Smoking Quit Rates Among Menthol vs Nonmenthol Smokers: Implications Regarding a US Ban on the Sale of Menthol Cigarettes

Abstract

Background

A ban on the sale of menthol cigarettes in the United States is currently under consideration. A justification is that menthol cigarettes are harder to quit, particularly for African American smokers who use menthols much more frequently than White smokers, but epidemiologic data are limited.

Methods

In a cohort of 16 425 mostly low-income African American and White current cigarette smokers enrolled during 2002-2009, we computed smoking quit and reuptake rates at 3 follow-ups conducted means of 4.6, 7.7, and 11 years after entry. Generalized estimation equations were used to compute odds ratios (ORs) and 95% confidence intervals (CIs) for quitting and resuming smoking for menthol vs nonmenthol smokers adjusted for race, age, education, income, and smoking pack-years.

Results

Crude annual quit rates among current smokers were 4.3% for menthol and 4.5% for nonmenthol smokers, with adjusted odds ratios of quitting for menthol vs nonmenthol smokers of 1.01 (95% CI = 0.91 to 1.11) overall, 0.99 (95% CI = 0.87 to 1.12) among African American smokers, and 1.02 (95% CI = 0.88 to 1.20) among White smokers. Crude annual smoking reuptake rates were somewhat higher among menthol smokers (8.4%) than nonmenthol smokers (7.1%), with an adjusted odds ratio of 1.19 (95% CI = 0.97 to 1.47), but net quit rates remained similar (OR = 1.01, 95% CI = 0.90 to 1.13 overall; OR = 1.00, 95% CI = 0.86 to 1.15 among African American participants; and OR = 1.04, 95% CI = 0.87 to 1.24 among White participants).

Conclusions

This large-scale prospective survey revealed similar quit rates among menthol and nonmenthol smokers. Results contribute to policy discussions, especially if, as a meta-analysis suggests, lung cancer risk is higher for nonmenthol smokers and a ban leads menthol smokers to switch to nonmenthol cigarettes.

Cigarette smoking is recognized as the primary cause of premature mortality in the United States and other countries (1,2). Since the initial Surgeon General’s report (1) in 1964 classifying smoking as a human carcinogen, the list of diseases among smokers has steadily increased (2). Hence, strategies aimed at the cessation of smoking among smokers and inhibition of smoking initiation among nonsmokers, especially among youth and young adults, are crucial to reducing the health burden of smoking.

The US Food and Drug Administration recently proposed banning the sale of menthol cigarettes in the United States as one means of combating the adverse health effects of smoking. The attention to menthol cigarettes, rather than all cigarettes, is related to Food and Drug Administration oversight of cigarette additives. Scientific rationale for a focus on menthols included the possibility that menthols may be more addictive with lower smoking cessation rates (3). In an updated literature review and meta-analysis, no statistically significant differences in quit rates were found between menthol and nonmenthol smokers overall, but among African American smokers, cessation was less likely among menthol users (4).

The Southern Community Cohort Study (SCCS) provides a unique opportunity to evaluate smoking cessation rates among low-income populations, particularly with regard to potential differences between African American and White smokers. SCCS participants include large numbers of African American smokers, who mainly smoke menthols but with a sizeable minority smoking nonmenthols, and of White smokers, who mainly smoke nonmenthols but with a sizeable minority smoking menthols, all of whom are under systematic active follow-up. We thus assessed whether SCCS menthol smokers had quit smoking at lower rates than nonmenthol smokers, as described below.

Methods

The Study Population

Description of the enrollment, characteristics, and follow-up of participants in the SCCS are provided in detail elsewhere (5-7). In brief, during 2002-2009, approximately 86 000 adults aged 40-79 years were enrolled in the cohort across 12 Southern states (AL, AR, FL, GA, KY, LA, MS, NC, SC, TN, VA, and WV) and consented to long-term follow-up. Most (85%) were recruited from community health centers, institutions providing basic health and preventive care in underserved areas, so that the cohort comprised many individuals with low income and education levels. The cohort was intentionally designed to recruit a high percentage (two-thirds) of African American participants to help remedy underrepresentation of African American people in previous large-scale epidemiologic studies. At community health centers, personal interviews were conducted to ascertain information about the demographic, medical, lifestyle, and other characteristics of the participants. Detailed smoking histories were obtained, including age started, age quit for former smokers, amount and duration of smoking, and usual type of cigarette (menthol or nonmenthol) smoked. Written informed consent was obtained from all participants at study entry. The study protocol was reviewed and approved by institutional review boards of the Vanderbilt University Medical Center and Meharry Medical College.

