Association of Device Type, Flavors, and Vaping Behavior with Tobacco Product Transitions among Adult Electronic Cigarette Users in the United States

Abstract

Background:

Few studies assess whether electronic cigarette (e-cigarette) device characteristics or flavors impact longitudinal patterns of cigarette and e-cigarette use.

Design:

We examined data from waves 2–4 of the Population Assessment of Tobacco and Health study (2014–2018). Among adult (≥18y) current e-cigarette users at wave 2 who were current smokers (dual users; n=1,759) and former smokers (exclusive e-cigarette users; n=470), we classified participants into four use patterns at wave 3 (~12-months later) and wave 4 (~24 months later): (1) dual use of e-cigarettes and cigarettes; (2) exclusive cigarette smoking; (3) exclusive e-cigarette use; (4) non-use of both products. We used multinomial logistic regression to assess correlates of changing use patterns at 24 months, relative to no change, adjusting for socio-demographic factors.

Results:

At 24 months, 26.5% of baseline exclusive e-cigarette users, and 9% of baseline dual users, abstained from both vaping and smoking. Participants who vaped non-tobacco flavors (vs. tobacco flavors), and used refillable tank or modifiable devices (vs. disposable, cartridges, and other devices) were less likely to transition to non-use of both products, and to exclusive cigarette smoking. Baseline daily vaping (vs. non-daily) was positively associated with exclusive e-cigarette use at 24 months for baseline daily cigarette smokers, but negatively associated with exclusive e-cigarette use and non-use of both products at 24 months for baseline non-daily smokers.

Conclusions:

Non-tobacco flavors, daily vaping, and modifiable e-cigarette devices may help some smokers abstain from cigarette smoking via transitioning to exclusive e-cigarette use, but are also associated with ongoing exclusive e-cigarette use.

INTRODUCTION

In the United States (US), >90% of adult electronic cigarette (e-cigarette) users are current or former combustible tobacco smokers.¹ Most adults vape e-cigarettes flavored to taste like mint, menthol, or sweet flavors, though cigarette smokers are more likely than non-smokers to vape tobacco flavors.² Substantial heterogeneity exists in e-cigarette devices, which include but are not limited to disposables, cartridges, refillable tank systems, and newer “pod-mod” devices.^3,4

Understanding how e-cigarette use changes over time, and how patterns of use relate to combustible cigarette smoking is critical to informing tobacco product regulations. Most studies on adult e-cigarette use evaluate e-cigarettes as a smoking cessation aid.^5–15 Fewer studies examine how e-cigarette use evolves over time concurrently with combustible cigarette smoking, and whether there are product characteristics or behaviors associated with more desired trajectories (e.g., abstinence from both products). Recent analyses of the Population Assessment of Tobacco and Health (PATH) study indicate >60% of US e-cigarette users discontinue use after two years.¹⁶ Studies from other global populations similarly demonstrate that vaping is a dynamic behavior.^17,18 However, cigarette smokers are more likely to persistently use e-cigarettes than non-smokers.¹⁶ Though some studies show e-cigarette use is associated with cigarette smoking relapse among former cigarette smokers,^19,20 few studies evaluate longer-term vaping patterns among former smokers. In addition, the impact of modifiable e-cigarette characteristics, including flavors or device type, on e-cigarette and cigarette use patterns is understudied.

Observational studies report that smokers who vape characterizing flavors (non-tobacco flavors) are more likely to reduce cigarette consumption or abstain from smoking,^21–25 though some find no association. ^26,27 Refillable devices or tank systems might be more effective than less customizable devices in helping smokers quit or reduce their cigarette exposure, due to increased nicotine delivery and a personalized vaping experience.^28,29 Generating evidence on e-cigarette product characteristics associated with tobacco use patterns is critical as the FDA evaluates the public health impact of e-cigarette products through premarket authorization applications.³⁰ Further, bans on flavored tobacco products (including e-cigarettes) are being considered and implemented across the US.³¹ In January 2020 the FDA banned all cartridge-based e-cigarettes with flavors other than tobacco or menthol. Understanding longitudinal patterns of smoking and vaping over time, and whether modifiable e-cigarette characteristics help or hinder smokers to abstain from both cigarettes and e-cigarettes is critical to fully evaluate the public health impact of e-cigarettes and inform tobacco product regulations.

We aimed to quantify patterns of concurrent e-cigarette and combustible cigarette use over time among current and former combustible cigarette smokers in the US and to assess e-cigarette correlates of different use patterns. Using prospective cohort data from the PATH study, we assessed patterns of e-cigarette use and cigarette smoking over two years. We then tested whether e-cigarette product characteristics and behaviors are associated with longitudinal patterns of use.

