 is composed of multiple
mortality, circulatory disease, and respiratory disease mesh blocks). We considered a day to be affected by LFS
hospitalisations, and asthma emergency department whenever the daily PM2·5 on any given SA2 was above the
presentations (outcome o), by state and year; exposed long-term (2000–20) 95th percentile for that SA2 month,
population (Popd) by Statistical Area Level 2 (SA2; a method that has already been validated in Australian

e611 www.thelancet.com/planetary-health Vol 5 September 2021

 Articles

settings.2,31,32 Additionally, we excluded 105 days having to LFS were zero when there was no area burned). Area
high particulate matter levels due to the presence of dust burnedft,t represented the total area burned by fire type (ft)
storms between 2002 and 2018.33 The wood heater smoke aggregated for time (t). β1,ft,t represented the regression
mass contribution has been previously estimated for a coefficient or slope (in this context meaning health cost
series of sites across New South Wales for different per area burned, $ per hectare). We estimated this
periods between 1998 and 2015 (appendix pp 9–15). We relationship for all LFS days, WFS days, and PBS days.
used this data, to train a random forest model, and We also considered alter­native fire day type classifications
estimate the contribution of wood heater smoke to daily and temporal grouping variables as detailed in the
PM2·5 across the study period. We fitted the random sensitivity analyses.
forest model with the following explanatory variables: To estimate health costs per hectare for the regional
daily PM2·5, day of the week, longitude, latitude, solar analysis, we aggregated the 8 NPWS branches into
exposure, mean temper­ ature, year, daily change in 5 groups: (1) Greater Sydney and Blue Mountains,
temperature (maximum temperature minus minimum (2) Hunter Central Coast, (3) Northern branches (North
temperature), minimum temperature, day of the year, Coast and Northern Inland branches), (4) Southern
relative humidity, maximum temperature, month, branches (South Coast and Southern Ranges branches),
heating degree days, day, and rainfall. and (5) West branch. Additionally, we looked at corre­
To assess the association between health cost and lation matrices and did visual analyses to assess for inter-
burned area, we estimated an average cost per hectare by regional effects, that is, smoke produced in one region
fire type, by fitting a linear regression with the following affecting the population health of other regions. We
shape, and estimated the goodness of fit (measured estimated the Pearson correlation coefficient using
by R²): R (4.0.2), between area burned and health costs between
regions.
Health costft,t = β0 + β1,ft,t × Area burnedft,t
Sensitivity analyses
Health costft,t was the total estimated health cost for fire We considered the baseline incidence rates and the
type (ft) aggregated for time (t); β₀ represented the health risk coefficients, with 95% CI calculated for both
intercept (we assumed β₀=0; ie, health costs attributable variables. Because of the potential for misclass­ification

A Prescribed burns
1400 Number of fires
5 million
Area burned (hectares) Area burned (hectares) per fire season
1200
Number of fires per fire season

