Proportion of baseline menthol and nonmenthol smokers still smoking at each examination of The Coronary Artery Risk Development in Young Adults Study. AA indicates African American; EA, European American.
Age-adjusted prevalence of coronary calcification among baseline menthol and nonmenthol smokers, The Coronary Artery Risk Development in Young Adults Study, 2000.
Ten-year change in forced expiratory volume in 1 second (FEV1) in relation to interval pack-year exposure to cigarette smoke among baseline menthol and nonmenthol smokers, The Coronary Artery Risk Development in Young Adults Study, 1985-1995.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Pletcher MJ, Hulley BJ, Houston T, Kiefe CI, Benowitz N, Sidney S. Menthol Cigarettes, Smoking Cessation, Atherosclerosis, and Pulmonary FunctionThe Coronary Artery Risk Development in Young Adults (CARDIA) Study. Arch Intern Med. 2006;166(17):1915–1922. doi:10.1001/archinte.166.17.1915
Copyright 2006 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2006
African American smokers are more likely to experience tobacco-related morbidity and mortality than European American smokers, and higher rates of menthol cigarette smoking may contribute to these disparities.
We prospectively measured cumulative exposure to menthol and nonmenthol cigarettes and smoking cessation behavior (1985-2000), coronary calcification (2000), and 10-year change in pulmonary function (1985-1995) in African American and European American smokers recruited in 1985 for the Coronary Artery Risk Development in Young Adults Study.
We identified 1535 smokers in 1985 (972 menthol and 563 nonmenthol); 89% of African Americans preferred menthol vs 29% of European Americans (P<.001). After adjustment for ethnicity, demographics, and social factors, we found nonsignificant trends in menthol smokers toward lower cessation (odds ratio [OR], 0.71; 95% confidence interval [CI], 0.49-1.02; P = .06) and recent quit attempt (OR, 0.77; 95% CI, 0.56-1.06; P = .11) rates and a significant increase in the risk of relapse (OR, 1.89; 95% CI, 1.17-3.05; P = .009). Per pack-year of exposure, however, we found no differences from menthol in tobacco-related coronary calcification (adjusted OR, 1.27; 95% CI, 1.01-1.60 for menthol cigarettes and 1.33; 95% CI, 1.06-1.68 for nonmenthol cigarettes per 10–pack-year increase; P = .75 for comparison) or 10-year pulmonary function decline (adjusted excess decline in forced expiratory volume in 1 second, 84 mL; 95% CI, 32-137 for menthol cigarettes and 80 mL; 95% CI, 30-129 for nonmenthol cigarettes, per 10–pack-year increase; P = .88 for comparison).
Menthol and nonmenthol cigarettes seem to be equally harmful per cigarette smoked in terms of atherosclerosis and pulmonary function decline, but menthol cigarettes may be harder to quit smoking.
Cigarette smoking is common, highly addictive, and harmful, causing approximately 440 000 deaths per year in the United States.1,2 African Americans have disproportionately higher rates of smoking-related illnesses, particularly cancer and cardiovascular disease, despite generally lighter smoking patterns.3,4 For a variety of historical and cultural reasons, including targeted advertising by the tobacco industry,5,6 African American smokers are much more likely to smoke menthol cigarettes than European American smokers (approximately 70% vs 30%).3,7 If menthol cigarettes were more harmful than nonmenthol cigarettes, the higher exposure to menthol cigarette smoke among African American smokers could help explain racial/ethnic disparities in disease rates.4
Menthol is a physiologically active compound that could enhance tobacco smoke–mediated harm by several mechanisms. The monocyclic terpene alcohol, which occurs naturally in plants of the Mentha genus, forms the major constituent of peppermint oil and produces a sensation of coolness and local anesthesia by direct interaction with peripheral nerve endings.8 It also increases transdermal and transmucosal permeability to drugs,8 increases breath holding while smoking,9 decreases the metabolism of nicotine,10 and when burned may directly produce carcinogenic compounds such as benzo[a]pyrene.11 Some clinical studies show higher rates of lung cancer12 and more difficulty quitting13,14 among smokers of menthol cigarettes, but others15- 19 do not show these associations, and no previous studies, to our knowledge, have examined the effects of mentholation on tobacco smoke–mediated atherosclerosis or lung damage.4,20
The Coronary Artery Risk Development in Young Adults (CARDIA) Studyprovides a unique opportunity to evaluate the long-term effects of smoking menthol cigarettes. The CARDIA Study has collected detailed longitudinal data on smoking habits during 15 years of follow-up in a large cohort of young African American and European American men and women in 4 US cities. The CARDIA Study also measured pulmonary function directly on 2 occasions 10 years apart and coronary calcification (a marker for atherosclerosis) at the year 15 examination. To determine whether menthol cigarettes are harder to quit smoking or more harmful than nonmenthol cigarettes, we measured the association between menthol cigarette exposure and smoking cessation behavior, coronary calcification, and pulmonary function decline in CARDIA smokers.
