- Email: [email protected]
Metal induced inhalation exposure in urban population: A probabilistic approach
Atmospheric Environment 128 (2016) 198e207
Contents lists available at ScienceDirect
Atmospheric Environment journal homepage: www.elsevier.com/locate/atmosenv
Metal induced inhalation exposure in urban population: A probabilistic approach Kamila Widziewicz a, *, Krzysztof Loska b a b
Institute of Environmental Engineering, Polish Academy of Sciences, M. Skłodowska-Curie 34, 41-819 Zabrze, Poland Faculty of Energy and Environmental Engineering, The Silesian University of Technology, Konarskiego 18 A, 44-100 Gliwice, Poland
h i g h l i g h t s We model health risks caused by inhalation exposure to airborne metals. We examine changes in the level of risk between PM2.5 and PM10-bound metals. PM2.5-bound metals poses a higher risk than PM10 e associated one. Infants and children are especially sensitive to airborne metals. Risk value depend on PM composition, as well as the variability of exposure.
a r t i c l e i n f o
a b s t r a c t
Article history: Received 1 September 2015 Received in revised form 22 December 2015 Accepted 24 December 2015 Available online 29 December 2015
The paper was aimed at assessing the health risk in the populations of three Silesian cities: Bielsko-Biała, Cze˛ stochowa and Katowice exposed to the inhalation intake of cadmium, nickel and arsenic present in airborne particulate matter. In order to establish how the exposure parameters affects risk a probabilistic risk assessment framework was used. The risk model was based on the results of the annual measurements of As, Cd and Ni concentrations in PM2.5 and the sets of data on the concentrations of those elements in PM10 collected by the Voivodship Inspectorate of Environmental Protection over 2012e2013 period. The risk was calculated as an incremental lifetime risk of cancer (ILCR) in particular age groups (infants, children, adults) following Monte Carlo approach. With the aim of depicting the effect the variability of exposure parameters exerts on the risk, the initial parameters of the risk model: metals concentrations, its infiltration into indoor environment, exposure duration, exposure frequency, lung deposition efficiency, daily lung ventilation and body weight were modeled as random variables. The distribution of inhalation cancer risk due to exposure to ambient metals concentrations was LN (1.80 106 ± 2.89 106) and LN (6.17 107 ± 1.08 106) for PM2.5 and PM10-bound metals respectively and did not exceed the permissible limit of the acceptable risk. The highest probability of contracting cancer was observed for Katowice residents exposed to PM2.5 e LN (2.01 106 ± 3.24 106). Across the tested age groups adults were approximately one order of magnitude at higher risk compared to infants. Sensitivity analysis showed that exposure duration (ED) and body weight (BW) were the two variables, which contributed the most to the ILCR. © 2015 Elsevier Ltd. All rights reserved.
Keywords: Cancer risk Monte Carlo Particulate matter Metals Silesia region Probabilistic assessment
1. Introduction In recent years, there has been an increasing concern about the exceedance of the permissible concentrations of particulate matter in atmospheric air. The results of epidemiological research reveal
* Corresponding author. E-mail address: [email protected] (K. Widziewicz). http://dx.doi.org/10.1016/j.atmosenv.2015.12.061 1352-2310/© 2015 Elsevier Ltd. All rights reserved.
that exposure to PM causes an increase in mortality rate and incidence of vascular and respiratory diseases, including cancer (Pope et al., 2011). Despite ample scientific evidence confirming the correlation between persistent exposure to PM and the increase in malignant cancer incidence, including lung cancer, it is still unknown, which compounds of the particles are responsible for inducing carcinogenesis (Harrison et al., 2004). Studies into environmental hazard indicate however that PM induced cancers can be attributed to its metallic constituents ('t Mannetje et al., 2011;
K. Widziewicz, K. Loska / Atmospheric Environment 128 (2016) 198e207
Yaman, 2012). In Poland, three metals: lead, cadmium, nickel and one semimetal, arsenic, are monitored in the air. Three of the above elements: As, Ni and Cd have been classified as factors with sufficient evidence of carcinogenicity in humans (IARC, 2014). Most of the available information about the health effects of those elements is derived from studies, which monitored As, Cd and Ni contents in PM10 (Sadovska, 2012; Trojanowska and Swietlik, 2012). Scientific investigations do not usually deal with the health effects of exposure to the above metals based on their concentrations in PM2.5. This results from the fact that before 2008, PM2.5 was not required to be monitored. However, since PM2.5 constitutes approximately 70% of PM10 mass (Gomis cek et al., 2004) and is deposited in the deepest areas of the respiratory track, it probably enables more accurate prediction of risk (WHO, 2013). The assessment of the health effects of exposure to PM and metals in urban areas is relevant because of their numerous emission sources and the size of the population exposed. Since metals are widely present in urban air and their carcinogenic properties are widely known, the assessment of their harmfulness to public health is necessary. The methodological difficulty in assessing the inhalation hazard caused by airborne carcinogenic metals lies in the complexity of their effects on a population, such as variability of metal concentrations in air, indoor/outdoor pollution ratio, exposure duration, physiological parameters (lung ventilation rate influencing deposition efficiency in the respiratory) and variability of individual characteristics in a population exposed (age and gender) (Biesiada, 2001). Those factors affect the amount of metals taken in, thus significantly determining the final risk. Monte Carlo technique involving the so called Probability Risk Assessment allows the variability of those risk parameters to be taken into account, while assessing risk and determining their effect on the final health risk value. Such approach gives more complex risk characteristics and constitute a source of additional information for risk managers. 1.1. Objectives The primary objective of this study was: to assess the inhalation risks to the residents of three Silesian cities e exposed to PM2.5 and PM10 associated metals by using Monte Carlo approach (1), to determine the influence of exposure parameters variability on health risk value (2), to identify those exposure parameters, which are the key contributors to over all risk value (3). 2. Materials and methods 2.1. Characteristics of the study area The Upper Silesian Region (Fig. 1) is one of the most polluted areas in Poland and is home to numerous extractive, electrical, power engineering and metallurgical industries. Its nature based on raw materials and power production, as well as developed transport infrastructure, makes it occupy the first place in terms of excessive PM concentrations. The emission of PM dust in this area ˛ skie, accounts for 22.14% of the over all national emission (Sla 2020þ).
199
sites are shown in Fig. 1. More detailed information concerning sampling sites along with samplers specification can be found in the Supplementary Materials. It must be highlighted that PM2.5 and PM10 data presented in this work in fact come from different sampling sites, where those fractions were collected separately. Throughout the work inhalation hazards were therefore evaluated independently for PM2.5 and PM10. 2.2.1. PM10 Datasets containing PM10 and associated metal concentrations were downloaded from the Silesian Air Monitoring database (www.katowice.pios.gov.pl/slmonpow), operated by the Chief Inspectorate for Environmental Protection (CIEP). The CIEPs laboratories carry out a routine assessment of metal contents in PM10. Metals assays are regularly done in two-week cumulative samples (14 PM10 filters in total). Thus, 26 measurements of metal concentrations in atmospheric air are conducted each year (Table 5). 2.2.2. PM2.5 Fine particles were collected using low volume samplers (2.3 m3/h) equipped with a PM2.5 measuring head onto Ø47 mm Whatman quartz fiber filters. Diurnal samples were sampled at Bielsko-Biała and Cze˛ stochowa urban background sites and additionally at Katowice traffic site. The mass of the sampled particles was determined gravimetrically (Radwag microbalance, resolution 1 mg) according to the BS EN 14907:2006 standard. Before each weighing, the filters were conditioned for at least 48 h at the air temperature of 20 ± 1 C and air relative humidity of 50 ± 5%. 2.3. Metals extraction and samples preparation Single quarter of each PM2.5 filter was digested in a Milestone microwave digestion system (MLS 1200 Mega oven) using 2 cm3 of 65% HNO3 (Baker). The digestion power was held at 1000 W over all digestion stages (time, 6 min). After cooling, the samples were transferred into flasks and made up to 25 cm3 with ultrapure water. Samples were transferred into plastic bottles and kept in ventilated room until the instrumental analysis. Determination of the total arsenic was carried out after As(V) reduction into As(III). For that purpose three remaining quarters of each filter were cut down to pieces smaller than 1 cm in length and placed in a tube. In next step exactly 5 cm3 of HCl (36% w/v), 2.5 cm3 of urea (25% w/v) and l cm3 of KI (25% w/v) was added to this tube. After 30 min (prereduction time), solution was filled to 25 cm3 with ultrapure water, filtrated through 0.45 mm cellulose filter and aspirated to HG-AAS. The complete reduction of the arsenous hydride was achieved in VGA reaction loop by the addition of 10 M HCl and sodium tetrahydroborate (III) in 0.05% (w/v) NaOH. 2.4. Instrumental analysis The instrumental analysis was conducted by means of GT-AAS method (in case of Ni and Cd) and by using HG-AAS (for As determination). All metal concentrations were calculated on PM2.5 mass collected and normalized by the volume of air passed through the filters during the sampling period. Calibration solutions were prepared daily by using Merck single-element stock solutions of 1000 ppm.
2.2. Sampling protocol 2.5. QA/QC In the presented study authors used the data on PM10 and associated metals concentrations collected by the Voivodship Inspectorate of Environmental Protection in Katowice (VIEP Katowice) and their own measurements of the above elements in PM2.5 for the same period (May 2012eApril 2013). Locations of the sampling
Blank filters were mineralized following the same digestion procedure as tested filters. Blank samples have been found to have non-detectable levels of elements of interest. Certified reference material NIES Urban Aerosol was used to verify the accuracy and
200
K. Widziewicz, K. Loska / Atmospheric Environment 128 (2016) 198e207
Fig. 1. Study area and sampling location. The map on the right shows the average concentration of PM2.5 in Europe e concentration above 30 mg/m3 are colored in violet (based on EEA data, reference year 2010). The contour map on the left presents the borders of Silesia administration district with marked sampling site positions. 1a,1b: Bielsko-Biała, 2a,2b: Cze˛ stochowa, 3a,3b: Katowice (Directive, 2008/50/EC), where symbols a and b denotes PM2.5 and PM10 collecting points. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article). This is an original source file from http://www.eea.europa.eu/data-and-maps/figures/pm2-5-annual-average-2010 website
Finf e PM infiltration rate into indoor environment. For j ¼ 1, j ¼ 3 and j ¼ 5 time interval the following distributions were used: in case of PM2.5-bound metals Finf ¼ LN (0.58; 0.11) and for PM10-bound metals ¼ LN (0.39; 0.11) for j ¼ 2,4 the Finf ¼ 1, Di e particles deposition efficiency in the respiratory track (calculated based on MPPD V2.11 modeling results) for specific age group, t e residence time in a j time interval (hr), IRij e inhalation rate in a given age group at a specific time interval (m3 day1), where j denotes the level of daily activity: j ¼ 1 (sleep or nap activity), j ¼ 2, 3, 4 (light activity), j ¼ 5 (sedentary/passive activity) following U.S. EPA 2011.
precision of the analytical method (Table 1). 2.6. Daily exposure level for inhalation Since there is no information regarding metals absorption in the human respiratory, it was assumed that the active dose of inhaled metals is equal to its airborne concentration and calculated following equation (1). To determine the daily changes in metal exposure levels, a day was divided into the 5 time intervals j ¼ 1, 2, 3, 4 and 5, which corresponded to the following hour intervals: 00:00e06:00, 07:00e08:00, 08:00e16:00, 16:00e18:00 and 18:00e24:00, where j ¼ 1 denotes the time of a nap/sleep spent at home; j ¼ 2 time a man spent on a way to school/work; j ¼ 3 an 8-h work or school time; j ¼ 4 time a man spent on a way from school/ work and finally j ¼ 5 denotes a period of evening relaxation at home.
