Cigarette Smoking Relapse Among People Who Switched to E-cigarettes or Other Tobacco Products

Abstract

Introduction

Little is known about how quitting and switching to other tobacco products affects cigarette smoking relapse. Additionally, there is no consensus on the best definition of relapse.

Aims and Methods

Respondents who smoked cigarettes at baseline, had quit by the first follow-up, and completed the second follow-up survey in the Population Assessment of Tobacco and Health Study were included. We employed multivariable logistic regressions to assess the association between non-cigarette tobacco use at follow-up 1 and smoking relapse risk at follow-up 2, considering three definitions of relapse.

Results

54.8% of the 1686 respondents who had recently quit smoking cigarettes relapsed according to Measure I (any smoking in the past 12 months), 40.3% using Measure II (any smoking in the past 30 days), and 30.1% using Measure III (smoked on ≥3 days in the past 30 days). Compared with no tobacco use at follow-up 1, any tobacco use was associated with increased relapse risk using Measure I (adjusted risk difference [aRD] = 7.14, CI [0.64 to 13.64]). The association was nonsignificant using Measures II (aRD = -0.53, CI [-6.62 to 5.56]) or III (aRD = −4.11, CI [−9.28 to 1.06]). Similarly, exclusive e-cigarette use was significantly associated with Measure I, but not with Measures II or III.

Conclusions

Compared with respondents who had recently quit and were tobacco-free at follow-up 1, those who switched to non-cigarette tobacco products may be more likely to slip but showed no difference in the likelihood of relapse when employing a relapse definition less strict than any smoking in the past 12 months.

Implications

The distinction between a slip and relapse needs to be considered more carefully, with researchers undertaking further studies that help us contemplate how we should define relapse. People who recently quit smoking cigarettes and switched to non-cigarette tobacco products (including e-cigarettes) may be more likely to slip compared to people who recently quit and were tobacco-free, but they are not clearly more likely to relapse.

Introduction

Although the prevalence of cigarette smoking has steadily decreased in the United States,¹ 11.2% of U.S. adults (27.6 million) reported current cigarette smoking in 2022.² Smoking remains the leading cause of preventable death in the United States.

Most people who smoke want to quit smoking, and many use multiple methods in their quit attempts.^3–5 Around a quarter of individuals who smoke cigarettes reported using e-cigarettes to quit in their most recent quit attempt.⁶ Many studies, including clinical trials^7–9 and population-based observations,^10–12 have examined the effectiveness of using e-cigarettes for smoking cessation. A recently updated Cochrane review concluded that there is high-certainty evidence that e-cigarettes increase quit rates compared to nicotine replacement therapy.¹³ In addition, frequent use of e-cigarettes is associated with higher odds of cessation.^10,14,15

Most cessation studies have focused on quitting smoking per se, without considering the nature of relapse that defines failure to maintain cessation. Despite effective smoking cessation aids and treatments, relapse is common.^16–18 In 2018, more than half of adults who smoke cigarettes had made a quit attempt but only 8% were successful in quitting for 6–12 months.¹⁹

A few longitudinal studies have examined the association between e-cigarette use and cigarette smoking relapse.^20–24 While e-cigarette use was consistently associated with higher smoking relapse among individuals who quit smoking long-term,^20–22 the association is less clear among individuals who recently quit smoking cigarettes, usually defined as those who had quit in the past year. Some researchers reported a positive association,²² but others found the association to be nonsignificant.^20,21,23

These studies have adopted a variety of relapse definitions, ranging from the most restrictive definition, any smoking in the last 12 months,²³ to the most lenient, ≥1 cigarette per day.²⁴ There is no consensus on the best definition of relapse, and it remains unknown how different definitions affect the association reported. Another deficiency in the literature is the lack of attention to multiple quitting methods. Many people who smoke cigarettes use several methods during their most recent quit attempt.⁶ Simply comparing those who switched to e-cigarettes with those who did not switch overlooks the behavior of using multiple tobacco products to quit, leading to potentially biased results. One study addressed this issue by comparing people who quit smoking and switched to non-cigarette tobacco products with their tobacco-free counterparts.²³ Lastly, it is important to distinguish switching to combustible tobacco products from switching to noncombustible products. Individuals who quit smoking cigarettes and switch to smoking non-cigarette combustibles may have limited health gains due to cigarette-like toxic chemical exposure due to the combustion of tobacco.

