You are on page 1of 8

Science of the Total Environment 407 (2009) 4894–4901

Contents lists available at ScienceDirect

Science of the Total Environment


j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / s c i t o t e n v

Exposure to fine particulate matter in ten night clubs in Athens Greece: Studying the
effect of ventilation, cigarette smoking and resuspension
Christos Halios a,⁎, Mat Santamouris a, Ageliki Helmi a, Maria Kapsalaki b, Maria Saliari a,
Anastasia Spanou a, Dimitris Tsakos a
a
Department of Environmental Physics and Meteorology, Faculty of Physics, University of Athens, Building Phys 5, University Campus, Athens, 15784 Greece
b
Faculty of Engineering, University of Porto, MIT-Portugal

a r t i c l e i n f o a b s t r a c t

Article history: The aim of the present work is to study the occupants' exposure to fine particulate concentrations in ten
Received 22 August 2008 nightclubs (NCs) in Athens, Greece. Measurements of PM1 and PM2.5 were made in the outdoor and indoor
Received in revised form 25 January 2009 environment of each NC. The average indoor PM1 and PM2.5 concentrations were found to be 181.77 μg m− 3 and
Accepted 10 February 2009
454.08 μg m− 3 respectively, while the corresponding outdoor values were 11.04 μg m− 3 and 32.19 μg m− 3.
Available online 4 June 2009
Ventilation and resuspension rates were estimated through consecutive numerical experiments with an indoor
air quality model and were found to be remarkably lower than the minimum values recommended by national
Keywords:
Indoor exposure standards. The relative effects of the ventilation and smoking on the occupants' exposures were examined
Resuspension using multiple regression techniques. It was found that given the low ventilation rates, the effect of smoking as
Ventilation well as the occupancy is of the highest importance. Numerical evaluations showed that if the ventilation rates
Nightclubs were at the minimum values set by national standards, then the indoor exposures would be reduced at the 70%
PM1 of the present exposure values.
PM2.5 © 2009 Elsevier B.V. All rights reserved.

1. Introduction Sidestream tobacco smoke is defined as the undiluted plume coming


from the smoldering end of the cigarette, mainstream smoke is the
A great number of scientific studies have shown that adverse health undiluted puff of smoke that is drawn through the cigar or cigarette
effects are associated with the public exposure to high particulate and then exhaled by the smoker, and ETS emissions defined as the
concentrations as well as to short-time variations of particulate matter mass of particles that have come to be dispersed in a previously
even at low concentrations (Brunekreef et al., 1995). It is also evident pollutant-free room just after a cigarette (or cigar) has been smoked.
that long-term exposure to low particulate concentrations is asso- During the last years several studies have examined indoor air
ciated with increased rates of bronchitis and reduced lung function quality in bars, clubs and pubs focusing mainly on the relationship
(WHO, 2000). Even though the vast majority of epidemiological between the smoking status and the indoor concentrations found in
studies have used particle metrics such as PM10 (particles with these places. For example Arundel et al. (1987) estimated the non
aerodynamic diameter less than 10 μm), black smoke or sulfate smokers average exposure to particulate ETS to be approximately three
concentration (Harrison and Yin, 2000), the health significance of fine to five times higher in bars than in other indoor environments, while
particles is now well established (Schwartz et al., 1996; Pope et al., Ballesta et al. (2006) in a one day cross sectional study in six European
2000; McDonnell et al., 2000). Previous studies have shown that the cities found the highest benzene levels within bars and motor vehicles.
fraction of the inhaled particulate matter deposited in the respiratory A number of other studies have shown that indoor air quality with
system of the lungs peaks for particles with aerodynamic diameter of respect to respirable particles, polycyclic aromatic hydrocarbons, vola-
2.5 μm (Chan and Lippman, 1980; Mitsakou et al., 2005). tile organic compounds and other ETS related constituents have been
Quite a few number of studies have focus on the determination of significantly reduced after the ban of smoking within bars and pubs in
the particle distributions that are emitted from the cigarettes burning Boston (Repace et al., 2006), Texas (Waring and Siegel, 2007),
(Keith and Derrick 1960; Chang et al. 1985; Ueno and Peters 1986; Delaware (Repace, 2004), Ireland (Goodman et al., 2007), New
Chung and Dunn-Rankin 1996; Klepeis et al., 2003). In general these Zealand (Wilson et al., 2007), Finland (Johnsson et al., 2006) and in
studies have focused on the size distribution of mainstream, side- Italy (Tominz et al., 2006). Therefore it is well established that the
stream cigarette smoke or environmental tobacco smoke (ETS). exposure within bars contributes largely to particulate exposure for
people who are attending these places.
⁎ Corresponding author. Tel.: +30 210 7276847; fax: +30 210 7295282.
In the absence of smoking ban, adequate ventilation should
E-mail addresses: c_halios@phys.uoa.gr (C. Halios), pds07005@fe.up.pt operate properly. In general, the significance of adequate ventilation
(M. Kapsalaki). in the overall indoor air quality is well established (Santamouris et al.,

0048-9697/$ – see front matter © 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.scitotenv.2009.02.023
C. Halios et al. / Science of the Total Environment 407 (2009) 4894–4901 4895

Fig. 1. Map of Athens. Bullets and numbers indicate the locations of the NCs. In the upper photograph the greater Athens Area is presented.