EI ¼
5 X j¼1
1 t $IRij Finf $Di $C $ 24 ij
(1)
where: EI e the daily exposure level for inhalation in a given age group (ng d1), i ¼ 1, 2, 3 refers to a specific age group: infants (1), children (2), adults (3), C e concentration of an element in air (ng m3),
Each variable in exposure equation was modeled as a specific probability distribution (PDFs) (Table 3). Because the risk model was intended to apply to general population of Silesia the exposure was calculated in age specific manner (infants 0e1, children 2e20, adults 21e76) including human activity during the day (Fig. 2). Following Klepeis et al. (2001) we assumed that Silesia residents spends on average 87% of their time indoors (the sum of the 1-st, 3rd and 5-th time interval). Keeping in mind that indoor pollution strictly differs from outdoor one and that atmospheric air can be considered as a substantial source of pollution for the closed spaces the scenario was classified into two parts: one is exposure to indoor metals of outdoor origin; the second one is exposure to outdoor metals directly. Due to the lack of data concerning the migration of
Table 1 Analytical results for certified reference material NIES Urban Aerosol, (mg kg1). Element
Measured value mean ± SD [mg/kg]
Certified value mean ± SD [mg/kg]
Recovery [%]
RSD [%]
As Cd Ni
80.4 ± 3.8 5.9 ± 0.3 52.1 ± 3.3
90.2 ± 10.7 5.6 ± 0.4 63.8 ± 3.4
89 107 82
4.82 4.94 6.37
K. Widziewicz, K. Loska / Atmospheric Environment 128 (2016) 198e207
B_PM2.5
B_PM10 20
As+Cd+Ni [ng/d]
As+Cd+Ni [ng/d]
20 15 10 5
15 10 5
0
0 0
7
12
18
0
7
infant
12
18
0
7
child
12
18
0
7
adult
12
18
0
7
infant
18
0
7
12
18
adult
Time [h]
Cz_PM10
Cz_PM2.5 20
As+Cd+Ni [ng/d]
20
As+Cd+Ni [ng/d]
12 child
Time [h]
15 10 5
15 10 5
0
0 0
7
12
18
0
7
infant
12
18
0
7
child
12
18
0
7
adult
12
18
0
7
infant
12
18
0
7
child
Time [h]
12
18
adult
Time [h]
K_PM10
K_PM2.5 20
As+Cd+Ni [ng/d]
20
As+Cd+Ni [ng/d]
201
15 10 5
15 10 5 0
0 0
7
12 infant
18
0
7
12
18
child
0
7
12
0
18
7
12
18
0
infant
adult
7
12
18
child
0
7
12
18
adult
Time [h]
Time [h]
Fig. 2. Daily inhaled mass of metals bound to PM2.5 and PM10, calculated using PRA approach for Bielsko-Biała (B), Cze˛ stochowa (Cz) and Katowice (K) residents.
particles into the indoor environment designated specifically for Silesia region, to determine the differences in exposure level between indoor and outdoor environment we used the infiltration factors (Finf) derived from Ji and Zhao (2015). Those authors reviewed the available literature to determine the PM10 and PM2.5 infiltration rates meet typically in urban areas around the world (US, Europ, China). By averaging these results we modeled the Finf distributions independently for PM2.5 and PM10. Some simplification has been however made in presented work by treating those coefficients (determined strictly for PM) as an approximation of PM-bound metals infiltration rates (Table 3). Since metals penetrate into the human respiratory with different efficiency, depending on the size of its PM carriers we also include this additional aspect in our calculations using the particlesize dependent deposition rate designated independently for PM2.5 and PM10. Deposition was calculated in age specific manner using MPPDV 2.11 software. Results were presented as an average regional depositions for head e H; tracheobronchial e TB; and pulmonary e P regions, which have important functional distinctions. It was assumed that the mass percentage of metals that are inhaled and deposited within the lower respiratory tract is equal to the total amount of its PM carriers, which reaches TB and P regions. The calculated deposition rates (Fig. 3) were averaged between the
age groups (infant, child, adult) and described in terms of probability distributions (Table 3).
2.7. Cancer risk estimates The calculation of the health effect involved the multiplication of EI by carcinogenic slope factor (CSF) following equation (2):
R¼
1=3 EI $EF$ED$ CSFi BW 70 BW$AT
cf
(2)
where: R e incremental individual lifetime cancer risk (ILCR) resulting P from a specific metal dose R ¼ Ri for i-th element; EI e daily 1 exposure level (ng day ), CSFi e slope factor for i-th element (mg kg1 day1)1, EF e exposure frequency (day year1) (Chen and Liao, 2006), ED e exposure duration (year) (GUS, 2013), AT e averaging time for carcinogens AT ¼ 76 (year) 365 (day year1) (RAGS, 1989), BW e body weight (kg) (U.S. EPA, 2011) and cf e conversion factor (106) (mg ng1). Inhalation Unit Risk multiplication: IUR (mg m3)1 1000 (mg mg1) (BW(kg)/IR (m3 d1)) was necessary to convert IUR to CSF (mg/kg d)1 (RAGS, 1989).
202
K. Widziewicz, K. Loska / Atmospheric Environment 128 (2016) 198e207
Infant
10 8 6 4 2 0 Cd
Ni
PM2.5
As
Cd
Ni
16
18 16 14 12 10 8 6 4 2 0
Metal dose [ng/d]
Metal dose [ng/d]
Metal dose [ng/d]
12
As
Adult
Child
14
12 10 8 6 4 2 0
As
PM10
14
Cd
Ni
As
PM2.5
Cd
As
Ni
Cd
Ni
As
PM2.5
PM10
Cd
Ni
PM10
Fig. 3. Boxplots showing metal doses to which the Silesia residents are exposed to through inhalation to its ambient concentrations. The bottom and top of the box are 25 and 75 percentiles, the band inside the box is median, while the ends of the whiskers represents 2.5 and 97.5 percentiles of dose distribution.
Instead of 70 e year human lifespan for AT, the average life expectancy in the Silesian district (76 yr) was used (GUS, 2013). All exposure variables and parameters were treated probabilistically according to Table 3. The incomplete toxicological data on the CSF did not allow to treat this variable as a probability distribution, hence CSF was modeled as a point estimate (Table 2).
both PM fractions. PM2.5 and PM10 fractions contained the largest amounts of arsenic followed by nickel and cadmium. Those concentrations however did not exceed the permissible metal standards described in the Directive 2008/50/EC.
3.2. Daily exposure metals concentrations 2.8. Characteristics of risk and uncertainty analysis The risk was calculated as a sum of hazards for particular metals and then converted into population risk e an expected number of additional cancer incidence, respectively for adults: P P20 ILCR ¼ 76 i¼21 Ri=56; children: ILCR ¼ i¼2 Ri=19 and infants: ILCR ¼ P1 i¼0 Ri (GUS, 2013). The final risk was referred to the acceptable risk threshold. The quantitative assessment of the cancer risk was carried out employing a uniform MC simulation by using Crystal Ball software (Decisioneering, 1996). In order to ensure the stability of the numerical output of the model, the number of simulation runs was increased gradually. Therefore, 1.000, 5.000 and 10.000 independent iterations were carried out. The expected risk interval was established at 97.5 and 2.5 quartiles of distribution. It has been found that the number of 2000 iterations was enough to obtain the risk distribution stability. 3. Results 3.1. The concentration of airborne As, Cd and Ni The inhalation risk from airborne metals was calculated taking into account the temporal and spatial variability of its PM carriers. The annual mean concentration of PM2.5 were 36 mg/m3 for BielskoBiała, 32 mg/m3 for Cze˛ stochowa and 37 mg/m3 for Katowice, while for PM10 is was higher and reached 45 mg/m3; 38 mg/m3 and 45 mg/ m3 respectively (Tables 4 and 5) and exceeded the permissible standards at all sampling sites (Regulation of the Minister of Environment (2010)). The most frequent excessive 24-h PM concentrations were found in Katowice, where PM vehicular emissions account for a considerable part of the total air pollution (WIOS, 2012). An increase in PM2.5 and PM10 concentrations significantly correlated with an increase in airborne metal concentrations. Tables 4 and 5 give an average contents of metals associated with
The daily inhalation exposure (ng d1) from annual concentrations of As, Cd and Ni was calculated employing equation (1). Tables 4 and 5 show the analytical results of fitting the probability distributions to the concentration data based on the chi-squared test (p < 0.05). All the distributions were right-skewed, typical of log-normal distribution. Based on this distributions the MC simulations enabled to plot a probability density function for daily metal doses (EI). The results and related uncertainties from those simulations are presented in Fig. 3. A significant differences were observed when comparing daily exposures between PM2.5 and PM10 fractions. Much smaller discrepancies were noted, when examining exposure in terms of age sensitivity. The medians of the metals doses inhaled together with PM2.5 were 5.25; 7.44 and 5.81 (ng d1) for infants, children and adults, respectively, and higher than those associated with PM10 fraction e 2.14; 3.03 and 1.05 (ng d1), respectively. The exposure to PM2.5 associated metals was quite similar in each age group, although slightly lower in case of infants. The infants have generally lower values of IR than those of children and adults (Table 3), which gives an explanation for infants who are at lower risks. This is although not true in case of PM10 e related metals exposure, where the lowest EI values concern adults. Such distribution of the exposure is influenced by lung deposition ratio e significantly lower in case of PM10 particles (Fig. 4). Results obtained from MPPD modeling indicate that deposition efficiency varied within the particle size, human age and respiratory compartment. Fine (PM2.5) particles deposition was much higher compared to PM10. The result shows very slight variation in PM2.5 deposition efficiency among different age-groups: 0.37; 0.45 and 0.35 for infants, children and adults respectively. Bigger differences were observed in case of PM10: 0.35; 0.30 and 0.047 respectively (deposition efficiency in adults was approximately ten times lower compared to other age-groups). Analyzing the PM mass distribution between specific areas of respiratory, it was observed that PM2.5 deposition was the highest
Table 2 The inhalation unit risk and cancer potency factor for Monte Carlo analysis. Element
Inhalation unit risk (mg m3)1
Arsenic Cadmium Nickel
4.3 103 1.8 103 2.4 104
a
a
Cancer slope factor (mg/kg d)1 15.1 6.3 0.84
CSF was adjusted by BW and IR for inhalation unit risk value from Integrated Risk Information System of U. S. EPA.
K. Widziewicz, K. Loska / Atmospheric Environment 128 (2016) 198e207
203
Table 3 Resident equation inputs for 1D Monte Carlo analysis. PDF Symbol C
Unit
Type 3
(ng m
)
LN
Area
Element
B
Receptor
K
PM10 a,b
Finf D
e e
LN LN
EF ED
(day year1) year
LN UN
AT IR
d (m3 d1)
PE PE
infant
LN
child
LN
adult
LN
infant child adult
kg
PM2.5
As Cd Ni As Cd Ni As Cd Ni
Cz
BW
Activity
infant child adult infant child adult Sleep Light Sedentary Sleep Light Sedentary Sleep Light Sedentary
(1.7; 3.1) (0.9; 2.8)a,b (1.9; 2.0)a,b (2.4; 2.9)a,b (1.1; 2.5)a,b (2.0; 2.1)a,b (2.4; 1.9)a,b (1.2; 2.3)a,b (2.4; 1.6)a,b (0.58; 0.11)a,b (0.4; 0.07)a,b (0.4; 0.06)a,b (0.3; 0.02)a,b (252; 1.01)a,b (0e1)c,d (0e21)c,d (0e76)c,d 76e*365 6.6 15.8 6.7 (9.4; 0.8)a,b (22.5; 1.4)a,b (9.5; 0.8)a,b (10.1; 0.5)a,b (22.8; 0.9)a,b (10.1; 0.5)a,b (6.8; 1.9)a,b (34.0; 24.8)a,b (79.1; 5.4)a,b
(1.1; 1.4)a,b (0.4; 1.7)a,b (1.1; 1.4)a,b (1.6; 1.8)a,b (0.5; 2.1)a,b (1.4; 1.9)a,b (1.5; 1.8)a,b (0.7; 2.3)a,b (1.3; 1.6)a,b (0.39; 0.11)a,b (0.3; 0.11)a,b (0.3; 0.06)a,b (0.1; 0.01)a,b
LN e log-normal distribution, UN e uniform distribution, PE e point estimate. a Geometric mean. b Geometric SD. c Minimum. d Maximum. e The average lifespan averaged for males and females (GUS, 2013), B e Bielsko-Biała, Cz e Cze˛ stochowa, K e Katowice.
