Inhalation Toxicology, 2010, 1–9, Early Online
ORIGINAL ARTICLE
Direct measurement of toxicants inhaled by water pipe
users in the natural environment using a real-time in situ
sampling technique
Inhalation Toxicology Downloaded from informahealthcare.com by 124.197.58.145 on 11/09/10
For personal use only.
M. Katurji, N. Daher, H. Sheheitli, R. Saleh, and A. Shihadeh
Aerosol Research Laboratory, Mechanical Engineering Department, American University of Beirut, Beirut, Lebanon
Abstract
While narghile water pipe smoking has become a global phenomenon, knowledge regarding its toxicant content
and delivery, addictive properties, and health consequences is sorely lagging. One challenge in measuring toxicant content of the smoke in the laboratory is the large number of simplifying assumptions that must be made to
model a “typical” smoking session using a smoking machine, resulting in uncertainty over the obtained toxicant
yields. In this study, we develop an alternative approach in which smoke generated by a human water pipe user
is sampled directly during the smoking session. The method, dubbed real-time in situ sampling (RINS), required
developing a self-powered portable instrument capable of automatically sampling a fixed fraction of the smoke
generated by the user. Instrument performance was validated in the laboratory, and the instrument was deployed
in a field study involving 43 ad libitum water pipe use sessions in Beirut area cafés in which we measured inhaled
nicotine, carbon monoxide (CO), and water pipe ma’ssel-derived “tar.” We found that users drew a mean of 119 L of
smoke containing 150 mg of CO, 4 mg of nicotine, and 602 mg of ma’ssel-derived “tar” during a single use session
(mean duration = 61 min). These first direct measurements of toxicant delivery demonstrate that ordinary water
pipe use involves inhaling large quantities of CO, nicotine, and dry particulate matter. Results are compared with
those obtained using the Beirut method smoking machine protocol.
Keywords: Particulate matter; personal exposure; hookah; shisha; tobacco smoke; CO; nicotine
Introduction
While over the past decades, a large evidence base has been
built about the constituents, addictive properties, and deleterious health effects of cigarette smoke, comparatively little is
known about the nature and health consequences of smoke
produced using the narghile water pipe (also known as “shisha,” “hookah”; see Figure 1). This knowledge gap has become
particularly salient in the past decade with a rapid and global
rise in water pipe prevalence (Baska et al., 2008; Eissenberg
et al., 2008; El Roueiheb et al., 2008; Jawaid et al., 2008; Pärna
et al., 2008; Primack et al., 2009, 2010; Cobb et al., 2010), possibly ushered in by the widespread commercial introduction
and modern marketing of prepackaged units in the 1990s of a
tobacco-based product known as ma’ssel (Maziak et al., 2004).
Ma’ssel is a sweetened, flavored mixture with ∼25% tobacco
content (Rees et al., 2007) sold specifically for consumption
in water pipes. As it is smoked, it releases an aroma similar to
caramelizing sugar in addition to the banana, apple, cherry,
cola, vanilla, or other essences added to flavor the product.
A weak regulatory environment has allowed a burgeoning
industry to continually introduce new ma’ssel products (e.g.
see Khalil et al., 2009) at a rate that far outpaces advancement
in the science. Basic questions regarding the constituents of
the smoke, far less the long-term health consequences of its
inhalation, remain open. One open question in this regard
is “what is an appropriate method to estimate inhaled water
pipe smoke toxicants?” Methods for measuring cigarette
smoke toxicant content commonly rely on machines to smoke
cigarettes under prescribed regimens that reflect to varying
extents smoker puffing behavior. While smoking machine
methodology is used for product compliance testing and
for generating smoke for detailed chemical and biological
assays, its use for estimating toxicant intake by real smokers
has been questioned (e.g. Kozlowski et al., 1980, 1982a, b;
Address for Correspondence: Alan Shihadeh, Aerosol Research Laboratory, Mechanical Engineering Department, American University of Beirut, P.O. Box No. 11070236, Riad El-Solh, Beirut 1107-2020, Lebanon. E-mail: as20@aub.edu.lb
(Received 23 July 2010; revised 12 September 2010; accepted 13 September 2010)
ISSN 0895-8378 print/ISSN 1091-7691 online © 2010 Informa Healthcare USA, Inc.
DOI: 10.3109/08958378.2010.524265
http://www.informahealthcare.com/iht
2 M. Katurji et al.
mouthpiece
tobacco
charcoal
head
hose
Inhalation Toxicology Downloaded from informahealthcare.com by 124.197.58.145 on 11/09/10
For personal use only.
body
bowl
water
Figure 1. Schematic of a narghile water pipe. The head, body, water bowl,
and hose are the primary elements from which the water pipe is assembled. Tobacco is loaded into the head, and burning charcoal is placed on
top of the tobacco. When a user inhales from the mouthpiece, air and hot
charcoal fumes are forced through the tobacco, raising its temperature,
and generating the desired smoke. The smoke exits from the bottom of the
head, into the body, and through the bubbler and hose to the user. The
water pipe illustrated here, and used in this study, is configured for use
with sweetened and flavored tobacco, known as ma’ssel. When ma’ssel is
used, a relatively deep (∼3 cm) head is filled with 10–20 g of the flavored
tobacco mixture and covered with an aluminum foil sheet that is perforated for air passage. Burning charcoal is placed on top of the aluminum
foil to provide the heat needed to generate the smoke (reprinted from
Daher et al., 2010).
