Health risk assessment

of air pollution

General p rin c ip les

Health risk assessment
of air pollution

General p rin c ip les

Abstract
An air pollution health risk assessment (AP-HRA) estimates the health impact to be expected from mea-
sures that affect air quality, in different socioeconomic, environmental, and policy circumstances. As
such, it is an important tool for informing public policy decisions. This document introduces the concept
of AP-HRA, describes in broad terms how the health risks of outdoor air pollution and its sources are
estimated, and gives an overview of the general principles for the proper conduct of an AP-HRA for
various scenarios and purposes. The information is aimed at a broad audience of readers who do not
need to know how to apply the tools, but seek a general understanding of the concepts, scope and
principles of AP-HRA.

Keywords
AIR POLLUTION - adverse effects
AIR POLLUTION - analysis
RISK ASSESSMENT - methods
ENVIRONMENTAL MONITORING

Citation advice
Health risk assessment of air pollution – general principles. Copenhagen: WHO Regional Office for Europe; 2016.

Address requests about publications of the WHO Regional Office for Europe to:
Publications
WHO Regional Office for Europe
UN City
Marmorvej 51
DK-2100 Copenhagen Ø, Denmark
Alternatively, complete an online request form for documentation, health information, or for permission
to quote or translate, on the Regional Office website (http://www.euro.who.int/pubrequest).

ISBN 978 92 890 51316

© World Health Organization 2016

All rights reserved. The Regional Office for Europe of the World Health Organization welcomes requests for permission to reproduce
or translate its publications, in part or in full.

The designations employed and the presentation of the material in this publication do not imply the expression of any opinion
whatsoever on the part of the World Health Organization concerning the legal status of any country, territory, city or area or of its
authorities, or concerning the delimitation of its frontiers or boundaries. Dotted lines on maps represent approximate border lines for
which there may not yet be full agreement.

The mention of specific companies or of certain manufacturers’ products does not imply that they are endorsed or recommended
by the World Health Organization in preference to others of a similar nature that are not mentioned. Errors and omissions excepted,
the names of proprietary products are distinguished by initial capital letters.

All reasonable precautions have been taken by the World Health Organization to verify the information contained in this publication.
However, the published material is being distributed without warranty of any kind, either express or implied. The responsibility for
the interpretation and use of the material lies with the reader. In no event shall the World Health Organization be liable for damages
arising from its use. The views expressed by authors, editors, or expert groups do not necessarily represent the decisions or the
stated policy of the World Health Organization or any other institution mentioned in the report.

Text editing: Pat Butler

Design: Christophe Lanoux, Paris, France
Layout: Edb&Rdb di Daniela Berretta, Rome, Italy
Photo copyrights: cover and p. viii © Shutterstock-Robert F. Balazik; p. 22 © Shutterstock; p. 26 © Emilio M. Dotto; p. 28 © Fotolia.

iv
ii
 Contents

Acknowledgements............................................................................................ iv

Abbreviations..................................................................................................... v

Executive summary............................................................................................  vi

1. Introduction.................................................................................................... 1

1.1 Scope and purpose of this publication................................................ 1

1.2 Background.......................................................................................... 1

1.3 What is a health risk assessment?....................................................... 3

2. Definition of the policy question..................................................................... 5

3. What information is needed to conduct an AP–HRA?.................................... 6

3.1 Planning the health risk assessment.................................................... 6

3.2 Estimating population exposure to air pollutants................................. 8

3.3 Estimating the health risk..................................................................... 8

3.4 Quantifying the health impact.............................................................. 9

4. Uncertainty in AP–HRA................................................................................... 11

5. Tools available................................................................................................ 13

6. Conclusions.................................................................................................... 17

References......................................................................................................... 18

Annex 1. Useful AP–HRA resources................................................................... 21

Annex 2. The Aphekom project.......................................................................... 22

Annex 3. Global Burden of Disease...................................................................  25

Annex 4. BenMAP............................................................................................... 28

iii
Acknowledgements

This publication was developed on the basis of discussions at the WHO Expert
Meeting on Methods and Tools for Assessing the Health Risks of Air Pollution at Local,
National and International Levels, held in Bonn, Germany, on 12–13 May 2014. It was
prepared by Susan Anenberg (Environmental Health Analytics, LLC, Washington, DC,
United States of America), Marie-Eve Héroux (WHO Regional Office for Europe, Bonn,
Germany) and Susann Wothe (née Henschel) (Federal Institute for Occupational Safety
and Health (BAuA), Dortmund, Germany) as part of the work of the Task Force on Health
Aspects of Air Pollution under the Joint WHO/United Nations Economic Commission
for Europe (UNECE) Convention on Long-range Transboundary Air Pollution, in
accordance with the Memorandum of Understanding between UNECE and the WHO
Regional Office for Europe. The Regional Office thanks the Swiss Federal Office for the
Environment, the German Federal Ministry for the Environment, Nature Conservation,
Building and Nuclear Safety, the French Ministry of Social Affairs and Health and the
United States Environmental Protection Agency for their financial support of the work
of the Task Force. The work is coordinated by the WHO Regional Office’s European
Centre for Environment and Health, Bonn, Germany.
The experts who participated in the Expert Meeting are thanked for their invaluable
contributions, as well as their advice on a variety of health risk assessments. Menno
Keuken (TNO, Utrecht, the Netherlands) and Patrick Kinney (Columbia University,
New York, United States of America) provided comments on a draft version of this
publication.

iv
 Abbreviations

Aphekom Improving Knowledge and Communication for Decision

Making on Air Pollution and Health in Europe (project)
AP-HRA air pollution health risk assessment
AQG air quality guidelines
BenMAP Environmental Benefits Mapping and Analysis Program
CRA comparative risk assessment
CRF concentration–response function
DALY disability-adjusted life-year
EC European Commission
EPA Environmental Protection Agency
EU European Union
GBD Global Burden of Disease
HIA health impact assessment
NO2 nitrogen dioxide
NOx nitrogen oxides
O3 ozone
PM particulate matter
PM10 particulate matter with an aerodynamic diameter below 10 μm
PM2.5 particulate matter with an aerodynamic diameter below 2.5 μm
REVIHAAP review of evidence on health aspects of air pollution (project)
RR relative risk
SO2 sulfur dioxide
SOx sulfur oxides
TFH Task Force on Health Aspects of Air Pollution
YLD years lost due to disability
YLL years of life lost

v
Executive summary

This publication introduces the concept of air pollution health risk assessment (AP-
HRA), describes in broad terms how the health risks of outdoor air pollution and its
sources are estimated, and gives an overview of the general principles for the proper
conduct of an AP-HRA for various scenarios and purposes. The target audience
includes policy-makers at the local, national and international level, and other users of
health risk estimates in agencies, and research and advocacy groups.
Because the publication was prepared in the context of the work of the UNECE
Convention on Long-range Transboundary Air Pollution, the emphasis is mainly on
European and North American tools and references.
The main purpose of an AP-HRA is to estimate and communicate the health impact
of exposure to air pollution or changes in air pollution in different socioeconomic,
environmental, and policy circumstances. In many countries, an AP-HRA is formally
required as part of the decision-making process for new programmes, projects,
regulations, or policies that potentially have an effect on air quality. In other countries,
where assessments were previously limited to qualitative descriptions, scientific
advances are now allowing more detailed quantitative analyses of the health risks of
air pollution. Thus an increasing number of AP-HRAs are being carried out for a variety
of policy scenarios and geographical and time scales, using different methods.
A number of tools are available to conduct an AP-HRA. These tools are associated
with different workloads and require different levels of expertise. In selecting a tool,
it is important to first define the policy question to be answered and the audiences
to be informed. The technical needs of the assessment context, such as the relevant
pollutants, geographical scale and data requirements, should then be considered.
There are a number of online tools for AP-HRAs with a range of technical and
operational characteristics, and incorporating functions, equations, and often datasets.
Using these tools for AP-HRAs leads to better consistency, comparability and quality
assurance. In selecting an AP-HRA, the aim should be to maximize scientific rigour
within the resources available.
The first step of an AP-HRA is to assess the exposure of the target population to
specific air pollutants. Monitoring data may be used to estimate the past and current
exposure to air pollution for populations living near the monitoring site. In addition,
air quality modelling is often used to estimate differences in exposure for different
socioeconomic and environmental conditions within the geographical area of interest,
and to predict changes in exposure in future policy scenarios.
The second step of an AP-HRA is to estimate the health risk associated with the
exposure to air pollution. This requires the use of concentration–response functions
(CRFs), which quantify the health impact per concentration unit of a particular air
pollutant. Typically, these CRFs will have been established in epidemiological studies.
Results of AP-HRAs are often reported in terms of numbers of attributable deaths
or cases of disease, years of life lost, disability-adjusted life years, or change in life
expectancy attributable to exposure or a change in exposure to air pollution. These

vi
health impacts can then be used to evaluate costs and benefits of policy change
in monetary terms. An important limitation of this component of the AP-HRA is the
availability of baseline public health statistics for the targeted population.
The third step of an AP-HRA is to quantify and express the uncertainty of the generated
estimate of health impact. This is an important and integral component of the results,
and it is vital to ensure both that the main message is not lost and that the results
produced are understandable by policy-makers and others who do not necessarily
have a technical background or expertise in AP-HRA. The use of expert judgement
(consensus) on the level of confidence of the results is recommended. In addition,
the involvement of communication experts may be considered, to ensure effective
communication of the AP-HRA results.

vii
viii
x
1. Introduction
1.1 Scope and purpose of this publication
This publication provides a general the emphasis is therefore mainly on
introduction to the concept of air pollution European and North American tools and
health risk assessment (AP-HRA) and references.
the estimation of health risks from air
For more information, the report of a WHO
pollution and its sources, and highlights
expert meeting on AP-HRA, together with
general principles for conducting an
all background papers, is available online
AP-HRA in various policy scenarios
(Regional Office for Europe, 2014).
(WHO Regional Office for Europe, 2014).
Examples are presented illustrating The target audience for this publication
different aspects of the topics discussed includes policy-makers at local, national
in relation to different policy questions. and international level and other users
This publication has been prepared in of health risk information from various
the context of the UNECE Convention on sectors in agencies and research and
Long-range Transboundary Air Pollution; advocacy groups.

