Adult smokers’ Complete Switching Away from Cigarettes at 6, 9, and 12 Months after Initially Purchasing a JUUL e-Cigarette

Abstract Background: Electronic nicotine delivery systems (ENDS) have the potential to benefit public health if smokers completely switch from cigarettes to ENDS. Methods: A cohort of adult smokers (age ≥ 21) who purchased a JUUL Starter Kit was followed for 12 months after the initial purchase. We defined “switching” as past-30-day abstinence from smoking (even if JUUL use had stopped), and analyzed factors associated with achievement and maintenance of self-reported repeated point-prevalence switching (RPPS) at three follow-ups, i.e., no past-30-day smoking at months 6, 9, and 12. Results: RPPS was reported by 21.6% of the 12,537 evaluable smokers. Smokers with lighter smoking history and lower cigarette dependence at baseline were more likely to report RPPS. RPPS was also associated with daily use of JUUL (Month-3: OR = 2.32, 95% CI = 2.02–2.68; Month-6: OR = 1.73, 95% CI = 1.42–2.10), and with greater subjective reinforcing effects from JUUL use (assessed by the mCEQ, Month-3: OR = 1.46, 95% CI = 1.38–1.56; Month-6: OR = 1.11; 95% CI = 1.02–1.20). Even among smokers who did not meet the criteria of RPPS (i.e., who smoked at least once), 35.5% reported past 30-day point-prevalence switching on at least one follow-up, and their cigarette consumption was substantially reduced compared to baseline. Conclusions: Approximately one-fifth of adult smokers reported 30-day point-prevalence abstinence at 6, 9, and 12 months after purchasing JUUL. Greater use of JUUL and stronger subjective reinforcing effects were associated with nonsmoking, validating the potential for ENDS to substitute for smoking, with potential for positive impacts on individual and population health.


Introduction
Although smoking rates have declined in the US and internationally, cigarette smoking continues to be the leading cause of preventable morbidity and mortality worldwide (World Health Organization, 2017).The health benefits of smoking cessation are time-dependent: some benefits are seen immediately within hours of cessation, but most benefits accrue over years (U.S.Department of Health & Human Services, 2010).Not surprisingly, smoking cessation remains highly challenging for many smokers, even with pharmacological and behavioral support and strong motivation to stop smoking (Dobbie et al., 2015;Jackson et al., 2019).
A substantial proportion of adult smokers are not ready to quit at any given time (Babb et al., 2017) and the majority of individual quit attempts are unsuccessful (Babb et al., 2017;Creamer et al., 2019).Compared to smoking combustible cigarettes, switching completely to electronic nicotine delivery systems (ENDS) significantly reduces exposure to smoking-related carcinogens and toxicants that are primarily produced by the combustion of tobacco (Akiyama & Sherwood, 2021;Cohen et al., 2021;Goniewicz et al., 2017;Holt et al., 2023;Pulvers et al., 2020).Accordingly, switching to ENDS has emerged as a potential harm reduction strategy for smokers who would not otherwise quit smoking in the near term (Balfour et al., 2021;Warner, 2019).
A growing body of evidence demonstrates that adult smokers can switch to ENDS (Abi Nehme et al., 2022;Baig & Giovenco, 2020;Brouwer et al., 2022;Farsalinos & Niaura, 2020;Glasser et al., 2021;Harlow et al., 2021;2023;Kalkhoran et al., 2020;Shafie-Khorassani et al., 2023), with a 30-day point-prevalence (i.e., no smoking in the past 30 days) at one year as a typical end-point (Brouwer et al., 2022;Kalkhoran et al., 2020).However, to fully take advantage of the potential benefits of ENDS, complete switching would need to be sustained for an extended period of time.Additionally, most of these naturalistic longitudinal studies examined switching in samples using a heterogeneous variety of ENDS, which can vary in nicotine delivery as well as other product features (Hajek et al., 2020;Voos et al., 2019), for varying periods of time.
In a recent study, Goldenson et al. (Goldenson et al., 2021) followed a cohort of smokers using the same ENDS product-JUUL, which is a closed-system pod-based ENDS brand widely used in the US (King et al., 2018).Participants in the Adult JUUL User Switching and Smoking Trajectories (ADJUSST) study were enrolled when they had just purchased a JUUL Starter Kit (JSK) for the first time, and were followed for 12 months without intervention and without being provided with any products.About half of the baseline adult established smokers reported no past-30-day smoking 12 months after their JSK purchase.This was defined as "switching" even if JUUL use had been stopped (by analogy to attributing quitting to NRT even after NRT has been stopped, see Measures).About one third of the sample reported switching at both the 9 and 12 month follow-ups (Goldenson et al., 2021).However, the predictors and dynamic processes associated with achieving and maintaining the status of complete switching over a longer period of time remain unknown.Using the ADJUSST study data, the present secondary analysis examines factors associated with self-reported repeated point-prevalence of switching (RPPS) at three time-points over a period of 6 months at the end of a year, at months 6, 9, and 12 after the initial JSK purchase.Analyses consider predictors from baseline characteristics, from smoking and JUUL use patterns at a prior assessment at 3 months, and predictors of continued abstinence among those who were abstinent at 6 months.