In this study, we examined how switching to e-cigarettes or other non-cigarette tobacco products affects cigarette smoking relapse among people who recently quit smoking cigarettes, considering the issues mentioned above. Consistent with previous studies,^20–24 switching does not identify when the non-cigarette product was first used. While some people who smoke cigarettes may have started using non-cigarette tobacco products before quitting cigarette smoking, others may have started after quitting. Note that it is common for people quitting smoking to have a period of dual use with the alternative product before ceasing cigarette smoking. We used 4 annual waves (waves 1–4) of the Population Assessment of Tobacco and Health (PATH) Study from 2013 to 2018 to construct two independent cohorts of respondents who recently quit smoking cigarettes, defined as those who smoked cigarettes at the baseline wave and did not smoke cigarettes at follow-up 1, a year later. Then we assessed their cigarette smoking relapse behaviors at follow-up 2, another year later, adopting alternative definitions of relapse. Based on their non-cigarette tobacco use behaviors at follow-up 1, we divided these respondents who recently quit smoking cigarettes into different user groups. Controlling for multiple independent risk factors identified by previous studies,²⁵ we examined whether individuals who recently quit smoking cigarettes and switched to non-cigarette tobacco products were more or less likely to relapse to cigarette smoking by follow-up 2 than individuals who had recently quit who were tobacco-free, considering three different definitions of relapse. We also examined if the frequency of tobacco product use affects the relapse risk.

Methods

Sample

The PATH Study is a national longitudinal cohort study that surveys a stratified sample of the civilian, noninstitutionalized population to examine tobacco use behaviors, attitudes and beliefs, and tobacco-related outcomes. The first wave was initiated in September 2013 and annual or biennial follow-up surveys have been conducted since. Follow-up surveys for individual participants were conducted during a target data collection period, starting 1 month before the anniversary month of the individual’s prior survey and ending 1 to 2 months after the anniversary month.²⁶ The weighted response rate for the wave 1 adult survey was 74.0%, with follow-up response rates at 83.2%, 78.4%, and 73.5% for waves 2, 3, and 4, respectively.²⁶

Following definitions from prior studies,^20,23 we included adults, ages 18 years, who: (1) had participated in three consecutive surveys that were one year apart (baseline, follow-up 1, and follow-up 2); (2) had an established smoking behavior at baseline (ie, they had smoked at least 100 cigarettes in their lifetime and currently smoked cigarettes every day or some days); and (3) reported no current cigarette smoking at follow-up 1. We refer to them as individuals who recently quit smoking cigarettes. Note that these individuals may have used other tobacco products at baseline. We identified 1945 respondents who recently quit smoking cigarettes at follow-up 1 and 259 were excluded due to the lack of follow-up 2 data. Those lost to follow-up showed similar characteristics to the ones included in the sample (Tables S1 and S2). Our sample consisted of 1686 respondents who recently quit smoking cigarettes, with 905 from cohort 1 (waves 1–3) and 781 from cohort 2 (waves 2–4). By investigating their use of non-cigarette tobacco products at follow-up 1 and their cigarette relapse behaviors at follow-up 2, we evaluated whether non-cigarette tobacco product use affects subsequent smoking relapse, considering three different definitions of relapse.

Measures

Smoking Relapse

We adopted three binary measures of smoking relapse at follow-up 2,^24,27 ranging from the strictest definition to successively more lenient ones that allow for occasional slips: (1) any cigarette smoking in the past 12 months (Measure I); (2) any cigarette smoking in the past 30 days (Measure II), which allows for slips prior to the most recent month; and (3) having smoked on ≥3 days in the past 30 days (Measure III), which allows for lapsing up to 2 days in the most recent month.