METHODS

We analyzed public use data from the PATH study, an ongoing US nationally representative prospective cohort study on tobacco product usage and health.³² At wave 1 (2013–2014), PATH sampled 32,320 adults using a four-stage stratified area probability sample design. Interviews were conducted approximately 12 months apart using computer assisted telephone interviewing (wave 1: 2013–2014, n=32,320; wave 2: 2014–2015, n=28,362; wave 3: 2015–2016, n=28,148; wave 4: 2016–2018, n=27,757). Details on study design and response rates are published elsewhere.³² We selected wave 2 as study baseline because of greater available detail on device type and flavors.

Sample

Our sample included current and former cigarette smokers who were current e-cigarette users at baseline. Current smokers had smoked >100 cigarettes in their lifetime, and currently smoked every-day or some days. Former smokers had smoked >100 cigarettes in their lifetime, and currently smoked not at all at baseline. All eligible participants used any e-cigarette product (e-cigarette, vape pen, e-hookahs, e-cigars, etc.) every day or some days. We conducted preliminary analyses to examine timing of e-cigarette initiation relative to cigarette smoking status at baseline (Supplementary Tables 1-3). The majority of current smokers initiated e-cigarettes after combustible cigarettes (90%), and few former smokers initiated vaping after quitting combustible cigarette smoking (1.8%). Due to small numbers of never smokers who were current e-cigarette users (n=193), we did not include this group in analyses. We excluded participants missing wave 3 or 4 interview data in order to evaluate tobacco use transitions at 12 and 24 months (n=600). The final analytic sample included 2,229 e-cigarette users, including 470 exclusive e-cigarette users who were former cigarette smokers, and 1,759 dual users of e-cigarettes and cigarettes (Supplementary Figure 1). Hereafter, we refer to our two samples as “exclusive e-cigarette users”, and “dual users”.

Measures

E-cigarette use and cigarette smoking status

At wave 3 (~12 months from baseline), and wave 4 (~24 months from baseline), we classified participants into four mutually exclusive outcomes based on current e-cigarette and cigarette use (currently vaping/smoking every day or some days). Outcome categories included dual use of e-cigarettes and cigarettes, exclusive cigarette smoking, exclusive e-cigarette use, or non-use of both products.

E-cigarette product characteristics and behavior

Our correlates of interest were e-cigarette device type, flavors, vaping frequency, and time to first e-cigarette puff after waking. For device type, we first classified participants as e-cigarette users, or other electronic nicotine delivery system (ENDS) users. Participants were classified as e-cigarette users if they reported using an “e-cigarette (including vape pens and personal vaporizers).” Participants were classified as other ENDS users if they did not use an e-cigarette, but used an e-cigar, e-hookah, e-pipe, or another ENDS product. We then further classified e-cigarette devices as disposable, cartridge-based, or a refillable tank/modifiable (mod) system.

For flavor use, participants reported whether they used an e-liquid in the past 30-days flavored to taste like “menthol, mint, clove, spice, fruit, chocolate, alcoholic drinks, candy, or other sweets.” As in prior PATH studies,^2,25 participants responding yes were “non-tobacco flavor” users, and those responding no were “tobacco-flavor” users. Participants classified as using tobacco-flavors may actually use unflavored e-liquids (e.g., propylene glycol/vegetable glycerin base only), though these are rare and not captured by PATH.^2,33 We then further categorized non-tobacco flavors into mutually exclusive groups of “mint/menthol only”, “sweet flavors only“, or “both mint/menthol and sweet flavors”.

Participants who used e-cigarettes daily or 30 days/month were classified as daily users. Participants also reported how soon after waking they took their first e-cigarette puff. Responses were categorized as ≤5 minutes, 6–30 minutes, and >30 minutes. Table 1 contains definitions for each e-cigarette correlate.

Table 1.