Materials and methods
This is a secondary analysis of the ADJUSST study (Shiffman et al., 2021), a naturalistic longitudinal follow-up of US adults 21+ who purchased a JSK (a JUUL device, a USB charging dock, and 4 JUUL pods), either online or in retail outlets between June and October 2018 and were invited to participate in an online longitudinal survey study.After a baseline assessment, participants were invited to complete follow-ups at 1, 2, 3, 6, 9, and 12 months.Participants were eligible for all follow-ups, regardless of smoking, JUUL use, or completion of prior surveys.The study was a noninterventional naturalistic longitudinal follow-up: participants were not recruited or selected based on having any behavior-change goal (such as reducing or stopping smoking), no goals were communicated to them, no intervention was offered, and participants were not provided with JUUL products, but could purchase them in the marketplace and use them (or not) as they saw fit.The study was approved by the Advarra IRB (https://www.advarra.com),and participants provided informed consent and were compensated $30 for each completed survey.Study methods have been detailed in a previous publication (Shiffman et al., 2021), as have a number of substantive analyses (Goldenson et al., 2021;Kim et al., 2021;Selya et al., 2021).
A detailed analysis of patterns of missed follow-ups in ADJUSST concluded that the potential for outcome bias due to missing data was limited (Shiffman et al., 2021).Specifically, there was very little difference in demographics or smoking history between participants who completed all follow-ups, no follow-up, or some follow-ups.Further, differences that were observed did not consistently suggest that participants with missing data were less likely to switch.Also, in a re-contact study with month-12 non-responders, almost half reported they had switched away from smoking, and very few reported that their non-response was due to having "started smoking" or "stopped using JUUL" (each less than 5%); rather, 81% reported exclusively survey-related reasons (i.e., missing the survey widow, compensation too small, etc.) for non-response (Shiffman et al., 2021).

Sample
Study eligibility criteria were: (1) age 21 or older, (2) a permanent resident of the US, (3) a first-time purchaser of JSK within the past 7 days, (4) neither employee nor family member of employees of Juul Labs, Inc. or PAX Labs, Inc.The present analyses focus on 12,537 ADJUSST enrollees who were established smokers at baseline (smoked in the past 30 days, smoked every day or some days, and had smoked 100+ cigarettes) whose RPPS status could be ascertained (as described below).

