Environment, Green Space and Pollution
Risk of acute respiratory infection from crop
burning in India: estimating disease burden and
economic welfare from satellite and national
health survey data for 250 000 persons
Mohammed Tajuddin Khan,
and Samuel P Scott
Department of Global Health, University of Washington, Seattle, WA, USA,
Department of Economics,
Oklahoma State University, Stillwater, OK, USA,
South Asia Ofﬁce, International Food Policy Research
Institute, New Delhi, India,
Agriculture for Nutrition and Health, International Food Policy Research
Institute, Washington, DC, USA and
Poverty Health and Nutrition Division, International Food Policy
Research Institute, Washington, DC, USA
*Corresponding author. Poverty Health and Nutrition Division, International Food Policy Research Institute, 1201 Eye St,
NW, Washington, DC 20005-3915, USA. E-mail: email@example.com
Editorial decision 30 January 2019; Accepted 14 February 2019
Background: Respiratory infections are among the leading causes of death and disability
globally. Respirable aerosol particles released by agricultural crop-residue burning (ACRB),
practised by farmers in all global regions, are potentially harmful to human health. Our ob-
jective was to estimate the health and economic costs of ACRB in northern India.
Methods: The primary outcome was acute respiratory infection (ARI) from India’s fourth
District Level Health Survey (DLHS-4). DLHS-4 data were merged with Moderate-Resolution
Imaging Spectroradiometer satellite data on ﬁre occurrence. Mutually adjusted generalized
linear models were used to generate risk ratios for risk factors of ARI. Overall disease
burden due to ACRB was estimated in terms of disability-adjusted life years.
Results: Seeking medical treatment for ARI in the previous 2 weeks was reported by 5050
(2%) of 252 539 persons. Living in a district with intense ACRB—the top quintile of ﬁres
per day—was associated with a 3-fold higher risk of ARI (mutually adjusted risk ratio
2.99, 95% conﬁdence interval 2.77 to 3.23) after adjustment for socio-demographic and
household factors. Children under 5 years of age were particularly susceptible (3.65, 3.06
to 4.34 in this subgroup). Additional ARI risk factors included motor-vehicle congestion
(1.96, 1.72 to 2.23), open drainage (1.91, 1.73 to 2.11), cooking with biomass (1.73, 1.58 to
1.90) and living in urban areas (1.35, 1.26 to 1.44). Eliminating ACRB would avert 149 thou-
sand disability-adjusted life years lost per year, valued at US$1.529 billion over 5 years.
Conclusions: Investments to stop crop burning and offer farmers alternative crop-
residue disposal solutions are likely to improve population-level respiratory health and
yield major economic returns.
CThe Author(s) 2019. Published by Oxford University Press on behalf of the International Epidemiological Association . 1113
This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs licence (http://creativecommons.org/licenses/
by-nc-nd/4.0/), which permits non-commercial reproduction and distribution of the work, in any medium, provided the original work is not altered or transformed in any way,
and that the work is properly cited. For commercial re-use, please firstname.lastname@example.org
International Journal of Epidemiology, 2019, 1113–1124
Advance Access Publication Date: 28 February 2019
Downloaded from https://academic.oup.com/ije/article-abstract/48/4/1113/5366950 by University of Washington user on 23 January 2020
Key words: Respiratory health, agriculture, disease burden, India, air pollution
Respiratory infections are the most common chronic disease
of children globally, are a leading cause of death in develop-
ing countries and make a large contribution to the overall
burden of disease as measured by disability-adjusted life
years (DALYs) lost.
Air pollution is a recognized contrib-
utor to respiratory disease, as airborne fine particulate mat-
) from the burning of solid fuels, vehicle exhaust,
windblown soil, construction and other sources can pene-
trate deep into lung tissue, triggering an inflammatory cas-
cade and oxidative stress.
exposure has been linked
to increased asthma-related emergency-room visits and hos-
progression of carotid intima-medial thick-
greater chronic obstructive pulmonary disease
and reduced life expectancy.
In India in 2015,
exposure to outdoor air pollution was found to be the third
leading risk factor (after high blood pressure and high fast-
ing plasma glucose) contributing to mortality among 79
behavioural, environmental and metabolic factors.
study found that 12.5% of the total deaths in India in 2017
were attributable to air pollution.
Delhi was the state with
the highest annual population-weighted mean PM
lowed by Uttar Pradesh, Bihar and Haryana in north India.
Whereas indoor air pollution due to burning of solid fuels in
poor states like Uttar Pradesh and Bihar were important fac-
tors, the ambient particulate-matter pollution was highest in
the north Indian states of Uttar Pradesh, Haryana, Delhi,
Punjab and Rajasthan.
Delhi, India’s large capital city and home to 25 million
residents, is experiencing a public-health emergency due to
high levels of air pollution. Delhi was the most polluted
large city in the world in 2016, with an average annual
of 122 mgm
(micrograms per cubic meter)—12
times the World Health Organization (WHO)’s recom-
mended target of 10 mgm
Air pollution in Delhi is par-
ticularly extreme during the winter months—a period
when farmers in the neighbouring upwind states of
Haryana and Punjab—where the burden of outdoor air
pollution is also high
—practise agricultural crop-residue
Banned in November 2015 by the National Green
ACRB is still widely practised due to weak en-
forcement of the ban, political economy issues and lack of
alternatives to burning among poor farmers.
alone, an estimated 44–51 million metric tonnes of residue
are burned each year, with rice being the primary source.
Winds carry suspended particles hundreds of miles, gener-
ating a thick cloud of smog above northern India visible by
satellite. Recently, the contribution of ACRB in north-
western India to air pollution in Delhi has been quantified,
with estimates ranging between 7 and 78% of PM
enhancements during the burning season being attributable
Moreover, previous work has shown that
ACRB results in an unrecoverable decrease in pulmonary
function among children aged 10–13 years.
ferent sources of outdoor air pollution, ACRB was respon-
sible for an estimated 66 200 deaths in 2015 in India.
addition to affecting human health, ACRB deteriorates soil
fertility, releases greenhouse gases that contribute to global
warming and results in the loss of biodiversity.