4 million
1000

800 3 million

600
2 million
400
1 million
200

0 0

B Wildfire
1400 5 million
Area burned (hectares) per fire season

1200
Number of fires per fire season

4 million
1000

800 3 million

600
2 million
400
1 million
200

0 0
1

02

04

05

06

0
–0

–1
–0

–1
–0

–1

–1

–1
–1

–2
–1
–0

–0

–1

–1

–1
–

–
–

10

16
11

12

14
00

09
06

13

15

17

18
01

02

04

19
03

05

07

08

20

20
20

20

20
20

20

20
20

20

20
20
20

20
20

20
20

20

20

20

Fire season

Figure 2: Number of fires and area burned by fire season

www.thelancet.com/planetary-health Vol 5 September 2021 e612

 Articles

bias of LFS day types, we tested two alternative

LFS days AU$ million (95% CI) Proportion Mean daily cost
of costs (AU$/LFS day) classifications of days: (1) season, because certain seasons
are strongly dominated by either WFS (summer), PBS
Total 2037 (100·0%) 2013·4 (718·2–3353·9) 100·0% 988 410
(autumn and spring), or wood heater smoke (winter);
By fire type (random forest)
and (2) daily area burned by wildfires and prescribed
PB 826 (40·5%) 360·8 (129·8–596·0) 17·9% 436 848
burns (simple categorical classification). For the simple
WF 1211 (59·5%) 1652·6 (588·4–2757·9) 82·1% 1 364 620
categorical classification, we assigned a category to each
By fire type (random forest; only fires with correct dates)
day according to the estimated proportion of area burned
PB 758 (37·2%) 337·2 (121·3–557·0) 16·9% 444 830
by a wildfire and by a prescribed burn: (1) wildfire 0–10%,
WF 1175 (57·7%) 1659·0 (590·7–2768·5) 83·1% 1 411 873
prescribed burn 90–100%; (2) wildfire 10–50%, pre­
By fire type (categorical)
scribed burn 50–90%; (3) wildfire 50–90%, prescribed
WF 0–10%, PB 90–100% 615 (30·2%) 255·0 (91·7–421·2) 12·7% 414 630 burn 10–50%; and (4) wildfire 90–100%, prescribed
WF 10–50%, PB 50–90% 208 (10·2%) 104·1 (37·5–172·0) 5·2% 500 629 burn 0–10%. For example, a simple category of wild­
WF 50–90%, PB 10–50% 212 (10·4%) 80·8 (29·1–133·4) 4·0% 381 246 fire 0–10% and prescribed burn 90–100% would be
WF 90–100%, PB 0–10% 1002 (49·2%) 1573·4 (559·9–2627·3) 78·1% 1 570 298 interpreted as having less than 10% of total daily area
By season burned by wildfires and more than 90% of the daily area
Summer 383 (18·8%) 1016·4 (360·7–1701·6) 50·5% 2 653 742 burned by prescribed burns.
Autumn 585 (28·7%) 291·9 (105·0–482·3) 14·5% 499 002 Because of the possible errors associated with wrong or
Winter 566 (27·8%) 146·7 (52·9–241·9) 7·3% 259 186 missing dates in the fires database, and the presence of
Spring 503 (24·7%) 558·4 (199·6–928·0) 27·7% 1 110 125 time lags between smoke production, exposure, and
By fire season onset of symptoms, we varied the time grouping variable
2000–01 43 (2·1%) 49·1 (17·7–81·2) 2·4% 1 142 848 to account for multiple days (eg, regression and
2001–02 79 (3·9%) 229·6 (81·7–384·0) 11·4% 2 906 702 correlation results were aggregated by day, week, month,
2002–03 153 (7·5%) 287·1 (103·0–476·2) 14·3% 1 876 430 year, and fire season), and we included a sensitivity
2003–04 116 (5·7%) 105·1 (37·9–173·5) 5·2% 906 338 analysis that assessed the effect of excluding fires with
2004–05 80 (3·9%) 42·5 (15·3–70·2) 2·1% 531 352 wrong or missing dates.
2005–06 56 (2·7%) 16·7 (6·0–27·6) 0·8% 298 729
2006–07 86 (4·2%) 67·8 (24·4–112·4) 3·4% 788 913 Role of the funding source
2007–08 49 (2·4%) 6·1 (2·2–10·1) 0·3% 124 691 The funder of the study had no role in study design, data
2008–09 66 (3·2%) 26·3 (9·5–43·6) 1·3% 399 175 collection, data analysis, data interpretation, or writing of
2009–10 87 (4·3%) 36·4 (13·1–60·2) 1·8% 418 193 the report.
2010–11 52 (2·6%) 7·6 (2·7–12·6) 0·4% 146 349
2011–12 78 (3·8%) 10·9 (3·9–18·0) 0·5% 139 490 Results
2012–13 88 (4·3%) 16·9 (6·1–27·9) 0·8% 191 913 A total of 15 536 fires (9638 [62·0%] wildfires) burning
2013–14 94 (4·6%) 75·4 (27·0–124·8) 3·7% 801 964 11 940 699 hectares (10 767 390 [90·2%] by wildfires)
2014–15 74 (3·6%) 21·7 (7·8–35·7) 1·1% 292 579
were active between July 1, 2000, and June 30, 2020
2015–16 87 (4·3%) 67·7 (24·3–112·1) 3·4% 778 197
(appendix p 18). 5001 (32·2%) fires, which burned
2016–17 75 (3·7%) 15·1 (5·4–25·0) 0·8% 201 503
2 881 092 (24·1%) hectares, had wrong or missing dates
2017–18 101 (5·0%) 47·5 (17·1–78·4) 2·4% 469 831
in the NPWS database. Summer and spring periods
accounted for most (11 621 [74·8%]) fires, which
2018–19 246 (12·1%) 60·5 (21·7–100·0) 3·0% 246 024
accounted for 10 258 381 (85·9%) hectares of the area
2019–20 327 (16·1%) 823·3 (291·4–1380·6) 40·9% 2 517 625
burned. The 2002–03 fire season had the highest
By branch
number of all fires (1592 [10·2%]), but the 2019–20 fire
Blue Mountains branch 748 (36·7%) 68·3 (24·3–114·1) 3·4% 91 274
season, which was dominated by unprecedented
Interbranch impacts 1613 (79·2%) 676·8 (242·0–1125·1) 33·6% 419 567
wildfires (figure 2) had the largest area burned
Greater Sydney branch 323 (15·9%) 818·2 (291·8–1362·4) 40·6% 2 533 021
(5 million hectares [41·9%]) and the highest mean area
Hunter Central Coast 645 (31·7%) 235·3 (84·1–391·2) 11·7% 364 737
branch
burned of 6692 hectares per fire (or 15 123 hectares per
North Coast branch 451 (22·1%) 9·7 (3·4–16·5) 0·5% 21 550
fire day). Both number of fires and area burnt were
Northern Inland branch 728 (35·7%) 41·6 (14·7–70·0) 2·1% 57 189
highest in the NPWS Northern Inland branch, followed
by the North Coast branch in the number of fires and
South Coast branch 463 (22·7%) 108·6 (38·7–181·3) 5·4% 234 612
by the Blue Mountains branch in terms of area burned.
Southern Ranges branch 343 (16·8%) 45·1 (15·7–76·9) 2·2% 131 343
Population weighted PM2·5 exposure across all fire
Data are n (%), unless otherwise stated. LFS= landscape fire smoke. PB=prescribed burns. WF=wildfires. seasons was higher for LFS days compared with all days,
with an estimated daily average of 6·8 µg/m³ (SD 9·2)
Table 1: Summary statistics of estimated LFS-related health costs
attributable to LFS (appendix p 19). On average, LFS-
related PM2·5 was higher for WFS during summer and
spring and was highest in the 2001–02, 2002–03, and