3.3. Inhalation incremental cancer risk
in the P region e 0.325, and the smallest in TB region e 0.089, while in the case of coarse particles the overwhelming particles mass was deposited in the head (H) region e with the 0.66 efficiency, and lowest in the pulmonary e 0.02. The total deposition efficiency for TB and P region, averaged over all analyzed age groups was approximately 40% greater in case of PM2.5 than for PM10. Pulmonary deposition was higher for children compared to adults, because of the higher physical activity of children and thus increased lung ventilation. It should be therefore noted that children are potentially more susceptible to particulate exposures.
The ILCR for the residents of particular cities was calculated following equation (2). Risk analysis was performed independently for metals associated with PM2.5 and PM10 fractions, taking into account its origin (exposure to ambient concentrations/exposure to indoor concentrations). The shape of risk distribution was lognormal in all the groups tested, however due to the complicated risk scenario, this article does not demonstrate a graphic diagram of this predicted PDFs. Instead the results are presented in the form of LN (geometric mean ± geometric standard deviation). Considering
Table 4 Statistical summary of airborne PM2.5 (mg m3) and the associated metal concentrations (ng m3). B e Bielsko-Biała, Cz e Cze˛ stochowa, K e Katowice. Sampling site
B
Cz
K
gm
geometric mean,
Parameter
PM2.5 As Cd Ni PM2.5 As Cd Ni PM2.5 As Cd Ni gsd
Statistics
Goodness-of-fit test
gm
gsd
Median
Min
Max
N
Skewness
Kurtozis
Distribution
36.2 1.7 0.9 1.9 32.0 2.4 1.1 2.0 37.2 2.4 1.2 2.4
38.1 3.1 2.8 2.0 32.4 0.9 2.5 2.1 28.7 1.9 2.3 1.6
21.4 0.82 0.28 1.30 23.0 1.53 0.44 1.38 28.2 1.88 0.56 2.00
1.4 0.1 0.1 0.1 2.0 0.1 0.1 0.3 1.4 0.3 0.02 0.7
247.2 22.8 16.9 96.1 309.8 36.6 10.6 88.1 224.4 7.4 12.0 12.2
352
2.1 2.7 3.9 13.4 3.5 2.7 2.1 8.8 2.5 0.7 3.2 2.4
5.4 8.9 21.4 209.0 19.7 11.2 6.5 92.7 10.2 0.23 13.1 7.9
e LN
e LN
e LN
p
c2 a ¼ 0.995
6 4 1
<0.0001 <0.0001 <0.0001
1.63 0.71 0.0039
5 8 1
<0.0001 <0.0001 <0.0001
1.14 2.73 0.0039
10 5 8
<0.0001 <0.0001 <0.0001
3.94 1.14 2.73
df
geometric SD, CV-coefficient of variation, Min-minimum, Max-maximum, df-degrees of freedom, p-probability value, c2-chi square test value.
204
K. Widziewicz, K. Loska / Atmospheric Environment 128 (2016) 198e207
Table 5 Statistical summary of airborne PM10 (mg m3) and the associated metal concentrations (ng m3). B e Bielsko-Biała, Cz e Cze˛ stochowa, K e Katowice. Sampling site
B
Cz
K
gm
geometric mean,
Parameter
Statistics
PM10 As Cd Ni PM10 As Cd Ni PM10 As Cd Ni gsd
Goodness-of-fit test
gm
gsd
Median
Min
Max
N
Skewness
Kurtozis
Distribution
42.8 1.1 0.4 1.1 38.4 1.6 0.5 1.4 45.3 1.5 0.7 1.3
39.3 1.4 1.7 1.4 28.4 1.8 2.1 1.9 33.2 1.8 2.3 1.6
28.0 0.68 0.09 0.68 31.0 1.06 0.12 0.83 37.0 0.96 0.20 0.82
5.0 0.9 0.2 1.0 5.0 0.9 0.2 1.0 5.0 0.9 0.2 1.0
256.0 2.6 1.5 4.7 285.0 8.9 2.9 11.0 286.0 6.3 2.6 5.8
326 26
2.2 2.2 1.9 4.1 3.1 3.1 1.9 3.1 2.7 2.0 0.8 2.7
6.2 2.8 4.4 16.0 17.9 11.2 3.8 9.1 12.3 4.5 0.4 8.0
e LN
367 26
363 26
e LN
e LN
df
p
c2 a ¼ 0.995
1 2 1
<0.0001 <0.0001 <0.0001
0.0039 0.10 0.0039
2 4 2
<0.0001 <0.0001 <0.0001
0.10 0.71 0.10
3 8 2
<0.0001 <0.0001 <0.0001
0.35 2.73 0.10
geometric SD, CV-coefficient of variation, Min-minimum, Max-maximum, df-degrees of freedom, p-probability value, c2-chi square test value.
Deposition fraction
0.5
0.4
0.3
0.2
0.1
0 3 mth
21 mth
23 mth
28 mth
3 yr
8 yr
9 yr
14 yr
18 yr
21 yr
human age TB_PM2.5
P_PM2.5
TB_PM10
P_PM10
Fig. 4. Averaged regional deposition of PM2.5 and PM10 particles.
outdoor/indoor metals sources in shaping the distribution of health risks among Silesia residents it was shown that outdoor exposures are potentially more harmful. The total ILCR averaged for the populations of Bielsko-Biała, Cze˛ stochowa and Katowice due to outdoor-originated PM2.5-bound metals in the indoor environment was LN (9.5 107 ± 1.5 106), LN (1.29 106 ± 2.06 106), LN (1.33 106 ± 2.09 106), while for PM10-bound metals it was LN (2.41 107 ± 4.9 107), LN (3.44 107 ± 5.76 107), LN (3.45 107 ± 5.80 107). The dispersion of risks arranged a bit differently in case of exposure to outdoor originated metals: for ambient PM2.5-bound metals LN (1.40 106 ± 2.16 106), LN (1.91 106 ± 2.94 106), LN (1.96 106 ± 3.02 106), while in case of PM10-bound metals LN (4.58 107 ± 7.54 107), LN (6.47 107 ± 1.06 106), LN (6.56 107 ± 1.07 106). The obtained findings e even two times higher risk in case of direct exposure to ambient metals concentrations suggest that indoor (closed environment) play a significant role in protecting occupants against exposure to air pollutants (this observation couldn't be generalized to indoor environments affected by internal emissions sources). As the size distribution shifts toward smaller particles, the infiltration factor increases. This easily explains the differences in lifelong exposures between PM2.5 and PM10-bound metals. Taking into account the age-dependent sensitivity it was shown that inhalation risks between analyzed age groups arranged accordingly to the daily metal doses (Table 6). The inhalation ILCR values for specific age groups was highest in case of children exposed to PM2.5-bound metals of outdoor origin LN
(9.25 107 ± 1.44 106), while the smallest risks concerned infants exposed to indoor concentrations of PM10-bound metals (1.66 108 ± 2.78 108). All 97.5% probabilities of Total ILCRs were below 106, which indicate acceptable probability distributions for each of the concerned age groups.
3.4. Sensitivity analysis The findings from sensitivity analysis are demonstrated in the diagrams depicting the percentage contribution of particular exposure variables to the total variance of predicted risk. The variables which exerted the strongest effect on the health risk for all the age groups included exposure duration (ED) and body weight (BW). Their approximate percentage contribution ranged from 28.1% to 44.6% for ED and from 23.1% to 32.8% for BW depending on the scenario (Fig. 5). The variation in the concentration-response estimates was to a very small extent owned to differences in metals migration into indoor environment. The uncertainties brought by metals infiltration into residences (Finf parameter), were close to 0.2% and therefore negligible with respect to variability affecting personalhealth responses. Together with other parameters, like deposition efficiency and inhalation rate, Finf accounts for only 10% of total risk variance.
K. Widziewicz, K. Loska / Atmospheric Environment 128 (2016) 198e207
205
Table 6 Predicted PDF of total incremental lifetime cancer risk for infants, children and adults. PM fraction
Age group
Total ILCR from outdoor exposure LN (gm, gsd)
PM2.5
Infant Child Adult Infant Child Adult
(6.76 (8.63 (9.25 (3.31 (4.02 (1.92
PM10
gm
geometric mean,
gsd
108 107 107 108 107 107
± ± ± ± ± ±
1.12 1.69 1.44 5.19 8.31 2.99
Total ILCR from indoor exposure LN (gm, gsd)
107) 106) 106) 108) 107) 107)
(4.48 (5.76 (5.75 (1.66 (2.04 (9.06
108 107 107 108 107 108
± ± ± ± ± ±
7.51 1.44 9.15 2.78 4.02 1.44
108) 106) 107) 108) 107) 107)
geometric SD.
PM2.5 others
PM10
10.3
BW_adult
others
1.2
B_PM2.5_As_child
9.3
K_PM10_As_child
1
2.2 ED_adult
BW_child
12.3
23.1
ED_child
BW_child
28.1
ED_adult
35.1
0
10
20
30
40
Contribution to the variance view [%]
32.8
ED_child
44.6
0
10
20
30
40
50
Contribution to the variance view [%]
Fig. 5. Results of sensitivity analysis for 1-D MCA (indoor exposure scenario).