1.2 Background
Air pollution is an important determinant of Figure 1. Air pollution health pyramid
health (WHO Regional Office for Europe,
2006). Numerous epidemiological studies
have found an association between air
pollution and a wide range of adverse
health effects in the general population;
the effects have ranged from subtle
subclinical effects to premature death
as shown in Figure 1 (Samet & Krewski,
2007). premature death
hospital admission severity
Some groups – for example older adults, of
children, pregnant women and people emergency room visit
effect
with an underlying disease, such as physician office visit
asthma – may be more at risk, and may reduced physical activity
develop more severe health effects more medication use
quickly when exposed to air pollution. In
respiratory symptoms
addition, certain groups may be exposed
to higher levels of outdoor air pollution, impaired lung function
e.g. people living near busy traffic routes subclinical (subtle) effects
or those in specific occupational or
socioeconomic groups (WHO Regional proportion of population affected
Office for Europe, 2005).
Pollution in ambient air is generally a Source: adapted from Samet & Krewski (2007),
complex mixture. Consequently, the reproduced by permission of Taylor & Francis Ltd.
adverse health impacts observed in

1
 epidemiological studies and attributed Europe, 2013). Different particle sizes,
to an individual air pollutant may actually composition, or characteristics can be
be partly due to other pollutants in related to specific emission sources
the mixture. The air pollutants often better than other air pollutants and
investigated in these studies – particulate may therefore be considered a (more)
matter (PM), black carbon, ozone (O3), suitable indicator. Thus, PM10 may be an
nitrogen dioxide (NO2), nitrogen oxides appropriate indicator when considering
(NOx), sulfur dioxide (SO2), carbon the impact of resuspension of road dust,
monoxide, heavy metals or black smoke while black carbon is a more sensitive
– may be proxies for the air pollutant indicator for exhaust emissions from road
mixture. This issue is particularly relevant traffic (Keuken et al., 2012). It is therefore
in relation to the health impact of exposure important in an AP-HRA to select the
to PM in ambient air. PM originates appropriate pollutants for the sources
from primary emissions (e.g. soot from that are relevant to the exposure of the
combustion sources, sea salt and soil targeted population. PM2.5 has been
from wind-driven resuspension) and investigated in many epidemiological
formation of secondary particles in the studies, and has been shown to be a
atmosphere. PM may be characterized robust indicator of risk associated with
in terms of the mass concentration of exposure to PM from diverse sources
particles smaller than 2.5 µm (PM2.5) or and in different environments (Lim et
10 µm (PM10), the number of particles al., 2013). Figure 2 shows a schematic
(ultrafine), or the chemical composition overview of the relative sizes of PM10 and
(e.g. black carbon, organic compounds PM2.5 in relation to a human hair and fine
and heavy metals). Epidemiological and beach sand (US EPA, 2008).
toxicological evidence shows that PM
While much has been done to improve air
mass (PM2.5, PM10) comprises fractions
quality and, consequently, human health
with varying types and degrees of
in many parts of the world, evidence for
health effects (WHO Regional Office for

Figure 2. Schematic overview of the relative size of particulate

pollution, PM10 and PM2.5

PM2.5 μm
Combustion particles, organic
HUMAN HAIR
compounds, metals, etc.
50–70 μm
<2.5 μm (microns) in diameter
(microns) in diameter

PM10
Dust, pollen, mold, etc.
<10 μm (microns) in diameter

90 μm (microns) in diameter
FINE BEACH SAND

Source: US EPA, 2008.

2
adverse health effects persists at levels local, national, and global policies to
below the current air quality standards reduce air pollution.
and historically low levels of air pollution
An AP-HRA can aid this process by
in many countries. In addition, air
answering specific policy questions.
pollution is of increasing concern in many
Indeed, in many countries it is required as
developing countries, where emissions
part of the decision-making process for
have been rising in the absence of strict
new programmes, projects, regulations,
air quality policies. This has resulted in
and policies aimed at improving air quality
several episodes of poor air quality, in
that may affect air quality as a side-
particular in urban areas (Health Effects
effect. In many other countries, it may be
Institute, 2010). Quantitative estimates
conducted as part of an assessment or
of the health impact of air pollution have
research project, even though there is no
become increasingly important, to allow
legal requirement (WHO Regional Office
policy-makers and other stakeholders
for Europe, 2014).
to devise and implement more effective

1.3 What is a health risk assessment?

A health hazard can be defined as a 2010). It should also be noted that AP-
source of risk to human health or well- HRAs generally include only the subset
being (Department of Health, 2006). A of health impacts that can be quantified,
health risk assessment is the scientific and do not deal with other health effects
evaluation of potential adverse health for which no CRF is available. Ideally, to
effects resulting from human exposure to protect public health, an AP-HRA should
a particular hazard. In the context of this be as inclusive as possible; however, in
publication, the health hazard of interest most cases, it is likely that the HRA will
is air pollution. underestimate the actual risk.
An AP-HRA aims to estimate the risks As an analytical tool, an AP-HRA can
of past, current or future exposure to air be used as part of a comprehensive
pollution and of changes in exposure that assessment of the health impacts of
may result from planned policies or other policies, programmes, and projects
modifications of air quality (Department of that affect environmental conditions – a
Health, 2006; HIP, 2014). An AP-HRA may health impact assessment (HIA). AP-
be quantitative or qualitative; it generally HRA and HIA are different concepts,
assesses (i) the amount of air pollution although the two terms are sometimes
present, i.e. pollutant concentrations, (ii) used interchangeably. An HIA may
the amount of contact (exposure) of the be defined as follows (WHO Regional
targeted population, and (iii) how harmful Office for Europe, 1999): “a combination
the concentration is for human health, i.e. of procedures, methods and tools by
the resulting health risks to the exposed which a policy, programme or project
population (WHO, 2010). The estimates may be judged as to its potential effects
provided by an AP-HRA are intended to on the health of a population, and the
inform the decisions of policy-makers or distribution of those effects within the
other stakeholders. population.” An HIA identifies appropriate
actions to manage those effects (Quigley
The required input data for an AP-
et al., 2006).
HRA (e.g. air pollution, baseline health
statistics, CRFs of air pollutants) are While an HRA tends to look into
not always available, and many risk particular hazards and their effects on
assessments have to be based on human health, an HIA takes a broader
estimates or judgements of some of the perspective. For example, when planning
data inputs or characterizations. As a the construction of a new industrial site
result, HRA outcomes generally have in or near a city, an HIA would look into
associated uncertainties, which should not only the specific risks associated
be characterized as far as possible (WHO, with possible air pollutants, but also

3
 issues such as noise and soil and water accumulation of hazards with already
pollution, as well as the potential impact existing hazards in the area.
on the population of the city through,
The focus of this publication is on AP-
for example, the influx of construction
HRA considering the definitions provided
workers, employment opportunities for
above. An overview of the steps involved
inhabitants of the city, and the possible
in an AP-HRA is given in Figure 3.

Figure 3. Overview of an AP-HRA process (Quigley et al., 2006; US

EPA, 2012; WHO Regional Office for Europe, 2014a)

Define the policy question with respect to risk to health of exposure to
the health hazard (air pollution)

Plan the health risk assessment
Questions ➔ subsequent actions
Who is at risk? ➞ Define the population and geographical scope
How can exposure best be described? ➞ Describe the spatial resolution and selection of
air pollutants of concern
What are the health effects? ➞ Identify relevant long– and short–term effects

Select appropriate AP–HRA tool or model for the specific assessment context

Conduct the AP–HRA
 Input for the tool

Population data 
(census, estimates or Population estimates
projections)

Air quality data  Input for the tool

(measured or modelled) Exposure estimates 

Baseline rates of  
Health data Health function  health outcomes
(recorded or projections) (concentration-response in the population
relationship for a particular studied
health outcome)

Output of the tool
Adverse health effects
Quantification of health risk
from exposure to
ambient air

ESTIMATES AND RESPONSES TO POLICY
QUESTIONS TO ALLOW INFORMED DECISION

4
2. Definition of the
policy question
As outlined above, the main purpose of or selected economic sectors, and
an AP-HRA is to answer policy questions what are the benefits of policies
about the likely health impacts of planned related to them?
policies or modifications of air quality.
4. What are the human health impacts
AP-HRAs are often used to answer of current policy or implemented
the following policy questions (WHO action?
Regional Office for Europe, 2014).
5. What are the policy implications of
1. What is the public health burden the uncertainties of the assessment?
associated with current levels of air
pollution? The results of an AP-HRA can be used
in an estimation of the economic value of
2. What are the human health benefits health benefits resulting from a change in
associated with changing an air policy. Some AP-HRA tools incorporate
quality policy or applying a more this step. The knowledge gained
stringent air quality standard? through an AP-HRA can also be used to
improve policies, such as increasing the
3. What are the human health impacts
stringency of air quality standards.
of emissions from specific sources

5
 3. What information is
needed to conduct an
AP-HRA?
3.1 Planning the health risk assessment
Figure 4 shows a schematic decision-tree response functions). The selection of the
for an AP-HRA process, covering definition method may depend on data availability
of the policy question, determination of or may determine the data requirements.
the availability of data and resources, In addition, different tools will entail
and selection of appropriate methods different workloads and require different
and tools. Input data are required on, levels of expertise. A detailed overview of
for example: (1) the level of air pollution, the various tools available is presented in
(2) the exposed population, and (3) the section 5.
health outcome affected (concentration–

Figure 4. Decision-tree showing the sequence of choices and feedback

loops when conducting an AP-HRA

What is the policy question

Reframe the policy question
to be addressed?