e613 www.thelancet.com/planetary-health Vol 5 September 2021

 Articles

Prescribed burns Wildfire

AU$/100 000 people

per fire season
No data
<$0·30 million
$0·30–0·60 million
$0·61–1·00 million
$1·01–2·00 million
$2·01–3·00 million
$3·01–4·00 million
>$4·00 million

Figure 3: Map of average estimated health costs by statistical area level 2 and fire type
Health costs were measured as AU$ per 100 000 people per fire season.

2019–20 fire seasons. There was also spatial variation, For New South Wales as a whole and when aggregating
with some NPWS branches such as the North Coast, the health costs and area burned by fire season, we estimated
Southern Ranges and the Greater Sydney, having higher an average cost of $105 per hectare for all LFS days
LFS-related PM2·5 than the rest. The long-term population ($104 per hectare for WFS and $477 per hectare for PBS),
weighted annual mean PM2·5 (all sources) for NSW was but these results were highly influenced by the 2002–03
around 6·6 µg/m³. On average, wildfires contributed and 2019–20 fire seasons (figure 4). When these two fire
0·47 µg/m³ to annual PM2·5 exposure per year. PM2·5 seasons were excluded the difference between WFS and
exposure was 0·94 µg/m³ in the 2001–02, 1·18 µg/m³ in PBS is reduced substantially, but our estimates still found
the 2002–03, and 3·97 µg/m³ in the 2019–20 fire seasons. that the average smoke-related health costs of prescribed
yThe 2019–20 fire season was by far the most burning was 1·4-times higher than that of wildfires
burdensome ($823·3 million [95% CI 291·4–1380·6]; ($478 per hectare for prescribed burns vs $344 per hectare
40·9% of total costs) followed by the 2002–03 fire season for wildfires). Although these results changed when
($287·1 million [103·0–476·2]; 14·3% of total costs). As excluding fires with incorrect dates the differences were
expected, the two most affected regions had the largest minor, with average health costs of prescribed burns
populations in New South Wales, the Greater Sydney being 4·0-times higher than that of wildfires considering
branch ($818·2 million [291·8–1362·4]) and the Hunter all fire seasons, and 1·2-times higher when excluding the
Central Coast branch ($235·3 million [84·1–391·2]). 2002–03 and 2019–20 fire seasons (table 2). When
$676·8 million (33·6% [242·0–1125·1]) were inter-branch increasing or changing the grouping variable (year,
effects (ie, fires in some regions produced smoke that month, week) the goodness of fit (measured by the R²
affected other, probably adjacent, regions). Health costs squared) was reduced, but the difference in costs was
(normalised per 100 000 people) and fire season varied sustained. The average cost of days classified as wild­
across New South Wales and by fire type; costs were fire 0–10%, prescribed burns 90–100% versus wild­
historically higher for wildfires, with an estimated fire 90–100%, prescribed burns 0–10% (when excluding
$1·6 million per 100 000 people and fire season, compared the 2002–03 and 2019–20 fire seasons) was cost 1·2-times
with prescribed burns, $0·33 million per 100 000 people more ($440 per hectare for wildfire 0–10%, prescribed
and fire season (figure 3). The highest wildfire costs were burns 90–100% vs $356 per hectare for wildfire 90–100%,
seen in the Southern Ranges ($2·68 million per prescribed burns 0–10%). All fires in the autumn cost
100 000 people and fire season) and the Northern Inland 1·1-times more than all fires in the summer ($503 per
branches ($2·47 million per 100 000 people and fire hectare in the Autumn vs $455 per hectare in the summer;
season). The highest costs for prescribed burning were appendix pp 27–28).
reported in Northern Inland ($360 000 per 100 000 people We found that fire activity, PM2·5 exposure, and smoke-
and fire seasons) and the Greater Sydney branch ($341 000 related health effects have different distributions across
per 100 000 people and fire season; appendix p 21). NPWS branches. The highest average costs (when

www.thelancet.com/planetary-health Vol 5 September 2021 e614

 Articles

A B
All LFS days All LFS days
1000 200
y=105·x, R2=0·914 y=370·x, R2=0·797
(AU$ million)