4. Discussion The MC technique employed herein to assess health risk enabled not only a more accurate calculation of its value, but also helped acquire relevant information on its distribution in particular age groups. The probabilistic analysis revealed that cancer risk in the tested population did not exceed the permissible level of 106e104. A higher risk of LN (1.80 106 ± 2.89 106) was found for metal exposure associated with PM2.5, while the value for PM10 reached LN (6.17 107 ± 1.08 106). Similar results were obtained by Sadovska (2012), who calculated an inhalation risks for Silesia neighbors e the population of Ostrava Karvina Coal Basin (an industrial region in the Czech Republic) exposed to airborne As, Cd and Ni. It must be emphasized that the statement about greater health hazard caused by PM2.5-bound metals is biased by the fact that PM fractions were collected at different sampling sites. A direct comparison of inhalation risks due to PM2.5 and PM10 exposure was therefore not possible. Literature data however indicate a more deleterious effect of PM2.5 fraction compared to PM10, which is probably caused by the larger active surface of grains, favoring metal concentration (Onat et al., 2012). This phenomenon is particularly alarming when considering the fact that fine particles account for even 90% of airborne particles (Salonen and Pennanen, 2007). The most popular approach in estimating human exposure to air pollution is treating the ambient concentrations as surrogates for real exposure estimation (Zhang et al., 2009; Zhou and Zhao, 2012), however this approach can result in a significant exposure misclassification and risk overestimation (Payne-Sturges et al., 2004). Since human population most time spend indoors (Zhou and Zhao, 2012) a special attention must be paid to the reliable estimation of I/O pollution ratio. In this work the exposure to As, Cd and Ni concentrations was therefore calculated in terms of its infiltration efficiency into indoor environment. Due to lack of data concerning metals concentrations in indoor/outdoor relationship, rny et al., 1995), we made an determined specifically for Silesia (Go assumption that indoor metals concentrations are mostly governed by the outdoor emissions (Table 3), which situation is typical for regions attributed to the excessive industrial activities (Nazir et al.,
2011). Obtained results showed that risks posed by ambient metals concentrations were almost two times greater compared to the risk revealed in case of chronic exposure to indoor metal pollution and additionally suggest an important role of buildings in protecting people from air pollution. Further research must be although provide in order to define to which extent risks are governed not only by metals infiltration into indoor environment but also by the existence of indoor emission sources (metals pollution generated by smoking; burning fossil fuels in domestic stoves; cooking etc.). The second factor controlling metals exposure is its deposition efficiency in the respiratory track. Since metals are adsorbed on PM particles, its deposition rate depends on PM concentration in ambient air, size of those particles and human ventilation rate. The dose of inhaled metals should be therefore calculated as the product of metals concentration, time which individuals spent in a specific environment and pulmonary ventilation rate (including metals deposition in particular area of the respiratory track). Presented data indicate that aerosol particles with a diameter less than 10 mm are mostly deposited in the thoracic region, while PM2.5 reach the pulmonary region. Health effects depends not only on the amount of the inhaled particles but also on the adsorbed substances they carry into specific regions of the lungs. Hence, the determination of metals deposition in the lung, their uptake, redistribution, storage, and removal is very important considering the range of harmful effects. In this work we observed a strong impact of particles size on its deposition. Coarse particles (PM10) deposition in the lower respiratory track was significantly lower compared to fine mode particles (PM2.5). The presented approach enabled to gain insights on the effects of particles size and human age on the PM deposition efficiency. It was shown that the total PM2.5 deposition efficiency was highest in case of children and for PM10 in case of infants. Only particles <2.5 mm seem to be influenced by the enhanced pulmonary deposition rate because particles larger 10 mm efficiently deposit in the upper airways. An increase of risk was correlated was not only with the total metals contents and its deposition in the respiratory track but also with the age. Infants appeared to be the most susceptible to the metals present in the air. The cancer risk for that age group was higher than the one for adults by 77% and 81.8% for PM2.5 and PM10,
206
K. Widziewicz, K. Loska / Atmospheric Environment 128 (2016) 198e207
respectively. The stratification of risk in particular age groups was also observed by Trojanowska and Swietlik (2012) who investigated the inhalation exposure to As, Cd and Ni in the largest Polish conurbations. Their findings indicate that children are the most exposed age group in the population and inhale approx. 50% more air per kg of body mass compared to adults. Thus, children and infants should avoid spending a lot of time outdoors, particularly when doing intense physical activity, which increases lung ventilation. In mathematical terms, the parameter, which mostly differentiated health risk between the age groups was the exposure duration (Table 3). So far, a sole criterion for selecting a suitable distribution of exposure duration has not been established. A lot of authors employing MC technique treat this variable as a point estimate (de Oliveira et al., 2012; Chen and Liao, 2006), roughly equal to the average lifespan of an inhabitant. This work employed the exposure duration determined by GUS (2013) data, however, it should be noted that selection of another distribution would lead to significantly different results. The considerable effect of ED on the final risk is confirmed by the results of sensitivity analysis (Fig. 5). The contribution of this variable to the total variance of model output was much bigger than the contribution of the “metal concentration” variable. Therefore to increase the accuracy of the results, efforts should focus on a better definition of ED distribution. Although physiological parameters as well as exposure time significantly affects carcinogenic hazards a key issue when estimating cancer risk caused by certain air pollutants is however the slope factor. Since a big differences exists in the order of cancer potency between single pollutants, the risk value is mostly governed by these one, which has the highest CSF value. Literature data indicates that CSF is the most highly uncertain parameters in risk analysis models, which occupied even more than 70% of the total variance of ILCR (Chen and Liao, 2006). It is because of the incomplete or even ambiguous epidemiological evidence, and uncertainties connected with extrapolation of animal data to human data (Biesiada and Bubak, 2001). To overcome these uncertainties it is suggested to treat CSF as a variable or fuzzy parameter (Mofarrah and Husain, 2010). In order to greatly improve the accuracy of the risk assessment, improving the CSF values might be much more effective than improving the accuracy of the exposure concentrations (Chen and Liao, 2006). Since CSF values are usually similar to the upper limits of estimates, the authors suspect that the actual health risk resulting from the inhalation exposure to As, Cd and Ni in Silesia may be lower than the one calculated. Further research into the determination of inhalation risk in Silesia should be aimed at determining probability distribution of particular exposure parameters, so that they could best reflect the heterogeneity of its population. 5. Conclusions It has been found that As, Cd and Ni concentrations measured in Silesia region do not exceed the permissible concentrations stipulated in the EU Directive, and the additional cancer risk faced by the population was lower than 106, indicating no significant cancer risks. This, in terms of public healthcare, does not necessitate any countermeasures to reduce the emissions of the metals into the atmospheric air. The risk analysis conducted by PRA also proved that PM2.5-bound metals poses a higher health risk, and both adults and children are the most susceptible to this pollutants. The total ILCR averaged for the populations of Bielsko-Biała, Cze˛ stochowa and Katowice calculated using PRA, lower by even order of magnitude than ILCR values calculated deterministically, confirmed that the variability of exposure parameters in the population exerts a significant effect on health risk. Knowledge of its strength is
particularly important when it affects the legitimacy of taking suitable countermeasures to minimize the risk. Funding sources The research was financed by “DoktoRIS” e Program for Innovative Silesia 2007e2013, co-financed by the European Union under the European Social Fund. http://www.doktoris.us.edu.pl/ projekt/projekt.php. Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.atmosenv.2015.12.061. References Biesiada, M., 2001. Simulations in health risk assessment. Int. J. Occup. Med. Environ. Health 14, 397e402. Biesiada, M., Bubak, A., 2001. Basics of Environmental Health Risk Assessment. Training Materials ‘Theory and Practice of Environmental Impact Assessments on Health’. Institute of Occupational Medicine and Environmental Health in Sosnowiec, pp. 29e38. BS EN 14907:2006. Ambient air quality e Standard gravimetric method for determining the mass fraction of PM2. 5 particulate matter. Chen, S.C., Liao, C.M., 2006. Health risk assessment on human exposed to environmental polycyclic aromatic hydrocarbons pollution sources. Sci. Total Environ. 366, 112e123. de Oliveira, B.F.A., Ignotti, E., Artaxo, P., Saldiva, P.H.D., Junger, W.L., Hacon, S., 2012. Risk assessment of PM2.5 to child residents in Brazilian Amazon region with biofuel production. Environ. Health 11, 301e309. Decisioneering, 1996. Crystal BallTM Fusion Edition 11.1.2.3.000 Inc. (Denver, CO). Directive 2008/50/EC of the European Parliament and of the Council of 21 May 2008 on ambient air quality and cleaner air for Europe. Gomis cek, B., Hauck, H., Stopper, S., Preining, O., 2004. Spatial and temporal variations of PM1, PM2.5, PM10 and particle number concentration during the AUPHEPdproject. Atmos. Environ. 38, 3917e3934. rny, R.L., Je˛ drzejczak, A., Pastuszka, J.S., 1995. Dust particles and metals in outGo door and indoor air of Upper Silesia. Ann. Natl. Inst. Hyg. 46, 151e161. GUS, 2013. Life expectancy tables of Poland 2012. Warszawa 2013. Harrison, R.M., Smith, D.J.T., Kibble, A.J., 2004. What is responsible for the carcinogenicity of PM2.5? Occup. Environ. Med. 61, 799e805. IARC, 2014. Monographs e Classifications e by Group (Last Update 31 March 2014). http://monographs.iarc.fr/ENG/Classification/ClassificationsGroupOrder.pdf. Ji, W., Zhao, B., 2015. Estimating mortality derived from indoor exposure to particles of outdoor origin. Plos One 10, 1e15. Klepeis, N.E., Nelson, W.C., Ott, W.R., Robinson, J.P., Tsang, A.M., Switzer, P., Behar, J.V., Hern, S.C., Engelmann, W.H., 2001. The National Human Activity Pattern Survey (NHAPS): a resource for assessing exposure to environmental pollutants. J. Expo. Analysis Environ. Epidemiol. 11, 231e252. Mofarrah, A., Husain, T., 2010. Modeling for Uncertainty Assessment in Human Health Risk Quantification: a Fuzzy Based Approach. International Environmental Modelling and Software Society (iEMSs). International Congress on Environmental Modelling and Software. Nazir, R., Shaheen, N., Shah, M.H., 2011. Indoor/outdoor relationship of trace metals in the atmospheric particulate matter of an industrial area. Atmos. Res. 101, 765e772. Onat, B., Sahin, U.A., Bayat, C., 2012. Assessment of particulate matter in the urban atmosphere: size distribution, metal composition and source characterization using principal component analysis. J. Environ. Monit. 14, 1400e1409. Payne-Sturges, D.C., Burke, T.A., Breysse, P., Diener-West, M., Buckley, T.J., 2004. Personal exposure meets risk assessment: a comparison of measured and modeled exposures and risks in an urban community. Environ. Health Perspect. 112, 589e598. Pope, C.A., Burnett, R.T., Turner, M.C., Cohen, A., Krewski, D., Jerrett, M., Gapstur, S.M., Thum, M.J., 2011. Lung cancer and cardiovascular disease mortality associated with ambient air pollution and cigarette smoke: shape of the exposure-response relationships. Environ. Health Perspect. 119, 1616e1621. RAGS, 1989. Risk Assessment Guidance for Superfund Volume I Human Health Evaluation Manual (Part A), EPA/540/1e89/002, Washington, D.C. Office of Emergency and Remedial Response U.S. Environmental Protection Agency. Regulation of the Minister of Environment of 26 January 2010 on the reference values for certain substances in the air (Journal of Laws of 2010, No. 16, item. 87). 't Mannetje, A., Bencko, V., Brennan, P., Zaridze, D., Szeszenia-Dabrowska, N., , E., Cassidy, A., Mates, D., Foretova, L., Rudnai, P., Lissowska, J., Fabi anova Janout, V., Fevotte, J., Fletcher, T., Boffetta, P., 2011. Occupational exposure to metal compounds and lung cancer. Results from a multi-center case-control study in Central/Eastern Europe and UK. Cancer Causes Control 22, 1669e1680. Sadovska, V., 2012. Health risk assessment of heavy metals adsorbed in particulates.
K. Widziewicz, K. Loska / Atmospheric Environment 128 (2016) 198e207 World Acad. Sci. Eng. Technol. 6, 170e173. Salonen, R.O., Pennanen, A., 2007. the impact of fine particles on health. In: Herring, Peter (Ed.), Views and Conclusions from the FINE Particles eTechnology. Environment and Health Technology Programme. Tekes. ˛ skie. Regional Development Strategy of the Voivodeship of Silesia 2020þ. Sla https://rpo.slaskie.pl/zalaczniki/2014/01/17/1389967464.pdf Trojanowska, M., Swietlik, R., 2012. Heavy metals cadmium, nickel and arsenic environmental inhalation hazard of residents of Polish cities. Environ. Med. 15, 33e41. U.S. EPA, 2011. (U.S. Environmental Protection Agency). Exposure Factors Handbook: 2011 Edition. Report EPA/600/R-090/052F, Washington, DC. WHO, 2013. Health Effects of Particulate Matter. WHO Report. http://www.euro. who.int/__data/assets/pdf_file/0006/189051/Health-effects-of-particulatematter-final-Eng.pdf.
207
2012. State of the Environment in the Silesia District in 2012. Library Of WIOS, Environmental Monitoring 2013, pp. 1e140. http://www.katowice.pios.gov.pl/ monitoring/raporty/2012/raport2012.pdf (20.08.2015). Yaman, M., 2012. In: Khare, Mukesh (Ed.), Comprehensive Comparison of Trace Metal Concentrations in Inhaled Air Samples (pp. 29e67) in Air Pollution e Monitoring, Modelling, Health and Control. InTech, p. 254. ISBN 978-953-510381-3. Zhang, Y., Tao, S., Shen, H., Ma, J., 2009. Inhalation exposure to ambient polycyclic aromatic hydrocarbons and lung cancer risk of Chinese population. Proc. Natl. Acad. Sci. U. S. A. 106, 21063e21067. Zhou, B., Zhao, B., 2012. Population inhalation exposure to polycyclic aromatic hydrocarbons and associated lung cancer risk in Beijing region: contributions of indoor and outdoor sources and exposures. Atmos. Environ. 62, 472e480.