Hammond et al., 2006; Burns et al., 2008; Marian et al., 2009).
Among other reasons, idiosyncrasies in user behavior make
representing the smoking session with a machine difficult
(e.g. repeated lighting of a pipe or cigar, varying ventilation
blocking by lips or fingers, adjusting charcoal on a water pipe
head). In addition, reproducing the smoking device itself is
challenging when nonstandard tobacco products are used
(e.g. roll-your-own cigarettes, pipes). With the water pipe,
variability is endemic; the amount of tobacco and charcoal
loaded, the density of the packed tobacco, the number of
users sharing it, the degree to which fresh air infiltrates into
the hose (Saleh and Shihadeh, 2008), and other variables
can affect toxicant yields and are difficult to characterize and
standardize for the laboratory.
One accepted alternative to smoking machine methods
has been chemical analysis of spent cigarette butts (e.g.
Kozlowski et al., 1982a, b; Green et al., 1985; Watson et al.,
2004; O’Connor et al., 2007; see Pauly et al., 2009). This
indirect method is based on correlation between quantities of toxicants trapped in the spent cigarette butt and
quantities delivered during smoking, and the method has
many practical advantages for standardized cigarettes, ecological validity foremost among them. Some disadvantages
include inability to measure gas-phase toxicants such as
CO and volatile aldehydes and imprecision in estimates of
toxicant trapping efficiency of the cigarette butt.
Another alternative is to directly sample humangenerated smoke in real time. In this study, we explore the
use of a sampling technique in which a small proportion
of each puff generated by a human is directly captured
for chemical analysis. During each puff, a smoke sample
is drawn through a glass fiber filter and exhausted into a
sealed inert bag for off-line chemical analysis of the particle- and gas phases. While development of such a real-time
in situ sampling (RINS) approach was motivated by difficulty accounting for water pipe smoker idiosyncrasies in
the analytical laboratory, RINS is in principle applicable to
any smoking device. Its principle advantage is that it makes
possible the assessment of toxicant intake by real smokers
in their natural settings, with no need for idealizations of the
smoker or the device. It also allows for simultaneous measurement of smoking topography.
This study reports the design and performance characterization of a RINS instrument as well as measurements
of smoking topography, CO, nicotine, and ma’ssel-derived
“tar” delivery of a sample of 43 narghile water pipe use sessions in Beirut area cafés. A comparison is made with previously measured toxicant yields obtained with a smoking
machine.
Materials and methods
Instrument description
A schematic of the RINS instrument connected to a water
pipe is given in Figure 2. RINS automatically samples a fraction of the smoke flowing through the mouthpiece whenever a smoker inhales from it. The sampled smoke is drawn
through a glass fiber filter by a digitally controlled miniature
diaphragm pump that exhausts into an inert bag. The inhaled
and sampled flow rates are monitored and recorded continuously by the RINS control software. At the end of each smoking session, the filter pad and contents of the bag are analyzed
to assess toxicant exposure, and smoker topography (e.g.
number of puffs, puff volume, duration, interpuff interval;
see Shihadeh et al., 2004) is recorded. An important feature
of this system is its portability. Except for a laptop computer
and the modified water pipe hose, the entire setup, including
a battery power supply, is contained in a small laptop computer case and is convenient to setup wherever water pipes
are normally used.
Key variables in the operation are the mainstream smoke
flow rate drawn by the smoker in the mouthpiece, Q, and
the minor flow drawn by the instrument just upstream of
the mouthpiece, q. The fraction sampled, f, is defined as the
ratio q/Q. Water pipe users generate and inhale a smoke
stream characterized by a time-varying flow rate and chemical composition. To ensure that a correctly weighted sample
Inhaled waterpipe smoke toxicants 3
mouthpiece/
probe assembly
glass fiber filter
flexible
tube
Carrying case
∆P
obstruction flow
meter
pressure
transducer
∆P
look up
control
signal
∆V
pusation
damper
flow rate
Inhalation Toxicology Downloaded from informahealthcare.com by 124.197.58.145 on 11/09/10
For personal use only.
Tedlar bag
time
save smoker
and sample
flow rates
USB daq
control
voltage
op-amp
miniature
vacuum
pump
pulsation
damper
hot wire
mass flow
sensor
battery
power
supply
to laptop
computer
Figure 2. Schematic of the real-time in situ sampling (RINS) instrument connected to a water pipe. (See colour version of this figure online at www.
informahealthcare.com/iht)
is collected from this unsteady flow, the sampling flow rate
must continuously vary in direct proportion to the mainstream flow rate. For the purposes of the present study, a
nominal desired sample yield of 25 mg of total particulate
matter (TPM) requires a sampled fraction of ∼2%, given an
average expected mainstream smoke TPM yield of 1400 mg
per water pipe use session (Shihadeh and Saleh, 2005).
