Relative risk calculations

Introduction
The assessment of the burden and impacts of air pollution on health is operated through a
methodology that will be implemented in AirQ+ and it is briefly described in the following text.
Before introducing the methodology and its formulas, it is important to introduce some definitions
that provide the basis for the formulas that will be used in AirQ+. One section summarizes data
needs and gives an example of calculations.

Air pollution is usually investigated in two separate fields: ambient (outdoor) air pollution and
household (indoor) air pollution. Furthermore, effects on health are usually considered for long-term
(for example, yearly) and short-term exposure (for example, daily concentrations). The impacts
considered include mortality, total and cause-specific and other relevant health outcomes, such as
hospital admissions for cardiovascular diseases, hospital admissions for respiratory diseases. Age-
specific impacts on mortality are calculated through the use of Life Tables using estimators such as
years-of-life-lost).

Three scenarios for resulting health effects are typically investigated by users:

1. partial reduction of exposure levels, e.g. 10% reduction of annual mean PM10;
2. reduction of exposure to a specific target level, e.g. reduction of annual mean of PM2.5 to
10μg/m3 as suggested by WHO air quality guidelines;
3. absolute reduction of exposure, e.g. 5 ppb decrease in the annual mean for ozone.

Ambient air pollution is discussed first, then household air pollution. The life table calculations for
AirQ+ are described in a separate document, AirQ_plus_lifeTableModule_V01.docx.

Definitions
For defined air pollutants (risk factors1) AirQ+ provides the associated relative risks (RR) of exposure.
For a population exposed to a risk factor the RR is defined as…

Equation 1

𝐼𝑒
𝑅𝑅 =
𝐼𝑐𝑓

Ie : incidence rate2 in the population exposed (e.g. number of new cases per 100 000 population per
year).

1
Such as particulate matter, particle sizes less or equal to 10 or 2.5 µm (PM 10, PM2.5), nitrogen dioxide (NO2),
ozone (O3) and black carbon (BC).

1
30 July 2015
Icf: incidence rate in the counterfactual population The counterfactual population is a population
that is comparable in composition to the exposed population, but is not exposed itself to the risk
factor.

In the scope of AirQ+ mortality rates (death rates) and morbidity rates are nothing but special
incidence rates.

Example of RR: RR for active smoking (AS) mortality due to lung cancer (LC), number of deaths per
year per 100 000 population (d):

Equation 2

𝐿𝐶
𝐿𝐶
𝑑𝐴𝑆
𝑅𝑅𝐴𝑆 = 𝐿𝐶
𝑑𝑛𝑜𝑛 𝑠𝑚𝑜𝑘𝑒𝑟𝑠

With a, b, c, and d are the number of cases (deaths) defined in the following population risk table…

Lung cancer
Risk Present Absent
Smoker a b a+b
Non-smoker c d c+d

… the smoking related relative risk for lung cancer is calculated as:

𝐿𝐶
𝑎/(𝑎 + 𝑏)
𝑅𝑅𝐴𝑆 =
𝑐/(𝑐 + 𝑑)

RRAS is the excess risk for smokers to die of lung cancer as compared to non-smokers. If there were
𝐿𝐶
no excess risk, RRAS would be equal to 1 (but there would still be cases of LC)3. Where 𝑑𝐴𝑆 is the
𝐿𝐶
probability of death for lung cancer for active smokers and 𝑑𝑛𝑜𝑛 𝑠𝑚𝑜𝑘𝑒𝑟𝑠 is the probability of death
for lung cancer for non smokers. If for example a = 10, b = 90, c = 1, and d = 99, then the relative risk
of cancer associated with smoking would be equal to 10. Smokers would develop lung cancer ten
times as likely as non-smokers .

2
Incidence rates are defined for the (sub-) group of population in question, in particular, if age groups are
𝑛_𝑐𝑎𝑠𝑒𝑠𝑎𝑔𝑒 𝑔𝑟𝑜𝑢𝑝
used, 𝐼𝑎𝑔𝑒 𝑔𝑜𝑢𝑝 = ⁄𝑛𝑜_𝑜𝑓_𝑝𝑒𝑜𝑝𝑙𝑒 This means that 𝐼𝑎𝑙𝑙 𝑎𝑔𝑒𝑠 ≠ ∑ 𝑎𝑙𝑙 𝑎𝑔𝑒 𝐼𝑎𝑔𝑒 𝑔𝑟𝑜𝑢𝑝
𝑎𝑔𝑒 𝑔𝑟𝑜𝑢𝑝 𝑔𝑟𝑜𝑢𝑝𝑠

since the denominators are all different.