Follow-up surveys were conducted among SCCS participants; the first during 2008-2012, the second during 2012-2015, and the third during 2015-2017. Questionnaires mailed to the participants in each survey included questions about whether the participant was a current smoker.

From the baseline interview, all persons who identified themselves as current cigarette smokers were classified as menthol or nonmenthol smokers based on the question “Are the cigarettes you usually smoke menthol?” Those with missing information on cigarette type were excluded. Among these smokers, 16 425 individuals who completed the first follow-up survey an average 4.6 years after cohort entry were identified and classified as a continuing or a quit smoker based on their response to the current smoking question in this follow-up survey. Among these persons, 10 015 completed the second follow-up an average 3.1 years later and were classified as a continuing or a quit smoker based on their response to the smoking question in the second follow-up survey. Finally, among these 10 015 individuals, 6599 completed the third follow-up survey an average of 3.3 years later and were classified as a continuing or quit smoker based on their response to the smoking question in the third follow-up survey.

Statistical Analysis

In initial statistical analyses, we computed crude quit rates among smokers who had quit in each of the 3 intervals studied (baseline to first follow-up, first to second follow-up, and second to third follow-up). These crudes quit rates were computed and annualized by dividing the numbers of quitters across the interval by the person years of follow-up within the interval. Because some who quit smoking relapsed and again took up the habit, we similarly computed crude smoking reuptake rates by follow-ups 2 and 3 among those who, respectively, reported that they had quit smoking at follow-ups 1 and 2. We also estimated crude net cumulative annual quit rates, considering that some quitters resumed smoking, by decreasing annual quit rates by the annual smoking reuptake rates.

For the current longitudinal data with 3 repeated measurements, we used weighted generalized estimating equations based on inverse probability weighting (8,9) to estimate the adjusted odds of quitting or smoking resumption for menthol vs nonmenthol smokers while considering the within-participant correlation and handling missing data under the missing at random assumption induced by dropouts during the follow-up. The model selection, including variable selection and the “working” correlation structure selection, was based on the quasi-likelihood under the independence model criterion, with a smaller quasi-likelihood under the independence value representing a better model (10). The covariates included in the final models for the association between menthol use and quitting smoking or for smoking reuptake among those who had quit in follow-ups 1 or 2 were age at the follow-ups, race, education, income, and cumulative pack-years of smoking (with education and income assessed at the baseline time of cohort entry and pack-years calculated using data obtained at baseline as well as during follow-up). The dropout missingness was considered to be affected by race, education, income, age at the follow-up interviews, the times of the follow-up, and participants’ smoking quitting status in the previous follow-up intervals. Education, income levels, and cumulative pack-years had a small percentage of missing values, so we used multiple imputation with 15 imputations to enable us to include the participants with missing covariate values. Finally, we combined the weighted generalized estimating equation results from 15 imputed datasets with the SAS MIANALYZE procedure.

We also carried out a meta-analysis across epidemiologic studies that had computed relative risks of lung cancer among smokers of menthol cigarettes vs nonmenthol cigarettes. Included were studies reported in a previous meta-analysis (11) as well as 2 newer investigations (7,12), one of which (7) was based on SCCS data. Relative risk estimates and their 95% confidence intervals (CIs) were obtained from the published reports and graphically displayed in a forest plot. Summary meta-analytic relative risk estimates were obtained using both fixed and random effect models, and a heterogeneity statistic (I2) and its χ² test value were used to assess consistency across the studies.

Results were considered statistically significant if P values were less than .05 or 95% confidence intervals excluded the null value 1.0.

Results

A total of 16 425 SCCS participants reported that they were current smokers of either menthol (10 830) or nonmenthol (5595) cigarettes at entry into the cohort during 2002-2009 and completed the first follow-up survey during 2008-2012. Table 1 shows the numbers of these smokers who reported that they had quit smoking at follow-up 1 as well as the numbers of smokers at the start of follow-ups 1 and 2 who reported that they had quit at follow-ups 2 and 3. The crude annual quit rates in each interval were similar for menthol and for nonmenthol smokers (4.3% and 4.5% per year overall, respectively; crude rate ratio = 0.97).

