Articles

Electronic cigarettes for smoking cessation: a randomised

controlled trial
Christopher Bullen, Colin Howe, Murray Laugesen, Hayden McRobbie, Varsha Parag, Jonathan Williman, Natalie Walker

Summary
Background Electronic cigarettes (e-cigarettes) can deliver nicotine and mitigate tobacco withdrawal and are used by Published Online
many smokers to assist quit attempts. We investigated whether e-cigarettes are more eective than nicotine patches September 7, 2013
http://dx.doi.org/10.1016/
at helping smokers to quit. S0140-6736(13)61842-5
See Online/Comment
Methods We did this pragmatic randomised-controlled superiority trial in Auckland, New Zealand, between Sept 6, http://dx.doi.org/10.1016/
2011, and July 5, 2013. Adult (18 years) smokers wanting to quit were randomised (with computerised block S0140-6736(13)61534-2
randomisation, block size nine, stratied by ethnicity [Mori; Pacic; or non-Mori, non-Pacic], sex [men or women], National Institute for Health
and level of nicotine dependence [>5 or 5 Fagerstrm test for nicotine dependence]) in a 4:4:1 ratio to 16 mg nicotine Innovation, School of
Population Health, The
e-cigarettes, nicotine patches (21 mg patch, one daily), or placebo e-cigarettes (no nicotine), from 1 week before until
University of Auckland,
12 weeks after quit day, with low intensity behavioural support via voluntary telephone counselling. The primary Auckland, New Zealand
outcome was biochemically veried continuous abstinence at 6 months (exhaled breath carbon monoxide (C Bullen MBChB, C Howe PhD,
measurement <10 ppm). Primary analysis was by intention to treat. This trial is registered with the Australian New V Parag MSc, N Walker PhD);
Health New Zealand, Lyttelton,
Zealand Clinical Trials Registry, number ACTRN12610000866000.
Christchurch, New Zealand
(M Laugesen MBChB); Wolfson
Findings 657 people were randomised (289 to nicotine e-cigarettes, 295 to patches, and 73 to placebo e-cigarettes) and Institute of Preventive
were included in the intention-to-treat analysis. At 6 months, veried abstinence was 73% (21 of 289) with nicotine Medicine, UK Centre for
Tobacco Control Studies,
e-cigarettes, 58% (17 of 295) with patches, and 41% (three of 73) with placebo e-cigarettes (risk dierence for
Queen Mary University of
nicotine e-cigarette vs patches 151 [95% CI 249 to 551]; for nicotine e-cigarettes vs placebo e-cigarettes 316 London, Charterhouse Square,
[95% CI 229 to 861]). Achievement of abstinence was substantially lower than we anticipated for the power London, UK
calculation, thus we had insucient statistical power to conclude superiority of nicotine e-cigarettes to patches or to (H McRobbie MBChB); and
Department of Public Health
placebo e-cigarettes. We identied no signicant dierences in adverse events, with 137 events in the nicotine and General Practice,
e-cigarettes group, 119 events in the patches group, and 36 events in the placebo e-cigarettes group. We noted no University of Otago,
evidence of an association between adverse events and study product. Christchurch, New Zealand
(J Williman PhD)

Interpretation E-cigarettes, with or without nicotine, were modestly eective at helping smokers to quit, with similar Correspondence to:
Dr Christopher Bullen, The
achievement of abstinence as with nicotine patches, and few adverse events. Uncertainty exists about the place of
National Institute for Health
e-cigarettes in tobacco control, and more research is urgently needed to clearly establish their overall benets and Innovation, School of Population
harms at both individual and population levels. Health, The University of
Auckland, Private Bag 92019,
Auckland 1142, New Zealand
Funding Health Research Council of New Zealand. c.bullen@nihi.auckland.ac.nz

Introduction dimensions of smoking. However, a trial in 300 smokers

Since their launch in 2004, electronic cigarettes unwilling to quit showed low rates of cessation at
(e-cigarettes), a diverse range of battery operated devices 12 months for nicotine e-cigarettes and placebo
that vaporise nicotine for inhalation, have been purchased e-cigarettes.9 E-cigarettes also have potential to harm:
by millions of people.1 Many smokers use e-cigarettes to researchers have detected toxins in e-cigarette uid and
help them quit (27% of those making a quit attempt in vapour,10 but at much the same concentrations as with
the UK, in May, 20132), and sales are increasing so rapidly NRT and lower than in cigarette smoke;11 a review
that some analysts predict that they will surpass cigarette deemed e-cigarettes to be very unlikely to pose
sales within a decade.1 signicant risks to smokers.12
The place of e-cigarettes in tobacco control is contro- In this trial we aimed to assess whether e-cigarettes
versial,3,4 and there is a paucity of reliable data to inform with cartridges containing nicotine (nicotine e-cigarette)
debate. Available research suggests that e-cigarettes were more eective for smoking cessation than nico-
have the potential to assist smokers to quit or reduce tine patches, and included a blind comparison with
smoking: surveys show that many smokers try e-cigarettes containing no nicotine (placebo e-cigarette).
e-cigarettes for these reasons,5,6 and studies show that We hypothesised that nicotine e-cigarettes would be
e-cigarettes are capable of delivering nicotine into the more eective than patches and placebo e-cigarettes for
bloodstream and attenuating tobacco withdrawal as smoking reduction, tobacco dependence, and relief of
eectively as nicotine replacement therapy (NRT).7,8 Use withdrawal symptoms, and that they would have no
of e-cigarettes also simulates behavioural and sensory greater risk of adverse events than nicotine patches.

www.thelancet.com Published online September 7, 2013 http://dx.doi.org/10.1016/S0140-6736(13)61842-5 1

