The Impact of Nicotine Reduction on Cigarette Reinforcement Value Within a Marketplace Containing Alternative Nicotine Delivery Products: A Secondary Analysis of a Multi-Site Trial

Abstract

Introduction:

A mandated reduction in the nicotine content of cigarettes would likely improve public health. Prior research has shown that assignment to very low nicotine content (VLNC) cigarettes reduces the reinforcement value of cigarettes. However, no studies have evaluated changes in reinforcement for smoking after experience with VLNC cigarettes in a marketplace where noncombusted nicotine products are available.

Aims and Methods:

Adult participants who smoke cigarettes (n = 438) were randomized 1:1 to VLNC or normal nicotine content cigarettes for 12 weeks. Participants purchased products using points valued at $1 from an experimental tobacco marketplace that contained study cigarettes and noncombusted nicotine products. At the 12-week visit, participants completed hypothetical purchase tasks for each product. Demand parameters were calculated, including intensity (consumption when product is free), breakpoint (price when consumption is reduced to zero), P_max (price that produces maximum daily spending), O_max (maximum amount spent in a single day), and α (rate of change in consumption across the demand curve).

Results:

Assignment to VLNC cigarettes reduced demand for study cigarettes across all parameters, and reduced demand for usual brand cigarettes for intensity and breakpoint, but not other parameters. Assignment to VLNC cigarettes increased O_max for e-cigarettes, but not other demand parameters. There were no significant differences in demand parameters for other products.

Conclusions:

A mandated reduction in cigarette nicotine content is likely to reduce the reinforcement value of cigarettes in a marketplace where noncombusted products are available, which may drive reductions in the prevalence of smoking and shift people who smoke toward noncombusted products.

Implications:

This is the first study to show that assignment to VLNC cigarettes along with access to noncombusted nicotine products reduces reinforcement value for both low nicotine and normal nicotine cigarettes, and may increase the reinforcement value of e-cigarettes. These results suggest that a mandated reduction in nicotine will decrease the prevalence of smoking and may increase the use of other noncombusted products.

Introduction

A mandated reduction in the nicotine content of cigarettes would likely improve public health by reducing the prevalence of smoking.¹ Clinical trials in which participants who smoke are provided with very low nicotine content (VLNC) cigarettes have demonstrated that assignment to VLNC cigarettes reduces the number of cigarettes smoked per day, reduces nicotine dependence, and increases the ability to quit smoking.^2–6 In the United States, the Food and Drug Administration published a Proposed Rule in 2025 for a mandated reduction in the nicotine content of cigarettes to a minimally addictive or nonaddictive level.⁷ In New Zealand, a mandated reduction in cigarette nicotine content was passed through legislation in 2022 and later repealed by a new coalition government.⁸

Nicotine is the primary reinforcing constituent in cigarettes,⁹ and a mandated reduction in the nicotine content of cigarettes may reduce smoking behavior by reducing their reinforcement value.^10,11 One tool for assessing reinforcement value is the cigarette purchase task (CPT).^12,13 The CPT asks people who smoke to report the number of cigarettes they would purchase across a range of hypothetical prices. Responses on the cigarette purchase task can be plotted on a curve (ie, demand curve) which can be used to calculate parameters related to reinforcement value. A modified version of the cigarette purchase task has also been used to assess the reinforcement value of non-cigarette tobacco products (ie, product purchase task).¹⁴ Parameters calculated from the CPT or product purchase task include: intensity, or the maximum consumption reported when the product is free; P_max, or the price that produces maximum daily spending; O_max, or the maximum amount of money that would be spent in a single day; breakpoint, or the price at which consumption is reduced to 0; and α, or the rate of change in consumption across the demand curve.¹¹