The CARDIA Study is a longitudinal study of risk factors for coronary artery disease in 5115 African American and European American women and men aged 18 to 30 years and healthy at the time of enrollment in 1985. After informed consent was obtained from participants and approval was provided by the institutional review board at each site (Oakland; Chicago, Ill; Minneapolis, Minn; and Birmingham), participants underwent a baseline examination and then follow-up examinations at years 2, 5, 7, 10, and 15, with 74% retention of the surviving cohort at year 15 (2000). Details of the study design, recruitment, and procedures have been published elsewhere.21,22 For this investigation, we identified CARDIA smokers and measured associations between menthol/nonmenthol exposure and smoking cessation behaviors during follow-up, the prevalence of coronary calcification in 2000, and changes in pulmonary function test results between 1985 and 1995.
Current smoking, number of cigarettes smoked per day, and menthol preference (“Is [your current brand of cigarettes] mentholated or nonmentholated?”) were assessed at each CARDIA examination. These data, along with data on past years of smoking at baseline, were used to estimate cumulative exposure to cigarettes in terms of pack-years. We partitioned pack-year exposure into menthol pack-years and nonmenthol pack-years, assuming that participants smoked only menthol or nonmenthol cigarettes at any one time.
The following smoking cessation behaviors were assessed at each examination: not currently smoking, recent quit attempts (“Have you [tried, made any attempts] to quit smoking in the past [2, 3, 5] years?”), and cessation if recent quit attempt (successful smoking cessation among participants who reported a recent quit attempt). We also examined longitudinal patterns of cessation behavior, including sustained smoking cessation (no current smoking the past 2 times they were examined in The CARDIA Study) and documented relapse (baseline smokers who reported no current smoking at a subsequent examination and then current smoking the final time they were examined).
Coronary calcification was measured in consenting CARDIA participants in 2000. Participants underwent computed tomography scanning using an electron beam scanner (GE Imatron C-150; GE Healthcare, Chalfont St Giles, England) or a multidetector scanner (GE LightSpeed; GE Healthcare; or Siemens VZ series; Siemens AG, Munich, Germany). A committee of expert cardiologists, radiologists, and a physicist developed a scanning protocol to standardize scan acquisition across these slightly different technologies. Two scans were obtained for each participant using a hydroxyapatite model for standardization. Scans were electrocardiographically gated at 80% (GE Imatron) or 50% (GE LightSpeed and Siemens VZ) of the R-R interval, with an image thickness of 3 (GE Imatron) or 2.5 (GE LightSpeed and Siemens VZ) mm, and completed within 100 (GE Imatron), 520 (GE LightSpeed), or 360 (Siemens VZ) milliseconds. Specialized image-processing software was used to identify all potential calcific foci composed of at least 4 adjacent pixels (an area ≥1.87 mm2) with a density greater than 130 Hounsfield units. Each potential focus was then confirmed or deleted by a blinded cardiovascular radiologist based on knowledge of coronary artery anatomy, and the presence or absence of coronary calcium was determined. Between- and within-reader reproducibility was high.23
Pulmonary function testing was performed in 1985 and in 1995 using a Collins Survey 8-L water-sealed spirometer and an Eagle II microprocessor (Warren E. Collins Inc, Braintree, Mass). Trained technicians performed frequent calibrations of the instruments and coached participants through 3 to 5 trials of forced expiration to obtain 3 trials meeting minimal quality standards (no clear errors in execution) and showing reproducibility of forced vital capacity (FVC) and forced expiratory volume in 1 second (FEV1) within 5% and 100 mL, respectively. Among the multiple expiratory trials recorded for each participant, we used maximum FEV1, maximum FVC, and maximum midexpiratory flow (MMEF, defined as the average flow between the points at which 25% and 75% of the total FVC has been expired) for analyses per CARDIA protocol. The primary outcome was the change in FEV1 between the 1985 and 1995 examinations; secondary analyses examined change in FVC and MMEF.
