From Dr F:
Dated Aug 28th 2017
***************************************************************
Abstract
Purpose. In 2015, a study identified 5 to 15-fold higher levels of formaldehyde emissions
from an old-generation e-cigarette working at 5.0 V compared to
tobacco cigarettes. We
set to replicate this study using the same e-cigarette equipment and
e-liquid, while
checking for the generation of dry puffs.
Design. Experienced e-cigarette users (n=26) took 4 s puffs at different voltage settings
and were asked to report the generation of dry puffs. Formaldehyde emissions were
measured at both realistic and dry puff conditions.
Results. Dry puffs were detected at ≤4.2 V by 88% of participants; thus, 4.0 V was
defined as the upper limit of realistic use. Levels ranged from 3.4 (SE=2.2) μg/10 puffs at
3.3 V to 718.2 (SE=58.2) μg/10 puffs at 5.0 V. The levels detected at 4.0 V were 19.8
(SE=5.6) μg/10 puffs. At 4.0 V, the daily exposure to formaldehyde from consuming 3 g
of liquid with the device tested would be 32% lower compared to smoking 20
tobacco
cigarettes.
Conclusions. The high levels of formaldehyde emissions that were reported in a previous
study were caused by unrealistic use conditions that create the unpleasant taste of dry
puffs to e-cigarette users and are thus avoided.
Keywords. Smoking, electronic cigarette, formaldehyde, dry puff, aerosol
Highlights
Formaldehyde is produced by thermal degradation in e-cigarettes
Dry puffs result from overheating and create an unpleasant taste that users avoid
In realistic conditions, formaldehyde in e-cigarettes is lower than cigarette smoke
High levels of formaldehyde are produced in unrealistic (dry puff) conditions
Dry puffs should be avoided in the laboratory setting
Introduction
Electronic cigarettes (e-cigarettes) were introduced in the last decade as smoking
alternatives. A growing body of evidence suggests that, although less harmful than
smoking, they are not risk-free (Farsalinos and Polosa, 2014). Research has focused,
among others, on the levels of toxic aldehyde emissions from electronic cigarettes.
Thermal degradation of the main ingredients of e-cigarettes, propylene glycol and
glycerol, can result in the formation of formaldehyde (Bekki et al., 2014). Formaldehyde
is also emitted in tobacco cigarette smoke (Counts et al., 2005). Goniewicz et al. (2014)
reported that the levels of formaldehyde in e-cigarette aerosol were approximately 9
times lower compared to tobacco cigarettes. However, Jensen et al. (2015) measured
formaldehyde emissions from e-cigarettes and reported that the levels emitted were much
higher than from tobacco cigarettes at high power (high voltage) settings, resulting in 5 to
15-fold higher formaldehyde-attributed cancer risk compared to smoking. The media
release (Portland State University, 2015) received worldwide media attention (e.g.
http://www.dailymail.co.uk/wires/afp/article-2920762/Formaldehyde-e-cigarettes-boostcancer-
risk.html). The authors tested two voltage settings (3.3 V and 5.0 V) and found
formaldehyde emissions at 5.0 V only, but they did not control for the development of
dry puffs, an unpleasant aversive taste resulting from overheating of the liquid, which the
users avoid (Farsalinos et al., 2015a).
The dry puff phenomenon, first described in the scientific literature in 2013 (Farsalinos et
al., 2013; Romagna et al., 2013), is common knowledge and experience among ecigarette
users and has been presented in detail elsewhere (Farsalinos et al., 2015a). In
brief, it represents an unpleasant change in the taste of the e-cigarette puff and is related
to overheating and thermal degradation of e-cigarette liquid ingredients. It results from
too much energy delivered to the atomizer, too much power and/or long puff duration or
when not enough liquid is present in the atomizer. Since this is an organoleptic
parameter, it is by definition subjective and can only be detected when reported by ecigarette
users. One study showed substantially elevated formaldehyde emission from ecigarettes
under dry puff conditions compared to realistic use settings (Farsalinos et al.,
2015a).