Definitions for measures of e-cigarette characteristics and behaviors

E-cigarette Correlate	Definition
E-cigarette Device Type
Disposable e-cigarette	Participant uses an e-cigarette device that:
	• Is not rechargeable
	• Is not refillable
	• Does not use cartridges
	• Does not use tank system
Cartridge-based e-cigarettes	Participant uses an e-cigarette device that:
	• Is rechargeable
	• Is either refillable or not refillable
	• Uses cartridges
	• Does not use a tank system
Tank or Mod e-cigarettes	Participant uses an e-cigarette device that
	• Is rechargeable
	• Is refillable
	• Does not use cartridge
	• Uses a tank system or does not use a tank system
Other electronic nicotine delivery systems (ENDS)	Uses an ENDS product at baseline that is not an e-cigarette. Includes e-cigars, e-hookahs, e-pipes, or any other ENDS. Data not available on whether other ENDS product is disposable, cartridge-based, rechargeable, or refillable.
Flavor use
Tobacco flavor	Did not use an e-liquid in the past 30-days flavored to taste like mint, menthol, clove, spice, fruit, chocolate, alcoholic drinks, candy, or other sweets.
Mint/menthol flavor only (non-tobacco flavor)	Used e-liquid in the past 30-days flavored to taste like mint or menthol. Did not use any sweet flavor.
Sweet flavor only (non-tobacco flavor)	Used e-liquid in the past 30-days flavored to taste like clove, spice, fruit, chocolate, alcoholic drinks, candy, or other sweets. Did not use mint or menthol flavor.
Mint/menthol and Sweet flavors (non-tobacco flavor)	Used e-liquid in the past 30-days flavored to taste like mint or menthol, and used e-liquid flavored to taste like clove, spice, fruit, chocolate, alcoholic drinks, candy, or other sweets.
E-cigarette frequency
Daily e-cigarette users	Reported using e-cigarettes daily, or reported using e-cigarettes 30-days in the past month.
Non-daily e-cigarette users	Reported using e-cigarettes <30 days in the past month.
Minutes to first e-cigarette puff after waking
≤5 minutes	Takes first puff from e-cigarette 5 minutes or less after waking.
6–30 minutes	Takes first puff from e-cigarette 6–30 minutes after waking.
>30 minutes	Takes first puff from e-cigarette more than 30 minutes after waking

Covariates

Socio-demographic covariates were age (18–34, 35–54, ≥55 years), sex (male, female), race/ethnicity (non-Hispanic Black, non-Hispanic Other, Hispanic, Non-Hispanic white), poverty status (<100% vs. ≥100% the federal poverty limit), and education (<High School or general educational development (GED) certification, High School or some college, college or advanced degree).

Additional covariates were markers of nicotine dependence and reasons for using e-cigarettes. These included other tobacco use (past 30-day use of smokeless tobacco, snus, hookah, pipes, or cigars/cigarillos) and past 12-month use of nicotine replacement therapy (NRT) (nicotine patch, gum, inhaler, nasal spray, lozenge or pill.) For dual users, we examined cigarette smoking intensity (1–9, 10–19 ≥20 cigarettes/day) and frequency (daily vs. non-daily), time to first cigarette after waking (≤5, 6–30 ≥30 minutes), using e-cigarettes to cut-down on cigarette smoking (yes/no), and plans to quit smoking cigarettes (yes/no). For exclusive e-cigarette users, we examined duration since quitting smoking cigarettes (≤1, 2–12, >12 months), smoking intensity when a smoker (1–9, 10–19 ≥20 cigarettes/day), and using e-cigarettes as an alternative to quitting all tobacco (yes/no). Covariates were measured at wave 2 except for poverty status (wave 1).

Statistical Analysis

We analyzed exclusive e-cigarette and dual user samples separately. We conducted descriptive statistics of baseline covariates and e-cigarette correlates, and examined the distribution of e-cigarette use and cigarette smoking at 12 and 24 months. We used multinomial logistic regression to estimate odds ratios and 95% confidence intervals for the association between e-cigarette correlates and patterns of e-cigarette use and cigarette smoking at 24 months, with no change (i.e., remaining exclusive e-cigarette user or dual user) as the reference. Final models adjusted for age, sex, race/ethnicity, poverty status, and education.

In sensitivity analyses, we repeated analyses: (1) stratifying by duration since quitting smoking for exclusive e-cigarette users (<12 months vs. ≥12 months), and cigarette smoking frequency for dual users (daily vs. non-daily); (2) adjusting for markers of nicotine dependence and reasons for using e-cigarettes; (3) in subgroups of participants with no missing data (complete case analysis); (4) mutually adjusting for flavor use and device type in vaping frequency models and mutually adjusting for device type in flavor use models.

Analyses were conducted using STATA SE 16. We used multiple imputation with chained equations and 5 imputed datasets to impute missing covariate and outcome data. Missing data ranged from <1% (vaping frequency) to 9% (time to first e-cigarette). All analyses were weighted using PATH wave 4 longitudinal sample weights and the STATA svy commands to adjust for complex survey design and non-response over follow-up. Weighting ensures estimates are representative of the non-institutionalized US civilian population at the time of wave 4 (2016–2018). Confidence intervals for descriptive statistics and the complete case analysis were derived using replicate weights and the balanced repeated replication (BRR) method with the Fay adjustment (ρ = 0.3). Standard confidence intervals were used for primary regression estimates as BRR methods are not compatible with STATA multiple imputation commands. As recommended by the American Statistical Association, we do not rely on null hypothesis testing, and instead interpret estimates by their magnitude and precision.³⁴ This research was deemed exempt by the Boston University Medical Center Institutional Review Board.

