Impact of nicotine reduction in cigarettes on smoking behavior and exposure: Are there differences by race/ethnicity, educational attainment, or gender?

Abstract

Background:

Lowering nicotine in cigarettes may reduce smoking prevalences; however, it is not known whether an immediate or gradual reduction in nicotine is the optimal approach for all population groups.

Objectives:

We examined whether the optimal approach to nicotine reduction depended on the education, gender, or race of people who smoke and whether the optimal approach differentially benefited people who smoke based on their education, gender, or race.

Methods:

Secondary analysis was conducted on a randomized clinical trial (N=1,250) comparing (1) immediate reduction from 15.5 to 0.4 mg of nicotine per gram of tobacco(mg/g);(2) gradual reduction to 0.4 mg/g;(3) control group with normal nicotine cigarettes(15.5 mg/g). Outcomes included cigarettes per day(CPD), carbon monoxide(CO), total nicotine equivalents(TNE), 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanol and its glucuronides(NNAL), phenanthrene tetraol(PheT), N-Acetyl-S-(2-cyanoethyl)-L-cysteine(CEMA). Data were analyzed as area under the curve(AUC).

Results:

Results were presented by education (High school[HS] or less n=505, more than HS n=745), gender (males n=701, females n=549), and race (Black participants n=373,White participants n=758). Regardless of education, gender, and race, CPD, CO, TNE, NNAL, PheT, and CEMA were lower in immediate versus gradual nicotine reduction. Comparing immediate versus the control, outcomes were lower for all subgroups; however, the magnitude of the effect for TNE varied by race. Specifically, geometric mean of the AUC of TNE in immediate versus gradual was 49% lower in Black participants and 61% lower in White participants (p-value=0.047).

Conclusions:

Immediately reducing nicotine in cigarettes has the potential to benefit people who smoke across lower and higher educational attainment, male and female gender, and Black and White race.

INTRODUCTION

While cigarette smoking has declined in the United States (US) since the 1960s, an estimated 34 million adults still smoked cigarettes in 2019. (Cornelius et al., 2020) Certain characteristics of the U.S. population are associated with elevated prevalences of cigarette smoking and/or smoking-related disease and death. For instance, data from the 2018 National Health Interview Survey (NHIS) shows a smoking prevalence among adults with lower educational attainment (i.e., high school degree, General Educational Development (GED), or less) of 20% compared with 6% among those with higher educational attainment (i.e., at least a bachelor’s degree). NHIS data also shows that the prevalence of current smoking among males was higher than among females (16% versus 12%). (Creamer et al., 2019) While the smoking prevalence in Black adults is similar to White adults at 15%, (Creamer et al., 2019) Black adults are significantly more likely to die from smoking-related diseases than White adults (Centers for Disease Control and Prevention) and this particular disparity is likely a result of inequitable systems (i.e., structural racism) that can harm the health of Black individuals in a variety of ways including inadequate access to quality health care, diminished health behaviors as coping mechanisms, and targeted marketing from the tobacco industry. (Egede and Walker, 2020) Innovative approaches to reduce smoking and its associated health burdens are urgently needed. (Hatsukami and Carroll, 2020)

One potential approach is to reduce the levels of the addictive drug nicotine in the cigarette itself. In 2018, the U.S. Food and Drug Administration, which has the authority to set tobacco product standards for the protection of public health in the US, issued an Advanced Notice of Proposed Rulemaking for limiting nicotine in cigarettes and potentially other combusted products to minimally addictive levels. (U.S. Food and Drug Administration, 2018) At present, there is a growing evidence base suggesting that reducing levels of nicotine found in conventional, commercially-available cigarettes (i.e., 15.5 mg of nicotine per gram of tobacco) by approximately 95% (i.e., to 0.4 mg/g) could benefit the health of people who smoke.(Berman and Glasser, 2019; White et al., 2020) Benefits may be conveyed via reductions in the number of cigarettes smoked and subsequently, reduced exposure to harmful constituents, as well as by increasing number of smoke free days, quit attempts, and likelihood of quitting success. (Donny et al., 2015; Hatsukami et al., 2010; Hatsukami et al., 2018; Higgins et al., 2020)

Research on the best approach to reducing nicotine in cigarettes has also been conducted. A 20 week randomized clinical trial compared three groups – those who immediately switched from smoking cigarettes with nicotine from 15.5 mg/g to 0.4 mg/g, those who underwent gradual (i.e., step-wise) nicotine reduction over the course of the 20 weeks to 0.4 mg/g, and a control group who experienced no nicotine reduction (i.e., smoked cigarettes with 15.5 mg/g during all 20 weeks). (Hatsukami et al., 2018) The study found that, among the overall sample, immediate nicotine reduction resulted in significantly reduced cigarettes smoked and biomarkers of exposure in people who smoke relative to gradual reduction or the control condition, as well as a greater number of smoke free days. (Hatsukami et al., 2018) The results led to the recommendation for setting a date at which all cigarettes would not be able to exceed a certain threshold of nicotine (e.g., 0.4 mg/g nicotine). (Hatsukami et al., 2018)