 Articles

Methods 2 weeks; and those with poorly controlled medical dis-

Study design and participants orders, allergies, or other chemical dependence. Partici-
We did this three parallel group, randomised controlled pants were mailed study information, and consent forms
trial in Auckland, New Zealand. First randomisation was to sign and return. The Northern X Regional Ethics
on Sept 6, 2011, and last follow-up was on July 5, 2013. Committee approved the study (Number NTX/10/11/111);
The published protocol describes procedures in detail.13 the Standing Committee on Therapeutic Trials approved
In brief, people were eligible if they were aged 18 years the use of nicotine e-cigarettes because they were not
or older, had smoked ten or more cigarettes per day for permitted for sale in New Zealand, but could be imported
the past year, wanted to stop smoking, and could provide for personal use or research.
consent. We recruited via community newspapers,
inviting people to call the study centre for eligibility Randomisation and masking
prescreening, done by research assistants, who also Callers who met the inclusion criteria and gave demo-
completed follow-up assessments. We excluded pregnant graphic details and information about nicotine depen-
and breastfeeding women; people using cessation drugs dence (Fagerstrm test for nicotine dependence [FTND]14)
or in an existing cessation programme; those reporting were randomised by the study statistician (VP) in a
heart attack, stroke, or severe angina in the previous 4:4:1 ratio to nicotine e-cigarettes, patches, or placebo

1293 assessed for eligibility

636 excluded
202 did not meet inclusion criteria
162 declined to participate
272 other reasons
657 randomised

289 assigned to nicotine e-cigarettes 295 allocated to nicotine patches 73 allocated to placebo e-cigarettes
289 received allocated intervention 295 received allocated intervention 73 received allocated intervention

13 lost to follow-up 27 lost to follow-up 6 lost to follow-up

3 discontinued 16 discontinued 1 discontinued

273 assessed on quit date 252 assessed on quit date 66 assessed on quit date

13 lost to follow-up 17 lost to follow-up 4 lost to follow-up

3 discontinued

260 assessed at 1 month follow-up 232 assessed at 1 month follow-up 62 assessed at 1 month follow-up

15 lost to follow-up 6 lost to follow-up 3 lost to follow-up

2 discontinued

245 assessed at 3 month follow-up 224 assessed at 3 month follow-up 59 assessed at 3 month follow-up

2 lost to follow-up 8 lost to follow-up 2 lost to follow-up

1 discontinued 1 discontinued
1 death

241 assessed at 6 month follow-up 215 assessed at 6 month follow-up 57 assessed at 6 month follow-up

289 included in analysis (ITT) 295 included in analysis (ITT) 73 included in analysis (ITT)

Figure 1: Trial prole

11 protocol violations occurred in the nicotine e-cigarettes group (three pregnancies, seven no biochemical validation, one undisclosed medication ineligibility).
11 protocol violations occurred in the patches group (four pregnancies, four no biochemical validation, three undisclosed medication ineligibility). Three protocol
violations occurred in the placebo e-cigarettes group (one no biochemical validation, two undisclosed medication ineligibility). ITT=intention to treat.

2 www.thelancet.com Published online September 7, 2013 http://dx.doi.org/10.1016/S0140-6736(13)61842-5