Prior research has demonstrated that nicotine reduction can reduce the reinforcement value of VLNC cigarettes,^11,15 and that reinforcement value may also be reduced for normal nicotine content (NNC) usual brand cigarettes following experience with VLNC cigarettes as part of randomized controlled trials.^4,11 Indeed, participants with exposure to VLNC cigarettes in clinical trials smoke fewer cigarettes per day even after returning to usual brand cigarettes.¹⁶ Reduced demand for usual brand suggests that extended experience smoking VLNC cigarettes reduces cigarette dependence, reducing demand for usual brand by the end of the trial. However, these trials have generally asked people who smoke not to use any nicotine or tobacco products other than their assigned study cigarettes. In the United States, non-cigarette tobacco products have been increasing in prevalence over the last decade,¹⁷ and evidence suggests that use of these products may increase following the implementation of a nicotine reduction policy.^5,18 Thus, it is important to understand the change in reinforcement value of cigarettes and non-cigarette tobacco products following a nicotine reduction policy in the context of continued access to other sources of nicotine. The present article reports results from a secondary analysis of a multi-site trial in which people who smoke were assigned to either VLNC or NNC cigarettes in a marketplace where non-cigarette, noncombusted nicotine products were also available. The primary paper from this trial has already been published, and the present secondary analysis reports results from hypothetical purchase tasks completed by participants for assigned study cigarettes, their usual brand of study cigarettes, and five noncombusted nicotine products.⁵

Methods

Participants

Participants (n = 438) were recruited primarily through social media, internet, and flyers, at six sites: University of Minnesota, University of California San Francisco, Duke University, Brown University, University of Pennsylvania, and Wake Forest University. Eligibility criteria included: ≥ the legal age to purchase tobacco (which increased from 18 to 21 during the study at all sites), 5–40 cigarettes smoked per day, and an expired carbon monoxide (CO) level of at least 10 ppm or a urinary cotinine level indicative of nicotine use on a test strip. Participants were ineligible if they exclusively used roll-your-own cigarettes or were unable to identify a preferred manufactured cigarette brand during baseline, had exposure to investigational cigarettes within the last 2 years, reported a serious or unstable psychiatric disorder or medical disease, reported excessive alcohol drinking or problems with alcohol or drug use, or were pregnant, breastfeeding, or planning to become pregnant.

Study Design

For a detailed description of the study methods, see the primary manuscript from the clinical trial, which includes the primary outcomes of the clinical trial including changes in tobacco product use and biomarkers during the trial.⁵ Briefly, participants were randomized 1:1 to either VLNC or NNC study cigarettes, stratified by site. Participants in both groups were provided with study cigarettes that were either menthol or non-menthol based on their preference, except in San Francisco where a ban on menthol cigarettes was imposed.

Participants in the trial purchased products from an experimental tobacco marketplace (ETM) and were told to use exclusively the products they purchased from the ETM for the duration of the trial. In brief, participants were given points to use to purchase products. The number of points was based on their baseline tobacco use, reflecting typical expenditure, with an additional point provided each day to allow for product experimentation. All products were discounted to 66% of local retail prices to discourage participants from purchasing products outside of the ETM. Unused points were banked and converted to cash at the end of the study. The marketplace included cigarettes, e-cigarettes, oral tobacco products like snuff, snus, and nicotine pouches, and nicotine replacement therapies. Brands and flavors were representative of those with the greatest market share or purchases based on a survey of retailers at each site. Noncombusted products available in the ETM were updated yearly to reflect changes in the real-world marketplace. A variety of e-cigarette types and flavors were provided.

The protocol was divided into three phases. In phase 1, participants smoked their usual brand cigarettes (2 weeks). In phase 2, participants purchased products from the ETM, and the cigarettes available in the marketplace were participants’ usual brand cigarettes (2 weeks). In phase 3, the cigarettes available in the marketplace were participants’ assigned study cigarettes (VLNC or NNC, 12 weeks). Participants attended weekly clinic visits for the first 4 weeks of phase 3, and then biweekly visits for the next 8 weeks. During the COVID-19 pandemic, the third weekly visit was removed and all in-person clinic visits were converted to virtual-plus-curbside visits. A final follow-up visit occurred at 16 weeks post randomization (not included in the present article).