What is the most appropriate Go back and select

method or tool for addressing a different tool
this question?

Do I have the data and There is an alternative tool

resources to use this that can be supported by
method/tool? the available.

Yes No The data available are not

suitable for the selected
tool

What are the best data to There is no data available

use?

There is no alternative tool

that can be supported by
Select best data the available.
supporting the chosen tool
Can new data be collected?

Conduct AP-HRA Yes No

6
The following factors should be considered with respect to data needs and availability,
depending on the question to be answered (WHO Regional Office for Europe, 2014).

1) The policy question and the event 2) The selection of the tool will define
or condition of interest will define the data needs with respect to the
the data needs with respect to the following.
following.
• What spatial resolution of the air
• Who is affected? For example, is pollution and population data is
there a need to consider specific needed? Will census data be needed
subpopulations in the targeted for a city, a specific location in a city,
population, such as certain age groups a specific region or the country as a
(children, elderly), people susceptible whole?
to specific diseases, particular
• What temporal resolution of the air
occupational or socioeconomic
pollution data is needed: hourly, daily
groups?
or annual averages?
• How are people affected by air
• What temporal resolution of the health
pollution? What health outcome will be
data is needed? Is there a need, for
assessed in the AP-HRA, e.g. mortality,
example, for daily number of incidents
hospital admissions, incidence rate of
or hospital admissions, daily mortality
a specific disease or work loss?
or annual mortality?
• Which key pollutant indicators are
• Are the data needed for the AP-HRA
to be considered to describe the
available? Is there a database that
exposure and estimate the health risk
meets the data needs, such as baseline
for a specific population in a specific
public health and population data?
situation? This will depend on whether
the policy question is concerned with, o Are measurements of air pollution
for example, air pollution emissions available, or modelled data? Are
from specific sources, the effect of the tools available to model relevant
implementation of specific legislation exposure data in a temporal and
or air quality in general. If measured spatial resolution that allows the
air quality data are used, what type exposure of the affected population
of measurement data is needed, to be described? Are the data
e.g. urban background levels, traffic of adequate quality, were proper
emissions or stationary industrial monitoring protocols used and did
measurements? they undergo a quality assurance or
control process?
• What is the spatial resolution of the
issue or question to be assessed in • Have there been previous studies
the AP-HRA: one specific city, specific describing the concentration–response
locations within a city, multiple cities, relationship for the health outcomes of
a region, the whole country or an even interest?
broader area? • How many people in the population of
• What is the temporal resolution of the interest are affected by a specific health
issue or question to be assessed in outcome caused by air pollution?
the AP-HRA: specific decades, years, What are the baseline statistics for
seasons or days; the period before, the health outcome, i.e. how much of
during or after a specific event; or a the observed adverse health effect or
comparison of historical data with change in a specific health outcome
future projections? can be assigned to air pollution or to
changes in air quality? Are there data
on a control area or control population
that can be used for comparison?

7
 3) Once the desired data have been This involves the following steps.
identified, the availability of the data
a. Estimate the exposure for the assessed
has to be assessed.
population.
If all the desired data are available, it is
b. Use the exposure estimates and
possible to proceed to the next step.
baseline health outcome rates as input
Otherwise either more data will have to
data for a function describing the
be gathered, a different tool will have
concentration–response relationship.
to be selected or a different policy
This will allow the health risk associated
question will have to be asked.
with the estimated exposure to be
assessed for the population.
4) The available, compiled data on
exposure to air pollution, health and c. As an optional additional step, an
population are then used to assess economic evaluation to quantify the
the health impacts associated with monetary cost or benefits of the health
exposure to air pollutants with respect impacts may be conducted. Some AP-
to the specific policy question to be HRA tools incorporate this step.
answered.

3.2 Estimating population exposure to air

pollutants
Data on population exposure to combination with existing ground-based
air pollutants generally come from monitoring has made information on
monitoring by local or national key air pollutant indicators increasingly
institutions. Estimations of population available, including in some of the most
exposure based on measured air pollution highly polluted and data-poor regions
data are often limited by the restricted (Brauer et al., 2012; Hoek et al., 2008;
geographical and time coverage. In Paciorek & Liu, 2012; UNECE, 2010; van
addition, it may be difficult to reconcile Donkelaar et al., 2010).
data from different locations since
When estimating the change in
measurements are often made using
population exposure as a result of a
different procedures and techniques
hypothetical change in emissions or
(WHO Regional Office for Europe, 2014).
pollutant concentrations, monitoring
Even the methods used in one location
data may be used as a baseline level. Air
may change with time. Recent progress
quality modelling is, however, needed to
in combining satellite remote sensing,
estimate future concentration changes
global chemical transport modelling,
resulting from policies and technological
land use regression models and high-
innovations.
resolution local dispersion models in

3.3 Estimating the health risk

The risk of air pollution to health in a visit, hospital admission) occurring in a
population is usually represented by a population exposed to a higher level of
concentration–response function, which air pollution relative to that in a population
is typically based on Relative Risk (RR) with a lower exposure level. Typically,
estimates derived from epidemiological RR is expressed as the proportional
studies. The RR estimate describes increase in the assessed health outcome
the likelihood of an adverse health associated with a given increase in
outcome (e.g. premature death, heart pollutant concentrations in μg per m3
attack, asthma attack, emergency room or parts per billion (ppb) (Katsouyanni,

8
2003). It is important to note that the RR and health care systems may be very
estimate cannot be assigned to a specific different in other places, and this may
person; it describes risk in a defined affect the CRF.
population, not individual risk (Australian
All these factors mean that, in certain
Department of Health, 2012; McAuley &
assessment contexts, the absence of
Hrudey, 2006).
direct epidemiological evidence about
In order to provide useful advice aimed the health risk of exposure to air pollution
at answering a specific question, the AP- is an important limitation. In some of
HRA assesses a specific health endpoint the most highly polluted regions of the
or set of health endpoints in a specific world, there is a severe lack of direct
population. The analysis does not cover epidemiological evidence. Studies are
the full range of possible adverse health urgently needed in these areas, because
effects in all possible groups of the the health response per unit change
population. in air pollution at such high levels may
differ from that seen in countries with
The CRFs used in AP-HRA tools are
lower pollution levels. For regions with
typically based on the epidemiological
limited or no epidemiological evidence,
evidence available for a specific health
information from studies in other parts
outcome. Some are based on evidence
of the world may be used to conduct an
from experiments in which people or
AP-HRA. However, such extrapolated
animals are deliberately exposed to
information may not accurately describe
a pollutant (WHO Regional Office for
the concentration–response relationship
Europe, 2014). The CRF may therefore
in the region to be assessed, leading to
be refined as new scientific evidence
uncertainties in the results (see section
becomes available. For some specific
4.1) (WHO Regional Office for Europe,
health endpoints or air pollutants, the
2014).
available data may be limited or old and
no longer considered appropriate, so Some tools allow the user to select
that it may not be possible to describe the CRFs to be applied to the specific
the concentration–response relationship. assessment, whereas in other cases
For example, the likely health risks the recommended CRFs are directly
of exposure to ultrafine particles are embedded in the tool.
currently not considered, as there is no
When generating and communicating
reliable CRF available (Hoek et al., 2010).
AP-HRA results for a specific health
In some cases, CRFs available may not
endpoint, it should be kept in mind that
be appropriate for very high and very low
the effects of long-term exposure are
concentrations. Finally, it is important
much greater than those observed for
to note that most studies have been
short-term exposure (WHO Regional
carried out in Europe and North America.
Office for Europe, 2013).
Pollution levels, chemical composition

3.4 Quantifying the health impact

Results of AP-HRAs are often reported status of a population (Murray & Lopez,
in terms of number of the attributable 2013). It is important to note that these
deaths or cases of disease, years of life metrics provide expected values for a
lost (YLL), disability-adjusted life years whole population and cannot be applied
(DALYs), or change in life expectancy to individuals in that population.
attributable to total exposure to air
pollution or a change in exposure (WHO Number of attributable deaths or
Regional Office for Europe, 2014)). cases of disease. This is calculated
These metrics aggregate different types as the difference in number of deaths
of health impact and can be used to or cases of diseases between the
highlight different aspects of the health incidence/rate at the exposure measured