Sex, ethnicity, and date of birth were recorded at baseline. Educational grade attained, family income, current alcohol consumption (number of drinks per week), physical activity level (Likert-type scale), and marital, employment, and insurance status were measured by means of self-report at each examination (except income is not available from the baseline examination). Glucose intolerance (defined by the use of diabetes mellitus medication or by a fasting blood glucose level >110 mg/dL [>6.1 mmol/L]), systolic and diastolic blood pressure, plasma levels of low- and high-density lipoprotein cholesterol and triglycerides, and body mass index (calculated as weight in kilograms divided by the square of height in meters) were measured directly.
For the first 3 cessation behavior outcomes (not currently smoking, recent quit attempt, and cessation if recent quit attempt), we performed repeated-measures analyses using logistic models that included each examination of each participant as a separate observation, and we used robust estimates of standard errors to take into account clustering by participant. For the longitudinal cessation pattern analyses (sustained smoking cessation and documented relapse outcomes), we used logistic regression with 1 observation per participant.
We used logistic regression to assess the independent contributions of menthol and nonmenthol exposure to the risk of having coronary calcium. We simultaneously estimated odds ratios (ORs) for each additional 10–pack-year increase in cumulative lifetime exposure to menthol and nonmenthol cigarettes. To compare the putative effects of menthol vs nonmenthol cigarettes, we used Wald tests of the null hypothesis that the coefficients for menthol and nonmenthol exposure were equal.
For the pulmonary function analysis, we used linear regression to assess the independent contributions of menthol and nonmenthol exposure to change in pulmonary function (FEV1, FVC, and MMEF) in the 10 years between 1985 and 1995. We simultaneously estimated coefficients for each additional 10–pack-year increase in the interval exposure (during the same 10 years) to menthol and nonmenthol cigarettes. To compare the putative effects of menthol vs nonmenthol cigarettes, we used Wald tests of the null hypothesis that the coefficients for menthol and nonmenthol exposure were equal. For each regression, we present a series of models sequentially adjusting for demographic and socioeconomic factors, habits, and potential mediators related to each outcome.
Of the original 5115 CARDIA participants, 74% reported for the 2000 examination. To take into account differential dropout we estimated visit-specific probabilities of participating in any given examination using smoking status, age, sex, ethnicity, education, income, and marital status measured at the previous examination. We used these probabilities, inverted, as probability weights in all the models except the sustained smoking cessation and documented relapse analyses (in which the outcome is not tied to a particular CARDIA examination). We also performed unweighted analyses and found nearly identical results in every case. All the analyses were executed using Stata 8.0 statistical software (StataCorp, College Station, Tex).
Of 5115 CARDIA participants enrolled in 1985, 1544 reported current smoking; 972 current smokers (63%) preferred menthol cigarettes and 563 (36%) nonmenthol. Menthol preference was unknown for 9 smokers (0.6%). Menthol preference was stable (11%-12% switched preference during follow-up) and was strongly related to ethnicity, with 89% of African Americans preferring menthol compared with only 29% of European Americans (P<.001). Baseline menthol preference was also associated with younger age, female sex, lower educational level, unemployment, lower alcohol intake, higher body mass index, and fewer cigarettes smoked per day (Table 1).
Baseline menthol smokers were more likely to still be smoking during follow-up examinations than baseline nonmenthol smokers (69% vs 54% in 2000; P<.001). However, stratification by ethnicity attenuates this association, as African Americans are more likely to smoke menthol cigarettes and less likely to quit smoking during follow-up (Figure 1). Before adjustment, menthol smokers were less likely to be noncurrent smokers at follow-up examinations (OR, 0.61; 95% confidence interval [CI], 0.49-0.76); after adjustment, the association is weaker (OR, 0.90; 95% CI, 0.68-1.19) (Table 2).
Among smokers who tried to quit, menthol seemed unrelated to quitting (adjusted OR, 1.00; 95% CI, 0.71-1.42), but menthol was associated with a lower likelihood of trying to quit in the first place (adjusted OR, 0.77; 95% CI, 0.56-1.06; P = .11). In longitudinal analyses, menthol smokers were less likely to exhibit sustained smoking cessation (adjusted OR, 0.71; 95% CI, 0.49-1.02; P = .06) and nearly twice as likely to relapse after an examination during which they reported no current smoking (adjusted OR, 1.89; 95% CI, 1.17-3.05; P = .009) (Table 2). Results were similar among African Americans and European Americans and after additional adjustment for cigarettes smoked daily at baseline.