Contents lists available at ScienceDirect
Atmospheric Environment journal homepage: www.elsevier.com/locate/atmosenv
Metal induced inhalation exposure in urban population: A probabilistic approach Kamila Widziewicz a, *, Krzysztof Loska b a b
Institute of Environmental Engineering, Polish Academy of Sciences, M. Skłodowska-Curie 34, 41-819 Zabrze, Poland Faculty of Energy and Environmental Engineering, The Silesian University of Technology, Konarskiego 18 A, 44-100 Gliwice, Poland
h i g h l i g h t s We model health risks caused by inhalation exposure to airborne metals. We examine changes in the level of risk between PM2.5 and PM10-bound metals. PM2.5-bound metals poses a higher risk than PM10 e associated one. Infants and children are especially sensitive to airborne metals. Risk value depend on PM composition, as well as the variability of exposure.
a r t i c l e i n f o
a b s t r a c t
Article history: Received 1 September 2015 Received in revised form 22 December 2015 Accepted 24 December 2015 Available online 29 December 2015
The paper was aimed at assessing the health risk in the populations of three Silesian cities: Bielsko-Biała, Cze˛ stochowa and Katowice exposed to the inhalation intake of cadmium, nickel and arsenic present in airborne particulate matter. In order to establish how the exposure parameters affects risk a probabilistic risk assessment framework was used. The risk model was based on the results of the annual measurements of As, Cd and Ni concentrations in PM2.5 and the sets of data on the concentrations of those elements in PM10 collected by the Voivodship Inspectorate of Environmental Protection over 2012e2013 period. The risk was calculated as an incremental lifetime risk of cancer (ILCR) in particular age groups (infants, children, adults) following Monte Carlo approach. With the aim of depicting the effect the variability of exposure parameters exerts on the risk, the initial parameters of the risk model: metals concentrations, its infiltration into indoor environment, exposure duration, exposure frequency, lung deposition efficiency, daily lung ventilation and body weight were modeled as random variables. The distribution of inhalation cancer risk due to exposure to ambient metals concentrations was LN (1.80 106 ± 2.89 106) and LN (6.17 107 ± 1.08 106) for PM2.5 and PM10-bound metals respectively and did not exceed the permissible limit of the acceptable risk. The highest probability of contracting cancer was observed for Katowice residents exposed to PM2.5 e LN (2.01 106 ± 3.24 106). Across the tested age groups adults were approximately one order of magnitude at higher risk compared to infants. Sensitivity analysis showed that exposure duration (ED) and body weight (BW) were the two variables, which contributed the most to the ILCR. © 2015 Elsevier Ltd. All rights reserved.
Keywords: Cancer risk Monte Carlo Particulate matter Metals Silesia region Probabilistic assessment
1. Introduction In recent years, there has been an increasing concern about the exceedance of the permissible concentrations of particulate matter in atmospheric air. The results of epidemiological research reveal
* Corresponding author. E-mail address: [email protected] (K. Widziewicz). http://dx.doi.org/10.1016/j.atmosenv.2015.12.061 1352-2310/© 2015 Elsevier Ltd. All rights reserved.
that exposure to PM causes an increase in mortality rate and incidence of vascular and respiratory diseases, including cancer (Pope et al., 2011). Despite ample scientific evidence confirming the correlation between persistent exposure to PM and the increase in malignant cancer incidence, including lung cancer, it is still unknown, which compounds of the particles are responsible for inducing carcinogenesis (Harrison et al., 2004). Studies into environmental hazard indicate however that PM induced cancers can be attributed to its metallic constituents ('t Mannetje et al., 2011;
K. Widziewicz, K. Loska / Atmospheric Environment 128 (2016) 198e207
Yaman, 2012). In Poland, three metals: lead, cadmium, nickel and one semimetal, arsenic, are monitored in the air. Three of the above elements: As, Ni and Cd have been classified as factors with sufficient evidence of carcinogenicity in humans (IARC, 2014). Most of the available information about the health effects of those elements is derived from studies, which monitored As, Cd and Ni contents in PM10 (Sadovska, 2012; Trojanowska and Swietlik, 2012). Scientific investigations do not usually deal with the health effects of exposure to the above metals based on their concentrations in PM2.5. This results from the fact that before 2008, PM2.5 was not required to be monitored. However, since PM2.5 constitutes approximately 70% of PM10 mass (Gomis cek et al., 2004) and is deposited in the deepest areas of the respiratory track, it probably enables more accurate prediction of risk (WHO, 2013). The assessment of the health effects of exposure to PM and metals in urban areas is relevant because of their numerous emission sources and the size of the population exposed. Since metals are widely present in urban air and their carcinogenic properties are widely known, the assessment of their harmfulness to public health is necessary. The methodological difficulty in assessing the inhalation hazard caused by airborne carcinogenic metals lies in the complexity of their effects on a population, such as variability of metal concentrations in air, indoor/outdoor pollution ratio, exposure duration, physiological parameters (lung ventilation rate influencing deposition efficiency in the respiratory) and variability of individual characteristics in a population exposed (age and gender) (Biesiada, 2001). Those factors affect the amount of metals taken in, thus significantly determining the final risk. Monte Carlo technique involving the so called Probability Risk Assessment allows the variability of those risk parameters to be taken into account, while assessing risk and determining their effect on the final health risk value. Such approach gives more complex risk characteristics and constitute a source of additional information for risk managers. 1.1. Objectives The primary objective of this study was: to assess the inhalation risks to the residents of three Silesian cities e exposed to PM2.5 and PM10 associated metals by using Monte Carlo approach (1), to determine the influence of exposure parameters variability on health risk value (2), to identify those exposure parameters, which are the key contributors to over all risk value (3). 2. Materials and methods 2.1. Characteristics of the study area The Upper Silesian Region (Fig. 1) is one of the most polluted areas in Poland and is home to numerous extractive, electrical, power engineering and metallurgical industries. Its nature based on raw materials and power production, as well as developed transport infrastructure, makes it occupy the first place in terms of excessive PM concentrations. The emission of PM dust in this area ˛ skie, accounts for 22.14% of the over all national emission (Sla 2020þ).
199
sites are shown in Fig. 1. More detailed information concerning sampling sites along with samplers specification can be found in the Supplementary Materials. It must be highlighted that PM2.5 and PM10 data presented in this work in fact come from different sampling sites, where those fractions were collected separately. Throughout the work inhalation hazards were therefore evaluated independently for PM2.5 and PM10. 2.2.1. PM10 Datasets containing PM10 and associated metal concentrations were downloaded from the Silesian Air Monitoring database (www.katowice.pios.gov.pl/slmonpow), operated by the Chief Inspectorate for Environmental Protection (CIEP). The CIEPs laboratories carry out a routine assessment of metal contents in PM10. Metals assays are regularly done in two-week cumulative samples (14 PM10 filters in total). Thus, 26 measurements of metal concentrations in atmospheric air are conducted each year (Table 5). 2.2.2. PM2.5 Fine particles were collected using low volume samplers (2.3 m3/h) equipped with a PM2.5 measuring head onto Ø47 mm Whatman quartz fiber filters. Diurnal samples were sampled at Bielsko-Biała and Cze˛ stochowa urban background sites and additionally at Katowice traffic site. The mass of the sampled particles was determined gravimetrically (Radwag microbalance, resolution 1 mg) according to the BS EN 14907:2006 standard. Before each weighing, the filters were conditioned for at least 48 h at the air temperature of 20 ± 1 C and air relative humidity of 50 ± 5%. 2.3. Metals extraction and samples preparation Single quarter of each PM2.5 filter was digested in a Milestone microwave digestion system (MLS 1200 Mega oven) using 2 cm3 of 65% HNO3 (Baker). The digestion power was held at 1000 W over all digestion stages (time, 6 min). After cooling, the samples were transferred into flasks and made up to 25 cm3 with ultrapure water. Samples were transferred into plastic bottles and kept in ventilated room until the instrumental analysis. Determination of the total arsenic was carried out after As(V) reduction into As(III). For that purpose three remaining quarters of each filter were cut down to pieces smaller than 1 cm in length and placed in a tube. In next step exactly 5 cm3 of HCl (36% w/v), 2.5 cm3 of urea (25% w/v) and l cm3 of KI (25% w/v) was added to this tube. After 30 min (prereduction time), solution was filled to 25 cm3 with ultrapure water, filtrated through 0.45 mm cellulose filter and aspirated to HG-AAS. The complete reduction of the arsenous hydride was achieved in VGA reaction loop by the addition of 10 M HCl and sodium tetrahydroborate (III) in 0.05% (w/v) NaOH. 2.4. Instrumental analysis The instrumental analysis was conducted by means of GT-AAS method (in case of Ni and Cd) and by using HG-AAS (for As determination). All metal concentrations were calculated on PM2.5 mass collected and normalized by the volume of air passed through the filters during the sampling period. Calibration solutions were prepared daily by using Merck single-element stock solutions of 1000 ppm.
2.2. Sampling protocol 2.5. QA/QC In the presented study authors used the data on PM10 and associated metals concentrations collected by the Voivodship Inspectorate of Environmental Protection in Katowice (VIEP Katowice) and their own measurements of the above elements in PM2.5 for the same period (May 2012eApril 2013). Locations of the sampling
Blank filters were mineralized following the same digestion procedure as tested filters. Blank samples have been found to have non-detectable levels of elements of interest. Certified reference material NIES Urban Aerosol was used to verify the accuracy and
200
K. Widziewicz, K. Loska / Atmospheric Environment 128 (2016) 198e207
Fig. 1. Study area and sampling location. The map on the right shows the average concentration of PM2.5 in Europe e concentration above 30 mg/m3 are colored in violet (based on EEA data, reference year 2010). The contour map on the left presents the borders of Silesia administration district with marked sampling site positions. 1a,1b: Bielsko-Biała, 2a,2b: Cze˛ stochowa, 3a,3b: Katowice (Directive, 2008/50/EC), where symbols a and b denotes PM2.5 and PM10 collecting points. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article). This is an original source file from http://www.eea.europa.eu/data-and-maps/figures/pm2-5-annual-average-2010 website
Finf e PM infiltration rate into indoor environment. For j ¼ 1, j ¼ 3 and j ¼ 5 time interval the following distributions were used: in case of PM2.5-bound metals Finf ¼ LN (0.58; 0.11) and for PM10-bound metals ¼ LN (0.39; 0.11) for j ¼ 2,4 the Finf ¼ 1, Di e particles deposition efficiency in the respiratory track (calculated based on MPPD V2.11 modeling results) for specific age group, t e residence time in a j time interval (hr), IRij e inhalation rate in a given age group at a specific time interval (m3 day1), where j denotes the level of daily activity: j ¼ 1 (sleep or nap activity), j ¼ 2, 3, 4 (light activity), j ¼ 5 (sedentary/passive activity) following U.S. EPA 2011.
precision of the analytical method (Table 1). 2.6. Daily exposure level for inhalation Since there is no information regarding metals absorption in the human respiratory, it was assumed that the active dose of inhaled metals is equal to its airborne concentration and calculated following equation (1). To determine the daily changes in metal exposure levels, a day was divided into the 5 time intervals j ¼ 1, 2, 3, 4 and 5, which corresponded to the following hour intervals: 00:00e06:00, 07:00e08:00, 08:00e16:00, 16:00e18:00 and 18:00e24:00, where j ¼ 1 denotes the time of a nap/sleep spent at home; j ¼ 2 time a man spent on a way to school/work; j ¼ 3 an 8-h work or school time; j ¼ 4 time a man spent on a way from school/ work and finally j ¼ 5 denotes a period of evening relaxation at home.