RESULTS

Dual users had lower income and education and were more likely to report other tobacco use than exclusive e-cigarette users (Table 2). Exclusive e-cigarette users were more likely to use tank/mod style e-cigarettes and vape non-tobacco flavors, while dual users were more likely to use disposable or cartridge-based systems and vape tobacco flavors.

Table 2.

Baseline Characteristics of 2,229 Exclusive E-cigarette Users and Dual Users of E-cigarettes and Cigarettes, PATH 2014–2018

	Exclusive E-cigarette Users (n=470)		Dual users (n=1,759)
	n	% (95% CI)	n	% (95%CI)
Age, years
18–34	224	41.6 (36.2–47.1)	902	46.2 (43.3–49.1)
35–54	161	36.5 (30.7–42.2)	623	38.2 (35.3–41.1)
≥55	85	21.9 (17.1–26.7)	234	15.6 (13.5–17.6)
Female	224	41.0 (35.1–46.9)	901	46.6 (44.2–49.0)
Race/Ethnicity				9.8 (8.3–11.2)
Non-Hispanic Black	35	7.6 (4.2–10.9)	174
Non-Hispanic Other	38	7.3 (4.1–10.6)	145	7.1 (5.7–8.5)
Non-Hispanic White	341	75.5 (70.3–80.7)	1,225	73.4 (71.0–75.7)
Hispanic	56	9.6 (6.9–12.2)	215	9.8 (8.2–11.3)
<100% Federal Poverty Level	132	23.9 (19.0–28.8)	801	41.1 (38.5–43.6)
Education
<HS or GED	75	16.2 (12.1–20.4)	494	26.3 (24.0–28.6)
HS or Some college	324	69.5 (64.9–74.0)	1,083	61.7 (59.0–64.4)
BA or Advanced degree	71	14.3 (10.8–17.8)	182	12.0 (10.3–13.7)
Past 30-day other tobacco use a	102	20.3 (16.1–24.5)	689	36.8 (34.4–39.2)
Past 12-month nicotine replacement therapy	45	9.6 (6.6–12.7)	202	11.4 (9.9–12.8)
Device type
Disposable	38	8.4 (5.3–11.5)	304	17.5 (15.7–19.3)
Cartridges	120	26.2 (20.7–31.7)	660	38.3 (35.7–40.8)
Tank system/Mod	281	59.1 (53.0)	645	36.4 (33.9–39.0)
Other ENDS	31	6.3 (3.6)	150	7.8 (6.2–9.3)
Flavor of e-cigarette e-liquid (past 30-days)
Mint/menthol	50	11.2 (70.4–15.4)	212	12.3 (10.6–14.0)
Sweet flavors	220	46.0 (40.4–51.6)	468	26.4 (24.0–28.8)
Mint/menthol and sweet flavors	57	12.0 (8.9–15.2)	147	7.8 (6.2–9.3)
Tobacco flavor	143	30.7 (25.6–35.8)	932	53.5 (50.0–56.9)
Use e-cigarettes daily	286	61.5 (56.0–67.0)	308	18.7 (16.5–20.8)
Smoke combustible cigarettes daily			1,285	72.3 (69.5–75.0)
Minutes to first e-cigarette puff after waking
≤5	77	15.0 (11.4–18.7)	336	19.5 (17.3–21.6)
6–30	147	34.9 (28.8–40.9)	435	24.8 (22.3–27.2)
>30	246	50.1 (44.4–55.8)	988	55.7 (52.8–58.7)
Minutes to first cigarette after waking
≤5			412	23.5 (21.0–26.0)
6–30			616	34.3 (31.6–37.1)
>30			731	42.2 (39.4–44.9)
Months since quitting smoking cigarettes
≤1	81	15.5 (11.4–19.6)
2–12	168	35.2 (30.2–40.2)
>12	221	49.3 (43.4–55.2)
Cigarettes smoked per day b
1–10	244	47.2 (41.2–53.2)	1,009	56.4 (53.3–59.4)
11–19	60	13.8 (9.0–18.6)	290	16.0 (13.8–18.2)
≥20	166	39.0 (34.3–43.7)	460	27.6 (24.7–30.5)
Using e-cigarettes as an alternative to quitting tobacco altogether	337	70.9 (65.4–76.4)
Using e-cigarettes to cut-down on cigarette smoking			1,184	67.3 (66.6–70.7)
Plans to ever quit smoking cigarettes			1,449	83.0 (80.9–85.1)

All covariates measured at wave 2 except for poverty status, which is derived from wave 1.