•Burning of agricultural crop residue to clear ﬁelds is a major contributor to air pollution. When rice farmers in north-
western India burn their ﬁelds, ﬁne particulate matter (PM
) concentrations in Delhi, the highly populated capital city
located downwind of burning areas, spike to about 20 times beyond the World Health Organization’s threshold for
•Our results suggest that living in areas where crop burning is intense—measured using daily satellite imaging data over a
5-month period—is associated with a 3-fold higher risk of acute respiratory infection—one of the leading global causes of
lost disability-adjusted life years. Children are particularly susceptible to the health effects of crop burning.
•Solutions to eliminate crop burning exist but require further investments. We found that crop-burning abatement
would be highly cost-effective and, in northern India, would avert disability-adjusted life years equivalent to
US$1.529 billion over a 5-year period.
•Reducing crop burning would beneﬁt human health.
1114 International Journal of Epidemiology, 2019, Vol. 48, No. 4
Downloaded from https://academic.oup.com/ije/article-abstract/48/4/1113/5366950 by University of Washington user on 23 January 2020
The Indian government has demonstrated an interest in
combatting air pollution and respiratory illness but so far
has fallen short in addressing the air-quality crisis. India
was the first country to develop national targets aimed at
reducing deaths from non-communicable diseases (NCDs)
including respiratory illness, by 2025, following the
WHO’s Global Action Plan for the Prevention and Control
of NCDs 2013–2020.
However, a major barrier to
further political action against crop burning is the lack of
rigorous evidence linking this practice to health outcomes.
To our knowledge, there are no available estimates of the
economic costs and societal disease burden associated with
crop burning. Therefore, we sought to estimate the risk of
acute respiratory infection (ARI) attributable to crop burn-
ing, among other socio-demographic and household fac-
tors, and to quantify the costs to society of this practice.
Data on ARI were obtained from the fourth round of
India’s District Level Household Survey (DLHS-4)
national demographic health survey sampled to be repre-
sentative at district and state levels. Data were used from
households in three Indian states—Haryana (in northern
India), Andhra Pradesh and Tamil Nadu (both in southern
India)—interviewed between September 2013 and
February 2014, the two southern states serving as compa-
rators without ACRB. In DLHS-4, household heads were
asked whether any household member had suffered from
an illness in the previous 15 days. If the response was ‘yes’,
information on illness type was collected for affected per-
sons. For individuals who had reported symptoms of ARI
in the previous 15 days, a follow-up question was asked
about whether and where the individual received medical
treatment. Our primary outcome was seeking treatment
for ARI in the previous 15 days at a private or public medi-
cal facility among those who also reported ARI symptoms.
Data on ACRB were obtained from the National
Aeronautics and Space Administration (NASA) Moderate-
Resolution Imaging Spectroradiometer (MODIS) Terra
satellite Fire Information for Resource Management
System (FIRMS) database.
The MODIS fire locations
provide daily information on spatial and temporal fire dis-
tribution. Each hotspot/active fire detection represents the
centre of a 1-km
area. Using Global Positioning System
coordinates and Arc Geographic Information System soft-
ware (Environmental Systems Research Institute), we
mapped the fires to state district centroids and boundaries.
The number of fires recorded by MODIS were counted
and summed by district and day. Thus, the primary explan-
atory variable used in the study was the number of fires
recorded per district per day from 1 September 2013 to 28
February 2014—a timeframe selected to be inclusive of the
period before and after peak ACRB. Additional risk factors
from DLHS-4 included age, sex, urban or rural residence,
education in years, access to electricity, source of cooking
fuel (biomass or other), whether a water-purification
method was used, water-drainage infrastructure (open or
closed drain), number of rooms in the home and whether
the kitchen was inside the home. Motor-vehicle congestion
was measured using an index ranging from 0 to 1 for the
number of vehicles per square kilometre at the district
level, using data from India’s 2011 Census.
was constructed as [(vehicle density in district J – minimum
vehicle density in sample)/vehicle density range]. A value
of 0 implies that the district has the lowest vehicle density
among all districts in the sample. A value of 1 is typical of
congested urban districts.
A secondary contributor to poor air quality in northern
India is the burning of firecrackers during the Hindu festi-
val of lights, Diwali, which occurs between mid-October
In Delhi, the time around Diwali is
associated with sharp increases in concentrations of respi-
rable particulate matter, total suspended particulate mat-
ter, sulphur dioxide and surface ozone.
examined the period after Diwali—3–10 November in
2013—as an additional ARI risk factor.
Individual-level data from DLHS-4 were merged with
MODIS data on the number of fires, by district and survey
date. The merged data set was a high-frequency panel with
daily temporal and spatial variations on exposure to
ACRB at the district level. Since ARI was reported for the
previous 15 days, data were merged using a recall-adjusted
survey date (interview date minus 7 days). The final sample
size was 252 539 individuals.
Visualization of ACRB and ARI co-occurrence
MODIS data on the number of fires were mapped for
Haryana—a state known for endemic ACRB. Reported
treatment seeking for ARI and ACRB occurrence was plot-
ted on a shared timescale to determine whether the two in-
dependently observed time series moved in tandem during
the study period. Separate plots were constructed for
Haryana (high occurrence of ACRB) vs southern states
(low occurrence of ACRB).