150
Health costs

750
500 100
250 50
0 0
0 2500 5000 7500 0 100 200 300 400

WFS days 200 WFS days

1000
y=104·x, R2=0·938 y=344·x, R2=0·783
(AU$ million)

750 150
Health costs

500 100
250 50
0 0
0 2500 5000 7500 0 100 200 300 400

PBS days PBS days

60 60
(AU$ million)

y=477·x, R2=0·785
Health costs

y=478·x, R2=0·785
40 40
20 20
0 0
0 30 60 90 120 0 30 60 90 120
Area burned (thousand hectares) Area burned (thousand hectares)

C D
All LFS days All LFS days
1000
y=124·x, R =0·902
2
150 y=410·x, R2=0·579
(AU$ million)

750
Health costs

100
500
250 50
0 0
0 2000 4000 6000 0 100 200 300

WFS days WFS days

1000
y=123·x, R2=0·921 150 y=403·x, R2=0·511
(AU$ million)

750
Health costs

500 100

250 50
0 0
0 2000 4000 6000 0 100 200 300

PBS days PBS days

80 80
y=493·x, R2=0·733 y=492·x, R2=0·733
(AU$ million)

60 60
Health costs

40 40
20 20
0 0
0 30 60 90 120 0 30 60 90 120
Area burned (thousand hectares) Area burned (thousand hectares)

Figure 4: Linear relationship between estimated health costs and total area burn by type of LFS day
(A) All LFS days (i), WFS days (ii), and PBS days (iii) across all fire seasons (2000–01 to 2019–20). (B) All LFS days (i), WFS days (ii), and PBS days (iii) across all fire seasons,
excluding the 2002–03 and 2019–20 fire seasons. (C) All LFS days (i), WFS days (ii), and PBS days (iii) across all fire seasons with the correct dates, and D) All LFS days (i),
WFS days (ii), and PBS days (iii) across all fire seasons with the correct dates, excluding 2002–03 and 2019–20 fire seasons. The linear equation is presented in AU$ and
hectares. Shaded areas are 95% CI.

excluding the 2002–03 and 2019–20 fire seasons) occurred For NPWS branches, a cost per area burned was
in the Greater Sydney and Blue Mountains branches ($494 estimated whenever LFS-related health costs were
per hectare) and the Hunter Central Coast branch ($1022 estimated on the same day a fire was occurring in the
per hectare; appendix p 22). For all NPWS branches, same branch. Therefore, it should be noted that these
except for the West branch, average PBS costs were higher results are probably an underestimation because the
than WFS (appendix p 31), with the largest difference inter-branch effects are not included. There were
reported in the Southern branches, where PBS was periods of time when landscape fires occurred in some
7·1-times more expensive than WFS ($321 per hectare for branches of New South Wales and smoke-related health
PBS vs $45 per hectare for WFS; appendix pp 24–31). impacts (health costs) were observed in other branches