Results represent unweighted frequencies and weighted percentages. Confidence intervals derived from replicate weights using the balanced repeated replication method with the Fay adjustment (ρ = 0.3)

^aIncludes use of smokeless/snus tobacco, hookah, traditional/filtered cigars, cigarillos, pipe tobacco

^bRefers to cigaretts smoked per day when a smoker for exclusive e-cigarette users

Patterns of E-cigarette Use and Cigarette Smoking at 12 and 24 months

Among exclusive e-cigarette users at baseline, 55% remained exclusive e-cigarette users, 22% were non-users of cigarettes and e-cigarettes, 15% were dual users, and 8% were exclusive cigarette smokers at 12 months (Figure 1). At 24 months, the percentage of exclusive e-cigarette users who remained exclusive e-cigarette users decreased to 44%, while 27% became non-users of both products, 14% changed to dual use, and 15% relapsed to exclusive cigarette smoking. Transitions differed by duration since quitting combustible smoking (Supplementary Figure 2). By 24 months, recent cigarette quitters (<12 months) were less likely than longer-term quitters (≥12 months) to remain exclusive e-cigarette users (35% vs. 54%) and to transition to non-use of both products (22% vs. 31%), and more likely to relapse to dual use (21% vs. 6%) and exclusive cigarette smoking (22% vs. 9%).

Figure 1.

Patterns of e-cigarette use and cigarette smoking at 12 and 24 months among former and current smokers who use e-cigarettes at baseline, PATH 2014–2018

Among dual users at baseline, 35% remained dual users, 5% were exclusive e-cigarette users, 7% quit both products, and 53% transitioned to exclusive cigarette smoking by 12 months. At 24 months, 26% of dual users remained dual users, 5% were exclusive e-cigarette users, 9% quit both products, and 60% transitioned to exclusive cigarette smoking. Transitions differed by smoking frequency (Supplementary Figure 3). By 24 months, daily cigarette smokers were less likely than non-daily smokers to transition to exclusive e-cigarette use (3% vs.11%) and non-use of both products (7% vs. 15%), and more likely to return to exclusive cigarette smoking (65% vs. 46%).

Correlates of Tobacco Use Transitions among Exclusive E-cigarette Users at Baseline

Transition to non-use of e-cigarettes and cigarettes at 24 months

Exclusive e-cigarette users who used disposable e-cigarettes, cartridge-based devices, or other ENDS compared to those who used tank/mods were more likely to transition to non-use of e-cigarettes and cigarettes at 24 months, relative to remaining an exclusive e-cigarette user (OR disposable: 2.68, 95% CI: 1.08–6.65; OR cartridge: 1.97, 95% CI: 1.03–3.78; OR other ENDS: 8.21, 95% CI: 2.54–26.6) (Table 3). Compared to tobacco flavors, any non-tobacco flavor use was inversely associated with non-use of both products at 24 months (OR: 0.41, 95% CI: 0.22–0.74). Adjusted odds ratios were similar for mint/menthol only (OR: 0.40, 95% CI: 0.15–1.04), sweet flavors only (OR: 0.42, 95% CI: 0.21–0.81), and both mint/menthol and sweet flavors (OR: 0.41, 95% CI: 0.16–1.02). Daily vaping (vs. non-daily) was inversely associated with non-use of both products at 24 months (OR: 0.06, 95% CI: 0.03–0.11).

Table 3.

Correlates of e-cigarette use and cigarette smoking patterns at 24-months among 470 baseline exclusive e-cigarette users, PATH, 2014–2018

	Non-User vs. Exclusive E-cigarette (n=132)		Exclusive Cigarette vs. Exclusive E-cigarette (n=76)		Dual User vs. Exclusive E-cigarette (n=66)
E-cigarette Characteristics and Behaviors	ORa	95% CI	ORa	95% CI	ORa	95% CI
Device type
Disposable	2.68	1.08–6.65	2.43	0.80–7.30	0.25	0.30–1.98
Cartridge	1.97	1.03–3.78	2.32	1.05–5.14	0.90	0.42–1.93
Tank/mod	REF	REF	REF	REF	REF	REF
Other ENDS	8.21	2.54–26.60	1.72	0.43–6.95	1.67	0.34–8.07
Flavors
Tobacco Flavor	REF	REF	REF	REF	REF	REF
Non-Tobacco Flavor	0.41	0.22–0.74	0.51	0.25–1.03	1.43	0.64–3.18
Mint/Menthol	0.40	0.15–1.04	0.83	0.25–2.66	1.34	0.36–4.90
Sweet flavor	0.42	0.21–0.81	0.50	0.22–1.11	1.51	0.64–3.57
Mint/Menthol and Sweet flavor	0.41	0.16–1.02	0.40	0.13–1.18	1.35	0.45–4.01
Vaping frequency
Daily vaping	0.06	0.03–0.11	0.19	0.09–0.40	0.50	0.24–1.03
Non-daily vaping	REF	REF	REF	REF	REF	REF
Minutes to first e-cigarette puff
≤5	REF	REF	REF	REF	REF	REF
6–30	0.54	0.21–1.42	0.84	0.30–2.36	0.62	0.23–1.65
>30	2.04	0.79–5.27	2.02	0.72–5.61	0.97	0.42–2.22

Abbreviations: CI, confidence interval; OR, odds ratio. Percentages are weighted using PATH longitudinal sample weights.