2007). Hyvarinen et al. (2006) reported improvements to ETS levels tration in indoor (Lee et al., 1999) transportation (Leutwyler et al.,
when the ventilation system is improved and highlights that bars and 2002), and outdoor (Hitchins et al., 2000) microenvironments. The
nightclubs require higher ventilation rates than restaurants when interval time log was set to 1 s. Then the logged data were averaged to
smoking is permitted. Several European countries have adopted 1-minute values and the outliers (i.e. values greater than the average
workplace bans on smoking that include the hospitality industry value plus three times the standard deviation) of each data set were
venues such as pubs, bars and restaurants, and legislation regarding excluded from the data sets. The sampling protocol was scheduled as
adequate ventilation conditions in order to reduce exposure of the follows: one experiment was performed at each night club and lasted
customers of these venues to ETS. However, in the Greece there is no for 3 h. Outdoor PM1 concentrations were measured for half an hour,
specific legislation controlling exposure to ETS in the workplace, but the next half hour the outdoor PM2.5 concentrations were measured,
there is a legislation concerning the minimum requirements of clear and then two sequential sets of indoor PM1 and PM2.5 measurements
air per person in public venues. The minimum required ventilation were performed, 1 hour each. Experiments were performed during
rate is 51 m3/h per person and the respective desirable ventilation working days' nights (Monday to Friday nights). The exact dates of the
rate is 85 m3/h per person. experiments are presented in Table 1. Experiments were conducted
The aim of the present work is to assess the indoor air quality in 10 during March, April and May of 2007. The outdoor measurements were
nightclubs in Athens, Greece with respect to fine particulate matter made in front of the nightclubs. Instruments were positioned inside, on
and study the occupants' exposure to particulate concentrations. an average height of 0.8 m above the tables and at least 3 m from the
Furthermore, the relative effect of the major factors that influence the entrance, in the centre of the nightclub when possible, and at least 2 m
occupants' exposures is examined. from any potential pollutant source (cigarettes, etc.). The time evolution
of the outdoor and indoor concentrations in NC2 can be seen in Fig. 2.
2. Materials and methods The technique used by the DustTrak aerosol monitor differs
substantially from validated gravimetric methods. The DustTrak
For the purposes of the present work particulate measurements in Aerosol Monitor uses 90-deg light scattering to measure the mass
ten nightclubs (hereafter NCs) were carried out. The NCs are situated in concentration of particles in an air stream that passes through an
the Greater Athens Area and their exact locations are depicted in Fig. 1. impactor assembly. The amount of scattered light is proportional to the
The NCs were selected randomly from the official list of business in the volume concentration of the aerosol. DustTrak measurements can be
Athens area. Nine of them are located close to the center of the city and influenced by aerosol composition and size distribution. Relationships
one at a suburb at the northeastern side of the city. Even though nine between concentrations obtained from DustTrak and gravimetric
out of the ten NCs were equipped with ducted ventilation systems the methods have been reported in several studies and are summarized
mechanical ventilation systems did not work during the experiments. in Table 2. We include studies that compare DustTrak measurements
In Greece, the mechanical ventilation systems of the nightclubs very
often are off. The number of occupants as well as the number of
smokers was recorded during the experiments and in Table 1 the
Table 1
general characteristics of the nightclubs are presented. General characteristics of the NCs.
Measurements of the PM1 and PM2.5 indoor and outdoor fraction of
the atmospheric aerosol were made with a DustTrak portable Aerosol NC1 NC2 NC3 NC4 NC5 NC6 NC7 NC8 NC9 NC10

monitor (TSI model 8520). DustTrak is a real-time laser photometric Date of 2/3 8/3 15/3 21/3 27/4 3.5 4/5 7/5 8/5 10/5
measurements
instrumentation for the determination of aerosol mass concentra-
A (m2) 339 795 388 592 360 271 838 475 400 348
tions, thus capable of measuring short-term exposure level. The V (m3) 280 700 408 960 400 220 1100 375 312 360
measurement is performed using a light scattering technique. # of occupants 138 128 78 105 54 81 88 42 49 49
Different impactors are available for the inlet of DustTrak allowing # of smokers 95 95 53 75 32 45 51 20 16 16
measurements of PM2.5 and PM1. Recently, DustTrak aerosol monitor A is the average indoor surfaces, V is the total volume of the NCs, # of occupants and # of
has been applied for the determination of particulate mass concen- smokers are the number of occupants and the number of smokers in each NC respectively.
4896 C. Halios et al. / Science of the Total Environment 407 (2009) 4894–4901

Fig. 2. Temporal evolution of the indoor and outdoor PM1 and PM2.5 concentrations in NC2.

with measurements obtained from gravimetric methods in the general their study: PM2.5 (corrected)=0.24 PM2.5 (measured)+5.24 (μg m− 3).
indoor and ambient environment. Gravimetric methods include The only study that reports comparisons for the PM1 concentrations
measurements with personal exposure monitors (P.E.M.) and standard measured with the DustTrak is Georgalas et al. (2006), and we
gravitational methods (S.G.M.). corrected the ambient and indoor PM1 concentrations by applying the
Even though it is clear that in all cases DustTrak and gravimetric relationship given in their study PM1 (corrected)= 0.12 PM1 (mea-
concentrations are highly correlated, PM2.5 measurements from sured) + 3.08 (μg m− 3).
DustTrak are overestimations when comparing with the reference Theoretical simulations were conducted with the aid of the well-
methods. In the majority of these studies, PM2.5 concentrations known indoor air quality model MIAQ (Multichamber Indoor Air
obtained by the DustTrak in the ambient and indoor smoke-free Quality model), in order to calculate the ventilation and resuspension
environments are overestimations by a factor of 3 (Kurmi et al., 2008; rates within each NC. Resuspension is the result of the aerodynamic
Ramachandran et al., 2000; Valente et al., 2007; Yanosky et al., 2002). lift forces, which are associated with turbulent bursts and are
On the other hand, Lee et al. (2001) reported DustTrak concentrations competing over the adhesion forces acting on the particle over time
1.5 fold higher than the reference method, but they do not comment (Cleaver and Yates, 1973). The calculations involved consecutive
on the type of aerosol they used for the comparison. Chang et al. theoretical calculations with varying ventilation and resuspension
(2001) report differences of the order of 2. Following the majority of rates during each theoretical simulation, until the model-predicted
the studies shown in Table 2, which report that PM2.5 measurements concentrations coincide with the measured concentrations. MIAQ is a
from DustTrak are 3-fold the measurements from reference methods, general mathematical model for aerosol dynamics in indoor air
we corrected the measured ambient PM2.5 concentrations by applying (Nazaroff and Cass, 1989). It accounts for the effects of ventilation,
the relationship given by Ramachandran et al. (2000) for the ambient filtration, direct emission, deposition onto surfaces, and coagulation
environment: PM2.5 (corrected) = 0.33 PM2.5 (measured). The only for particles. Model results have been validated with experimental
study that reports comparison in an indoor environment reserved for data (Nazaroff and Cass, 1986; Nazaroff and Cass, 1989; Drakou et al.,
smokers is Valente et al. (2007). Therefore, the measured indoor 1998). Details regarding the numerical simulations are presented in
concentrations were corrected by applying the relationship given in section 3.2.