The prevalence of coronary calcification (14% overall in these smokers) was strongly associated with cumulative exposure to tobacco smoke, but the association seemed to be equivalent among baseline menthol smokers and nonmenthol smokers (Figure 2). Logistic models showed no difference in the associations between menthol and nonmenthol exposure and coronary calcification before or after adjustment for demographics, socioeconomic status, or other habits. Further adjustment for physiologic mediators of coronary disease led to attenuation in each smoking–coronary calcium association consistent with partial mediation by those measured factors (Table 3). Results were similar in African Americans and European Americans.
The FEV1 declined by an average of 180 mL between 1985 and 1995 in CARDIA smokers, and the magnitude of decline was strongly associated with interval exposure to tobacco smoke among menthol and nonmenthol smokers (Figure 3 and Table 4). Regression models showed no detectable difference in the magnitude of association between menthol- and nonmenthol-associated decline in FEV1, FVC, or MMEF before or after adjustment. European Americans tended to have larger declines associated with menthol cigarettes (P = .05 for FEV1), whereas African Americans showed the opposite tendency (P = .35 for FEV1).
In this longitudinal analysis of smoking behavior across 15 years, we looked for differences between menthol and nonmenthol smokers in smoking cessation behavior, coronary calcification, and pulmonary function decline. We found some evidence that menthol smokers are less likely to attempt cessation, more likely to relapse after successfully quitting, and less likely to report sustained smoking cessation than nonmenthol smokers. Per pack-year of exposure, however, we found no evidence that menthol cigarettes are more harmful than nonmenthol cigarettes; both types of cigarettes were equally associated with coronary atherosclerosis and accelerated pulmonary function decline.
The finding that menthol smokers are less likely to quit and more likely to relapse is consistent with some but not all previous studies. One cross-sectional study of African American patients demonstrated more recent relapse and a trend toward shorter duration of cessation during the last quit attempt in menthol vs nonmenthol smokers, “suggest[ing] that African-American menthol smokers are less successful with smoking cessation.”24(p1208) Two other articles13,14 by the same group analyzed quit rates among African American smokers after enrollment in a randomized trial of bupropion for smoking cessation and found a crude association between menthol preference and lower cessation at 6 to 7 weeks that was more pronounced among persons assigned to active bupropion treatment and that persisted after adjustment among patients younger than 50 years (OR, 2.02; 95% CI, 1.03-3.95). An analysis of the Community Intervention Trial for Smoking Cessation (COMMIT),19 however, found no association between menthol and quit rates at 5 years after adjusting for ethnicity, and a case-control study25 designed to examine risk factors for cancer found no relationship between current (vs former) smoking and menthol preference. Menthol inhibits the metabolism of nicotine,10 leading to higher levels of nicotine per cigarette smoked, and menthol smokers tend to show higher levels of addiction in terms of time to first cigarette in the morning13 despite lower volume and frequency of puffs26 and lower numbers of cigarettes smoked per day seen in the present study and others.7,25,26 These findings suggest that menthol cigarettes may be harder to quit smoking, but uncertainty about this point remains, in part because of the difficulty and large sample sizes required to tease apart the effects of ethnicity and menthol preference, which are highly correlated.
Smoking is known to be a major risk factor for coronary artery disease,27- 29 and the present findings demonstrate a strong dose-response relationship with coronary calcification, a marker of atherosclerosis. Per cigarette, however, menthol and nonmenthol exposure seem to be equally harmful. Previous studies of menthol vs nonmenthol cigarette smoking show no effects of mentholation on blood pressure30 or heart rate,30,31 although one study32 showed a small difference in heart rate associated with menthol candy/tea ingestion. We are unaware of any previous studies of menthol and atherosclerosis. Although the present study does not rule out a difference in thrombosis or other nonatherosclerotic mechanisms leading to coronary events, we found no evidence that it plays a role in coronary heart disease disparities between African Americans and European Americans.
