The effect of e-cigarettes on smoking cessation and cigarette smoking initiation: An evidence-based rapid review and meta-analysis

INTRODUCTION The contribution made by e-cigarettes to smoking cessation continues to be controversial. Reports suggest that teenagers are becoming increasingly addicted to e-cigarettes and that e-cigarette use in adolescents is associated with subsequent cigarette smoking. METHODS Systematic searches of eleven databases were conducted (January 2015 to June 2020). Systematic reviews, randomized controlled trials (RCTs) and cohort studies comparing e-cigarettes with placebo e-cigarettes, nicotine replacement therapy (NRT) or no e-cigarette use were included. The two primary outcomes were smoking cessation among smokers and smoking initiation among non-smoking teenagers. The secondary outcome was adverse events. Data were synthesized using risk ratio (RR) or adjusted odds ratio (AOR) with 95% confidence interval (CI). RESULTS Six systematic reviews, 5 RCTs and 24 cohort studies were identified. For smoking cessation, findings from 4 systematic reviews indicated that e-cigarettes contributed to cessation while one found the opposite. Meta-analysis of 5 RCTs suggested that e-cigarettes were superior to NRT or placebo for smoking cessation (RR=1.55; 95% CI: 1.00–2.40; I2=57.6%; low certainty; 5 trials, n=4025). Evidence from 9 cohort studies showed that e-cigarette use was not associated with cessation (AOR=1.16; 95% CI: 0.88–1.54; I2=69.0%; n=22220). Subgroup analysis suggested that intensive e-cigarette use may be associated with cessation. In terms of smoking initiation, adolescents who ever used e-cigarettes had a greater risk for smoking initiation than non-users (AOR=2.91; 95% CI: 2.61–3.23; I2=61.0%; 15 trials, n=68943), the findings were consistent with one included systematic review. No serious adverse events were reported in the included studies. CONCLUSIONS Low certainty evidence suggests that e-cigarettes appear to be potentially effective for smoking cessation. The use of e-cigarettes in adolescents may be associated with smoking initiation. No serious adverse events were reported.


INTRODUCTION
smoking cessation tools 4 . Reports have suggested that achieving abstinence by using e-cigarettes was comparable to nicotine replacement therapy (NRT), and e-cigarettes also appear to effectively decrease cigarette consumption 5,6 . This has resulted in a dramatic increase in the use of e-cigarettes worldwide. The estimated global value of e-cigarette sales was up to US$3.5 billion in 2015 7 . A national large-scale survey from 28 EU member states reported that there were about 48.5 million e-cigarette users in 2016 8 .
Increasingly teenagers are becoming addicted to e-cigarettes. Between 2016 and 2017, the proportion of e-cigarette users among those aged 11 to 16 years increased from 7% to 11%, respectively, across the UK 9 . The proportion of US high school teenagers who had tried e-cigarettes in the last month rose from 1.5% to 16.0% between 2011 and 2015, a more than 10-fold increase 10 . In 2019, it was estimated that among high school students in the US, 27.5% (95% CI: 25.3-29.7) were current e-cigarette users 11 . Worse still, many studies appear to report that use of e-cigarettes by young people increases the risk of subsequent cigarette smoking 12 , and that minors are an emerging new smoking population 13 .
Concerns have been expressed about the safety of e-cigarettes. The US Food and Drug Administration (FDA) reported that e-cigarettes contained some toxic heavy metals and the concentration of heavy metals released by e-cigarettes was much higher than conventional cigarettes 14 . These were associated with the development of cancer 15 and the occurrence of coronary events 16 . In light of the above, the FDA banned flavored e-cigarette sales in January 2020, as flavored e-cigarettes had the most appeal to teenagers 17 . On 1 November 2019, China as the birthplace of e-cigarettes announced that the online purchase of e-cigarettes would be completely banned in China 18 . Previously, two randomized controlled trials (RCTs) 5,6 indicated that e-cigarettes were potentially effective in smoking cessation among adult smokers. Published studies on cigarette smoking initiation associated with e-cigarette use among adolescents have been primary studies and there has been limited evidence synthesis. There are few evidence-based studies that have comprehensively evaluated the benefits and risks of e-cigarettes. Therefore, a rapid review was conducted to evaluate the effects of e-cigarettes on smoking cessation among smokers and the risks for smoking initiation among non-smoking adolescents, and their safety.

METHODS
Due to the urgent need to inform ongoing policy, a rapid review methodology was employed 19 . This type of simplified systematic review is helpful in providing a timely synthesis for decision makers. Streamlined methods are usually used in a rapid review, which usually includes limiting retrieval dates and databases 19 . This rapid review and meta-analysis are reported following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 20 statement. A PRISMA statement with a checklist of items that should be included in reports of systematic reviews is included in the Supplementary file.