EI ¼
5 X j¼1
1 t $IRij Finf $Di $C $ 24 ij
(1)
where: EI e the daily exposure level for inhalation in a given age group (ng d1), i ¼ 1, 2, 3 refers to a specific age group: infants (1), children (2), adults (3), C e concentration of an element in air (ng m3),
Each variable in exposure equation was modeled as a specific probability distribution (PDFs) (Table 3). Because the risk model was intended to apply to general population of Silesia the exposure was calculated in age specific manner (infants 0e1, children 2e20, adults 21e76) including human activity during the day (Fig. 2). Following Klepeis et al. (2001) we assumed that Silesia residents spends on average 87% of their time indoors (the sum of the 1-st, 3rd and 5-th time interval). Keeping in mind that indoor pollution strictly differs from outdoor one and that atmospheric air can be considered as a substantial source of pollution for the closed spaces the scenario was classified into two parts: one is exposure to indoor metals of outdoor origin; the second one is exposure to outdoor metals directly. Due to the lack of data concerning the migration of
Table 1 Analytical results for certified reference material NIES Urban Aerosol, (mg kg1). Element
Measured value mean ± SD [mg/kg]
Certified value mean ± SD [mg/kg]
Recovery [%]
RSD [%]
As Cd Ni
80.4 ± 3.8 5.9 ± 0.3 52.1 ± 3.3
90.2 ± 10.7 5.6 ± 0.4 63.8 ± 3.4
89 107 82
4.82 4.94 6.37
K. Widziewicz, K. Loska / Atmospheric Environment 128 (2016) 198e207
B_PM2.5
B_PM10 20
As+Cd+Ni [ng/d]
As+Cd+Ni [ng/d]
20 15 10 5
15 10 5
0
0 0
7
12
18
0
7
infant
12
18
0
7
child
12
18
0
7
adult
12
18
0
7
infant
18
0
7
12
18
adult
Time [h]
Cz_PM10
Cz_PM2.5 20
As+Cd+Ni [ng/d]
20
As+Cd+Ni [ng/d]
12 child
Time [h]
15 10 5
15 10 5
0
0 0
7
12
18
0
7
infant
12
18
0
7
child
12
18
0
7
adult
12
18
0
7
infant
12
18
0
7
child
Time [h]
12
18
adult
Time [h]
K_PM10
K_PM2.5 20
As+Cd+Ni [ng/d]
20
As+Cd+Ni [ng/d]
201
15 10 5
15 10 5 0
0 0
7
12 infant
18
0
7
12
18
child
0
7
12
0
18
7
12
18
0
infant
adult
7
12
18
child
0
7
12
18
adult
Time [h]
Time [h]
Fig. 2. Daily inhaled mass of metals bound to PM2.5 and PM10, calculated using PRA approach for Bielsko-Biała (B), Cze˛ stochowa (Cz) and Katowice (K) residents.
particles into the indoor environment designated specifically for Silesia region, to determine the differences in exposure level between indoor and outdoor environment we used the infiltration factors (Finf) derived from Ji and Zhao (2015). Those authors reviewed the available literature to determine the PM10 and PM2.5 infiltration rates meet typically in urban areas around the world (US, Europ, China). By averaging these results we modeled the Finf distributions independently for PM2.5 and PM10. Some simplification has been however made in presented work by treating those coefficients (determined strictly for PM) as an approximation of PM-bound metals infiltration rates (Table 3). Since metals penetrate into the human respiratory with different efficiency, depending on the size of its PM carriers we also include this additional aspect in our calculations using the particlesize dependent deposition rate designated independently for PM2.5 and PM10. Deposition was calculated in age specific manner using MPPDV 2.11 software. Results were presented as an average regional depositions for head e H; tracheobronchial e TB; and pulmonary e P regions, which have important functional distinctions. It was assumed that the mass percentage of metals that are inhaled and deposited within the lower respiratory tract is equal to the total amount of its PM carriers, which reaches TB and P regions. The calculated deposition rates (Fig. 3) were averaged between the
age groups (infant, child, adult) and described in terms of probability distributions (Table 3).
2.7. Cancer risk estimates The calculation of the health effect involved the multiplication of EI by carcinogenic slope factor (CSF) following equation (2):
R¼
1=3 EI $EF$ED$ CSFi BW 70 BW$AT
cf
(2)
where: R e incremental individual lifetime cancer risk (ILCR) resulting P from a specific metal dose R ¼ Ri for i-th element; EI e daily 1 exposure level (ng day ), CSFi e slope factor for i-th element (mg kg1 day1)1, EF e exposure frequency (day year1) (Chen and Liao, 2006), ED e exposure duration (year) (GUS, 2013), AT e averaging time for carcinogens AT ¼ 76 (year) 365 (day year1) (RAGS, 1989), BW e body weight (kg) (U.S. EPA, 2011) and cf e conversion factor (106) (mg ng1). Inhalation Unit Risk multiplication: IUR (mg m3)1 1000 (mg mg1) (BW(kg)/IR (m3 d1)) was necessary to convert IUR to CSF (mg/kg d)1 (RAGS, 1989).
202
K. Widziewicz, K. Loska / Atmospheric Environment 128 (2016) 198e207
Infant
10 8 6 4 2 0 Cd
Ni
PM2.5
As
Cd
Ni
16
18 16 14 12 10 8 6 4 2 0
Metal dose [ng/d]
Metal dose [ng/d]
Metal dose [ng/d]
12
As
Adult
Child
14
12 10 8 6 4 2 0
As
PM10
14
Cd
Ni
As
PM2.5
Cd
As
Ni
Cd
Ni
As
PM2.5
PM10
Cd
Ni
PM10
Fig. 3. Boxplots showing metal doses to which the Silesia residents are exposed to through inhalation to its ambient concentrations. The bottom and top of the box are 25 and 75 percentiles, the band inside the box is median, while the ends of the whiskers represents 2.5 and 97.5 percentiles of dose distribution.
Instead of 70 e year human lifespan for AT, the average life expectancy in the Silesian district (76 yr) was used (GUS, 2013). All exposure variables and parameters were treated probabilistically according to Table 3. The incomplete toxicological data on the CSF did not allow to treat this variable as a probability distribution, hence CSF was modeled as a point estimate (Table 2).
both PM fractions. PM2.5 and PM10 fractions contained the largest amounts of arsenic followed by nickel and cadmium. Those concentrations however did not exceed the permissible metal standards described in the Directive 2008/50/EC.
3.2. Daily exposure metals concentrations 2.8. Characteristics of risk and uncertainty analysis The risk was calculated as a sum of hazards for particular metals and then converted into population risk e an expected number of additional cancer incidence, respectively for adults: P P20 ILCR ¼ 76 i¼21 Ri=56; children: ILCR ¼ i¼2 Ri=19 and infants: ILCR ¼ P1 i¼0 Ri (GUS, 2013). The final risk was referred to the acceptable risk threshold. The quantitative assessment of the cancer risk was carried out employing a uniform MC simulation by using Crystal Ball software (Decisioneering, 1996). In order to ensure the stability of the numerical output of the model, the number of simulation runs was increased gradually. Therefore, 1.000, 5.000 and 10.000 independent iterations were carried out. The expected risk interval was established at 97.5 and 2.5 quartiles of distribution. It has been found that the number of 2000 iterations was enough to obtain the risk distribution stability. 3. Results 3.1. The concentration of airborne As, Cd and Ni The inhalation risk from airborne metals was calculated taking into account the temporal and spatial variability of its PM carriers. The annual mean concentration of PM2.5 were 36 mg/m3 for BielskoBiała, 32 mg/m3 for Cze˛ stochowa and 37 mg/m3 for Katowice, while for PM10 is was higher and reached 45 mg/m3; 38 mg/m3 and 45 mg/ m3 respectively (Tables 4 and 5) and exceeded the permissible standards at all sampling sites (Regulation of the Minister of Environment (2010)). The most frequent excessive 24-h PM concentrations were found in Katowice, where PM vehicular emissions account for a considerable part of the total air pollution (WIOS, 2012). An increase in PM2.5 and PM10 concentrations significantly correlated with an increase in airborne metal concentrations. Tables 4 and 5 give an average contents of metals associated with
The daily inhalation exposure (ng d1) from annual concentrations of As, Cd and Ni was calculated employing equation (1). Tables 4 and 5 show the analytical results of fitting the probability distributions to the concentration data based on the chi-squared test (p < 0.05). All the distributions were right-skewed, typical of log-normal distribution. Based on this distributions the MC simulations enabled to plot a probability density function for daily metal doses (EI). The results and related uncertainties from those simulations are presented in Fig. 3. A significant differences were observed when comparing daily exposures between PM2.5 and PM10 fractions. Much smaller discrepancies were noted, when examining exposure in terms of age sensitivity. The medians of the metals doses inhaled together with PM2.5 were 5.25; 7.44 and 5.81 (ng d1) for infants, children and adults, respectively, and higher than those associated with PM10 fraction e 2.14; 3.03 and 1.05 (ng d1), respectively. The exposure to PM2.5 associated metals was quite similar in each age group, although slightly lower in case of infants. The infants have generally lower values of IR than those of children and adults (Table 3), which gives an explanation for infants who are at lower risks. This is although not true in case of PM10 e related metals exposure, where the lowest EI values concern adults. Such distribution of the exposure is influenced by lung deposition ratio e significantly lower in case of PM10 particles (Fig. 4). Results obtained from MPPD modeling indicate that deposition efficiency varied within the particle size, human age and respiratory compartment. Fine (PM2.5) particles deposition was much higher compared to PM10. The result shows very slight variation in PM2.5 deposition efficiency among different age-groups: 0.37; 0.45 and 0.35 for infants, children and adults respectively. Bigger differences were observed in case of PM10: 0.35; 0.30 and 0.047 respectively (deposition efficiency in adults was approximately ten times lower compared to other age-groups). Analyzing the PM mass distribution between specific areas of respiratory, it was observed that PM2.5 deposition was the highest
Table 2 The inhalation unit risk and cancer potency factor for Monte Carlo analysis. Element
Inhalation unit risk (mg m3)1
Arsenic Cadmium Nickel
4.3 103 1.8 103 2.4 104
a
a
Cancer slope factor (mg/kg d)1 15.1 6.3 0.84
CSF was adjusted by BW and IR for inhalation unit risk value from Integrated Risk Information System of U. S. EPA.
K. Widziewicz, K. Loska / Atmospheric Environment 128 (2016) 198e207
203
Table 3 Resident equation inputs for 1D Monte Carlo analysis. PDF Symbol C
Unit
Type 3
(ng m
)
LN
Area
Element
B
Receptor
K
PM10 a,b
Finf D
e e
LN LN
EF ED
(day year1) year
LN UN
AT IR
d (m3 d1)
PE PE
infant
LN
child
LN
adult
LN
infant child adult
kg
PM2.5
As Cd Ni As Cd Ni As Cd Ni
Cz
BW
Activity
infant child adult infant child adult Sleep Light Sedentary Sleep Light Sedentary Sleep Light Sedentary
(1.7; 3.1) (0.9; 2.8)a,b (1.9; 2.0)a,b (2.4; 2.9)a,b (1.1; 2.5)a,b (2.0; 2.1)a,b (2.4; 1.9)a,b (1.2; 2.3)a,b (2.4; 1.6)a,b (0.58; 0.11)a,b (0.4; 0.07)a,b (0.4; 0.06)a,b (0.3; 0.02)a,b (252; 1.01)a,b (0e1)c,d (0e21)c,d (0e76)c,d 76e*365 6.6 15.8 6.7 (9.4; 0.8)a,b (22.5; 1.4)a,b (9.5; 0.8)a,b (10.1; 0.5)a,b (22.8; 0.9)a,b (10.1; 0.5)a,b (6.8; 1.9)a,b (34.0; 24.8)a,b (79.1; 5.4)a,b
(1.1; 1.4)a,b (0.4; 1.7)a,b (1.1; 1.4)a,b (1.6; 1.8)a,b (0.5; 2.1)a,b (1.4; 1.9)a,b (1.5; 1.8)a,b (0.7; 2.3)a,b (1.3; 1.6)a,b (0.39; 0.11)a,b (0.3; 0.11)a,b (0.3; 0.06)a,b (0.1; 0.01)a,b
LN e log-normal distribution, UN e uniform distribution, PE e point estimate. a Geometric mean. b Geometric SD. c Minimum. d Maximum. e The average lifespan averaged for males and females (GUS, 2013), B e Bielsko-Biała, Cz e Cze˛ stochowa, K e Katowice.