 Articles

e-cigarettes, with computerised block randomisation, dependence (according to the Glover-Nilsson smoking
block size nine, stratied by: ethnicity (Mori; Pacic; or behavioural questionnaire, GN-SBQ).20
non-Mori, non-Pacic), sex (men or women), and level The primary outcome was continuous smoking abstin-
of nicotine dependence (>5 or 5 FTND). It was not ence (self-reported abstinence over the whole follow-up
feasible to mask participants to allocation to patch or period, allowing 5 cigarettes in total21), 6 months after
e-cigarettes. Research assistants undertaking outcome quit day, veried at that point in time by exhaled breath
assessments used a list generated by the trial database carbon monoxide measurement (<10 ppm), using Bedfont
giving no indication of product allocation. Micro Smokerlyzers (Bedfont Scientic, Maidstone, UK).
Carbon monoxide tests were administered by research
Procedures assistants at the University of Aukland; participants were
Elusion e-cigarettes are among the e-cigarette market not paid for testing, but received transportation costs.
leaders in Australasia; in New Zealand, nicotine e-cigarettes Secondary outcomes assessed at 1, 3, and 6 months post
are not permitted to be sold, but nicotine-free e-cigarettes quit day were: continuous abstinence, 7 day point
are widely available for sale and identical in appearance to prevalence abstinence (proportion reporting no smoking
nicotine versions. We commissioned analyses of these of tobacco cigarettes, not a pu, in the past 7 days),
e-cigarettes: the liquid was free of diethylene glycol (a toxin number of tobacco cigarettes smoked per day, proportion
detected in uid in one brand of e-cigarettes10); nicotine of participants reducing tobacco smoking, time to relapse
cartridges (labelled 16 mg) contained 1016 mg nicotine to tobacco smoking, number of patches or cartridges
per mL; and placebo cartridges contained no nicotine. used, use of other cessation treatments, withdrawal symp-
Vapour analyses done midway through the trial (using toms, stage of addiction,19 smoking latency,22 and adverse
Goniewicz and colleagues methodology15) showed that events. Data collection continued as scheduled if
300 pus from one nicotine e-cigarette cartridge delivered participants discontinued study treatments.
36 mg nicotine, equivalent to smoking between one and
ve tobacco cigarettes. The rst 20 participants randomised Statistical analysis
to the nicotine e-cigarettes group were invited to take part A sample size of 657 (292 in the nicotine e-cigarettes
in testing, and four completed the testing regimen. In group, 292 in the patches group, 73 in the placebo
these four participants, who had been using the nicotine
e-cigarettes for at least 1 week, plasma nicotine con- Nicotine Patches Placebo
centrations were sampled every 10 min for 1 h, and peaked e-cigarettes (n=295) e-cigarettes
(n=289) (n=73)
at 10 min after commencement of product use at
34 ng/mL, a median increase from baseline of 21 ng/mL. Age (years) 436 (127) 404 (130) 432 (124)
We chose nicotine patches (21 mg/24 h) for comparison Women 178 (62%) 182 (62%) 45 (62%)
with e-cigarettes because they are the most popular NRT Ethnicity*
product in New Zealand,16 have proven eectiveness,17 and New Zealand Mori 95 (33%) 95 (32%) 23 (32%)
few known adverse events.17 Non-Mori 194 (67%) 200 (68%) 50 (68%)
Participants allocated to patches were sent exchange Education below year 12 or no qualication 150 (52%) 123 (42%) 38 (52%)
cards in the mail redeemable for patches from com- Average number of cigarettes (including RYO) smoked 184 (72) 176 (60) 177 (56)
munity pharmacies, with instructions to use patches per day
daily, from 1 week before until 12 weeks after their chosen Age started smoking (years) 156 (47) 152 (38) 157 (51)
quit day, consistent with smoking cessation guidelines.18 Number of years smoking continuously 259 (131) 235 (129) 248 (137)
We also supplied vouchers to these participants to cover Type of tobacco usually smoked
dispensing costs. Participants in both e-cigarettes groups Factory made only 167 (58%) 167 (57%) 47 (64%)
were couriered an e-cigarette, spare battery and charger, RYO only 92 (32%) 92 (31%) 21 (29%)
and cartridges (with labels masked to nicotine content), Both 30 (10%) 35 (12%) 5 (7%)
plus simple instructions to use them as desired from Lives with other smokers 151 (52%) 149 (51%) 42 (58%)
1 week before until 12 weeks after their chosen quit day. At least 1 quit attempt in past 12 months 158 (55%) 169 (57%) 39 (53%)
All randomised participants were referred (by fax or by a FTND score 56 (20) 55 (20) 55 (20)
scanned request) to Quitline, who called the participants FTND >5 (high dependence) 157 (54%) 162 (55%) 40 (55%)
to oer telephone-based behavioural support. Participants GN-SBQ score 201 (79) 201 (84) 214 (86)
who declined or did not call back were still able to access Self-ecacy to quit 37 (10) 37 (09) 36 (10)
other Quitline support, such as Txt2Quit (a free SMS AUTOS total score 226 (72) 231 (76) 234 (73)
support service). Quitline provided us with reports to
Data are mean (SD) or n (%). RYO=roll your own (loose tobacco) cigarettes. FTND=Fagerstrm test of nicotine
monitor usage. After randomisation, additional baseline
dependence. GN-SBQ: Glover-Nilsson smoking behavioural questionnaire. AUTOS=autonomy over smoking scale;
data were collected: education, smoking and quitting higher scores indicate greater dependence. *All non-Mori ethnicity categories aggregated as non-Mori.25 Age 16 or
history, quitting self-ecacy, medication, withdrawal 17 years. Self-ecacy to quit=belief in ability to quit this time, measured on scale of 1 to 5, 1=very low, 5=very high.
symptoms and stage of addiction (according to the
Table 1: Baseline characteristics of participants
autonomy over smoking scale, AUTOS),19 and behavioural

www.thelancet.com Published online September 7, 2013 http://dx.doi.org/10.1016/S0140-6736(13)61842-5 3

 Articles

Nicotine Patches (n=295) Dierence Relative risk Risk dierence

e-cigarettes p value (95% CI) (95% CI)
(n=289)
Continuous abstinence
1 month 67 (232%) 47 (159%) 003 146 (104 to 204) 725 (084 to 1366)
3 months 38 (131%) 27 (92%) 012 144 (090 to 233) 400 (110 to 910)
6 months (primary outcome) 21 (73%) 17 (58%) 046 126 (068 to 234) 151 (249 to 551)
Sensitivity analyses for 6 months continuous abstinence data
Complete case analysis* 21/241 (87%) 17/215 (79%) 076 110 (060 to 203) 080 (427 to 587)
Per-protocol analysis 1 21/231 (91%) 15/207 (72%) 048 125 (066 to 237) 184 (328 to 696)
Per-protocol analysis 2 20/211 (95%) 13/151 (86%) 078 110 (057 to 214) 087 (510 to 684)
Per-protocol analysis 3 12/147 (82%) 12/138 (87%) 087 094 (044 to 202) 054 (700 to 592)
Including not biochemically veried 30 (104%) 21 (71%) 016 146 (086 to 249) 326 (132 to 784)
Repeated measures analysis||
Overall treatment eect 005 161 (100 to 257)
1 month eect 0004 187 (123 to 285)
3 months eect 012 152 (089 to 258)
6 months eect 021 146 (081 to 262)
7 day point prevalence abstinence
1 month 69 (239%) 51 (173%) 005 138 (100 to 191) 659 (005 to 1313)
3 months 62 (215%) 50 (170%) 017 127 (091 to 177) 450 (188 to 1088)
6 months 61 (211%) 46 (156%) 009 135 (096 to 191) 552 (075 to 1179)

All analyses are intention to treat unless otherwise specied (assumes participants with missing smoking status were smoking). Data are n (%) or n/N (%) unless otherwise
specied. *Complete case analysis: excludes 128 participants with missing 6 month visits (withdrawn or lost to follow-up; 48 in nicotine e-cigarettes group and 80 in patches
group), and includes 456 participants (241 in nicotine e-cigarettes group and 215 in patches group). Per-protocol analysis 1: excludes protocol violations: pregnancy, death,
quitters who did not have biochemical verication, undisclosed medication ineligibility, withdrew, and lost to follow-up at 6 months. Per-protocol analysis 2: excludes
protocol violations from per-protocol analysis 1 plus: cross-overs, use of other or combined nicotine replacement therapy products, and use of non-nicotine replacement
therapy (eg, varenicline). Per-protocol analysis 3: excludes protocol violations from per-protocol analysis 2 plus: participants still using product to which they were
randomised at 6 months. Continuous abstinence including not biochemically veried: eight participants in nicotine e-cigarettes group: one moved, two refused, four did
not attend appointment, one adverse event (birth) did not want to attend; four participants in patches group: one moved, three refused. ||Output for repeated measures
analysis is dierence in least squares means, not relative risk.