Self-reported repeated point-prevalence of switching (RPPS) at months 6, 9, and 12
Switching at each follow-up was defined as a "No" response to the question "In the past 30 days, have you smoked a cigarette, even one or 2 puffs?"RPPS was defined as reporting no smoking at all three consecutive follow-up assessments, at 6, 9, and 12 months.As more than 98% of the sample reported that they have already used JUUL prior to the determination of RPPS, the definition did not take JUUL use at those months into account (We use the term "switching" because participants switched from smoking cigarettes to using JUUL (for at least some of the time) and because terms such as "quitting" or "cessation" are commonly associated with FDA-approved smoking cessation aids, such as NRT.As JUUL is a consumer nicotine product and not a smoking cessation aid, we believe it is important to use different terminology so JUUL is not misperceived as a smoking cessation aid).Although RPPS may include some participants who smoked in non-survey months, repeated reports of past-month cigarette abstinence have been used as a proxy of sustained cigarette abstinence (Kalkhoran et al., 2020) and mutually exclusive multi-year behavioral trajectories in the PATH study (Stanton et al., 2020;Taylor et al., 2020).The 6-month point was selected as the start of the RPPS interval because previous data showed that rates of past-30-day switching were rising steeply up to that point (Goldenson et al., 2021).Therefore, starting at an earlier time-point would exclude many participants who took longer to switch (which was particularly likely among those with heavier smoking history), but might nevertheless achieve RPPS.
Those reporting no past-30-day smoking at all three time-points were deemed to have achieved RPPS; those who reported any smoking at any of the three time-points were deemed to not have achieved RPPS, regardless of their smoking status or missingness at other time-points.Participants who reported abstinence at one or two time-points, but had a missing status for the others, or were missing all three time-points, were deemed unevaluable.See Table S1.
The surveys captured cigarette smoking and JUUL use behaviors, but not the use of noncigarette combusted tobacco products (i.e., cigar, cigarillo, hookah, etc.).Although multiple tobacco product use without cigarettes is uncommon, (Kasza et al., 2017) the possibility of other combustible product use and its impact on harm reduction potential of ENDS cannot be assessed (see Limitations).

JUUL use behaviors
Participants' JUUL use and related responses were measured at each follow-up.Analyses examined the 3-and 6-month data in relation to later RPPS.Daily JUUL use (all 30 days vs. less than 30 days) was assessed, as was the number of JUUL sessions (defined as the use of around 15 puffs, or 10 min (Foulds et al., 2015)) on days used JUUL.Dependence on JUUL was evaluated with the 16-item Adult Tobacco Dependence Index (TDI), which is validated for assessing dependence on ENDS (Strong et al., 2017;2020).Participants also completed the modified Cigarette Evaluation Questionnaire (mCEQ, adapted for JUUL), which has previously been used to assess subjective responses to ENDS (DiPiazza et al., 2020;Steinberg et al., 2014).Preliminary analyses showed that the subscales of mCEQ (satisfaction, psychological rewards, enjoyment of respiratory tract sensations, reduction in cigarette cravings) were highly correlated (averaging r = 0.54) and constituted a unidimensional scale (first principal component eigenvalue was 2.64, accounting for 66% of the variance; all others were <0.60).This presented a collinearity problem for their simultaneous inclusion in a multivariable model.Therefore, subscales were averaged to represent the overall positive reinforcing effects of JUUL use in the multivariate models.

Covariates
Sociodemographic factors analyzed included race/ethnicity, gender, age, household income, educational attainment, and marital status (see Table 1).Baseline smoking history assessed smoking frequency (number of days smoked in the past 30 days), cigarettes per day (averaged over 30 days), cigarette dependence (using the TDI), baseline readiness to quit smoking (planning to quit in the next 30 days [yes/no] (Prochaska & Velicer, 1997)), and years of regular smoking.

Analyses
Associations of sociodemographic factors, smoking history, JUUL use behaviors, and RPPS were initially assessed using univariate logistic regression, and then in three multivariable models predicting RPPS from different sets of regressors: (1) "Baseline model" adjusted for baseline sociodemographic and smoking characteristics; (2) "Month-3 model" added month-3 JUUL use behaviors to the Baseline model, assessing what month-3 behaviors predicted subsequent achievement of RPPS; and (3) "Month-6 model" added month-6 JUUL use behaviors to the Baseline model, assessing what month-6 behaviors predicted subsequent maintenance of switching among those who were already not smoking at 6 months.Models assessing JUUL use behaviors were based on past-months users of JUUL at the respective month (95.7% at Month-3 and 93.4% at Month-6).Multivariable models adjusted for factors demonstrating a significant association with RPPS at the univariate level.In this cohort of JUUL purchasers, those not using JUUL, especially while achieving RPPS, are of unique interest.Characteristics of this subgroup are described in the Supplementary Material.
A secondary analysis of 6,521 participants who were still smoking at month 3 assessed associations of characteristics and subsequent RPPS.Another important subset is those who were smoking daily at baseline, especially those who are in their 30s and older, as ENDS use is more common among young adult smokers (Kalkhoran et al., 2020).Factors associated with RPPS in this subgroup are assessed in the supplementary analyses as a robustness-check.Furthermore, among 9,829 participants who did not achieve RPPS, average cigarette-per-day (CPD) were evaluated in those who reported switching at one or two of the relevant follow-ups (but not all three, N = 3,488), compared to those who did not (N = 6,341).All statistical analyses were conducted in R version 4.0.2(R Core Team, 2020).

