Effect of unguided e-cigarette provision on uptake, use, and smoking cessation among adults who smoke in the USA: a naturalistic, randomised, controlled clinical trial

Study design

The trial recruited adult smokers in 11 cities across the US and randomized them to receive a free 4- week supply of e-cigarettes (n = 427) or not (n = 211). No additional product was offered beyond one month, though e-cigarette use was measured throughout follow-up, which occurred via phone surveys occurred at Weeks 1, 2, 3, 4, 8, 12 and 24. E-cigarette devices and tanks were purchased directly from NJoy with grant funds. All study procedures were approved by the Medical University of South Carolina (MUSC) Institutional Review Board (IRB) and was registered at ClinicalTrials.gov (NCT03453385). Findings below are presented in concordance with CONSORT reporting guidelines.

Participants

Eligibility criteria included: a) age 21+ (abiding by tobacco sales requirements), b) current smoker of ≥5 cigarettes per day for ≥1 year, c) no known recent history of cardiovascular distress, d) neither pregnant nor breastfeeding, and e) no current use of smoking cessation medication. Inclusion was further restricted to participants with minimal history of e-cigarette use, defined as not having purchased e-cigarettes in the prior six months, nor having ever used a tank system weekly, nor having used any e-cigarette of any kind in the preceding six months. Participants were recruited through online methods (e.g., Craigslist), which provided a direct link to an secure online platform (e.g., REDCap) for initial screening. Upon initial eligibility, individuals provided their contact information and were subsequently consented through one of two channels, available upon preference. The first option allowed for mailed consents (postage-paid return reply); the second was via synchronous teleconsent; both provided opportunity to discuss study details and ask any questions with study staff prior to consent. Upon receipt of the mailed consent packet or completion of teleconsent, participants were scheduled for a baseline (Week 0) intake call, during which they were randomized to assigned group. As a remote trial, the unique details of screening and consent are provided elsewhere.¹⁶

Among the 11 cities selected for participant recruitment, the local Charleston SC metro area was included, with the intent to recruit a sub-sample of participants (n = 120) who could come into the research lab to provide biochemical verification of smoking behavior. However, the COVID-19 pandemic significantly compromised these plans. University shut-downs and delayed re-opening disallowed many in-person study visits, and collection of biomarkers was not feasible as intended. Thus, while the trial included a local sub-sample, there was insufficient data to report on biomarker outcomes (57 biospecimen samples at Week 4, and 39 at Week 24, across both groups).

Randomization

Randomization occurred via stratified, mixed block design, stratifying on local vs. national enrollment and on desire to quit smoking: 0–6 vs. 7–10 on a VAS scale. Randomization was 2:1 (e-cigarette:control) to increase the precision of parameter estimates around e-cigarette product uptake. The randomization allocation was created by the study statistician and uploaded into REDCap so that the research staff team was blinded to the sequence and could randomize individuals as they were eligible for participation. Randomization did not include any masking.

Procedures

The general aim within the e-cigarette group was to approximate the real-world scenario in which smokers are exposed to e-cigarette products and self-decide if and how to use them. The guiding intent was to use the most efficient, non-modifiable product on the market at the time of study initiation, with the goal of optimizing user satisfaction and avoiding the possibility of tank adulteration [spilling, adding more]. NJoy's pre-filled tank and battery was selected for this purpose, coinciding with the FDA decision to use it as a prototype for a Standardized Research E-Cigarette (SREC) for research purposes. The NJoy device was a closed tank system, sufficiently powered (1000 mAh) with 3 ml pre-filled nicotine (15 mg/ml). Participants could choose up to 2 flavors among 5 offered: tobacco, menthol, blue/blackberry (one flavor), apple melon, or iced fruit, and could change selection at 2nd shipment.

The product was provided in two ∼2-week shipments: 30 tanks total, under the rationale that one month would allow for an adequate acclimation period in which participants could determine if and how they would use it. In rare instances where shipments were not received (n = 14; 3%), additional product was sent in deviation from the protocol. Adjunctive messaging, provided via phone at baseline call and within mailing of product, gave cursory background and suggestions on how it could be used but emphasized that choice on if or how to use e-cigarettes was completely up to the participant. Full messaging is provided within Supplemental Materials, but specific excerpts include:

• We will provide an e-cigarette device for you to use over the next four weeks. We'll send you everything you need via mail, and you'll receive it in a few days. … The most important thing to tell you now is that you can use these e-cigarettes however you like, as much or as little as you like. You can reduce your smoking, quit smoking entirely, or just use e-cigarettes in places where you cannot smoke … You are not required to use the e-cigarettes, but you might want to give them a try.