At present, it is not known whether an immediate or gradual reduction in nicotine in cigarettes is the optimal temporal approach for all subgroups of the U.S. population. To work towards answering this important question, our first objective was to assess whether the effects of the temporal approach to nicotine reduction (i.e., immediate versus gradual) were moderated by the educational attainment, race, or gender of persons who smoke. Should policymakers decide to implement a nicotine reduction policy, it would also be important to understand whether the benefits of the optimal approach would be distributed similarly across subgroups of the U.S. population compared to smoking normal nicotine content cigarettes. Thus, the second objective was to evaluate whether the optimal temporal approach determined through the first objective, was moderated by educational attainment, race, or gender of persons who smoke when compared to no nicotine reduction in cigarettes, which is the current regulatory landscape in which no reduced nicotine content standard has been enacted in cigarettes.

METHODS

Study design and participant eligibility

We conducted a secondary analysis of data from a randomized, parallel, double-blind trial among people who smoke cigarettes (N=1,250). While main findings from the trial have been published (Hatsukami et al., 2018), this will be the first analysis to examine the data by educational attainment, race, or gender –which was a secondary aim of the trial. The trial was conducted at 10 U.S. sites between 2014 and 2016. Each of the 10 sites obtained approval for this study from their respective institutional review board. The study protocol was also reviewed by the US Food and Drug Administration. To participate, all participants provided informed consent prior to enrolling into the study.

Participant eligibility criteria included the following: being at least 18 years of age; breath alcohol level <0.02%; self-report smoking 5 or more cigarettes per day (CPD); biochemical confirmation of active smoking (expired carbon monoxide [CO] > 8ppm or urinary cotinine > 1000 ng/ml); not breastfeeding, pregnant or planning to become pregnant; stable mental or physical health conditions as determined by a study physician; negative urinary toxicology test for illicit drug use with the exception of cannabis; no intentions to quit smoking in the next 30 days; did not use roll-your own cigarettes exclusively; did not use tobacco products other than machine-manufactured cigarettes for > 9 days of the past 30 days; and no previous use of reduced nicotine content study cigarettes. Additional information on the study methods and procedures can be found elsewhere. (Hatsukami et al., 2018)

Study procedures

All participants completed a 2-week baseline phase during which they smoked their usual brand of cigarettes. Afterwards, participants were randomly assigned to one of the following conditions in a 2:2:1 ratio for a duration of 20 weeks: (1) immediate nicotine reduction to 0.4 mg/g, (2) gradual reduction where nicotine content was reduced every four weeks until reaching 0.4 mg/g (nicotine contents were 15.5, 11.7, 5.2, 2.4, and 0.4 mg/g), or (3) usual nicotine content control (15.5 mg/g). Study cigarettes used across all conditions after randomization were provided by the National Institute on Drug Abuse of the National Institutes of Health and have been described previously. (Donny et al., 2015)

After randomization, participants attended a weekly clinic visit for the first 4 weeks. Afterwards, visits were held biweekly. At clinic visits, participants were provided free study cigarettes in an amount that was twice their baseline cigarette use to allow for an increase in smoking behavior or in the event of a missed clinic visit. Participants received standardized counseling on the importance of smoking only study cigarettes and were encouraged to be honest if they smoked non-study cigarettes. During visits, all participants completed surveys, and provided CO samples and first morning void urine samples for outcome assessment (described below). Throughout the study, participants used a phone-based interactive voice response system to report the number of study and non-study cigarettes smoked per day.

To increase compliance with smoking only their assigned study cigarettes, participants, regardless of condition, were told that a urine sample collected at each visit would be randomly selected for analysis to determine whether they had used non-study cigarettes and that a bonus payment would be made if their urine showed that they were only smoking study cigarettes. In reality, bonus payments were provided to participants in the immediate and gradual reduction conditions when the participant’s urine total nicotine equivalent levels (TNE)—the gold standard measure for internal nicotine dose—were ≤12 nmol/mL when smoking the 0.4 mg of nicotine cigarette (i.e., weeks 18 and 20 for both groups). This threshold allowed minimal use of non-study cigarettes. (Denlinger et al., 2016) All participants in the control condition received bonuses.

Measures

Moderators

For the present study, we were interested in the following as potential moderators which were assessed via self-report in a baseline survey: 1) educational attainment based on response to the following question: “What is the highest level of education you completed?” (response options: 8^th grade or less, some high school, high school grad/GED, some college/2-year, college grad/4-year, and graduate/professional). For the present study, participants who reported their highest education as 8^th grade or less, some high school, or high school grad/equivalent were compared to participants who reported their highest education as college/2-year, college grad/4-year, or graduate/professional; 2) Gender based on participants being asked to select their gender (response options: male, female); 3) Race based on participants being asked to select their race with the option to select more than one (response options: White, Black, Asian, American Indian/Alaska Native, Native Hawaiian/Pacific Islander, and Other race). For the present study, participants who selected only White race were compared to participants who selected only Black race because less than 10% of participants identified as being one of the other race groups or of multiple races.