The study by Jensen et al. generated some controversy and several letters to the editor
suggested that the findings of very high levels of formaldehyde emissions could be
explained by overheating the liquid (Bates and Farsalinos, 2015; Kershaw, 2015; Nitzkin
et al., 2015). However, until now experimental evidence substantiating that dry puffs
were the reason for the high formaldehyde emissions was lacking. Therefore, the purpose
of the current study was to clarify this issue by measuring formaldehyde emissions using
the same e-cigarette device, atomizer and liquid at different voltage settings after
verifying and differentiating between realistic and dry puff conditions. Additionally, the
levels of formaldehyde emitted from the e-cigarette tested were compared with data on
formaldehyde emissions from tobacco cigarettes.
Methods
Equipment and participants
After contacting the authors of the original study, we obtained the same e-cigarette
equipment and liquid. The equipment used was CE4 top coil atomizer, Innokin iTaste VV
V3.0 variable voltage battery device and Halo Café Mocha liquid with 6 mg/mL nicotine
concentration. The CE4 atomizer represents an outdated design which, to the best of our
knowledge, is not currently available in Europe. Thus, it was purchased from China.
Twenty six adult experienced daily nicotine-containing e-cigarette users were recruited to
identify the generation of dry puffs. All participants were former smokers and were using
e-cigarettes for at least 2 months. When asked, they all knew the phenomenon of dry
puffs which was described by them as an unpleasant “burning” taste related to liquid
overheating. For the experimental session, they took 5-7 puffs of 4 s duration and 30 s
interpuff interval at varying voltage settings and reported whether the characteristic
change in taste associated with dry puffs was detected. A preliminary assessment by two
members of the research team (experienced e-cigarette users) identified the upper limit of
realistic puffing conditions at approximately 4.0 V. To make the duration of the
experiment acceptable and limit total nicotine intake, participants tested the devices
starting at 3.6 V and with increments of 0.2 V until the time they identified dry puffs.
Each session was accompanied by 5-10 minutes resting period, during which the
participants did not use their own e-cigarette. Participants were blinded to the power
setting and the e-cigarette battery screen was covered with black tape. The device was not
tested in random order of voltage settings because experienced e-cigarette users would
easily identify the increased or decreased aerosol yield associated with substantial
increases or decreases in voltage. When dry puffs were identified, each participant
retested the device after 15-20 minutes of resting time. Initially, the same voltage that
resulted in dry puffs was applied; if dry puffs were detected, then they tested the device at
0.2 V lower setting, while if dry puffs were not detected they retested the device at 0.2 V
higher setting. Findings from this session were used to determine the voltage associated
with dry puffs. Each participant used his own atomizer since the mouth piece of the
atomizer was non-removable. The study conforms to the Declaration of Helsinki for
research involving human subjects and was approved by the institutional review board.
Written informed consent was signed by the volunteers before participating to the study.
Aerosol collection and formaldehyde measurements
Aerosol collections were performed at different voltage settings using a smoking machine
and 2 impingers (connected in series) containing a solution of 2,4-dinitrophenylhydrazine
(2,4-DNPH) and acetonitrile. The puffing regime used was 60 mL puff volume, 4 s puff
duration and 30 s interpuff interval. In total, 50 puffs were collected per sample. Three
unused CE4 atomizers were used and two collections per atomizer were performed at
each voltage setting (total of six repetitions per voltage setting). Blank air samples were
simultaneously collected in different impingers to measure environmental (room air)
levels of formaldehyde; these levels were subtracted from the levels in the collected
aerosol. Formaldehyde was measured by High Performance Liquid Chromatography
using a previously validated protocol with slight modifications (Farsalinos et al., 2015;
Cooperation Centre for Scientific Research Relative to Tobacco, 2013).
Statistical analysis
Formaldehyde levels were expressed as μg/10 puffs, with mean value and standard error
(SE) reported. Liquid consumption per puff was expressed as mg/puff. Voltage settings
were reported in the study by Jensen et al. (2015). However, as explained previously
(Farsalinos et al., 2015a), power settings are more appropriate when assessing the energy
delivered to the atomizer; thus, both voltage and power settings are presented here.
Comparison in liquid consumption per puff and formaldehyde levels between different
voltage settings was performed by one-way analysis of variance (ANOVA) with post-hoc
Bonferroni correction. Analyses were performed with SPSS v22.0. A P value of < 0.05
was considered statistically significant.