International Journal of Epidemiology, 2019, Vol. 48, No. 4 1115
Downloaded from https://academic.oup.com/ije/article-abstract/48/4/1113/5366950 by University of Washington user on 23 January 2020
Estimating the association between ACRB
Associations between risk factors and reported ARI were
assessed through unadjusted and mutually adjusted
generalized linear models (GLMs), as ARI satisfies the
epidemiologic ‘rare disease assumption’.
To test for the
association of ACRB with ARI, for individual i belonging
to household h in district s and survey date t, we ran GLM
models using Equation (1):
ARIihst ¼aþACRBstb1þXihstXþihst :(1)
In the unadjusted model, ACRBst and each risk factor in
Xihst were individually related to ARI, whereas, in the mutu-
ally adjusted model, all risk factors were considered simulta-
neously. ACRBst is the daily district-level exposure to intense
ACRB, Xihst is the set of risk factors that an individual was ex-
posed to and ihst is the individual-specific error term. The rel-
ative risk ratio b1 represents the ratio of the risk (probability)
of ARI in the exposed group to the risk of ARI in the reference
Exposure to pollution from ACRB was deemed in-
tense if the household member was interviewed on a day
when MODIS detected 100 or more fires (the top quintile of
fires per day per district) in their district of residence.
To test the sensitivity of our estimates from Equation (1),
we ran regressions on sub-samples of young children
(under 5 years), the elderly (over 60 years) and place of
residence (rural vs urban). Using Equation (1), we also
estimated models using the natural logarithm of the num-
ber of fires per day in a district as a continuous explanatory
variable to obtain the marginal effect of a 1% increase in
daily fire occurrence.
Health and economic benefits to society from
In estimating the disease burden attributable to ACRB, we
considered the total population in three Indian states likely
to be affected by this practice: Haryana, Punjab and Delhi.
These states were chosen based on previous ACRB stud-
and from satellite images of the fires (Figure 1).
Maps showing district-level population density, population
of young children and urbanization in the studied states
can be found in Supplementary Figure 1, available as
Supplementary data at IJE online.
Four steps were taken to estimate the direct disease bur-
den attributable to ACRB and firecracker burning. First,
the total estimated number of cases of ARI that would be
averted if exposure to these risk factors were eliminated in
the population, or the population attributable fraction
(PAF), was calculated using the ‘punaf’ STATA routine.
Second, state-specific DALY rates attributable to ARIs
were obtained for Haryana, Punjab and Delhi from India
state-level Disease Burden Initiative Collaborators.
these DALY rates were multiplied by our estimated PAF
values for ACRB or firecrackers and by state population to
estimate total DALYs attributable to these two risk factors
in each state. For Punjab, to account for the higher inci-
dence of ACRB in this state relative to Haryana, the DALY
rates were apportioned based on the ratio of percentage
area burned in Punjab relative to Haryana (67% in Punjab
vs 29% in Haryana, i.e. 2.3 times).
Fourth, total DALYs
saved were converted to per-year economic benefits by
multiplying the state-specific per-capita gross domestic
product (GDP) by the total DALYs attributable to ACRB
For firecrackers, we also subtracted
US$1 billion (US$200 million per year over 5 years)—the
estimated loss in revenue from the sale of firecrackers dur-
Benefits were compounded for 5 years using
a 3% discount rate, as is standard practice when account-
ing for financial returns at a future date.
Crop burning in Haryana during October and November
in 2013 was most intense in the northern districts and was
primarily practised in districts where paddy cultivation
was also practised (Figure 2).
Figure 1. Typical crop-residue-burning episode and affected geographi-
cal area. State borders are overlaid on a MODIS satellite image from 5
November 2012. Active ﬁres are shown as dots (NASA 2012).
1116 International Journal of Epidemiology, 2019, Vol. 48, No. 4
The frequency of reported ARI symptoms in Haryana
closely paralleled the number of fires observed by the
MODIS satellite in this state, with ARI symptoms being
more frequently reported in urban than in rural areas
(Figure 3). In south Indian states (Andhra Pradesh and
Tamil Nadu), where ACRB is not practised and firecracker
burning during Diwali is much less prevalent, ARI fre-
quency and the number of fires were low.
In Haryana, 5.4% of surveyed individuals reported
suffering from ARI symptoms in the previous 15days,
whereas the reported ARI symptoms in southern states
were only 0.1% (Table 1). Among those who reported
suffering from ARI, 83% also reported receiving treat-
ment for ARI at a private or public medical facility.
Whereas high-intensity fire exposure was virtually absent
in south India, 17.5% of individuals in Haryana lived in a
district where 100 or more fires per day were observed by
satellite. In Haryana, compared with southern states,
more households cooked with biomass (15.3 vs 0.2%)
and drained water into an open drain (90.6 vs 70.4%)
and fewer households treated water before drinking (15.8
Living in a district with intense ACRB was the leading
risk factor for ARI, with 100 fires per day in the district
being associated with a 3-fold higher risk of ARI in the mu-
tually adjusted model (adjusted risk ratio, 95% confidence
interval: 2.99, 2.77 to 3.23) (Figure 4). Other risk factors
included the week after Diwali (2.45, 2.21 to 2.72), being
less than 5 years of age (2.21, 2.17 to 2.61), living in a dis-
trict with high motor-vehicle congestion (1.96, 1.71 to
2.23), draining water into open drains (1.91, 1.73 to 2.11),
using biomass for cooking (1.73, 1.58 to 1.90), living in
Figure 3. Temporal association between incidence of agricultural crop-residue burning and acute respiratory infection among Indians. Dashed lines
in lower two panels indicate urban areas and solid lines indicate rural areas. Data on number of ﬁres were sourced from NASA-MODIS-FIRMS.
on reported ARI were sourced from Indian DLHS-4.
ARI, acute respiratory infection.
Figure 2. Average number of ﬁres per day during October and November
2013 in districts in Haryana. Fire data from NASA-MODIS-FIRMS, 2013.