Baseline characteristics and JUUL use behaviors of the sample
RPPS across months 6, 9, and 12 was evaluable for 12,537 participants.As in the overall ADJUSST sample (Shiffman et al., 2021), evaluable participants were majority non-Hispanic White, male, and daily smokers (Table 1).On average, they were 32.9 years old, smoked for 12.8 years, and smoked 11.4 cigarettes per day at baseline.Prior to Month 6 (cut-off for the definition of RPPS), 98.5% of the sample reported that they had used JUUL.
Overall, 21.6% met the criteria for 6-month RPPS, i.e., reported past-month complete switching at months 6, 9, and 12; the remainder reported smoking (at least a puff) on at least one of those follow-ups (see Table S1).
Table 1 compares the characteristics of participants by their RPPS status.On a univariate basis, compared to those who smoked at ≥1 follow-up at 6, 9, or 12 months, those reporting RPPS were younger, had higher household income, and were more likely to be married.At baseline, those reporting RPPS had a lighter smoking history and profile: they had a shorter history of regular smoking, smoked less frequently and fewer cigarettes-per-day, had a lower level of cigarette dependence, and were more likely to be ready to quit.Those with RPPS were significantly more likely to report JUUL use prior to Month 6, although JUUL use was nearly universal across both RPPS statuses (99.7% and 98.1%).Focusing on Month 3, several parameters of JUUL use were prospectively associated with subsequent RPPS across months 6 to 12.Those who at 3 months were daily (vs.non-daily) JUUL users, used JUUL more times per day, were more dependent on JUUL, and rated Preliminary analyses examined the use of other brand ENDS and its effects on RPPS.Using other ENDS exclusively (without JUUL) was uncommon in this sample (between 0.1% at Month 1 and 3.4% at Month 12), and their odds of later RPPS were not different from those not using ENDS at all across Months 1 to 3 (all ps > 0.05).Therefore, subsequent analyses did not consider other ENDS use.

Factors associated with later RPPS: particicpant characteristics and JUUL use behaviors
The first multivariable model with baseline sociodemographic and smoking characteristics ("Baseline model, " see Table 2) largely confirmed the associations between smoking history and RPPS seen in univariate analyses.Those with lighter smoking history and patterns were more likely to achieve RPPS.With the multivariable adjustment, the associations of age and cigarette dependence with RPPS were no longer statistically significant.
In the Month-3 multivariable model, some JUUL use characteristics were no longer significant when mutually adjusted, but those who used JUUL daily still had over twice the odds of reporting RPPS (OR = 2.32), and those reporting greater reinforcing effects with JUUL use were also more likely to achieve RPPS.Daily JUUL use, higher JUUL dependence, and greater reinforcing effects predicted later RPPS among baseline daily smokers and those older than 30 (Table S2).Being switched at month 3 was strongly correlated with later RPPS, as 53.5% of those already switched at month 3 continued to report RPPS (Table 1), while only 8.1% of month-3 smokers achieved subsequent RPPS.Factors associated with RPPS among month-3 past-30-day smokers were examined separately (see below).
The "Month-6 model" tested how 6-month JUUL use characteristics among those switched at month 6 related to reported abstinence at both 9 and 12 months (i.e., RPPS).For these smokers, who already achieved past-30-day nonsmoking at month 6, most baseline sociodemographic and smoking characteristics were no longer associated with the maintenance of switching.However, daily JUUL use and stronger reinforcing effects of JUUL at month 6 continued to predict the maintenance of switched status, reaching RPPS through month 12.