Table 1.

Numbers of menthol and nonmenthol smokers and crude annual smoking quit rates among smokers at the start of the 3 survey intervals and overall

Interval	Menthol			Nonmenthol			Crude quit rate ratio
	Total	Quitters	Interval annual quit ratea, %	Total	Quitters	Interval annual quit ratea, %
Baseline to follow-up 1	10 830	2159	4.2	5595	1019	4.2	0.98
Follow-up 1 to follow-up 2	5306	686	4.2	2841	385	4.5	0.94
Follow-up 2 to follow-up 3	3328	584	5.4	1686	306	5.5	0.98
Total			4.3			4.5	0.97

Numbers of quitters divided by person years of follow-up in interval (or overall, for total).

The multivariable analyses indicated that adjusted odds ratios for quitting smoking increased with advancing age, increasing household income, and college education; were higher among women than men; decreased with increasing pack-years of smoking; and were not strongly related to race (Table 2). The overall odds ratio for quitting among menthol vs nonmenthol smokers was 1.01 (95% CI = 0.91 to 1.11). Similar associations were observed for African American (OR = 0.99, 95% CI = 0.87 to 1.12) and White (OR = 1.02, 95% CI = 0.88 to 1.20) participants. Table 3 shows crude overall cumulative quit rates as well as adjusted odds ratios for each of African American women and men and White women and men. The quit rates among men were somewhat lower for menthol than nonmenthol smokers, but none of the differences among the 4 race–sex groups were statistically significant, with the confidence limits about the odds ratios all well including the null value of 1.0.

Table 2.

Adjusted odds ratios (95% CI) for quitting smoking according to menthol vs nonmenthol cigarette type and covariates

Indicator	ORa (95% CI)
Cigarette type
Nonmenthol	1.00 (Referent)
Menthol	1.01 (0.91 to 1.11)
Sex
Female	1.00 (Referent)
Male	0.89 (0.82 to 0.96)
Race
African American	0.93 (0.83 to 1.03)
White	1.00 (Referent)
Age, y	1.05 (1.05 to 1.06)
Pack-years smoking per 5 pack-years	0.93 (0.92 to 0.94)
Education
Less than high school education	1.00 (Referent)
High school education	1.03 (0.94 to 1.14)
Some post high school	1.05 (0.95 to 1.17)
College	1.25 (1.08 to 1.45)
Income
<$15 000	1.00 (Referent)
$15 000-$24 000	1.09 (0.99 to 1.19)
$25 000-$49 000	1.32 (1.17 to 1.48)
≥$50 000	1.42 (1.20 to 1.67)

Odds ratio and confidence interval for quitting associated with the indicator variable adjusted for all other variables. CI = confidence interval; OR = odds ratio.

Table 3.

Numbers of menthol and nonmenthol smokers, crude cumulative annual quit rates, and adjusted odds ratios for quitting in menthol vs nonmenthol smokers by race and sex

Group	Menthol		Nonmenthol		Crude rate ratio	Adjusted OR for quitting (95% CI)c
	No.a	Cumulative annual quit rateb, %	No.a	Cumulative annual quit rateb, %
Total	10830	4.3	5595	4.5	0.97	1.01 (0.91 to 1.11)
African American women	5285	4.7	884	5.0	0.94	1.02 (0.86 to 1.21)
African American men	4330	3.9	818	4.6	0.85	0.95 (0.78 to 1.15)
White women	887	4.5	2463	4.2	1.07	1.06 (0.88 to 1.27)
White men	328	3.9	1430	4.5	0.87	0.89 (0.65 to 1.21)

Numbers of smokers at baseline to follow-up 1. CI = confidence interval; OR = odds ratio.

Cumulative annual quit rates across all study periods (sum of numbers of quitters across the 3 study intervals divided by sum of person years of follow-up across the 3 study intervals).

Odds ratio for quitting smoking for menthol vs nonmenthol smokers, adjusted for covariates listed in Table 2 (except race and sex for race-sex–specific ORs).