e615 www.thelancet.com/planetary-health Vol 5 September 2021

 Articles

(appendix pp 32–34). For example, a high correlation

Variable All fire All fire seasons excluding
was observed between the area burned in the Greater seasons 2002–03 and 2019–20
Sydney and Blue Mountains branches with costs in the
By fire type (random forest)
Hunter Central Coast branch (correlation 0·83),
PB PBS costs (AU$ per hectare) $477 $478
Northern branch (correlation 0·91), Southern branches
WF WFS costs (AU$ per hectare) $104 $344
(correlation 0·95), and West branch (correlation 0·93).
PBS to WFS ratio 4·6 1·4
By fire type (random forest; fires with correct dates)
Discussion
PB PBS costs (AU$ per hectare) $493 $492
To our knowledge, this is the first study to thoroughly
WF WFS costs (AU$ per hectare) $123 $403
assess the relative contribution of PM2·5 associated with
PBS to WFS ratio 4·0 1·2
smoke from wildfires and prescribed burns to health
costs. Previously, studies have focused on assessing the PB=prescribed burns. PBS=prescribed burning smoke. WF=wildfire. WFS=wildfire smoke.
impact of LFS on air quality, the study of specific events,
Table 2: Summary of PBS and WFS health costs per hectare and PBS to WFS cost ratio
or the estimation of the total LFS health burden. We
found that over the past 20 years in New South Wales,
fire history has been dominated by wildfires, accounting association between the percentage of monitor-days
for more than 62% of fire events and more than 90% of exceedances (>35 µg/m³) and area burned for California
area burned. Wildfires occurred mainly during spring and Washington, with exceedance frequency also related
and summer, with high fire severity and number to fire activity across states.35 These results are in line
variability through time, and the 2002–03 and 2019–20 with our findings, with higher PM2·5 exposure levels for
fire seasons being the most severe on record in New wildfires compared with prescribed burns. Our results
South Wales. We estimated these fires caused substantial also suggest that health costs driven by PM2·5 exposure
increases in PM₂·₅ exposure, equivalent to more than the are coupled to area burned in both wildfires and
1 µg/m³ annual average for the 2002–03 and 2019–20 fire prescribed burns, with a considerable contribution of
seasons. We calculated that wildfires account for most of PM2·5 exposure due to inter-branch fire activity (ie, fires
the health costs (82·1%), but that per hectare health costs occurring in some NPWS branches and PM2·5 exposure
were larger for prescribed fires. observed in other branches).
Previous studies have assessed the impact of fires on In Australia, Price and colleagues11 assessed the effect
society, including the smoke-related health burden and of two prescribed burns, with burned areas of 52 hectares
costs.2,20,21 In the USA, 30% of the population has an and 700 hectares. They found that the small burn
estimated annual LFS-related PM2·5 exposure of between produced larger spikes in PM2·5 closer to the burn (about
0·56 µg/m³ and 1·5 µg/m³, with 10% of the population 500 m downwind), whereas the larger event might have
exposed to more than 1·5 µg/m³ LFS-related PM2·5.20,34 In produced important air pollution increases in locations
Canada, a large proportion of the population is exposed up to 14 kms away. The smoke levels from these
to wildfire season, which lasts from May to September; prescribed fires was found to be severe, reaching levels of
average LFS-PM2·5 exposure is reported to be between above 1000 µg/m³, and daily averages of above 200 µg/m³,
0·2 µg/m³ and 1 µg/m³, with exposures of more than far exceeding national air quality standards. The authors
5–10 µg/m³ reported for some years.21 In Australia, we suggest that prescribed fires can probably produce high
estimated an average annual exposure to PM2·5 due to local effects and lower regional level effects, both
wildfires to be 0·47 µg/m³, which increased to 3·97 µg/m³ increasing with fire size. This situation outlined by Price
for the 2019–20 fire season. and colleagues11 has been observed in Australia, with
Other studies have assessed the relative impact of some prescribed burning programmes producing similar
wildfires and prescribed fires on air quality. A systematic large air pollution episodes to those observed for serious
review of studies from the USA found that PM2·5 wildfire events.14
exposure from wildfires was significantly lower than Generally, LFS-related health costs estimated by
prescribed burns, but these findings were attributed to different studies are not comparable because of existing
differences in the methods between studies, because differences in the magnitude of exposed populations,
particulate matter was measured further away for the duration of fires (or studies), and the values used
wildfires and closer for prescribed burns.12 These results to quantify and monetise health impacts and costs.
were not in keeping with our findings or those of other One way to compare health-related costs across different
studies. Jaffe and colleagues35 found that for 2017, contexts is by normalising these values by a constant
maximum daily PM2·5 values were higher for wildfires amount of population, generally 100 000 people
(125–550 µg/m³) compared with prescribed burns (ie, similar to a per capita estimate). In our study,
(29–49 µg/m³) for the top five states for annual area we estimated average LFS-related health costs of
burned, with both fire types having a similar magnitude $1·61 million per 100 000 people fire season for wildfires
of area burned (260–640 thousand hectares per year).35 ($0·33 million per 100 000 people fire season for
Between 2004 and 2018, there seemed to be a positive prescribed burns), which is lower than estimates for

