Methodology

Overview of potential sources of emissions of indoor air pollutants

Pollutants measured in the indoor air originate from both indoor and outdoor sources. It is widely recognised that the most important indoor sources of pollution in schools are building materials, furnishings, cleaning products, toiletries, stationery and human activities. The indoor air can also be polluted by unfiltered outdoor air containing pollutants emitted primarily by traffic and industrial facilities.

Sources of information

Valuable information was provided via two questionnaires. The classroom questionnaire was used to obtain information on furnishings, consumer products, cleaning products, stationery etc., as well as occupants’ behaviour and indoor activities. The school questionnaire was used to gather information on building characteristics (building materials, floor covering, classroom size and air volume, size of openable windows etc.).

Preparation of the in-depth analysis

The indoor air pollution data collected in the course of the two SEARCH projects were available for the in-depth analysis. The starting point was to identify the extremely high levels of indoor air pollution in the classrooms. These extremely high values clearly had to be in the range of outliers, thus the first step was to define those outliers.

Several approaches to identifying outliers can be found in the literature, with the recommendation to select the most appropriate method according to the subject. For the in-depth analysis of the SEARCH database, the Tukey method was considered the most appropriate.

A summary of the relevant findings in the literature, the regulatory framework for IAQ, figures, the statistical evaluation process, and the associations between environmental parameters and children’s health status can be found by clicking on the link below.


Additional information

Summary of relevant findings in the literature

In the last few decades, the issue of indoor air quality has become increasingly prominent. Several studies have focused on the evaluation of indoor air pollution levels, the detection of sources and the determination of the health effects of indoor air pollutants.

Research has been carried out in various indoor environments (office, home, public buildings), although studies on the exposure of children to pollution, especially in the indoor school environment, have been limited. A brief overview of air quality measurements carried out in classrooms is provided below, in order to contribute to the assessment of the large amount of SEARCH data.

J.M. Daisey et al. (2003) reviewed the literature on indoor air quality in schools up to 1999, comprising more than 300 articles. Some of the reported data are highlighted here.

Mean concentrations of CO2 varied between 1,420 and 1,850 ppm in two schools in Sweden and were in the range of 500 to 1,500 ppm (mean = 1,000 ppm) in 11 Danish schools. The level of CO2 pollution ranged from about 400 to 5,000 ppm (mean = 1,480 ppm) in nine schools in the US. It was also concluded that CO2 concentrations exceeded the 1,000 ppm ventilation standard of the American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE) in 74 percent of the rooms.

With respect to concentrations of formaldehyde in classrooms, the average pollution level in schools was 0.01 ppm (12.5 µg/m3) before and during teaching time. Mean concentrations of formaldehyde reported in 10 schools in Milan, Italy, and 10 schools in Paris, France, were also around 0.05 ppm (62.5 µg/m3).

Concentrations of formaldehyde were monitored in 104 classrooms in California, and indoor pollution levels were found with an average concentration of 21 ppb (26 µg/m3) and a maximum of 98 ppb (122 µg/m3) (ASHRAE).

Indoor air quality was evaluated by H. Fromme et al. (2007) in 64 schools in and around Munich, Germany, during the 2004–2005 winter season. In 60 percent of the classrooms, the daily median CO2 concentration exceeded 1,500 ppm.

The most commonly known VOC pollutants (benzene, toluene and formaldehyde) were measured during three seasons (spring, winter and autumn) in Izmir, Turkey. The 95th percentile of contaminants was 29 µg/m3 (benzene), 87 µg/m3 (toluene) and 106 µg/m3 (formaldehyde). According to S.C. Sofuoglu et al. (2011), formaldehyde was considered the most important compound due to its high chronic toxic and carcinogenic risk levels.

A one-year sampling programme was carried out in order to investigate the VOCs pollution level in one school (four classrooms) in Seoul, Korea, in 2010–2011 (Lim-Kyu Lee et al. 2012). The concentrations varied between 0.7 and 12; 33 and 3,757; 56 and 117; and 76 and 3,978 µg/m3 for benzene, toluene, xylenes and ethylbenzene respectively.

