Personal Characteristics Associated With Switching From Cigarettes to Noncombustible Tobacco and Nicotine Products Among U.S. Adults: Findings From PATH Study Waves 1–5

Abstract

Introduction

Reducing the disease burden from tobacco smoking may encompass switching to noncombustible (NCs), along with cessation. This study evaluates factors associated with switching to NCs (e-cigarettes, smokeless, and snus) versus continued smoking, complete cessation, or dual use.

Aims and Methods

Population Assessment of Tobacco and Health adult data were analyzed in three 2-year wave pairs across 2013–2019 to assess product transitions among people who exclusively smoked tobacco. Generalized estimating equations examined demographics, smoking characteristics, perceptions, and messaging exposure as predictors of transitions from cigarette smoking.

Results

Ages 18–44 years (ref = 65+; aORs = 3.79–5.92), males (aOR = 1.18, 95% CI [1.01, 1.37]), and perceiving NCs as less harmful than smoking (ref = not; aOR = 1.47, 95% CI [1.28, 1.69]) were positively associated with switching to NCs versus continued smoking, while being Black (ref = White; aOR = 0.36, 95% CI [0.26, 0.48]) or Hispanic (ref = non-Hispanic; aOR = 0.59, 95% CI [0.45, 0.78]) were negatively associated. Ages 18–64 years (aORs = 2.49–5.89), noticing promotional ads (ref = not; aOR = 1.34, 95% CI [1.14, 1.58]), daily light or heavy smoking (ref = nondaily; aORs = 2.33–3.72), and smoking within 30 minutes of waking (ref=>30 minutes; aOR = 1.47, 95% CI [1.18, 1.85]) were positively associated with switching to NCs versus cessation, while being Black (aOR = 0.55, 95% CI [0.38, 0.74]) or Hispanic (aOR = 0.52, 95% CI [0.38, 0.71]) were negatively associated. Incomes of $10 000–≥$100 000 (ref ≤ $10 000; aORs = 2.08–3.13) and daily heavy smoking (aOR = 1.96, 95% CI [1.11, 3.48]) were positively associated with switching to NCs versus dual use, while being male (aOR = 0.44, 95% CI [0.29, 0.65]) was negatively associated.

Conclusions

Transitioning to NCs was unequally distributed among sociodemographic groups and smoking characteristics. The mere availability of NCs may not promote health equity. Continued market and regulatory efforts that promote both smoking cessation and transitioning to NCs among various populations may promote public health gains.

Implications

This study examines factors associated with transitioning from smoking to noncombustible tobacco product use. We examined three different product use scenarios that may be associated with varied levels of exposure to harm. We observed that younger ages and perceiving NCs as less harmful than cigarettes were more likely to transition to NCs as compared to continued smoking and cessation, while Black or Hispanic individuals were less likely to transition to NCs. This highlights the need for more focused harm reduction efforts for racial and ethnic minority populations as a complementary strategy to cessation to reduce health disparities from smoking.

Introduction

Cigarette smoking continues to be the leading preventable cause of morbidity and mortality and is responsible for over 480 000 deaths in the United States annually.¹ Although complete cessation of tobacco product use would be best for reducing smoking-related burdens, relatively few smokers are able to achieve cessation. In 2017, approximately 7 in 10 smokers reported desire to quit smoking and over 50% reported at least one cessation attempt in the past year.² However, less than 10% of smokers report successful smoking cessation annually.³ Alternatively, smoking-related harm can be reduced by completely switching to nicotine- or tobacco-containing products that are less harmful as compared with cigarette smoking, such as noncombustible products (NCs).^4–7

U.S. Food and Drug Administration’s Center for Tobacco Products (FDA CTP) comprehensive plan for tobacco and nicotine regulation notes that products fall on a continuum of risk, where combustible cigarettes are most harmful followed by NC products such as e-cigarettes and smokeless tobacco, and nicotine replacement products considered being least harmful.⁸ There has been recent attention to ensuring that adults who smoke understand the relative risks of tobacco products.^9,10 While nicotine replacement products are FDA-approved for cessation, their use is not encouraged as a long-term substitute for cigarettes.¹¹ Use of products such as e-cigarettes and smokeless tobacco opens avenues for people who smoke and want to continue using nicotine but reduce the health risks associated with cigarettes. Data from the U.S. and other countries suggest that e-cigarettes may effectively serve as long-term and potentially less harmful substitutes for cigarettes and aid in smoking cessation.^12–14 For instance, completely switching to e-cigarettes can reduce users’ exposure to harmful toxicants compared to cigarette smoking.¹⁵ Similarly, studies from Sweden suggest the role of smokeless tobacco products, such as snus, for harm reduction among people who use combustible tobacco.¹⁶