With a characteristic flow dimension of 1 cm in the water
pipe hose and f = 1.95%, the probe diameter for required for
isokinetic sampling (see Brockman, 2001) is calculated as
1.4 mm. Conservatively assuming a flow rate and particle size
of 20 lpm and 1 μm, respectively, this probe diameter yields
an inlet Stokes number of ∼0.01 indicating that particle inertia will not affect probe aspiration efficiencies (Brockman,
2001). Therefore, the instantaneous free stream and sampling probe velocities can differ significantly without particle
size sampling biases, meaning that f need not be restricted
to the value corresponding to isokinetic sampling for this
probe size. It is important, however, that f remains constant
and known.
To achieve this, the RINS controller continuously monitors
the flow rate in the hose using an obstruction meter previously developed to measure smoking topography (Shihadeh
et al., 2005). Based on the input f, the controller generates a
pump control voltage that yields the desired instantaneous
sampling flow rate, q(t) = f·Q(t). The actual resulting flow rate
generated by the sampling pump is measured by a 3 msec
response hot wire mass flow sensor located upstream of the
pump. The signals of the hotwire flow sensor and the hose
obstruction meter are stored as the smoking session proceeds, allowing for postprocessing of smoking topography
data as well as numerical integration of the sampling flow rate
signal to determine the total sampled volume. This algorithm,
coded in Labview©, is run at 10 Hz, and is interfaced via a
USB data acquisition and control card (NI USB-6008) whose
analog output signal is current-amplified by an external opamp and sent to the miniature sampling pump, as shown in
Figure 2.
A key element of the instrument is its mouthpiece/sampler assembly. As shown in Figure 2, the sampling probe
4 M. Katurji et al.
Inhalation Toxicology Downloaded from informahealthcare.com by 124.197.58.145 on 11/09/10
For personal use only.
and filter are mounted in a wye configuration with the
mouthpiece. The probe is made of seamless thin-walled stainless steel hypodermic tubing, while the holder is machined
from polycarbonate plastic. The probe is concentric with the
hose, and the inlet is located far enough upstream that the
flow is unaffected by the bend leading to the mouthpiece,
according to a 3-D flow computer simulation implemented in
Fluent©. The mouthpiece is made of wood and was part of an
original water pipe hose. The added weight of the assembly is
∼100 g. Incorporation of the filter holder into the mouthpiece
minimizes sample transport losses. Particle diffusion and settling in the probe section were calculated (Brockman, 2001)
over a range of particle sizes and were found to be negligible
for the relevant flow conditions.
Laboratory validation
The performance of the RINS was characterized by connecting
it to a water pipe that was smoked using a smoking machine
(Shihadeh and Azar, 2006). Ma’ssel-derived “tar,” TPM, and
displaced volume measured using RINS and the smoking
machine were compared over a range of programmed smoking conditions. Displaced volumes were also compared on
an individual puff-by-puff basis to check that a proportional
volume of each individual puff had indeed been sampled.
Twenty-one smoking sessions were executed, with average
flow rates varying from 9 to 15 lpm, puff durations from 2 to
5 sec, total number of puffs drawn from 50 to 182, and with f
set to the isokinetic sampling condition (1.95%).
For each of the smoking sessions, the water pipe was prepared using 10 g of “Two Apples” (Nakhla brand) tobacco
mixture and a single easy light charcoal briquette (Three
Kings, Holland) was lit and placed on top of the water pipe
head at the beginning of the smoking session. For sessions
that involved >105 puffs, a second half-briquette was placed
on the water pipe head at the 105th puff. Particulate matter
exiting the water pipe mouthpiece was collected on 47-mm
glass fiber filters (Pall, Type A/E), while that sampled by the
RINS instrument was collected on 25-mm glass fiber filters
(Pall, Type A/E). Ma’ssel-derived “tar” concentrations for the
mainstream and sample stream were determined by pre- and
postweighing the filters on a 0.1 mg balance, analyzing water
content using a modified Karl-Fischer titration method, and
dividing the dry masses (= total – water) by the measured
volumes drawn through the mouthpiece and sampler, respectively. Other details regarding the machine-smoking protocol
used in the study are given in Shihadeh and Saleh (2005).
Actual volume sampled through the probe, Vsampled = ∫ q (t ) dt
during machine smoking was compared with desired sample volume, calculated as Videal = f ∫ Q (t ) dt , where q(t) is the
instantaneous probe flow rate reported by the RINS hot wire
flow sensor and Q(t) is the flow rate reported by the smoking
machine.
Field study
A field study was conducted to measure inhaled ma’sselderived “tar”, nicotine, CO, and smoking topography parameters for water pipes smoked by human users in their natural
settings. We sampled 49 water pipe use sessions during the
period September 2006 through September 2007 in two
popular cafés in Beirut, Lebanon where narghile water pipes
and food are served. Operator error or equipment malfunction resulted in discarding the data of six of these sessions.
The primary sampling location, located on the Ras Beirut
coastal road, was chosen because of its convenient proximity
to the American University of Beirut and because the relaxed,
informal outdoor setting of the café lent itself to our ability to
conduct the study without disrupting business. In addition,
the café caters to a range of ages and socioeconomic statuses,
and is a popular destination for narghile café smoking with
a view of the sea.
To obtain a random sample of water pipe use sessions,
we asked the food servers to inform us whenever a ma’ssel
water pipe was ordered. Once notified, we approached the
café patron and described the nature of the study and the
equipment involved, and asked for permission to sample
her/his smoke. Sixty-five percent of those we approached
agreed.