Outcomes

To evaluate differences in smoking behavior and exposure between conditions, the following outcomes collected throughout the 20 week trial were examined: total CPD (study and non-study cigarettes), any cigarette free days (yes or no), number of cigarette free days, CO, TNE, urinary 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanol and its glucuronides (total NNAL; metabolites of a tobacco-specific nitrosamine), urinary phenanthrene tetraol (PheT; a biomarker of exposure to polycyclic aromatic hydrocarbons), and urinary N-Acetyl-S-(2-cyanoethyl)-L-cysteine (CEMA; a mercapturic acid metabolite of acrylonitrile).

Covariates

We evaluated several covariates measured at baseline by educational attainment, gender, and race. Covariates included socio-demographic variables presented in Table 1 and several measures related to smoking behavior including cigarette dependence assessed by the Fagerström Test for Cigarette Dependence (FTCD; scored 1 to 10 with higher values indicative of stronger dependence) (Fagerström, 2011); serum nicotine metabolite ratio (NMR) with larger values indicating a faster rate of nicotine metabolism which has been associated with greater smoking intensity (Carroll et al., 2020); duration of regular smoking defined as the participant’s age at baseline minus participant’s age started smoking regularly.

Table 1.

Participant socio-demographics, smoking behavior characteristics, and biomarkers of exposure at baseline by moderator subgroups

	≤ High school (N = 505)	> High school (N=745)	p-value	Male (N = 701)	Female (N=549)	p-value	Black (N=373)	White (N = 758)	p-value
Age (years), mean (SD)	45.36 (13.04)	44.93 (13.62)	0.5774	43.75 (13.34)	46.84 (13.26)	<.0001	47.20 (10.76)	44.96 (14.35)	).0078
Male sex, %	61.39	52.48	0.0019	NA	NA		58.18	54.35	).2237
Race, %			<.0001			0.2579
White	49.49	70.08		59.88	64.19		NA	NA
Black	44.24	21.04		31.54	28.94		NA	NA
Other	6.26	8.88		8.58	6.86		NA	NA
Hispanic ethnicity, %	4.95	5.50	0.6680	5.56	4.92	0.6126	0.54	4.88	<.0001
Education, %						0.0019			<.0001
≤ High school	NA	NA		44.22	35.52		58.71	32.32
> High school	NA	NA		55.78	64.48		41.29	67.68
Employment status, %			<.0001			0.6683			<.0001
Employed (full/part-time)	34.85	51.54		46.22	42.99		37.27	47.36
Unemployed	32.28	15.03		20.97	23.32		30.83	18.07
Disability	13.66	7.92		10.13	10.38		15.01	8.44
Other	19.21	25.50		22.68	23.32		16.89	26.12
Years smoking, mean (SD)	27.60 (13.40)	26.66 (13.81)	0.2285	25.54 (13.77)	28.95 (13.26)	<.0001	27.43 (1176)	27.83 (14.38)	).6472
Menthol smoker, %	56.24	38.66	<.0001	44.08	47.91	0.1779	89.01	25.07	<.0001
Serum NMR, GM (IQR)	0.38 (0.23, 0.48)	0.40 (0.27, 0.54)	0.1477	0.37 (0.24, 0.48)	0.42 (0.28, 0.59)	0.0054	0.36 (0.20, 0.44)	0.41 (0.29,0 0.55)	).0051
FTCD, mean (SD)	5.67 (1.92)	5.10 (2.22)	<.0001	5.42 (2.12)	5.22 (2.11)	0.0989	5.35 (1.92)	5.38 (2.20)	).8299
CPD, mean (SD)	18.23 (8.74)	17.62 (8.86)	0.2308	18.9 6 (9.21)	16.48 (8.07	<.0001	14.91 (7.62)	19.64 (9.11)	<.0001
CO, mean (SD)	19.03 (8.85)	19.17 (9.71)	0.8069	19.26 (9.25)	18.92 (9.53)	0.5166	16.52 (8.26)	20.66 (9.72)	<.0001
TNE (nmol/mg), GM (IQR)	55.02 (40.77, 84.55)	58.63 (41.33, 89.78)	0.1037	51.33 (38.49, 77.55)	65.52 (46.51, 101.07)	<.0001	44.55 (34.42, 67.73)	65.87 (46.84, 97.36)	<.0001
NNAL (pmol/mg), GM (IQR)	1.28 (0.78, 2.20)	1.23 (0.70, 2.42)	0.4418	1.11 (0.66, 2.07)	1.44 (0.90, 2.67)	<.0001	1.01 (0.65, 1.73)	1.46 (0.87, 2.69)	<.0001
PheT (pmol/mg), GM (IQR)	2.20 (135, 3.55)	2.37 (1.43, 3.88)	0.0942	2.16 (1.32, 3.50)	2.49 (1.47, 4.05)	0.0010	1.59 (0.91, 2.54)	2.76 (1.72, 2.49)	<.0001
CEMA (pmol/mg), GM (IQR)	0.63 (0.42, 1.01)	0.69 (0.45, 1.14)	0.0696	0.62 (0.40, 1.02)	0.72 (0.48, 1.16)	0.0010	0.54 (0.38, 0.88)	0.75 (0.50, 1.17)	<.0001