Subgroup of those reporting RPPS and not using ENDS at month 6
Among those with RPPS, 93 (3.4%) participants reported not using ENDS (neither JUUL nor other brand) at Month 6.Compared to those using ENDS, they were more likely to be younger and have a shorter smoking history.Even within this subgroup, 94% reported using ENDS at Month 1 and 31% at Month 12. Thus, these participants were using ENDS intermittently, rather than completely discontinuing them (see Supplementary Material on JUUL and ENDS use).

Analyses among those smoking at month 3
As noted above, only a small minority (8.1%) of those still smoking at month 3 subsequently reported RPPS (Table 1).To understand factors associated with such conversion, we examined predictors of RPPS among month-3 smokers (i.e., those not switched at 3 months).As seen in Table 3, on a univariate basis, month-3 smokers who later showed RPPS had a lighter smoking profile (less smoking and lower dependence) at baseline and were also smoking less at month 3.As in the overall sample, those who used JUUL daily had more JUUL-use sessions per day, reported higher dependence, and perceive the reinforcing effects of JUUL more strongly were more likely to subsequently convert to RPPS.When mutually adjusted in multivariable analyses, baseline smoking characteristics were rendered non-significant, but participants who smoked more days per month at month 3 continued to be less likely to subsequently achieve RPPS.Also, participants who at month 3 were using JUUL daily, and those who reported greater dependence on JUUL, were more likely to achieve RPPS.

Cigarette consumption among smokers who did not achieve RPPS
As seen in Table S1, the cohort that did not achieve RPPS was heterogeneous, with some having reported switching at one or two time-points during months 6, 9, and 12, while others had not reported any switching in this time frame.To better understand the smoking behavior of these two cohorts, nH: non-Hispanic, HS: High school, tDi: adult tobacco Dependence index.Factors denoted "not adjusted" were not included in the multivariable models as no association with RPPS was shown at the univariate level. 1 a higher score indicates stronger agreement.
2 assessed for the past-month JuuL users at month 3 (N = 9719, 95.7% of 10153 participants with evaluable RPPS status and valid JuuL use data. 3 One session was defined as "the use of around 15 puffs, or 10 min". 4 average score of satisfaction, psychological reward, enjoyment of sensation, and craving reduction domains of the mceQ. 5assessed for the those who reported past-month JuuL use and nonsmoking at month 6 (N = 3774).univariate analyses within this sample are not shown.note: Odds ratio (95% confidence interval), boldface indicates statistical significance (p < 0.05).
we compared the average CPD for those who were smoking at months 6, 9, and 12, respectively.Table S3 shows that those who were not smoking at one or two follow-ups were lighter smokers at baseline.Both of the cohorts reported reducing their cigarette consumption (compared to baseline) considerably, but the reduction was greater among those who had switched away from smoking on at least one of the follow-ups.Among those who did not switch, the median cigarette reduction was 60%, and 60% of participants had reduced their cigarette consumption at least by half.In the group that switched once or twice, the median reduction was 90%, and >80% had reduced their consumption by half or more.When they were smoking, this group's cigarette consumption averaged <3 cigarettes per day across all three follow-ups.