The sampling instrument was then attached to the water
pipe, and we instructed the café patron to smoke “as usual”
and to summon us directly or through a food server whenever s/he finished smoking. We then withdrew to a distant
but within-view table in the café, where we read and ordered
food to blend into the environment and to ensure that
occupying a table was not an economic burden on the café.
Because the patron usually had company, and because of the
leisurely disposition of the study personnel as they ate and
read at a distant table, we believe that the study had minimal
impact on the manner in which the water pipe was smoked,
apart from potential interference caused by the feel of the
mouthpiece sampler itself.
At the end of the smoking session, we returned to the
participant’s table, analyzed the smoke vapors for CO content, and removed the particulate filter from the instrument.
The filter was placed in a numbered, airtight container and
wrapped in aluminum foil to prevent light from degrading
the sample. At the end of the sampling day, all the filter
samples were taken to the laboratory, weighed, and either
analyzed for moisture (in order to determine ma’ssel-derived
“tar” = TPM – moisture) using a modified Karl-Fischer technique (see Shihadeh and Saleh, 2005 for details) or stored
in a freezer at −4°C until they were chemically analyzed for
nicotine content. Filters analyzed for nicotine were extracted
in ethyl acetate and toluene and analyzed by GC-MS as
described in Shihadeh and Saleh (2005).
An electrochemical sensor was used to measure the carbon monoxide concentration in the sample bag, and the
inhaled CO was then calculated by multiplying the measured CO concentration by the measured volume of smoke
drawn.
Because the participants in the field study were anonymous and because their participation in the study could not
place them at any reasonable risk, the field study protocol
was deemed IRB exempt in accordance with US Code 45 CFR
46.101(b)(2).
Inhaled waterpipe smoke toxicants 5
Laboratory validation
Measurements comparing session-integrated mainstream
and sampler-drawn volumes and particulate matter concentrations for the 21 machine-smoking sessions described
above resulted in an average error in sampled volume of
∼0.1% and an average error in particulate mass concentration of 1.1%. Both sampled volume and sampled particulate
mass concentration showed high correlation (r > 0.97) with
the smoking machine measures.
Typical traces of smoking machine and sampling probe
flow rates are shown in Figure 3 for four consecutive
machine-drawn puffs. The sampler reproduces the patterns
generated by the machine in detail, but exhibits a small
time lag. Because of the small Stokes numbers, however,
the instantaneous difference in flow velocities between the
mainstream and probe resulting from the lagging response
are unimportant from a sampling efficiency perspective for
the bulk of each puff.
While minimal variation between sampler and mainstream flow velocities is desirable, the more important criterion for the current application is that each puff drawn by
the smoker be proportionally sampled. In Figure 4, the volume drawn through the probe is plotted against that drawn
through the mouthpiece on a puff-by-puff basis for a single
machine-smoking session that “played back” a previously
recorded 1-h smoking event by a 26-year-old male smoker in
a café (Shihadeh and Azar, 2006). As shown in the figure, the
sampled and mouthpiece puff volumes are linearly related
(r > 0.99), indicating the sampler’s proportional sampling
accuracy over a wide range of conditions. The measured
sampling fraction was f = 0.0194, an error of <1% relative to
the programmed value of 0.0195.
smoking machine
flow rate (arbitrary units)
Inhalation Toxicology Downloaded from informahealthcare.com by 124.197.58.145 on 11/09/10
For personal use only.
Results
0
5
10
REALTIME
15
20
time (sec)
25
30
35
Figure 3. Time traces for smoking machine and real-time in situ sampler
(RINS) flow rate (scaled by 1/f) when the RINS instrument is attached to
the machine-smoked water pipe. The machine was programmed to follow
pre-recorded realistic human puffing pattern.
Water pipe use session characteristics
Table 1 summarizes the characteristics of the use sessions
recorded in the field study. As shown, >80% of the water pipes
were smoked by single users, with the remainder smoked by
dyads. “Two Apples” flavored tobacco was the most common
flavor used, followed by grape, mint, and rose. More than
60% of the water pipes sampled were used by self-reported
water pipe-only smokers; the remainder was smoked by users
reporting regular cigarette consumption in addition to water
pipe use.
Puff topography
Table 2 summarizes the smoking topography data. The data
are broadly consistent with previous data we have gathered,
though it should be noted that the previous 171 puff standard
smoking session derived in Shihadeh et al. (2004) was based
on extrapolation from topography records of only the first
30 min of the smoking sessions sampled. The current data,
in contrast, are based on the entire water pipe use session,
regardless of its duration. It should also be noted that four of
the smokers we sampled (all singletons) drew >400 puffs in
0.045
0.04
y = 0.0194x
0.035
Sampled volume (L)
Data analysis
Mean and 95% confidence intervals were computed for all
measures. Confidence intervals were computed using twotailed Student’s t-distribution. Probability values below 0.05
were taken to indicate a statistically significant difference.
Pearson product correlation coefficients, r, were also tested
for significance using two-tailed t-distribution.