Companies can apply to FDA CTP for their products to be granted modified risk tobacco product (MRTP) status, which permits evidence-based claims that a particular product poses less health risk or exposure to harmful toxicants than other products. To date, only a handful of products have received MRTP authorization, including several snus products, heated tobacco products, very-low nicotine content cigarettes, and one moist snuff product. Although NCs have the potential to reduce smoking-related harms when used as a complete substitute, dual use of cigarettes and NCs may diminish the benefits of these products.^17,18 With the availability of a variety of product options in an evolving marketplace, there is a critical need to understand the use trajectories a person who smokes may follow over time and their implications for individual and public health. One transition is continued smoking or use of other combustible products, in which case users remain exposed to the harmful constituents of tobacco combustion. By contrast, the ideal transition is complete cessation, as stopping tobacco use is the best way to limit further harmful exposure to tobacco or nicotine. A third transition is completely switching to NCs, which may reduce but not eliminate exposure to harmful constituents. A fourth transition is the dual use of combustibles and noncombustibles, which may be associated with varied degrees of exposure to nicotine and tobacco combustion byproducts.¹⁹ Despite a wide variety of ongoing research on use transitions, there are certain gaps that remain.^20–27 First, while studies often look at transition with specific alternative products (eg, e-cigarettes), noncombustibles are not often considered together as a product class.^21,22 However, it is valuable to consider noncombustibles as a larger product class to determine if regulations that promote noncombustible use among people who smoke may be appropriate for the protection of public health. Additionally, existing research focuses on consecutive wave transitions (typically separated by ~1 year), but little work has been done to assess transitions over a larger gap between baseline and follow-up.

Additionally, there is limited understanding of specific factors that are associated with transitions among people who smoke such as personal characteristics (sociodemographic characteristics, tobacco use history), beliefs (harm perceptions), and environmental influences (exposure to messaging and promotions); what is known focuses on individual products and not the broad spectrum of harm. For instance, researchers have highlighted the independent role of cigarette dependence, exposure to noncombustible product advertisements and harm perceptions of intentions to try e-cigarettes among those who smoke.^28–30 Additionally, studies have shown that Black, Hispanic, and low-income smokers may be less likely to adopt e-cigarette use and perceive e-cigarettes as less harmful than smoking,^31,32 which may have implications for health equity. A greater understanding of factors that are associated with different use transitions, using different timepoints and noncombustibles as a broad product class, can help in constructing interventions and policies that can promote tobacco harm reduction and/or cessation. Thus, the purpose of this study is to evaluate characteristics that predict the tobacco use transitions of people who smoke (ie, continued smoking, switch to NC, dual use, and cessation) using longitudinal population data from the Population Assessment of Tobacco and Health (PATH) Study.

Methods

Data Source

The PATH Study is a U.S. nationally representative, longitudinal cohort study. The data were obtained from wave 1 (October 2013–December 2014), wave 2 (October 2014—October 2015), wave 3 (October 2015—October 2016), wave 4 (December 2016—January 2018), and wave 5 (2 years later from December 2018—November 2019). The PATH Study collects self-reported information about tobacco use behaviors and attitudes as well as health effects via computer-assisted self-interviews conducted in both English and Spanish. For adult (age 18 or older) interviews, the overall weighted response rates at waves 1— 5 were 74.0%, 83.2%, 78.4%, 73.5%, and 69.4%, respectively.³³ Additional details regarding the PATH Study design, methods, and overall demographic and tobacco use distributions are published elsewhere.^34–37 PATH Study public use files (IPSCR_36498_V17)³⁸ were used for this analysis. Three wave pairs were constructed to standardize a 2-year time difference between baseline and follow-up: Waves 1–, waves 2–4, and waves 4–5. Wave 6 was not included in this study because of changes in interview protocols,³⁹ differential impacts of COVID-19 in different areas, and other federal policies (eg, Tobacco 21, enforcement on unauthorized flavored e-cigarette cartridges) that may introduce noise and complicate data interpretation.

Measures

Tobacco Use

Participants included those who used combustible products exclusively at each baseline wave of each wave pair (waves 1, 2, and 4). Those who endorsed use of any combustible tobacco products (cigarettes, cigar, cigarillo, filtered cigar, pipe, or hookah) at baseline waves and answered “no” to use of any other products were defined as exclusive combustible tobacco product users. Cigarette use was defined as having smoked 100 + lifetime cigarettes and currently smoking everyday or some days. Other combustible products and NCs were defined by ever engaging in regular use and currently using everyday or somedays. At the follow-up wave of each wave pair (waves 3, 4, and 5), participants were grouped into continued smoking, exclusive NC use, dual use of combustibles and NCs, or complete cessation. Exclusive or combined current use of cigarettes and other combustibles were defined as continued smoking. Exclusive or combined current use of e-cigarettes, smokeless tobacco, and snus were defined as exclusive NC use. Combined use of any combustibles and NCs was defined as dual use. No reported use of any tobacco or nicotine products was defined as cessation. We decided to combine e-cigarettes, SLT, and snus under single group of NCs. This is because there is existing research on these individual product transitions but limited knowledge of their combined effect on transition.^21,22 Moreover, it allows for an observation of a overall shift in the continuum of harm and switching to reduced-harm tobacco products. E-cigarettes, snus, and smokeless tobacco are all non-combustible, generally accepted as less harmful than cigarettes (though to different degrees), and some snus and SLT products have received MRTP marketing authorization, suggesting that the FDA CTP believes these products can reduce risks compared to smoking.