3.3. Inhalation incremental cancer risk
in the P region e 0.325, and the smallest in TB region e 0.089, while in the case of coarse particles the overwhelming particles mass was deposited in the head (H) region e with the 0.66 efficiency, and lowest in the pulmonary e 0.02. The total deposition efficiency for TB and P region, averaged over all analyzed age groups was approximately 40% greater in case of PM2.5 than for PM10. Pulmonary deposition was higher for children compared to adults, because of the higher physical activity of children and thus increased lung ventilation. It should be therefore noted that children are potentially more susceptible to particulate exposures.
The ILCR for the residents of particular cities was calculated following equation (2). Risk analysis was performed independently for metals associated with PM2.5 and PM10 fractions, taking into account its origin (exposure to ambient concentrations/exposure to indoor concentrations). The shape of risk distribution was lognormal in all the groups tested, however due to the complicated risk scenario, this article does not demonstrate a graphic diagram of this predicted PDFs. Instead the results are presented in the form of LN (geometric mean ± geometric standard deviation). Considering
Table 4 Statistical summary of airborne PM2.5 (mg m3) and the associated metal concentrations (ng m3). B e Bielsko-Biała, Cz e Cze˛ stochowa, K e Katowice. Sampling site
B
Cz
K
gm
geometric mean,
Parameter
PM2.5 As Cd Ni PM2.5 As Cd Ni PM2.5 As Cd Ni gsd
Statistics
Goodness-of-fit test
gm
gsd
Median
Min
Max
N
Skewness
Kurtozis
Distribution
36.2 1.7 0.9 1.9 32.0 2.4 1.1 2.0 37.2 2.4 1.2 2.4
38.1 3.1 2.8 2.0 32.4 0.9 2.5 2.1 28.7 1.9 2.3 1.6
21.4 0.82 0.28 1.30 23.0 1.53 0.44 1.38 28.2 1.88 0.56 2.00
1.4 0.1 0.1 0.1 2.0 0.1 0.1 0.3 1.4 0.3 0.02 0.7
247.2 22.8 16.9 96.1 309.8 36.6 10.6 88.1 224.4 7.4 12.0 12.2
352
2.1 2.7 3.9 13.4 3.5 2.7 2.1 8.8 2.5 0.7 3.2 2.4
5.4 8.9 21.4 209.0 19.7 11.2 6.5 92.7 10.2 0.23 13.1 7.9
e LN
e LN
e LN
p
c2 a ¼ 0.995
6 4 1
<0.0001 <0.0001 <0.0001
1.63 0.71 0.0039
5 8 1
<0.0001 <0.0001 <0.0001
1.14 2.73 0.0039
10 5 8
<0.0001 <0.0001 <0.0001
3.94 1.14 2.73
df
geometric SD, CV-coefficient of variation, Min-minimum, Max-maximum, df-degrees of freedom, p-probability value, c2-chi square test value.
204
K. Widziewicz, K. Loska / Atmospheric Environment 128 (2016) 198e207
Table 5 Statistical summary of airborne PM10 (mg m3) and the associated metal concentrations (ng m3). B e Bielsko-Biała, Cz e Cze˛ stochowa, K e Katowice. Sampling site
B
Cz
K
gm
geometric mean,
Parameter
Statistics
PM10 As Cd Ni PM10 As Cd Ni PM10 As Cd Ni gsd
Goodness-of-fit test
gm
gsd
Median
Min
Max
N
Skewness
Kurtozis
Distribution
42.8 1.1 0.4 1.1 38.4 1.6 0.5 1.4 45.3 1.5 0.7 1.3
39.3 1.4 1.7 1.4 28.4 1.8 2.1 1.9 33.2 1.8 2.3 1.6
28.0 0.68 0.09 0.68 31.0 1.06 0.12 0.83 37.0 0.96 0.20 0.82
5.0 0.9 0.2 1.0 5.0 0.9 0.2 1.0 5.0 0.9 0.2 1.0
256.0 2.6 1.5 4.7 285.0 8.9 2.9 11.0 286.0 6.3 2.6 5.8
326 26
2.2 2.2 1.9 4.1 3.1 3.1 1.9 3.1 2.7 2.0 0.8 2.7
6.2 2.8 4.4 16.0 17.9 11.2 3.8 9.1 12.3 4.5 0.4 8.0
e LN
367 26
363 26
e LN
e LN
df
p
c2 a ¼ 0.995
1 2 1
<0.0001 <0.0001 <0.0001
0.0039 0.10 0.0039
2 4 2
<0.0001 <0.0001 <0.0001
0.10 0.71 0.10
3 8 2
<0.0001 <0.0001 <0.0001
0.35 2.73 0.10
geometric SD, CV-coefficient of variation, Min-minimum, Max-maximum, df-degrees of freedom, p-probability value, c2-chi square test value.
Deposition fraction
0.5
0.4
0.3
0.2
0.1
0 3 mth
21 mth
23 mth
28 mth
3 yr
8 yr
9 yr
14 yr
18 yr
21 yr
human age TB_PM2.5
P_PM2.5
TB_PM10
P_PM10
Fig. 4. Averaged regional deposition of PM2.5 and PM10 particles.
outdoor/indoor metals sources in shaping the distribution of health risks among Silesia residents it was shown that outdoor exposures are potentially more harmful. The total ILCR averaged for the populations of Bielsko-Biała, Cze˛ stochowa and Katowice due to outdoor-originated PM2.5-bound metals in the indoor environment was LN (9.5 107 ± 1.5 106), LN (1.29 106 ± 2.06 106), LN (1.33 106 ± 2.09 106), while for PM10-bound metals it was LN (2.41 107 ± 4.9 107), LN (3.44 107 ± 5.76 107), LN (3.45 107 ± 5.80 107). The dispersion of risks arranged a bit differently in case of exposure to outdoor originated metals: for ambient PM2.5-bound metals LN (1.40 106 ± 2.16 106), LN (1.91 106 ± 2.94 106), LN (1.96 106 ± 3.02 106), while in case of PM10-bound metals LN (4.58 107 ± 7.54 107), LN (6.47 107 ± 1.06 106), LN (6.56 107 ± 1.07 106). The obtained findings e even two times higher risk in case of direct exposure to ambient metals concentrations suggest that indoor (closed environment) play a significant role in protecting occupants against exposure to air pollutants (this observation couldn't be generalized to indoor environments affected by internal emissions sources). As the size distribution shifts toward smaller particles, the infiltration factor increases. This easily explains the differences in lifelong exposures between PM2.5 and PM10-bound metals. Taking into account the age-dependent sensitivity it was shown that inhalation risks between analyzed age groups arranged accordingly to the daily metal doses (Table 6). The inhalation ILCR values for specific age groups was highest in case of children exposed to PM2.5-bound metals of outdoor origin LN
(9.25 107 ± 1.44 106), while the smallest risks concerned infants exposed to indoor concentrations of PM10-bound metals (1.66 108 ± 2.78 108). All 97.5% probabilities of Total ILCRs were below 106, which indicate acceptable probability distributions for each of the concerned age groups.
3.4. Sensitivity analysis The findings from sensitivity analysis are demonstrated in the diagrams depicting the percentage contribution of particular exposure variables to the total variance of predicted risk. The variables which exerted the strongest effect on the health risk for all the age groups included exposure duration (ED) and body weight (BW). Their approximate percentage contribution ranged from 28.1% to 44.6% for ED and from 23.1% to 32.8% for BW depending on the scenario (Fig. 5). The variation in the concentration-response estimates was to a very small extent owned to differences in metals migration into indoor environment. The uncertainties brought by metals infiltration into residences (Finf parameter), were close to 0.2% and therefore negligible with respect to variability affecting personalhealth responses. Together with other parameters, like deposition efficiency and inhalation rate, Finf accounts for only 10% of total risk variance.
K. Widziewicz, K. Loska / Atmospheric Environment 128 (2016) 198e207
205
Table 6 Predicted PDF of total incremental lifetime cancer risk for infants, children and adults. PM fraction
Age group
Total ILCR from outdoor exposure LN (gm, gsd)
PM2.5
Infant Child Adult Infant Child Adult
(6.76 (8.63 (9.25 (3.31 (4.02 (1.92
PM10
gm
geometric mean,
gsd
108 107 107 108 107 107
± ± ± ± ± ±
1.12 1.69 1.44 5.19 8.31 2.99
Total ILCR from indoor exposure LN (gm, gsd)
107) 106) 106) 108) 107) 107)
(4.48 (5.76 (5.75 (1.66 (2.04 (9.06
108 107 107 108 107 108
± ± ± ± ± ±
7.51 1.44 9.15 2.78 4.02 1.44
108) 106) 107) 108) 107) 107)
geometric SD.
PM2.5 others
PM10
10.3
BW_adult
others
1.2
B_PM2.5_As_child
9.3
K_PM10_As_child
1
2.2 ED_adult
BW_child
12.3
23.1
ED_child
BW_child
28.1
ED_adult
35.1
0
10
20
30
40
Contribution to the variance view [%]
32.8
ED_child
44.6
0
10
20
30
40
50
Contribution to the variance view [%]
Fig. 5. Results of sensitivity analysis for 1-D MCA (indoor exposure scenario).