0.03
0.025
0.02
0.015
0.01
0.005
0
0
0.5
1
1.5
2
2.5
Smoking machine puff volume (L)
Figure 4. Real-time in situ sampling versus smoking machine-drawn puff
volume for a 1-h human-mimic smoking session. Each circle represents a
single puff event of a 182-puff human-mimic smoking session. (See colour
version of this figure online at www.informahealthcare.com/iht)
Table 1. Water pipe session and user characteristics for the café sampling
campaign of September 2006–September 2007.
Water pipe session and user characteristics
Number of singleton sessions
36
Number of dyad sessions
7
Tobacco flavor smoked (apple/grape/mint/rose)
22/12/5/4
Median age (interquartile range)
26 (23–33)
Male/female
34/16
Median water pipe use frequency (/month)
8
Percent users who use water pipe at least 20 times/month
30%
Median years of water pipe use (interquartile range)
6 (4–9)
A total of 49 smoking sessions were recorded. The total number of users
was greater than the number of smoking sessions because there were 7
dyads. No sessions had more than two persons sharing a narghile water
pipe.
6 M. Katurji et al.
Nicotine
Mean nicotine intake for 17 determinations was 4.82 ± 1.07 mg.
Nicotine intake increased with inhaled volume, though
the correlation was slightly short of 95% significance
(Figure 6; P < 0.064). Mean nicotine concentration was
0.034 ± 0.008 mg/L of smoke.
Water and ma’ssel-derived “tar”
Water content was determined for 11 filter samples and
found to be highly correlated to TPM (P < 0.0001), as
Table 2. Summary of smoking topography results (mean ± 95% confidence
interval).
Current study
Shihadeh et al. (2004)
Total smoking time
64.0 ± 8.4
61
(min)
Number of puffs
220 ± 34
171
Smoke drawn (L)
130 ± 21
91
Puff duration (sec)
2.83 ± 0.27
2.6
Puff volume (mL)
626 ± 88
530
Interpuff interval (sec)
17.4 ± 3.0
17
10
y = 0.028x
8
6
4
2
0
0
50
100
150
200
250
300
Inhaled volume (L)
Figure 6. Inhaled nicotine versus inhaled volume (r = 0.459, P < 0.064).
Filled triangle corresponds to previously reported mean nicotine yield
from smoking machine measurements (Shihadeh and Saleh, 2005). (See
colour version of this figure online at www.informahealthcare.com/iht)
400
1000
y = 1.170x
y = 0.464x
800
Water-(mg)
300
CO (mg)
Inhalation Toxicology Downloaded from informahealthcare.com by 124.197.58.145 on 11/09/10
For personal use only.
Toxicant intake
Carbon monoxide
The mean carbon monoxide intake for 43 determinations was
150 ± 26 mg (mean ± 95% CI). Individual session CO data are
plotted as a function of inhaled volume in Figure 5. Data were
highly correlated (P < 0.0001) with inhaled volume, with an
average slope of 1.17 mg CO per liter of smoke.
shown in Figure 7. Using the fitted slope of 46.4%, inhaled
ma’ssel-derived “tar” was computed for the 41 TPM determinations and found to be 640 ± 119 mg per water pipe use
session. Directly measured (Tarm = TPM − water) and computed (Tarc =0.464 × TPM) ma’ssel-derived “tar” were found
to be highly correlated with inhaled volume (P < 0.0001
and P < 0.001, respectively). Computed ma’ssel-derived
“tar” intake versus inhaled volume is plotted in Figure 8;
an average slope of 4.8 mg ma’ssel-derived “tar” per liter of
smoke is found. It should be noted that due to the differing
combustion conditions and materials involved (e.g. water
pipe ma’ssel typically contains 25% tobacco whereas the
cigarette rod is largely tobacco-derived), ma’ssel-derived
“tar” should not be equated with cigarette “tar”; they likely
have very different chemical compositions as suggested
by the widely differing nicotine:”tar” ratios apparent in
Table 3.
Nicotine (mg)
a sitting, skewing the average puff number. Excluding these
four sessions, the average is 181 puffs per session. Pending
verification in a controlled clinical study, current results
provide preliminary data supporting the notion that puffing
behavior is not greatly altered by the incorporation of the
RINS sampler mouthpiece.
200
600
400
100
200
0
0
50
100
150
200
250
300
Inhaled volume (L)
Figure 5. Inhaled CO versus inhaled volume (r = 0.858, P < 0.0001). Filled
triangle corresponds to previously reported mean CO yield from smoking
machine measurements (Shihadeh and Saleh, 2005). (See colour version
of this figure online at www.informahealthcare.com/iht)
0
0
400
800
1200
1600
2000
TPM (ng)
Figure 7. Measured water versus total particulate matter (TPM) (r > 0.99,
P < 0.0001). (See colour version of this figure online at www.informahealthcare.com/iht)
Inhaled waterpipe smoke toxicants 7
This work was undertaken to address challenges faced in
measuring toxicant intake by real, idiosyncratic users of a
smoking device. A self-powered portable instrument was
developed, which enables sampling the smoke at the mouthpiece as it is generated by real users in natural settings. This
direct approach allows for dispensing many idealizations
inherent in current smoking machine or spent cigarette butt
methodologies for estimating toxicant intake.