SD: standard deviation; IQR: interquartile range; NMR: Nicotine metabolite ratio; FTCD: Fagerstrom Test for Cigarette Dependence; CPD: Cigarettes per day; TNE: total nicotine equivalents; NNAL: urinary 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanol and its glucuronides; PheT: phenanthrene tetraol; CEMA: N-Acetyl-S-(2-cyanoethyl)-L-cysteine

Statistical Analyses

As previously described, (Hatsukami et al., 2018) missing data were imputed by the Markov Chain Monte Carlo–based multiple imputation method. (Little and Rubin, 2019; Schafer, 1997) Imputation was performed for each condition (i.e., immediate, gradual, control) separately, as described previously. (Hatsukami et al., 2018) Data were analyzed as area under the curve (AUC) to assess CO, total CPD, and urinary biomarker outcomes resulting from an immediate as opposed to gradual reduction in nicotine content of cigarettes. The AUC was calculated using the trapezoidal rule for the imputed data and then scaled by follow-up time (i.e., time-scaled AUC), and hence the unit of AUC is the same as the unit of its respective exposure variable. Linear regression was used to analyze the AUC (or log AUC for analyses of the biomarkers of exposure to help correct for the skewness of these measures). For non-transformed AUC, the treatment effects are presented as adjusted mean difference (MD) in AUC; for log AUC, the treatment effects are presented as the adjusted ratio of geometric means (RGM), which was calculated as the exponential of the adjusted MD in log AUC. The dichotomous outcome of any cigarette free days was analyzed using logistic regression and reported as the estimated odds ratio (OR). The count variable for the number of cigarette free days was evaluated by the negative binomial regression and the results presented as the estimated incidence rate ratios (IRRs).

All models were run separately for each moderator of interest (i.e., educational attainment, gender, race) where the model included an interaction term between condition and the moderator. Models were adjusted for the baseline level of the variable (except for the outcomes any cigarette free days and number of cigarette free days), study site, and baseline characteristics that differed by condition (i.e., employment, FTCD, and NMR). (Hatsukami et al., 2018) Additionally, models were adjusted for covariates that differed across levels of a given moderator variable based on a p-value<0.2 (see Table footnotes for covariates adjusted for in analyses as they differed based on the given moderator). For each treatment condition, Fisher’s exact test compared the 20 week compliance to the moderators. A p-value <0.05 was considered statistically significant for interaction effects. All statistical analysis was conducted using SAS version 9.4 (SAS Institute Inc., Cary NC).

RESULTS

Table 1 presents socio-demographics and baseline smoking behavior characteristics of participants (N=1,250) by the moderators of interest. Regarding smoking behavior characteristics at baseline and when compared with participants with higher educational attainment, participants with lower educational attainment were significantly more likely to report preference for menthol cigarettes and had a higher average cigarette dependence score based on FTCD. With respect to gender, male participants when compared with female participants smoked a greater number of cigarettes per day on average, had a shorter average duration of smoking, had a lower average nicotine metabolite ratio (, and a lower average urinary TNE level. Compared to White participants, Black participants smoked fewer CPD on average, were more likely to report preference for menthol cigarettes and to have lower average levels of the following: nicotine metabolite ratio, CO, and TNE. Baseline characteristics across condition (i.e., immediate, gradual, control) have been presented previously. (Hatsukami et al., 2018)

Table 2 presents results from the adjusted AUC analyses testing educational attainment, gender, and race as moderators of the relationship between immediate versus gradual nicotine reduction and the outcomes CO, CPD, and urinary biomarkers of tobacco exposure. In all subgroups, the AUC for CO, CPD, and urinary biomarkers was significantly lower in the immediate reduction group than the gradual reduction group, though moderation analyses supported that the magnitude of the treatment effect varied significantly by subgroup in some cases. Specifically, there was a significant interaction between educational attainment and intervention on total NNAL. For those with lower educational attainment, the geometric mean of the AUC of total NNAL in the immediate group was 16% lower than in the gradual group, whereas a 29% difference was observed for those with higher educational attainment (RGM: 0.84 versus 0.71; p-value=0.030). In other words, the effect of immediate versus gradual reduction in reducing total NNAL was larger for the higher versus lower educational attainment subgroup. Regarding gender, male versus female participants had a significantly smaller mean difference in AUC for CO when comparing immediate versus gradual reduction (MD: −3.40 versus −5.10; p-value=0.028). Also, moderation was supported for TNE where the geometric mean of the AUC of TNE in the immediate group was 34% lower than in the gradual group for male participants, whereas 45% lower for female participants (RGM: 0.66 versus 0.54; p-value=0.041). Regarding race, the mean difference in AUC when comparing immediate versus gradual reduction was significantly smaller for Black participants than White participants for CO (MD: −2.48 versus −4.70; p-value=0.006) and CPD (MD: −4.01 versus −5.64; p-value=0.045). Also, the geometric mean of the AUC of TNE in the immediate group was 29% lower than in the gradual group for Black participants, whereas 43% lower for White participants (RGM: 0.71 versus 0.57; p-value=0.034). See Supplementary Table 1 for the results from unadjusted modeling.