Regulation of indoor air quality

Although outdoor air pollution levels have been regulated for decades, air quality in the indoor environment is not sufficiently regulated. Only partial regulations and indoor air guidelines have been developed to date. The World Health Organization (WHO) European Centre for Environment and Health, Bonn Office, supported by the WHO Regional Office for Europe, published guidelines in 2010 related to nine indoor air pollutants, including benzene, NO2 and formaldehyde.

Several countries have also developed their own sets of rules on indoor air pollution. Table A15 shows the WHO guidelines as well as recommendations set out by various organisations for levels of other pollutants investigated in the SEARCH initiative.

Table A15 Guidelines and recommendations for concentrations of indoor air pollutants

POLLUTANT

VALUE

AVERAGING TIME

REFERENCE

CH2O

(µg/m3)

100

30 minutes

WHO Guidelines for Indoor Air Quality: Selected Pollutants, 2010

123

50

1 hour

8 hours

Residential Indoor Air Quality

Guidelines, Health Canada

NO2

(µg/m3)

200

40 40

1 hour

annual

1 week

WHO Guidelines for Indoor Air Quality: Selected Pollutants, 2010

Residential Indoor Air Quality Guidelines, Health Canada

Benzene (µg/m3)

No safe level of exposure can be recommended

 

WHO Guidelines for Indoor Air Quality: Selected Pollutants, 2010

2

not specified

Decree of the Flemish Government of June 11, 2004, providing for measures aimed at controlling health risks caused by indoor pollution (2004)

CO2 (µg/m3)

1,025

in METs=1*

American Society of Heating, Refrigerating and Air-Conditioning Engineers

PM10 (µg/m3)

40

24 hours

Decree of the Flemish Government of June 11, 2004, providing for measures aimed at controlling health risks caused by indoor pollution (2004)

Toluene (µg/m3)

2,300

1 year

Residential Indoor Air Quality Guidelines, Health Canada

Science Assessment Document: Toluene. Ottawa: Minister of Health, 2011 (cited by Amanda J. et al.

300

precautionary value (RW 1)

Federal Environmental Agency of Germany (Umweltbundesamt – UBA) Germany, 2011

3,000

health hazard for sensitive people (RW 2) especially

15,000

2,300

8 hours

24 hours

Residential Indoor Air Quality Guidelines, Health Canada

Science Assessment Document: Toluene. Ottawa: Minister of Health, 2011.

*METs: metabolic equivalent of task, i.e. energy cost of physical activities

REMARKS:

In the framework of the European Union INDEX project, the working group recommended 30 µg/m3 (30 minutes) for formaldehyde and 40 µg/m3 (one week) for NO2 as indoor guidelines.

In the case of xylenes, a long-term indoor exposure limit equal to 200 μg/m3 has been established (ASHRAE 2004).

To date there is no regulation for the indoor air environment for ethylbenzene.

Evaluation of indoor air quality

In practice, indoor exposure levels are assessed on the basis of existing guidelines and recommendations. Unfortunately, the indoor air pollution measured during SEARCH I and II cannot be evaluated in this way due to the differences between the sampling times used in the SEARCH initiative and those specified in the guidelines/recommendations. The guidelines can therefore be considered as indicative only.

A comparison of the SEARCH I and II environmental data with reference values makes possible a general characterisation of the magnitude of IAQ problems.

In-depth analysis

Selection of extremely high values in the data

One precondition for any in-depth analysis is the existence of a reliable and precise database, including the appropriate selection of outliers and extremely high data values as well.

It is not a straightforward task to find the outliers in a database. Outliers may be the result of errors (measurement errors, clerical errors, data entry errors etc.), but they may also be real values.

Before determining the group of outliers, the entire database was revised in order to identify unrealistic values. Following the review of the database, only two pieces of data (17,040 µg/m3 CO2 and 401 µg/m3 toluene) were deleted. The high CO2 indoor concentration seemed unlikely, while the large value for toluene was outdoor data. The other data were considered to be valid values.

The limit lines (inner and outer fences) for the selection of extreme data for each pollutant were calculated using Q3 values and the interquartile ranges shown in Table A16.

