Cigarettes, Tobacco Dependence, and Smoking Cessation: Project MOM Final Report Lorraine R. Reitzel, Ph.D., The University of Texas, MD Anderson Cancer Center Background and Analysis Objectives Background. Secondary analyses of data are needed to assist the Tobacco Product Scientific Advisory Committee with their recommendations to the Food and Drug Administration (FDA) regarding the impact of menthol in cigarettes on public health. Study Questions and Hypotheses. This study was designed to answer the following two questions among a sample of racially/ethnically diverse pregnant women who quit smoking due to pregnancy and who wanted to remain quit postpartum: 1. Is menthol cigarette use associated with tobacco dependence? and 2. Does menthol cigarette use predict postpartum smoking abstinence? These analyses were exploratory in nature and not guided by specific hypotheses. Methods and Approach Study Population. Project MOM was a randomized clinical trial designed to test the efficacy of a motivationally-based treatment for smoking relapse prevention. Participants were 1 racially/ethnically diverse adult women (32% Black, % Latina, 36% White, 2% Other) from the Houston, Texas metropolitan area who were in their th 33 rd week of pregnancy at the time of enrollment. Participants stopped smoking either during their pregnancy or within 2 months prior to becoming pregnant and wanted to remain quit postpartum. Participants reported smoking an average of at least 1 cigarette daily for the year prior to pregnancy. Women reporting a high-risk pregnancy were excluded. Data Source. Participants were proactively recruited from within the Houston metropolitan area through a local health care system, and via newspaper, radio, bus, and clinic advertisements. Participants were enrolled from 7. Data were collected at enrollment and Weeks 8 and 26 postpartum. Questionnaires were administered and completed via computer. Data collection was completed by 9. The outcome paper (Reitzel et al., ) contains information about participant attrition and flow through the study. Measures. Our Independent Variable (IV) for both study questions was a single item, administered at enrollment that asked the participants if their regular brand of cigarettes was menthol or non-menthol. Results yielded a binary predictor variable (non-menthol vs. menthol; reference group = non-menthol). Two participants did not answer this item, and missing values were maintained in analyses. For study question 1, the Dependent Variable (DV) was tobacco dependence. The primary measure of tobacco dependence was the Wisconsin Inventory of Smoking Dependence Motives (WISDM-68; Piper et al., 4). The WISDM-68 is a comprehensive, multi-dimensional measure of dependence that yields an overall smoking dependence score (WISDM-68 total score) as well as subscale scores for critical dimensions of dependence, including non-physical indices of dependence [e.g., Affiliative Attachment, Automaticity, Social/Environmental Goads; (Piper et al., 4)]. Higher scores on the WISDM-68 are indicative of greater tobacco dependence. Secondary measures of tobacco dependence included single items assessing the: 1) number of cigarettes smoked per day prior to quitting, and 2) the time to the first cigarette of the day prior to quitting (< minutes vs. > minutes; reference group = < minutes). All dependence variables were collected at enrollment. For study question 2, the DV was continuous abstinence from smoking, which was defined as a self-report of no smoking (not even a puff) since the delivery date and biochemical verification at 8 and 26 weeks postpartum through expired carbon monoxide levels < ppm (Hajek et al., 1) and/or a cotinine value of < ng/ml (McBride et al., 1999). The percentage of participants with missing smoking status at these time points was 21% (week 8) and 28% (week 26). However, an intention-to-treat procedure was followed, whereby those lost to follow-up
were considered relapsed. Thus, there were effectively no missing data for this variable. Because the focus was on continuous abstinence, relapse at Week 8 automatically resulted in classification as relapsed at Week 26. Unadjusted and adjusted analyses were performed for each study question. Covariates relevant to study question 1 were: participant age, race/ethnicity (Hispanic, Black, White, Other), partner status (single/widowed/divorced vs. married/living with partner), income (<$,/year vs. >$,/year annual household income) and educational achievement (< high school/ged vs. > high school/ged). Covariates relevant to study question 2 were those indicated above plus the treatment group assignment, number of cigarettes smoked per day, and time to the first cigarette of the day (< minutes or > minutes). All covariates were collected at enrollment. The only covariate with missing data was income (n = 27 missing), which were maintained in the adjusted analyses. 