The present article focuses on the data collected from purchase tasks within the trial. In phase 2 (baseline, prior to randomization) and week 12, participants completed hypothetical purchase tasks for different tobacco products. At baseline, participants completed the CPT asking about usual brand cigarettes, and at week 12 participants completed the CPT for both usual brand and study cigarettes. At baseline and week 12, participants also completed five purchase tasks for different noncombusted nicotine products: e-cigarettes, smokeless tobacco or snus, nicotine replacement therapy lozenges, nicotine replacement therapy gum, and nicotine pouches. For each purchase task, participants were instructed to think about how they were feeling at that moment, while assuming that: (1) they had the same income and savings they had at that time and no access to any cigarettes or nicotine products other than the product being offered in that purchase task, (2) they could use the product without restrictions and without factoring in what might happen in the next 24 hours related to their participation in the study, and (3) they could not save or stockpile products. Each purchase task asked how many units (cigarettes, e-cigarettes, dips, pouches, or lozenges) they would purchase for use in 24 hours if they were free, 2¢, 5¢, 10¢, 20¢, 30¢, 40¢, 50¢, 60¢, 70¢, 80¢, 90¢, $1.00, $2.00, $3.00, $4.00, and $5.00. The e-cigarette version instructed participants to assume that each e-cigarette was about 10 puffs.

Participants were compensated for participation in the study and received compensation for unspent points in the marketplace. The protocol was approved by the site Institutional Review Boards and written informed consent was obtained from participants prior to study participation.

Statistical Analyses

We focused our analyses on three research questions. (1) Did assignment to VLNC cigarettes reduce demand for VLNC cigarettes or usual brand cigarettes? (2) Did assignment to VLNC cigarettes change demand for noncombusted tobacco products? and (3) Did assignment to VLNC cigarettes increase the proportion of people who self-reported that they would not purchase study cigarettes or usual brand cigarettes at any price above $0.00?

We assessed the impact of assignment to VLNC cigarettes on demand parameters for study cigarettes, usual brand cigarettes, and all noncombusted nicotine products using the following demand parameters: intensity (Q₀), O_max, P_max, breakpoint, and α, as previously described.¹¹ All parameters except α could be obtained empirically from the data. To estimate α, a nonlinear least-squares (NLS) regression was fitted for each participant:

$$𝒬={𝒬}_0\times {10}^{k\left(\exp \left(-\alpha {𝒬}_{0}C\right)-1\right)}$$

except when a participant’s data did not follow the assumed exponential curve; specifically, when (1) the number of reported products used was the same across all prices, or (2) participants reported they would not use any products at any price above $0.00. Excluding participants in category (2) excludes those who were most impacted by a reduction in nicotine content. Thus, to include as many participants as possible, results focus on empirical parameters, other than α. The parameter k was determined by subtracting the log10-transformed average consumption at the highest price ($5 per unit) from that at the lowest price ($0) used in curve-fitting. Descriptive statistics that are robust to extreme values (median and interquartile range (IQR)) are presented for each demand parameter and these parameters were compared between two treatment groups by using the nonparametric Wilcoxon rank-sum test.

We also estimated the treatment effect on α using a nonlinear mixed model similar to those proposed by Kaplan et al.,¹⁹ but with two additional fixed effects, the treatment effects on log₁₀ Q₀ and log₁₀α. The nonlinear mixed model allows the utilization of all participants’ data, including participants who reported zero consumption across all prices. The nonlinear mixed model generally found the same pattern of results as the nonlinear least-squares model described above. However, the default model (with the most flexible, positive definitive variance–covariance structure for the two random effects for log₁₀Q₀ and log₁₀α) only converged for some of the products (possibly because of the larger number of parameters in our models than those in Kaplan et al.¹⁹), and when the default model did not converge, more restrictive (ie, fewer parameters) models were utilized, for example, the diagonal or identity variance–covariance structure or single random-effect models. Thus, for the purposes of the present article, we relied primarily on the nonlinear least-squares regression described above as our primary model.