^aAdjusted for age, sex, race/ethnicity, education, poverty level. Outcome reference is exclusive e-cigarette use at 24-month follow-up (n=196). Sample interpretation of OR: Those with the characteristic at baseline were more likely (OR>1.0) or less likely (OR<1.0) than those without the characteristic to transition to the specified tobacco use pattern relative to remaining an exclusive e-cigarette user at 24 months.

Transition to dual use at 24 months

Daily vaping (vs. non-daily) was inversely associated with dual use at 24 months relative to remaining an exclusive e-cigarette user (OR: 0.50, 95% CI: 0.24–1.03).

Transition to exclusive cigarette smoking at 24 months

Cartridge-based devices compared to tank/mods were positively associated with exclusive cigarette smoking at 24 months versus remaining an exclusive e-cigarette user (OR: 2.32, 95% CI: 1.05–5.14). Non-tobacco flavors (vs. tobacco-flavors, OR: 0.51, 95% CI: 0.25–1.03), and daily vaping (vs. non-daily, OR: 0.19, 95% CI: 0.09–0.40) were inversely associated with exclusive cigarette smoking at 24 months. Estimates for specific flavor types were imprecise, however mint/menthol flavors conferred the weakest association (OR: 0.83, 95% CI; 0.25–2.66), with the strongest inverse association for both mint/menthol and sweet flavors (OR: 0.40, 95% CI: 0.13–1.18).

Correlates of Tobacco Use Transitions among Dual Users at Baseline

Transition to non-use of e-cigarettes and cigarettes at 24 months

Dual users who used disposable devices compared to tank/mod devices were more likely to report non-use of both products at 24 months relative to remaining a dual user (OR: 2.44, 95% CI: 1.31–4.57) (Table 4). Non-tobacco flavor use (vs. tobacco flavor) was inversely associated with transition to non-use of both products (OR: 0.43, 95% CI: 0.28–0.67). The strongest inverse associations were for mint/menthol (OR: 0.33, 95% CI: 0.16–0.68) and both mint/menthol and sweet flavors (OR: 0.30, 95% CI: 0.14–0.62).

Table 4.

Correlates of e-cigarette use and cigarette smoking patterns at 24-months among 1,759 baseline dual users of e-cigarettes and cigarettes, PATH, 2014–2018

	Non-User vs. Dual User (n=154)		Exclusive Cigarette vs. Dual User (n=1,055)		Exclusive E-cigarette vs. Dual User (n=94)
E-cigarette Characteristics	ORa	95% CI	ORa	95% CI	ORa	95% CI
Device type
Disposable	2.44	1.31–4.57	2.46	1.67–3.62	0.30	0.10–0.89
Cartridge	1.47	0.89–2.45	1.42	1.07–1.90	0.83	0.48–1.44
Tank/mod	REF	REF	REF	REF	REF	REF
Other ENDS	1.76	0.73–4.29	2.18	1.07–4.43	0.45	0.12–1.66
Flavors
Tobacco Flavor	REF	REF	REF	REF	REF	REF
Non-Tobacco Flavor	0.43	0.28–0.67	0.51	0.39–0.67	1.32	0.78–2.25
Mint/Menthol	0.33	0.16–0.68	0.65	0.43–0.99	0.95	0.41–2.18
Sweet flavor	0.54	0.32–0.90	0.50	0.37–0.69	1.63	0.91–2.92
Mint/Menthol and Sweet flavor	0.30	0.14–0.62	0.34	0.21–0.55	0.89	0.40–2.00
Vaping frequency
Daily vaping	0.91	0.51–1.62	0.38	0.28–0.53	3.77	2.17–6.52
Non-daily vaping	REF	RF	REF	REF	REF	REF
Minutes to first e-cigarette puff
≤5	REF	REF	REF	REF	REF	REF
6–30	0.73	0.31–1.76	0.83	0.52–1.32	2.53	1.10–5.81
>30	1.16	0.52–2.57	1.30	0.89–1.90	1.40	0.63–3.10

Abbreviations: CI, confidence interval; OR, odds ratio.

^aAdjusted for age, sex, race/ethnicity, education, poverty level. Outcome reference is dual use at 24-month follow-up (n=456). Sample interpretation of OR: Those with the characteristic at baseline were more likely (OR>1.0) or less likely (OR<1.0) than those without the characteristic to transition to the specified tobacco use pattern relative to remaining a dual user at 24 months.