1 Model Specification and Analysis. All analyses were completed in SAS, version 9.1. Because analyses were largely exploratory and not guided by specific hypotheses, the threshold used for statistical significance was p<. (i.e., no corrections were made for multiple analyses). For Study Question 1, the association between menthol use and tobacco dependence was assessed in regression-based models (SAS GLM for continuous outcomes and SAS LOGISTIC for the binary outcome). For Study Question 2, the effect of menthol use on continuous smoking abstinence was assessed using continuation ratio (CR) logit models (PROC GENMOD). CR logit models are appropriate when ordered categories (e.g., relapsed at week 8, abstinent at week 8 but relapsed at week 26, and abstinent through week 26) represent a progression through stages (Agresti, 2; Bender & Benner, ; McGowan, ). The CR logit models operate by modeling the conditional probability of being abstinent at the current assessment point given that a participant has been abstinent through the most recent assessment point. Findings Participant characteristics by menthol use status are in Table 1. users were younger than non-menthol users. There were also significant differences between menthol and nonmenthol users by race/ethnicity, partner status, income, and educational achievement. These variables, among others, were included as covariates in adjusted models. Study Question 1. cigarette use was not significantly associated with the WISDM-68 total score in unadjusted analyses (β = -.4, SE =.; t = -.28; p =.78; n = 249) or in analyses adjusted for age, race/ethnicity, partner status, income, and educational achievement (β =., SE =.19; t =.24; p =.81; n = 222). We also examined the association between menthol use and each of the 13 subscales of the WISDM-68 to determine if menthol use was significantly associated with any of the specific components of tobacco dependence. In unadjusted analyses, menthol cigarette use was significantly associated with WISDM-68 Cue Exposure/Associative Processes (β = -.2, SE =.21; t = -2.; p =.1; n = 249) and Tolerance (β =.38, SE =.; t = 1.93; p =.; n = 249). Effect sizes, calculated using Cohen s d, were considered small (Cue Exposure/Associative Processes = -.31; Tolerance =.24). Specifically, menthol cigarette use was associated with less dependence in response to cue exposures/associative processes, but more tolerance-related dependence, relative to nonmenthol use. However, these significant associations were not maintained in adjusted analyses. cigarette use was significantly associated with the number of cigarettes smoked per day in unadjusted analyses (β = -1.91, SE =.96; t = -1.98; p =.; n = 249; effect size = -.), but not adjusted analyses (β = -.38, SE = 1.; t = -.33; p =.74; n = 222). Finally, menthol cigarette use was significantly associated with the time to the first cigarette of the day in unadjusted analyses [β = -.31, SE =.16; χ 2 = 3.86; p =.; n = 249; OR =.73 (.4-1.)], but not adjusted analyses (β = -.19, SE =.21; χ 2 = -.84; p =.36; n = 222). 1 Please note that analyses were also performed that excluded income as a covariate due to excessive missing values for this variable. However, the pattern of results remained unchanged.
Table 1. Participant Characteristics by Use Status: Project MOM Covariates/Outcome (n=1) Non- (n=124) Total (N=249) M(SD)/n(%) M(SD)/n(%) M(SD)/n(%) Age* 23.6 (4.6).7 (.8) 24.7 (.3) Race/Ethnicity** Latina 33 (26.4) 42 (33.9) 7 (.1) Black 7 (6.) 11 (8.9) 81 (32.) Other 2 (1.6) 3 (2.4) (2.) White (16.) 68 (4.8) 88 (.3) Partner Status* Single/Divorced/Widowed 7 (.6) 36 (29.) 93 (37.3) Married/Living with Partner 68 (4.4) 88 (71.) 6 (62.7) Annual Household Income* < $,/year 48 (43.2) 26 (23.4) 74 (33.3) >= $,/year 63 (6.8) 8 (76.6) 148 (66.7) Educational Achievement* < High school/ged 29 (23.2) 16 (12.9) (18.1) >= High school/ged 96 (76.8) 8 (87.1) 4 (81.9) WISDM-68 Total 2.7 (1.2) 2.8 (1.2) 2.7 (1.2) Affiliative attachment 1.8 (1.2) 1.8 (1.2) 1.8 (1.2) Automaticity 2.7 (1.) 2.6 (1.) 2.7 (1.) Loss of control 2. (1.2) 2.2 (1.) 2.1 (1.4) Behavioral choice/melioration 2. (1.1) 2. (1.1) 2. (1.1) Cognitive enhancement 2.4 (1.6) 2.3 (1.) 2.4 (1.) Craving 2.8 (1.6) 2.7 (1.6) 2.8 (1.6) Cue exposure/associative processes* 3.1 (1.6) 3.6 (1.7) 3.4 (1.7) Negative reinforcement 3. (1.7) 3.6 (1.8) 3. (1.8) Positive reinforcement 2.8 (1.) 2.9 (1.6) 2.8 (1.6) Social/environmental goads 4.1 (2.1) 4.1 (2.) 4.1 (2.1) Taste/sensory processes 3. (1.6) 3. (1.7) 3. (1.6) Tolerance* 2.6 (1.6) 2.3 (1.) 2.4 (1.6) Weight control 2.3 (1.8) 2.6 (1.9) 2.4 (1.9) Cigarettes per day* 9.2 (6.7) 11.1 (8.4).2 (7.6) Time to first cigarette* > minutes 92 (73.6) 4 (83.9) 196 (78.7) <= minutes 33 (26.4) (16.1) 3 (21.3) Note: ANOVAs or chi-square tests were used to evaluate significant differences; * p <., ** p <.1 Study Question 2. Unadjusted continuous abstinence rates by menthol cigarette use status are displayed in Figure 1. users had lower rates of continuous abstinence than nonmenthol users at both follow-up points. However, menthol cigarette use did not significantly predict continuous abstinence from smoking in analyses adjusted for time and treatment group (β = -.17, SE =.23; χ 2 =.4; p =.46; n = 338), or in analyses adjusted for age, race/ethnicity,