Table 2.

Evaluating education, gender, and race as effect modifiers of the relationship between the treatments immediate versus gradual nicotine reduction and expired carbon monoxide (CO), cigarettes per day (CPD), and biomarkers of tobacco exposure analyzed as area under the curve (AUC) for the 20 week intervention period using multiple imputation for missing values

	≤ high school education	> high school education		Male gender	Female gender		Black	White
	Mean differencea	Mean difference	Interaction p-value	Mean difference	Mean difference	Interaction p-value	Mean difference	Mean difference	Interaction p-value
CO ppm	−3.91 (−5.11, −2.72)	−4.34 (−5.32, −3.37)	0.565	−3.40 (−4.46, −2.33)	−5.10 (−6.22, −3.99)	0.028	−2.48 (−3.82, −1.13)	−4.70 (−5.67, −3.73)	0.006
	<0.001	<0.001		<0.001	<0.001		<0.001	<0.001
CPD	−4.54 (−5.69, 3.40)	−5.63 (−6.59, −4.67)	0.140	−5.55 (−6.54, −4.57)	−4.66 (−5.75, −3.57)	0.213	−4.01 (−5.35, −2.67)	−5.64 (−6.60, −4.69)	0.045
	<0.001	<0.001		<0.001	<0.001		<0.001	<0.001
	Geometric mean ratiob	Geometric mean ratio	Interaction p-value	Geometric mean ratiob	Geometric mean ratio	Interaction p-value	Geometric mean ratiob	Geometric mean ratio	Interaction p-value
TNE (nmol/mg)	0.67 (0.57, 0.79)	0.56 (0.49, 0.64)	0.090	0.66 (0.57, 0.76)	0.54 (0.46, 0.62)	0.041	0.71 (0.59, 0.85)	0.57 (0.50, 0.65)	0.034
	<0.001	<0.001		<0.001	<0.001		<0.001	<0.001
NNAL (pmol/mg)	0.84 (0.74, 0.95)	0.71 (0.64, 0.78)	0.030	0.79 (0.71, 0.88)	0.71 (0.63, 0.80)	0.185	0.85 (0.74, 0.98)	0.74 (0.67, 0.82)	0.115
	0.006	<0.001		<0.001	<0.001		0.024	<0.001
PheT (pmol/mg)	0.87 (0.80, 0.95)	0.89 (0.83, 0.94)	0.720	0.88 (0.82, 0.95)	0.88 (0.82, 0.95)	0.950	0.91 (0.83, 1.01)	0.88 (0.83, 0.94)	0.530
	0.001	0.001		0.001	0.002		0.073	<0.001
CEMA (pmol/mg)	0.70 (0.61, 0.79)	0.62 (0.56, 0.69)	0.163	0.67 (0.60, 0.74)	0.63 (0.55, 0.71)	0.443	0.71 (0.62, 0.83)	0.63 (0.57, 0.70)	0.170
	<0.001	<0.001		<0.001	<0.001		<0.001	<0.001

^a:interpreted as the larger (more negative) difference indicating a greater reduction in the immediate versus gradual nicotine reduction condition.

^b:interpreted as a smaller ratio indicating a greater decrease in the immediate versus gradual nicotine reduction condition.

Parentheses indicated 95% confidence intervals; CO: Carbon monoxide; CPD: Cigarettes per day; TNE: total nicotine equivalents; NNAL: urinary 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanol and its glucuronides; PheT: phenanthrene tetraol; CEMA: N-Acetyl-S-(2-cyanoethyl)-L-cysteine; Education models are adjusted for the baseline of the dependent variable, site, employment, baseline FTND, baseline NMR, gender, race, menthol smoking, and urinary TNE at baseline (adjusted for creatinine); Gender models are adjusted for the baseline of the dependent variable, site, employment, baseline FTND, baseline NMR, age education, CPD at baseline, smoking duration, menthol smoking, CO at baseline, Urinary TNE at baseline (adjusted for creatinine); Race models are adjusted for the baseline of the dependent variable, site, employment, baseline FTND, baseline NMR, age education, CPD at baseline, menthol smoking, CO at baseline, Urinary TNE at baseline (adjusted for creatinine)

Table 3 presents results from the adjusted AUC analyses testing educational attainment, gender, and race as moderators of the relationship between the immediate versus no nicotine reduction and the outcomes CO, total CPD, and biomarkers of tobacco exposure. Regardless of educational attainment, gender, or race, the AUC for CO, CPD, and urinary biomarkers was significantly lower in the immediate nicotine reduction group than the no nicotine reduction group. Moderation analyses supported that the magnitude of the treatment effect significantly varied by race for the outcome TNE. Specifically, for Black participants, the geometric mean of the AUC of TNE in the immediate group was 49% lower than in the gradual group, whereas for White participants, the geometric mean of the AUC of TNE in the immediate group was 61% lower than in the gradual group (RGM: 0.51 versus 0.39; p-value=0.047). In other words, immediate versus gradual reduction had a stronger effect in reducing TNE in White versus Black participants. Neither educational attainment nor gender moderated any of these relationships (interaction p-values > 0.05). See Supplementary Table 1 for the results from unadjusted modeling.