Tobacco Product Use

At each wave, participants were asked, “Do you now smoke/use [tobacco product]?” For each tobacco product, including cigarettes, e-cigarettes, cigars, pipe, hookah, smokeless tobacco, dissolvable tobacco, and snus, we categorized participants as current users of that product if they answered “every day” or “some days” and daily users if they answered “every day.”

We divided respondents who recently quit smoking cigarettes into different groups based on their tobacco product use at follow-up 1: (1) tobacco-free (no current use of any tobacco product); (2) any tobacco use (use of any non-cigarette tobacco products); (3) any combustible use (use of any combustible tobacco product with or without noncombustible products); (4) exclusive noncombustibles use (exclusive use of noncombustible products excluding e-cigarettes); and (5) exclusive e-cigarette use. We also separately identified the current and daily use of at least one tobacco product for the latter 4 groups.

Study Covariates

We included an extensive list of covariates to adjust for sociodemographics, psychological status, environmental factors, use of any FDA-approved smoking cessation aids, the existence of any heart disease/respiratory disease/cancer, perceptions of harmfulness, and cigarette consumption history. Sociodemographic variables were age, sex, race, ethnicity, education, and household income at baseline. A 3-level severity measure (low, moderate, and high) was used for both baseline internalizing and externalizing mental health problems to assess psychological status (Table S3).²⁸ Environmental factors included baseline exposure to secondhand smoke and smoke-free home at follow-up 1. FDA-approved cessation aid use at follow-up 1 was identified by any current use of Chantix, varenicline, Wellbutrin, Zyban, bupropion, nicotine patch, gum, inhaler, nasal spray, lozenge, or pill. We included the existence of any lifetime diagnosis of heart disease, respiratory disease, or cancer at baseline. The baseline perceived harmfulness of cigarettes and relative perceived harmfulness of e-cigarettes were also included. Cigarette consumption history was captured by baseline daily cigarette consumption, a 16-item baseline tobacco dependence index,²⁹ and the duration of cigarette abstinence at follow-up 1.

Analysis

We conducted multivariable logistic regressions to examine the association between non-cigarette tobacco product use at follow-up 1 and cigarette smoking relapse by follow-up 2 among respondents who recently quit smoking cigarettes, controlling for the covariates identified above. Separate regressions were performed comparing people who recently quit smoking cigarettes by follow-up 1 and were tobacco-free (1) with those who used any tobacco product and (2) with those who used any combustibles, used noncombustibles exclusively, and used e-cigarettes exclusively. The analyses were performed using Stata version 18, with Fay’s method of balanced repeated replication to estimate variance. We incorporated longitudinal survey weights using Stata’s “svy” command, adjusting for variable probabilities of selection, differential nonresponse rates, and possible deficiencies in the sampling frame.²⁶

We report our main results as adjusted risk differences and absolute risks using Stata’s “margins” command. Due to the high prevalence of smoking relapse, an effect or association that appears small according to ratios may be substantial in terms of absolute risks. We therefore decided to report risk differences based on recommendations from Holmberg and Anderson.³⁰

Sensitivity Analysis

We conducted several sensitivity analyses: (1) including only respondents who had 3 or more years of fairly regular smoking, based on previous literature^20,23; (2) adding another smoking relapse measure defined as current daily or non-daily smoking^20,22,31; (3) examining the association between any e-cigarette use at follow-up 1 (with or without other tobacco products) and cigarette smoking relapse; and (4) removing longitudinal survey weights.