Transition to exclusive cigarette smoking at 24 months

Dual users who used disposable, cartridge-based, or other ENDS devices compared to tank/mods were more likely to report exclusive cigarette smoking at 24 months relative to remaining a dual user (OR disposable: 2.46, 95% CI: 1.67–3.62; OR cartridge: 1.42, 95% CI: 1.07–1.90; OR other ENDS: 2.18, 95% CI: 1.07–4.43). Any non-tobacco flavor use (vs. tobacco flavor) was inversely associated with exclusive cigarette smoking relative to remaining a dual user (OR: 0.51, 95% CI: 0.39–0.67). The strongest inverse association was for both mint/menthol and sweet flavors (OR: 0.34, 95% CI: 0.21–0.55). Daily vaping (vs. non-daily) was inversely associated with exclusive cigarette smoking at 24 months (OR: 0.38, 95% CI: 0.28, 0.53).

Transition to exclusive e-cigarette use at 24 months

Using a disposable vs. tank/mod devices was inversely associated with transitioning to exclusive e-cigarette use relative to remaining a dual user (OR: 0.30, 95% CI: 0.10–0.89). Daily vs. non-daily vaping (OR: 3.77, 95% CI: 2.17–6.52) and waiting 6–30 minutes to smoke cigarettes after waking vs. <5 minutes (OR: 2.53, 95% CI: 1.10–5.81) were positively associated with exclusive e-cigarette use at 24 months relative to remaining a dual user.

Sensitivity Analyses

Among exclusive e-cigarette users, similar trends were seen when stratifying models by duration since quitting smoking (Supplementary Table 4). Among dual users, there was a noticeable difference in the association between vaping frequency and tobacco use transitions by cigarette smoking frequency (Supplementary Table 5). Among dual users who smoked cigarettes daily, daily vaping was positively associated with transition to exclusive e-cigarette use and negatively associated with transitioning to exclusive cigarette smoking relative to remaining a dual user. Among non-daily smokers, daily vaping was inversely associated with transitioning to non-use and exclusive e-cigarette use, but positively associated with transition to exclusive cigarette smoking relative to remaining a dual user. Associations persisted after adjustment for markers of nicotine dependence and reasons for using e-cigarettes (Supplementary Tables 6 and 7), and when restricting to participants with no missing data (Supplementary Tables 8 and 9). Mutual adjustment did not markedly alter results (Supplementary Table 10).

DISCUSSION

In this nationally representative cohort study of current and former cigarette smokers using e-cigarettes, longitudinal patterns of e-cigarette and cigarette use differed by cigarette smoking status and e-cigarette device characteristics. Approximately one quarter of exclusive e-cigarette users, and 9% of dual users of e-cigarettes and cigarettes, abstained from both vaping and cigarette smoking by 24 months. Vaping daily, non-tobacco flavors, and refillable tank/mod devices were associated with reduced likelihood of exclusive cigarette smoking and increased likelihood of exclusive e-cigarette use at 24 months among baseline exclusive e-cigarette and dual users.

Among exclusive e-cigarette users at baseline, 44% continued exclusive e-cigarette use after two years. However, among dual users at baseline, only 26% maintained dual use at two years, with 60% returning to exclusive cigarette smoking and just 5% transitioning to exclusive e-cigarette use. These findings highlight important differences in vaping patterns by cigarette smoking status: e-cigarette use remained relatively stable for exclusive e-cigarette users who were former smokers, while dual users were more likely to switch patterns of use. Our findings are supported by prior research showing vaping is highly dynamic for current smokers.^{17,18,35–37} We demonstrate exclusive vaping was the most common use pattern for former smokers who use e-cigarettes. These results highlight that vaping may not be a temporary smoking intervention, but rather a new adopted behavior once cessation is achieved. Our results corroborate other research showing many cigarette smokers continue vaping after quitting cigarette smoking.^5,38 Importantly, the percentage of participants who quit both vaping and smoking increased over time. In addition, exclusive e-cigarette users who had quit smoking less than twelve months from baseline, and dual users who were daily cigarette smokers were less likely to achieve abstinence from both products at two years. Our findings highlight the need to track both e-cigarette use and cigarette smoking over time, and to identify factors that facilitate smokers to quit both products.