Totals of 1868 and 1585 SCCS participants who completed follow-ups 2 and 3, respectively, had reported in the follow-up 1 and 2 surveys that they had quit smoking. Crude cumulative annual smoking resumption rates among these quitters were 8.4% among those who quit menthol cigarettes and 7.1% among those who quit nonmenthols, with an adjusted odds ratio for smoking reuptake of 1.19 (95% CI = 0.97 to 1.47) (Table 4). The table also shows that adjusted odds ratios for smoking reuptake among menthol vs nonmenthol smokers were higher for African American than White participants, but for none of the 4 race–sex groups did the 95% confidence interval exclude the null value of 1.0.

Table 4.

Numbers of menthol and nonmenthol smokers who reported quitting at follow-up 1 or 2, crude cumulative annual smoking reuptake rates following quitting, and adjusted odds ratios for smoking reuptake in menthol vs nonmenthol smokers overall and by race and sex

Group	Menthol		Nonmenthol		Crude rate ratio	Adjusted OR for resuming smoking (95% CI)c
	No.a	Cumulative annual reuptake rateb, %	No.a	Cumulative annual reuptake rateb, %
Total	1680	8.4	868	7.1	1.19	1.19 (0.97 to 1.47)
African American women	916	8.2	173	6.0	1.36	1.35 (0.94 to 1.97)
African American men	560	9.7	119	8.4	1.16	1.22 (0.83 to 1.81)
White women	158	6.2	351	7.3	0.85	1.09 (0.71 to 1.67)
White men	46	7.1	225	6.9	1.03	1.06 (0.54 to 2.09)

Numbers of participants who quit smoking at follow-up 1 or follow-up 2. CI = confidence interval; OR = odds ratio.

Cumulative annual reuptake rates across the follow-up 1-2 and follow up 2-3 intervals, computed by dividing the numbers of quitters who resumed smoking by the sum of person years of follow-up of these quitters across these 2 study intervals.

Odds ratio and 95% confidence interval for quitting smoking for menthol vs nonmenthol smokers, adjusted for covariates listed in Table 2 (except race and sex in the race-sex–specific odds ratios).

Not shown in Table 4, there was a difference in smoking reuptake rates when follow-up was short term after quitting (ie, smoking resumption at follow-up 2 or follow-up 3, respectively, after newly reporting quitting in follow-up 1 or follow-up 2) than longer term after quitting (ie, smoking resumption at follow-up 3 after reporting quitting at follow-up 1). Crude annual reuptake rates were 10.3% and 9.2% for short-term quitters of menthol and nonmenthol cigarettes (adjusted OR = 0.96, 95% CI = 0.76 to 1.20) and much lower at 3.2% and 1.9% (OR = 1.93, 95% CI = 1.02 to 3.65) for longer-term quitters. Follow-up 3 also enabled examination of quit rates among the relatively small number of smokers at baseline who quit at follow-up 1 and resumed smoking at follow-up 2 (n = 370), compared with the participants (n = 4644) who were continuing smokers throughout (ie, at baseline, follow-up 1, and follow-up 2). Crude annual quit rates during the interval between follow-up 2 and follow-up 3 were higher among those who had quit before (12.0% for menthol and 14.0% for nonmenthol) than among those who smoked throughout (4.9% for menthol and 5.0% for nonmenthol).

Decrementing the crude cumulative overall annual quit rates by the annual smoking reuptake rates yielded net annual quit rate estimates of 4.0% for menthol and 4.2% for nonmenthol smokers overall, 4.3% vs 4.7% for African American women, 3.5% vs 4.2% for African American men, 4.3% vs 3.9% for White women, and 3.6% vs 4.2% for White men. In the multivariable weighted GEE analysis, adjusted odds ratios for net quitting among menthol vs nonmenthol smokers were 1.01 (95% CI = 0.90 to 1.13) overall, 1.00 (95% CI = 0.86 to 1.15) for African American participants, and 1.04 (95% CI = 0.87 to 1.24) for White participants.

Figure 1 updates a previous meta-analysis (11) on lung cancer risks among menthol vs nonmenthol cigarette smokers. As shown in Figure 1, most studies found lower lung cancer risks, though not statistically significantly lower as judged by the 95% confidence intervals overlapping the null value of 1.0 among menthol vs nonmenthol smokers. Overall, a statistically significantly lower risk of lung cancer was seen among menthol smokers (summary OR = 0.88, 95% CI = 0.81 to 0.96), with no evidence of statistically significant heterogeneity among the studies.