Results

Adjusting for longitudinal survey design, 42.0% (95% CI: 39.1% to 45.0%) of individuals who recently quit smoking cigarettes were female, 48.0% (95% CI: 45.0% to 51.1%) were between 18 and 35, and the majority (79.3%; 95% CI: 76.2% to 82.2%) were White. Around a third (34.0%; 95% CI: 31.0% to 37.0%) reported current tobacco use at follow-up 1, with 11.8% (95% CI: 9.9% to 14.0%) using any combustibles, 4.4% (3.0% to 6.3%) exclusively using noncombustibles (excluding e-cigarettes), and 17.0% (14.6% to 19.7%) exclusively using e-cigarettes at follow-up 1. 19.7% (95% CI: 17.4% to 22.2%) reported daily tobacco use, with 2.7% (95% CI: 1.9% to 3.9%) smoking combustibles, 2.8% (95% CI: 2.0% to 3.8%) using exclusively noncombustibles (excluding e-cigarettes), and 14.0% (95% CI: 11.8% to 16.5%) using exclusively e-cigarettes (Table S4). More than half (54.8%; 95% CI: 52.2% to 57.5%) of individuals who recently quit smoking cigarettes relapsed according to Measure I, 40.3% (95% CI: 37.3% to 43.3%) using Measure II, and 30.1% (95% CI: 27.2% to 33.1%) using Measure III.

Table S5 shows the baseline characteristics of individuals who recently quit smoking cigarettes, by their non-cigarette tobacco product use status at follow-up 1. Compared with individuals who recently quit smoking cigarettes and did not use any tobacco product, those who used non-cigarette tobacco at follow-up 1 were younger, more likely to be male or non-Hispanic, with higher educational attainment and household income, less likely to use FDA-approved cessation aids, more likely to have non-smoke-free homes, and to perceive e-cigarettes to be less harmful than cigarettes. Individuals who recently quit smoking cigarettes and used non-cigarette tobacco products at follow-up 1 also reported higher tobacco dependence index scores, 45.0 (95% CI: 41.8 to 48.2) versus 34.2 (95% CI: 31.9 to 36.6), more cigarettes smoked daily at baseline, 8.4 (95% CI: 7.2 to 9.6) versus 7.0 (95% CI: 6.3 to 7.7), and more exposure to secondhand smoke in the past 7 days, 13.2 hours (95% CI: 11.0 to 15.4) versus 10.2 hours (95% CI: 9.0 to 11.4).

Table 1 presents the prevalence of cigarette smoking relapse at follow-up 2 by non-cigarette tobacco product use at follow-up 1. The prevalence of smoking relapse among individuals who currently used non-cigarette tobacco, compared with their tobacco-free counterparts, was significantly higher using Measure I, 60.1% (95% CI: 54.6% to 65.5%) versus 52.1% (95% CI: 48.8% to 55.4%), but similar using Measure II, 41.2% (95% CI: 36.0% to 46.4%) versus 39.8% (95% CI: 36.5% to 43.3%), or Measure III, 29.2% (95% CI: 24.5% to 33.9%) versus 30.6% (95% CI: 27.5% to 33.8%). Similar patterns were found comparing individuals who recently quit smoking cigarettes and were tobacco-free at follow-up 1 with those reporting current exclusive e-cigarette use, daily tobacco use, and daily combustible use at follow-up 1.

Table 1.

Prevalence of Smoking Relapse at Follow-up 2 by Tobacco Product Use at Follow-up 1 Among Respondents Who Recently Quit Smoking Cigarettes

	Prevalence of smoking relapse Weighted % (95% CI)
Tobacco product use at follow-up I	Relapse measure Ia n = 957	Relapse measure IIb n = 719	Relapse measure IIIc n = 549
Overall	54.8 (52.2 to 57.5)	40.3 (37.3 to 43.3)	30.1 (27.2 to 33.1)
Tobacco-free (REF)	52.1 (48.8 to 55.4)	39.8 (36.5 to 43.3)	30.6 (27.5 to 33.8)
Current tobaccod	60.1 (54.6 to 65.5)	41.2 (36.0 to 46.4)	29.2 (24.5 to 33.9)
Current combustiblese	59.4 (48.7 to 55.5)	41.0 (32.5 to 49.4)	24.0 (17.1 to 30.8)
Current exclusive noncombustiblesf	58.5 (45.5 to 71.4)	34.0 (20.5 to 47.6)	23.9 (13.4 to 34.4)
Current exclusive e-cigarette	61.0 (54.4 to 67.6)	43.3 (36.8 to 49.8)	33.7 (27.7 to 39.6)
Daily tobaccod	59.6 (53.9 to 65.4)	42.2 (37.0 to 47.4)	31.3 (26.5 to 36.2)
Daily combustiblese	68.9 (54.2 to 83.7)	55.7 (40.1 to 71.3)	34.9 (19.3 to 50.5)
Daily exclusive noncombustiblesf	52.0 (34.4 to 69.5)	37.3 (19.7 to 54.9)	29.9 (14.3 to 45.4)
Daily exclusive e-cigarette	59.2 (52.3 to 66.1)	40.3 (34.3 to 46.2)	30.4 (24.6 to 36.2)