Participants who vaped sweet or mint/menthol flavors (vs. tobacco flavors), and those that used refillable tank/mod devices (vs. cartridges, disposables, and other ENDS products), were less likely to abstain from vaping and smoking over follow-up, but also less likely to relapse to exclusive cigarette smoking. Dual users vaping disposable e-cigarettes were less likely than those using tank/mod devices to substitute exclusive vaping for dual use. Non-tobacco flavors and tank devices may aid in smoking abstinence by making vaping more palatable and customizable.^21–24,39 However, flavors and more customizable features may also encourage continued vaping after quitting smoking. Some customizable e-cigarettes have the ability to deliver greater nicotine exposure than disposables or cartridge-based devices, leading to sustained use. Other studies report flavored tobacco use is associated with sustained or more frequent use of the same product.^40,41 Our results and prior literature indicate a public health tradeoff: there may be e-cigarette characteristics that appeal to smokers and help smokers switch to e-cigarettes and maintain smoking abstinence that simultaneously make it harder to quit vaping in the long-term.

In this study, greater vaping frequency was associated with smoking abstinence through e-cigarette substitution, but lower likelihood of abstaining from both vaping and smoking. Importantly, daily vaping was only beneficial for daily cigarette smokers, and was inversely associated with smoking abstinence for non-daily cigarette smokers. Prior studies find e-cigarettes may be more efficacious for cigarette cessation with greater vaping frequency.^8,42 Our results indicate that frequent vaping may only benefit daily cigarette smokers. Daily and non-daily smokers are likely distinct populations^43,44 and should be assessed separately in studies of e-cigarette use and smoking abstinence.

Our findings have implications for tobacco regulations. Given that 70% of former smokers and 47% of current smokers in our nationally representative sample used a characterizing (non-tobacco) flavor, there may be tradeoffs of banning flavored vaping options. On the one hand, flavors may help with smoking abstinence by making e-cigarettes more appealing and palatable. On the other hand, flavorings may contribute to greater long-term usage of e-cigarettes, and restricting flavors may help reduce the percentage of the population addicted to e-cigarettes. In addition, prior research suggests e-cigarette flavors appeal to youth,⁴⁵ and thus attention should be given to the impact of regulating flavors on youth populations. Similar tradeoffs apply to regulation of customizable devices, which may be associated with greater intensity of youth cigarette smoking.⁴⁶ Results indicate a need for studies that examine differences between e-cigarette product characteristics rather than homogenizing e-cigarettes as one single product.

Our study has limitations. Outcomes were defined as current use, however some participants may have used e-cigarettes or cigarettes between surveys, but not at the time of data collection. We did not define the temporal ordering of e-cigarette use relative to cigarette smoking.

However, most adults initiate e-cigarettes to help quit or reduce cigarette smoking,^21,47,48 and so we expect most dual users initiated cigarettes prior to e-cigarettes, and most exclusive e-cigarette users initiated e-cigarettes prior to quitting smoking (See Supplementary Tables 1-3). Due to small numbers, we were unable to evaluate product characteristics with greater granularity or in combination. There may be residual confounding and results should be interpreted as associations rather than effects. We excluded participants missing follow-up data; if loss to follow-up is associated with product characteristics and tobacco use outcomes, this may induce selection bias.⁴⁹ Importantly, the tobacco product landscape has evolved since the study period, and we do not capture more current products such as pod-mod e-cigarettes.⁴ Finally, our results apply to adults in the US, and should not necessarily be generalized to youth populations who have different vaping motivations and behaviors, or to other countries with different tobacco products, policies and norms.

This study revealed that many former smokers who used e-cigarettes continued to vape after two years, while most current smokers returned to exclusive cigarette smoking. Current and former smokers who vaped with greater frequency, used non-tobacco flavors, and used refillable tank or mod devices were less likely to transition to exclusive cigarette smoking, but were also less likely to abstain from vaping over follow-up. When making regulatory decisions, policy makers will need to consider the trade-off between preventing more smokers and youth from becoming addicted to vaping, versus potentially helping smokers quit smoking or maintain smoking abstinence through e-cigarette substitution.

WHAT THIS PAPER ADDS:

• Few studies examine how e-cigarette use evolves over time concurrently with combustible cigarette smoking, and whether there are modifiable e-cigarette product characteristics associated with more desired trajectories (e.g., abstinence from both products).

• We used three waves of the nationally representative Population Assessment of Tobacco and Health (PATH) study and found that after two years, 44% of exclusive e-cigarette users at baseline continued exclusive e-cigarette use. Among dual users at baseline, 26% maintained dual use at two years, and 60% returned to exclusive cigarette smoking.

• The percentage of participants who abstained from both vaping and smoking increased over time. Exclusive e-cigarette users who had quit cigarette smoking <12 months from baseline (vs. =12 months), and dual users who were daily cigarette smokers (vs. non-daily) were less likely to achieve abstinence from both products at two years.

• Vaping daily, flavors, and tank/modifiable devices were associated with lower likelihood of abstaining from both vaping and cigarette smoking, and lower likelihood of returning to exclusive cigarette smoking after two years.