3
Theoretically it is also possible for a RR to be less than one, this means that exposure leads to a reduction in
risk. The reason for this potential confusion is the epidemiological use of the phrase “exposure”, which differs
from its everyday use, where “exposure” comes with a negative connotation. The epidemiological use is purely
technical and therefore a population can be exposed to beneficial factors as well. Examples are exposure to
medical treatment or physical activity. In the case for air pollution, since there is no beneficial health effects
known, RRs are always greater than one.

2
30 July 2015
Ideally, the counterfactual population is not exposed to the risk factor. Like in the example: only
smokers are exposed to AS. Non-smokers are also exposed, but to second hand smoke (SHS), which
is studied using distinct SHS relative risks.

The concept of a counterfactual is needed in studying air pollution since usually the whole
population is exposed. Therefore, no studies are available for comparable populations with and
without exposure. For air pollution, analysis is carried out asking: how does the actual situation
relate to one with a lower exposure level? Any specific counterfactual (exposure) depends on
specific assumptions, like the lowest value for which data is available, or the natural level of
exposure, or an exposure level under which no adverse health effects are expected to occur.

From epidemiological studies we have several RRs available both for long-term and short term
effects and various health outcomes.

In general, AirQ+ supplies stratified relative risks:

Equation 3

𝑅𝑅𝑝𝑜𝑙𝑙𝑢𝑡𝑎𝑛𝑡 = 𝑅𝑅𝑝𝑜𝑙𝑙𝑢𝑡𝑎𝑛𝑡 (level of exposure, health outcome, age, sex)

This means that RRs are built as specific functions of the level of exposure, the health outcome, age
and sex of the population. A health outcome can be mortality or a specific type of disease, like lung
cancer (LC). Intervals for age groups are typically five years. RRs stratified by sex (male, female) is
only available for indoor air pollution (IAP). In any case, AirQ+ should be capable of calculating
integrated effects for all ages and all sexes.

Ambient air pollution: Formulas

For outdoor air pollution, the whole population is assumed to be exposed. Population attributable
fraction (PAF) estimates the proportion of cases that are prevented if the exposure were zero. In
case the whole population is exposed…

Equation 4

𝑅𝑅 − 1
𝑃𝐴𝐹 =
𝑅𝑅

Example: dispersed particulate matter (PM) is an air pollutant that has a RRLC greater than one for
lung cancer. If the level of exposure were reduced in a population to a (counterfactual) level where
RR is equal to one, PAF would be zero. Please note that there is still a risk to die of lung cancer due
to causes other than exposure to PM in that population.

If only a fraction p of a population is exposed, as in the case of AS, the formula is generalized to:

Equation 5

𝑝(𝑅𝑅 − 1)
𝑃𝐴𝐹 = ,0 < 𝑝 ≤ 1
𝑝(𝑅𝑅 − 1) + 1

3
30 July 2015
Household air pollution: Formulas

Equation 5 is also applied in the case of studying the effect of indoor air pollution, since under
certain conditions frequency and levels of exposure of females to indoor air pollution are higher
than for men.

The percentage of the population exposed to household air pollution can be calculated by country
(or by city if local data are available). Relative risks are calculated separately for men, women and
children, based on the integrated exposure-response functions (IER) for all diseases but COPD. The
suggested counterfactual concentration is between 5.8 and 8.8 μg/m3 (the concentration under
which no adverse health effects are expected). The country population attributable fractions for
ALRI, COPD, LC, stroke and IHD are calculated using Equation 4; p is the percentage of the population
exposed to that level of air pollution, i.e. the percentage of the population using solid fuels for
cooking.

Table 1

Household air pollution RRs

Disease RR (95% CI) women RR (95% CI) men

ALRI 2.9 (2.0-3.8) for children
COPD 2.3 (1.7-3.1) 1.9 (1.2-3.1)
Lung Cancer 2.3 (1.5-2.8) 1.9 (1.4-2.3)
IHD (1.4-2.2) (1.4-2.2)
Stroke (1.4-2.4) (1.3-2.4)
Source: WHO ([Link]

If the whole population is exposed, then p = 1 and Equation 5 becomes Equation 4.