Eligibility criteria
Systematic reviews based on RCTs or comparative observational studies (cohort and cross-sectional studies), parallel group RCTs and controlled cohort studies were included. The study population comprised adult smokers and non-smoking adolescents, who had no serious diseases or pregnancy. The study interventions were e-cigarettes. The controls referred to placebo e-cigarettes (without nicotine), NRT or no treatment. The primary outcomes were smoking cessation among smokers and smoking initiation among adolescents. The secondary outcome was occurrence of adverse events. Experimental studies and studies that failed to report the minimal required information were excluded. There was no need for approval from an ethics committee or agreement of participants in this study as data were extracted from publications that were in the public domain. electronic cigarettes, electronic nicotine delivery systems, vape, and vaping. An example of the PubMed search strategy is given in the Supplementary file.

Study selection and data extraction
After removing duplicates, two authors (SJZ and FLB) independently screened studies by title and abstract, according to the eligibility criteria. In the full text screening process, uncertainty and insufficient information was determined for eligibility through full texts. Disagreements were resolved by discussion or arbitrated by a methodologist (JPL). Reasons for excluding studies were recorded at this stage. After identification of studies, data were extracted independently by two authors (SJZ and FLB) and included: characteristics of study design (study type, setting, and funding); details of PICO (participants, interventions, comparisons, and outcomes); and follow-up and adjusted factors. Any disagreements were resolved by discussion or thirdparty adjudication.

Quality assessments
The methodological quality assessment of each included study was performed independently by two review authors (FLB and SJZ). Cochrane Risk of Bias tool (ROB) 21 was employed to assess whether the risk of bias of RCTs was low, high or unclear according to its seven domains. A Measurement Tool to Assess Systematic Reviews (AMSTAR-2) 22 which has 16 items with 7 key items was employed to assess the quality of the included systematic reviews; the overall confidence for each review was assessed as high, moderate, low or critically low. The quality of cohort studies was assessed by Newcastle-Ottawa Scale (NOS) 23 , which is composed of 8 items and includes three subscales: selection of studies, comparability of studies, and the ascertainment of exposure. The maximum score is 9 points and any study scoring >5 points was considered of moderate to high quality 23 . Grading of Recommendations Assessment, Development and Evaluation (GRADE) 24 was employed to assess the certainty of evidence from RCTs in five domains (risk of bias, directness, precision, consistencies, and publication bias).

Data synthesis
Qualitative and quantitative methods were adopted to synthesize the findings. The findings from systematic reviews were narratively described. Data are presented as risk ratios (RRs) or adjusted odds ratio (AORs) with 95% confidence intervals (CIs). Meta-analysis was performed by Stata version 14.0 software when the trials had acceptable heterogeneity and similarities in clinical characteristics. The random effects model (REM) was utilized in the meta-analysis to consider potential sources of clinical heterogeneity. The I 2 statistic was employed to assess statistical heterogeneity 21 . When p<0.10 and I 2 >50%, the heterogeneity may be considered as high 21,25 . To explain heterogeneity, we predefined the subgroup analysis by the frequency of e-cigarette use: intensive use (daily or regular use for at least one month) versus intermittent or irregular use. Sensitivity analysis was employed to test the robustness of the results when the primary outcomes were statistically different for the following three methodological domains: reported clear randomization concealment or not; placebo used or not; and reported lost to follow-up or not. Funnel plots were employed to detect the possibility of publication bias if ≥10 studies were included in a meta-analysis.

Screening
Initially, 1620 records were retrieved and 926 duplicates were removed. In all, 612 records were excluded by scanning the title and abstract due to irrelevant studies, protocols, clinical studies, casecontrol studies or cross-sectional surveys, commentary of included studies, and studies that provided only abstracts. This left 82 remaining records, of these, 47 studies were excluded through full-text screening due to ineligible study design, uncontrolled cohort studies or lack of cessation outcomes. Finally, 35 studies were included after full-text screening. Of these studies, findings from 6 systematic reviews [26][27][28][29][30][31] were narratively described. The screening process is shown in Figure 1.

Description of studies
The characteristics of the included studies are shown in Tables 1 and 2. Included in the 35 studies identified were: 6 systematic reviews 26-31 , 5 RCTs 32-36 , and 24 cohort studies  . These studies were classified into two categories based on the two primary outcome measures. Nineteen studies [26][27][28][29][30][32][33][34][35][36][37][38][39][40][41][42][43][44][45] examined the effects of e-cigarette use on smoking cessation, and the remaining 16 studies 31,46-60 explored the causality between e-cigarette use and smoking initiation among teenagers. The study population included adult smokers and non-smoking adolescents and young adults. The definition of e-cigarette use was 'current use', 'ever use' or 'past 30-day use'. In terms of the included 6 systematic reviews, 5 reviews 26-30 focused on smoking cessation among adult smokers, and included either RCTs, cohort studies or cross-sectional studies. Another systematic review 31 was based on 9 cohort studies and explored the association between e-cigarette use and subsequent smoking initiation.