To understand whether assignment to VLNC cigarettes increased self-reported non-smoking at any price above $0.00 at week 12, a chi-square test with continuity correction was used to compare the proportion of participants who reported zero consumption of product at all prices above $0.00 between the VLNC and NNC groups at week 12, accompanied with an estimated odds ratio (OR) and 95% confidence interval (CI).

For all analyses, we excluded participants who did not complete the week 12 purchase tasks. Multiple imputation-based logistic regression and median regression, which included participants who did not complete week 12, were performed as a sensitivity analysis for binary and continuous outcomes, respectively. Furthermore, the Wilcoxon rank-sum test was used to compare baseline purchase task demand parameters (week 0) between those who completed week 12 and those who did not. All analyses were performed using SAS 9.4 (SAS Institute Inc., Cary, NC).

Results

Of the 438 randomized participants (220 VLNC, 218 NNC), 166 people in the VLNC group and 173 people in the NNC group completed the week 12 purchase task assessments. Baseline purchase task demand parameters were not significantly different for any product between those who completed week 12 and those who did not (all ps > .05).

The Impact of Nicotine Reduction on Demand for Study Cigarettes and Usual Brand Cigarettes

Demand curves plotting the median responses for VLNC and NNC groups across prices are shown for both study cigarettes and usual brand cigarettes in Figure 1. Median demand parameters obtained empirically (intensity, breakpoint, P_max, and O_max) and from model-based estimates (α) are shown in Table 1 for both study cigarettes and usual brand cigarettes. Empirical parameters include all participants who completed the week 12 assessments, while α estimates are only calculated for participants whose demand curves follow an exponential curve, as described above. For study cigarettes, α is reported for 72 and 135 participants in the VLNC and NNC groups, respectively, and for usual brand cigarettes, α is reported for 118 and 148 participants in the VLNC and NNC groups, respectively. Assignment to VLNC cigarettes, compared with assignment to NNC cigarettes, was associated with reduced demand for study cigarettes across all parameters, including intensity, breakpoint, P_max, O_max, and α (higher α = lower demand). Assignment to VLNC cigarettes also reduced demand for usual brand cigarettes across most parameters, including intensity, breakpoint, and P_max, but not O_max or α.

Figure 1.

Box plots showing the range, median value (line within the box), and interquartile range (box) for purchasing for VLNC and NNC groups across prices on the CPT for study cigarettes (A) and usual brand cigarettes (B). Mean consumption at each price is shown in the line series.

Table 1.

Median and Interquartile Ranges (IQR) for Demand Parameters for Versions of the CPT Completed for Both Study Cigarettes and Usual Brand Cigarettes

Parameter	VLNC		NNC		p-valuec
	Median	IQR	Median	IQR
Study cigarettes
Intensity	0.50	0.00–10.00	15.00	6.00–20.00	<.0001
Breakpoint	0.01	0.00–0.60	0.60	0.20–2.00	<.0001
P max	0.00	0.00–0.40	0.40	0.20–0.70	<.0001
O max	0.00	0.00–4.00	5.00	0.70–10.00	<.0001
α a	0.07	0.03–0.15	0.03	0.02–0.08	<.001
Usual brand cigarettes
Intensity	10.00	1.00–20.00	15.00	9.00–20.00	.002
Breakpoint	0.70	0.02–2.00	0.90	0.60–3.00	.025
P max	0.50	0.00–0.70	0.50	0.40–0.80	.013
Omax	4.65	0.00–10.00	6.00	3.00–12.00	.16
α b	0.04	0.02–0.07	0.03	0.02–0.05	.19

Demand parameters at Week 12 for the VLNC and NNC groups for both study cigarettes (top) and usual brand cigarettes (bottom). Intensity, breakpoint, P_max, and O_max were obtained empirically from the data. Alpha (α) was obtained by fitting the data from each participant to a nonlinear least-squares regression after excluding data from some participants who did not follow an exponential curve and could not be fit to the data. Higher α is associated with lower demand.