General formulas

In a specific population the number of cases or the fraction of all cases attributable to a specific
pollutant and health outcome is of interest to calculate the attributable burden (AB):

Equation 6

𝐴𝐵 = 𝐵𝑜𝐷 ∙ 𝑃𝐴𝐹

The burden of disease (BoD) is the total burden of a specific health outcome. The same formula
holds for the attributable deaths (AD). For example, let’s consider a population with 100 deaths in a
given year due to stroke in the age group 40-45. How many are attributable to PM2.5?

The yearly average level of exposure is 28 µg/m3. The relative risk (Equation 3) RRPM2.5 (28 µg/m3,
stroke, 40-45 years)4
µg
𝑅𝑅𝑃𝑀2.5 (28 , stroke, 40 to 45 years) = 𝟏. 𝟖𝟑
𝑚3

4
The red numbers are the example values taken from Table 3 to Table 5 in the next section.

4
30 July 2015
AirQ+ implements alternative methods for RR values, which can be chosen by the user. RR values are
either calculated by AirQ+ based on defined formulas or ranges of values (see Annex 1) or picked
from a table that contains about 9 000 RR-values in steps of 1 µg/m3 for all health outcomes and age
groups. Table 3 is based on such a RR value table5.

The BoD in this case is the total number of deaths by stroke in that age group is (males plus females)

𝑑(𝑠𝑡𝑟𝑜𝑘𝑒, 40 𝑡𝑜 45 𝑦𝑒𝑎𝑟𝑠) = 𝟏𝟎𝟎

Thus, according to Equation 4 and Equation 6 (see Table 3 and Table 4 for the values):

1.83 − 1
𝐴𝐵𝑃𝑀2.5 (𝑠𝑡𝑟𝑜𝑘𝑒, 40 𝑡𝑜 45 𝑦𝑒𝑎𝑟𝑠) = 100 ∙ = 45.4
1.83

This means that approximately 45 out of 100 deaths for stroke in the 40-45 age group are
attributable to the exposure to air pollution measured as an average yearly value of 28 µg/m3 PM2.5.

The total attributable burden for stroke is the sum over all age groups

𝐴𝐵𝑃𝑀2.5 (𝑠𝑡𝑟𝑜𝑘𝑒) = ∑ 𝐴𝐵𝑃𝑀2.5 (𝑠𝑡𝑟𝑜𝑘𝑒, 𝑎𝑔𝑒 𝑔𝑟𝑜𝑢𝑝) = 𝟒𝟕𝟕𝟓

𝑎𝑙𝑙 𝑎𝑔𝑒 𝑔𝑟𝑜𝑢𝑝𝑠

This means repeating the same calculation for all the age groups, considering the respective values
of mortality and RRs and sum them up.

Finally, to assess the total burden for PM2.5 the total number of deaths for all causes in that year
needs to be known. In this example dtotal = 122 168. The total AB for PM2.5 (see table 3) is the sum
overall health outcomes6:

𝐴𝐵𝑃𝑀2.5 = ∑ ∑ 𝐴𝐵𝑃𝑀2.5 (ℎ𝑒𝑎𝑙𝑡ℎ 𝑜𝑢𝑡𝑐𝑜𝑚𝑒, 𝑎𝑔𝑒 𝑔𝑟𝑜𝑢𝑝) = 𝟏𝟖 𝟓𝟓𝟎

𝑎𝑙𝑙 ℎ𝑒𝑎𝑙𝑡ℎ 𝑎𝑙𝑙 𝑎𝑔𝑒
𝑜𝑢𝑡𝑐𝑜𝑚𝑒𝑠 𝑔𝑟𝑜𝑢𝑝𝑠

Thus, the total attributable burden of disease fraction (ABF) due to PM2.5 exposure in that specific
population is

18 550
𝐴𝐵𝐹𝑃𝑀2.5 = = 15.2%
𝟏𝟐𝟐 𝟏𝟔𝟖

The following Equations 7-9 summarize the previous calculations. The last two sums in general
notation:
5
In practice there are two ways to consider effects on health due to change of concentrations of an air
pollutant. One can consider the change to linear and “the available analyses suggest that it is reasonable to
use linear CRFs to assess risks within Europe, given the expected levels of PM 2.5 in 2020. This is especially the
case for all-cause mortality. For more specific causes of death, a supra-linear function, steeper in lower
concentrations, may fit the data slightly better” (WHO, 2013: 14). So, a second possibility is to consider
changing RRs values according to concentration values.
6
Health outcomes for PM2.5 would be acute lower respiratory infections (ALRI), ischemic heart disease (IHD),
chronic obstructive pulmonary disease (COPD), stroke and lung cancer (LC).