9
 over a specific period and that at baseline average duration of the case until
exposure, e.g. difference between remission or death (years) and a disability
current disease incidence and historical weight factor that reflects the severity of
incidence or projected future incidence, the disease on a scale from 0 (perfect
or total health risk (in relation to zero health) to 1 (dead) (WHO, 2014). The
exposure or to some assumed threshold GBD 2010 study used an updated life
value) (WHO Regional Office for Europe, expectancy standard for the calculation
2014). of YLL and based the YLD calculation
on prevalence rather than incidence
Years of life lost. YLL is a measure of the
(WHO 2014). The Prevalence YLD were
years of life lost as a result of premature
estimated by multiplying the number of
death. In simplified terms, the calculated
prevalent cases by the disability weight
number of deaths attributable to changes
factor.
in exposure to air pollution is multiplied by
the standard life expectancy at the age Disability-adjusted life years. One
at which death occurs. In some cases, DALY is one lost year of healthy life. The
social value weights are also applied sum of DALYs across a population – the
(WHO, 2014). Social value weights include burden of disease – can be thought of
disability weights (used in the calculation as a measurement of the gap between
of YLD (see below)), time discounting actual health status and an ideal situation
and age weights, which assign different in which the entire population lives to
values to the time lived at different ages, an advanced age, free of disease and
to reflect varying societal roles and disability. Total DALYs for a particular
changing levels of dependence with age disease or health condition in a population
(Murray, 1994; WHO, 2014). Discounting are calculated as the sum of YLL and
takes into account whether a year of YLD (WHO, 2014; Murray & Lopez, 2013).
healthy life gained now is worth more
These estimates of impacts can be
to society than one gained sometime in
used for further estimation of the
the future, while age weighting reflects
monetary costs and benefits in a health
the fact that lost years of healthy life are
benefits analysis. Some AP-HRA tools
valued more at some ages than others
incorporate this economic valuation step
(Murray & Acharya, 1997; SA Health,
or it is conducted in a separate step
2003). An example of a study that applied
after the AP-HRA. This health benefits
social value weights is the Global Burden
analysis yields the economic value of the
of Disease Study.
change in health impacts. If it is a positive
Years lost due to disability. YLD change, then it is considered a benefit.
measures years lost due to disability. It If it is negative, it could be considered
is estimated by multiplying the number a cost. This economic valuation is not
of incident cases of a particular health considering the implementation costs of
outcome in a particular period by the the policy.

10
4. Uncertainty in
AP-HRA
The uncertainty of an assessment is a number of countries are combined.
related to a lack of knowledge about one In addition, uncertainty arises when
or more components of the assessment projections are made of population
(US EPA, 2011). Uncertainty analysis size and deaths in the future.
is an instrumental part of any scientific
3. Pollution exposure level
analysis, and is usually limited to
components that are already identified as Because there is no full geographical
uncertain (“known unknowns”). coverage of ground monitors, most AP-
HRAs rely to some extent on modelling
It is a challenging yet important task to
to estimate exposure. Modelling is
find a balance between the complexity
also needed for estimates of future
of information and tools used and the
exposure based on predicted changes
need to produce understandable results
in air pollution as a result of new policies
for policy-makers and others who do not
or technological improvements. Since
necessarily have a technical background
air quality models are based on a set
or expertise in the field (WHO Regional
of assumptions, it is not possible to
Office for Europe, 2014).
be certain that the estimated exposure
The key sources of uncertainty in an AP- coincides with the actual ambient
HRA are listed below (WHO Regional concentrations in a given location.
Office for Europe, 2014). Even if there could be full coverage by
ground monitors, all AP-HRAs assume
1. Air pollutants exist as a complex that either measurements made at a
mixture specific location or model estimates
Despite great improvements in the of average exposure over a particular
science underlying AP-HRAs, it is still area are representative of the exposure
not possible to know with complete of the targeted population. Even if
certainty the effects of air pollution population exposure is well estimated,
on health (WHO Regional Office for individual exposures can vary
Europe, 2014). There is a considerable substantially, as a result of differences
body of evidence from epidemiological in concentrations at different places
studies in various parts of the world as well as individuals’ own activity
documenting a wide range of adverse patterns. Personal monitoring is
health effects associated with ambient generally necessary to assess
air pollution. However, the observed individual-level risks.
adverse effects attributed to an 4. The concentration–response
individual air pollutant may actually be function
(partly) attributable to other pollutants
in the mixture which are correlated with CRFs are derived from epidemiological
the assessed pollutant (WHO Regional studies, in which assumptions made
Office for Europe, 2013). The resulting during the analysis inevitably introduce
uncertainty in the outcome of the AP- some uncertainty into the results. In
HRA may be considered “unknown addition, epidemiological evidence
unknowns” and is not included in this on air pollution is scarce or absent
document. in some parts of the world. Most
epidemiological studies have been
2. Baseline disease burden conducted in developed countries,
The number of deaths or cases of and the range of exposures studied
disease may be uncertain for a variety does not necessarily represent what is
of reasons, especially when data from observed around the world.

11
 For many public health and policy of air pollution is not an uncertainty in
decisions, the mortality attributable the same sense as those previously
to ambient air pollution has to be discussed. The results of the AP-HRA
considered in the context of mortality are sensitive to this choice, but they are
due to other factors. A comparative not themselves made uncertain by it.
risk assessment (CRA) is a type of
6. Deliberate simplifications of the
AP-HRA that provides comparable
model
estimates for the various risk factors;
it requires that consistent approaches Practical considerations may require
are used to estimate the various risks the use of a simplified model, which
(WHO Regional Office for Europe, can lead to increased uncertainty
2014). The GBD 2010 project, (WHO, 2005).
coordinated by the Institute for Health
Metrics and Evaluation (IHME), carried It is important that these various sources
out a CRA to compare the burden of uncertainty are acknowledged and
of disease from various risk factors, described as fully as possible, to give a
including air pollution (Lim et al., 2013). sense of the precision of the estimates.
The project developed integrated Uncertainties arise mainly from the
exposure–response (IER) functions current understanding of CRFs and from
that combined evidence from studies exposure estimates. In addition, there
of ambient air pollution, second-hand are probably components of uncertainty
smoke, household air pollution and that are not recognized. Despite these
active smoking to estimate risk from uncertainties, an AP-HRA can provide
ambient air pollution over the entire useful and valid results. Therefore,
range of exposure (Lim et al., 2013; while it is important to communicate
WHO Regional Office for Europe, uncertainties, the message should be
2014). balanced to ensure that decision-makers
5. The counterfactual level of air take the results of the AP-HRA seriously.
pollution The multiple sources of uncertainty may
be quantified by Monte Carlo simulations
The counterfactual level of air pollution
or probabilistic simulations, as was done
is a baseline or reference exposure
in the GBD 2010 study (Lim et al., 2013).
against which the health impacts of
air pollution are calculated (WHO,
2014b). This level of air pollution may Confidence in AP-HRA estimates
be defined differently in different The results of the uncertainty analysis
AP-HRAs, depending on the policy are usually presented as ranges, with
question to be answered. It may, for the focus being on confidence intervals
example, be defined as the national around the mean (WHO Regional Office
air quality standard, the WHO air for Europe, 2014). Conventionally, 95%
quality guideline (AQG) level, the confidence intervals are used to provide
natural level (i.e. without anthropogenic an estimate of the precision of the results.
influence) or the lowest level observed This interval is the range of values within
in epidemiological studies (WHO which there is a 95% probability that
Regional Office for Europe, 2014). the true value lies (Scott, 2008; US EPA,
Uncertainty in the counterfactual level 2015).
may be due to imperfect knowledge
about the exact effect of some previous For example, as outlined in Annex 3,
policy change or a theoretical minimum the uncertainty in the estimates in the
level of pollution (WHO Regional Office GBD 2010 study (Lim et al., 2013) was
for Europe, 2014). It should be noted presented as numerical ranges, i.e.
that choosing a counterfactual level confidence intervals around the mean.

12
 5. Tools available
Computer-based tools are now available Automated tools are typically preloaded
that automate the process of an AP-HRA. with health and demographic data and
These tools offer several advantages to concentration-response functions, and
the practitioner and end-user, including some allow for user-specified inputs.
simplicity (lowering the barrier to However, the tools vary in many aspects,
conducting assessments), consistency, and analysts should choose the one that
comparability among assessments, and most closely matches the context of the
quality assurance. assessment.
Most tools use similar approaches, In preparation for a WHO Expert Meeting
relying on epidemiologically derived in May 2014, health risk assessment
concentration–response functions and tools were surveyed to ascertain their
population-level exposure estimates technical and operational characteristics.
to determine the proportion of cases of Information was collected on 12 air
a particular health effect that may be pollution health risk assessment
attributable to a change in air quality. tools (Table 1). Detailed tables of tool

Table 1. Air pollution health impact assessment tools

Tool Developing Geographical scope Health endpoint

institution addressed1
Global (42 cities, additional 3000
AirCounts Abt Associates Mortality
under development)
World Health Any population with specified size,
AirQ2.2 (update under development) Mortality and morbidity
Organization mortality and morbidity characteristics
French Institute of Public Global (current version
Aphekom Mortality and morbidity
Health Surveillance focuses on Europe)
Northern hemisphere, continental
Economic Valuation of Air Pollution (EVA) Aarhus University Mortality and morbidity
(e.g. Europe), national, city
EcoSense University of Stuttgart Europe Mortality and morbidity
Environmental Benefits Mapping and
US Environmental Protection Continental USA and China pre-
Analysis Program – Community Edition Mortality and morbidity
Agency defined; any other as defined by user
(BenMAP-CE)
Environmental Burden of Disease (EBD) World Health
Global Mortality and morbidity
Assessment tool for ambient air pollution Organization
GMAPS2 World Bank Global Mortality and morbidity
Can be used anywhere where there
Institute of Occupational is background mortality data and
IOMLIFET Mortality and morbidity
Medicine measured or predicted pollutant
concentrations
US Environmental Protection
Under development for all countries
Rapid Co-benefits Calculator Agency, Stockholm Mortality
globally
Environment Institute
SIM-Air Urban emissions Asia, Africa, Latin America Mortality
European Commission Joint
TM5-FASST Global (56 source regions) Mortality and morbidity
Research Centre
1 Morbidity may include, for example, cardiovascular diseases, respiratory diseases, hospital admissions, emergency room admissions,
days of restricted activity, and work loss days. Not all tools address all morbidity outcomes.
2 The model itself is no longer actively maintained and therefore no longer available for download.