^aVLNC n = 72, NNC n = 135.

^bVLNC n = 118, NNC n = 148.

^cSensitivity analysis using median regression with multiple imputed data (whenever converged) show similar results.

The Impact of Nicotine Reduction on Demand for Non-cigarette Nicotine and Tobacco Products

For non-cigarette nicotine and tobacco products, empirical parameters include all participants who completed the week 12 assessments, while α estimates are only calculated for participants whose demand curves follow an exponential curve, as described above. Assignment to VLNC cigarettes, compared with NNC cigarettes, was associated with higher O_max for e-cigarettes (Figure 2, VLNC median: 2.30, IQR: 0.00–9.00, NNC median: 0.20, IQR: 0.00–5.00, p = .034), but not intensity, breakpoint, P_max, or α (ps > .05). α calculations for e-cigarettes included 90 and 74 people in the VLNC and NNC groups, respectively. Demand parameters did not differ significantly for other noncombusted tobacco products, including nicotine lozenge, nicotine gum, smokeless tobacco, or nicotine pouches (ps > .05).

Figure 2.

Box plots demonstrating the range, median value (line within the box), and interquartile range (box) purchasing for VLNC and NNC groups across prices on the purchase task for e-cigarettes. The task instructed participants to assume that 1 e-cigarette = 10 puffs. Mean consumption at each price is shown in the line series.

The Impact of VLNC Cigarettes on Self-Reported Non-purchasing of Products at Any Price Greater Than $0.00

Assignment to VLNC cigarettes, compared with NNC cigarettes, was associated with a higher proportion of people who self-reported that they would not smoke any study cigarettes at any price above $0.00 (Figure 3). In the VLNC and NNC groups, 54.2% and 19.1% of people reported they would purchase 0 cigarettes at all prices above $0.00, respectively (OR: 5.02, 95% CI: 3.09% to 8.18%, p < .0001). Assignment to VLNC cigarettes also increased the proportion of people who self-reported that they would not smoke any usual brand cigarettes at any price above $0.00. In the VLNC and NNC groups, 25.9% and 11.6% of people reported they would purchase 0 usual brand cigarettes at all prices above $0.00, respectively (OR: 2.67, 95% CI: 1.50% to 4.78%, p < .001). There were no statistically significant differences between VLNC and NNC groups in the proportion of people who self-reported zero purchasing of e-cigarettes, nicotine lozenges, nicotine gum, smokeless tobacco, or nicotine pouches at any price above $0.00. The sensitivity analysis using multiple imputed data showed similar results (p < .0001 and = .001 for study cigarettes and usual brand cigarettes, respectively).

Figure 3.

The percentage of participants in the VLNC and NNC groups who reported they would not purchase any product for all prices above $0.00 (ie, self-reported non-purchasing) at the week 12 visit. Significant differences indicated by *.

Discussion

In the present secondary analysis, assignment to VLNC cigarettes in a context where other noncombusted sources of nicotine were available reduced the reinforcement value of smoking study cigarettes across multiple demand parameters, including intensity, O_max, P_max, breakpoint, and α. These data suggest that reductions in reinforcement value may drive the changes in smoking behavior and increased ability to quit smoking that have been observed in VLNC clinical trials. The present analysis is the first report of reduced cigarette reinforcement when people who smoke VLNC cigarettes also have access to noncombustible alternative sources of nicotine, demonstrating that nicotine obtained from another source does not function to maintain the reinforcement of cigarettes.