www.thelancet.com/planetary-health Vol 5 September 2021 e616

 Articles

Canada (AU$3 million per 100 000 people) and the USA Additionally, we excluded days potentially affected by
($6·1 million per 100 000 people).20,21 The main dust storms and the estimated contribution of wood
difference between our study and the estimates for the heater use on increased PM2·5 during colder days. If this
USA and Canada probably lies in the magnitude of the was not done, the smoke-related health costs would have
VSL used to monetise mortality. These studies used a been overestimated. Nevertheless, our approach was
VSL of CA$6·5 million (at the 2007 value of the conservative with respect to the identification of LFS days
Canadian dollar) and US$10·1 million (at the 2010 because, by considering only the days with the highest
value of the US dollar), substantially higher than the 5% of PM2·5 exposure (excluding days affected by dust
AU$4·5 million used for this study. storms), we excluded a considerable number of days that
The main strengths of this study relate to the could have been affected by fire smoke but had a
application of a quantitative health impact assessment in moderate level of pollution. Finally, we did several
a setting ideal for this analysis: New South Wales. sensitivity analyses to test the robustness of our results to
New South Wales has forests close to many population the approach we used to characterise the relationship
centres; a longstanding programme of prescribed between LFS attributable health costs and fire activity,
burning; a reliable fire history database, with more than using different fire type classification methods and
20 years of data; a well developed air quality monitoring different temporal levels of aggregation. Across all
network; and up to date demographic and health combinations, we consistently observed that the per
statistical information. hectare health costs for prescribed burns were higher
Like all health and economic modelling studies, this than those for wildfires.
analysis has some inherent assumptions, uncertainties, The historical health burden calculated in this study is
and limitations that are well recognised and have been probably an underestimation of the real impact. We used
discussed elsewhere.36 Our aim was to use consistent and short-term health risk coefficients (lower than those used
accepted methods to compare the relative health costs of for chronic exposure) usually used for acute exposure,
the two types of landscape fires. This required using although our results show that the LFS-related PM2·5
some additional approaches, which have additional exposure translated to relevant annual PM2·5 values, with
specific limitations: the use of administrative fire data more than 30% of days within the study period being
(and its inherent data errors) to identify individual fire affected by LFS. Additionally, our analysis only includes
events, the association of fire type days with identified the effect of PM2·5 on health, but other pollutants—such
LFS-affected days, and the estimation of a relationship as ozone—might also produce a relevant smoke-related
between LFS-attributable health costs and fire activity health burden.37
measured by area burned. The potential errors present in The future will probably bring an increase in frequency
the NPWS fire history database could lead to the wrong and magnitude of wildfires in different regions across the
classification of fire type days and the overestimation or globe with increasing pressure to better manage the
underestimation of area burned for each of these fire inherent risks. Interventions used to reduce wildfire risk
types. We sought to overcome this limitation by doing affect future fuel availability, wildfire occurrence, and
some additional analyses and presenting our results wildfire magnitude. Although prescribed fires might be an
excluding fires with incorrect original dates. As can be effective fire management tool, the burn window for a safe
seen in our analysis, the exclusion of fires with wrong and effective application of prescribed burns might be
dates did not have a substantial effect on the results. reduced or changed in seasonality. Many factors—such as
Additionally, we did our analysis at different aggregate fire intensity, fuel availability, and type and fire duration—
temporal levels (days to years) and classified fire type have an effect on smoke volume, dispersion, and com­
days under alternative categorical variables (simple position and, as a result, on health impacts.14 Nevertheless,
categorical classification based on area burned by fire our results show a high correlation between estimated
type) and season (spring, summer, autumn, winter) that health costs and area burned, and the most interesting and
might be representative of wildfires and prescribed surprising result of this study is that per hectare health
burns, and observed consistency in our results. The costs are higher for prescribed burns than for wildfires.
association of fire type days with identified LFS affected This can be due to various reasons, including the larger
days was one of the most complex methodological magnitude of wildfire surface, proximity to people, or
decisions we had to make for this analysis. If we chose to because prescribed burns tend to occur in weather more
analyse individual fires, we would have been incapable of conducive to produce smoke accumulation.11,12 Prescribed
attributing smoke exposure to many of these fires, burns are usually done at interfaces between rural and
particularly because substantial population exposure due urban environments; the aim of these burns is to protect
to multiple fire events happening across the landscape the built environ­ ment. Therefore, prescribed burns
and not just individual fires could occur. Therefore, probably produce a substantial population exposure to air
instead of trying to identify how specific fire events pollution. Crawford and colleagues13 found that smoke
affected PM2·5 in surrounding locations, we assigned the particulate concentrations on winter days in Sydney, NSW,
most likely smoke source (WFS or PBS) to each day. Australia, were 11-times higher when the atmosphere was