5
30 July 2015
Equation 7

𝐴𝐵𝑝𝑜𝑙𝑙𝑢𝑡𝑎𝑛𝑡 (ℎ𝑒𝑎𝑙𝑡ℎ 𝑜𝑢𝑡𝑐𝑜𝑚𝑒) = ∑ 𝐴𝐵𝑝𝑜𝑙𝑙𝑢𝑡𝑎𝑛𝑡 (ℎ𝑒𝑎𝑙𝑡ℎ 𝑜𝑢𝑡𝑐𝑜𝑚𝑒, 𝑎𝑔𝑒 𝑔𝑟𝑜𝑢𝑝)

𝑎𝑙𝑙 𝑎𝑔𝑒 𝑔𝑟𝑜𝑢𝑝𝑠

and

Equation 8

𝐴𝐵𝑝𝑜𝑙𝑙𝑢𝑡𝑎𝑛𝑡 = ∑ ∑ 𝐴𝐵𝑝𝑜𝑙𝑙𝑢𝑡𝑎𝑛𝑡 (ℎ𝑒𝑎𝑙𝑡ℎ 𝑜𝑢𝑡𝑐𝑜𝑚𝑒, 𝑎𝑔𝑒 𝑔𝑟𝑜𝑢𝑝)

𝑎𝑙𝑙 ℎ𝑒𝑎𝑙𝑡ℎ 𝑎𝑙𝑙 𝑎𝑔𝑒
𝑜𝑢𝑡𝑐𝑜𝑚𝑒𝑠 𝑔𝑟𝑜𝑢𝑝𝑠

For calculation of the attributable fraction (ABF) the total burden by all causes (in the example
above: total number of deaths from non-external7 all causes) needs to be available:

Equation 9

𝐴𝐵𝑝𝑜𝑙𝑙𝑢𝑡𝑎𝑛𝑡
𝐴𝐵𝐹𝑝𝑜𝑙𝑙𝑢𝑡𝑎𝑛𝑡 =
𝑇𝑜𝑡𝑎𝑙 𝐵𝑜𝐷

Relation between RR for all ages and RR by age groups

Sometimes relative risks are not available by age groups, but as a single number for all ages. For the
whole population and a certain health outcome Equation 6 is applied with BoD being the number of
all cases in the population. If a specific age group is being investigated, RRall ages is used for that age
group, too.

Note: in case age group specific RRs are available, RRall ages needs to be supplied separately since, due
to the definition of RR involving incidences (Equation 1), there is no simple relationship between
1
RRall ages and age group specific relative risks, like: 𝑅𝑅𝑎𝑙𝑙 𝑎𝑔𝑒𝑠 ≠ ∑ 𝑅𝑅𝑎𝑔𝑒 𝑔𝑟𝑜𝑢𝑝
𝑛 𝑎𝑙𝑙 𝑎𝑔𝑒 𝑔𝑟𝑜𝑢𝑝𝑠

7
External causes are all kinds of “accidents”, such as traffic mortality or drownings.

6
30 July 2015
Structure of BoD data for RR calculations
There are two basic methods for the determination of the RRs8:

1. RRs implemented as analytical functions;

2. RR values pre-calculated in simulations and tabulated in units by one per health outcome
(Integrated exposure risk functions (IER), worksheet 4.) IER RRs in Excel file, provided by
WHO).

The two analytical functions are defined in Bart Ostro: Outdoor Air Pollution.

Table 2

Analytical RR functions (excerpt), X is pollutant concentration, Xo the value of the

counterfactual

Source: B. Ostro [Link]

Equation 10

𝑅𝑅 = exp[𝛽 ln(𝑋 − 𝑋0 )]

Equation 11

𝑅𝑅 = [(𝑋 + 1)/(𝑋0 + 1)]𝛽

⇔ RR = exp[β(ln(X+1) - ln(𝑋0 +1))]

8
Excel file: 2B_RR_forumlas_and_tables.xslm

7
30 July 2015
The two parameters, β and X0, depend on the type of pollutant, the health outcome and
(sometimes) age; they are provided by WHO.