Table 2
Studies that report comparison between concurrent measurements from DustTrak and gravimetric methods.

Study Reference Integration Number of Season Aerosol Correlation Calibration


method time observations coefficient factorsa
Chang et al. (2001) P.E.M.b 12 h 35 Summer PM2.5 ambient 0.93 0.48x
Chang et al. (2001) P.E.M.b 12 h 41 Winter PM2.5 ambient 0.9 0.49x
Kurmi et al. (2008) S.G.M.c 24 h 16 N/A PM2.5 indoor (rural) 0.69 0.31x
Kurmi et al. (2008) S.G.M.c 24 h N/A PM2.5 indoor (urban) 0.95 0.35x
Lee et al. (2001) S.G.M.d N/A ≈5 Summer PM2.5 0.96 0.66x
Ramachandran et al. (2000) S.G.M.e 24 h N20 Summer PM2.5 ambient 0.96 ≈ 0.33x
Ramachandran et al. (2000) P.E.M.f 24 h N50 Summer PM2.5 indoor smoke free 0.87 ≈ 0.51x
Valente et al. (2007) S.G.M.g 8h N/A Summer and winter PM2.5 indoor smoking 0.87 0.24x + 5.23
Valente et al. (2007) S.G.M.g 8h N/A Summer and winter PM2.5 indoor smoke free 0.87 0.37x + 3.5
Yanosky et al. (2002) S.G.M.h 24 h 17 Winter PM2.5 indoors (empty) 0.93 0.33x + 2.25
Georgalas et al. (2006) S.G.M.i 24 h N40 N/A PM1 ambient 0.97 0.12x + 3.08
a
Factors that the measured DustTrak data (x) should be multiplied with, in order to obtain the corrected values.
b
Personal exposure monitors, MSP Corp.
c
In accordance with the MDHS 14/3 (HSE 2000).
d
Partisol Sampler (Model 2000-H, Rupprecht and Patashnich Co., Inc).
e
Andersen RAAS 2.5-300 sampler.
f
MSP Inc, Minneapolis.
g
Skypost PM TCR-Tecora).
h
PQ200 Ambient Fine Particulate Sampler (BGI Inc.).
i
Partisol Sampler 2025.
C. Halios et al. / Science of the Total Environment 407 (2009) 4894–4901 4897

3. Results

3.1. Concentrations

In Figs. 3 and 4 the box plots of the measured outdoor


concentrations of PM1 and PM2.5 are presented and in Figs. 5 and 6
the box plots of the respective indoor concentrations are shown.
Measured indoor particulate concentrations are particularly high,
reaching values as high as 0.97 mg m− 3 and 2.05 mg m− 3 for PM1 and
PM2.5 respectively, while the average measured values in the NC are
186.86 μg m− 3 and 484.6 μg m− 3 for PM1 and PM2.5 respectively. In
NC 10 there were no available outdoor measurements. The average
values of the particulate concentrations between the different NCs
show significant variability. The highest indoor average PM1 concen-
tration was observed in NC4 (394.47 μg m− 3) and the lowest in NC6
(93.9 mg m− 3) while the highest indoor average PM2.5 concentration
was observed in NC1 (598.3 μg m− 3) and the lowest in NC9 (145.54 μg
m− 3). Thus, it is apparent that the mechanisms that control the indoor
particulate concentrations NCs show significant variation in the
different NCs.

Fig. 4. Box plot of PM2.5 out.


3.2. Numerical simulations

During the numerical experiments the measured outdoor parti- rate, assuming that each smoker smokes three cigarettes per hour.
culate concentrations and the indoor particulate mass emission rate Initial indoor particulate concentrations were assumed to be the
due to cigarette smoking were set as input values to the model. The measured concentrations during the first set of measurements.
particulate mass emission rate due to the cigarette smoking was During these theoretical simulations resuspension cannot be expli-
calculated as following: according to Klepeis et al. (2003), the citly calculated by the model. Rather, it can be indirectly included in our
particulate mass emission rate yielding from one cigarette is calculations by considering an extra production term. Therefore in the
0.91 mg min− 1 assuming that all of the emitted aerosol mass is in theoretical simulations an extra production term is included that we
the PM1 aerosol fraction. The recorded average number of the smokers assume that mainly corresponds to resuspension. In Schneider et al.
in each NC was multiplied with the one cigarette particulate emission (1999) the resuspension rate was calculated as a function of the aerody-
namic diameter and two constants a (=1.8) and b (=7.2 × 10− 10 s− 1).
From this relationship, the ratio of the resuspension production for the
PM1 and PM2.5 − 1 fraction of the atmospheric aerosol was estimated to
be 15.43. Then consecutive numerical experiments with varying
ventilation and resuspension production rates were performed, until
the next hour indoor concentrations calculated by the model were equal
to the concentrations measured during the second set of measurements.
During the numerical experiments, the following processes were taken
into account: transport from outdoors to the indoor environment,

Fig. 3. Box plot of PM1 out. Fig. 5. Box plot of PM1 in.
4898 C. Halios et al. / Science of the Total Environment 407 (2009) 4894–4901

and PM2,5 − 1 fraction of the atmospheric aerosol. The average produc-


tion rate with respect to PM2.5 obtained during these theoretical
experiments was 9.3 mg min− 1. These values are very high bearing in
mind that the average production rates from the cigarettes smoking
were calculated to be 2.5 mg min− 1. The cigarette production rates
were calculated with the methodology described in Section 3.2 and
concern PM1 particles.