Discussion
The analyses presented here assessed the predictors of self-reported past-30-day abstinence from smoking in 3 consecutive surveys at months 6, 9, and 12 in a sample of adult smokers who purchased a JSK.In the evaluable sample, 21.6% reported such repeated point prevalence switching away from smoking.While considerably lower than the single point-prevalence switching rates of 47 ~ 51% at those months (Goldenson et al., 2021), this is a substantial proportion of these smokers, particularly considering that half of them were not ready to quit smoking when they first purchased a JSK (Table 1).Also, even among those not ready to quit at baseline, 18% reported RPPS over the last 6 months of the following year.The data also indicate that over three-quarters of those who achieved RPPS during months 6 to 12 had already switched away from smoking at month 3, suggesting that switching was often more prolonged than was captured by the RPPS end-point used here.
As expected, individuals with a lighter smoking history (lower baseline smoking frequency, quantity, duration, and dependence) were more likely to achieve RPPS, consistent with predictors of clinical smoking cessation (Borland et al., 2010;Hyland et al., 2004;2006;Kasza et al., 2020;Vangeli et al., 2011).Studies on switching with ENDS also showed similar results: lighter and less dependent smokers at baseline were more likely to report complete cigarette abstinence (Abi Nehme et al., 2022;Kalkhoran et al., 2020;Shafie-Khorassani et al., 2023) and reduced their cigarette consumption more greatly (Rubenstein et al., 2021).As heavier smokers may take longer to achieve complete switching (Goldenson et al., 2021), further follow-up beyond the initial first year would be informative.Two demographic factors also emerged as predictors in adjusted models as married smokers and those with higher income were more likely to achieve RPPS.This association has also been observed in literature on smoking cessation (Kasza et al., 2020;Vangeli et al., 2011) and switching (Harlow et al., 2023;Kalkhoran et al., 2020), and likely reflects the availability of social support and other psychosocial resources.Collectively, the profile of those switching to JUUL is in line the previous literature and furthers the current understanding on the specific group of smokers who may benefit from adopting ENDS.
Interestingly, the strongest predictors of subsequent RPPS reflected how participants were using and reacting to JUUL.Those who at 3 months were using JUUL daily and using it more times per day, who scored higher on JUUL dependence, and who reported greater reinforcing effects from using JUUL were more likely to subsequently achieve RPPS, and their effects remained consistent across various subgroups such as baseline daily smokers.Similarly, even among those who were already switched at month 6, daily use of  JUUL and greater perceived reinforcing effects increased the odds of maintaining the switched status at months 9 and 12.These findings are consistent with prior studies that have similarly reported that daily ENDS use is associated with switching (Chang et al., 2023;Farsalinos & Niaura, 2020;Glasser et al., 2021;Harlow et al., 2021;Kalkhoran et al., 2020).Previous research has shown that switchers (i.e., former smokers who are currently using ENDS) report that the subjective sensations from ENDS were important in their transition from cigarette smoking to exclusive ENDS use (DiPiazza et al., 2020;Evans et al., 2020).Indeed, users' subjective responses to ENDS, such as reduction in cigarette craving and ENDS liking, were associated with higher odds of switching and greater smoking reduction (Gades et al., 2020;Goldenson et al., 2021).The present finding extends that association to RPPS, consistent with the hypothesized role of ENDS in displacing cigarette smoking (Abrams et al., 2018;Balfour et al., 2021;Gray et al., 2005;Warner, 2019).
Previous studies have suggested the potential role of vape shop employees in informing consumers of factors associated with switching (Burbank et al., 2016;Galimov et al., 2021).Communicating these findings formally, e.g., in product labeling, remains a challenge under the regulatory framework governing ENDS, but avenues for doing so would be a useful topic for future research and regulatory discussion.
Collectively, the present findings suggest the role of ENDS/JUUL use, and its potential to substitute for cigarettes, as an important factor in switching away from smoking.This is consistent with the conceptualization of ENDS as substitutes for cigarettes, and with Abrams et al. 's conceptual model (Abrams et al., 2018), which emphasizes that ENDS need to provide satisfaction sufficient to promote and reinforce use in order to compete with cigarettes and facilitate switching (see also Gray et al. (2005)).The importance of 'abuse liability' of ENDS in facilitating smokers' switching has been similarly recognized by the Food and Drug Administration's Center for Tobacco Products (U.S. Food & Drug Administration, 2019).
It was notable that among the participants who did not achieve RPPS, but who had switched at one or more follow-ups at 6, 9, and 12 months, reported much-reduced cigarette consumption, with over 80% reporting reductions of at least 50%.Some of these participants may have been getting close to achieving abstinence at all three time points.The substantial reductions in cigarette consumption observed in those who did not achieve RPPS are consistent with Selya et al.'s report (2021) that most ADJUSST participants who were engaged in dual-use had reduced their cigarette consumption by at least 50%, a degree of reduction that has been associated with very substantial reductions in exposures to smoking-related toxicants (Akiyama & Sherwood, 2021;Cohen et al., 2021;Morris et al., 2022), and has been associated with reductions in disease risk, particularly for cancer (Akiyama & Sherwood, 2021;Goniewicz et al., 2017;Pulvers et al., 2020).