Participants were also told that e-cigarettes were ‘likely safer’ than combustible cigarettes (the most conservative, IRB-approved messaging at the time of study onset), and, in response to e-cigarette or vaping use-associated lung injury (EVALI) which arose as a concern midway through the study, brief caution was expressed. While provision of free product is not naturalistic, the trial's design allowed for the assessment of naturalistic outcomes of product use when cost is not a barrier.

Participants in the no-product control group did not receive any product but were compensated nominally more for study participation than the e-cigarette group ($320 vs. $240) to offset concerns about attrition. Note that a placebo comparison is specifically not warranted in a naturalistic product use design. A study of e-cigarette provision, inclusive of any expectancy effects, is properly controlled via no provision of product.

Outcomes

The study's primary purpose was to assess naturalistic uptake and usage of the e-cigarette. Thus, descriptive information is provided in regard to: a) any use and b) number of days of use and c) puffing episodes on days of use, each assessed weekly for the first month and then episodically thereafter. Puffing episodes were guided by precedent from other studies,¹⁷ and explained to participants as “the number of times per day that you use the e-cigarette (one time consists of about 15 puffs or 10 min of use).” While these uptake outcomes pertain primarily to the e-cigarette group, control participants could also use e-cigarettes, on their own accord, and thus, these outcomes are reported as well.

While not a cessation-driven trial, the study was powered on these outcomes under the rationale that: a) cessation might naturally occur, b) it would be missed opportunity to be under-powered to detect any between-group differences, and c) readers would be interested in these study outcomes in comparison to other trials. Thus, while cessation was not the primary objective, it did serve as the outcome on which the study was powered (see below). Additional measures of cessation included: a) 7-day abstinence at each weekly visit, b) ‘floating abstinence;’ i.e., having ever achieved 7-days of non-smoking, throughout follow-up,¹⁸ and c) incidence of 24-hr quit attempts. Other related outcomes include changes in cigarette smoking and smoking reduction, changes in motivation and confidence to quit smoking (MTQ, CTQ, both assessed via 0–10 scales where 0 = none and 10 = maximum confidence/motivation), product dependence, and product co-use. Cigarette and e-cigarette dependence is based on the Penn State Nicotine Dependence Index,¹⁹ with separate but similar wording for each, allowing for comparison across products (range 0–20 for each). Consistent with national recommendations for trial outcomes¹⁸ and early trials of e-cigarettes,^6,7,20 smoking reduction is operationalized as achieving a ≥50% reduction in cigarettes per day (CPD) since baseline (% yes), thought to signify a meaningful reduction in toxicant exposure; i.e., harm.

Sample size determination

Per above, sample size estimations were based on cigarette cessation: 7-day point prevalence abstinence at 6-month follow-up. Base rates came from prior nationwide trials of similar design conducted by our group, in which 6-month abstinence rates within the control group ranged 4–13%.^21,22 Other estimates suggest ∼5% for population-based quitting.²³ Ultimately, a base rate of 8% was chosen. Anticipated quit rates in the e-cigarette group were derived from the most relevant study published at the time of study design: an uncontrolled observational cohort (i.e., naturalistic) study of smokers given e-cigarettes for up to 6 months,²⁴ in which 23% (intent-to-treat; ITT) and 33–39% (respondent only analyses) were quit at 6 months. Since the current trial provided e-cigarettes for 1 month only, a lower quit rate was anticipated. The trial was thus powered to detect a difference in rates of abstinence of 8% vs. 16% (RR = 2.0), requiring 600 participants (400 e-cigarette vs. 200 control). Inflating by 10% to account for potential attrition, the final anticipated sample size was estimated at 660 participants. The final actual sample size was 638 (97% of target); unavailability of product supply in the final months of recruitment precluded additional enrollment.