Table 2: Continuous smoking abstinence and 7 day point prevalence, nicotine e-cigarettes versus patches

e-cigarettes group) conferred 80% power, with two- including baseline values. We also did per-protocol
sided p=005, to detect an absolute dierence of 10% in analyses for the primary outcome, in which participants
quit rates between the nicotine e-cigarettes group and with major protocol violations (eg, cross-over treat-
patches group (1:1 ratio), and a 15% dierence between ments, withdrawals, and loss to follow-up) were
the nicotine e-cigarettes group and placebo e-cigarettes excluded. We assessed consistency of eects for pre-
group (4:1 ratio), with expected quit rates of 15% in the specied subgroups (men vs women, ethnicity [Mori vs
placebo e-cigarettes group and 20% in the patches non-Mori]) using tests for heterogeneity. Secondary
group (based on meta-analyses of NRT trials).23 We analyses were done with overall cessation rates
used SAS (version 93) for analyses. The primary corrected for discordance between reported and veried
analyses used the intention-to-treat approach (partici- cessation. We used Kaplan-Meier curves and the log-
pants with unknown smoking status were assumed to rank test for analyses of time to relapse. Adverse events
be smoking). We calculated quit rates, relative risks were dened according to international guidelines,
(RR), and absolute risks for nicotine e-cigarettes versus categorised by CB (masked to intervention product) as
patches, and for nicotine e-cigarettes versus placebo related or unrelated to the intervention, and analysed
e-cigarettes. We compared treatment groups using as serious or non-serious, by treatment group and
tests, with multivariate regression adjusting for other association with study treatment, in line with recom-
variables as appropriate. The proportions of participants mended best practice.24
with signicantly reduced smoking consumption of at This trial is registered with the Australian New Zealand
least 25% and 50% were calculated using the same Clinical Trials Registry, number ACTRN12610000866000.
methods. Change from baseline in each of the repeated
AUTOS measures and cigarettes smoked per day (in Role of the funding source
non-abstainers) were analysed using mixed models The sponsor of the study had no role in study design,
with a compound symmetry covariance structure data collection, data analysis, data interpretation, or

4 www.thelancet.com Published online September 7, 2013 http://dx.doi.org/10.1016/S0140-6736(13)61842-5

 Articles

Nicotine e-cigarettes Placebo e-cigarettes Dierence Fishers Relative risk Risk dierence
(n=289) (n=73) exact p value (95% CI) (95% CI)
Continuous abstinence
1 month* 67 (232%) 12 (164%) 021 141 (081 to 246) 674 (306 to 1654)
3 months* 38 (131%) 5 (68%) 014 192 (078 to 470) 630 (068 to 1328)
6 months (primary outcome) 21 (73%) 3 (41%) 044 177 (054 to 577) 316 (229 to 861)
Sensitivity analyses for 6 months continuous abstinence data
Complete case analysis 21/241 (87%) 3/57 (53%) 059 166 (051 to 536) 345 (335 to 1025)
Per-protocol analysis 1 21/231 (91%) 3/54 (56%) 059 164 (051 to 529) 353 (362 to 1068)
Per-protocol analysis 2 20/211 (95%) 2/46 (43%) 036 218 (053 to 900) 513 (197 to 1223)
Per-protocol analysis 3 12/147 (82%) 1/30 (33%) 070 245 (033 to 1813) 483 (297 to 1263)
Including not biochemically 30 (104%) 4 (55%) 026 189 (069 to 521) 490 (139 to 1120)
veried||
Repeated measures analysis**
Overall treatment eect 013 191 (083 to 437)
1 month eect 009 180 (090 to 361)
3 months eect 016 200 (076 to 528)
6 months eect 023 192 (065 to 566)
7 day point prevalence abstinence
1 month* 69 (239%) 12 (164%) 017 145 (083 to 253) 744 (238 to 1726)
3 months* 62 (215%) 12 (164%) 034 131 (074 to 229) 501 (472 to 1474)
6 months* 61 (211%) 16 (219%) 088 096 (059 to 157) 081 (1140 to 978)

All analyses are intention to treat unless otherwise specied (assumes all participants with missing smoking status were smoking). Data are n (%) or n/N (%)
unless otherwise specied. *Dierence from 2 test. Complete case analysis: excludes 64 participants with missing 6 month visits (withdrawn or lost to
follow-up; 48 in nicotine e-cigarettes group and 16 in placebo e-cigarettes group) and includes 298 (241 in nicotine e-cigarettes group and 57 in placebo
e-cigarettes group). Per-protocol analysis 1: excludes protocol violations: pregnancy, death, quitters who did not have biochemical verication at 6 months,
undisclosed medication ineligibility, withdrew, and lost to follow-up at 6 months. Per-protocol analysis 2: excludes protocol violations from per-protocol
analysis 1 plus: cross-overs, use of other or combined nicotine replacement therapy products, and use of non-nicotine replacement therapy (eg, varenicline).
Per-protocol analysis 3: excludes protocol violations from per-protocol analysis 2 plus: participants still using product to which they were randomised at
6 months. ||Continuous abstinence including not biochemically veried: eight participants in nicotine e-cigarettes group who reported quitting did not attend
for biochemical verication (one moved, two refused, four did not attend appointment, one adverse event [birth] did not want to attend); one participant in the
placebo e-cigarettes group did not attend appointment. **Output for repeated measures analysis is dierence in least squares means (not relative risk).