The performance of the instrument was assessed by
attaching it to a machine-smoked water pipe, and the
instrument was deployed in a field campaign in Beirut
cafés, sampling 43 water pipe use sessions. The results are
summarized in Table 3. It can be seen that compared with
a single cigarette, a water pipe use session involves much
greater smoke volume, and delivers roughly 20 times the
dry particulate matter (ma’ssel-derived “tar” + nicotine),
seven times the CO, and double the nicotine delivered by a
1600
y = 0.464x
1200
Tar (mg)
Inhalation Toxicology Downloaded from informahealthcare.com by 124.197.58.145 on 11/09/10
For personal use only.
Discussion
800
400
0
0
50
100
150
200
250
300
Inhaled Volume (L)
Figure 8. Inhaled ma’ssel-derived “tar” versus inhaled volume (r = 0.588,
P < 0.001). Filled triangle corresponds to previously reported mean nicotine yield from smoking machine measurements (Shihadeh and Saleh,
2005). (See colour version of this figure online at www.informahealthcare.
com/iht)
Table 3. Summary of smoking topography and toxicant intake results
(mean ± 95% confidence interval).
Smoking machine
(Beirut method;
Cigarette
RINS (current
Shihadeh and
(Djordjevic
study)
Saleh, 2005)
et al., 2000)
Smoke drawn (L)
130 ± 21
91
0.523
TPM (mg)
1193 ± 223
1380 ± 93
NR
640 ± 119
802 ± 86
29
“Tar” (mg)a
CO (mg)
150 ± 26
143 ± 11
22.5
Nicotine (mg)
4.82 ± 1.07
2.96 ± 0.16
2.39
Data from a previous smoking machine study using the Beirut method
is shown for comparison, as is mean data for medium-yield cigarettes,
machine-smoked in accordance with measured human topography
parameters.
a
Water pipe “tar” and cigarette “tar” likely differ in chemical composition
though both are measured in the same manner.
single medium-yield cigarette. These are the first reported
measurements of toxicant delivery by narghile water pipes
in human-generated smoke.
Data presented in Figures 5, 6, and 8 demonstrate the
intuitive result that the greater the smoke volume the greater
the toxicant intake. Absent in other information, the equations given in the figures can be used to estimate intake
from smoke volume measurements. Because the large puff
volumes recorded by RINS indicate that the smoke must be
directly inhaled into the lung rather than held in the mouth,
the reported toxicant intake are best regarded as the actually
inhaled rather than the delivered yields commonly reported
for the cigarette. It is also worth noting that the wide range
of inhaled volumes over which ma’ssel-derived “tar” appears
to increase with volume (Figure 8) illustrates a basic difference with cigarette smoking. Whereas the cigarette is typically
smoked until the tobacco rod is consumed, in the case of the
water pipe the “tar” data suggest that most often the water
pipe session ends well before the ma’ssel charge is spent. This
provides a rationale for water pipe smoking machine protocols to base endpoints on number of puffs drawn rather than
a metric related to the consumption of the tobacco.
The results also support the plausibility of the smoking
machine protocol presented by Shihadeh and Saleh (2005),
which was based on topography measurements in water
pipes smoked by a similar population, namely young adults
in Beirut cafés. In brief, the machine protocol, referred to
below as the Beirut method, consists of smoking 10 g of
ma’ssel with 1.5 easy light charcoal briquettes through a water
pipe smoking machine programmed to produce 171 puffs of
2.6 sec duration, 17 sec interpuff interval, and 530 mL volume.
It can be seen in Table 3 that there were no significant differences in mean CO, TPM, and ma’ssel-derived “tar” generated
by real smokers and by machine using the Beirut method.
Nicotine intake, on the other hand, was ∼60% greater with
human smokers than measured with the Beirut method.
Comparison between the toxicant yields of the Beirut
method and human smokers is better contextualized when
the machine yield data are plotted with the RINS data in
Figures 5, 6, and 8 (see filled triangles). It can be seen that
the method results in CO (Figure 5) and nicotine (Figure 6)
yields that are centrally located in the cluster of values generated by real smokers at the nominal volume of 91 L. The
ma’ssel-derived “tar” yield (Figure 8) is near the upper limit of
those of real smokers in the vicinity of 91 L, but still plausible.
Because >90% of the CO originates in the charcoal (Monzer
et al., 2008), whereas ma’ssel-derived “tar” and nicotine derive
from the ma’ssel (Shihadeh, 2003), we believe that the good
match between the machine smoking and human data for
CO yields suggests that the quantity of charcoal prescribed
by the Beirut method is appropriate.
While the Beirut method appears to give results that are
consistent with human-produced smoke in Beirut cafés, the
method may not be relevant for other population groups or
in other settings. Recent measurements of water pipe smoking topography in clinics in Richmond, Virginia (Eissenberg
and Shihadeh, 2009) and Aleppo, Syria (Maziak et al., 2009),
Inhalation Toxicology Downloaded from informahealthcare.com by 124.197.58.145 on 11/09/10
For personal use only.
8 M. Katurji et al.
for example, demonstrate that mean water pipe topography
parameters can vary widely by setting and/or location. This
fact further highlights the utility of a portable technology in
which toxicant yields can be directly measured in a variety of
communities and age groups without need for developing a
costly machine-based smoking protocol for each.