Table 3.

Evaluating education, gender, and race as effect modifiers of the relationship between the immediate nicotine reduction versus no nicotine reduction and expired carbon monoxide (CO), cigarettes per day (CPD), and biomarkers of tobacco exposure analyzed as area under the curve (AUC) for the 20 week intervention period using multiple imputation for missing values

	≤ high school education	> high school education		Male gender	Female gender		Black	White
	Mean differencea	Mean difference	Interaction p-value	Mean difference	Mean difference	Interaction p-value	Mean difference	Mean difference	Interaction p-value
CO ppm	−3.78 (−5.30, −2.26)	−3.41 (−4.67, −2.15)	0.711	−2.94 (−4.25, −1.63)	−4.23 (−5.63, −2.83)	0.188	−2.60 (−4.31, −0.89)	−3.72 (−4.97, −2.48)	0.297
	<0.001	<0.001		<0.001	<0.001		0.003	<0.001
CPD	−5.34 (−6.83, −3.84)	−5.44 (−6.67, −4.21)	0.914	−6.10 (−7.37, −4.83)	−4.55 (−5.95, −3.14)	0.109	−4.30 (−6.07, −2.52)	−5.78 (−7.00, −4.55)	0.182
	<0.001	<0.001		<0.001	<0.001		<0.001	<0.001
	Geometric mean ratiob	Geometric mean ratio	Interaction p-value	Geometric mean ratiob	Geometric mean ratio	Interaction p-value	Geometric mean ratiob	Geometric mean ratio	Interaction p-value
TNE (nmol/mg)	0.46 (0.38, 0.56)	0.39 (0.33, 0.46)	0.220	0.46 (0.39, 0.55)	0.37 (0.31, 0.44)	0.062	0.51 (0.41, 0.64)	0.39 (0.33, 0.45)	0.047
	<0.001	<0.001		<0.001	<0.001		<0.001	<0.001
NNAL (pmol/mg)	0.75 (0.64, 0.87)	0.62 (0.54, 0.70)	0.068	0.70 (0.61, 0.80)	0.62 (0.54, 0.72)	0.285	0.77 (0.65, 0.93)	0.64 (0.56, 0.72)	0.075
	<0.001	<0.001		<0.001	<0.001		0.005	<0.001
PheT (pmol/mg)	0.83 (0.75, 0.92)	0.89 (0.82, 0.97)	0.265	0.87 (0.80, 0.95)	0.86 (0.78, 0.95)	0.889	0.91 (0.81, 1.03)	0.85 (0.78, 0.93)	0.360
	<0.001	0.008		0.002	0.003		0.147	<0.001
CEMA (pmol/mg)	0.76 (0.65, 0.88)	0.76 (0.65, 0.88)	0.174	0.71 (0.63, 0,81)	0.68 (0.58, 0.79)	0.667	0.79 (0.66, 0.95)	0.68 (0.60, 0.77)	0.205
	<0.001	<0.001		<0.001	<0.001		0.011	<0.001

^a:interpreted as the larger (more negative) difference indicating a greater reduction in the immediate versus no nicotine reduction (control) condition.

^b:interpreted as a smaller ratio indicating a greater decrease in the immediate versus no nicotine reduction (control) condition.

Parentheses indicated 95% confidence intervals; CO: Carbon monoxide; CPD: Cigarettes per day; TNE: total nicotine equivalents; NNAL: urinary 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanol and its glucuronides; PheT: phenanthrene tetraol; CEMA: N-Acetyl-S-(2-cyanoethyl)-L-cysteine; Education models are adjusted for the baseline of the dependent variable, site, employment, baseline FTND, baseline NMR, gender, race, menthol smoking, and urinary TNE at baseline (adjusted for creatinine); Gender models are adjusted for the baseline of the dependent variable, site, employment, baseline FTND, baseline NMR, age education, CPD at baseline, smoking duration, menthol smoking, CO at baseline, Urinary TNE at baseline (adjusted for creatinine); Race models are adjusted for the baseline of the dependent variable, site, employment, baseline FTND, baseline NMR, age education, CPD at baseline, menthol smoking, CO at baseline, Urinary TNE at baseline (adjusted for creatinine)

Table 4 presents results from the adjusted analyses testing educational attainment, gender, and race as moderators of the relationship between immediate versus gradual or immediate versus no nicotine reduction and the outcomes any cigarette free days and number of cigarette free days. Comparing immediate versus gradual, immediate nicotine reduction resulted in greater proportion of any cigarette free days and a greater mean number of cigarette free days regardless of educational attainment, gender, and race. Neither educational attainment nor gender moderated any of these relationships (interaction p-values > 0.05). Race was a significant moderator for the outcome number of cigarette free days whereby White participants when compared to Black participants had 2.98 times (95% CI: 1.14, 7.75; p-value=0.025) the rate ratio of cigarette-free days for immediate versus gradual nicotine reduction. See Supplementary Table 2 for unadjusted number of cigarette free days. Comparing immediate versus no nicotine reduction, immediate nicotine reduction resulted in greater proportion of any cigarette free days and a greater mean number of cigarette free days regardless of educational attainment, gender, and race. None of the tests for moderation were significant (all interaction p-values > 0.05).