Predictor Variables

Predictor variables included sociodemographics, smoking characteristics (smoking intensity, nicotine dependence), NC harm perceptions, and exposure to warning labels and point-of-sale promotions. Sociodemographics included PATH-derived variables such as age, sex (male, female), race (White alone, Black alone, all others), ethnicity (non-Hispanic, Hispanic), education (less than high school, high school, some college, college graduate), and income (<$10 000, $10 000–24 999, $25,000–49 999, $50 000–99 999, and >$100 000). Smoking intensity was coded from original variables “Do you now smoke cigarettes (everyday/someday/not at all)” and “Average number of cigarettes now smoked each day.” Participants who reported daily smoking more than 10 cigarettes per day were categorized as “daily heavy smokers” and those with 10 or less were categorized as “daily light smokers,”⁴⁰ whereas those who reported smoking somedays were grouped as “non-daily” smokers. Time to first cigarette was obtained from ‘Number of minutes from waking to smoking first cigarette’ and was dichotomized into those smoking within the first 30 minutes and those smoking 30 minutes or later. Time to first cigarette was chosen as a single-item indicator of dependence, as earlier time to first cigarette is associated with biomarkers of tobacco exposure and cessation success.⁴¹ For relative NC harm perceptions, original variables for “harmfulness of using e-cigarettes compared to smoking cigarettes” and “harmfulness of using smokeless tobacco compared to smoking cigarettes” (response options, 1= less harmful, 2= About the same, 3= more harmful) were used to create a single variable. A common lower NC harm perceptions (yes/no) variable was obtained by recoding responses as “yes” if participants responded to either of the above questions as “1(less harmful),” and “no” if they responded to both as “2 (about the same) or 3 (more harmful).” Exposure to messaging was assessed through noticing pack warning labels and noticing promotions at point of sale. Frequency of noticing warning labels on cigarette packs was obtained from the original variable ‘How often have you noticed health warnings on cigarette packages in past 30 days? (1= Never to 5= very often). The response options in were recoded as “never =1, occasional= 2-3, and often = 4-5.” Noticing tobacco promotions in stores was obtained from an original derived variable “Noticed tobacco ads or promotions on store windows or inside stores where tobacco is sold in the past 6 months (yes/no).”

Data Analysis

Descriptive analyses were used to characterize transition rates from baseline exclusive combustible use to follow-up continued smoking, exclusive NC use, dual use, or cessation for the three wave pairs: Waves 1– (n = 8371), waves 2–4 (n = 7811), and waves 4–5 (n = 8235).

Three weighted Generalized Estimating Equation (GEE) models assessed the correlates of switching from exclusive cigarette smoking (baseline) to exclusive NC use (follow-up) (1) compared to continued smoking (n = 5933), (2) compared to tobacco cessation (n = 2223), and (3) compared to dual-use (n = 1120). Predictor variables (baseline) included sociodemographics (age, education, income, sex, race, and ethnicity), smoking characteristics (smoking intensity, TTFC), NC harm perception, exposure to warning labels, and exposure to tobacco promotions. A binomial distribution of outcome variables was specified with logit link and an unstructured correlation matrix. Wave 5 all-wave weights for the wave 1 cohort were used for all analyses. Analyses were conducted using Stata, version 16.0 (StataCorp LP), using the svy suite of commands and adapting SAS macro code to run weighted GEE analyses and calculate adjusted odd ratios (aORs) and 95% CIs.⁴² Variances were estimated using balanced repeated replications with Fay’s adjustment set to 0.3 to increase estimate stability. Statistical significance was determined at p < .05. Missing data for all variables was coded as a separate sub-group named “missing” and used in total sample for frequency calculation. However, for the variables where “missing” sub-group was comparable to other groups in size, it was included in the analysis as its own sub-group (education, income, race, and ethnicity), otherwise, it was excluded from the analysis. Depending on the analysis model, there were missing data for sociodemographic characteristics such as education (0.43%–0.52%), income (6.06%–6.10%), race (0.87%–2.42%), and ethnicity (0.99%–1.48%). Missing data for other variables included noticing health warnings (0.16%–0.24%), noticing promotions at the point of sale (0.17%–0.29%), harm perceptions (3.31%–6.76%), and smoking intensity (7.08%–17.68%). Because smoking intensity was based on cigarette smoking, any individuals who exclusively used non-cigarette combustibles (ie, cigar, cigarillo, filtered cigar, pipe, and hookah) were excluded from the GEE models. Data with large missingness may have been due to skip patterns or restricting data for models to exclusive cigarette smokers.