Apart from the small number of toxicants measured,
there are two main limitations of this study. First, mouthpiece alterations to accommodate RINS may impact the feel
of the water pipe to the user and thereby modify the manner
in which s/he smokes. Second, smoke particle deposition in
the topography sensor located in the hose may negatively
bias the quantity of toxicants reaching the mouthpiece,
and hence, the sampler. Both of these limitations should
be addressed in future studies. While the current study did
not examine PAH and volatile aldehydes, we have previously found these in high quantities in machine-generated
smoke (Shihadeh and Saleh, 2005; Al Rashidi et al., 2008;
Sepetdjian et al., 2008). Challenges remain in validating the
accuracy of the analytical techniques used to measure trace
chemical species when applied to the relatively small smoke
samples generated by RINS. Some trade-offs may be necessary between collecting sufficiently large smoke sample
for accurate chemical analyses and affecting the feel of the
smoking device to the user, since the greater the sampled
fraction, the greater the work needed by the user to draw
a puff.
Conclusions
We have demonstrated RINS, a new technology that enables
direct measurement of smoking topography and toxicant
intake for water pipe users in natural environments. Using
it, we have found that in a single use session, water pipe
smokers inhale large quantities of smoke, ma’ssel-derived
“tar,” CO, and nicotine, which, given the large number of
variables, are remarkably consistent with previous estimates
obtained using the Beirut method smoking machine protocol. Because RINS involves no idealizations about smoker
behavior or product preparation, the results presented here
are perhaps the most robust evidence to date that water pipe
smoking entails inhaling large quantities of toxicants. The
results presented here are consistent with a growing body of
studies on narghile water pipe toxicant content and health
effects, which have unanimously pointed to the hazardous nature of first- and second-hand narghile water pipe
smoking.
Acknowledgements
Mr. Joseph Nassif and his staff at the AUB Engineering Shops
are gratefully acknowledged for their assistance in fabricating the RINS instrument. Mr. Samir Berjaoui provided key
insights in refining the controller algorithm and troubleshooting the electronic circuitry. Dr. Thomas Eissenberg
provided insightful comments on an early draft of the
manuscript.
Declaration of interest
This study was supported by grants from the American
University Research Board and Research for International
Tobacco Control, a secretariat of the IDRC (Canada). The
authors have no competing interests related to this work.
References
Al Rashidi M, Shihadeh A, Saliba NA. 2008. Volatile aldehydes in the mainstream
smoke of the narghile waterpipe. Food Chem Toxicol 46:3546–3549.
Baska T, Pudule I, Tilgale N, Warren CW, Lee J, Lea V, Jones NR. 2008. Smoking
tobacco in waterpipes among adolescents in Europe: the case of Latvia
and Slovakia. Tob Control 17:432.
Brockman J. 2001. Sampling and transport of aerosols, in: Barron P, Willeke K
(Eds.), Aerosol Measurement: Principles, Techniques and Applications,
2nd ed., Wiley-Interscience, New York, pp. 143–190.
Burns DM, Dybing E, Gray N, Hecht S, Anderson C, Sanner T, O’Connor R,
Djordjevic M, Dresler C, Hainaut P, Jarvis M, Opperhuizen A, Straif K. 2008.
Mandated lowering of toxicants in cigarette smoke: a description of the
World Health Organization TobReg proposal. Tob Control 17:132–141.
Cobb C, Ward KD, Maziak W, Shihadeh AL, Eissenberg T. 2010. Waterpipe
tobacco smoking: an emerging health crisis in the United States. Am J
Health Behav 34:275–285.
Daher N, Saleh R, Jaroudi E, Sheheitli H, Badr T, Sepetdjian E, Al Rashidi M,
Saliba N, Shihadeh A. 2010. Comparison of carcinogen, carbon monoxide,
and ultrafine particle emissions from narghile waterpipe and cigarette
smoking: sidestream smoke measurements and assessment of secondhand smoke emission factors. Atmos Environ 44:8–14.
Djordjevic MV, Stellman SD, Zang E. 2000. Doses of nicotine and lung carcinogens delivered to cigarette smokers. J Natl Cancer Inst 92:106–111.
Eissenberg T, Shihadeh A. 2009. Waterpipe tobacco and cigarette smoking:
direct comparison of toxicant exposure. Am J Prev Med 37:518–523.
Eissenberg T, Ward KD, Smith-Simone S, Maziak W. 2008. Waterpipe tobacco
smoking on a U.S. College campus: prevalence and correlates. J Adolesc
Health 42:526–529.
El-Roueiheb Z, Tamim H, Kanj M, Jabbour S, Alayan I, Musharrafieh U. 2008.
Cigarette and waterpipe smoking among Lebanese adolescents, a crosssectional study, 2003–2004. Nicotine Tob Res 10:309–314.
Green CR, Conrad FW, Bridle JK, Borgerding MF. 1985. A liquid chromatography procedure for analysis of nicotine on cellulose acetate filters. Beitr
Tabakforsch Int 13:11–16.
Hammond D, Fong GT, Cummings KM, O’Connor RJ, Giovino GA, McNeill A.
2006. Cigarette yields and human exposure: a comparison of alternative
testing regimens. Cancer Epidemiol Biomarkers Prev 15:1495–1501.