Note on slang: Ostro refers to the first equation as “linear” and the second one as “log-linear”.
Instead, AirQ 2.2 used a “really” linear function for BoD calculation, i.e. RR = 1 + β(X-X0), and
Equation 10 for life tables. In order to clearly distinguish between the three cases, we use the terms
“exponential linear” and “exponential log-linear” in the Excel file. The “really” linear function is
provided in the Excel file, too, but it is not intended to implement it in AirQ+ since for small values of
the exponent results are comparable to those from Equation 10 due to the Taylor series expansion of
the exp-function.

In order to calculate the burden of disease due to a certain pollutant according to Equation 7 to
Equation 8 the following data needs to be available:

 the level of exposure;

 the relative risk function for the pollutant (Equation 3);
 the BoD distribution by age groups and causes (e.g. number of deaths by cause);
 the total BoD (e.g. number of deaths by all causes).

The following example is based on a real sample population. It illustrates how calculations are
performed and potential constraints due to a lack of information, for example, for many age groups
there is no health outcome specific data. The sample population is exposed to PM2.5 at a level of
28µg/m3 (Table 3 to Table 5). In this example, there’s no additional risk due to exposure to PM2.5 for
concentrations equal or below 5 µg/m3, which is the counterfactual in this case. The actual RR values
would be provided by WHO as a table.

ALRI is of special relevance for young children, that’s why only the age group 0-4 years is of
relevance. The other four outcomes, COPD, IHD, Stroke and LC are only considered for adults (ages
of 25 years or older). Two of those, COPD and LC, have RR values for all ages only, which are applied
to the respective age groups for calculating the number of attributable deaths.

Table 3
3
Relative risks for PM2.5 exposure of 28 µg/m , type of health outcome and age group. Note that age groups start at age
25 years; for some age groups, RR values are not available.

Age group ALRI9 COPD IHD Stroke LC

AllAge 1.21 1.14 1.26 1.39 1.18
25-29 2.13 2.08
30-34 2.03 2.00
35-39 1.93 1.91
40-44 1.83 1.83
45-49 1.65 1.77
50-54 1.60 1.69
55-59 1.54 1.61
60-64 1.47 1.54
65-69 1.40 1.42
70-74 1.34 1.37

9
ALRI is for 0-4 years of age only.

8
30 July 2015
75-79 1.28 1.32
80+ 1.22 1.26

Table 4

Number of annual deaths in a sample population, type of health outcome and age groups. Highlighted cells indicate
values relevant for the computations.

Age group ALRI COPD IHD Stroke LC

Up to 4 33 1 0 - -
5-9 1 0 1 0 -
10-14 2 0 0 0 -
15-19 3 0 6 2 -
20-24 15 3 20 4 1
25-29 26 4 81 14 1
30-34 82 7 168 25 3
35-39 175 13 254 42 10
40-44 102 18 521 100 37
45-49 116 48 1 065 223 110
50-54 169 95 1 889 443 277
55-59 136 110 2 782 729 449
60-64 113 173 3 993 1 159 495
65-69 43 193 3 979 1 345 435
70-74 42 339 6 787 2 444 462
75-79 30 506 9 917 3 749 335
80+ 43 620 23 082 8 545 335
Total 1 132 2 131 54 545 18 825 2 951

Table 5
3
Attributable deaths for a sample population due to ambient air pollution at PM2.5 = 28 µg/m

ALRI COPD IHD Stroke LC

Up to 4 6 - - - -
5-9 - - - - -
10-14 - - - - -
15-19 - - - - -
20-24 - - - - -
25-29 - 1 43 7 0
30-34 - 1 85 13 1
35-39 - 2 123 20 2
40-44 - 2 236 45 6
45-49 - 6 420 97 17
50-54 - 12 708 181 42
55-59 - 14 976 276 68
60-64 - 21 1 277 406 76
65-69 - 24 1 137 398 66
70-74 - 42 1 722 660 70

9
30 July 2015
75-79 - 62 2 169 909 51
80+ - 76 4 162 1 763 51
Total 6 261 13 057 4 775 450
Number of attributable deaths, all health outcomes: 18 549
Total number of deaths in the sample population by all causes in the relevant age groups: 122 168

In this example, the total number of deaths (122 168) corresponds to the total BoD of Equation 9. It
is the sum of all annual deaths in the sample population for the age groups 0-4 years plus 25 years
and older.

Data input
The data can be entered by the user, but some defaults will be available after installation, for
example relative risk values and total BoD by country.