Results

Of the 8371 people who exclusively smoked at baseline wave 1, 6187 (73.91%) reported continued smoking, 1026 (12.26%) reported exclusive noncombustible use, 227 (2.71%) reported dual use, and 931 (11.12%) reported cessation at follow-up wave 3. Of those who transitioned to exclusive NC use at wave 3, 87.44% used e-cigarettes, 11.88% used SLT, and 3.10% used snus (with 2.41% using multiple products). Similarly, of the 7811 people who exclusively smoked at baseline wave 2, 5934 (75.97%) reported continued smoking, 913 (11.69%) reported exclusive NC use, 178 (2.28%) reported dual use, and 786 (10.06%) reported cessation at follow-up wave 4. Of those who transitioned to exclusive NC use at wave 4, 82.05% used e-cigarettes, 17.48% used SLT, and 4.20% used snus (with 3.73% using multiple products). Finally, from the 8235 people who exclusively smoked at baseline wave 4, 5921 (71.90%) reported continued smoking, 1207 (14.66%) reported exclusive NC use, 245 (2.98%) reported dual use, and 862 (10.47%) reported cessation at wave 5 (Table 1). Of those who transitioned to exclusive NC use at wave 5, 91.94% used e-cigarettes, 8.06% used SLT, and 3.10% used snus (with 3.10% using multiple products). Table S1 shows descriptive statistics for all predictors by each transition category.

Table 1.

Distribution of Participants at Baseline and at Follow up Waves by Tobacco Product Use

Baseline	Follow up	n	%
W1 combustible use (n = 8371)	W3 continued smoking	6187	73.91
	W3 NC use	1026	12.26
	W3 dual use	227	2.71
	W3 cessation	931	11.12
W2 combustible use (n = 7811)	W4 continued smoking	5934	75.97
	W4 NC use	913	11.69
	W4 dual use	178	2.28
	W4 cessation	786	10.06
W4 combustible use (n = 8235)	W5 continued smoking	5921	71.90
	W5 NC use	1207	14.66
	W5 dual use	245	2.98
	W5 cessation	862	10.47

Model 1 (Table 2) assessed the likelihood of switching to NCs compared to continued smoking at follow-up among those reporting exclusive smoking at baseline waves. Ages 18–24 years (aOR = 5.92, 95% CI [3.55, 9.87]); 25–44 years (aOR = 3.79, 95% CI [2.31, 6.21]) as compared to 65 years and above, some college education (aOR = 1.46, 95% CI [1.15, 1.87]) compared to less than high school, income of ≥$100 000 (aOR = 1.59, 95% CI[1.17, 2.15]) compared to <$10 000, males (aOR = 1.18, 95% CI [1.01, 1.37]) compared to females, and those who perceived NCs as less harmful (aOR = 1.47, 95% CI [1.28, 1.69]) as compared to not, showed a significant positive association with switching to NCs compared to continued smoking. Participants who are Black (aOR = 0.36, 95% CI [0.26, 0.48]) compared to White, and who are Hispanic (aOR = 0.59, 95% CI [0.45, 0.78]) compared to non-Hispanic, reported a reduced likelihood of switching to NCs compared to smoking. Additionally, participants at baseline wave 4 had an increased likelihood of switching to NCs compared to continued smoking at follow-up (aOR = 1.28, 95% CI [1.08, 1.52]) compared to participants at baseline wave 1.

Table 2.

Predictors for Transition to Noncombustibles Versus Continued Smoking

Switching to noncombustibles vs. continued smoking		aOR	95% CI		p-value
			LL	UL
Wave	Wave 1	Ref
	Wave 2	0.96	0.84	1.10	.59
	Wave 4	1.28	1.08	1.52	.005
Age (years)	18–24	5.92	3.55	9.87	.001
	25–44	3.79	2.31	6.21	<.001
	45–64	1.61	0.98	2.67	.06
	65 and above	Ref	—	—	—
Education	Less than high school	Ref	—	—	—
	High school	1.18	0.93	1.50	.17
	Some college	1.46	1.15	1.87	.002
	College grad	1.17	0.84	1.64	.35
	Missing	1.35	0.50	3.63	.59
Income (U.S. dollars)	<$10 000	Ref	—	—	—
	$10 000–24 999	1.09	0.89	1.34	.86
	$25 000–49 999	1.07	0.86	1.34	.64
	$50 000–99 999	1.13	0.89	1.44	.30
	≥$100 000	1.59	1.17	2.15	.003
	Missing	1.01	0.73	1.40	.96
Race	White alone	Ref	—	—	—
	Black alone	0.36	0.26	0.48	<.001
	All others	0.87	0.65	1.16	.34
	Missing	0.26	0.09	0.79	.017
Ethnicity	Non-Hispanic	Ref	—	—	—
	Hispanic	0.59	0.45	0.78	<.001
	Missing	0.71	0.28	1.81	.47
Sex	Male	1.18	1.01	1.37	.037
	Female	Ref	—	—	—
Notices health warning	Often	0.90	0.76	1.06	.21
	Occasional	1.13	0.94	1.36	.19
	Never	Ref	—	—	—
Notices promotional ads	Yes	1.11	0.97	1.27	.21
	No	Ref	—	—	—
Lower NC harm perceptions	Yes	1.47	1.28	1.69	<.001
	No	Ref	—	—	—
Time to first cigarette (min)	Less than 30 min	0.97	0.82	1.13	.67
	30 min or more	Ref	—	—	—
Smoking intensity	Daily heavy smoker	1.11	0.89	1.38	.35
	Daily light smoker	1.09	0.89	1.33	.41
	Non-daily smoker	Ref	—	—	—