Prevalence was calculated without adjusting for any covariates.

p-value was calculated by comparing the prevalence of smoking relapse between the specific tobacco use group and the no tobacco use group. Figures are bolded if p < .05.

^aAny cigarette smoking in the past 12 months.

^bAny cigarette smoking in the past 30 days.

^cSmoked cigarettes on ≥3 days in the past 30 days.

^dTobacco products included e-cigarettes, cigars, pipe, hookah, smokeless tobacco, dissolvable tobacco, and snus.

^eCombustible tobacco products included cigars, pipe, and hookah.

^fNoncombustible tobacco products included smokeless tobacco, dissolvable tobacco, and snus.

Adjusted risk differences (aRDs) of relapse at follow-up 2 by current non-cigarette tobacco product used at follow-up 1 among individuals who recently quit smoking cigarettes, controlling for all the covariates noted previously, are reported in Table 2. Using Measure I, individuals who recently quit smoking cigarettes and were current non-cigarette tobacco users at follow-up 1 had significantly higher risks of relapse compared with those who were tobacco-free (aRD = 7.14 percentage points; 95% CI: 0.64 to 13.64). However, the risk differences became smaller and nonsignificant when we used Measure II to assess smoking relapse (aRD = −0.53 percentage point; 95% CI: −6.62 to 5.56) or Measure III (aRD = −4.11 percentage points; 95% CI: −9.28 to 1.06). We found similar results comparing individuals who recently quit smoking cigarettes and exclusively used e-cigarettes with those who were tobacco-free, with aRDs decreasing from 8.39 percentage points (95% CI: 0.03 to 16.76) using Measure I, to 0.70 percentage points (95% CI: −7.10 to 8.50) using Measure II and −1.60 percentage points (95% CI: −8.38 to 5.17) using Measure III. The aRDs were nonsignificant for individuals who used any combustibles or used noncombustibles exclusively.

Table 2.

Adjusted Risk Differences of Relapse at Follow-up 2 Among Respondents Who Recently Quit Smoking Cigarettes, by Current Tobacco Product Use at Follow-up 1

Current use at follow-up 1	aRDa % points, 95% CI	p-value	Risk w/o tobacco use (%)	Risk with use (%)
Cigarette relapse measure I b
Tobacco-free (REF)			54.74 (50.89 to 58.59)
Any tobaccoe	7.14 (0.64 to 13.64)	.03		61.88 (57.04 to 66.72)
Any combustiblesf	6.13 (−2.46 to 14.71)	.16		60.94 (53.09 to 68.79)
Any exclusive noncombustiblesg	4.36 (−8.30 to 17.01)	.50		59.17 (47.23 to 71.12)
Any exclusive e-cigarettes	8.39 (0.03 to 16.76)	.05		63.21 (56.50 to 69.92)
Cigarette relapse measure II c
Tobacco-free (REF)			42.21 (38.23 to 46.19)
Any tobaccoe	−0.53 (−6.62 to 5.56)	.86		41.68 (36.56 to 46.80)
Any combustiblesf	2.09 (−6.47 to 10.65)	.63		44.42 (36.63 to 52.21)
Any exclusive noncombustiblesg	−13.13 (−29.51 to 3.26)	.12		29.21 (13.20 to 45.21)
Any exclusive e-cigarettes	0.70 (−7.10 to 8.50)	.86		43.03 (36.21 to 49.86)
Cigarette relapse measure IIId
Tobacco-free (REF)			33.64 (30.16 to 37.12)
Any tobaccoe	−4.11 (−9.28 to 1.06)	.12		29.53 (24.84 to 34.22)
Any combustiblesf	−5.72 (−14.19 to 2.74)	.18		27.98 (19.89 to 36.08)
Any exclusive noncombustiblesg	−12.96 (−26.88 to 0.96)	.07		20.75 (6.69 to 34.81)
Any exclusive e-cigarettes	−1.60 (−8.38 to 5.17)	.64		32.11 (26.04 to 38.17)

Figures are bolded if p < .05.