This study has several limitations that should be taken into account.The ADJUSST study had no control group or randomization, precluding strong causal inferences.Nevertheless, the consistent associations of greater JUUL use and JUUL reinforcement with RPPS suggest that JUUL use may play a role in RPPS.ADJUSST study did not collect whether participants used other tobacco products.Cigarettes account for more than 80% of all combustible product use (Cornelius et al., 2022), and as such, are of critical importance for population health.While it is possible that participants stopped smoking cigarettes but substituted them with non-cigarette combustible products (i.e., cigar, cigarillo, hookah, etc.), using multiple tobacco products without cigarettes is highly uncommon (Kasza et al., 2017), and combinations of ENDS and non-cigarette product(s) accounted for only a small proportion of multiple tobacco product use.Nevertheless, participants' other tobacco use status remains unknown, and is acknowledged as a limitation of this study.RPPS was defined based on self-report, without biochemical verification, consistent with other observational studies on switching in population-based data (Abi Nehme et al., 2022;Baig & Giovenco, 2020;Brouwer et al., 2022;Farsalinos & Niaura, 2020;Glasser et al., 2021;Harlow et al., 2021;2023;Kalkhoran et al., 2020;Shafie-Khorassani et al., 2023) and a naturalistic randomized controlled trial (Carpenter et al., 2023).
As in many longitudinal surveys, some participants did not complete all follow-up surveys, resulting in missing data.In some cases, missing data made it impossible to establish participants' RPPS status over all three follow-ups.While there is a possibility of selective drop-out (i.e., those who switched continued responding while those who returned to smoking stopped responding to the survey), analyses of survey response in ADJUSST suggests that the potential for bias from differential response rates was limited (Shiffman et al., 2021).Nevertheless, it remains a limitation.In ADJUSST, the vast majority of non-response was due to survey-logistics reasons such as survey window being closed or survey taking too long, rather than avoidance of reporting smoking (Shiffman et al., 2021).Previous literature describes various strategies which may improve the retention rates of longitudinal studies (Teague et al., 2018).In this case, survey logistics-related improvements, such as sending reminders more frequently, making them more attention-grabbing, arranging questions in the order of importance and developing more efficient skip patterns to prevent fatigue, may be able to promote participation and provide more complete understanding of the behavioral trajectory.While a similar definition has been used to assess multi-year sustained switching with PATH data by Kalkhoran and colleagues (Kalkhoran et al., 2020).RPPS does not necessarily imply continuous abstinence over the 6-month period.Conversely, even the reports of smoking at a follow-up do not necessarily imply a full return to smoking, as even reporting one puff of smoking in the month preceding any of the three assessments assigned the participants in the 'non-RPPS' group.A dynamic analysis of more detailed trajectories might provide useful additional insight.
Participants were recruited based on their purchase of JSK, so may not necessarily represent the general population of cigarette smokers or even all ENDS users, which are often defined on the basis of any use in the past 30 days and thus including experimental triers (Brouwer et al., 2022;Kalkhoran et al., 2020).Purchase of a JSK likely indicates a degree of commitment to JUUL use, and this may help explain higher rates of complete switching at 12 months, compared to other studies (Abi Nehme et al., 2022;Farsalinos & Niaura, 2020;Harlow et al., 2023;Johnson et al., 2019).The study also followed smokers from their first JSK purchase, so may better capture trajectories over time.Many longitudinal studies start with many ENDS users who are dual users at baseline and therefore may omit smokers who already have successfully switched and overrepresent those unable to switch.
This study focused on users of JUUL, a particular pod-based nicotine-salt ENDS product, and it is not clear how results might apply to other ENDS, which are highly heterogeneous in nicotine delivery and other factors (Hajek et al., 2020;Voos et al., 2019).Although some participants reported using non-JUUL brands of ENDS, using other ENDS without JUUL was extremely rare, considering the ADJUSST cohort was recruited based on the purchase of JUUL.Furthermore, devices used were heterogeneous in brand, device specification and offerings.Some studies have reported that JUUL can deliver nicotine more effectively than other ENDS (Hajek et al., 2020;Phillips-Waller et al., 2021), which has been suggested as an important feature of ENDS that facilitates smokers to switch (Abrams et al., 2018;Hajek et al., 2020;Phillips-Waller et al., 2021).Thus, inferences on other brand ENDS cannot be drawn, and the generalizability of the findings to the range of other ENDS products is unknown.
The study also had multiple strengths.This study extended follow-up analyses from a single time-point to RPPS over a six-month period.We analyzed a large sample of baseline smokers who had just purchased JUUL, addressing the limitations of some previous studies, including small sample sizes.The participants represent real-world adopters of ENDS, and were followed forward from their initial purchase, making the results of this study more readily interpretable in terms of smokers' trajectories, and perhaps more applicable to real-world use.The analyses included details of smoking history and JUUL use patterns that enabled informative predictive models.