e617 www.thelancet.com/planetary-health Vol 5 September 2021

 Articles

stable than when it was not.13 Stable days are preferred for 4 Cascio WE. Wildland fire smoke and human health.
Sci Total Environ 2018; 624: 586–95.
prescribed burning because the presence of light winds
5 Reid CE, Brauer M, Johnston FH, Jerrett M, Balmes JR, Elliott CT.
reduces the risk of an uncontrolled fire. In New South Critical review of health impacts of wildfire smoke exposure.
Wales, to effectively reduce the area burned by wildfires an Environ Health Perspect 2016; 124: 1334–43.
area about three-times the size needs to be burned through 6 Rice MB, Henderson SB, Lambert AA, et al. Respiratory impacts of
wildland fire smoke: future challenges and policy opportunities.
prescribed fires.28 It is expected that in this setting, the An official American Thoracic Society workshop report.
implementation of a prescribed burning programme that Ann Am Thorac Soc 2021; 18: 921–30.
successfully reduces the area burned by wildfires will 7 Fernandes PM, Botelho HS. A review of prescribed burning
effectiveness in fire hazard reduction. Int J Wildl Fire 2003;
produce an overall increase in smoke-related harms. 12: 117–28.
The health effects of LFS should be adequately 8 Hunter ME, Robles MD. Tamm review: The effects of prescribed
incorporated into the assessment of the trade-offs fire on wildfire regimes and impacts: a framework for comparison.
For Ecol Manage 2020; 475: 118435.
between wildfires and prescribed burns. Alternatives to
9 Haikerwal A, Reisen F, Sim MR, et al. Impact of smoke from
prescribed burning in areas close to human populations— prescribed burning: is it a public health concern?
such as herbivory, the manual removal of fuels, vegetation J Air Waste Manag Assoc 2015; 65: 592–98.
modification, and the creation of so called green 10 Price OF, Bradstock RA. The spatial domain of wildfire risk and
response in the wildland urban interface in Sydney, Australia.
firebreaks38—will probably have major economic benefits, Nat Hazards Earth Syst Sci 2013; 13: 3385–93.
provide similar levels of risk mitigation, and lower smoke- 11 Price OF, Horsey B, Jiang N. Local and regional smoke impacts
related health harm outweighing the additional costs of from prescribed fires. Nat Hazards Earth Syst Sci 2016; 16: 2247–57.
these alternatives compared with prescribed burning. 12 Navarro KM, Schweizer D, Balmes JR, et al. A review of community
smoke exposure from wildfire compared to prescribed fire in the
The likely increase in frequency and intensity of fire United States. Atmosphere 2018; 9: 185.
seasons, coupled with changes in the seasonality and 13 Crawford J, Chambers S, Cohen D, Williams A, Griffiths A,
availability of days with ideal conditions for prescribed Stelcer E. Assessing the impact of atmospheric stability on locally
and remotely sourced aerosols at Richmond, Australia, using
fires, will increase the complexity and risk trade-offs Radon-222. Atmos Environ 2016; 127: 107–17.
associated with prescribed burning programmes in the 14 Williamson GJ, Bowman DMJS, Price OF, Henderson SB,
future. This requires an increase in the extent and Johnston FH. A transdisciplinary approach to understanding the
health effects of wildfire and prescribed fire smoke regimes.
diversity of fuel management programmes, and systems Environ Res Lett 2016; 11: 125009.
for better mitigation of the adverse population health 15 Clarke H, Penman T, Boer M, et al. The proximal drivers of large
impacts of smoke from prescribed burns. fires: a pyrogeographic study. Front Earth Sci 2020; 8: 90.
16 Kupfer JA, Terando AJ, Gao P, Teske C, Hiers JK. Climate change
Contributors projected to reduce prescribed burning opportunities in the
NB-A and FHJ conceived the paper. NB-A contributed to the methods, south-eastern United States. Int J Wildl Fire 2020; published online
conducted the analyses, and drafted the manuscript. FHJ contributed to June 1. https://doi.org/10.1071/WF19198.
the health assessment methods and analyses. DMJSB contributed to the 17 Clarke H, Tran B, Boer MM, Price O, Kenny B, Bradstock R. Climate
fire-related methods and analyses. AJP contributed to the health economic change effects on the frequency, seasonality and interannual
methods and analysis. OP and SS contributed to the fire-related data variability of suitable prescribed burning weather conditions in
collection and analysis. HC contributed to the overall methodological south-eastern Australia. Agric For Meteorol 2019; 271: 148–57.
approach. GS contributed to the statistical and machine learning methods 18 Héroux ME, Anderson HR, Atkinson R, et al. Quantifying the
used. NB-A and GS had access to all the data. All authors helped revise health impacts of ambient air pollutants: recommendations of a
the final version of this article and were responsible for the decision to WHO/Europe project. Int J Public Health 2015; 60: 619–27.
submit the manuscript. 19 Borchers Arriagada N, Palmer AJ, Bowman DM, Johnston FH.
Exceedances of national air quality standards for particulate matter
Declaration of interests in Western Australia: sources and health-related impacts. Med J Aust
We declare no competing interests. 2020; 213: 280–81.
Data sharing 20 Fann N, Alman B, Broome RA, et al. The health impacts and
All data, codes, and materials used in the analysis are available from the economic value of wildland fire episodes in the U.S.: 2008–2012.
Sci Total Environ 2017; 610–11: 802–09.
authors upon reasonable request.
21 Matz CJ, Egyed M, Xi G, et al. Health impact analysis of PM2·5 from
Acknowledgments wildfire smoke in Canada (2013–2015, 2017–2018). Sci Total Environ
NB-A is supported by a Tasmania Graduate Research Scholarship, and 2020; 725: 138506.
Top-up Scholarships from Asthma Australia and the New South Wales 22 Johnston FH, Henderson SB, Chen Y, et al. Estimated global
Bushfire Risk Management Research Hub. HC is supported by the mortality attributable to smoke from landscape fires.
New South Wales Bushfire Risk Management Research Hub. FHJ is Environ Health Perspect 2012; 120: 695–701.
supported by a fellowship from the Select Foundation. We thank the 23 Broome RA, Johnston FH, Horsley J, Morgan GG. A rapid
Department of Planning, Industry and Environment of New South Wales, assessment of the impact of hazard reduction burning around
Sydney, NSW, Australia, for providing the available hourly and daily air Sydney, May 2016. Med J Aust 2016; 205: 407–08.
quality data since 2000 and for providing funds to support this research 24 Filkov AI, Ngo T, Matthews S, Telfer S, Penman TD. Impact of
via the New South Wales Bushfire Risk Management Research Hub. Australia’s catastrophic 2019/20 bushfire season on communities
and environment. Retrospective analysis and current trends.
References J Saf Sci Resil 2020; 1: 44–56.
1 Xu R, Yu P, Abramson MJ, et al. Wildfires, global climate change, 25 Godfree RC, Knerr N, Encinas-Viso F, et al. Implications of the
and human health. N Engl J Med 2020; 383: 2173–81. 2019–2020 megafires for the biogeography and conservation of
2 Johnston FH, Borchers-Arriagada N, Morgan GG, et al. Australian vegetation. Nat Commun 2021; 12: 1023.
Unprecedented health costs of smoke-related PM₂.₅ from the 26 Borchers-Arriagada N, Palmer AJ, Bowman DM, Morgan GG,
2019–20 Australian megafires. Nat Sustain 2020; 4: 42–47. Jalaludin BB, Johnston FH. Unprecedented smoke-related health
3 Johnston FH. Bushfires and human health in a changing burden associated with the 2019–20 bushfires in eastern Australia.
environment. Aust Fam Physician 2009; 38: 720–24. Med J Aust 2020; 213: 282–83.