Table 3: Continuous abstinence and 7 day point prevalence, nicotine e-cigarettes versus placebo e-cigarettes

writing of the report. The corresponding author had full 100 Nicotine EC
Patches
access to all the data in the study and had nal Placebo EC
responsibility for the decision to submit for publication.
80
Probability of continuous abstinence (%)

Results
Of 1293 people who were assessed, 657 were eligible for
inclusion in the study (gure 1). 289 people were assigned 60
to nicotine e-cigarettes, 295 to patches, and 73 to placebo
e-cigarettes. Participants baseline characteristics were
evenly balanced between treatment groups (table 1). 40
Overall, loss to follow-up was 22%: 17% (48 of 289) in the
nicotine e-cigarettes group, 27% (80 of 295) in the patches
group, and 22% (16 of 73) in placebo e-cigarettes group. 20

Veried continuous abstinence at 6 months after quit

day was highest in the nicotine e-cigarettes group (73%),
followed by the patches group (58%), and placebo 0
0 50 100 150 200
e-cigarettes group (41%; tables 2, 3). Achievement of Duration between quit date and relapse date (days)
Number at risk
abstinence was substantially lower than we anticipated, Nicotine EC 289 108 77 64 5
thus we had insucient statistical power to conclude Patches 295 68 51 43 5
Placebo EC 73 21 16 13 2
superiority of nicotine e-cigarettes to patches or to
placebo e-cigarettes. 7 day point prevalence abstinence Figure 2: Kaplan-Meier analysis of time to relapse
was closer to our estimate of 20%, and the RR suggested EC=e-cigarettes.

www.thelancet.com Published online September 7, 2013 http://dx.doi.org/10.1016/S0140-6736(13)61842-5 5

 Articles

a dierence in favour of nicotine e-cigarettes, but was not signicantly higher than in the placebo e-cigarettes
signicant at 6 months. Repeated measures analyses at 1 group (45%; p=008).
month and overall also showed a benet of nicotine Over 6 months, AUTOS scores in the e-cigarettes
e-cigarettes compared with patches (table 2). However, groups halved from baseline compared with a decrease
both the point prevalence and repeated measures analy- of a third in the patches group (data not shown). The
ses used self-reported cessation. Subgroup analyses dierence between the nicotine e-cigarettes group and
stratied by sex or ethnicity showed no signicant patches group in total AUTOS score reduction from
dierences in primary outcome (data not shown). baseline to 6 months was signicant (156, p=002), but
Quit rates were initially high then decreased in all the dierence between the nicotine e-cigarettes group
groups (gure 2). Most participants relapsed within and placebo e-cigarettes group was not signicant (134,
50 days. Among those who relapsed, median time to p=019). Behavioural dependence, as measured by GN-
relapse in the nicotine e-cigarettes group was 35 days SBQ, was balanced at baseline, with 36% (105 of 289) of
(95% CI 1556), more than twice as long as in the patches participants in the nicotine e-cigarettes group, 37%
group (14 days, 95% CI 818, p<00001) or placebo (109 of 295) in the patches group, and 42% (31 of 73) in
e-cigarettes group (12 days, 534, p=009). Mean cigarette the placebo group scoring strong or very strong
consumption decreased by two cigarettes per day more dependence, but we identied no association between
in the nicotine e-cigarettes group than the patches group score and outcome (data not shown).
(p=0002; table 4). In the nicotine e-cigarettes group, A higher number and proportion of adverse events
57% of participants reduced daily cigarettes by at least occurred in the nicotine e-cigarettes group than in the
half at 6 monthsa signicantly greater proportion than patches group (table 5); however, we identied no
in the patches group (41%; p=00002) and non- evidence of an association with study product, and the
event rate was not signicantly dierent (incidence rate
Nicotine Patches Dierence ratio for nicotine e-cigarettes vs patches 105, 95% CI
e-cigarettes (nicotine e-cigarettespatches) 082134, p=07).
Mean SE Mean SE Mean SE p value Adherence to study treatments was signicantly higher
in the nicotine e-cigarettes group compared with the
Overall 111 04 91 04 20 05 <00001
patches group (p<00001 at each follow-up assessment)
1 month 129 04 105 04 24 06 <00001
and with the placebo e-cigarettes group (p<00001 at each
3 months 108 04 91 04 17 06 0006
follow-up assessment): at 1 month post quit day, 78% (203
6 months 97 04 77 04 19 06 0002
of 260) of participants in the nicotine e-cigarettes group
*For those reporting smoking at least one cigarette in past 7 days. and 82% (51 of 62) of those in the placebo e-cigarettes
group were using the allocated product, compared with
Table 4: Change from baseline in cigarettes consumed per day during follow-up period, nicotine
e-cigarettes and patches* 46% (107 of 232) of those allocated to patches. By
3 months, 51% (126 of 245) participants in the nicotine
e-cigarettes group and 53% (31 of 59) of those in the
Nicotine e-cigarettes Patches Placebo e-cigarettes placebo e-cigarettes group were still using allocated
N % N % N % treatments, compared with only 18% (40 of 224) of those
Total 137 100% 119 100% 36 100%
in the patches group; at 6 months, 29% (71 of 241) of the
Event type
nicotine e-cigarettes group and 35% (20 of 57) of the
placebo e-cigarettes group persisted with e-cigarette use,
Serious* 27 197% 14 118% 5 139%
with only 8% (17 of 215) of those in the patches group still
Any non-serious event 110 803% 105 882% 31 861%
using patches. Among those in the nicotine e-cigarettes
Relation to study treatment
group veried as abstinent, 38% (eight of 21) still used
Denitely 0 1 08% 0
e-cigarettes at 6 months; among non-quitters, 29% (63 of
Probably 1 07% 1 08% 1 28%
220) still used e-cigarettes (whether nicotine e-cigarettes
Possibly 5 36% 4 34% 1 28%
or placebo e-cigarettes is unclear). Since average daily use
Unrelated 131 956% 113 950% 34 944%
was low, some participants could have been using
107 participants in the nicotine e-cigarettes group had a total of 137 events. 96 participants in the patches group had a cartridges allocated at randomisation, others might have
total of 119 events. 26 participants in the placebo group had a total of 36 events. Event rate was 08 events per person purchased cartridges online. Participants using nico-
month in nicotine e-cigarettes group and patches group, and 09 in placebo e-cigarettes group. The dierence
between the rates in the nicotine e-cigarettes group and patches group were not signicant (incidence rate ratio 105,
tine e-cigarettes reported having used an average of
95% CI 082134, p=07). *Serious adverse event by convention includes: death (n=1, in nicotine e-cigarettes group), 13 cartridges per day at 1 month, 11 per day at 3 months,
life threatening illness (n=1, in nicotine e-cigarettes group), admission to hospital or prolongation of hospital stay and 07 per day at 6 months; in the placebo group
(12% of all events in nicotine e-cigarettes group, 8% in patches group, and 11% in placebo e-cigarettes group),
participants reported using 11 cartridges per day at
persistent or signicant disability or incapacity, congenital abnormality, medically important (6% of all events in
nicotine e-cigarettes group, 4% in patches group, and 3% placebo e-cigarettes group). No serious adverse events in 1 month, 12 per day at 3 months, and 07 per day at
any groups were related to product use. 6 months. Nicotine patches were used as instructed (an
average of one per day). Few participants used other
Table 5: Adverse events by type (serious or non-serious) and relation to study treatment
cessation products: at 6 months, in both the nicotine