13
 characteristics can be found elsewhere • Exposure characterization. Most
(WHO Regional Office for Europe, 2014) tools rely on air quality modelling to
and examples provided in Annex. Most of estimate exposure, although some
these tools are available without charge. may also be able to take observations
from air quality monitors. Some
Often the first factor that must be
of these tools use full air quality
considered is the tool’s geographical
modelling, which accounts for the
scope, or the spatial coverage or extent
complex atmospheric chemistry and
of the tool as currently configured.
transport governing air pollution and
Geographical scope is distinct from
simulates the influence of emission
spatial resolution, which is the degree
controls on air pollution levels. When
of granularity allowed by the tool. For
air quality modelling is unavailable,
example, a tool with global scope may
reduced-form tools can generate
have a national-scale resolution, city-
broad-scale estimates of the impact of
scale resolution, or a gridded resolution.
air pollution from built-in relationships
The characteristics of the available tools between emissions and the exposure
should also be considered and matched metric (often concentration) derived
against the needs of the assessment from externally conducted air quality
context to select the most appropriate model simulations. Care must be taken
one for addressing the policy question. to match the spatial resolution of the
The key technical factors of the tools are assessment context, the air quality
listed below. model, and the epidemiological inputs
• Pollutants addressed. Most tools are to the health impact function as closely
preconfigured to assess the effects of as possible.
PM (PM2.5 and PM10) and ozone. Some • Data sources. Health impact
also include NOx, sulfur oxides (SOx), assessments typically rely on
carbon monoxide, heavy metals or information about population size and
black smoke. characteristics (e.g. age distribution),
• Health outcomes quantified. All baseline mortality and disease
the tools reviewed assess impact on incidence rates (usually derived from
premature mortality in terms of the country statistics), and concentration–
number of excess or avoided deaths. response functions (usually derived
Many tools can also quantify the from epidemiological studies). Some
number of life years lost, disability- tools are flexible enough to allow users
adjusted life years (healthy life years to input data from any source. Others
lost) and cases of disease (e.g. chronic are preconfigured with data from
obstructive pulmonary disorder). specific sources, and users must decide
whether those datasets are appropriate
• Resolution. Some tools assign air for their assessment context.
quality values to a grid, which divides
the geographical scope into cells The available tools also have a range
(either uniform or variable in shape). of operational characteristics, and
Population exposure and health their abilities and constraints, including
impacts are quantified separately resources and expertise, should be
for each cell. Other tools assign air considered. The key operational
quality data to areas within geopolitical characteristics are listed below.
boundaries, such as countries, • Format. Some tools are client-based
provinces, and cities. Ideally, the spatial software programs, which have to be
resolution of the tool should match the downloaded and installed by users.
spatial resolution of the assessment These tools include extensive datasets
context (e.g. a tool with city-level or of health impact functions, population,
finer resolution should be used to and health data, which may be modified;
assess the effects of air pollution in these tools are generally complicated
cities). and users may need to invest time

14
 and resources in learning how to use considerations when conducting AP-
them. Other tools depend on external HRAs: (1) the user input required
software of general application (e.g. to characterize exposure (pollutant
Microsoft Excel), which is generally emissions or concentration levels); and
accessible to most users but may need (2) spatial resolution (e.g. regional (more
to be purchased. Since many analysts than one country), national, or city-level).
are familiar with Microsoft Office, A third factor that often needs to be
extra training may not be necessary. considered is the pollutants addressed
A few tools are web-based, allowing by the tool. Table 2 thus indicates, on
users to generate air pollution health the basis of these three factors, which
impact estimates without downloading available tools are appropriate in the
or installing a program. Web-based context of a given assessment. Analysts
tools may be most accessible to non- would then need to ensure that the other
technical users, particularly in countries technical and operational characteristics
that lack the resources to conduct of the tools are consistent with their
full-scale, detailed, and refined health needs and capabilities.
impact assessments. Some tools offer
The use of Table 2 can best be
online tutorials and training workshops
demonstrated through several fictional
(e.g. BenMAP-CE, IOMLIFET, SIM-Air).
examples.
• Complexity. The tools described
1. An international development
here vary in technical complexity and
organization is interested in estimating
accessibility. Users will need to find a
the health benefits of PM2.5 reductions
balance between their ability to deal
associated with improved public
with technical complexity and the level
transport systems in cities. It has
of specificity called for in the policy
projections of emission reductions
context.
but does not have the resources
• Degree of peer-review and use to simulate the resulting PM2.5
in policy settings. Analysts should concentration changes. Table 2
consider whether the tool has been indicates the tools that can (a) read
peer-reviewed, the extent to which in emission estimates, (b) have city-
it has been used to inform policy, level resolution, and (c) quantify PM2.5
and whether it is open-source or health impacts: AirCounts, SIM-Air,
proprietary. Some tools have been Aphekom, and EcoSense.
externally peer-reviewed and have
2. An analyst working for an environment
been used extensively in support of
ministry in a developed country wants
national air quality regulations (e.g.
to assess the national health benefits
US EPA National Ambient Air Quality
of a power plant emission regulation
Standards). A critical advantage of
that is expected to reduce both
open-source tools is that they are
PM2.5 and ozone. She has resources
fully transparent, allowing analysts
to estimate emissions and simulate
to evaluate the underlying algorithms
concentration changes. Table 2 shows
and datasets used to calculate impact
that she can use any of the tools
(Anenberg et al., 2015).
that read in concentrations, except
• Degree of maintenance. Analysts EBD which does not address ozone:
should consider whether the tool is BenMAP-CE, AirQ2.2, IOMLIFET, and
maintained as a living tool, with updates EVA.
of datasets and methods over time, or
3. An analyst working for an environment
is fixed. The data inputs required for air
ministry in a developing country wants
pollution health impact assessments
to estimate the national PM2.5-related
should be updated over time to reflect
health benefits of adopting new diesel
changes in the science.
vehicle emission standards, but has no
Table 2 classifies the surveyed tools resources to run air quality modelling.
according to two of the most common The analyst might consider using a

15
Table 2. Classification of available tools according to user input needed to
characterize exposure, spatial resolution and pollutants dealt with

User input Emissions Concentration

needed to
characterize
exposure1

Spatial Regional National City Any

resolution

Pollutant PM2.5 Ozone Other PM2.5 Ozone Other PM2.5 Ozone Other Any

AirCounts SIM-Air - SIM-Air Co- Co- TM5- AirCountsTM Aphekom SIM-Air BenMAP-CE
(PM10) benefits benefits FASST SIM-Air EcoSense (PM10) AirQ2.2
Calculator Calculator (NOx, SOx) Aphekom EcoSense IOMLIFET
TM5- TM5- EcoSense EcoSense (NOx, SOx, EVA
FASST FASST (NOx, SOx, CO, heavy EBD (no ozone)
EcoSense EcoSense CO, heavy metals,
metals, dioxins,
dioxins, radio-
radio- nucleotides)
nucleotides)

1 Tools that read in emissions datasets are often considered “reduced-form” tools, as they can generate broad-scale estimates of the impact of
air pollution from built-in relationships between emissions and the exposure metric (often concentrations) derived from externally conducted
air quality model simulations. Tools that read in concentrations require the analyst to generate concentration datasets externally (either from air
monitoring or air quality modelling simulations). One tool (GMAPS) reads in economic and climate indicators from a reduced-form econometric
model and is not included in this table.

tool that reads in emissions, runs at using a reduced-form tool that uses
a national resolution, and quantifies emission data to forecast impact may be
PM2.5 health impacts: Co-benefits sufficient. Reduced-form tools use built-
Calculator, TM5-FASST, or EcoSense. in parameterizations, avoiding the need
for expensive and resource-intensive
The various AP-HRA tools have made
chemical transport modelling. For
it easier for analysts to respond to a
example, a reduced-form tool would be
range of policy questions by conducting
helpful in estimating the health benefits
different types of assessment in a
of different approaches to emission
consistent and reliable manner. While
reduction in countries where regional air
analysts should strive to use the
quality modelling is not available. Even
most technically sound methods for
where high quality data exist, reduced-
conducting assessments (e.g. using air
form tools can be used to screen a large
quality modelling to simulate changes in
number of scenarios, to determine which
pollutant concentration associated with a
should be evaluated in greater detail. In
certain reduction in emissions), technical
general, analysts should use the AP-HRA
refinement often comes at the expense
tools that provide the maximum degree
of accessibility (because air quality
of technical rigour within the resources
modelling is technically demanding and
available.
resource-intensive). In some cases,