Quality assessment
There were no 26 or one 28 non-critical weakness in two systematic reviews, and these were rated as high confidence according to AMSTAR-2. Three 29-31 were rated as low confidence due to the lack of a protocol and inadequate details of the included studies. One 27 was of critically low confidence as it lacked a protocol and the lists of excluded studies. Details are shown in Table 3. In terms of the RCTs, the random sequence generation and allocation concealment for 3 RCTs 32-34 were well described and were judged as having a low risk of bias. Blinding of the participants and personnel was only reported for one RCT 33 and was rated as 'low' in the blinding domain. Apart from this, the risk of performance bias was rated as 'high' in the other 4 RCTs 32,34-36 , but the outcome assessment was blinded in these 4 RCTs and were rated as having low risk of detection bias. Intention-to-treat analysis was conducted in 5 RCTs 32-36 , and the attrition rate was <10%; all RCTs were assessed as having low risk of attrition bias. In terms of reporting bias, 3 RCTs 34-36 were assessed as 'low' since the protocols were registered and the consistency between the outcomes described in protocol and actual outcomes in the results; the remaining 2 RCTs 32,33 were assessed as 'unclear' due to lack of a registered protocol. The baseline data were comparable for 4 RCTs 32,34-36 , one 35 was rated as 'unclear' due to the lack of baseline data. The risk of bias summary is shown in Figure 2. The quality of 24 cohort studies scored between 5 and 7 according to the NOS grading due to the higher number of dropouts and self-reported outcomes. Overall, the quality was considered to be satisfactory. Details are shown in Table 4.

Effects of interventions Smoking cessation
Findings from 5 systematic reviews A Cochrane systematic review 28 compared e-cigarettes with placebo e-cigarettes on smoking cessation, but only 2 RCTs 5,6 were eligible for the meta-analysis. The abstinence rates for at least 6 months were 4% (placebo group) and 9% (e-cigarette group), suggesting that e-cigarettes were more likely to favor cessation (RR=2.29; 95% CI: 1.05-4.96; low certainty; 2 trials, n=662). This was consistent with 3 systematic reviews 27,29,30 . Another systematic review 26 reported that the odds of cessation were 28% lower in the e-cigarette group than the non-use group (OR=0.72; 95% CI: 0.57-0.91; 20 trials, n=355011).

Smoking initiation
Findings from one systematic review Only one systematic review 31 based on cohort studies involving 17389 young people aged 14-30 years was identified. This review indicated that ever e-cigarette users were more likely to initiate cigarette smoking at follow-up than never users (23.2% vs 7.2%) (AOR=3.5; 95% CI: 2.38-5.16; I 2 =56.0%; 7 trials, n=8759).  Findings from 15 cohort studies Fifteen cohort studies 46-60 were included. The pooled results suggested that ever e-cigarette users were more likely to initiate smoking than non-e-cigarette users (AOR=2.91; 95% CI: 2.61-3.23; I 2 =61.0%; 15 trials, n=68943) (Supplementary file, Figure S1). Subgroup analysis on the frequency of e-cigarette use was not available, since only one trial 52 reported the AOR of intensive and intermittent e-cigarette use. Sensitivity analysis was also unavailable due to insufficient data. The funnel plot based on the AORs of smoking initiation (Supplementary file, Figure S2) appeared to be asymmetrical, suggesting that there was potential publication bias.

Adverse events
Of the 35 included studies, adverse events were reported in 5 studies [27][28][29]32,34 . Two systematic reviews 27,28 reported minor adverse short-term events related to e-cigarettes. These included irritation of the mouth and throat, cough and respiratory diseases, but long-term safety is still unknown. Two studies 29,32 reported that there was no statistical difference in adverse events between the e-cigarette group and control group. One RCT 34 reported that adverse events were significantly lower in the e-cigarette group (6.7%) compared with the NRT group (17.3%); adverse events included oral pain, cough, headache, and nausea. No serious adverse events were reported in the included trials.