Table A16 Outlier boundaries (µg/m3) for each pollutant

POLLUTANT

INNER FENCE

OUTER FENCE

CO2

3,292

4,576

PM10

185

276

Benzene

9.5

14

Toluene

29

44

Ethylbenzene

3.3

4.8

Xylenes

18

27

NO2

35

50

CH2O

21

31

For CO2, benzene, toluene, ethylbenzene and formaldehyde, concentrations above the outer fence were used in the in-depth analysis. Some of the data exceeded the outer fence for PM10, NO2 and xylenes, therefore data higher than the suspected outliers were integrated into the analysis.

Indoor and outdoor air pollution

Due to the process of infiltration, indoor air quality is affected by ambient air pollution to some extent. In order to understand the relationship between indoor and outdoor concentrations, it is important to make a more accurate judgment about the associations between classroom indicators (pollution sources) and indoor pollution levels.

Figure A11

Ratio of indoor and outdoor mean pollution levels

presents the ratio of mean indoor and outdoor concentrations of each pollutant. Mean values were calculated from the whole dataset of the 10 SEARCH countries.

The ratio <1 shows that NO2 and PM10 primarily originate from outdoors. In the case of benzene, as the ratio is very close to the value 1, it was assumed that benzene could be emitted from both indoor and outdoor sources.

Emission sources for the other measured pollutants can be looked for principally in the classrooms.

 

The averages for indoor and outdoor pollution levels are presented for each country in Figures A12 to A19: CO2

(
Figure A12)

Average indoor and outdoor levels of CO2 per country

, PM10

(
Figure A13)

Average indoor and outdoor levels of PM10 per country

, benzene

(
Figure A14)

Average indoor and outdoor levels of benzene per country

, toluene

(
Figure A15)

Average indoor and outdoor levels of toluene per country

, ethylbenzene

(
Figure A16)

Average indoor and outdoor levels of ethylbenzene per country

, xylenes

(
Figure A17)

Average indoor and outdoor levels of xylenes per country

, NO2

(
Figure A18)

Average indoor and outdoor levels of NO2 per country

and formaldehyde

(Figure A19).

Average indoor and outdoor levels of formaldehyde per country



In each country, NO2 was definitely considered as an outdoor pollutant. In six of the countries, outdoor NO2 concentrations were two to four times higher than indoor concentrations. The concentrations of CO2 in the air inside the classrooms exceeded the values measured in the ambient air (eight times higher in Tajikistan, and four times higher in the other countries). There was a relatively large variation in indoor formaldehyde pollution between the countries. The highest level of pollution was found in the schools in Italy. In five countries, slightly higher levels of benzene were measured indoors than in the ambient air. According to the measurements, mean concentrations of PM10 outdoors in half of the SEARCH countries exceeded the pollution level measured inside the classrooms. This was especially true for schools in Bosnia and Herzegovina, where ambient air pollution was two times higher than the level of pollution indoors. Compared to the outdoor air, ethylbenzene pollution was higher in classrooms in all the SEARCH countries except Serbia, where the indoor mean concentration was lower than in the ambient air. In the case of toluene and xylenes, pollution levels were higher indoors than outdoors. The smallest difference between indoor and outdoor mean concentrations of toluene were found in Belarus and Italy. The highest concentrations of xylenes in the ambient air compared to the indoor level in classrooms were found in Hungary.

Indoor concentrations of toluene, xylenes and ethylbenzene were far higher than the respective outdoor concentrations, indicating the presence of significant indoor sources of these pollutants (including building, furnishing and cleaning products and products used in children’s daily activities).

Statistical processing

The second step in the in-depth analysis was the detailed statistical processing of the measurement results.

One of the statistical methods for presenting the differences or relationships between two groups out of the whole indoor air database is to use data below the outer fence, which can be considered as the remaining dataset. This subgroup comprises around 98 percent of the entire database.

A second subgroup is the set of extremely high values above the outer fence. These data represent around 2 percent of the entire database.

The in-depth analysis was carried out using:

  • the most characteristic parameters of both datasets performed by descriptive statistics; and
  • the relationship between different variables carried out by logistic regression, such as school building characteristics and indoor air pollution, and the association between indoor air pollution and the characteristic health conditions of children.

Descriptive statistics

Indoor air pollution data collected in the 10 SEARCH countries were divided into two groups — a set of remaining data and a subgroup of extremely high value data. Both datasets were characterised and presented using descriptive statistical tools, such as measures of central tendency, spread, correlation and frequency distribution.