3.2.3. Sensitivity to the ventilation


In order to further investigate the effect of the ventilation on the
occupants' exposures to particulate concentrations, two sets of
numerical evaluations were performed with the MIAQ model, using
the same input values for the outdoor concentrations that were used
during the numerical simulations described in Section 3, but under
different ventilation rates. During the first scenario the values of the
ventilation rates were equal to the minimum values required by the
national standards (51 m3 h− 1 per person). The initial indoor
concentrations used during the simulations were again the PM1 and
PM2.5 concentrations measured during the first set of the scheduled
measurements. The ventilation rates during these numerical simula-
tions and the obtained indoor concentrations after 1 h, are presented
in Table 4. It is clear that even though the PM1 and PM2.5 indoor
concentrations are still higher than the outdoor values, they are
substantially reduced comparing with the measured values. Thus, the
Fig. 6. Box plot of PM2.5 in.
new average PM1 concentrations is 50.8 μg m− 3 and the maximum
value is 72.1 μg m− 3, while the measured values were 181,8 μg m− 3 h,
transport from the indoor to the outdoor environment, deposition on and 394.5 μg m− 3 respectively.
indoor surfaces, coagulation and direct production. Deposition calcula- During the theoretical simulations of the second scenario, the
tions are performed by the model assuming a laminar flow. Details are ventilation rates were set to be equal with the average required by the
given in Nazaroff and Cass (1989). The model calculates coagulation in national standards (85 m3 h− 1 per person). The ventilation rates in
two stages: initially the collision frequency between two particles is each club during these simulations and the calculated concentrations,
determined and then these probabilities are integrated in order to obtain are presented in Table 4. It can be seen that these concentrations were
the growth and loss rates. (Nazaroff and Cass, 1989). further reduced, but still quite higher from the outdoor concentrations,
indicating that the ventilation rates are not quite sufficient for the total
removal of the indoor polluted air to the outdoor environment.
3.2.1. Ventilation rates
Ventilation rates are usually measured by means of the tracer gas
decay method and/or the blower door decay method (Sfakianaki et 3.3. Statistical analysis
al., 2007), but they could not be directly measured in the NCs due to
the high occupancy. In order to indirectly estimate the ventilation In order to assess the relative importance of the various processes
rates in the different NCs, numerical experiments using the MIAQ that influence the occupants' exposure to fine particulates within the
numerical model were performed. NCs, we performed the statistical analysis that follows. Initially the
In Table 3 the estimated production rates due to cigarette smoking Pearson correlation coefficient between the indoor PM1 and PM2.5
(with respect to PM1), the production rates due to resuspension concentrations and selected variables that were considered of
(referring to PM1 and PM2.5 fraction) and the obtained ventilation rates importance were calculated. The selected variables were:
at each NC are presented. It is apparent that the calculated ventilation
rates are very low when comparing with the minimum ventilation – the measured outdoor PM1 and PM2.5 concentrations that contribute
rates indicated from the national standards (51 m3 h− 1, Technical to indoor exposure by entering from the outdoor environment
Chamber of Greece), with the average value to be 13.0 m3 h− 1 per indoors through the transport process
person and the maximum ventilation rate to be 22.1 m3 h− 1 per person – the number of occupants as were recorded during the experiments,
in NC8. that potentially can contribute to indoor exposure with various
activities like resuspension of the deposited particulate matter
3.2.2. Production from resuspension – the percentage of the smokers (% smokers) within each NC,
During the theoretical simulations, the production due to – the ratio of the indoor surfaces to the total volume of each NC (A/V
resuspension was estimated and is presented in Table 3 for the PM1 ratio), which may be an indicator of the deposition rate. The

Table 3
Input values used during the numerical experiments and obtained results.

NC1 NC2 NC3 NC4 NC5 NC6 NC7 NC8 NC9 NC10
λ (h− 1) 7.3 1.5 1.8 0.5 1.7 4.9 0.7 2.5 2.8 2.6
λ (m3 h− 1 per person) 14.8 8.2 9.2 4.1 12.8 13.3 8.2 22.1 17.8 19.0
P (μg min− 1) 5034 4348 1831 3661 2441 2380 3204 915 824 778
R (PM1) (μg min− 1) 1814 810 286 175 239 259 486 226 64 32
R (PM2.5 − 1) (μg min− 1) 29200 12500 4400 1600 3700 4000 7500 3500 990 500

λ is the calculated from the numerical experiments ventilation rates, P is the estimated mass emission rates due to cigarette smoking. R (PM1) is the production due to resuspension
for the PM1 fraction and R (PM2.5) is the production due to resuspension for the PM2.5 fraction of the atmospheric aerosol.
C. Halios et al. / Science of the Total Environment 407 (2009) 4894–4901 4899

