Background/aim Asthma is a burdensome disease which has been cited as the most common chronic disease in children. Traffic-related air pollution (TRAP) may be an important exposure in its development. Bradford, UK, is a multi-ethnic deprived city suffering from asthma rates higher than national and regional averages. TRAP is of concern to local communities and is thought to contribute to the asthma burden. Methods We estimated childhood population exposure to traffic-related NOx and NO2 at the smallest census tract level in Bradford using three validated exposure assessment methods: two novel full-chain models linking transport, two different emission models and air dispersion models and one land-use regression (LUR) model, which was developed as part of ESCAPE and which also provided estimates for other pollutants including PM2.5, PM10 and black carbon. We extracted national and local childhood asthma incidence rates and used meta-analytic exposure-response functions. We calculated the relative risk and population attributable fraction of childhood asthma attributable to each pollutant. We estimated asthma cases attributable to each pollutant and exposure combination. Results Depending on the exposure assessment method used and the pollutant studied, the estimated TRAP-attributable asthma cases varied between 279 and 687 annually, representing between 15% to 38% of all asthma cases in the city. The health impacts estimated were sensitive to the exposure assessment method used, the pollutant selected in the analysis but, differently from the initial hypothesis, not to the vehicle emission factors used in the full-chain models. Conclusion TRAP is estimated to cause a large, but preventable, childhood asthma burden. This study is the first study undertaking full-chain health impact assessment that considers the full-chain from source, through exposure pathway to outcome. The study also adds to the scarce literature exploring the impacts of different exposure assessments on the estimated burden of disease.
2. Previous HIA of TRAP and Childhood Asthma
âą Only four relevant studies, coming from the same
research group (Perez et al., 2009, Perez et al., 2013,
KĂŒnzli et al., 2008, Perez et al., 2012)
âą Three from California and one European (10 cities)
âą Exposure to TRAP was characterized by proximity to
major roadways
âą Underlying exposure-response function sourced from
one Southern Californian cohort study (McConnell et
al. 2006)
âą All examined prevalent asthma
7. Traffic modelling
Simulation and Assignment of Traffic to Urban Road Networks model (SATURN)
4,500 road links
1,359 km coverage
3 simulation time periods (AM, Inter-peak, PM)
29. Baseline health data
Wheezing disorders
based on treatment
which identifies the
existence of at least
two drug prescriptions
indicated for the
treatment of asthma a
minimum of 1 week
and maximum of 12
months apart -- 442 per
10,000 person-years, by
the age of 7 years
30. Health impacts modelling
âą The population attributable fraction (PAF) -- the
proportional reduction in morbidity that would occur if
exposure to air pollution was reduced or to an alternative
ideal exposure scenario (zero)
âą PAF =
(Risk estimateexposure difference -1)/Risk estimateexposure difference
âą Attributable number of cases =
PAF*expected asthma cases due to all causes
âą Sum-up all attributable number of cases across all census
tracts -- separately for each pollutant
31. Health impact modelling â LUR model
137 asthma cases per 10,000 person-years, by the age of 18 years
Pollutant Attributable cases Attributable cases lower CI Attributable cases upper CI Percentage of all cases
NO2 433 191 571 24%
NOx 685 -278 1191 38%
PM2.5 485 180 728 27%
PM10 608 277 861 33%
BC 276 112 444 15%
32. Health impact modelling â LUR model
123 per 10,000 person-years, by the age of 7 years
Pollutant Attributable cases Attributable cases lower CI Attributable cases upper CI Percentage of all cases
NO2 389 171 513 24%
NOx 615 -250 1070 38%
PM2.5 435 162 653 27%
PM10 546 249 773 33%
BC 248 100 398 15%
33. Health impact modelling â LUR model
442 per 10,000 person-years, by the age of 7 years
Pollutant Attributable cases Attributable cases lower CI Attributable cases upper CI Percentage of all cases
NO2 1398 615 1844 24%
NOx 2209 -898 3843 38%
PM2.5 1564 582 2347 27%
PM10 1961 894 2276 33%
BC 891 360 1432 15%
34. Health impact modelling â ADMS
137 asthma cases per 10,000 person-years, by the age of 18 years
Pollutant Attributable cases Attributable cases lower CI Attributable cases upper CI Percentage of all cases
COPERT
NO2
321 139 428 18%
COPERT
NOx
530 -201 976 29%
PHEM
NO2
317 137 423 17%
PHEM
NOx
523 -198 966 29%
35. Health impact modelling â ADMS
137 asthma cases per 10,000 person-years, by the age of 18 years
Pollutant
Attributable
cases
Attributable
cases lower CI
Attributable cases
upper CI
Percentage of
all cases
Cases attributable
to traffic
COPERT
NO2
394 173 520 22% 128 (7%)
COPERT
NOx
638 -256 1125 35% 219 (12%)
+ minor roads and cold starts
36. Summary
âą First full-chain HIA of TRAP and childhood asthma with
validation, as possible, at each stage
âą Comparing it to commonly used approaches (LUR models)
âą These results indicate that air pollution is responsible for a
large (up to 38% of all cases), but preventable, asthma
burden
âą This burden is higher than previous studies estimates ~ 14%
(Perez et al. 2013)
âą Different exposure assessments (pollutant selections,
exposure models) result in different estimated burdens
âą Different emission factors did not result in different
estimated burdens