Statistical analysis

Measures of uptake are largely descriptive, focusing primarily on the e-cigarette group. Binary cessation outcomes were analyzed two ways: (1) coding missing data as returned to smoking (labeled as Missing = Smoking; M = S) as per suggested practice²⁵ and (2) utilizing generalized linear mixed models (GLMMs) with binary distribution, logit links, and random residual options to allow time points with missing data to be informed by prior collected data. No imputations were made for continuous outcomes. For M = S analyses, logistic regression models were used for each of the cessation outcomes noted above, with the main effect of intervention group included. Odds ratios (ORs) and 95% confidence intervals (95% CIs) were determined from the models for (1) point prevalence abstinence at weekly follow-up, (2) floating abstinence, (3) cumulative incidence of 24 hr QA's and (4) achieving 50% reduction in CPD. The GLMMs were created similarly with main effects for intervention group, time, and an interaction between time and group. Estimate statements from the models were used to determine group differences at each time point of interest and reported as ORs and 95% CIs. Note that both M = S and GLMM outcomes are shown in Table 2, but the text within Results reports M = S outcomes (outcome percents, ORs, 95% CIs) only. Per the original protocol (Supplemental Materials), covariates were not included in models unless baseline differences suggested need for otherwise.

General linear models (GLMs) were used to measure the relationships between group and time for continuous outcomes of interest such as CPD, MTQ, CTQ, and dependence. Interactions between time and group were initially included in models and were removed if not significant at an α = .1 level of significance, leaving the main effects of time and group.

Patterns of use and dual use over time are also largely descriptive, and in cases of missing data, the last observation was carried forward.

As noted above, the current trial stands in contrast with many existing RCTs, in that study recruitment was open to smokers across the motivational spectrum, without requirement to quit. Accordingly, many of the above outcomes were separately assessed among smokers high (7–10 on 10-pt scale) and low in motivation to quit (0–6), to determine if the effects were restricted to merely the former. Notably, the trial was not powered on these post-hoc sub-group differences; the intent herein is to compare effect sizes only.

Role of the funding source

The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. Authors MJC, AEW, and TTS had full access to the data throughout the study. All authors contributed to methods and manuscript editing. MJC had final responsibility for decisions on publication.

E-cigarette uptake

Fig. 2 shows the percentage of participants reporting use of an e-cigarette and frequency of use in each group. In the e-cigarette group, product use ranged from 60 to 70% during the 4-week sampling period and decreased over time (vs. 9–13% within no-product group during this same period). At 6-months, 44% of the e-cigarette sample group still reported use of an e-cigarette vs. 16% in the no-product control group. On average, frequency of use within the entire e-cigarette group (inclusive of non-users) ranged from 2.0 to 4.5 days per week during the initial four weeks, decreasing to two days per week at final follow-up. Among users only, frequency of use was more regular: 4–6 days per week throughout the entire course of the study. Frequency of e-cigarette use was much lower in the no-product control group. However, among users only, the daily usage levels were comparable to those in the e-cigarette group (Fig. 2). The same was true for amount of use within a day; i.e., puffing episodes (Supplemental Materials). Thus, participants in the e-cigarette group had a much higher uptake of the product overall (as expected), but among users, frequency and amount of use were similar, i.e., free product promoted increased access/use but did not increase frequency or amount of use. Another measure of uptake is product adoption; i.e., independent purchase of e-cigarettes at any time throughout the follow-up period. On this metric, there were no group differences on purchase of either another device or additional nicotine liquid (19% within e-cigarette group vs. 20% within no-product control group).

Between group comparisons on all measures of uptake were similar within each group of smokers who were highly motivated to quit vs. those who were not.