Conclusions
ENDS have emerged as potential substitutes for cigarette smokers who are otherwise not likely to quit in the near term.Complete switching away from cigarettes would reduce their exposure to combustion-related toxicants (Akiyama & Sherwood, 2021) and thus reduce smoking-related harm (McNeill et al., 2018;National Academies of Sciences, 2018).Approximately one-fifth of adult smokers who purchased JUUL at baseline reported complete switching at months 6, 9, and 12. Lighter baseline smoking and greater use of JUUL (especially daily use) and stronger perceived reinforcing effects of JUUL predicted RPPS.These results suggest that, consistent with models of ENDS' role in harm reduction, products such as JUUL may help smokers switch away from smoking, ultimately improving the population health.
defined as "the use of around 15 puffs, or 10 min." 3 assessed for the past-month JuuL users and smokers at month 3 (N = 6228).4 average score of satisfaction, psychological reward, enjoyment of sensation, and craving reduction domains of the mceQ.note: Odds ratio (95% confidence interval), boldface indicates statistical significance (p < 0.05).

Table 1 .
Sociodemographic factors, baseline smoking history, and JuuL use characteristics of baseline smokers, stratified by repeated point-prevalence of switching (N = 12537).
JUUL as more reinforcing (i.e., perceiving JUUL use satisfying and psychologically rewarding, enjoying the sensation in the throat, and having their cravings for cigarettes effectively alleviated) were more likely to achieve RPPS.Similarly, smokers who had already switched at month 6 were more likely to achieve RPPS over the following 6 months if, at 6 months, they used JUUL daily and found JUUL more reinforcing.Associations with a daily quantity of JUUL use and JUUL dependence were not statistically significant among those already switched at month 6.JUUL use behaviors at Months 9 and 12 remained consistent with earlier months in which those with RPPS demonstrated a more engaged pattern of JUUL use (see Supplementary Material on JUUL and ENDS use).

Table 1 .
JuuL use data) reported that they did not use JuuL in the past month.RPPS of Month-6 non-JuuL user group was 19.7%, which was higher than that of the nondaily users (16.0%, chi-squared test: p = 0.0202) but lower than that of daily users (35.1%, p < 0.001).Descriptive statistics on Month-6 JuuL use behaviors and missingness are calculated from 9564 past-month JuuL users at Month 6. count (valid percentage) for categorical variables; mean (standard deviation) for continuous variables; asterisks (*) represent significant univariate differences between RPPS group and smoker group.univariate associations of month-6 JuuL use behaviors and sustained switching were examined within the month-6 complete switcher group.See table 2 for odds ratios.continued.

Table 2 .
associations between participant characteristics and repeated point-prevalence of switching across months 6, 9, and 12: Results from univariate and multivariable modeling (N = 12537).