Results
Liquid consumption and formaldehyde emissions
Dry puffs were identified at 4.0 V (7.3 W) by 8 participants, 4.2 V (8.0 W) by 15 and at
4.4 V (8.4 W) by 3 participants. None of the participants was willing to try the device at
5.0 V, explaining that the expected taste would be really aversive. During the testing,
some atomizer units were either non-functional or were generating dry puffs at low
voltage settings, indicating that they were defective. These atomizers were replaced by
new units. Given that most e-cigarette users (88%) experienced the dry puff taste at 4.2
V, we consider 4.0 V as the maximum level associated with realistic use conditions.
Based on this, aerosol collections for formaldehyde measurements were performed at the
following voltage (power) settings: 3.3 V (5.0 W), 3.6 V (5.9 W), 4.0 V (7.3 W), 4.2 V
(8.0 W), 4.6 V (9.6 W), 4.8 V (10.5 W) and 5.0 V (11.4 W).
The amount of liquid consumption per puff at different voltage setting is displayed in
Figure 1. Liquid consumption ranged from 3.7 (0.3) mg at 3.3 V to 8.0 (0.5) mg at 5.0 V.
The differences between liquid consumption at different voltage settings were statistically
significant (one-way ANOVA: F = 17.1, P < 0.001). While a linear increase in liquid
consumption per puff was observed from 3.3 V to 4.0 V, the pattern was erratic at higher
voltage settings.
The levels of formaldehyde emissions are presented in Figure 2 together with the results
by Jensen et al. Formaldehyde levels ranged from 3.4 (2.2) μg/10 puffs at 3.3 V to 718.2
(58.2) μg/10 puffs at 5.0 V. The differences between formaldehyde levels at different
voltage settings were statistically significant (one-way ANOVA: F = 34.1, P < 0.001). In
the post-hoc analysis, the levels of formaldehyde were not significantly different at 3.3 V,
3.6 V, 4.0 V and 4.2 V; the aldehyde levels only increased at higher settings.
Formaldehyde was detected at the lowest voltage setting while Jensen et al. reported nondetected
levels (< 0.1 μg/10 puffs) at that setting. At the maximum voltage setting (5.0
V), the levels of formaldehyde we detected were 89% higher compared to the study by
Jensen et al. At the upper limit of realistic use conditions (4.0 V), formaldehyde was
found at levels of 19.8 (5.6) μg/10 puffs, which were 36-fold lower compared to the
levels at 5.0 V.
Comparison with tobacco cigarettes
Formaldehyde emissions from e-cigarettes were compared with tobacco cigarettes using
data from Counts et al. (2005), calculating the average levels of formaldehyde emissions
from 50 tobacco cigarette products under Health Canada Intense puffing regime. The
levels measured in that study were 74.0 (3.4) μg/cigarette. To be consistent with the
analysis by Jensen et al., 3 g e-cigarette liquid consumption was compared with 20
tobacco cigarettes (1480 μg/20 cigarettes formaldehyde). At the upper voltage setting of
realistic use conditions (4.0 V), the level of formaldehyde exposure from e-cigarette use
was 1005.4 μg/3g liquid, which is 32% lower compared to 20 tobacco cigarettes, At 5.0
V, a setting that was associated with dry puff conditions, the respective level was 27151.5
μg/3g liquid, which is 18.3-fold higher compared to smoking 20 tobacco cigarettes.
Discussion
This replication of a previous study demonstrates that high formaldehyde emissions in ecigarettes
can be found during testing in a laboratory setting. However, such levels are
caused by dry puffs due to overheated e-liquid and, thus, do not correspond to realistic
vaping conditions. The study shows the importance of ensuring that realistic use
conditions are used in laboratory studies when examining e-cigarette aerosol content.
Recently, Sleiman et al. (2017) reported formaldehyde emissions of 48,200 μg/g liquid at
4.8 V using a similar atomizer. This means that the consumption of 3 g of liquid would
be equivalent to smoking 1954 cigarettes in terms of formaldehyde exposure when
compared to the tobacco cigarette smoke levels reported by Counts et al. (2005). That
study too needs to be replicated under verified realistic use conditions to examine
whether such levels are associated with real-life exposure and could represent a
significant health risk for the users.
The atomizer used in this study is an outdated design, with the wick and coil head
positioned just under the mouthpiece (“top-coil”). This is an inefficient design,
preventing rapid delivery of liquid to the coil due to liquid movement against gravity.