The level of exposure needs to be calculated based on the air pollution data that the user has
available: two main possibilities are possible:

1) the input is one average value for a given period of time (that is a common case when a user
has no access to primary data from monitoring stations, satellite data, or dispersion
modelling;
2) the user has a complete time series of daily averages for one year; then AirQ+ calculates the
average yearly value. The original data can be a single table that covers data from one
source (for example one monitoring station) or multiple sources (two or more monitoring
stations). The user has to input one table of air pollution data. In the case of PM or NOx the
data input are the daily average values in µg/m3. The case of Ozone is different because the
concentration data are calculated as SOMO3510, Sum of Means over 35 PPB (parts per
billion) that corresponds to 70 µg/m3.

Ozone

Calculations for Ozone follow the same procedure used for PM, but the differences between the
average concentrations that the population is exposed and a counterfactual have not to be
considered in the formula because the levels of Ozone are calculated against a value, usually 35ppb.
The user should be interested in measuring the burden of Ozone when it has a certain value, for
example 45ppm. In this case the calculations of the exposure levels, measured by the level of
pollution, should be done outside.

10
𝑆𝑂𝑀𝑂35 = ∑𝑖 𝐴𝐵𝑝𝑜𝑙𝑙𝑢𝑡𝑎𝑛𝑡 𝑚𝑎𝑥 (0, 𝐶𝑖 − 35𝑝𝑝𝑚)
where Ci is the maximum daily 8-hour average concentration and the summation is from day i=1 to
365 per year.

10
30 July 2015
Annex 1

Recommended concentration – response functions for particulate matter, long-term exposure

(HRAPIE project: WHO Regional Office for Europe, 2013)

Table: RRs for air pollutants (PM10, PM2.5, NO2, O3 and BC).

Pollutant Health outcome Population Range of

metric concentration
RR (95% CI) per 10 μg/m3
PM2.5 Mortality, all-cause (natural) age 30+ years 1.062 All

Annual (1.040 – 1.083)

mean

PM2.5 Mortality, cerebrovascular disease age 30+ years CRFs used in the GBD All
(includes stroke), ischaemic heart 2010 study (see IER
Annual disease, chronic obstructive section)
mean pulmonary disease (COPD) and
trachea, bronchus and lung cancer

PM10 Postneonatal infant mortality, all- age 1 – 12 1.04 All

cause months)
Annual (1.02–1.07)
mean

PM10 Prevalence of bronchitis in children age 6 – 12 (or 6 – 1.08 All

18) years
(0.98 – 1.19)

PM10 Incidence of chronic bronchitis in age 18+ years 1.117 All

adults
(1.040 – 1.189)

Source: based on HRAPIE project: recommendations for concentration – response functions of

particulate matter, ozone and nitrogen dioxide (WHO Regional Office for Europe, 2013)

Recommended concentration – response functions for particulate matter, short-term exposure

Pollutant metric Health outcome Population Range of

concentration
RR (95% CI) per 10 μg/m3
PM2.5 Mortality, all-cause (natural) All ages 1.0123 All

daily mean (1.0045 – 1.0201)

PM2.5 Hospital admissions, All ages 1.0091 All

Cardiovascular diseases
daily mean (CVDs) (includes stroke) (1.0017-1.0166)

PM2.5 Hospital admissions, All ages 1.0190 (0.9982 – 1.0402) All

respiratory diseases
daily mean

11
30 July 2015
PM2.5 two-week Restricted activity days All ages 1.047 All
average, converted (RADs)
to PM2.5, annual (1.042–1.053)
average

PM2.5 two-week Work days lost, working-age age 20-65 years 1.046 All
average, converted population
to PM2.5, annual (1.039–1.053)
average

PM10 Incidence of asthma 5-19 years 1.028 All

symptoms in asthmatic
daily mean children (1.006–1.051)

Source: based on HRAPIE project: recommendations for concentration – response functions of

particulate matter, ozone and nitrogen dioxide (WHO Regional Office for Europe, 2013)

Recommended concentration – response functions for ozone: long-term

Pollutant metric Health outcome Population Range of

concentration
RR (95% CI) per 10 μg/m3
O3, summer Mortality, respiratory age 30+ years 1.014 >35 ppb
months (April– diseases
September), (1.005–1.024) (>70 μg/m³)
average of daily
maximum 8-hour
mean over 35 parts
per billion (ppb)

Source: based on HRAPIE project: recommendations for concentration – response functions of

particulate matter, ozone and nitrogen dioxide (WHO Regional Office for Europe, 2013)