Main findings
In the light of current policies on e-cigarettes in China and the US, it was timely to conduct a rapid review to comprehensively evaluate the benefits and risks of e-cigarettes. Thirty-five studies published from January 2015 to June 2020 were evaluated and included 6 systematic reviews, 24 cohort studies and 5 RCTs. The study population included adult smokers with or without intention to cease smoking and non-cigarette smoking adolescents. Four systematic reviews [27][28][29][30] indicated that e-cigarettes were superior to placebo for cigarette smoking cessation. Another systematic review 26 reported that the odds of cessation were 28% lower in the e-cigarette group than the nonuse group. Five RCTs (two 33,36 placebo or no treatment controlled and three 32,34,35 NRT controlled) reported that the abstinence rate in the e-cigarette group was 2.6% higher than that in the control group, suggesting that e-cigarettes may be more effective than NRT or placebo in achieving smoking cessation among adult smokers. However, this evidence was downgraded due to the small number of events and inconsistencies as assessed using GRADE. Nine cohort studies involving 22220 adult smokers found that e-cigarettes were not superior to non-e-cigarette use for cessation. Subgroup analysis suggested that intensive e-cigarette use (daily or regular use for at least one month) may contribute to cessation, while intermittent or irregular use did not. In light of the limited number of RCTs and the findings from cohort studies on smoking cessation, we could not draw robust conclusions from current evidence. In terms of smoking initiation, one systematic review 31 reported that ever or past 30-day e-cigarette use was more likely to initiate smoking among teenagers. Updated estimates (AORs) of 15 cohort studies 46-60 involving 68943 adolescents suggested that ever e-cigarette users were nearly 3 times more likely than non-users to begin smoking cigarettes. However, the evidence was limited due to potential bias (self-reported outcomes, high dropouts and variations in the length of follow-up). No serious adverse events were reported in the included studies.

Comparisons with other studies
Previous systematic reviews [26][27][28][29][30] have merely focused on the benefits of e-cigarettes for smoking cessation or the risks for smoking initiation 31 . Former systematic reviews were based on primary studies such as RCTs and cohort studies. We identified the latest systematic reviews, RCTs and cohort studies published in the last six years to provide comprehensive and rapid evidence for policy makers. This is the first rapid review to evaluate the benefits and risks of e-cigarettes, comprehensively. A newly published Cochrane systematic review 61 found that nicotine e-cigarettes were superior to placebo e-cigarettes or NRT for smoking cessation. This was consistent with the findings from 5 newly conducted RCTs 32-36 in this rapid review. In addition, this review conducted a subgroup analysis based on the frequency of e-cigarette use.

Implications
For the study design, future studies should consider the frequency of e-cigarette use, the concentration of nicotine in e-cigarettes and the type of e-cigarettes, since the quitting or initiating smoking effects of e-cigarette use could be potentially affected by those factors. E-cigarette products change very rapidly, especially the Juul type e-cigarettes, which use nicotine salt and can deliver a higher dose of nicotine 62 . Juul is the most popular e-cigarette in the US 63 . Therefore, it is essential to study new products in the future in terms of their effect on quitting or initiating cigarettes. The definition of smoking cessation or smoking initiation needs to be clarified, the current standards suggest that the duration is for at least 30 days cessation or smoking initiation at follow-up. Likewise, the effects of electronic cigarettes on smoking cessation among smokers with long-term or short-term smoking may vary. Hence, different smoking durations among smokers should be well reported and stratified in future studies. In terms of outcomes, objective measurements of smoking initiation rather than self-reported measurements are warranted. Cotinine as a biomarker 64 of smoke exposure can be used to predict smoking initiation among adolescents. Additionally, dropouts should be minimized as much as possible in future cohort studies. Last but not least, the long-term safety of e-cigarette use should also be a future focus. For policy makers, appropriate e-cigarette use may be potentially effective in smoking cessation for adult smokers. However, e-cigarette use in adolescents was potentially associated with subsequent smoking initiation, and therefore the sale of e-cigarettes to minors should be banned completely worldwide.

Strengths and limitations
Rapid review methodology was employed to provide timely evidence on e-cigarette use due to the urgent need to inform ongoing government policies. This is the first rapid review to evaluate the benefits and risks of e-cigarettes, comprehensively. The latest eligible systematic reviews, RCTs and cohort studies were all identified in order to provide a comprehensive evaluation on e-cigarette use. However, there are several limitations. First, similar to other rapid reviews, the search time was limited from January 2015 to June 2020, and some evidence may have been missed. The results for smoking initiation were at risk of publication bias. Second, though the evidence from CONCLUSIONS Low certainty evidence suggests that e-cigarettes appear to be potentially effective in smoking cessation for adult smokers. However, this beneficial effect needs to be further confirmed in large sample, well designed and fully reported trials. The use of e-cigarettes in adolescents may be associated with subsequent smoking initiation. No serious adverse events were reported in the included studies, however, the long-term safety of e-cigarettes should also be a future focus.