This can create conditions of insufficient liquid replenishment to the coil that can result in
overheating because the evaporation rate surpasses the liquid replenishment rate. The
non-linear change in liquid consumption per puff that was observed here under dry puff
conditions could be explained by this phenomenon. To the best of our knowledge, the last
atomizer produced with this top coil design was the CE5 atomizer, which was released in
2012 (see:
). Since then, atomizers
feature a more efficient bottom coil design and are also using cotton wick, which has
better sorptivity and is expected to improve the speed and efficiency of liquid supply to
the coil compared to silica wick. A recent study has shown that aldehyde emissions are
substantially lower, by up to 2 orders of magnitude, in bottom coil cotton atomizers
compared to older designs (Gilman et al., 2016). This indicates that the evolution and
advances in design and material has resulted in improved safety profile. This should be
taken into consideration by regulators. For example, in the US the current substantial
equivalence legislation allows products commercially available before February 15, 2007
to be available without applying for an order permitting their marketing. This means that
more harmful old-generation devices will be available while regulatory burdens will be
applied to newer-generation products if this legislation is implemented for e-cigarettes.
Substantial variability in formaldehyde emissions was observed between repetitions,
which was also observed in the studies by Jensen et al. (2015) and Gilman et al. (2016)
with the same atomizer, but was not seen in studies using different atomizers (Farsalinos
et al., 2015a; Gilman et al., 2016). Since we also noticed problems with functioning of
these atomizers, as described in the results section, we would advise e-cigarette users
against the use of these atomizers. It is possible that the variability in dry puff detection
by the participants could also be related to different characteristics of the atomizers
tested. Additionally, the value of using these atomizers for research purposes is
questionable, since this is an outdated product with inconsistent performance and is not
representative of the e-cigarette devices developed in the past few years.
When discussing the fact that e-cigarette users avoid dry puffs, a possibility was raised
that flavourings may mask the unpleasant taste and so increase the likelihood that they
would be inhaled (Pankow et al., 2015). The present findings show that dry puffs are
identified with flavoured e-liquid too, and this is in agreement with the common
experience of e-cigarette users most of whom use flavoured e-liquids and are well aware
of dry puffs. A recent study by Geiss et al. (2016) also used flavoured e-liquid and the
users were able to identify the burning taste of dry puffs. This phenomenon is not only
reported in the literature but represents common knowledge among e-cigarette users
(Farsalinos et al., 2015b). Therefore, it should be taken into consideration when assessing
emissions from e-cigarettes, especially thermal degradation products, in order to ensure
that realistic use conditions are tested and findings are relevant to true exposure of users.
A call to retract the paper by Jensen et al. was published in 2015 (Bates and Farsalinos,
2015), which was based on an alleged mispresentation of the calculated relative cancer
risk of e-cigarettes compared to tobacco cigarettes without ensuring that realistic use
conditions representative of true exposure of humans were tested. The study herein
provides experimental evidence to support the call for retraction, showing that blindly
testing e-cigarettes in the laboratory setting without evaluating realistic use is a serious
omission that can result in misleading conclusions about the risk to consumers compared
to smoking. In fact, such testing of e-cigarettes is not very different from overcooking
food to the point of becoming a inedible piece of charcoal and then assuming that
consumers would consume it and be exposed to the resulting carcinogenic compounds in
their daily routine. Accepting that e-cigarettes are less harmful than smoking (Farsalinos
and Polosa, 2014; Glasser et al., 2017), such an omission could result in unintendedly
misleading smokers into thinking that there is little to be gained by switching to ecigarettes.
A limitation of the current study is that it cannot determine a causal link between
formaldehyde emissions and dry puff detection. It is possible that other aldehydes, such
as acrolein which has an acrid smell, could be responsible for dry puff detection. More
studies are needed to determine this link. Additionally, it is important for future studies to
examine the inter-individual and intra-individual variability in dry puff detection,
preferably by using more consistent and reliable devices than the one tested herein.
In conclusion, the study shows the critical need to verify that realistic use conditions are
tested in laboratory studies of e-cigarette emissions. This would ensure that abuse of
devices in the laboratory setting is avoided and that findings have clinical relevance and
represent realistic exposure of e-cigarette users.