6 www.thelancet.com Published online September 7, 2013 http://dx.doi.org/10.1016/S0140-6736(13)61842-5

 Articles

e-cigarettes group and patches group, two participants

had used bupropion and ve had used varenicline in the Panel: Research in context
past month; in the placebo e-cigarettes group, three Systematic review
participants reported using varenicline. We searched Medline, PsycINFO, CINAHL, Embase, and the Cochrane library using the
Quitline support was accessed by fewer than half of terms e-cig* OR elect* cigar* OR electronic nicotine, for reports published between
participants: 40% (115 of 289) in the nicotine e-cigarettes Jan 1, 2005, and Aug 23, 2013. The strategy identied 186 articles, of which only one was a
group, 36% (106 of 295) in the patches group, and 36% randomised, placebo-controlled trial with a cessation endpoint measured at 6 months or
(26 of 73) in the placebo e-cigarettes groups, but a post- more.9 This previous trial,9 done between 2011 and 2012, recruited 300 adult Italian
hoc analysis showed no benet of use of support on smokers unwilling to quit, with 100 randomised to each of three groups: 72 mg nicotine
the primary outcome for participants in the nicotine cartridges for 12 weeks, 6 weeks of 72 mg cartridges followed by 6 weeks of 54 mg
e-cigarettes group (p=067) or patches group (p=016). cartridges, and 0 mg nicotine cartridges for 12 weeks. No behavioural support was
There was sustained enthusiasm for e-cigarettes: at provided but nine follow-up visits occurred, with carbon monoxide measures at each. The
1 month, 88% (230 of 260) of participants in the nicotine primary outcome was not clearly prespecied nor were calculations done to estimate
e-cigarettes group, and 92% (57 of 62) in the placebo power. Analysis was by intention to treat. At 12 months, 39% of participants had been lost
e-cigarettes group stated that they would recommend to follow-up, a potential source of bias. Of those assessed, 9% had quit (13%, 9%, and 4%
their allocated product to a friend wanting to quit, in the two nicotine e-cigarettes groups and placebo e-cigarettes groups, respectively) and
compared with 56% (130 of 232) of those in the patches reduction occurred in 10%, 9%, and 12%; none of the comparisons were statistically
group; at 6 months the gures changed little, being 85% signicant. The reliability of e-cigarettes was problematic. These results are much the same
(205 of 241), 88% (50 of 57), and 50% (107 of 215), as those reported in previous trials of unsupported pharmacotherapy with patches32 and
respectively. Among participants allocated to e-cigarettes, are similar to our trial ndings.
40% (96 of 241) liked their tactile, cigarette-like qualities,
sensory familiarity, perceived health benets, taste, Interpretation
absence of cigarette odour, and ease of use. In our study, e-cigarettes, with or without nicotine, were modestly eective at helping
smokers to quit. Nicotine e-cigarettes might be more eective or of similar eectiveness
Discussion to patches, but so far studies have not had sucient statistical power to draw more
13 weeks of nicotine e-cigarette use resulted in increased denitive conclusions. E-cigarette use was associated with few adverse events, similar to
smoking abstinence at 6 months compared with use of patches, but longer-term data are needed. Uncertainty exists about the place of
patches or placebo e-cigarettes, but these dierences were e-cigarettes in tobacco control, and more research is urgently needed to clearly establish
not statistically signicant. Nevertheless, the results were their overall benets and harms at both individual and population levels.
consistent across a range of analyses, and the 95% CIs do
not exclude an advantage. In post-hoc analyses using a
5% non-inferiority limit for the risk dierence (on the patches. Those who reported previously trying to quit
basis of a margin used in our non-inferiority smoking with patches or other forms of NRT (about 20% in the
cessation trial of cytisine26), nicotine e-cigarettes were at past year in each group) might have disadvantaged
least as eective as patches (the absolute risk dierence patches (by being more likely to give up on patches
for the primary outcome was 151 [95% CI 249 to 551]; subsequently); however, at 6 months the dierence
249 is within the margin of 5). Therefore, we conclude between the results of the intention-to-treat analysis
that among smokers wanting to quit, nicotine e-cigarettes and per-protocol analysis was minimal, suggesting this
might be as eective as patches for achieving cessation at bias was not a major issue.
6 months. We identied no dierence in adverse events Third, the modest abstinence rate for nicotine
with e-cigarettes compared with patches. e-cigarettes is much the same as quit rates shown in
The strengths of our study include use of a con- studies of NRT products used without behavioural
servative primary outcome measure, and rigorous trial support.27 Addition of more intensive support might have
conduct to mitigate risk of bias. We used a pragmatic improved quit rates, but it would also have misrepresented
design because we believe that an assessment of real- the typically low support environment in which most
world eectiveness of e-cigarettes is a priority for policy e-cigarette users attempt to quit. The modest abstinence
development, although it could be argued a trial of a rates might have been compounded by inadequate nico-
novel intervention should be more explanatory than tine replacement: as noted, the cartridges contained less
pragmatic in design. Our study had several limitations. nicotine than labelled, and delivery was inecient (not
First, the eect size and estimates of abstinence on uncommon in other early e-cigarette models15,28). Further-
which the study sample size was calculated were more, users consumed on average just over one cartridge
optimistic; hence, statistical power to detect dierences per day, delivering around only 20% of the nicotine
was reduced. Second, participants assigned to patches obtained from cigarette smoking.29 Although trials of the
had a higher loss to follow-up and withdrawal rate than eects of early e-cigarettes on withdrawal relief showed
those assigned to e-cigarettes. Some of the participants that low levels of nicotine delivery attenuated with-
might have agreed to take part in the study to try drawal symptoms,7,8 improved nicotine delivery by newer
e-cigarettes, and then lost interest when randomised to models of e-cigarettes provides greater withdrawal relief,