Pulmonary function decline and obstructive lung disease are also known to be strongly associated with tobacco smoke exposure,33,34 and the present study demonstrated this dose-response relationship. Per cigarette, however, menthol and nonmenthol exposure again seem to be equally harmful. There are few published studies of menthol and pulmonary function and no previous longitudinal studies of menthol and pulmonary function in humans, to our knowledge. One tobacco industry–funded study35 of rats exposed to menthol and nonmenthol cigarette smoke for 13 weeks showed histopathologic changes consistent with smoking that seemed to be equivalent between menthol- and nonmenthol-exposed rats. Another industry document36,37 hints at “an adverse effect on the respiratory function” associated with mentholation of cigarettes without providing details, but several small trials of menthol vapor inhalation (not in a cigarette) suggest improved mucociliary clearance,38 less airway reactivity,39 fewer wheezing episodes and less need for bronchodilator dosing,39 bronchodilation,40 and “easier breathing”41,42 among patients with asthma,39 chronic obstructive lung disease,38 or acute upper respiratory tract illness.40- 42 Menthol seems unlikely to be a contributor to the pulmonary function decline associated with tobacco smoke exposure.
This study is limited somewhat by sample size, particularly when we attempt to tease apart the effects of ethnicity and menthol preference. The limited numbers of European American menthol smokers (n = 189) and African American nonmenthol smokers (n = 95) make ethnicity-specific analyses and (to a lesser extent) adjusted analyses somewhat imprecise. Inherent random variation and measurement error also limit precision and bias measures of association toward the null so that we could have missed small differences between menthol and nonmenthol cigarettes. Loss to follow-up in this cohort may theoretically bias results, but selection bias does not occur with out differential effects simultaneously by menthol preference and the outcome, and sensitivity analyses show essentially no differences between analyses with and without inverse probability weighting despite the fact that we could predict dropout with some accuracy. Some investigators43 have raised concerns that individuals may switch to menthol cigarettes when they develop preclinical disease. This could theoretically bias the coronary calcification or pulmonary function results (but not the cessation results, which use only baseline menthol preference); however, this tendency would lead to a bias toward finding more harm from menthol, which is essentially a conservative bias in relation to these conclusions. There is also a concern that other cigarette characteristics (tar, nicotine content, or other additives) could be different in menthol cigarettes and thereby confound our results; we cannot adjust directly for these other characteristics owing to limited data in The CARDIA Study. We note, however, that the most popular menthol (Newport and Kool) and nonmenthol (Marlboro and Camel) cigarette brands44 are similar at least in terms of tar and nicotine content. Finally, this analysis does not address lung cancer, nonatherosclerotic heart disease, or other potential harms from cigarette smoking that may be facilitated or amplified by the presence of menthol. Ischemic heart disease and chronic airway obstruction, however, are the 2 most common causes of smoking-attributable mortality in the United States after lung cancer.1
In summary, the preference for mentholated cigarettes among US smokers is highly associated with ethnicity and seems to be relatively stable across time. Mentholation of cigarettes does not seem to explain disparities in ischemic heart disease and obstructive pulmonary disease between African Americans and European Americans in the United States but may partially explain lower rates of smoking cessation among African American smokers. It is possible, therefore, that switching from menthol cigarettes to nonmenthol cigarettes might facilitate subsequent smoking cessation, especially in African Americans, and thereby reduce tobacco-related health disparities. At a policy level, regulation of tobacco additives, such as menthol, has been proposed as a way to reduce tobacco addiction in the United States45 and as one step in a long-term strategy designed to replace tobacco with “clean” sources of nicotine46 to reduce the health consequences of nicotine addiction in the United States. Our results provide some support for this strategy, although the primary goal of public health officials, physicians, and patients should be to reduce all tobacco smoke exposure regardless of menthol content.
Correspondence: Mark J. Pletcher, MD, MPH, Departments of Epidemiology and Biostatistics and Medicine, University of California, San Francisco, 185 Berry St, Lobby 4, Suite 5700, San Francisco, CA 94143-0560 (email@example.com).
Accepted for Publication: May 25, 2006.
Author Contributions: Dr Pletcher had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analyses.
Financial Disclosure: None reported.
Funding/Support: The CARDIA Study is supported by contracts N01-HC-48047, N01-HC-48048, N01-HC-48049, N01-HC-48050, and N01-HC-95095 from the National Heart, Lung, and Blood Institute.
Role of the Sponsor: The CARDIA Study Publications and Presentations Committee and the National Heart, Lung, and Blood Institute reviewed and approved this manuscript, and the statistical analyses were verified independently by the CARDIA Coordinating Center.
Previous Presentation: An abstract containing a preliminary analysis of these data was presented at the American Heart Association's 45th Annual Conference on Cardiovascular Disease Epidemiology and Prevention; May 2, 2005; Washington, DC.