Significant reductions were observed across multiple demand parameters for VLNC cigarettes, suggesting that smoking behavior would likely be reduced following a mandated reduction in cigarette nicotine content, even if cigarettes were not available for free, as is the case in many trials. In the present analysis, O_max (maximum daily expenditure on cigarettes), P_max (price that produces O_max), and breakpoint (lowest price that suppressed cigarette consumption to zero) were also reduced, suggesting that VLNC consumption is likely to be reduced across a range of prices. Median breakpoint (price that suppressed consumption to 0) was $0.01 ($0.20 per pack), and more than half of study participants (54.2%) said they would not purchase study cigarettes at any price above $0.00, which may reflect quitting smoking because participants were told to assume they did not have access to any nicotine or tobacco products other than those offered in each purchase task. These data are consistent with data from one prior trial in which half of participants said if VLNC cigarettes were the only cigarettes available for purchase, 1 year later they would not smoke at all.¹¹ Furthermore, in the primary paper for this clinical trial, cigarettes were available within an ETM that asked participants to “purchase” cigarettes or other products using points that could later be exchanged for money, and there was a significantly higher rate of biochemically verified point prevalence abstinence in the VLNC group compared to the NNC group at week 12 (OR: 3.1, 95% CI: 1.69% to 5.96%).⁵ Together, these data suggest that many people who smoke would not pay the current market price for VLNC cigarettes and would opt to quit smoking altogether if a nicotine reduction policy for cigarettes were implemented in the current, real-world marketplace.

Reinforcement value was also reduced across several demand parameters for usual brand cigarettes, including intensity, breakpoint, and P_max, suggesting that reinforcement value for the smoking of any cigarettes may be reduced following a mandated nicotine reduction policy through reinforcement extinction, including NNC cigarettes. One common concern regarding a nicotine reduction policy is that it may drive an illicit market for NNC cigarettes.²⁰ However, if a nicotine reduction policy reduces demand for NNC cigarettes, as shown in the present analysis, the market for illicit cigarettes may be minimal, especially if noncombusted sources of nicotine are widely available.

In the primary outcomes paper, use of alternative nicotine and tobacco products increased in the VLNC group, with e-cigarettes being the most purchased non-cigarette product.⁵ The present analysis indicated that demand for noncombusted products remained low on the purchase task, but there was a significant increase in O_max for e-cigarettes (maximum money spent in a single day). However, there were no significant differences between the VLNC and NNC groups in demand parameters for other products. It is possible that a mandated reduction in the nicotine content of cigarettes may not increase the reinforcement value of noncombusted alternatives, but those products become relatively more reinforcing compared with cigarettes with low reinforcing value, driving choices to use those products within the trial. It is also possible that changes in demand for other products may take longer to emerge than the length of the trial (ie, 12 weeks).

One strength of the present analysis is the high degree of external validity in the design of the primary trial. Participants had access to assigned cigarettes in an unblinded manner, in the context of a diverse marketplace for 12 weeks. The present analysis is the first analysis of reinforcement value following extended exposure to unblinded VLNC cigarettes and an array of noncombusted tobacco products.

One limitation of the present analysis is that purchase tasks asked people who smoked to assume they had no access to other nicotine or tobacco products when completing the assessment. These instructions are standard for purchase task assessments,^11,13 but in this case, they may have reduced the impact of alternative product availability during the trial on purchase task responding. Prior research has shown that the availability of alternatives within the same session can decrease demand for cigarettes,²¹ and demand for VLNC cigarettes in this study may have been even further reduced in the VLNC group if participants had been told to assume they could also access noncombusted products.

In summary, these data are consistent with and expand upon a body of evidence from clinical trials that have documented the potential of a mandated nicotine reduction policy for cigarettes (and other selected combusted tobacco products) to improve public health by reducing smoking behavior and increasing smoking cessation. Given that cigarette smoking continues to be the leading cause of preventable death in the United States, a mandated reduction policy would likely improve public health.

Data Availability

De-identified data will be made available with a data dictionary and the protocol after the investigators have published their respective papers. Requests for the data should be made to DKH (hatsu001@umn.edu). The submission of an abstract that describes the purpose of the request, data needed, and analysis plan will be requested for record keeping. No financial support will be provided to the requestor.