International Journal of Epidemiology, 2019, Vol. 48, No. 4 1117
Table 1. Summary statistics for Indians surveyed in three states between September 2013 and February 2014
Haryana Andhra Pradesh and Tamil Nadu All
N90 327 162 212 252 539
Reported ARI in previous 2 weeks, % 5.4 0.1 2.0
N4832 224 5056
Treatment seeking by those who reported ARI
Treatment at a private facility 76.5 52.7 75.5
Treatment at a public facility 6.5 35.7 7.8
Treatment at home 3.5 3.1 3.4
No treatment 9.0 1.8 8.7
N90 327 162 212 252 539
Days when district had 100 or more ﬁres, % 17.5 0.0 6.3
1–7 days after Diwali, % 7.6 0.0 2.7
Children less than 5 years of age, % 8.0 7.1 7.4
Adults 60 years and older, % 10.1 12.4 11.6
Females, % 46.8 51.5 49.8
Urban residence, % 41.5 42.5 42.1
Less than 5 years of education, % 47.9 50.8 49.7
Motor-vehicle index 0.2 0.1 0.1
Household has electricity, % 97.2 98.6 98.1
Cooks with biomass, %15.3 0.2 5.6
Treats water before drinking, % 15.8 26.9 23.0
Drains water into open drain, % 90.6 70.4 77.6
Fewer than two rooms in house, % 39.7 46.1 43.8
Kitchen is inside house, % 69.3 69.4 69.4
Data were taken from the fourth round of the District Level Household Survey (International Institute for Population Sciences 2015) other than days when the
district had 100 or more ﬁres, which were derived from NASA-MODIS-FIRMS data (National Aeronautics and Space Administration).
Only individuals who reported ARI in the previous 2 weeks and sought treatment at a private or public medical facility were classiﬁed as having the outcome
Figure 4. Risk of acute respiratory infection among multiple determinants. 95% conﬁdence intervals are shown for unadjusted (grey) and mutually ad-
justed (black) risk ratios. The dashed vertical line at 1.0 on the x-axis indicates no risk difference between groups. The mutually adjusted model was
adjusted for all other factors shown. Data were taken from DLHS-4
other than living in intense crop-burning district (100 or more ﬁres per day),
which was derived from NASA-MODIS-FIRMS data,
and vehicle density, which was derived from India Census 2011 data.
1118 International Journal of Epidemiology, 2019, Vol. 48, No. 4
urban areas (1.35, 1.26 to 1.44) and being 60 years or
older (1.14, 1.03 to 1.25). Protective factors included hav-
ing access to electricity (0.40, 0.35 to 0.46), treating water
before drinking (0.75, 0.69 to 0.82), having fewer than
two rooms in the house (0.87, 0.82 to 0.93) and being fe-
male (0.92, 0.86 to 0.97).
Analysis on subgroups (Table 2) showed that children
under 5 years old are at particularly high risk of ARI from
ACRB (3.65, 3.06 to 4.34) and open drainage (2.85, 2.14
to 3.80). The elderly were particularly susceptible to ARI
associated with firecracker burning (2.53, 1.84 to 3.49)
and cooking with biomass (1.92, 1.44 to 2.55). Individuals
residing in urban areas were at a higher risk of ARI from
ACRB and Diwali, whereas those in rural areas were at
greater risk of ARI from open drains and cooking with bio-
mass. Morever, children and elderly living in urban areas
were at a higher risk for ARI associated with ACRB than
those living in rural areas. When looking at sex differences
in risk factors of ARI, motor-vehicle congestion was a
stronger risk factor among females (2.18, 1.80 to 2.64)
than among males (1.78, 1.49 to 2.14). Similarly, cooking
with biomass was a greater risk factor for females (1.84,
1.61 to 2.12) compared with males (1.64, 1.44 to 1.86).
Results using a continous explanatory variable to measure
ACRB were similar (Supplementary Table 1, available as
Supplementary data at IJE online).
We estimated that 14.4% of all ARI cases were attribut-
able to ACRB and 6.6% are attributable to burning fire-
crackers in 2013 (Table 3). For Haryana, Punjab and Delhi
combined, the total number of DALYs averted from elimi-
nating ACRB was estimated to be 149 thousand years, val-
ued at US$1.529 billion for 5 years. DALYs averted by
eliminating firecrackers were estimated at 42 thosuand
years, valued at US$357 million for 5 years.
Our study has three novel findings. First, exposure to
outdoor air pollution from intense ACRB. firecracker
burning and motor vehicles are three leading risk factors
for ARI in northern India. Second, children under 5 years
old are at high risk for ARI from ACRB, whereas other
leading household-level risk factors for ARI are exposure
to open drains, cooking with biomass and urban residence.
Third, the DALY benefits attributable to complete ACRB
and firecracker abatement in three states in northern
India are estimated at 149 and 42 thousand years, respec-
tively, valued at about US$1.9 billion cumulative for
Apart from ACRB and firecrackers, open drainage and
the use of biomass for cooking fuel were associated with
higher ARI, whereas access to electricity and water
filtration were protective factors. We also found a small
protective effect of having fewer rooms in the house; this
may just reflect poverty and the fact that poor households
have lower healthcare-seeking practices and thus were not
classified as having ARI in our analyses. The coefficients
were in the expected direction
and the relative
magnitude of the effects is important for policy focus.
Programmes and policies must simultaneously address in-
door and outdoor pollution, through a combination of
bans and agricultural subsidies; increasing access to im-
proved cooking fuels like liquefied petroleum gas, electrifi-
cation and promotion of induction cooking stoves; and the
construction of improved drainage systems for households.
In addition, behavioural-change communication cam-
paigns for the promotion of water treatment and reduction
of firecracker use are likely to strengthen macro-level poli-
cies and outcomes.