General indicators

General indicators (maximum values and the number of valuable data in each group) were calculated and are presented in Table A17.

Table A17 General indicators

Indicator

Group of

data

CO2

PM10

Benzene

Toluene

Ethyl-benzene

Xylenes

NO2

CH2O

Maximum µg/m3

Remaining

4,576

197

14

44

5

18

35

34

Extreme

5,265

318

30

667

13

69

54

86

Number of data

Remaining

309

355

365

365

362

365

355

366

Extreme

7

8

8

8

8

9

13

10

The maximum values for each pollutant are equal to the outer fences in the case of the remaining dataset.

Central values

The calculated central values are shown in Table A18.

Table A18 Central values (mean, median, mode) for two groups of indoor air pollutants (two datasets)

Indicator

Group of

data

CO2

PM10

Benzene

Toluene

Ethyl-
benzene

Xylenes

NO2

CH2O

Mean

µg/m3

Remaining

1,714

64

4

11

1.2

6

15

7

Extreme

4,324

227

22

145

5.8

27

43

70

Median

µg/m3

Remaining

1,492

53

3

8

1

5

13

6

Extreme

4,092

202

23

68

5

22

44

72

Mode

µg/m3

Remaining

833

58

4

4

0.6

3

14

2

Extreme

4,611

NA

NA

NA

5

NA

42

NA

NA: not applicable

Large differences in central values are absolutely acceptable due to the two different datasets.

Dispersion indicators

The statistics presented in Table A19 are the standard deviations (SD) and variances (V) calculated for all pollutants.

Table A19 Standard deviation and variance values for all indoor pollutants

Indicator

Group of

data

CO2

PM10

Benzene

Toluene

Ethyl-
benzene

Xylenes

NO2

CH2O

SD

µg/m3

Remaining

833

39

3

9

0.8

4

7

5

Extreme

576

55

5

212

2.4

15

5

12

Variance

Remaining

693,515

1,516

8

81

0.7

16

50

26

Extreme

279,967

3,062

23

45,047

7.4

220

27

155

The biggest difference between standard deviation and variance was observed in the case of toluene. This is a logical consequence of the two very high concentrations (667 and 114 g/m3).

Frequency distribution

Frequency distribution is a commonly used method to categorise indoor air data. It was advisable to use the entire indoor air quality dataset collected from the 10 SEARCH countries in order to show the extreme values as well.

Frequency distributions with five to eight categories, depending on the type of pollutant, are depicted in Figures A20 to A27: CO2

(Figure A20),

Absolute and relative frequency of CO2

PM10

(Figure A21),

Absolute and relative frequency of PM10

benzene

(Figure A22),

Absolute and relative frequency of benzene

toluene

(Figure A23),

Absolute and relative frequency of toluene

ethylbenzene

(Figure A24),

Absolute and relative frequency of ethylbenzene

xylenes

(Figure A25),

Absolute and relative frequency of xylenes

NO2

(Figure A26)

Absolute and relative frequency of NO2

and formaldehyde

(Figure A27).

Absolute and relative frequency of formaldehyde

The figures show both absolute and relative frequency distributions.



Figures A20 to A27 show that most indoor values of monitored pollutants are in the lowest range, except CO2 and PM10. In the case of CO2, half the data were between 1,100 and 2,000 µg/m3. In the case of PM10, 40 percent of the data are in the range of 40 to 100 µg/m3. The figures also show values that are in the range of outliers. Table A20 shows the cumulative relative frequency distribution for each component.

Table A20 Frequency values (µg/m3) and interquartile range for each indoor air pollutant

RANGE (%)

CO2

PM10

Benzene

Toluene

Ethyl-
benzene

Xylenes

NO2

CH2O

10

929

22

1

3

0.5

2

7

2

25 (Q1)

1,152

32

2

4

0.6

3

10

3

50 (Q2)

1,492

53

3

8

1.1

5

14

7

75 (Q3)

2,008

93

5

14

1.8

9

20

10

90

2,889

126

8

26

2.4

13

28

14

95

3,341

153

12

36

3.3

16

32

20

Measures of correlation

In order to describe the relationship between pollutants, the correlation coefficients were calculated (Table A21).