www.thelancet.com/planetary-health Vol 5 September 2021 e618

 Articles

27 Binskin M, Bennett A, Macintosh A. Royal Commission into 33 New South Wales Department of Planning Industry and
National Natural Disaster Arrangements Report. 2020 https:// Environment. Air quality annual report. 2020. https://www.
naturaldisaster.royalcommission.gov.au/publications/royal- environment.nsw.gov.au/topics/air/air-quality-annual-reports
commission-national-natural-disaster-arrangements-report (accessed Oct 15, 2020).
(accessed Nov 15, 2020). 34 Rappold AG, Reyes J, Pouliot G, Cascio WE, Diaz-Sanchez D.
28 Price OF, Bradstock RA. Quantifying the influence of fuel age and Community vulnerability to health impacts of wildland fire smoke
weather on the annual extent of unplanned fires in the Sydney exposure. Environ Sci Technol 2017; 51: 6674–82.
region of Australia. Int J Wildl Fire 2011; 20: 142. 35 Jaffe DA, O’Neill SM, Larkin NK, et al. Wildfire and prescribed
29 New South Wales Department of Planning Industry and burning impacts on air quality in the United States.
Environment. NPWS fire history - wildfires and prescribed burns. J Air Waste Manage Assoc 2020; published online April 2.
2020. https://data.nsw.gov.au/data/dataset/fire-history-wildfires- DOI:10.1080/10962247.2020.1749731.
and-prescribed-burns-1e8b6 (accessed July 6, 2020). 36 Johnston FH. Burning to reduce fuels: the benefits and risks of a
30 Breiman L. Random forests. Mach Learn 2001; 45: 5–32. public health protection strategy. Med J Aust 2020; 213: 246–48.
31 Johnston FH, Hanigan IC, Henderson SB, et al. Creating an 37 Ballesteros-González K, Sullivan AP, Morales-Betancourt R.
integrated historical record of extreme particulate air pollution Estimating the air quality and health impacts of biomass burning in
events in Australian cities from 1994 to 2007. northern South America using a chemical transport model.
J Air Waste Manag Assoc 2011; 61: 390–98. Sci Total Environ 2020; 739: 139755.
32 Hanigan IC, Morgan GG, Williamson GJ, et al. The biomass smoke 38 Bowman DMJS, Balch J, Artaxo P, et al. The human dimension of
validated events database. 2015. https://researchdata.edu.au/ fire regimes on Earth. J Biogeogr 2011; 38: 2223–36.
biomass-smoke-validated-events-database/934712 (accessed
Aug 22, 2021).

e619 www.thelancet.com/planetary-health Vol 5 September 2021