^aAdjusted risk difference. Adjusted for all study covariates at baseline: age, sex, race, ethnicity, education, household income, internalizing mental health problems, externalizing mental health problems, use of any FDA-approved smoking cessation aids (follow-up 1), exposure to secondhand smoke, smoke-free home (follow-up 1), lifetime diagnosis of heart disease/respiratory disease/cancer, perceived harmfulness of cigarettes, relative perceived harmfulness of e-cigarettes to cigarettes, daily cigarette consumption, tobacco dependence, and the duration of cigarette abstinence (follow-up 1).

^bAny cigarette smoking in the past 12 months.

^cAny cigarette smoking in the past 30 days.

^dSmoked cigarettes on ≥3 days in the past 30 days.

^eTobacco products included e-cigarettes, cigars, pipe, hookah, smokeless tobacco, dissolvable tobacco, and snus.

^fCombustible tobacco products included cigars, pipe, and hookah.

^gNoncombustible tobacco products included smokeless tobacco, dissolvable tobacco, and snus.

In Table 3, we compare individuals who recently quit smoking cigarettes who were tobacco-free with daily tobacco users at follow-up 1. The patterns are similar to those in Table 2, except for those who used combustibles daily. Individuals who used tobacco products daily reported significantly higher risks of smoking relapse using Measure I (aRD = 8.30 percentage points; 95% CI: 1.13 to 15.46), but the risk differences were nonsignificant using Measure II (aRD = 1.07 percentage points; 95% CI: −5.40 to 7.53) or Measure III (aRD = −2.77 percentage points; 95% CI: −9.07 to 3.54). Individuals who used combustibles daily had significantly higher risks of smoking relapse using Measures I (aRD = 20.31 percentage points; 95% CI: 5.85 to 34.77) and II (aRD = 18.56 percentage points; 95% CI: 1.05 to 36.06), but not Measure III (aRD = 6.00 percentage points; 95% CI: −11.94 to 23.94). For all three measures of relapse, individuals who reported exclusive use of noncombustibles daily or exclusive use of e-cigarettes daily showed no difference in smoking relapse risks compared with those who were tobacco-free. Complete regression results can be found in Tables S6-S9.

Table 3.

Adjusted Risk Differences of Relapse at Follow-up 2 Among Respondents Who Recently Quit Smoking Cigarettes, by Daily Tobacco Product Use at Follow-up 1

Daily use at follow-up 1	aRDa % points, 95% CI	p-value	Risk w/o tobacco use (%)	Risk with use (%)
Cigarette relapse measure I b
Tobacco-free (REF)			54.71 (50.85 to 58.58)
Daily tobaccoe	8.30 (1.13 to 15.46)	.02		63.01 (57.63 to 68.39)
Daily combustiblesf	20.31 (5.85 to 34.77)	.01		75.05 (61.44 to 88.67)
Daily exclusive noncombustiblesg	−0.95 (−15.62 to 13.72)	.90		53.79 (39.53 to 68.06)
Daily exclusive e-cigarettes	7.40 (−1.27 to 16.06)	.09		62.14 (55.07 to 69.20)
Cigarette relapse measure II c
Tobacco-free (REF)			42.17 (38.18 to 46.16)
Daily tobaccoe	1.07 (−5.40 to 7.53)	.74		43.24 (37.89 to 48.59)
Daily combustiblesf	18.56 (1.05 to 36.06)	.04		60.75 (43.41 to 78.08)
Daily exclusive noncombustiblesg	−5.65 (−22.84 to 11.54)	.52		36.54 (20.04 to 53.04)
Daily exclusive e-cigarettes	−1.24 (−8.76 to 6.29)	.75		40.95 (34.51 to 47.40)
Cigarette relapse measure III d
Tobacco free (REF)			33.61 (30.11 to 37.10)
Daily tobaccoe	−2.77 (−9.07 to 3.54)	.39		30.84 (25.28 to 36.40)
Daily combustiblesf	6.00 (−11.94 to 23.94)	.51		39.61 (22.04 to 57.18)
Daily exclusive noncombustiblesg	−3.97 (−21.24 to 13.29)	.65		29.63 (12.83 to 46.43)
Daily exclusive e-cigarettes	−4.55 (−11.71 to 2.60)	.21		29.05 (22.47 to 35.64)