Table 4.

Evaluating education, gender, and race as effect modifiers of the relationship between conditions and any cigarette-free days and number of cigarette free days during the 20 week study period (n=1001)

Immediate nicotine reduction versus gradual reduction		Immediate nicotine reduction versus no reduction (control)
Any cigarette free days	Number of cigarette free days	Any cigarette free days	Number of cigarette free days
Odds Ratio (95% CI)	IRR (95% CI)	Odds Ratio (95% CI)	IRR (95% CI)
p-value	p-value	p-value	p-value
>HS*Treat†	>HS*Treat†	>HS*Treat†	>HS*Treat†
0.77 (0.42, 1.41)	1.08 (0.45, 2.60)	0.64 (0.30, 1.38)	0.86 (0.28, 2.64)
p=0.398	p=0.866	p=0.257	p=0.794
Male*Treat†	Male*Treat†	Male*Treat†	Male*Treat†
0.86a (0.48, 1.54)	1.01b (0.44, 2.28)	0.92 (0.44, 1.94)	0.93 (0.30, 2.90)
p=0.600	p=0.989	p=0.824	p=0.895
White*Treat†	White*Treat†	White*Treat†	White*Treat†
1.04 (0.55, 1.97)	2.98 (1.14, 7.75)	1.13 (0.50, 2.60)	1.18 (0.33, 4.24)
p=0.913	p=0.025	p=0.766	p=0.797

^aInterpretation: When less than one, the moderator category shown (e.g., Male) has a lower odds and when greater than one, the moderator category shown has a greater odds than the reference moderator category (e.g., females) of having any cigarette-free day in the immediate versus gradual or control condition.

^bInterpretation: Ratio of the rate of cigarette-free day for immediate relative to gradual or control condition in the moderator category shown (e.g., male) compared to the reference moderator category (e.g., female). HS; high school

Education models are adjusted for site, employment, baseline FTND, baseline NMR, gender, race, menthol smoking, and urinary TNE at baseline (adjusted for creatinine); Gender models are adjusted for site, employment, baseline FTND, baseline NMR, age education, CPD at baseline, smoking duration, menthol smoking, CO at baseline, Urinary TNE at baseline (adjusted for creatinine); Race models are adjusted for site, employment, baseline FTND, baseline NMR, age education, CPD at baseline, menthol smoking, CO at baseline, Urinary TNE at baseline (adjusted for creatinine)

Regarding compliance to study cigarettes which was defined as a urinary TNE ≤12 nmol/mL when smoking the 0.4 mg of nicotine cigarette (i.e., measured at weeks 18 and 20 for immediate and gradual reduction groups), there were significant differences across moderator subgroups. See Supplementary Table 3 for the proportion compliant in each moderator subgroup. Specifically, among participants randomized to the immediate nicotine reduction condition, compliance to study cigarettes was lower among participants with low versus higher educational attainment (42.5% versus 59.6%; p=0.003), male versus female gender (44.9% versus 61.6%; p=0.002), and Black versus White race (40.4% versus 57.8%; p=0.005). Similar patterns in compliance by educational attainment, gender, and race were observed in participants randomized to the gradual reduction condition.

DISCUSSION

This paper addressed two important public health questions regarding the potential impact of a nicotine reduction policy in cigarettes. The first was whether the optimal temporal approach to nicotine reduction, either immediate or gradual, depended on the educational attainment, gender, or race of persons who currently smoke cigarettes. In answering this question, we were most concerned with identifying whether some subgroups benefited while other subgroups were harmed by a particular approach. The parent study found that immediate versus gradual nicotine reduction resulted in greater reductions in biomarkers of exposure among the overall sample.(Hatsukami et al., 2018) When examining moderators in the present study, we observed that an immediate versus gradual nicotine reduction resulted in greater reductions in smoking behavior and biomarkers of exposure, as well as more smoke-free days regardless of the level of education, gender, or race of persons who currently smoke cigarettes. These results provide additional evidence that an immediate nicotine reduction is likely more optimal than gradual nicotine reduction in the subgroups studied. The second question was whether the optimal temporal approach differentially benefited participants based on their educational attainment, gender, or race when compared to the control group, a group which mirrors the current regulatory landscape in which no reduced nicotine content standard has been enacted in cigarettes. The parent study found significant differences in immediate versus no nicotine reduction among the overall sample. When examining moderators in the present study, we observed that regardless of level of educational attainment, gender and race, the immediate nicotine reduction group versus no nicotine reduction group had fewer CPD and lower levels of biomarkers of exposure, as well as more cigarette free days. Thus, an immediate reduction in nicotine holds promise in reducing smoking behavior and harmful exposure across all moderator subgroups examined.