We found a relatively stronger effect of outdoor air pol-
lution (crop burning, firecrackers, motor vehicles) on ARI
compared with indoor air pollution (cooking with bio-
mass). Recently, a large-cluster randomized–controlled
trial in Malawi found no benefit of switching to cleaner
cooking stoves on child pneumonia.
The authors suggest
that daily exposure to pollution from other sources may
have negated the effect of the intervention, which only tar-
geted indoor pollution. There is a large ongoing effort by
the central Indian government to expand the use of liquid
petroleum gas for cooking to 80% of households by
and it will be of interest to assess the impact of
these policy efforts on health outcomes in the next few
In our subgroup analyses, we found that motor-vehicle
congestion was a larger risk factor for ARI in women com-
pared with men. This finding is in line with literature sug-
gesting that women may be more susceptible to pollutants
compared with men for biological and social reasons.
We also found that firecracker burning was a stronger risk
factor in urban compared with rural areas, but that open
drainage was a stronger risk factor in rural areas compared
with urban areas. These findings are as expected, given rel-
atively intense firecracker burning and less open drainage
in urban centres compared with rural areas.
Our findings are in line with previous investigations
reporting on exposure to environmental air pollution and
respiratory health. Wildfires, which are becoming increas-
ingly common with climate change, have been shown to
have a broad range of negative population-level health
impacts, including respiratory infection.
A few studies
have reported more frequent hospital visits for respiratory
infections following wildfires in the USA and Canada.
A study in Australia showed that an increase in PM
was associated with a 15% increase risk of
International Journal of Epidemiology, 2019, Vol. 48, No. 4 1119
Table 2. Relative risk models for acute respiratory infection on population subgroups
Children (<5 years) Elderly (>59.9 years) Rural Urban Females Males
ARR 95 % CI ARR 95 % CI ARR 95 % CI ARR 95 % CI ARR 95 % CI ARR 95 % CI
Intense ACRB, 0/1 3.646 [3.06, 4.34] 2.982 [2.34, 3.80] 2.913 [2.63, 3.22] 3.245 [2.88, 3.65] 3.076 [2.75, 3.45] 2.929 [2.64, 3.25]
Diwali week, 0/1 2.037 [1.56, 2.65] 2.532 [1.84, 3.49] 2.042 [1.76, 2.37] 2.949 [2.54, 3.42] 2.526 [2.16, 2.95] 2.397 [2.08, 2.76]
District vehicle index, 0–1 1.625 [1.16, 2.27] 1.807 [1.16, 2.81] 2.886 [2.45, 3.40] 1.002 [0.80, 1.26] 2.182 [1.80, 2.64] 1.784 [1.49, 2.14]
Exposed to open drain, 0/1 2.848 [2.14, 3.80] 1.636 [1.25, 2.13] 2.444 [2.09, 2.86] 1.558 [1.37, 1.77] 1.789 [1.56, 2.06] 2.025 [1.76, 2.33]
Cooks with biomass, 0/1 1.303 [1.03, 1.65] 1.918 [1.44, 2.55] 1.769 [1.59, 1.97] 1.216 [0.96, 1.54] 1.844 [1.61, 2.12] 1.639 [1.44, 1.86]
Urban residence, 0/1 1.511 [1.29, 1.76] 1.260 [1.03, 1.54] 1.423 [1.29, 1.56] 1.291 [1.18, 1.41]
Kitchen is inside house, 0/1 1.139 [0.95, 1.37] 1.156 [0.92, 1.45] 1.036 [0.94, 1.14] 0.990 [0.87, 1.12] 1.005 [0.90, 1.12] 1.049 [0.95, 1.16]
Female, 0/1 0.834 [0.72, 0.97] 0.901 [0.74, 1.09] 0.880 [0.81, 0.96] 0.956 [0.87, 1.05]
Fewer than two rooms in house, 0/1 0.768 [0.65, 0.91] 0.954 [0.78, 1.17] 0.890 [0.81, 0.97] 0.890 [0.80, 0.99] 0.849 [0.77, 0.94] 0.898 [0.82, 0.98]
Treats drinking water, 0/1 0.663 [0.54, 0.82] 0.775 [0.61, 0.99] 0.553 [0.47, 0.65] 0.868 [0.78, 0.96] 0.771 [0.68, 0.87] 0.739 [0.66, 0.83]
Has electricity, 0/1 0.348 [0.25, 0.49] 0.383 [0.26, 0.57] 0.409 [0.35, 0.48] 0.344 [0.27, 0.44] 0.422 [0.34, 0.52] 0.377 [0.31, 0.46]
Below 5 years of education, 0/1 0.793 [0.63, 0.99] 1.159 [1.05, 1.28] 1.040 [0.93, 1.16] 1.183 [1.07, 1.31] 1.036 [0.94, 1.14]
Children (<5 years), 0/1 2.152 [1.90, 2.43] 2.745 [2.39, 3.15] 2.128 [1.86, 2.43] 2.633 [2.32, 2.98]
Elderly (>59.9 years), 0/1 1.104 [0.97, 1.26] 1.175 [1.01, 1.36] 1.031 [0.89, 1.19] 1.233 [1.08, 1.41]
N18 760 29 282 146 163 106 376 125 866 126 673
ARR, mutually adjusted risk ratio, i.e. adjusted for all other factors shown; CI, conﬁdence interval. Data were taken from DLHS-4 (International Institute for Population Sciences 2015) other than intense ACRB (days
when district had 100 or more ﬁres), which was derived from NASA-MODIS-FIRMS data (National Aeronautics and Space Administration).