Figure 1.

Meta analysis forest plot of epidemiologic studies assessing relative risks (RRs) of lung cancer among menthol vs nonmenthol cigarette smokers. The error bars represent the 95% confidence intervals (CIs). Heterogeneity was assessed by the I² statistic and its statistical significance by a 9 d.f. χ² test.

Discussion

This prospective study of self-reported smoking habits, with relatively large numbers of both African American and White menthol and nonmenthol cigarette smokers reassessed for up to 15 years, found roughly similar smoking quit rates among menthol vs nonmenthol smokers. Overall, the adjusted odds of quitting were nearly identical (OR = 1.01, 95% CI = 0.91 to 1.11) for menthol and nonmenthol smokers. The similarity was observed among both African American and White participants and among men as well as women, and this similarity persisted over the duration of reassessing SCCS participants. However, there was a suggestion of higher smoking resumption rates for those who quit menthols, particularly for African American participants, although the differences in smoking reuptake rates between menthol and nonmenthol smokers who reported quitting were not statistically significant; net quit rates remained close.

Previous research on smoking quit rates among menthol smokers has provided inconsistent results. In a meta-analysis of 19 studies of North American smokers, no overall association between menthol cigarette use and quitting was observed (4). Included in the meta-analysis were available early SCCS data from our then-ongoing follow-up 1 survey (13). Although no overall association was detected in the meta-analysis, the authors reported a statistically significantly (P = .04) lower quit rate (OR = 0.88, 95% CI = 0.78 to 1.00) among African American smokers who smoked menthol vs nonmenthol cigarettes but not among White smokers (OR = 1.04, 95% CI = 0.96 to 1.13) (4). In our extended follow-up, we found no evidence of lower quit rates among African American menthol smokers after adjusting for covariates (OR = 0.99, 95% CI = 0.87 to 1.12). We did find an approximately 30% increase (OR = 1.30, 95% CI = 0.99 to 1.70) in the odds of resuming smoking for quitters of menthol vs nonmenthol cigarettes among African American participants, but even after taking this into account, adjusted net quit rates remained similar for African American menthol and nonmenthol smokers (adjusted OR = 1.00, 95% CI = 0.86 to 1.15). Although annual smoking reuptake rates were low among those who reported being a former smoker at 2 consecutive intervals, the higher resumption among menthol (3.2%) than nonmenthol (1.9%) smokers suggests that additional research involving 10 or more years of follow-up would be useful to monitor if this persists and higher percentages of quitters of menthol than nonmenthol cigarettes eventually resume smoking.

The large size, systematic prospective follow-up, and standardization of data collection for menthol vs nonmenthol smokers are study strengths in assessing racial differences in quit rates. The enrollment of both African American and White people from federally qualified health centers ensured that socioeconomic differences, which can confound assessment of racial differences, would be minimized because both African American and White participants were of generally similar (and low) socioeconomic status. Further, residual differences in education and income not accounted for in the study design could be adjusted in our statistical analyses so that our comparisons between African American and White smokers, we believe, were made on a more level playing field than is often possible elsewhere.

We acknowledge potential study weaknesses, including reliance solely on self-report of smoking cessation. We also asked only if the usual brand of cigarettes smoked were menthols, and some smokers may have switched or smoked both types. In the second follow-up survey, we again asked current smokers if the type of cigarette usually smoked was menthol and found only 7.2% of menthol and 8.1% of nonmenthol smokers reporting switching, consistent with previous reports that switching is uncommon (14). We also had no information on the number of quit attempts made by either menthol or nonmenthol smokers. More than 9000 African American menthol smokers were included in our analyses, a larger sample size than in any of the previous studies, but still with quit rates at less than 5% per year and annual smoking resumption rates in less than 10% of the quitters, the numbers for robust analyses of menthol vs nonmenthol differences in smoking reuptake were not large, especially for long-term quitters. Although large variations in response rates in the follow-up surveys by cigarette type were not observed, dropout rates of 35%-39% between our baseline and follow-up surveys are a concern. Our generalized estimation equation and multiple imputation sensitivity analyses, however, found that the odds ratios for quitting among menthol vs nonmenthol smokers were barely changed in analyses with and without accounting for missing data. The cohort studied also is not representative of the overall US population, because participants were recruited only from Southern states and the primary recruitment sources were community health centers, institutions providing services mainly to those of low income. Although generalizability is affected, internal validity in odds ratio estimation is maintained because we are comparing quit rates among a study base of participants, both menthol and nonmenthol smokers, who voluntarily enrolled in the follow-ups.