Model 2 (Table 3) assessed the likelihood of switching to NCs as compared with smoking cessation at follow-up among those reporting exclusive smoking at baseline waves. Ages 18–4 (aOR = 5.89, 95% CI [3.44, 10.08]); 25–44 (aOR = 4.61, 95%CI [2.77, 7.68]); 45–64 (aOR = 2.49, 95% CI [1.48, 4.18]) compared to 65 years and above, noticing tobacco advertisements (ref = not; aOR = 1.34, 95% CI [1.14, 1.58]), occasionally noticing health warnings (ref = never; aOR = 1.35, 95% CI [1.09, 1.67]), lower NC harm perceptions (aOR = 1.39, 95% CI [1.17, 1.65]) compared to not, daily heavy smoking (aOR = 3.72, 95% CI[2.90, 4.78]) and daily light smoking; (aOR = 2.33, 95% CI[1.85, 2.92]) compared to non-daily smoking and smoking within first 30 minutes of waking up (aOR = 1.47, 95% CI [1.18, 1.85]) compared to smoking at 30 minutes or later, were positively associated with switching to NCs compared to smoking cessation. Being a college graduate (aOR = 0.61, 95% CI [0.41, 0.91]) compared to less than high school was negatively associated with switching to NCs versus cessation. Participants who are Black (aOR = 0.53, 95% CI [0.38, 0.74]) compared to White, and Hispanic (aOR = 0.52, 95% CI [0.38, 0.71]) compared to non-Hispanic, reported a reduced likelihood of switching to NCs compared to cessation. Additionally, participants at baseline wave 4 had an increased likelihood of switching to NCs compared to cessation at follow-up (aOR = 1.37, 95% CI [1.11, 1.69]) compared to participants at baseline wave 1.

Table 3.

Predictors for Transition to Noncombustibles Versus Smoking Cessation

Switching to noncombustible vs. cessation		aOR	95% CI		p-value
			LL	UL
Wave	Wave 1	Ref
	Wave 2	1.07	0.91	1.26	.39
	Wave 4	1.37	1.11	1.69	.003
Age (years)	18–24	5.89	3.44	10.08	<.001
	25–44	4.61	2.77	7.68	<.001
	45–64	2.49	1.48	4.18	.001
	65 and above	Ref	—	—	—
Education	Less than high school	Ref	—	—	—
	High school	1.12	0.83	1.51	.45
	Some college	1.09	0.81	1.47	.58
	College grad	0.61	0.41	0.91	.015
	Missing	0.91	0.13	6.37	.92
Income (U.S. dollars)	<$10 000	Ref	—	—	—
	$10 000–24 999	1.09	0.84	1.42	.51
	$25 000–49 999	0.96	0.73	1.27	.79
	$50 000–99 999	0.91	0.67	1.25	.57
	≥$100 000	1.10	0.77	1.56	.61
	Missing	0.63	0.36	1.09	.096
Race	White alone	Ref	—	—	—
	Black alone	0.53	0.38	0.74	<.001
	All others	0.98	0.70	0.139	.92
	Missing	0.22	0.07	0.66	.007
Ethnicity	Non-Hispanic	Ref	—	—	—
	Hispanic	0.52	0.38	0.71	<.001
	Missing	0.59	0.16	2.05	.40
Sex	Male	1.14	0.94	1.39	.18
	Female	Ref	—	—	—
Notices health warning	Often	1.17	0.98	1.41	.087
	Occasional	1.35	1.09	1.67	.007
	Never	Ref	—	—	—
Notices promotional ads	Yes	1.34	1.14	1.58	<.001
	No	Ref	—	—	—
Lower NC harm perceptions	Yes	1.39	1.17	1.65	<.001
	No	Ref	—	—	—
Time to first cigarette (min)	Less than 30 mins	1.47	1.18	1.85	.001
	30 min or more	Ref	—	—	—
Smoking intensity	Daily heavy smoker	3.72	2.90	4.78	<.001
	Daily light smoker	2.33	1.85	2.92	<.001
	Non-daily smoker	Ref	—	—	—