Table 4 Table 6
Input values used during the numerical experiments for the calculation of indoor Multiple regression analysis regarding the occupants' exposure to PM1 concentrations.
concentrations under varying ventilation and predicted PM1 and PM2.5 indoor
concentrations after 1 h. Beta Std error B Std error t(13) p-level
of beta of B
NC1 NC2 NC3 NC4 NC5 NC6 NC7 NC8 NC9 NC10 Intercept 117.248 56.75 2.06 0.05
λ (h− 1) PM1 out 0.12 0.14 0.25 0.30 0.83 0.42
Scenario 1 29.1 9.5 9.1 5.6 6.6 20.4 4.5 5.4 8.2 7.1 % smokers 0.50⁎ 0.16⁎ 1.51⁎ 0.49⁎ 3.07⁎ 0.01
Scenario 2 48.6 15.8 15.2 9.3 11.1 34.0 7.5 9.1 13.6 11.8 A/V − 0.62⁎ 0.20⁎ −117.98⁎ 38.40⁎ −3.07⁎ 0.01
Ventilation rate 0.14 0.18 3.06 3.74 0.82 0.43
PM1 (μg m− 3)
⁎Marked correlations are significant at p b 0.05. R = 0.87, R2 = 0.76, Adjusted R2 = 0.69.
Scenario 1 58.6 68.6 48.4 50.1 72.1 44.8 59.2 41.1 29.4 36.3
Scenario 2 38.1 50.0 34.8 33.0 47.9 30.7 41.4 27.8 21.1 28.7

PM2.5 μg m− 3
Scenario 1 279.6 201.6 133.7 78.3 192.1 106.6 185.2 186.1 69.0 56.7 model is R2, sometimes called the coefficient of determination which
Scenario 2 173.1 138.8 92.4 52.2 135.8 71.7 133.2 136.8 52.3 44.8 is the proportion of the total variability in the dependent variable that
λ is the ventilation rate values. PM1 and PM2.5 are the PM1 and PM2.5 concentrations
is accounted for by the regression equation (Turalioglu et al., 2005).
respectively. For the purpose of the multiple regression analysis only the most
dominant factors were selected. Thus, regarding the PM1 concentra-
tions dependent variables were the outdoor PM1 concentrations, the
deposition rate is given by the product of the deposition velocity vd percentage of the smokers in the NCs, the A/V ratios and the
with the A/V ratio (Halios and Helmis, 2007) ventilation rates, while for the PM2.5 concentrations the corresponding
– the ventilation rate, as was calculated during the theoretical dependent variables were outdoor PM2.5 concentrations, the occu-
simulations that were described in 2.1 pants in each NC, the percentage of smokers, the A/V ratios and the
ventilation rates. Multiple regression analysis was obtained using
The calculated correlation coefficients are given in Table 5. The
Statistica 7.0 (Statsoft, Tulsa, OK, USA). The results of the multiple
significance of smoking to the indoor occupants' exposure to PM1
regressions concerning the PM1 and PM2.5 exposures are presented in
concentrations is clearly depicted (R = 0.71). The number of occu-
Table 6 and 7 respectively. The B-values correspond to the b-
pants in each NC is correlated mainly with the PM2.5 and PM2.5 − 1
coefficients of regression in Eq. (1). The Beta values are standardized
concentrations (R = 0.82 and 0.81 respectively) rather than the PM1
(adjusted) coefficients which result from the B-values after extracting
concentrations (R = 0.56). On the other hand it seems that the A/V
the average value and divide with the standard deviation of each
ratio contributes mainly to the reduction of the PM1 rather than the
variable (xi of Eq. (1)) and allow to compare the relative contribution of
PM2.5 concentrations (R = − 0.75 and −0.44 for the PM1 and PM2.5
each independent variable in the prediction of the dependent variable.
concentrations respectively). This may be attributed to the fact that
The results indicate that the multiple regression model that was
the effect of the deposition is analogous to the A/V ratio (the
obtained, explain a portion of about 70% of the observed variation in
deposition rate equals with the product of the deposition velocity and
the indoor PM1 concentrations and 75% of the observed variation in the
the A/V ratio (Nazaroff and Cass, 1986)) contributing this way to the
indoor PM2.5 concentrations. From Table 6 it can be clearly seen that
reduction of the indoor concentrations. The outdoor PM1 and PM2.5
the most significant variable for the occupants' exposure to PM1
seems to have negligible effect on the average indoor occupants'
concentrations is the percentage of smokers within the NCs (Beta value
exposures to PM1 and PM2.5, apparently due to the low ventilation
0.50), as well as the A/V ratio (Beta value −0.62). This may be
rates. The effect of the particulate transport from the indoor environ-
attributed to the fact that the majority of the smoking produced
ment to the outdoors is of significant importance (R = −0.74 and −0.65
particles are smaller than 0.5 μm. The most significant factor for the
for occupants' exposures to PM1 and PM2.5 respectively).
occupants' exposure to PM2.5 concentrations, as resulted from Table 7
In order to further study the relative importance of the various
is the occupants within the NCs (Beta value 0.94). It is of interest to
processes on the occupants' exposures to particulate pollution and to
notice that the effect of ventilation seems to be negligible for the
confirm the results so far obtained, multiple linear regression analysis
exposures to PM1 concentrations (Beta values 0.14), while is of some
was performed. Regression analysis is usually used to find the relation-
relative significance for the exposures to PM2.5 concentrations. This
ship between variables and if the variables are more than one, multiple
may be attributed to the fact that the calculated ventilation rates are
linear regression analysis is used and a general regression equation can
low in all NCs, thus the effect of the other parameters (with respect to
be expressed as
the PM1 exposures) is more important. On the other hand, the effect of
the deposition, (which is here examined through the A/V ratio) is of
Y = a + b1 x1 + b2 x2 + b3 x3 + :::bi xi ð1Þ significance for the exposure mainly to PM1 rather than the PM2.5
concentrations (Beta values − 0.62 and − 0.22 respectively) in
accordance with the results concluded from the analysis from the
where a is the constant of regression and b are the coefficients of
correlation coefficients.
regression. A generally used measure of the goodness of fit of a linear