Figures are bolded if p < .05.

^aAdjusted risk difference. Adjusted for all study covariates at baseline: age, sex, race, ethnicity, education, household income, internalizing mental health problems, externalizing mental health problems, use of any FDA-approved smoking cessation aids (follow-up 1), exposure to secondhand smoke, smoke-free home (follow-up 1), lifetime diagnosis of heart disease/respiratory disease/cancer, perceived harmfulness of cigarettes, relative perceived harmfulness of e-cigarettes to cigarettes, daily cigarette consumption, tobacco dependence, and the duration of cigarette abstinence (follow-up 1).

^bAny cigarette smoking in the past 12 months.

^cAny cigarette smoking in the past 30 days.

^dSmoked cigarettes on ≥3 days in the past 30 days.

^eTobacco products included e-cigarettes, cigars, pipe, hookah, smokeless tobacco, dissolvable tobacco, and snus.

^fCombustible tobacco products included cigars, pipe, and hookah.

^gNoncombustible tobacco products included smokeless tobacco, dissolvable tobacco, and snus.

Findings in sensitivity analyses are shown in Tables S10-S14. Adding the inclusion criterion of ≥3 years of fairly regular smoking, the patterns of results are similar to our main findings: some associations were significant or marginally significant using Measure I but all were nonsignificant using Measures II and III. The alternative relapse definition of current daily or non-daily cigarette smoking at follow-up 2 yielded similar findings as Measures II and III. The associations between any e-cigarette use (with or without other non-cigarette tobacco products) and smoking relapse were almost identical to those between exclusive e-cigarette use and relapse. Removing survey weights also yielded similar findings to our main results.

Discussion

Using the relapse definition of any cigarette smoking in the past 12 months (Measure I), respondents who recently quit smoking cigarettes and had switched to current non-cigarette tobacco product use or exclusive e-cigarette use had higher risks of relapse. Our findings using Measure I are consistent with a prior study that concluded switching to any tobacco product increased relapse risk.²³ However, if we allow for occasional slips, using Measures II and III, these individuals who recently quit smoking cigarettes and had switched showed no difference in smoking relapse risk. The contrasting findings using alternative definitions of smoking relapse highlight the importance of properly assessing and defining smoking relapse.

Another contribution of our study, compared with previous literature, is the separate evaluation of combustible and noncombustible tobacco products. Since most of the harm associated with cigarette smoking derives from the toxic chemical exposure produced by combustion,³² smoking other non-cigarette combustible tobacco products also generates substantial harm. Consuming noncombustible tobacco products (including e-cigarettes), on the other hand, does not involve combustion and is thus less harmful.^33,34 We found different patterns between combustibles and noncombustibles with smoking relapse. While daily use of any combustible product is statistically significantly associated with smoking relapse (Measures I and II), the exclusive use of noncombustibles or e-cigarettes is not.