Model 3 (Table 4) assessed the likelihood of switching to NCs as compared with dual use at follow-up among those reporting exclusive smoking at baseline. Incomes of $10 000–24 999 (aOR = 2.08, 95% CI [1.30, 3.33]), $25 000–49 999 (aOR = 2.45, 95% CI [1.41, 4.25]), $50 000–99 999 (aOR = 2.80, 95% CI [1.57, 4.99]), and >$100 000 (aOR = 3.13, 95% CI [1.46, 6.75]) compared to <$10 000, and daily heavy smoking (aOR = 1.96, 95% CI [1.11, 3.48]) compared to non-daily smoking, were significantly positively associated with switching to NCs compared to dual use. Being male (aOR = 0.44, 95% CI [0.29, 0.65]) compared to female was associated with reduced odds of transitioning to NCs compared to dual use.

Table 4.

Predictors for Transition to Noncombustibles Versus Dual Use

Switching to noncombustible vs. dual use		aOR	95% CI		p-value
			LL	UL
Wave	Wave 1	Ref
	Wave 2	1.20	0.77	1.83	.40
	Wave 4	1.01	0.65	1.57	.98
Age (years)	18–24	0.85	0.28	2.57	.78
	25–44	1.10	0.37	3.27	.86
	45–64	0.92	0.29	2.9	.89
	65 and above	Ref	—	—	—
Education	Less than high school	Ref	—	—	—
	High school	1.11	0.65	1.88	.71
	Some college	1.20	0.71	2.01	.50
	College grad	0.96	0.46	2.01	.92
	Missing	n/a
Income (U.S. dollars)	<$10 000	Ref	—	—	—
	$10 000–24 999	2.08	1.30	3.33	.002
	$25 000–49 999	2.45	1.41	4.25	.001
	$50 000–99 999	2.80	1.57	4.99	.001
	≥$100 000	3.13	1.46	6.75	.003
	Missing	1.24	0.56	2.75	.59
Race	White alone	Ref	—	—	—
	Black alone	0.60	0.32	1.09	.095
	All others	0.77	0.43	1.38	.38
	Missing	2.68	0.32	22.48	.36
Ethnicity	Non-Hispanic	Ref	—	—	—
	Hispanic	0.58	0.33	1.02	.06
	Missing	1.12	0.24	5.16	.89
Sex	Male	0.44	0.29	0.65	<.001
	Female	Ref	—	—	—
Notices health warning	Often	0.76	0.49	1.19	.23
	Occasional	0.69	0.42	1.11	.13
	Never	Ref	—	—	—
Notices promotional ads	Yes	1.01	0.71	1.43	.97
	No	Ref	—	—	—
Lower NC harm perceptions	Yes	0.84	0.58	1.23	.37
	No	Ref	—	—	—
Time to first cigarette (min)	Less than 30 min	0.67	0.45	1.00	.052
	30 min or more	Ref	—	—	—
Smoking intensity	Daily heavy smoker	1.96	1.11	3.48	.021
	Daily light smoker	1.45	0.85	2.46	.17
	Non-daily smoker	Ref	—	—	—

Discussion

The results of this study identify factors that are associated with switching to NC compared to other tobacco use trajectories among adults who smoked combustible tobacco. This study included a novel approach of combining multiple products into classes of combustibles and NCs. Existing research has often focused on individual product transitions, though there is limited knowledge of their combined effects to make broad observations in the shift of the tobacco continuum of harm and understand who is switching to reduced-harm products. Additionally, most existing research focuses on transitions between consecutive wave pairs (eg., waves 1 to 2), while this study assesses longer-term transition by separating baseline and follow-up by two waves (eg, waves 1 to 3). Third, we used GEE models to stack baseline to follow-up wave pairs, allowing us to combine multiple baseline to follow-up timepoints into a single analysis to understand person-based characteristics that predict transitions between product classes. Across all wave pairs, continued smoking was the most common transition and observed in over 70% of the sample. Transitioning from exclusive smoking to NCs, compared to continued smoking, was more likely from waves 4 to 5 than waves 1 to 3, which may indicate increased adoption of potentially less harmful alternatives in more recent years. A large majority of participants switched to e-cigarette use, while switching to SLT, snus, or multiple NCs was less common. Similarly, transitioning from exclusive smoking to cessation, compared to continued smoking, was more likely from waves 4 to 5 than waves 1 to 3. By contrast, transitioning from exclusive smoking to dual use, compared to continued smoking, did not differ by wave, indicating that dual-use trajectories have not become more common in recent years. Together, these transitions suggest that changes in the tobacco product landscape and tobacco regulation may lead to relatively more transitions in recent years to exclusive NC use or smoking cessation, compared to continued smoking, in recent years, which may represent a positive public health impact.