1120 International Journal of Epidemiology, 2019, Vol. 48, No. 4
respiratory infection among indigenous people living in
dry areas with frequent vegetation fires.
bacco smoke has also been related to ARI. The odds of
hospital admission for ARI were 1.55 times higher in chil-
dren under 2 years of age who were exposed to second-
hand tobacco smoke compared with children who were
Respiratory infections were found to be the single
greatest contributor to the global burden of disease attrib-
utable to second-hand smoke in children under 5 years of
Others have looked at ambient particulate-matter
concentration (from multiple sources) as a predictor of re-
spiratory outcomes. A meta-analysis of four studies in
areas with relatively low average annual PM
tions (12–25 lgm
, compared with 153 lgm
during the study period
) found a 12% increased risk of
ARI in children under 5 years of age per 10 lgm
A long-term study in the USA found that a
1-interquartile range increase in ozone predicted a 4%
increase in respiratory infection among children aged
Given that most studies relating poor air qual-
ity to respiratory health come from developed countries
with much lower ambient concentrations of pollutants as
well as less poverty and better health services, direct com-
parison to the currently studied population in northern
India may be limited.
Our study has several strengths. First, to our knowl-
edge, this is the only study to systematically estimate the ef-
fect of exposure to ACRB on ARI in India. Second, we
leveraged two high-resolution data sets: crop-burning data
from MODIS and data on reported ARI among all age
groups of men and women in Haryana from DLHS-4,
which has a large sample size, is representative at the dis-
trict level and were collected when ACRB peaks. Third, we
accounted for a comprehensive set of risk factors in our
mutually adjusted statistical model, thus our study pro-
vides an important benchmark in examining the relative
contribution of these risk factors for ARI. Fourth, we pro-
vide the first available estimates of ACRB abatement in
terms of DALYs and US dollars.
Our study is not without limitations. First, the DLHS
survey was not conducted in Punjab—the state with the
highest incidence of ACRB in October—thus we could not
estimate the contribution of ACRB on ARI in that state.
Second, to our knowledge, high-frequency, high-resolution
data on particulate-matter concentration in the ambient air
of Haryana or Punjab are not available; consequently, we
could not estimate the dose–response function linking
ACRB and the resulting increase in air pollution to health
outcomes. Third, in our mutually adjusted models, we con-
trolled for other factors that may also increase outdoor air
pollution. However, it is possible that we missed some fac-
tors, such as a fall in the ambient temperature, which may
worsen the effect of air pollution on population health.
Fourth, the effects of firecracker burning were estimated
by specifying a dummy variable for the week after Diwali.
The independent effects of firecracker burning would
Table 3. Economic beneﬁts of government actions to abate agricultural crop-residue burning and burning of ﬁrecrackers
Haryana Punjab Delhi Total
Row ACRB Firecrackers ACRB Firecrackers ACRB Firecrackers ACRB Firecrackers
1 DALY rates for ARI per 100,000
1311 1311 887 887 799 799
2 State population, millions
25.4 25.4 27.7 27.7 16.8 16.8
3 Proportion of ARI cases attributed
to risk factor
0.14 0.06 0.33 0.06 0.14 0.06 0.14 0.07
4 DALYs saved by eliminating risk
factor, thousand years
47.9 19.6 81.4 14.5 19.3 7.9 148.5 420
5 Per-capita state GDP, US$/person
2637.26 2637.26 1739.67 1739.67 4575.16 4575.16
6 Economic value of DALYs saved
per yr, US$ million/yr
126.2 41.7 141.6 15.2 88.2 26.1 356.0 83.1
7 Economic value of DALYs saved
over 5 yrs, US$million
542.0 179.1 607.9 65.4 378.7 112.2 1528.6 356.7
From Figure 4 in Dandona et al. (2017).
From Indian Population Census 2011 (Ofﬁce of the Registrar General & Census Commissioner 2011).
From PUNAF after estimating Equation (1), with the PAR for Punjab multiplied by 2.3 to account for higher ACRB in Punjab relative to Haryana (Lohan
et al. 2018).
Row 1 Row 2 Row 3.
From NITI Aayog (National Institution for Transforming India 2017).
Row 4 Row 5. One billion in economic beneﬁts subtracted due to losses faced by ﬁrecracker industry (KV Lakshmana 2017).
From Row 6 for 5 years disco unted at 3% per year.
DALY, disability-adjusted life years; GDP, gross domestic product.
International Journal of Epidemiology, 2019, Vol. 48, No. 4 1121
ideally be estimated in the absence of ACRB to avoid con-
founding, but this was not possible due to co-occurrence of
these events. Therefore, our estimate of the contribution of
firecracker burning is likely biased. Finally, we note that
low variability in the district vehicle index in the urban
sample prevents us from being able to accurately examine
the effect of motor-vehicle density on ARI in cities, and
data at a higher spatial resolution would allow better risk
assessment in this case.
We found that living in an area where crop burning is prac-
tised was a leading risk factor for respiratory disease in
northern India. Whereas the total burden of diseases from
air pollution declined between 1990 and 2016 due to
efforts to reduce the burning of solid fuel for household
use, outdoor air pollution increased by 16.6%.
that ACRB leads to over US$300 million losses in eco-
nomic value every year and a 3-fold risk of ARI to those
exposed in the general population. Any investments made
for the abatement of ACRB are likely to be highly favour-
able in terms of their return on investment, when com-
pared with other public-health interventions or
ACRB is a growing problem in India and in
The promotion of sustainable eco-
nomic development should include investments to stop
Supplementary data are available at IJE online.
We acknowledge the permission to use data and imagery from
LANCE FIRMS operated by the NASA/GSFC/Earth Science Data
and Information System (ESDIS). We thank Pramod Kumar Joshi
for his feedback on the manuscript prior to submission. This work
was supported by the Indian Council of Agricultural Research. The
funders had no role in study design, data collection and analysis, de-
cision to publish or preparation of the manuscript.
Conﬂict of interest: None declared.
1. Ferkol T, Schraufnagel D. The global burden of respiratory dis-
ease. Annals ATS 2014;11:404–06.