Table 7
Table 5 Multiple regression analysis regarding the occupants' exposure to PM2.5 concentrations.
Correlation coefficients (R) between the average calculated occupants'' exposures to
PM1 and PM2.5 concentrations and selected variables described in the text. Beta Std error B Std error t(13) p-level
of beta of B
a
PM1 out PM2.5 out % smokers # persons A/V Ventilation Intercept 189.83 159.29 1.26 0.23
PM1 0.01 −0.12 0.71⁎ 0.56⁎ − 0.75⁎ − 0.74⁎ PM2.5 out 0.00 0.14 0.02 0.59 0.03 0.98
PM2.5 −0.21 −0.09 0.64⁎ 0.82⁎ −0.44 − 0.65⁎ # persons 0.94⁎ 0.18⁎ 3.14⁎ 0.62⁎ 5.09⁎ 0.00
PM2.5−1 −0.28 −0.10 0.48⁎ 0.81⁎ −0.10 − 0.40 % smokers −0.12 0.20 −1.07 1.67 −0.64 0.54
A/V −0.22 0.18 −116.13 96.5 −1.20 0.25
⁎Marked correlations are significant at p b 0.05. Number of cases N = 19 (casewise
Ventilation rate −0.25 0.19 −15.13 11.12 −1.36 0.20
deletion of missing data).
a
The ventilation reported here is normalized per occupant (m3 h− 1 per person). ⁎Marked correlations are significant at p b 0.05. R = 0.90 R2 = 0.82, Adjusted R2 = 0.75.
4900 C. Halios et al. / Science of the Total Environment 407 (2009) 4894–4901