The significant associations between daily combustible tobacco use and smoking relapse (Measures I and II), which do not hold for current use, suggest that the frequency of tobacco product use matters for relapse. This dose–response relationship was also reported by previous studies on smoking relapse and the literature on smoking cessation. Regarding the latter, for example, many studies have reported that daily or frequent e-cigarette use increases the odds of smoking cessation.^10,12,14,15

A possible explanation for the inconsistent findings across alternative definitions of smoking relapse is the difference between a slip and a relapse. A slip is a limited episode of smoking over a limited time; it does not necessarily constitute a resumption of smoking. On the other hand, relapse is when an individual who has quit smoking returns to regular smoking.²⁷ Our Measure I, any smoking in the past 12 months, does not distinguish relapse from slips, while Measures II and III allow for occasional slips. Future studies are needed to evaluate and compare these alternative definitions of relapse with longer follow-up periods.

Another possible explanation lies in the assessment of relapse. Our measure of relapse by follow-up 2 captures the relapse outcomes of the most recent quit attempt. It is possible that some individuals who recently quit smoking cigarettes had already relapsed and tried quitting again between follow-up 1 and follow-up 2. Survey data with higher frequency, such as quarterly follow-ups, could examine smoking relapse behaviors more accurately.

People who quit cigarettes and experience nicotine withdrawal symptoms may use non-cigarette tobacco products as alternative nicotine sources. This may be especially true for individuals who smoke a lot of cigarettes or those with stronger nicotine dependence at baseline. Switching to exclusive noncombustibles, including e-cigarettes, permits the maintenance of nicotine needs without subjecting these individuals to combustion-level health risks. E-cigarettes, with more efficient nicotine delivery³⁵ and mimicking the sensory experiences of smoking a traditional cigarette,³⁶ may prevent some people who recently quit smoking cigarettes from relapsing.

Findings from this study also have important implications for the impact of e-cigarettes on smoking cessation. Smoking cessation is a complex and dynamic process. Many studies, both clinical trials and population-based observational studies, have shown that e-cigarette use, especially frequent e-cigarette use, can help people quit smoking cigarettes.¹⁴ However, if these people who recently quit smoking cigarettes are prone to relapse when they switch to e-cigarettes, then the overall effect of e-cigarettes on smoking cessation remains ambiguous. Our results, finding that e-cigarette use is only associated with relapse using Measure I but not with Measures II or III, suggest that people who recently quit smoking cigarettes and switched to e-cigarettes may be more likely to slip over the course of a year compared to people who recently quit who were tobacco-free, but they are not clearly more likely to relapse. As this study vividly demonstrates, the likelihood of relapse is clearly a function of how that behavior is defined.

This study has a few limitations. First, PATH data are self-reported, and thus subject to potential response bias. Another limitation is not including more recent waves of PATH (waves 5 or 6). Since these recent waves are biennial follow-ups instead of annual ones, we did not include them in our analyses. More research is needed to examine if the association between other tobacco use and smoking relapse has changed in more recent years and whether the association holds with longer follow-ups.

Conclusions

The distinction between a slip and relapse needs to be considered more carefully, with researchers undertaking further studies that help us contemplate how we should define relapse. This study demonstrates a striking contrast between the strictest of relapse definitions—any smoking within a 12-month period—and more “lenient” definitions that permit an occasional slip or two. At least for some people who smoke, by satisfying nicotine cravings, the availability of reduced-risk nicotine-delivery products (such as e-cigarettes) might represent the difference between a slip leading to relapse, and a slip not impeding long-term smoking cessation.

Supplementary Material

Supplementary material is available at Nicotine and Tobacco Research online.

Funding

Drs Mendez and Warner received support from the National Cancer Institute of the National Institutes of Health (NIH) and the Food and Drug Administration’s (FDA) Center for Tobacco Products (award number U54CA229974).

Declaration of Interest

The authors have no conflicts or disclosures. The authors have not published, posted or submitted any related papers from the same study.

Author Contributions

Ruoyan Sun (Conceptualization [lead], Formal analysis [lead], Software [lead], Writing—original draft [lead]), David Mendez (Conceptualization [supporting], Funding acquisition [equal], Writing—review & editing [equal]), and Kenneth Warner (Conceptualization [supporting], Funding acquisition [equal], Writing—review & editing [equal])

Data Availability

The datasets were derived from sources in the public domain: https://www.icpsr.umich.edu/web/NAHDAP/studies/36498