Model 1, which assessed predictors of those who switch to NCs versus continued smoking, represents potential engagement in tobacco harm reduction strategies. We observed that people who exclusively smoked at baseline and switched completely to NCs, compared to those who continued smoking, were younger, male, had relatively higher socioeconomic status, and a lower harm perception for NCs; while individuals who are Black or Hispanic had reduced odds of switching completely to NCs versus continued smoking These findings are consistent with research findings that individuals who are Black, Hispanic, or lower socioeconomic status are less likely to perceive e-cigarettes as less harmful than smoking and less likely to transition to their use.^31,32 Notably, many of these associations demonstrate that the availability of tobacco harm reduction products has not furthered health equity for groups that experience tobacco-related disparities, such as people from lower socioeconomic status, older age, and minority racial and ethnic populations.^43,44 Future work may benefit from further assessing reasons for the lack of engagement with tobacco harm reduction in these populations, which may help develop and test targeted educational and interventional efforts for populations that experience tobacco-related health disparities to promote transitioning to less harmful forms of tobacco for those unable or uninterested in quitting smoking. Similarly, future work may look at mediation or moderation models to further elucidate the interaction of these variables on tobacco use transitions.

Model 2, which assessed predictors of those who switched to NCs versus smoking cessation, is a more nuanced situation. For instance, the FDA must consider how possible MRTPs influence the increased or decreased likelihood of cessation among people who smoke.⁴⁵ That is, there may not be a population benefit if individuals choose not to quit because NC products are available, but there may be a population benefit if individuals who would not quit choose to switch to NC products to reduce tobacco-related harms. Some consistency with Model 1 was observed, in which switching to NCs was associated with younger ages and people who perceive NCs as less harmful than smoking; while individuals who are Black or Hispanic had lower odds of transitioning to NCs versus cessation. However, differences emerged between models regarding associations with college graduates, noticing ads/promotions at point of sale, and heavier smoking and dependence. The positive association between smoking intensity and switching to NCs (vs. cessation) and negative association with time to first cigarette suggest that those who are more heavily dependent may be more likely to adopt tobacco harm reduction than achieve smoking cessation. This result is consistent with previous research using national data finding that daily e-cigarette use was associated with greater likelihood of smoking cessation among individuals who had no plans to quit.^46,47 Regarding promotions at point of sale, it is possible that promotions that reduce price or highlight attractive features may affect someone who smoke’s choice to switch to an NC or quit. However, an important limitation is the unknown features of these promotions and what products were promoted.

Model 3, which assessed predictors of those who switched to NCs versus dual use, also likely represents a positive outcome of relatively reducing tobacco-related harms as dual use involves continued smoking and may increase nicotine dependence.⁴⁸ Significant associations observed in this model were positive associations with increasing income and daily heavy smoking, and negative associations with male participants. It is not fully clear why a consistent effect of income was found, though it is a social determinant of health and possible effects may be related to the association between income, and health literacy.⁴⁹ These results suggest that people with low income, males, and lighter smokers may be the most fruitful targets for educational interventions that encourage exclusive NC use versus dual use. It remains important to assess whether dual use continues long-term or is a transitional phase between smoking and exclusive NC use, as well as its implications for exposure to nicotine and potentially harmful constituents in tobacco.

Across three models we observed a greater association of younger age groups with transitioning to NCs compared to continued smoking and cessation, but not dual use. A possible explanation for this can be a greater flexibility for adapting to newer products (eg, e-cigarettes) in relatively younger populations as compared to older established cigarette smokers who may continue smoking or prefer cessation.^33,50 Being Black and Hispanic were negatively associated with switching to NCs versus continued smoking and cessation, but not dual use. This finding may be due to a lack of product availability or awareness about the harm reduction potential of noncombustibles among those communities. It has been well documented that tobacco companies differentially market their cigarettes and combustible products to minority groups.⁵¹ Additionally, rates of smoking decline are lower among non-Hispanic Blacks when compared to their counterparts.⁵² While these trends fall in line with several previous reports, they also highlight the need for more focused efforts in racial and ethnic minority groups about harm reduction.^53–55 Efforts to reduce targeted combustible product marketing and greater accessibility to accurate information about the use of NCs for harm reduction should be promoted, particularly considering the FDA Center for Tobacco Products recent emphasis on communicating the relative risks of tobacco products.⁹ Additionally, reducing the financial barriers and disparities based on accessibility to healthcare (lack of insurance) may be a source of useful and reliable information about the relative risks of products.¹⁰ Males were more likely to use NCs than continue smoking, but less likely to engage in exclusive NC than dual use. Thus, the first may promote harm reduction but continued efforts are needed to encourage males switching exclusively to NCs rather than dual use. We observed mixed results with education and income, but generally higher education and income were associated with greater switching to NC compared with smoking and dual-use, and not with cessation which possibly demonstrates uptake of tobacco harm reduction among individuals of higher socioeconomic status. Those who perceived NCs as less harmful were generally more likely to switch to NCs (except for dual use), which resonates well with previous reports associating reduced harm perception with increased use.^30,56–58 Dual use may also possibly represent a transient phase for those attempting to switch or achieve complete cessation. Lastly, we did not observe any association between noticing health warning labels on cigarettes with switching to NCs, suggesting that current health warning labels may not be sufficient to change the behaviors of people who currently smoke. In the current regulatory environment, as there are potential policy goals to restrict Menthol cigarettes or shift towards low nicotine-containing cigarettes, our results highlight the need for increased outreach to promote switching to reduced-harm alternatives among those who are unable or unwilling to quit nicotine completely. However, little is known about how the use transitions will evolve after the implementation of new policies that affect the tobacco marketplace, highlighting the need for research to investigate the intended and unintended consequences of these policies.