16
6. Conclusions
The characterization of health risks in a scale, to allow the most appropriate tool
population from ambient air pollution is to be chosen.
critical to the development of effective
It is often simpler to use an already
risk management policies and strategies
available automated tool than to develop
(Samet & Krewski, 2007).
a new model for each assessment; this
An AP-HRA can quantify the health also improves consistency, comparability
impact of air pollution or of changes among assessments, and quality
in air pollution resulting from different assurance. The available tools have a
socioeconomic, environmental, or policy range of technical characteristics (e.g.
circumstances. In many countries, an geographical scope, spatial resolution,
AP-HRA is formally required as part of pollutants addressed, health outcomes
the decision-making process for new quantified, method of characterizing
programmes, projects, regulations, exposure) and operational characteristics
and policies that may affect air quality. (tool format, complexity, degree of peer-
It is, therefore, important for decision- review). Users should try to choose
makers to understand why an AP- the tool that most closely matches
HRA is instructive, what resources and the characteristics of the assessment
institutions are needed for AP-HRA, and context. In general, users should choose
what the limitations of the assessment the AP-HRA tool that provides the
may be. Those conducting an AP-HRA maximum degree of technical rigour
need to understand how to do it, know within the resources available. AP-HRA
what data are available and needed results should be presented together
and where to find them, and how to with confidence intervals that take into
communicate the results. account the various possible sources
of error in the input parameters. It may
Various AP-HRA tools are currently
be difficult to present the AP-HRA
available. When selecting the most
results and the associated uncertainty
appropriate tool for the assessment
to decision-makers in an efficient
context, it is important first to define
way. Communication experts may
the policy question to be answered and
be able to help ensure more effective
the audiences to be informed. Next,
communication of the AP-HRA results
the technical needs of the assessment
from the technical experts to policy-
context need to be identified, such as the
makers and other stakeholders.
relevant pollutants and the geographical

17
 References
Anenberg SC (2015) Survey of ambient air pollution health risk assessment tools. Risk Analysis
(in press).
Aphekom (2011) Summary report of the Aphekom project 2008–2011 (http://www.aphekom.org/c/
document_library/get_file?uuid=5532fafa-921f-4ab1-9ed9-c0148f7da36a&groupId=10347).
Australian Department of Health (2012) Environmental health risk assessment. Guidelines
for assessing human health risks from environmental hazards [online]. Canberra, Australian
Government Department of Health.
Brauer M et al. (2012) Exposure assessment for estimation of the global burden of disease
attributable to outdoor air pollution. Environmental Science and Technology, 46: 652–660.
Burnett RT et al. (2014) An integrated risk function for estimating the global burden of disease
attributable to ambient fine particulate matter exposure. Environ Health Perspect, 122: 397–
403.
Chanel O et al. (2014) Economic valuation of the mortality benefits of a regulation on SO2 in 20
European cities. The European Journal of Public Health, 24(4):631–637.
Department of Health (2006) Health risk assessment in Western Australia. Perth, Government
of Western Australia (http://www.public.health.wa.gov.au/cproot/1499/2/Health_Risk_
Assessment.pdf).
Fann N, Risley D (2013) The public health context for PM2.5 and ozone air quality trends.
Air Quality, Atmosphere & Health, 6: 1–11 (http://link.springer.com/article/10.1007%2
Fs11869-010-0125-0).
Forouzanfar MH et al. (2015) Global, regional, and national comparative risk assessment of 79
behavioural, environmental and occupational, and metabolic risks or clusters of risks in 188
countries, 1990–2013: a systematic analysis for the Global Burden of Disease Study 2013. The
Lancet (http://dx.doi.org/10.1016/S0140-6736(15)00128-2).
Health Effects Institute (2010) Outdoor air pollution and health in the developing countries of
Asia: a comprehensive review. Special report 18. Boston, Health Effects Institute (http://pubs.
healtheffects.org/getfile.php?u=602).
HIP (2014) Frequently asked questions about integrating health impact assessment into
environmental impact assessment [online]. Human Impact Partners (http://www.epa.gov/
region9/nepa/PortsHIA/pdfs/FAQIntegratingHIA-EIA.pdf).
Hoek G et al. (2008) A review of land-use regression models to assess spatial variation of
outdoor air pollution. Atmospheric Environment, 42: 7561–7578.
Hoek G et al. (2010) Concentration response functions for ultrafine particles and all-cause
mortality and hospital admissions: results of a European expert panel elicitation. Environmental
Science and Technology, 44: 476–482.
Katsouyanni K (2003) Ambient air pollution and health. British Medical Bulletin, 68: 143-156
(http://bmb.oxfordjournals.org/content/68/1/143.full.pdf).
Keuken MP et al. (2012) Elemental carbon as an indicator for evaluating the impact of traffic
measures on air quality and health. Atmospheric environment, 61: 1–8.
Krewski D et al. (2009) Extended follow-up and spatial analysis of the American Cancer Society
study linking particulate air pollution and mortality. Res Rep Health Eff Inst, Report 140: 5–114;
discussion 115–36.
Le Tertre A et al. (2014) Impact of legislative changes to reduce the sulphur content in fuels in
Europe on daily mortality in 20 European cities: an analysis of data from the Aphekom project.
Air Quality, Atmosphere & Health, 7: 83–91.

18
Lim SS et al. (2013) A comparative risk assessment of burden of disease and injury attributable
to 67 risk factors and risk factor clusters in 21 regions, 1990–2010: a systematic analysis for
the Global Burden of Disease Study 2010. Lancet, 380: 2224–2260.
McAuley C, Hrudey SE. (2006) Towards meaningful stakeholder comprehension of sour gas
facility risk assessments. Journal of Environmental Engineering and Science, 5: 1–11 (http://
dx.doi.org/10.1139/s05-012).
Medina S et al. (2013) Quantifying the health impacts of outdoor air pollution: useful estimations
for public health action. J Epidemiol Community Health, 67: 480–3.
Murray CJ (1994) Quantifying the burden of disease: the technical basis for disability-adjusted
life years. Bulletin of the World Health Organization, 72: 429–445 (http://www.ncbi.nlm.nih.gov/
pmc/articles/PMC2486718/pdf/bullwho00414-0105.pdf).
Murray CJ, Acharya AK (1997) Understanding DALYs (disability-adjusted life years). J Health
Econ, 16: 703–30.
Murray CJL, Lopez AD (2013) Measuring the global burden of disease. New England Journal of
Medicine, 369: 448–457 (http://www.nejm.org/doi/full/10.1056/NEJMra1201534).
Paciorek CJ, Liu Y (2012). Assessment and statistical modeling of the relationship between
remotely sensed aerosol optical depth and PM2.5 in the eastern United States. Research report
167. Health Effects Institute (http://pubs.healtheffects.org/getfile.php?u=718).
Pascal M (2013) Assessing the public health impacts of urban air pollution in 25 European
cities: results of the Aphekom project. Science of the Total Environment, 449: 390–400.
Quigley R et al. (2006) Health impact assessment international best practice principles
[online]. Fargo, USA; International Association for Impact Assessment (http://www.iaia.org/
publicdocuments/special-publications/SP5.pdf?AspxAutoDetectCookieSupport=1).
SA Health (2003) The South Australian Burden of Disease study: background and discussion
papers. Discounting and age-weighting [online]. Adelaide; Government of South Australia
(http://www.sahealth.sa.gov.au/wps/wcm/connect/1d94008048002172bd3bfd7675638bd8/
Discounting+and+Age+Weighting-PIGR-SABoD-20110808.pdf?MOD=AJPERES&CACHEID=
1d94008048002172bd3bfd7675638bd8&CACHE=NONE).
Samet J, Krewski D (2007) Health effects associated with exposure to ambient air pollution.
Journal of Toxicology and Environmental Health, Part A 70, 227–242 (http://www.tandfonline.
com/doi/pdf/10.1080/15287390600884644).
Scott I (2008) Interpreting risks and ratios in therapy trials. Australian Prescriber, 31: 12–16.
US EPA (2008) Particulate matter (PM). Basic information [online]. US Environmental Protection
Agency (http://www.epa.gov/pm/basic.html; accessed 26 April 2015.
US EPA (2011) Exposure factors handbook: 2011 edition. Washington, DC; National Center for
Environmental Assessment (http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=20563).
US EPA (2014) Environmental Benefits Mapping and Analysis Program. Community edition
(v1.08). Research Triangle Park, NC (http://www.epa.gov/air/benmap).
US EPA (2015) Technology Transfer Network Air Toxics. 2005 National-Scale Air Toxics
Assessment. Glossary of key terms [online]. United States Environmental Protection Agency
(http://www.epa.gov/airtoxics/natamain/gloss1.html; accessed 8 January 2015.
US EPA (2012) Human health risk assessment [online]. United States Environmental Protection
Agency (http://epa.gov/riskassessment/health-risk.htm).
UNECE (2010). Hemispheric transport of air pollution 2010. Part A: Ozone and particulate matter.
Air Pollution Studies No. 17. New York and Geneva, United Nations Economic Commission for
Europe (http://www.unece.org/fileadmin/DAM/env/lrtap/Publications/11-22134-Part-A.pdf).
van Donkelaar A et al. (2010) Global estimates of ambient fine particulate matter concentrations
from satellite-based aerosol optical depth: development and application. Environmental health
perspectives, 118, 847.
WHO (2005). Workshop on guiding public health policy in areas of scientific uncertainty –
Workshop report, July 11–13, 2005. Ottawa, Canada; McLaughlin Centre for Population Health