www.thelancet.com Published online September 7, 2013 http://dx.doi.org/10.1016/S0140-6736(13)61842-5 7

 Articles

potentially enhancing cessation eectiveness.8 Trials of Acknowledgments

such second generation e-cigarettes are needed. The e-cigarettes and cartridges were Elusion brand products provided by
PGM International, New Zealand. PGM International had no role in the
We included the placebo e-cigarettes group to explore study design, data collection, data analysis, data interpretation, or writing
the role of behavioural replacement by e-cigarettes, of this report. We thank the participants, research assistants, our
independent of nicotine delivery in cessation.30 How- colleagues, the Health Research Council of New Zealand, PGM
ever, our study was underpowered to detect the small International, and New Zealand Quitline.
eect, and the GN-SBQ instrument, which purports to References
measure behavioural dependence but has not been 1 Purkayastha D. BAT ramps-up e-cigarette expansion as sales go up
in smoke international business times (July 31, 2013). http://www.
widely used in this context, might have been inadequate thefreelibrary.com/BAT Ramps-up E-cigarette Expansion as Sales
for this purpose. Go Up in Smoke.-a0338323170 (accessed Aug 13, 2013).
A third of the participants allocated to the e-cigarettes 2 West R. Smoking toolkit study: monthly tracking of key
performance indicators (July 20, 2012). http://www.
groups reported continued product use at 6 months, smokinginengland.info/latest-statistics/ (accessed Aug 9, 2013).
suggesting that they might have become long-term 3 Hajek P, Foulds J, Houezec JL, Sweanor D, Yach D. Should
e-cigarette users. Those who had relapsed to smoking e-cigarettes be regulated as a medicinal device? Lancet Respir Med
2013; 1: 42931.
but continued to use e-cigarettes (so called dual use) at 4 Cobb N, Cobb C. Regulatory challenges for rened nicotine
6 months had reduced cigarette consumption. Research products. Lancet Respir Med 2013; 1: 43133.
has shown higher cessation rates in people using NRT 5 Etter J-F. Electronic cigarettes: a survey of users. BMC Public Health
2010; 10: 231.
while still smoking;31 if e-cigarettes act in the same way
6 Dawkins L, Turner J, Roberts A, Soar K. Vaping proles and
this would be a positive feature. Further research is preferences: an online survey of electronic cigarette users. Addiction
needed to explore this area. 2013; 108: 111525.
Finally, as far as we are aware, our trial provides for the 7 Bullen C, McRobbie H, Thornley S, Glover M, Lin R, Laugesen M.
Eect of an electronic nicotine delivery device (e cigarette) on desire
rst time adverse event information for 657 people to smoke and withdrawal, user preferences and nicotine delivery:
randomly allocated to e-cigarettes or patches. The nding randomised cross-over trial. Tob Control 2010; 19: 98103.
of no signicant dierences in occurrence of adverse 8 Vansickel A, Eissenberg T. Electronic cigarettes: eective nicotine
delivery after acute administration. Nicotine Tob Res 2013;
events between groups over the duration of a standard 15: 26770.
NRT treatment course, and the further 3 months 9 Caponnetto P, Campagna D, Cibella F, et al. Eciency and safety of
monitoring, suggests such short-term e-cigarette use is an electronic cigarette (ECLAT) as tobacco cigarettes substitute:
a prospective 12-month randomized control design study. PloS One
of low risk. However, longer-term use requires more 2013; 8: e66317.
research (panel). 10 US Food and Drug Administration (FDA). Summary of results:
Our study has established benchmarks for performance laboratory analysis of electronic cigarettes conducted by FDA. http://
www.fda.gov/NewsEvents/PublicHealthFocus/ucm173146.htm
of nicotine e-cigarettes relative to NRT and placebo (accessed Aug 9, 2013).
e-cigarettes with which to design future, more adequately 11 Goniewicz M, Knysak J, Gawron M, et al. Levels of selected
powered trials. Our ndings point to potential for carcinogens and toxicants in vapour from electronic cigarettes.
e-cigarettes in regard to cessation eectiveness beyond Tob Control 2013; 6: 6.
12 Burstyn I. Peering through the mist: what does the chemistry of
that noted in the present study. Furthermore, because they contaminants in electronic cigarettes tell us about health risks?
have far greater reach1,2 and higher acceptability (as shown Technical report. http://publichealth.drexel.edu/SiteData/docs/
by the present study) among smokers than NRT, and ms08/f90349264250e603/ms08.pdf (accessed Aug 13, 2013).
13 Bullen C, Williman J, Howe C, et al. Study protocol for a
seem to have no greater risk of adverse eects, e-cigarettes randomised controlled trial of electronic cigarettes versus nicotine
also have potential for improving population health. patch for smoking cessation. BMC Public Health 2013; 13: 210.
Contributors 14 Heatherton T, Kozlowski L, Frecker R, Fagerstrom K. The
Fagerstrom test for nicotine dependence: a revision of the
CB, NW, HM, and ML conceived the original idea for the trial, and
Fagerstrom tolerance questionnaire. Br J Addict 1991; 86: 111927.
sought and obtained funding. CB, NW, HM, ML, CH, VP, and JW wrote
15 Goniewicz M, Kuma T, Gawron M, Knysak J, Kosmider L. Nicotine
the study protocol. CH managed the day-to-day running of the trial,
levels in electronic cigarettes. Nicotine Tob Res 2013; 15: 15866.
including all participant follow-up. VP did the data analyses. This Article
16 Price E, Allen M. New Zealand: eective access to tobacco
was written by CB with input from all coauthors. CB is guarantor for
dependence treatment. WHO, 2003. http://www.who.int/tobacco/
this Article. All authors read and approved the nal version. research/cessation/en/best_practices_new_zealand.pdf (accessed
Conicts of interest Sept 4, 2013).
We declare that we have received no support from any companies for the 17 Stead LF, Perera R, Bullen C, et al. Nicotine replacement therapy
submitted work and have no non-nancial interests that might be relevant for smoking cessation. Cochrane Database Syst Rev 2012;
to the submitted work. ML, via his company Health New Zealand, 11: CD000146.
previously did research funded by Ruyan (an e-cigarette manufacturer). CB 18 Ministry of Health. New Zealand smoking cessation guidelines.
and HM have done research on Ruyan e-cigarettes funded by Health New Wellington: Ministry of Health, 2007.
Zealand, independently of Ruyan. HM has received honoraria for speaking 19 DiFranza J, Wellman R, Ursprung W, Sabiston C. The autonomy
at research symposia, has received benets in kind and travel support over smoking scale. Psychol Addict Behav 2009; 23: 65665.
from, and has provided consultancy to, the manufacturers of smoking 20 Glover E, Nilsson F, Westin A, Glover P, Lain M, Persson B.
cessation drugs. NW has provided consultancy to the manufacturers of Developmental history of the Glover-Nilsson smoking behavioral
smoking cessation drugs, received honoraria for speaking at a research questionnaire. Am J Health Behav 2005; 29: 44355.
meeting and received benets in kind and travel support from a 21 West R, Hajek P, Stead L, Stapleton J. Outcome criteria in smoking
cessation trials: proposal for a common standard. Addiction 2005;
manufacturer of smoking cessation drugs. JW has provided consultancy to
100: 299303.
the manufacturers of smoking cessation medications.