Table A21 Pearson correlation coefficients
 

CO2

PM10

Benzene

Toluene

Ethyl-
benzene

Xylenes

NO2

CH2O

CO2

1.00

             

PM10

0.14

1.00

           

Benzene

0.23

0.38

1.00

         

Toluene

0.19

0.28

0.49

1.00

       

Ethyl-

benzene

0.11

0.19

0.37

0.26

1.00

     

Xylenes

0.06

0.15

0.37

0.26

0.91

1.00

   

NO2

-0.06

0.26

0.21

0.19

0.08

0.06

1.00

 

CH2O

0.04

0.04

0.01

0.01

0.11

0.10

-0.14

1.00

The strength of the linear dependence between the components was obtained by Pearson correlation. Table A21 shows that the strongest associations were between ethylbenzene and xylenes (r=0.91). The large correlation coefficient indicates that both pollutants are likely to have originated from the same sources. Benzene was moderately correlated with PM10 and with the other BTEX components (r=0.37 to 0.49).

Regression analysis

Logistic regression was considered appropriate to measure the relationship between indoor air quality concentrations and classroom characteristics, as well as between children's health status and their exposure indoors. To estimate the strength of the relationship, the odds ratios (ORs) and 95 percent confidence intervals (CIs) were computed.

Associations between indoor air concentrations and classroom characteristics

In the relationship between indoor air quality concentrations and classroom characteristics, the independent variables were the classroom parameters, while the indoor air concentrations were considered as dependent variables (divided into two subgroups).

Regression coefficients (OR) with 95 percent confidence intervals and their significance (significant codes: *** <0.001; ** <0.01; * <0.05) for those pollutants that were significantly associated with classroom characteristics are presented in Table A22.

Table A22 Significant associations between indoor air concentrations and classroom characteristics

Classroom characteristics

Pollutant

Odds ratio

95% confidence interval

Class floor level

Level -2 to level 4 vs. level -1 to level 1

NO2

9.19*

1.14 – 74.46

Flooring

Stone/concrete

NO2

5.34*

1.37 – 20.80

Stone with carpet

NO2

Benzene

6.47**

20.83**

1.67 – 25.00

2.28 – 190.47

Windows opened

Window surface >=2 m2

NO2

9.54*

1.13 – 80.73

Not opened

NO2

4.82*

1.18 – 19.73

Air conditioning

 

CO2

13.56*

1.23 – 149.93

Cleaning

Morning

CH2O

44.35***

5.02 – 391.94

Evening

CH2O

0.13*

0.03 – 0.69

Mop with bleach

Toluene

8.05**

1.87 – 34.61

Occupation density

<1.5 m2/person

Benzene

6.38*

1.04 – 39.06

Wall treatment

Water-resistant paint

Ethylbenzene

15.49*

1.78 – 134.74

The highest indoor level of NO2 indicated significant associations with almost all classroom characteristics:

  • floor level of the classroom on the ground floor and basement (OR: 9.19; 95% CI: 1.14–74.46)
  • openable windows >2m2 (OR: 9.54; 95% CI: 1.13–80.73)
  • closed windows (OR: 4.82; 95% CI: 1.18–19.73)
  • stone/concrete floor (OR: 5.34; 95% CI: 1.37–20.80)
  • stone floor covered with carpet (OR: 6.47; 95% CI: 1.67–25.00)

Extremely high benzene pollution levels measured in the classrooms were significantly associated with the presence of carpeted floors (OR: 20.83; 95% CI: 2.28–190.47) and with overcrowding (<1.5 m2/person) (OR: 6.38, 95% CI: 1.04–39.06).

Extremely high indoor concentrations of toluene were significantly associated with cleaning using mop and bleach (OR: 8.05, 95% CI: 1.87–34.61).

The use of water-resistant paint on the walls showed a significant impact on ethylbenzene, with results in the highest pollution range in the classroom air (OR: 15.49; 95% CI: 1.78–134.74).

The results indicate that morning cleaning in the classrooms (OR: 44.35; 95% CI: 5.02–391.94) was significantly associated with the highest levels of formaldehyde.