19
 Risk Assessment, Institute of Population Health, University of Ottawa in partnership with the
World Health Organization (http://www.who.int/peh-emf/meetings/WHO_Final_workshop_
report.pdf?ua=1).
WHO (2010) WHO human health risk assessment toolkit: chemical hazards. IPCS harmonization
project document; no.8. Geneva; World Health Organization (http://www.who.int/ipcs/
publications/methods/harmonization/toolkit.pdf?ua=1).
WHO (2014a) Health statistics and information systems. Metrics: disability-adjusted life year
(DALY) [online]. Geneva; World Health Organization (http://www.who.int/healthinfo/global_
burden_disease/metrics_daly/en/).
WHO (2014b) Burden of disease from ambient air pollution for 2012. Geneva; World Health
Organization (http://www.who.int/phe/health_topics/outdoorair/databases/AAP_BoD_results_
March2014.pdf).
WHO (2014c) WHO news release. 7 million premature deaths annually linked to air pollution.
Geneva (http://www.who.int/mediacentre/news/releases/2014/air-pollution/en/).
WHO Regional Office for Europe (1999) Health impact assessment: main concepts and
suggested approach [online]. Brussels, European Centre for Health Policy, WHO Regional
Office for Europe (http://www.euro.who.int/document/PAE/Gothenburgpaper.pdf).
WHO Regional Office for Europe (2005). Health effects of transport-related air pollution.
Copenhagen (http://books.google.ie/books?id=txw26P7Lb1oC&printsec=frontcover#v=onep
age&q&f=false).
WHO Regional Office for Europe (2006) Air quality guidelines for particular matter, ozone,
nitrogen dioxide and sulphur dioxide. Global update 2005. Copenhagen (http://www.euro.who.
int/__data/assets/pdf_file/0005/78638/E90038.pdf).
WHO Regional Office for Europe (2013). Review of evidence on health aspects of air pollution
(REVIHAAP). Technical report. Copenhagen (http://www.euro.who.int/en/health-topics/
environment-and-health/air-quality/publications/2013/review-of-evidence-on-health-aspects-
of-air-pollution-revihaap-project-final-technical-report).
WHO Regional Office for Europe (2014) WHO Expert Meeting. Methods and tools for assessing
the health risks of air pollution at local, national and international level. Copenhagen (http://
www.euro.who.int/__data/assets/pdf_file/0010/263629/WHO-Expert-Meeting-Methods-and-
tools-for-assessing-the-health-risks-of-air-pollution-at-local,-national-and-international-level.
pdf?ua=1).

20
 Annex 1. Useful
AP-HRA resources
AirCounts. http://www.aircounts.com/.
AirQ2.2. http://www.euro.who.int/en/health-topics/environment-and-health/air-quality/
activities/ tools-for-health-impact-assessment-of-air-quality-the-airq-2.2-software.
Aphekom. http://www.aphekom.org/web/aphekom.org/publications.
Co-benefits Calculator. Points of contact: Neal Fann (US Environmental Protection
Agency, [email protected]), Harry Vallack (Stockholm Environment Institute, harry.
[email protected]).
To obtain: via the points of contact.
EcoSense. Point of contact: Joachim Roos (University of Stuttgart, Joachim.Roos@ier.
uni-stuttgart.de).
To obtain: EcoSenseWeb provides a web interface for single source calculations
(http://ecosenseweb.ier.uni-stuttgart.de), which can be accessed for a small fee.
Environmental Burden of Disease (EBD) assessment tool for ambient air pollution.
To obtain: contact [email protected].
Economic Valuation of Air Pollution (EVA). To obtain: contact Professor Jørgen Brandt
([email protected]).
GBD. https://www.healthdata.org/gbd.
Global Model of Ambient Particulates (GMAPS). Points of contact: The World Bank,
World Development Indicators database. Note: The model itself is no longer actively
maintained and therefore no longer available for download.
IOMLIFET. http://www.iom-world.org/research/research-expertise/statistical-services/
iomlifet/.
SIM-Air. http://www.urbanemissions.info/.
TM5-FASST. Point of contact: Rita van Dingenen (European Commission Joint Research
Centre, [email protected]).
How to obtain: via the point of contact.
US EPA BenMAP-CE. http://www.epa.gov/air/benmap.
Ambient air pollution health impact assessment tools with national scope.
Air Quality Benefits Assessment Tool (AQBAT). To obtain: contact Stan Judek (stan.
[email protected]).
AP2 (formerly APEEP). https://sites.google.com/site/nickmullershomepage/home/ap2-
apeep-model-2.
Co-benefits Risk Assessment (COBRA) screening model. http://epa.gov/
statelocalclimate/resources/cobra.html.
Illness Costs of Air Pollution (ICAP). Points of contact: Canadian Medical Association
and Ontario Medical Association websites. Note: The model itself is no longer actively
maintained and therefore no longer available for download.
Integrated Transport and Health Impact Modelling Tool (ITHIM). Points of contact:
James Woodcock ([email protected]) and Marko Tainio ([email protected].
ac.uk).
To obtain: via the points of contact.

21
 Annex 2.
The Aphekom project
The Aphekom project (Chanel et al., How was the health risk estimated?
2014; Le Tertre et al., 2014; Medina
From available concentration response
et al., 2013; Pascal et al., 2013) was a
functions in the literature for non-
European multicity project that aimed
accidental, all-cause mortality from PM2.5
to provide new information and tools to
exposure.
allow: (a) decision-makers to set more
effective European, national and local
Results
policies on air pollution and health; (b)
health professionals to better advise A decrease in annual mean PM2.5 level
vulnerable individuals on air pollution; to 10 µg per m3 could add more than 6
and (c) all individuals to better protect months of life expectancy at age 30 in
their health from the effects of air half of the EU cities (Figure 5). Exceeding
pollution. Specifically, it aimed to answer the WHO guideline level on PM2.5 leads
the following two questions. to a burden on mortality of nearly 19 000
deaths per year. The associated costs
would reach €30 billion annually (Pascal
Question 1: What are the health benefits
et al., 2013).
of reducing air pollution to the WHO air
quality guideline values in 25 European
How was the uncertainty of the results
cities with a total of nearly 39 million
treated?
inhabitants?
Uncertainties in the CRF and the
Methods economic valuation were combined
in two different ways. Monte Carlo
• Air pollutant assessed: PM2.5. An
simulations were done to analyse the
assessment was made of the health
uncertainty in the HIA results and the
benefits that could be obtained if PM2.5
economic values.
concentrations were lowered to meet
the WHO air quality guideline value
of 10 µg per m3 (the counterfactual
level), with all other variables staying
constant.
• Geographical scope: 25 EU cities.
• Population data spatial resolution: City
level.

Tools
• Aphekom HIA guidelines and tools
(available at: http://www.aphekom.org/
web/aphekom.org/publications).

How was population exposure

estimated?
From urban background monitoring data
averaged across each individual city.
The study period 2004–2006 for baseline
scenario; monitoring data adjusted to
reflect hypothetical air quality policies.

22
Figure 5. (a) PM2.5 levels in 25 EU cities and (b) predicted gain in life
expectancy from complying with WHO guideline
Adapted from Aphekom (2011)

AQG
a) Bucharest 38.2
Budapest 33.7
Athens 29.4
Ljubijana 29.4
Barcelona 27
Valencia 23
Seville 22.9
Vienna 21.6
Granada 21.4
Rome 20.9
Brussels 19
Marseilles 18.5
Strasbourg 16.6
Lille 16.6
Lyons 16.5
Paris 16.4
Bilbao 15.7
Bordeaux 15.7
Rouen 15.3
Le Havre 14.5
Toulouse 14.2
London 13.1
Malaga 12.8
Dublin 10.5
Stockholm 9.4
0 5 10 15 20 25 30 35 40
Average PM25 µg/m3

b) Bucharest 22.1
Budapest 19.3
Athens 12.8
Ljubijana 14.5
Barcelona 13.7
Valencia 9.9
Seville 10.2
Vienna 9.3
Granada 9.3
Rome 11.5
Brussels 7
Marseilles 7.5
Strasbourg 5.7
Lille 5.8
Lyons 5.7
Paris 5.8
Bilbao 4.9
Bordeaux 5
Rouen 4.6
Le Havre 4.2
Toulouse 3.6
London 2.5
Malaga 2.2
Dublin 0.4
Stockholm 0
0 5 10 15 20 25
Gain in life expectancy (months)

23
 Question 2: Did policies designed to How was the health risk estimated?
reduce the sulfur content in certain liquid
(1) Poisson regression: City-specific risks
fuels improve air quality and subsequently
of death associated with changes in SO2
public health in 20 cities in the European
for periods prior to and following the
Union (EU)?
implementation of three EU directives. (2)
City-specific risk estimates pooled using
Methods
meta-regression. (3) HIA: estimation of
• Air pollutant assessed: SO2. An premature deaths avoided as a result
assessment was made of changes of changes in SO2 after implementation
in associations between daily stages compared with baseline levels;
concentrations in SO2 and daily economic evaluation.
mortality before and after the
implementation of legislative measures Results
regulating the sulfur content in certain
Overall outcomes were based on data
fuels.
from 20 collaborating EU cities from
• Geographical scope: 20 EU cities. 2000 onwards, compared with the period
prior to implementation of the directive
• Population data spatial resolution: City
due to reductions in SO2 concentrations:
level.
2212 lives were saved each year from
all causes (95% CI: 772–3663); annual
Tools
monetary savings were valued at €191.6
• Aphekom HIA guidelines and tools million (Chanel et al., 2014; Le Tertre et
(available at: http://www.aphekom.org/ al., 2014).
web/aphekom.org/publications).
How was the uncertainty of the results
How was population exposure treated?
estimated?
Monte Carlo simulations were done to
From urban background monitoring data analyse the uncertainty in the HIA results
averaged across each individual city. The and the economic values.
study period is 1990–2008.