Jawaid A, Zafar AM, Rehman TU, Nazir MR, Ghafoor ZA, Afzal O, Khan JA. 2008.
Knowledge, attitudes and practice of university students regarding waterpipe smoking in Pakistan. Int J Tuberc Lung Dis 12:1077–1084.
Khalil J, Heath RL, Nakkash RT, Afifi RA. 2009. The tobacco health nexus? Health
messages in narghile advertisements. Tob Control 18:420–421.
Kozlowski LT, Rickert WS, Pope MA, Robinson JC. 1982a. A color-matching
technique for monitoring tar/nicotine yields to smokers. Am J Public
Health 72:597–599.
Kozlowski LT, Rickert WS, Pope MA, Robinson JC, Frecker RC. 1982b. Estimating
the yield to smokers of tar, nicotine, and carbon monoxide from the ‘lowest yield’ ventilated filter-cigarettes. Br J Addict 77:159–165.
Kozlowski LT, Rickert WS, Robinson JC, Grunberg NE. 1980. Have tar and nicotine yields of cigarettes changed? Science 209:1550–1551.
O’Connor RJ, Kozlowski LT, Hammond D, Vance TT, Stitt JP, Cummings KM.
2007. Digital image analysis of cigarette filter staining to estimate smoke
exposure. Nicotine Tob Res 9:865–871.
Marian C, O’Connor RJ, Djordjevic MV, Rees VW, Hatsukami DK, Shields PG.
2009. Reconciling human smoking behavior and machine smoking patterns: implications for understanding smoking behavior and the impact
on laboratory studies. Cancer Epidemiol Biomarkers Prev 18:3305–3320.
Maziak W, Rastam S, Ibrahim I, Ward KD, Shihadeh A, Eissenberg T. 2009. CO
exposure, puff topography, and subjective effects in waterpipe tobacco
smokers. Nicotine Tob Res 11:806–811.
Maziak W, Ward KD, Afifi Soweid RA, Eissenberg T. 2004. Tobacco smoking
using a waterpipe: a re-emerging strain in a global epidemic. Tob Control
13:327–333.
Monzer B, Sepetdjian E, Saliba N, Shihadeh A. 2008. Charcoal emissions as a
source of CO and carcinogenic PAH in mainstream narghile waterpipe
smoke. Food Chem Toxicol 46:2991–2995.
Inhaled waterpipe smoke toxicants 9
Inhalation Toxicology Downloaded from informahealthcare.com by 124.197.58.145 on 11/09/10
For personal use only.
Pärna K, Usin J, Ringmets I. 2008. Cigarette and waterpipe smoking among
adolescents in Estonia: HBSC survey results, 1994–2006. BMC Public
Health 8:392.
Pauly JL, O’Connor RJ, Paszkiewicz GM, Cummings KM, Djordjevic MV, Shields
PG. 2009. Cigarette filter-based assays as proxies for toxicant exposure
and smoking behavior—a literature review. Cancer Epidemiol Biomarkers
Prev 18:3321–3333.
Primack BA, Fertman CI, Rice KR, Adachi-Mejia AM, Fine MJ. 2010. Waterpipe
and cigarette smoking among college athletes in the United States.
J Adolesc Health 46:45–51.
Primack BA, Walsh M, Bryce C, Eissenberg T. 2009. Water-pipe tobacco smoking among middle and high school students in Arizona. Pediatrics
123:e282–e288.
Rees V, Zuckerman R, Wayne G, Connolly G. 2007. Tobacco Waterpipe in the U.S.
Survey Profile of College-Age Users and Characteristics of Raw Mo’assel
Tobacco. The 13th Annual Meeting of the Society for Research on Nicotine
and Tobacco (SRNT), Austin, TX, February 21–24.
Saleh R, Shihadeh A. 2008. Elevated toxicant yields with narghile waterpipes
smoked using a plastic hose. Food Chem Toxicol 46:1461–1466.
Sepetdjian E, Shihadeh A, Saliba NA. 2008. Measurement of 16 polycyclic aromatic hydrocarbons in narghile waterpipe tobacco smoke. Food Chem
Toxicol 46:1582–1590.
Shihadeh A. 2003. Investigation of mainstream smoke aerosol of the argileh
water pipe. Food Chem Toxicol 41:143–152.
Shihadeh A, Antonios C, Azar S. 2005. A portable, low-resistance puff topography instrument for pulsating, high-flow smoking devices. Behav Res
Methods 37:186–191.
Shihadeh A, Azar S. 2006. A closed-loop control “playback” smoking machine
for generating mainstream smoke aerosols. J Aerosol Med 19:137–147.
Shihadeh A, Azar S, Antonios C, Haddad A. 2004. Towards a topographical
model of narghile water-pipe café smoking: a pilot study in a high socioeconomic status neighborhood of Beirut, Lebanon. Pharmacol Biochem
Behav 79:75–82.
Shihadeh A, Saleh R. 2005. Polycyclic aromatic hydrocarbons, carbon monoxide, “tar”, and nicotine in the mainstream smoke aerosol of the narghile
water pipe. Food Chem Toxicol 43:655–661.
Watson C, McCraw J, Polzin G, Ashley D, Barr D. 2004. Development of a method
to assess cigarette smoke intake. Environ Sci Technol 38:248–253.