4. Discussion and conclusions aerosol would be other than the value calculated here. For example, if
the actual values of the two constants were 20% higher than the values
The results of this study showed the presence of elevated levels of used for our calculations, then the ratio of the resuspension
particulate exposure in nightclubs. Recently, Valente et al. (2007) production terms for the PM1 and PM2.5 − 1 would be 26.6 instead of
reported similar values of fine particles (PM2.5) in pubs where 15.54.
smoking was allowed, in Italy, before the smoking ban (average By performing univariate and multivariate statistics to our data, we
concentration 368.1 μg m− 3). Repace et al. (2006) reported lower developed linear regression models in order to quantitatively examine
concentrations of fine particles (PM3.5) during a monitoring study in the relative contribution of the various mechanisms that control
seven Boston pubs before the smoking ban (average concentration occupants' exposures to fine particulate concentrations. Our results
179 μg m− 3), but the percentage of the smokers was significantly showed that 40% the occupants' exposures to PM1 concentrations is
small (11.65% in their study, 57% in the present study). Gee et al. controlled by the smoking activity, 40% by the process of deposition,
(2005) reported PM3.5 concentrations of the order of 115 μg m− 3 from while the ventilation process and the outdoor concentrations
a monitoring study conducted in 60 pubs and bars in Manchester, but contribute only 10% each. The density of occupancy within each
in their study different types of ventilation were available. The nightclub controls 60% of the exposure to PM2.5 concentrations.
exceptionally high concentrations found in the present study (average In the absence of smoking ban, the significance of ventilation is
PM1 and PM2.5 concentrations 181.8 μg m− 3 and 454 μg m− 3 can be crucial. By performing theoretical calculations it was apparent that if the
attributed mainly to the elevated presence of smokers, in conjunction ventilation rates were at the minimum values required by the national
with the very low ventilation rates, which were found to be 25% lower standard, then the average occupants' exposures to the indoor
than the ventilation required by the national standards. particulate concentrations would be reduced at the 70% of the present
The results presented in Section 3.2.2 showed high values of the exposure values. This reduction would be at the 80% of the present
resuspension rates of the PM1 and PM2.5 within the different NCs. values if the ventilation rates were at the values that are recommended
Resuspension of fine particles are reported in Thatcher and Layton by the national standards.
(1995), Gomes et al. (2007), Klinmalee et al. (2008), Ferro et al. On the overall, in the present work we found that the exposure to
(2004a,b), and Qian and Ferro (2008). Also, Long et al. (2001) particulate matter in ten nightclubs in Athens, Greece is exceptionally
reported evidence for PM2.5 resuspension during walking. The high when comparing with values reported in other studies
resuspension values obtained during our study are consistent with performed in similar indoor environments. By quantitatively examin-
the results reported by Ferro et al. (2004a,b) where the average PM2.5 ing the relative effect of the different processes to the exposure levels,
production rate due to resuspension was 450 μg min− 1 for two we found that the significance of the resuspension was most
persons walking in a room and 200 μg min− 1 for one person dancing prominent. An important output of the study was that given the
on rag. Data from Ferro et al. (2004a,b) were obtained during a study absence of the smoking ban, the increase of the ventilation rates up to
in a residence. Data regarding resuspension in microenvironments the levels set by national standards, would substantially reduce the
similar to the ones examined during our study are quite sparse in the occupants' exposure to particulate matter.
international literature. In two studies where resuspension was
estimated alongside with smoking (Ott et al., 1996; Repace and References
Lowry, 1982) it was estimated that an average of about 83–87% of the
Arundel A, Sterling T, Weinkam J. Never smoke lung cancer risks from exposure to
respirable suspended particles in taverns and dinner theater are particulate tobacco smoke. Environ Int 1987;13(6):409–26.
produced from smoking, while the remaining 13–17% is produced Ballesta PP, Field RA, Connolly R, Cao N, Baeza Caracena A, De Saeger E. Population
from cooking and resuspension. These results are quite different from exposure to benzene: one day cross-sections in six European cities. Atmos Environ
2006;40(18):3355–66.
the results obtained here, where 78% or the total emitted particulate Brunekreef B, Dockery DW, Krzyzanowski M. Epidemiologic studies on short-term
mass is attributed to resuspension and only 21% is produced from effects of low levels of major ambient air pollution components. Environ Health
smoking. The observed differences could be mainly attributed to the Persp 1995;103(Suppl 2):3–13.
Chan TL, Lippmann M. Experimental measurements and empirical modelling of the
enhanced mobility of the occupants in the nightclubs (i.e. moving,
regional deposition of inhaled particles in humans. Am Ind Hyg Assoc J 1980;41:
dancing) in contrast with the lower mobility in the taverns. 399–409.
The results presented in Section 3.3 depicted the significance of Chang Li-Te, Suh HH, Wolfson JM, Misra K, Allen GA, Catalano PJ, Koutrakis P. Laboratory
and field evaluation of measurement methods for one-hour exposures to 03, PM
deposition on the exposure levels in the NCs. The model calculations
2.5, and CO. J Air Waste Manage 2001;51:1414–22.
of the deposition were performed considering a laminar flow Chang PT, Peters LK, Ueno Y. Particle size distribution of mainstream cigarette smoke
condition. Bearing in mind the enhanced mobility of the people undergoing dilution. Aerosol Sci Tech 1985;4:191–207.
within the NCs, leading to a turbulent flow field, it is probable that the Chung I, Dunn-Rankin D. In situ light scattering measurements of mainstream and
sidestream cigarette smoke. Aerosol Sci Tech 1996;24(2):85–101.
deposition values are underestimations, leading to higher values of Cleaver JW, Yates B. Mechanism of detachment of colloidal particles from a flat substrate
the resuspension rates than the values calculated here. Detailed in a turbulent flow. J Colloid Interface Sci 1973;44(3):464–74.
measurements of the realistic flow conditions within NCs during Drakou G, Zerefos C, Ziomas I, Voyatzaki M. Measurements and numerical simulations
of indoor O-3 and NOx in two different cases. Atmos Environ 1998;32(4):595–610.
controlled experiments are needed in order to correctly assess the Ferro A, Kopperud RJ, Hildemann L. Source strengths for indoor human activities that
deposition effect. resuspend particulate matter. Environ Sci Technol 2004a;38:1759–64.
As discussed in Section 3.2, during the estimation of the Ferro AR, Kopperud RJ, Hildemann LM. Elevated personal exposure to particulate matter
from human activities in a residence. J Exp Anal Env Epid 2004b;14(SUPPL. 1):
resuspension production term, we employed the relationship pre- S34–40.
sented in Schneider et al. (1999) according to which the resuspension Gee IL, Watson AFR, Carrington J. The contribution of environmental tobacco smoke to
rate is calculated as a function of the particle's aerodynamic diameter indoor pollution in pubs and bars. Indoor Built Environ 2005;14(3–4):301–6.
Georgalas B, Grivas G, Chakoulakou A. Comparison between a direct reading real time
and two constants. This relationship was obtained by making a linear
aerosol monitor and a gravimetric instrument for PM1 measurements in the
regression of the log-transformed values of the data presented in atmosphere of Athens, Greece. 7th International Aerosol Conference, MN, USA;
Thatcher and Layton (1995). As this data concern particles less than 2006. p. 481–2.
Gomes C, Freihaut J, Bahnfleth W. Resuspension of allergen-containing particles under
20 μm, the calculated relationship is extrapolated for particles with
mechanical and aerodynamic disturbances from human walking. Atmos Environ
aerodynamic diameter greater 20 μm, and thus could be very 2007;41:5257–70.
uncertain. More specifically, Schneider et al. (1999) report that in Goodman P, Agnew M, McCaffrey M, Paul G, Clancy L. Effects of the Irish smoking ban on
rooms with high activity and no carpets, the resuspension rates could respiratory health of bar workers and air quality in Dublin pubs. Am J Resp Crit Care
2007;175(8):840–5.
be much larger. In such a case, the ratio of the resuspension Halios CH, Helmis CG. On the estimation of characteristic indoor air quality parameters
production terms for the PM1 and PM2.5 − 1 fraction of the atmospheric using analytical and numerical methods. Sci Total Environ 2007;381:222–32.
C. Halios et al. / Science of the Total Environment 407 (2009) 4894–4901 4901