The findings from this report are relevant to considerations of the ban on the sale of menthol cigarettes currently proposed by the US Food and Drug Administration. One of the rationales for the ban is the concern that menthols may be harder to quit (3). Our data suggest this may not be the case for all populations, especially low-income groups with high smoking prevalence.

We have also previously reported (7,13) that risk of lung cancer, one of the primary adverse outcomes of smoking, may be statistically significantly lower among menthol than nonmenthol smokers. As shown in Figure 1 examining lung cancer risks across 10 studies, statistically significantly lower risk of lung cancer is seen among menthol smokers (summary OR = 0.88, 95% CI = 0.81 to 0.96). The 2 most recent studies in our meta-analysis, one from the SCCS (7) and the other from a National Health and Nutrition Examination Survey cohort follow-up (12), both reported nearly 30% lower lung cancer risks among menthol than nonmenthol smokers. Because it is anticipated that should a ban go into effect, most menthol smokers may switch to nonmenthol cigarettes rather than quit, those who do switch may be increasing their risk of this common cancer. Data from Canada, where menthol cigarette bans were initiated in 2017, suggest that only a minority (20%-24%) of menthol smokers will quit smoking (15,16). If the existing epidemiologic data showing lower risk of lung cancer among menthol than nonmenthol smokers generally hold, then in the long term if high percentages of menthol smokers switch to nonmenthols, the ban could have the unintended consequence of a net increase rather than decrease in risk, at least for lung cancer. Such a possibility suggests that should the ban be implemented, it should be accompanied by awareness campaigns to offer menthol smokers smoking cessation support, perhaps akin to mass media campaigns to the network of state quitlines (1-800-QUIT-NOW—offering free counseling and nicotine replacement in all 50 states and some US territories). As noted earlier, we did not collect data on numbers of quit attempts. We also acknowledge that the SCCS provides no information on youth initiation of smoking and preference for menthols, one of the key considerations in menthol ban proposals (3,17). In December 1999, Congress passed and the president signed the Tobacco 21 amendment to the Federal Food, Drug and Cosmetic Act, prohibiting the sale of tobacco products to minors, which should mitigate this concern. Finally, although far beyond the scope of this report, we note that banning a product used primarily by 1 segment of the population (African American people) but not another (White people) also could have adverse enforcement as well as social and practical repercussions that are difficult to predict but are likely to be problematic.

In summary, the data presented herein suggest that one of the rationales (that menthols are harder to quit) for banning menthol cigarettes may not hold in some large segments of the American population. Taken with recent Canadian findings that menthol smokers tend to switch to nonmenthol cigarette brands rather than quit smoking altogether and with previous observations of higher lung cancer risks among nonmenthol smokers, our findings raise the unexpected hypothesis that menthol bans could result in a net long-term increase in cancer risk.

Funding

The Southern Community Cohort Study (SCCS) is funded by grant U01CA202979 from the National Cancer Institute (NCI) at the National Institutes of Health (NIH).

Notes

Role of the funder: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclosures: The authors have no disclosures to report. WJB, who is a JNCI AE and coauthor on this article, was not involved in the editorial review or decision to publish the manuscript.

Author contributions: Conceptualization (WB, MS, WZ); methodology (HM, WW, WB); resources (WB, MS, WZ); formal analysis (HM, WW); writing—original draft (WB); writing—review and editing (All authors); supervision (WB, MS, WZ).

Disclaimer: The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Acknowledgements: We thank Dr Hilary Tindle for her advice and guidance and manuscript review.

Data Availability

Individuals who wish to use data from the Southern Community Cohort Study (SCCS) for research may submit an application to Data and Biospecimen Use Committee for review. The data-sharing policy is described in the study web site www.southerncommunitystudy.org.