24
 Annex 3. Global
Burden of Disease
The GBD 2010 project, coordinated minimum and 5th percentile of the PM2.5
by IHME (Lim et al., 2013), was a CRA exposure distribution of the American
exercise to compare the burden of disease Cancer Society Cancer Prevention II
associated with various risk factors, cohort study (5.8 µg per m3 and 8.8
diseases and injuries. Specifically, for air µg per m3, respectively) (Burnett et al.,
pollution, it aimed to answer the following 2014; Krewski et al., 2009; Lim et al.,
two questions .1 2013).
• What is the risk to health from ambient
Tools
and household air pollution over the
entire global range of exposure? • Air pollution: TM5-FASST, satellite-
based estimates and the Greenhouse
• What are the impacts on health
Gas and Air Pollution Interactions and
of recently observed air pollution
Synergies (GAINS) emission inventory.
compared with the health burden of
other risk factors? • Health risk: integrated exposure–
response (IER) model.
Methods
How was population exposure
• Air pollution was assessed 3 different
estimated?
ways:
PM2.5, a common useful indicator • Ambient particulate air pollution
of risk associated with exposure to data integration: remote sensing
a mixture of pollutants from diverse estimates and chemical transport
sources and in different environments model simulations were averaged and
assessing (i) ambient PM pollution, and calibrated with available measurement
(ii) household air pollution from solid data in a single global regression
fuels; and ambient ozone, but in the model. The resulting estimates
context of this example the focus is on provided full global coverage, reduced
PM only. biases, high spatial resolution and
facilitated estimation of source and
• Geographical scope: global.
source-sector contributions to ambient
• Population data spatial resolution: concentrations.
population surveys and censuses.
• Household and ambient air pollution
• CRA approach: consistent methods were considered as separate risk
were used to estimate attributable factors for the global disease burden;
burden of disease for a variety of the degree of overlap in exposures
risk factors at global, regional, and from these two risk factors was also
national levels and their respective considered.
uncertainties.
• Impacts for ambient PM pollution were How was the health risk estimated?
calculated in relation to a counterfactual • IER functions were developed,
level, defined by a uniform distribution combining evidence from studies of
with lower and upper bounds at the ambient air pollution, second-hand

1 This annex highlights certain aspects of the GBD 2010 study that are relevant to the content of this publication.
The GBD 2013 data are now available as well (Forouzanfar et al., 2015).

25
 smoke, household air pollution and • Household air pollution was ranked
active smoking (Burnett et al., 2014). as 4th risk factor worldwide for 2010,
accounting for 4.5% (3.4–5.3%) of
• This was done because existing data
global DALYs (Figure 6) (Lim et al.,
covered only small concentration
2013).
ranges and no exposure-response
functions were available from studies • Ambient air pollution was ranked as
outside North America and Europe. 9th risk factor worldwide for 2010,
Existing exposure-response functions accounting for 3.1% (2.7–3.4%) of
could not be directly applied to global DALYs (Figure 6) (Lim et al.,
countries with high levels of air 2013).
pollution, e.g. in Asia (Lim et al., 2013).
IER assumptions: air pollution- How was the uncertainty of the results
attributable mortality is independent of treated?
other risk factors.
The GBD study used simulation
methods to incorporate uncertainty
Results from four sources: disease burden,
Ambient particulate matter pollution and pollution exposure level, response to the
household air pollution from solid fuels pollution, and the counterfactual level
were a significant risk factor for health in of air pollution (WHO Regional Office for
2010, contributing to premature deaths Europe, 2014).
worldwide.

26
Figure 6. Burden of disease attributable to 20 leading risk factors in 2010,
expressed as a percentage of global DALYs

High blood pressure

Tobacco smoking, including second-hand smoke
Alcohol use
Household air pollution from solid fuels
Diet low in fruits
High body-mass index
High fasting plasma glucose
Childhood underweight
Ambient particulate matter pollution
Physical inactivity and low physical activity
Diet high in sodium
Diet low in nuts and seeds
Iron deficiency
Suboptimal breastfeeding
High total cholesterol
Diet low in whole grains
Diet low in vegetables
Diet low in seafood omega-3 fatty acids
Drugs use
Occupational risk factors for injuries
-0.5 0 2 4 6 8
Disability-adjusted life-years (%)

Cancer Diabetes, urogenital, blood and endocrine Maternal disorders

Cardiovascular and circulatory diseases Muscoloskeletal disorders Neonatal disorders
Chronic respiratory diseases Other non-communicable diseases Nutritional deficiencies
Cirrhosis HIV/AIDS and tuberculosis Other communicable diseases
Digestive diseases Diarrhoea, lower respiratory infections and Transport injuries
Neurological disorders other common infectious diseases Unintentional injuries
Mental and behavioural disorders Neglected tropical diseases and malaria Intentional injuries
War and disaster

Source: reprinted from The Lancet, Vol 380, Lim et al., A comparative risk assessment of burden of disease and injury
attributable to 67 risk factors and risk factor clusters in 21 regions, 1990-2010: a systematic analysis for the Global Buren
of Disease Study 2010, p. 2244, Elsevier Limited (2013), with permission from Elsevier.

27
 Annex 4. BenMAP
The Environmental Benefits Mapping and to 2007. Data were interpolated using
Analysis Program (BenMAP) (US EPA, the VNA algorithm.
2014) supports regulatory development
in the United States of America. A study How was the health risk estimated?
was conducted in the USA by the US
• Baseline incidence: taken as 3-year
Environmental Protection Agency (Fann
average (2006–08) county-level all-
& Risley, 2013) using BenMAP to assess
cause mortality rates from the CDC-
the public health context for PM2.5 and
WONDER database.
ozone air quality trends. The study aimed
to answer the following question. • Available CRFs: two CRFs were selected
for each pollutant, to compensate for
• What is the level of premature mortality
strengths and weaknesses inherent in
incurred or avoided as a result of
the studies selected, e.g. one may have
changes in the level and distribution
considered a broader geographical
of PM2.5 and O3 in air nationwide in the
area and the other a larger population.
USA?
Results
Methods
• Estimated reductions in monitored
• Air pollutants assessed: PM2.5 and O3.
PM2.5 and O3 concentrations from 2000
• Geographical scope: country. to 2007 were associated with an annual
total of 22 000–60 000, and 880–4100,
• Population data spatial resolution: US
avoided premature deaths from all
census block-level populations.
causes, respectively (Figure 7) (Fann &
• Approach used involved the estimation Risley (2013).
of:
o spatial distribution of changes in How was the uncertainty of the results
ambient air quality resulting from treated?
past changes in air quality;
• Confidence intervals were estimated
o change in population exposure, and using a Monte Carlo analysis.
o health impacts (by applying CRFs
from the epidemiological literature to
the change in population exposure).

Tools
• Air pollution: Voronoi neighbour
averaging (VNA) algorithm.
• Population exposure and health risk:
BenMAP (US EPA, 2014).

How was population exposure

estimated?
• Measured air pollutant monitor
concentrations were used to create
a 12×12 km gridded spatial map of
concentrations for each year from 2000
(the counterfactual level of air pollution)

28
Figure 7. Annual incidence of premature mortality avoided or
incurred due to changing PM2.5 or ozone concentrations 2000-2007

Annual incidence of premature

mortality avoided or incurred
due to changing PM2.5
concentrations 2000-2007

Incurred Avoided
> 100

50 to 100

1 to 50

No change

1 to 50

50 to 100

100 to 150

150 to 200

200 to 250

> 250

Annual incidence of premature

mortality avoided or incurred
due to changing ozone
concentrations 2000-2007

Incurred Avoided
> 30
> 30

20 to 30

10 to 20

1 to 10

No change

1 to 10

10 to 20

20 to 30

Source: Fann & Risley (2013), with kind permission from Springer Science and Business Media.

29
The WHO Regional
Office for Europe

The World Health Organization (WHO) is a

specialized agency of the United Nations created
in 1948 with the primary responsibility for
international health matters and public health.
The WHO Regional Office for Europe is one of six
regional offices throughout the world, each with
its own programme geared to the particular health
conditions of the countries it serves.

Member States

Albania
Andorra
Armenia
Austria
Azerbaijan
Belarus
Belgium
Bosnia and Herzegovina
Bulgaria
Croatia
Cyprus
Czech Republic
Denmark
Estonia
Finland
France
Georgia
Germany
Greece
Hungary
Iceland
Ireland
Israel
Italy
Kazakhstan
Kyrgyzstan
Latvia
Lithuania
Luxembourg
Malta
Monaco
Montenegro
Netherlands
Norway
Poland
Portugal
Republic of Moldova
Romania
Russian Federation
San Marino
Serbia 9 789289 051316 >
Slovakia
Slovenia
Spain
Sweden
Switzerland
Tajikistan
The former Yugoslav
World Health Organization
Republic of Macedonia
Regional Office for Europe
Turkey
Turkmenistan UN City, Marmorvej 51, DK-2100 Copenhagen Ø, Denmark
Ukraine Tel.: +45 45 33 70 00/Fax: +45 45 33 70 01
United Kingdom Email: [email protected]
Uzbekistan Website: www.euro.who.int