Recommended concentration – response functions for ozone: short-term

Pollutant metric Health outcome Population Range of

concentration
RR (95% CI) per 10 μg/m3
O3, daily maximum Mortality, all (natural) causes All ages 1.0029 (1.0014–1.0043) >35 ppb
8-hour mean
(>70 μg/m³)

O3, daily maximum Mortality, all (natural) causes All ages 1.0029 >10 ppb
8-hour mean
(1.0014–1.0043) (>20 μg/m³)

O3, daily maximum Mortality, CVDs and All ages CVD: 1.0049 >35 ppb
8-hour mean respiratory diseases
(1.0013–1.0085); (>70 μg/m³)
respiratory: 1.0029
(0.9989–1.0070)

O3, daily maximum Mortality, CVDs and All ages CVD: 1.0049 (1.0013– >10 ppb
8-hour mean respiratory diseases 1.0085); respiratory:
1.0029 (0.9989–1.0070) (>20 μg/m³)

O3, daily maximum Hospital admissions, CVDs age 65+ years CVD: 1.0089 (1.0050– >35 ppb
8-hour mean (excluding stroke) and 1.0127); respiratory:
respiratory disease 1.0044 (1.0007–1.0083) (>70 μg/m³)

12
30 July 2015
O3, daily maximum Hospital admissions, CVDs age 65+ years CVD: 1.0089 (1.0050– >10 ppb
8-hour mean (excluding stroke) and 1.0127); respiratory:
respiratory disease 1.0044 (1.0007–1.0083) (>20 μg/m³)

O3, daily maximum Minor restricted activity days All ages 1.0154 (1.0060–1.0249) >35 ppb
8-hour mean (MRADs)
(>70 μg/m³)

O3, daily maximum Minor restricted activity days All ages 1.0154 (1.0060–1.0249) >10 ppb
8-hour mean (MRADs)
(>20 μg/m³)

Source: based on HRAPIE project: recommendations for concentration – response functions of

particulate matter, ozone and nitrogen dioxide (WHO Regional Office for Europe, 2013)

Recommended concentration – response functions for nitrogen dioxide: long-term

Pollutant metric Health outcome Population Range of

concentration
RR (95% CI) per 10 μg/m3
NO2, annual mean Mortality, all (natural) causes age 30+ years 1.041 >20 μg/m³

(1.019–1.064)

NO2, annual mean Prevalence of bronchitic Age 5–14 years 1.021 (0.990–1.060) per All
symptoms in asthmatic 1μg/m³ change in annual
children mean NO2

Source: based on HRAPIE project: recommendations for concentration – response functions of

particulate matter, ozone and nitrogen dioxide (WHO Regional Office for Europe, 2013) and Faustini
et al. (2014)

Recommended concentration – response functions for nitrogen dioxide: short-term

Pollutant metric Health outcome Population RR (95% CI) per 10 μg/m3 Range of
concentration

NO2, daily Mortality, all (natural) causes All ages 1.0027 All
maximum 1-hour
mean (1.0016 – 1.0038)

NO2, daily Hospital admissions, All ages 1.0015 All

maximum 1-hour respiratory diseases
mean (0.9992 – 1.0038)

NO2, daily Hospital admissions, All ages 1.0180 All

maximum 24-hour respiratory diseases
mean (1.0115 – 1.0245)

Source: based on HRAPIE project: recommendations for concentration – response functions of

particulate matter, ozone and nitrogen dioxide (WHO Regional Office for Europe, 2013)

Recommended concentration – response functions for black carbon: long-term

13
30 July 2015
Pollutant metric Health outcome Population Range of
concentration
RR (95% CI) per 10 μg/m3
BC, Annual mean Mortality, all (natural) causes age 30+ years 1.06 All

(1.04 – 1.09)

For Household air pollution, a table with data on the estimation of the yearly use of solid fuels is
needed for calculations. See example below.

Country Data source Population (%) using solid fuels

Year: 2013
Afghanistan Multi-level 80
Albania Multi-level 37
Algeria Multi-level <5
Andorra High-income <5
Angola Multi-level 54
Antigua and Barbuda Multi-level <5
… … …
… … …
… … …
… … …
Venezuela (Bolivarian Republic of) Multi-level <5
Viet Nam Multi-level 47
Yemen Multi-level 32
Zambia Multi-level 82
Zimbabwe Multi-level 71

14
30 July 2015