Harrison RM, Yin J. Particulate matter in the atmosphere: which particle properties are Ramachandran G, Adgate JL, Hill N, Sexton K, Pratt G, Bock D. Comparison of short term
important for its effects on health? Sci Total Environ 2000;249:85–101. variations (15-minute averages) in outdoor and indoor PM2.5 concentrations. J Air
Hitchins J, Morawska L, Wolff R, Gilbert D. Concentrations of submicrometre particles Waste Manage 2000;50:1157–66.
from vehicle emissions near a major road. Atmos Environ 2000;34:51–9. Repace JL, Lowrey AH. Tobacco smoke, ventilation, and indoor air quality. ASHRAE
Hyvarinen M, Johnsson T, Tuomi T, Mielo T, Reijula K. Control of exposure to Transactions 1982;88:895–914.
environmental tobacco smoke in restaurants and bars. Proceedings of Indoor Air Repace JL, Hyde JN, Brugge D. Air pollution in Boston bars before and after a smoking
2002. 9th International Conference on Indoor Air Quality and Climate. Montery, ban. BMC Public Health 2006. doi:10.1186/1471-2458-6-266.
California 30th June 5th July 2002: 150–55. Repace J. Respirable particles and carcinogens in the air of Delaware hospitality venues
Johnsson T, Tuomi T, Riuttala H, Hyvarinen M, Rothberg M, Reijula K. Environmental before and after a smoking ban. J Occup Environ Med 2004;46(9):887–905.
tobacco smoke in Finnish restaurants and bars before and after smoking restrictions Santamouris M, Argiroudis K, Georgiou M, Livada I, Doukas P, Assimakopoulos MN, et al.
were introduced. Ann Occup Hyg 2006;50(4):331–41. Indoor air quality in fifty residences in Athens. Int J Ventilation 2007;5(4):367–80.
Keith CH, Derrick JC. Measurement of the particle size distribution and concentration of Schneider T, Kildeso J, Breum NO. A two compartment model for determining the
cigarette smoke by the “conifuge”. J Colloid Sci 1960;15:340–56. contribution of sources surface deposition and resuspension to air and surface dust
Klinmalee A, Srimongkol K, Kim Oanh NT. Indoor air pollution levels in public buildings concentration levels in occupied rooms. Build Environ 1999;34:583–95.
in Thailand and exposure assessment. Environ Monit Assess 2008. doi:10.1007/ Schwartz J, Dockery DW, Neas LM. Is daily mortality associated specifically with fine
s10661-008-0507-z. particles? J Air Waste Mana 1996;46(10):927–39.
Klepeis NE, Apte MG, Gundel LA, Sextro RG, Nazaroff WW. Determining size-specific Sfakianaki A, Pavlou K, Santamouris M, Livada I, Assimakopoulos MN, Mantas P, et al. Air
emission factors for environmental tobacco smoke particles. Aerosol Ssi Tech tightness measurements of residential houses in Athens, Greece. Build Environ
2003;37:780–90. 2007. doi:10.1016/j.buildenv.2007.01.006.
Kurmi OP, Semple S, Steiner M, Henderson GD, Ayres JG. Particulate matter exposure Thatcher TL, Layton DW. Deposition, resuspension and penetration of particles within a
during domestic work in Nepal. Ann Occup Hyg 2008;52(6):509–17. residence. Atmos Environ 1995;29(13):1487–97.
Lee SC, Chang M, Chan KY. Indoor and outdoor air quality investigation at six residential Tominz R, Poropat C, Bovenzi M. Changes in PM10 and PM2.5 air levels in bars after the
buildings in Hong Kong. Environ Int 1999;25:489–96. enforcement of the smoking ban in the Italian legislation. Epidemiologia e
Lee SC, Li Wai Ming, Chan LY. Indoor air quality at restaurants with different styles of prevenzione 2006;30(6):325–33.
cooking in metropolitan Hong Kong. Sci Total Environ 2001;279:181–93. Technical Chamber of Greece. Building facilities: Components for the calculation of air
Leutwyler M, Siegmann K, Monn C. Suspended particulate matter in railway coaches”. conditioning loads in buildings. Technical Chamber of Greece (Technical Directive,
Atmos Environ 2002;36:1–7. T.O.T.E.E., 2425/86).
Long CM, Suh JH, Catalano PJ, Koutrakis P. Using time- and size-resolved particulate data Turalioglu FS, Nuhoglu A, Bayraktar H. Impacts of some meteorological parameters on
to quantify indoor penetration and deposition behavior. Environ. Sci. Technol SO2 and TSP concentrations in Erzurum, Turkey. Chemosphere 2005;59:1633–42.
2001;35:2089–99. Ueno Y, Peters LK. Size and generation rate of sidestream cigarette smoke particles.
McDonnell WF, Nishino-Ishikawa N, Petersen FF, Chen LH, Abbey DE. Relationships of Aerosol Sci Tech 1986;5:469–76.
mortality with the fine and coarse fractions of long-term ambient PM10 Valente P, Forastiere F, Bacosi A, Cattani G, Di Carlo S, Ferri M, et al. Exposure to fine and
concentrations in nonsmokers. J Expo Anal Env Epid 2000;10(5):427–36. ultrafine particles from secondhand smoke in public places before and after the
Mitsakou C, Helmis C, Housiadas C. Eulerian modelling of lung deposition with sectional smoking ban, Italy, 2005. Tobacco Control 2007;16:312–7.
representation of aerosol dynamics. J Aerosol Sci 2005;36:75–94. Waring MS, Siegel JA. An evaluation of the indoor air quality in bars before and after a
Nazaroff WW, Cass GR. Mathematical modelling of chemically reactive pollutants in smoking ban in Austin, Texas. J Expo Sci Env Epid 2007;17(3):260–8.
indoor air. Environ Sci Tech 1986;20(9):24–934. WHO Regional Publications. European Series, No. 91, 2000 Air Quality Guidelines for
Nazaroff WW, Cass GR. Mathematical modeling of indoor aerosol dynamics. Environ Sci Europe. second edition. Copenhagen: World Health Organization Regional Office
Tech 1989;24:66–7. for Europe; 2000.
Ott W, Switzer P, Robinson J. Particle concentrations inside a tavern before and after Wilson N, Edwards R, Maher A, Nathe J, Jalali R. National smokefree law in New Zealand
prohibition of smoking: evaluating the performance of an indoor air quality model. improves air quality inside bars, pubs and restaurants. BMC Public Health 2007.
J Air Waste Manage 1996;46:1120–34. doi:10.1186/1471-2458-7-85.
Pope CA, Hill RW, Villegas GM. Particulate air pollution and daily mortality on Utah's Yanosky JD, Williams PL, MacIntosh DL. A comparison of two direct-reading aerosol
Wasatch Front. Environ Health Persp 2000;107(7):567–73. monitors with the federal reference method for PM2.5 in indoor air. Atnos Environ
Qian J, Ferro AR. Resuspension of dust particles in a chamber and associated 2002;36:107–13.
environmental factors. Aerosol Sci Tech 2008;42:566–78.

You might also like