Smoking cessation, quit attempts, smoking reduction

As shown in Table 2, participants within the e-cigarette group were more likely to self-report cigarette abstinence after the 4-week trial period and at all follow-up assessments through six months. At final follow-up, 14% vs. 8% (M = S) of participants reported abstinence from cigarettes (OR = 1.79; 95% CI: 1.02–3.16). The cumulative incidence of quit attempts also favored the e-cigarette group throughout follow-up (26% vs. 19% at Week 24). Differences in floating abstinence, i.e., having ever achieved 7 days of non-smoking, were non-significant (e-cigarette: 17% vs. no-product control: 12%). GLM analyses revealed a significant group by time effect for changes in cigarette smoking (P < .0001), with a greater reduction in the e-cigarette group over time. A larger percentage of participants in the e-cigarette group achieved at least 50% reduction in CPD throughout follow-up compared to the no-product control group (Table 2; 28% vs. 18% at Week 24). Additional post-hoc analyses that further examined a) frequency of e-cigarette use and b) e-cigarette users vs. non-users, are presented in Supplemental Materials. At both Week 12 and Week 24, e-cigarette users in the e-cigarette group (at Week 4) had significantly higher rates of cessation-related behavior (quit attempts, point prevalence abstinenence, 50% reduction in CPD) as compared to Week 4 non-e-cigarette users in the no-product control group (Supplemental Table S5).

Post-hoc subgroup analyses (also in Table 2) demonstrated roughly comparable effects for each of the outcomes above among participants with low and high motivation to quit, as reported at baseline. While absolute rates of outcomes were lower among the low-motivated participants, the effect size (odds ratio) was comparable overall between low vs. high motivation. Thus, cessation effects were not restricted to smokers wanting to quit.

Motivation and confidence to quit smoking

GLM analyses revealed a significant group by time effect for both motivation (P < .0001) and confidence to quit smoking (P < .0001), with significant increases in each within the e-cigarette group, but not the no-product control group. The same was true among subgroups of low and high motivation to quit, except for CTQ among highly motivated smokers (time by group interaction not statistically significant).

Cigarette/E-cigarette dependence

Using the Penn State Nicotine Dependence Index (possible range 0–20), GLM analyses revealed a significant group by time effect for cigarette dependence (P < .0001), with significant decreases over time within e-cigarette sampling group but not no-product control group (Fig. 3). The same was true within sub-groups of smokers motivated to quit and not. Within the e-cigarette group alone, the same instrument allows for comparison of dependence across products (also Fig. 3). E-cigarette dependence increased slightly through Week 4 and then declined. At all timepoints, e-cigarette dependence was significantly lower than combustible cigarette dependence.

Concurrent/dual use

By Week 4, 68 (16%) of smokers within the e-cigarette group were exclusive smokers, 43 (10%) were exclusive e-cigarette users, 315 (74%) were co-using, and 1 (1%) was completely abstinent from both products. Among Week 4 mono-smokers (n = 68), 2 (2.9%) achieved complete abstinence (of both products) at Week 24, 5 (7.4%) had become dual users; 1 person (1.5%) became an exclusive e-cigarette user, but the majority (n = 60; 88.2%) maintained mono-smoking. Among exclusive e-cigarette users at Week 4 (n = 43), 13 (30.2%) achieved complete abstinence at Week 24, 2 (4.7%) returned to smoking and were dual using, 19 (44.2%) were still exclusive users of e-cigarettes, and 9 (20.9%) relapsed to exclusive smoking. Among dual users at Week 4 (n = 315), 18 (5.7%) achieved complete abstinence at Week 24, 187 (59.4%) were still dual using, 91 (28.9%) were relapsed to mono-smoking, and 19 (6.0%) became exclusive e-cigarette users. Full trajectories of use, across both groups, are provided within Supplemental Materials, as are rates and levels of smoking reduction among dual users.

Adverse events

Within the e-cigarette group, 180 people (42%) reported a total of 360 adverse events (AEs), of which 7 (2%) were severe, 113 (31%) were moderate, and 232 (64%) were mild (8 additional uncoded). The most common adverse events were cough (17%), headaches (12%), and increased phlegm (12%). There was one serious adverse event that was questionably associated with e-cigarette use: a participant-reported case of asthma-induced hospitalization, with “liquid in the lung”, possibly attributed to increased nebulizer use and/or e-cigarette use. Within the no-product control group, 86 people (41%) reported a total of 197 AEs, of which 7 (4%) were severe, 60 (30%) were moderate, and 124 (63%) were mild (6 additional uncoded). The most commonly reported AEs in the control group were cough (20%), increased phlegm (18%), and headaches (8.1%). The average (SD) number of AEs reported per person in the e-cigarette group was 0.9 (1.5), ranging from 0 to 7 AEs per person, as compared to the no-product control group with an average of 0.8 (1.3) AEs per person, ranging from 0 to 9.