Further interpretation of the results testing race as a moderator suggest that if an immediate nicotine reduction in cigarettes were implemented, Black adults who smoke cigarettes may experience smaller reductions in some biomarkers of exposure as compared to White adults who smoke cigarettes. For instance, when comparing immediate versus no nicotine reduction levels of TNE in participants who identified as Black decreased to a lesser extent than among participants who identified as White and total NNAL trended in a similar direction although this analysis did not achieve statistical significance. However, results on compliance demonstrated that Black participants were less compliant than White participants with exclusively smoking assigned study cigarettes (versus normal nicotine content cigarettes), which could contribute to smaller reductions in TNE in the study. If a nicotine reduction policy was implemented, normal nicotine cigarettes would not be readily available. It is possible, depending on the specific reasons why non-compliance occurred, which were not examined here, that Black adults who smoke may be more likely than White adults who smoke to seek alternatives to low nicotine cigarettes or quit tobacco altogether. If alternatives comprise tobacco products that have been proven to be less harmful than cigarettes then there is potential for this policy to help close the gap in Black to White smoking-related disparities. Future research should examine compliance to using study products when participants who smoke have access to both low nicotine cigarettes and other tobacco or nicotine products since this would better reflect the real-world marketplace and this two-pronged approach (e.g., providing both low nicotine cigarettes and other tobacco or nicotine products) was put forward by the U.S. FDA.(Gottlieb and Zeller, 2017)

Prior research among people who smoke and have lower educational attainment or psychiatric conditions found that reducing the nicotine content of cigarettes would benefit these population subgroups by reducing the addictive potential of cigarettes and a secondary analysis of the data found that male and female people who smoke would potentially benefit similarly (i.e., no gender interaction). (Higgins et al., 2017; Higgins et al., 2020; Streck et al., 2019) Likewise in the present study, there was no indication that male and female participants differed in their response to an immediate versus no nicotine reduction in cigarettes. We also found no indication that educational attainment moderated response to an immediate versus no nicotine reduction.

Thus, an immediate nicotine reduction in cigarettes may confer a similar benefit in terms of smoking behavior and exposure for those of lower versus higher educational attainment and for males and females. Results on compliance demonstrated that male participants and those with lower educational attainment were less compliant to smoking only study cigarettes compared with their counterparts. Similar to the interpretations for race, people of lower educational attainment and males may be more likely than their counterparts to seek other tobacco or nicotine products or become abstinent shall the FDA implement this policy.

The present study has several limitations that merit acknowledgement. First, because of the post-hoc nature of this study we may have not been powered to detect significant interactions and therefore, non-significant tests for interactions have the potential to be due to false negatives (i.e., Type II errors). However, given the sample sizes among the three groups in our study, the subgroup comparisons are still able to detect small effect sizes in the range of 0.2 to 0.3 with power of at least 80%. Second, while we focused on some population subgroups, small samples sizes limited our ability to study all population characteristics associated with elevated smoking behavior and/or smoking-related disease and death such as American Indian/Alaska Native race, and identifying as LGBTQ+. Future studies including these groups are recommended to expand upon the evidence base. Third, the entire study duration was 20 weeks and therefore the long-term effect of reduced nicotine content cigarettes in these populations remain uncertain. Fourth, study cigarettes were provided to participants at no cost. Therefore, the effects of participants needing to pay for these cigarettes is unclear.

Conclusions

The present study examined if educational attainment, gender, or race moderated the impact of a potential nicotine reduction policy in cigarettes—a necessary assessment for anticipating the full public health impact of a nicotine reduction policy. We found that regardless of educational attainment, gender, and race, participants assigned to immediate nicotine reduction had reduced smoking behavior, lower levels of biomarkers of exposure, and more smoke-free days when compared to either a gradual nicotine reduction or no nicotine reduction. However, only about one-half of participants in the immediate nicotine reduction condition were compliant to study cigarettes and compliance was lower among participants with lower versus higher educational attainment, male versus female gender, and Black versus White race. There is potential for this policy to help close the gaps in existent smoking disparities shall people of lower educational attainment, male gender, and Black race be more likely to seek alternatives to low nicotine cigarettes that are less harmful or quit using tobacco altogether. Future research should examine the impact of providing participants who smoke access to both low nicotine cigarettes and less harmful tobacco or nicotine products. In summary, the results suggest that an immediate reduction in the nicotine content of cigarettes has the potential to benefit people who smoke across low and higher educational attainment, male and female gender, and Black and White race.

Highlights.

• An immediate versus gradual nicotine reduction resulted in greater reductions in smoking behavior and biomarkers of exposure, as well as more smoke- free days regardless of participants’ education, gender, or race

• An immediate reduction (when compared to the present with no nicotine reduction) in nicotine holds promise in reducing smoking behavior and harmful exposure across all subgroups examined.

• There is potential for this policy to help close the gaps in existent smoking disparities by education and race shall the higher risk groups (i.e., people of lower educational attainment, Blacks) be more likely to seek alternatives to low nicotine cigarettes that are less harmful or quit using tobacco altogether.