8 www.thelancet.com Published online September 7, 2013 http://dx.doi.org/10.1016/S0140-6736(13)61842-5

 Articles

22 Ursprung W, Morello P, Gershenson B, DiFranza J. Development 28 Polosa R, Morjaria J, Caponnetto P, et al. Eectiveness and
of a measure of the latency to needing a cigarette. J Adolesc Health tolerability of electronic cigarette in real-life: a 24-month prospective
2011; 48: 33843. observational study. Intern Emerg Med 2013; published online
23 Fiore M, Jaen C, Baker T, et al. Treating tobacco use and July 20. DOI:10.1007/s11739-013-0977-z.
dependence: 2008 update. Rockville, MD: US Department of Health 29 Mariner D, Ashley M, Shepperd C, Mullard G, Dixon M. Mouth
and Human Services, Public Health Service, 2008. level exposure using analysis of lters from smoked cigarettes:
24 Schulz KF, Altman DG, Moher D. CONSORT 2010 Statement: A study of eight countries. Regul Toxicol Pharmacol 2011;
updated guidelines for reporting parallel group randomised trials. 61: S3950.
BMJ 2010; 340: c332. 30 Caponnetto P, Cibella F, Mancuso S, Campagna D, Arcidiacono G,
25 Cormack D, Robson C. Classication and output of multiple Polosa R. Eect of a nicotine-free inhalator as part of a
ethnicities: considerations for monitoring Mori health. Wellington: smoking-cessation programme. Eur Respir J 2011; 38: 100511.
Te Rp Rangahau Hauora a Eru Pmare, 2010. 31 Fagerstrom K, Tejding R, Westin A, Lunell E. Aiding reduction of
26 Walker N, Howe C, Bullen C, et al. Study protocol for a smoking with nicotine replacement medications: hope for the
non-inferiority trial of cytisine versus nicotine replacement therapy recalcitrant smoker? Tob Control 1997; 6: 31116.
in people motivated to stop smoking. BMC Public Health 2011; 32 Stead L, Perera R, Bullen C, et al. Nicotine replacement therapy for
11: 880. smoking cessation. Cochrane Database Syst Rev 2012; 11: CD000146.
27 Shiman S, Rolf C, Hellebusch S, et al. Real-world ecacy of
prescription and over-the-counter nicotine replacement therapy.
Addiction 2002; 97: 50516.

www.thelancet.com Published online September 7, 2013 http://dx.doi.org/10.1016/S0140-6736(13)61842-5 9