2. GBD 2015 LRI Collaborators, Forouzanfar M, Rao PC et al.
Estimates of the global, regional, and national morbidity, mor-
tality, and aetiologies of lower respiratory tract infections in 195
countries: a systematic analysis for the Global Burden of Disease
Study 2015. Lancet Infect Dis 2017;17:1133–61.
3. Chen J, Li C, Ristovski Z et al. A review of biomass burning:
emissions and impacts on air quality, health and climate in
China. Sci Total Environ 2017;579:1000–34.
4. Adar SD, Sheppard L, Vedal S et al. Fine particulate air pollution
and the progression of carotid intima-medial thickness: a pro-
spective cohort study from the multi-ethnic study of atheroscle-
rosis and air pollution. PLoS Med 2013;10:e1001430.
5. Bentayeb M, Simoni M, Baiz N et al. Adverse respiratory effects
of outdoor air pollution in the elderly. Int J Tuberc Lung Dis
6. Health Effects Institute. Burden of Disease Attributable to
Major Air Pollution Sources in India. Boston, MA: Health
Effects Institute, 2018.
7. India State-Level Disease Burden Initiative Air Pollution
Collaborators. The impact of air pollution on deaths, disease
burden, and life expectancy across the states of India: the Global
Burden of Disease Study 2017. Lancet Planet Health 2019;3:
8. WHO. WHO Global Urban Ambient Air Pollution Database
(Update 2016). Geneva, Switzerland: World Health Organization.
9. Dandona L, Dandona R, Kumar GA et al. Nations within a na-
tion: variations in epidemiological transition across the states of
India, 1990–2016 in the Global Burden of Disease Study. Lancet
10. Vasudeva V. Punjab farmers defy ban on stubble burning. The
Hindu. 2017. https://www.thehindu.com/news/national/other-
2262.ece (15 September 2018, date last accessed).
11. Lohan SK, Jat HS, Yadav AK et al. Burning issues of paddy resi-
due management in north-west states of India. Renew Sustain
Energy Rev 2018;81:693–706.
12. Ministry of New & Renewable Energy G of I. Biomass Resource
Atlas of India. Bangalore, India: CGPL, Indian Institute of
13. Cusworth DH, Mickley LJ, Sulprizio MP et al. Quantifying the
inﬂuence of agricultural ﬁres in northwest India on urban air pol-
lution in Delhi, India. Environ Res Lett 2018;13:044018.
14. Awasthi A, Singh N, Mittal S, Gupta PK, Agarwal R. Effects of
agriculture crop residue burning on children and young on PFTs
in North West India. Sci Total Environ 2010;408:4440–45.
15. WHO. Global action plan for the prevention and control of non-
communicable diseases 2013–2020. World Health Organ 2013;
16. International Institute for Population Sciences. District Level
Household Survey 2012–2013 (DLHS-4). Mumbai, India:
International Institute for Population Sciences, 2015.
17. National Aeronautics and Space Administration. Fire
Information for Resource Management System. USA: LANCE
18. Ofﬁce of the Registrar General & Census Commissioner.
Population Enumeration Data (Final Population). New Delhi,
India: ORGI DDU, 2011.
19. Kulshrestha UC, Nageswara Rao T, Azhaguvel S, Kulshrestha
MJ. Emissions and accumulation of metals in the atmosphere
due to crackers and sparkles during Diwali festival in India.
Atmos Environ 2004;38:4421–25.
20. Ganguly ND. Surface ozone pollution during the festival of
Diwali, New Delhi, India. J Earth Sci India 2009;2:224–29.
21. Mandal P, Prakash M, Bassin JK. Impact of Diwali celebrations
on urban air and noise quality in Delhi City, India. Environ
Monit Assess 2012;184:209–15.
1122 International Journal of Epidemiology, 2019, Vol. 48, No. 4
22. Perrino C, Tiwari S, Catrambone M, Torre SD, Rantica E,
Canepari S. Chemical characterization of atmospheric PM in
Delhi, India, during different periods of the year including
Diwali festival. Atmos Pollut Res 2011;2:418–27.
23. Schmidt CO, Kohlmann T. When to use the odds ratio or the rel-
ative risk? Int J Public Health 2008;53:165–67.
24. Northridge ME. Public health methods—attributable risk as a
link between causality and public health action. Am J Public
25. Subramanian M. Can Delhi save itself from its toxic air? Nature
26. Newson R. PUNAF : Stata Module to Compute Population
Attributable Fractions for Cohort Studies Fractions and Other
Scenario Comparisons. Boston, USA: Boston College
Department of Economics, 2016.
27. Brown DW. Economic value of disability-adjusted life years lost
to violence: estimates for WHO Member States. Rev Panam
Salud Publica 2008;24:203–09.
28. Lakshmana KV. SC ban on ﬁrecrackers: industry stares at Rs 1,
000-crore loss, layoffs. The Hindustan Times. 2017. https://
DaPf4O5DWggJI.html (15 September 2018, date last accessed).
29. Shekar M, Dayton Eberwein J, Kakietek J. The costs of stunting
in South Asia and the beneﬁts of public investments in nutrition.
Matern Child Nutr 2016;12:186–95.
30. Zheng X, Ding H, Jiang L et al. Association between air pollu-
tants and asthma emergency room visits and hospital admissions
in time series studies: a systematic review and meta-analysis.
PLoS One 2015;10:e0138146.
31. Jayachandran S. Air quality and early-life mortality: evidence
from Indonesia’s wildﬁres. J Human Resources 2009;44:916–54.
32. Singh P. Seasonal Exposure to Wildﬁres: Child Weight
Outcomes for India. New Delhi, India: India Statistical Institute,
2017, pp. 1–28.