Associations between children’s health status and concentrations of indoor air pollutants

Schoolchildren were categorised according to two groups: those exposed to extremely high levels of pollution; and those exposed to lower concentrations.

In the analysis, the independent variables were indoor air pollution values measured during teaching time. Children’s health-related symptoms were considered as dependent variables. Table A23 shows the health outcomes among schoolchildren.

Table A23 Significant associations between indoor concentrations of pollutants and children’s health status

Health outcomes

Pollutant

OR

CI

Allergy

NO2

Xylenes

1.61**

1.56*

1.17–2.21

1.06–2.28

- diagnosed allergy

NO2

Xylenes

1.73***

1.68**

1.26–2.38

1.13–2.49

- doctor-diagnosed allergy to dust mites

NO2

1.64*

1.06–2.52

- pollens excluding ragweed

NO2

1.70**

1.14–2.52

- doctor-diagnosed allergy to pollens excluding ragweed

NO2

1.81**

1.19–2.76

- all pollens

NO2

1.49*

1.01–2.20

- doctor-diagnosed allergy to all pollens

NO2

1.65*

1.09–2.49

- doctor-diagnosed allergy to animals

NO2

1.61*

1.07–2.41

Sleep disorder

Xylenes

1.82*

1.01–3.28

Fatigue

NO2

1.93***

1.40–2.65

Attention deficit disorder

NO2

1.82**

1.24–2.68

Irritability

NO2

1.66**

1.18–2.34

Anxiety

NO2

CH2O

1.66**

1.85**

1.10–2.49

1.07–3.20

Symptoms of depression

NO2

1.63**

1.20–2.23

Chronic cough symptoms in the last 12 months

CH2O

1.71*

1.09–2.68

Woken by wheezing in the last 12 months

Toluene

2.53***

1.51–4.26

Morning cough in autumn/winter

PM10

2.50***

1.64–3.81

Conjunctivitis

Ethylbenzene

Xylenes

2.14**

2.06*

1.25–3.67

1.18–3.58

Earache

Ethylbenzene

Toluene

1.73*

1.83**

1.08–2.77

1.22–2.73

Complications of sinusitis or earache

Ethylbenzene

Toluene

1,62*

1.63**

1.06–2.47

1.13–2.35

OR: odds ratio; 95% CI: confidence interval

Significance codes: *p<0.05 ** p<0.01 ***p<0.001

Children’s exposure to NO2 at extremely high levels in the classroom raised the risk of allergy (to OR: 1.61–1.73) fatigue (to OR: 1.93), attention deficit disorder (to OR: 1.82), irritability (to OR: 1.66), anxiety (to OR: 1.66) and symptoms of depression (to OR: 1.63).

It was found that extremely high PM10 concentrations were associated with an increased risk of regular morning coughing (to OR: 2.50; 95% CI: 1.64–3.81). This means that children exposed to PM10 pollution during teaching time were 2.5 times more likely to have a cough every morning compared to children who were less exposed.

The presence of toulene in the classroom air was significantly related to earache (OR: 1.83; 95% CI: 1.22–2.73), sinusitis or earache complications (OR: 1.33; 95% CI: 1.13–2.35), and being woken by wheezing in the last 12 months (OR: 2.53; 95% CI: 1.51–4.26).

The effect of indoor air pollution from ethylbenzene on the occurrence of conjunctivitis (OR: 2.14; 95% CI: 1.25–3.67), earache (OR: 1.73; 95% CI: 1.08–2.77), and sinusitis or earache complications (OR: 1.62; 95% CI: 1.06–2.47) in children was significant.

Children who were exposed to extremely high levels of indoor air pollution from xylenes were twice as likely to become ill with conjunctivitis and one and a half times more likely to suffer from an allergy or sleep disorder compared to children who were not exposed to such high levels of pollution in the classroom.

It was supposed that the highest concentrations of formaldehyde in the classroom air had an impact on children’s health. A significant association was found between children’s exposure and chronic cough symptoms in the last 12 months (OR: 1.71; 95% CI: 1.09–2.68) and anxiety (OR: 1.85; 95% CI: 1.07–3.20).




 
Ministero Dell'ambiente Italian Trust Fund