Several limitations of this study help to contextualize results. First, the limited assessment of demographic factors such as sex and race limits our ability to evaluate nuanced differences in minority populations that may experience tobacco-related health disparities. Similarly, we did not evaluate gender identity and sexual preferences, though sexual and gender minorities experience tobacco-related disparities.⁵⁹ Secondly, we focused on broad product classes (eg, combustible vs. NC) which contain a variety of products with different user bases, which may mask the effects of the products used less commonly, like SLT and snus. However, we also believe this may provide valuable data and future research may continue to assess both broad product classes (combustible vs. NC) as well as more granular product data. We were unable to include oral nicotine products (eg, pouches, non-medicinal nicotine gum) in our class of NCs because they were not measured in PATH waves 1–5. These products are rising in popularity and warrant similar investigations regarding combustible users who switch to oral nicotine products. Importantly, this study focused on tobacco use behavior as an outcome but did not assess the health impacts of these behaviors. Future research should further assess changes in biomarkers of exposure and harm, as well as meaningful health indicators, in these different transition groups to assess which trajectories result in clinically relevant reductions in harm. Additionally, we did not assess longer-term transitions and thus, do not have insight into the stability of these trajectories. It is essential to assess whether those who switch to NCs continue to use them, relapse back to smoking, or end up quitting tobacco permanently (particularly compared to those who achieved cessation in follow-up waves). Similarly, it would be useful to assess the long-term trajectories for those who engage in dual use to determine if this is a step on the way to switching/cessation or if it remains a long-term behavior. Other than long-term behaviors, there is also a valuable research direction in assessing the interaction of time with these predictors to better understand how these associations change as a function of time, particularly with more recent PATH waves in a changing world (eg, COVID-19) and regulatory environment (eg, Tobacco 21 laws). Finally, it would be useful to replicate these analyses using the youth sample to determine how findings may be similar or different for vulnerable, younger populations.

The paper highlights important individual characteristics that are associated with engagement in tobacco harm reduction among adults who smoked cigarettes. A better understanding of these factors can help improve population health by determining which populations may benefit from increased attention to tobacco harm reduction options, as well as populations that may benefit from more targeted cessation efforts. Notably, the availability of NCs may contribute to preexisting health inequities, as populations that experience tobacco-related disparities (eg, low socioeconomic status, Black and Hispanic populations) were generally less likely to transition to NCs. Continued research is needed in both tobacco harm reduction and cessation efforts to best facilitate the reduction of smoking-related harms on a population level.

Supplementary material

Supplementary material is available at Nicotine and Tobacco Research online.

Funding

Research reported in this publication was supported by NCI and FDA Center for Tobacco Products R21CA268198 (PI: NJF) and AS’s contribution was partly supported by R01DA054993. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH or the Food and Drug Administration,

Declaration of Interest

None.

Data Availability

PATH Study Data are publicly available at https://doi.org/10.3886/ICPSR36498.v19

Author Contributions

Akshika Sharma (Formal analysis [lead], Investigation [equal], Methodology [equal], Writing – original draft [lead], Writing – review & editing [equal]), Karin Kasza (Conceptualization [supporting], Data curation [equal], Formal analysis [equal], Funding acquisition [supporting], Investigation [equal], Methodology [equal], Validation [equal], Writing – review & editing [equal]), Richard O'Connor (Conceptualization [supporting], Data curation [equal], Funding acquisition [supporting], Investigation [equal], Methodology [supporting], Writing – review & editing [equal]), and Nicholas Felicione (Conceptualization [lead], Data curation [equal], Formal analysis [equal], Funding acquisition [lead], Investigation [equal], Methodology [lead], Project administration [lead], Supervision [equal], Writing – original draft [equal], Writing – review & editing [equal])