33. Zhang L, Liu Y, Hao L. Contributions of open crop straw burn-
ing emissions to PM2.5 concentrations in China. Environ Res
34. Page CM, Patel A, Hibberd PL. Does smoke from biomass fuel
contribute to anemia in pregnant women in Nagpur, India? A
cross-sectional study. PLoS One 2015;10:e0127890.
35. Zhou Y, Zou Y, Li X et al. Lung function and incidence of
chronic obstructive pulmonary disease after improved cooking
fuels and kitchen ventilation: a 9-year prospective cohort study.
PLoS Med 2014;11:e1001621.
36. Canc¸ado JED, Saldiva PHN, Pereira LAA et al. The impact of
sugar cane-burning emissions on the respiratory system of chil-
dren and the elderly. Environ Health Perspect 2006;114:725–29.
37. Schei MA, Hessen JO, Smith KR, Bruce N, McCracken J, Lopez
V. Childhood asthma and indoor woodsmoke from cooking in
Guatemala. J Expo Sci Environ Epidemiol 2004;14:S110–17.
38. Gurung A, Bell ML. Exposure to airborne particulate matter in
Kathmandu Valley, Nepal. J Expo Sci Environ Epidemiol 2012;
39. Liu LS. Exposure and Health Assessments of the Effects of
Agricultural Field Burning in Young Adults with Asthma Living
in Pullman Washington. Seattle, Washington: University of
40. Mortimer K, Ndamala CB, Naunje AW et al. A cleaner burning
biomass-fuelled cookstove intervention to prevent pneumonia in
children under 5 years old in rural Malawi (the Cooking and
Pneumonia Study): a cluster randomised controlled trial. Lancet
41. Times of India. PM Modi: 10 crore LPG connections given in 4
years against 13 crore in 6 decades. India News—Times of India.
articleshow/64350785.cms (15 September 2018, date last
42. Butter ME. Are women more vulnerable to environmental pollu-
tion? J Hum Ecol 2006;20:221–6.
43. Clougherty JE. A growing role for gender analysis in air
pollution epidemiology. Environ Health Perspect 2010;118:
44. Black C, Tesfaigzi Y, Bassein JA, Miller LA. Wildﬁre smoke ex-
posure and human health: signiﬁcant gaps in research for a grow-
ing public health issue. Environ Toxicol Pharmacol 2017;55:
45. Duclos P, Sanderson LM, Lipsett M. The 1987 forest ﬁre disaster
in California: assessment of Emergency Room visits. Arch
Environ Health 1990;45:53–58.
46. Rappold AG, Stone SL, Cascio WE et al. Peat bog wildﬁre smoke
exposure in rural North Carolina is associated with cardiopul-
monary emergency department visits assessed through syn-
dromic surveillance. Environ Health Perspect 2011;119:
47. Henderson SB, Brauer M, MacNab YC, Kennedy SM. Three
measures of forest ﬁre smoke exposure and their associations
with respiratory and cardiovascular health outcomes in a
population-based cohort. Environ Health Perspect 2011;119:
48. Reid CE, Jerrett M, Tager IB, Petersen ML, Mann JK, Balmes
JR. Differential respiratory health effects from the 2008 northern
California wildﬁres: a spatiotemporal approach. Environ Res
49. Hanigan IC, Johnston FH, Morgan GG. Vegetation ﬁre smoke,
indigenous status and cardio-respiratory hospital admissions in
Darwin, Australia, 1996–2005: a time-series study. Environ
50. US Ofﬁce on Smoking and Health. The Health Consequences of
Involuntary Exposure to Tobacco Smoke [Internet].Heal.
Consequences Involuntary Expo. to Tob. Smoke A Rep. Surg.
Gen. Centers for Disease Control and Prevention (US), 2006.
http://www.ncbi.nlm.nih.gov/pubmed/20669524 (18 December
2018, date last accessed).
51. Oberg M, Jaakkola MS, Woodward A, Peruga A, Pru¨ ss-Ustu¨n A.
Worldwide burden of disease from exposure to second-hand
smoke: a retrospective analysis of data from 192 countries.
Lancet (London, England) 2011;377:139–46.
52. World Health Organization. Ambient (Outdoor) Air Pollution
in Cities Database 2014. Geneva, Switzerland: WHO World
Health Organization, 2014. https://www.who.int/phe/health_
topics/outdoorair/databases/cities-2014/en/ (15 September 2018,
date last accessed).
53. Mehta S, Shin H, Burnett R, North T, Cohen AJ. Ambient partic-
ulate air pollution and acute lower respiratory infections: a
International Journal of Epidemiology, 2019, Vol. 48, No. 4 1123
systematic review and implications for estimating the global bur-
den of disease. Air Qual Atmos Health 2013;6:69–83.
54. Darrow LA, Klein M, Flanders WD, Mulholland JA, Tolbert PE,
Strickland MJ. Air pollution and acute respiratory infections
among children 0–4 years of age: an 18-year time-series study.
Am J Epidemiol 2014;180:968–77.
55. Hoddinott J, Alderman H, Behrman JR, Haddad L, Horton S.
The economic rationale for investing in stunting reduction.
Matern Child Nutr 2013;9:69–82.
56. Lomborg B. Global Problems Smart Solutions: Costs and
Beneﬁts. Cambridge, UK: Cambridge University Press, 2012.
57. Zhuang Y, Chen D, Li R et al. Understanding the inﬂuence of
crop residue burning on PM
China from 2013 to 2017 using MODIS data. Int J Environ Res
Public Health 2018;15:1504.
58. Stavrakou T, Mu¨ller JF, Bauwens M et al. Substantial underestima-
tion of post-harvest burning emissions in the north china plain
revealed by multi-species space observations. Sci Rep 2016;6:1–11.
1124 International Journal of Epidemiology, 2019, Vol. 48, No. 4