4. Discussion The MC technique employed herein to assess health risk enabled not only a more accurate calculation of its value, but also helped acquire relevant information on its distribution in particular age groups. The probabilistic analysis revealed that cancer risk in the tested population did not exceed the permissible level of 106e104. A higher risk of LN (1.80 106 ± 2.89 106) was found for metal exposure associated with PM2.5, while the value for PM10 reached LN (6.17 107 ± 1.08 106). Similar results were obtained by Sadovska (2012), who calculated an inhalation risks for Silesia neighbors e the population of Ostrava Karvina Coal Basin (an industrial region in the Czech Republic) exposed to airborne As, Cd and Ni. It must be emphasized that the statement about greater health hazard caused by PM2.5-bound metals is biased by the fact that PM fractions were collected at different sampling sites. A direct comparison of inhalation risks due to PM2.5 and PM10 exposure was therefore not possible. Literature data however indicate a more deleterious effect of PM2.5 fraction compared to PM10, which is probably caused by the larger active surface of grains, favoring metal concentration (Onat et al., 2012). This phenomenon is particularly alarming when considering the fact that fine particles account for even 90% of airborne particles (Salonen and Pennanen, 2007). The most popular approach in estimating human exposure to air pollution is treating the ambient concentrations as surrogates for real exposure estimation (Zhang et al., 2009; Zhou and Zhao, 2012), however this approach can result in a significant exposure misclassification and risk overestimation (Payne-Sturges et al., 2004). Since human population most time spend indoors (Zhou and Zhao, 2012) a special attention must be paid to the reliable estimation of I/O pollution ratio. In this work the exposure to As, Cd and Ni concentrations was therefore calculated in terms of its infiltration efficiency into indoor environment. Due to lack of data concerning metals concentrations in indoor/outdoor relationship, rny et al., 1995), we made an determined specifically for Silesia (Go assumption that indoor metals concentrations are mostly governed by the outdoor emissions (Table 3), which situation is typical for regions attributed to the excessive industrial activities (Nazir et al.,
2011). Obtained results showed that risks posed by ambient metals concentrations were almost two times greater compared to the risk revealed in case of chronic exposure to indoor metal pollution and additionally suggest an important role of buildings in protecting people from air pollution. Further research must be although provide in order to define to which extent risks are governed not only by metals infiltration into indoor environment but also by the existence of indoor emission sources (metals pollution generated by smoking; burning fossil fuels in domestic stoves; cooking etc.). The second factor controlling metals exposure is its deposition efficiency in the respiratory track. Since metals are adsorbed on PM particles, its deposition rate depends on PM concentration in ambient air, size of those particles and human ventilation rate. The dose of inhaled metals should be therefore calculated as the product of metals concentration, time which individuals spent in a specific environment and pulmonary ventilation rate (including metals deposition in particular area of the respiratory track). Presented data indicate that aerosol particles with a diameter less than 10 mm are mostly deposited in the thoracic region, while PM2.5 reach the pulmonary region. Health effects depends not only on the amount of the inhaled particles but also on the adsorbed substances they carry into specific regions of the lungs. Hence, the determination of metals deposition in the lung, their uptake, redistribution, storage, and removal is very important considering the range of harmful effects. In this work we observed a strong impact of particles size on its deposition. Coarse particles (PM10) deposition in the lower respiratory track was significantly lower compared to fine mode particles (PM2.5). The presented approach enabled to gain insights on the effects of particles size and human age on the PM deposition efficiency. It was shown that the total PM2.5 deposition efficiency was highest in case of children and for PM10 in case of infants. Only particles <2.5 mm seem to be influenced by the enhanced pulmonary deposition rate because particles larger 10 mm efficiently deposit in the upper airways. An increase of risk was correlated was not only with the total metals contents and its deposition in the respiratory track but also with the age. Infants appeared to be the most susceptible to the metals present in the air. The cancer risk for that age group was higher than the one for adults by 77% and 81.8% for PM2.5 and PM10,
206
K. Widziewicz, K. Loska / Atmospheric Environment 128 (2016) 198e207
respectively. The stratification of risk in particular age groups was also observed by Trojanowska and Swietlik (2012) who investigated the inhalation exposure to As, Cd and Ni in the largest Polish conurbations. Their findings indicate that children are the most exposed age group in the population and inhale approx. 50% more air per kg of body mass compared to adults. Thus, children and infants should avoid spending a lot of time outdoors, particularly when doing intense physical activity, which increases lung ventilation. In mathematical terms, the parameter, which mostly differentiated health risk between the age groups was the exposure duration (Table 3). So far, a sole criterion for selecting a suitable distribution of exposure duration has not been established. A lot of authors employing MC technique treat this variable as a point estimate (de Oliveira et al., 2012; Chen and Liao, 2006), roughly equal to the average lifespan of an inhabitant. This work employed the exposure duration determined by GUS (2013) data, however, it should be noted that selection of another distribution would lead to significantly different results. The considerable effect of ED on the final risk is confirmed by the results of sensitivity analysis (Fig. 5). The contribution of this variable to the total variance of model output was much bigger than the contribution of the “metal concentration” variable. Therefore to increase the accuracy of the results, efforts should focus on a better definition of ED distribution. Although physiological parameters as well as exposure time significantly affects carcinogenic hazards a key issue when estimating cancer risk caused by certain air pollutants is however the slope factor. Since a big differences exists in the order of cancer potency between single pollutants, the risk value is mostly governed by these one, which has the highest CSF value. Literature data indicates that CSF is the most highly uncertain parameters in risk analysis models, which occupied even more than 70% of the total variance of ILCR (Chen and Liao, 2006). It is because of the incomplete or even ambiguous epidemiological evidence, and uncertainties connected with extrapolation of animal data to human data (Biesiada and Bubak, 2001). To overcome these uncertainties it is suggested to treat CSF as a variable or fuzzy parameter (Mofarrah and Husain, 2010). In order to greatly improve the accuracy of the risk assessment, improving the CSF values might be much more effective than improving the accuracy of the exposure concentrations (Chen and Liao, 2006). Since CSF values are usually similar to the upper limits of estimates, the authors suspect that the actual health risk resulting from the inhalation exposure to As, Cd and Ni in Silesia may be lower than the one calculated. Further research into the determination of inhalation risk in Silesia should be aimed at determining probability distribution of particular exposure parameters, so that they could best reflect the heterogeneity of its population. 5. Conclusions It has been found that As, Cd and Ni concentrations measured in Silesia region do not exceed the permissible concentrations stipulated in the EU Directive, and the additional cancer risk faced by the population was lower than 106, indicating no significant cancer risks. This, in terms of public healthcare, does not necessitate any countermeasures to reduce the emissions of the metals into the atmospheric air. The risk analysis conducted by PRA also proved that PM2.5-bound metals poses a higher health risk, and both adults and children are the most susceptible to this pollutants. The total ILCR averaged for the populations of Bielsko-Biała, Cze˛ stochowa and Katowice calculated using PRA, lower by even order of magnitude than ILCR values calculated deterministically, confirmed that the variability of exposure parameters in the population exerts a significant effect on health risk. Knowledge of its strength is
particularly important when it affects the legitimacy of taking suitable countermeasures to minimize the risk. Funding sources The research was financed by “DoktoRIS” e Program for Innovative Silesia 2007e2013, co-financed by the European Union under the European Social Fund. http://www.doktoris.us.edu.pl/ projekt/projekt.php. Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.atmosenv.2015.12.061. References Biesiada, M., 2001. Simulations in health risk assessment. Int. J. Occup. Med. Environ. Health 14, 397e402. Biesiada, M., Bubak, A., 2001. Basics of Environmental Health Risk Assessment. Training Materials ‘Theory and Practice of Environmental Impact Assessments on Health’. Institute of Occupational Medicine and Environmental Health in Sosnowiec, pp. 29e38. BS EN 14907:2006. Ambient air quality e Standard gravimetric method for determining the mass fraction of PM2. 5 particulate matter. Chen, S.C., Liao, C.M., 2006. Health risk assessment on human exposed to environmental polycyclic aromatic hydrocarbons pollution sources. Sci. Total Environ. 366, 112e123. de Oliveira, B.F.A., Ignotti, E., Artaxo, P., Saldiva, P.H.D., Junger, W.L., Hacon, S., 2012. Risk assessment of PM2.5 to child residents in Brazilian Amazon region with biofuel production. Environ. Health 11, 301e309. Decisioneering, 1996. Crystal BallTM Fusion Edition 11.1.2.3.000 Inc. (Denver, CO). Directive 2008/50/EC of the European Parliament and of the Council of 21 May 2008 on ambient air quality and cleaner air for Europe. Gomis cek, B., Hauck, H., Stopper, S., Preining, O., 2004. Spatial and temporal variations of PM1, PM2.5, PM10 and particle number concentration during the AUPHEPdproject. Atmos. Environ. 38, 3917e3934. rny, R.L., Je˛ drzejczak, A., Pastuszka, J.S., 1995. Dust particles and metals in outGo door and indoor air of Upper Silesia. Ann. Natl. Inst. Hyg. 46, 151e161. GUS, 2013. Life expectancy tables of Poland 2012. Warszawa 2013. Harrison, R.M., Smith, D.J.T., Kibble, A.J., 2004. What is responsible for the carcinogenicity of PM2.5? Occup. Environ. Med. 61, 799e805. IARC, 2014. Monographs e Classifications e by Group (Last Update 31 March 2014). http://monographs.iarc.fr/ENG/Classification/ClassificationsGroupOrder.pdf. Ji, W., Zhao, B., 2015. Estimating mortality derived from indoor exposure to particles of outdoor origin. Plos One 10, 1e15. Klepeis, N.E., Nelson, W.C., Ott, W.R., Robinson, J.P., Tsang, A.M., Switzer, P., Behar, J.V., Hern, S.C., Engelmann, W.H., 2001. The National Human Activity Pattern Survey (NHAPS): a resource for assessing exposure to environmental pollutants. J. Expo. Analysis Environ. Epidemiol. 11, 231e252. Mofarrah, A., Husain, T., 2010. Modeling for Uncertainty Assessment in Human Health Risk Quantification: a Fuzzy Based Approach. International Environmental Modelling and Software Society (iEMSs). International Congress on Environmental Modelling and Software. Nazir, R., Shaheen, N., Shah, M.H., 2011. Indoor/outdoor relationship of trace metals in the atmospheric particulate matter of an industrial area. Atmos. Res. 101, 765e772. Onat, B., Sahin, U.A., Bayat, C., 2012. Assessment of particulate matter in the urban atmosphere: size distribution, metal composition and source characterization using principal component analysis. J. Environ. Monit. 14, 1400e1409. Payne-Sturges, D.C., Burke, T.A., Breysse, P., Diener-West, M., Buckley, T.J., 2004. Personal exposure meets risk assessment: a comparison of measured and modeled exposures and risks in an urban community. Environ. Health Perspect. 112, 589e598. Pope, C.A., Burnett, R.T., Turner, M.C., Cohen, A., Krewski, D., Jerrett, M., Gapstur, S.M., Thum, M.J., 2011. Lung cancer and cardiovascular disease mortality associated with ambient air pollution and cigarette smoke: shape of the exposure-response relationships. Environ. Health Perspect. 119, 1616e1621. RAGS, 1989. Risk Assessment Guidance for Superfund Volume I Human Health Evaluation Manual (Part A), EPA/540/1e89/002, Washington, D.C. Office of Emergency and Remedial Response U.S. Environmental Protection Agency. Regulation of the Minister of Environment of 26 January 2010 on the reference values for certain substances in the air (Journal of Laws of 2010, No. 16, item. 87). 't Mannetje, A., Bencko, V., Brennan, P., Zaridze, D., Szeszenia-Dabrowska, N., , E., Cassidy, A., Mates, D., Foretova, L., Rudnai, P., Lissowska, J., Fabi anova Janout, V., Fevotte, J., Fletcher, T., Boffetta, P., 2011. Occupational exposure to metal compounds and lung cancer. Results from a multi-center case-control study in Central/Eastern Europe and UK. Cancer Causes Control 22, 1669e1680. Sadovska, V., 2012. Health risk assessment of heavy metals adsorbed in particulates.
K. Widziewicz, K. Loska / Atmospheric Environment 128 (2016) 198e207 World Acad. Sci. Eng. Technol. 6, 170e173. Salonen, R.O., Pennanen, A., 2007. the impact of fine particles on health. In: Herring, Peter (Ed.), Views and Conclusions from the FINE Particles eTechnology. Environment and Health Technology Programme. Tekes. ˛ skie. Regional Development Strategy of the Voivodeship of Silesia 2020þ. Sla https://rpo.slaskie.pl/zalaczniki/2014/01/17/1389967464.pdf Trojanowska, M., Swietlik, R., 2012. Heavy metals cadmium, nickel and arsenic environmental inhalation hazard of residents of Polish cities. Environ. Med. 15, 33e41. U.S. EPA, 2011. (U.S. Environmental Protection Agency). Exposure Factors Handbook: 2011 Edition. Report EPA/600/R-090/052F, Washington, DC. WHO, 2013. Health Effects of Particulate Matter. WHO Report. http://www.euro. who.int/__data/assets/pdf_file/0006/189051/Health-effects-of-particulatematter-final-Eng.pdf.
207
2012. State of the Environment in the Silesia District in 2012. Library Of WIOS, Environmental Monitoring 2013, pp. 1e140. http://www.katowice.pios.gov.pl/ monitoring/raporty/2012/raport2012.pdf (20.08.2015). Yaman, M., 2012. In: Khare, Mukesh (Ed.), Comprehensive Comparison of Trace Metal Concentrations in Inhaled Air Samples (pp. 29e67) in Air Pollution e Monitoring, Modelling, Health and Control. InTech, p. 254. ISBN 978-953-510381-3. Zhang, Y., Tao, S., Shen, H., Ma, J., 2009. Inhalation exposure to ambient polycyclic aromatic hydrocarbons and lung cancer risk of Chinese population. Proc. Natl. Acad. Sci. U. S. A. 106, 21063e21067. Zhou, B., Zhao, B., 2012. Population inhalation exposure to polycyclic aromatic hydrocarbons and associated lung cancer risk in Beijing region: contributions of indoor and outdoor sources and exposures. Atmos. Environ. 62, 472e480.