partner status, income, and educational achievement, time, treatment group, cigarettes smoked per day, and time to the first cigarette of the day (β = -.19, SE =.; χ 2 =.42; p =.2; n = 4). Figure 1: Continuous Abstinence Rates by Use Status: Project MOM Post-hoc racial/ethnic subgroup analyses Because racial/ethnic differences in the relationships between: 1) menthol cigarette use and tobacco dependence, and 2) menthol cigarette use and smoking cessation were of particular interest to the Tobacco Product Scientific Advisory Committee, RTI International requested that we conduct post-hoc racial/ethnic subgroup analyses with Project MOM data. As such, we reran the analyses described in detail above for each racial/ethnic group separately (with the exception of the Other race/ethnic group, which had only participants). We were asked to take this approach to analyses rather than including an interaction term in the original models. These post-hoc analyses yielded a few significant findings. First, menthol use was significantly associated with the Craving subscale of the WIDSM-68 among Black women in the sample, but only in unadjusted analyses (β =.96, SE =.47; t = 2.4; p =.44; n = 81; effect size =.6). Black menthol users endorsed more Craving than Black non-menthol users (mean score = 2.82 versus 1.86). Second, menthol use was significantly associated with the number of cigarettes smoked per day among Black women in the sample, while controlling for age, partner status, income and educational level (β = 3.82, SE = 3.77; t = 2.31; p =.2; n = 71; effect size =.67). Black menthol users smoked more cigarettes per day than Black non-menthol users (8 versus cigarettes per day, on average). Third, results indicated that menthol cigarette use predicted continuous smoking abstinence among White women in unadjusted analyses [β = - 1.73, SE =.67; χ 2 = 6.63; p =.1; n = 1, OR =. (. -.)] and in analyses adjusted for age, partner status, income, and educational achievement, time, treatment group, cigarettes smoked per day, and time to the first cigarette of the day [(β = -1.62, SE =.76; χ 2 = 4.49; p =.3; n = 8, OR =.19 (.4 -.89)]. White menthol users were less likely to maintain continuous abstinence through postquit week 26 than White non-menthol users. Unadjusted continuous
abstinence rates by menthol cigarette use status for each racial/ethnic group (except Other race/ethnicity) are displayed in Figures 2a, 2b, and 2c. Figure 2a. Continuous Abstinence Rates by Use Status: Project MOM White Participants (N=88) Figure 2b. Continuous Abstinence Rates by Use Status: Project MOM Black Participants (N=81)
Figure 2c. Continuous Abstinence Rates by Use Status: Project MOM Latina Participants (N=7) Conclusions The purpose of this study was to examine the relationships between: 1) menthol cigarette use and tobacco dependence, and 2) menthol cigarette use and smoking cessation in Project MOM, a study comprised of a racially/ethnically diverse sample of pregnant/postpartum women who quit smoking as a result of pregnancy and who wanted to remain quit postpartum. Strengths of this study included the ability to adjust for several important covariates in evaluating the relationships of interest. Results indicated that although menthol cigarette use was significantly associated with some, but not all, indicators of tobacco dependence in unadjusted analyses, these significant associations were not maintained in adjusted analyses. Moreover, although menthol cigarette users had lower rates of continuous abstinence from smoking at both follow-up assessments, these differences were not significant in either unadjusted or adjusted analyses. Thus, our well-controlled analyses failed to support that menthol cigarette use significantly affected tobacco dependence or continuous smoking abstinence among our sample of pregnant/postpartum women. Results, however, should be considered in light of limited power to detect very small effects (e.g., % estimated power to detect an effect size of. for analyses involving WISDM-68, 22% estimated power to detect an odds ratio of 1.3 for analyses involving continuous abstinence). If very small effects were deemed clinically meaningful, a larger sample size would be needed to achieve adequate power to detect a true effect. The menthol and non-menthol users in Project MOM differed significantly from one another in age, race/ethnicity, partner status, income, and educational achievement. These sociodemographic variables might be important moderators to examine in future studies. Because racial/ethnic differences in the outcomes were of particular interest to the Tobacco Product Scientific Advisory Committee, we performed post-hoc analyses by racial/ethnic subgroup. Results of adjusted dependence analyses indicated significant differences in the number of (pre-quit) cigarettes smoked per day among Black menthol versus non-menthol users, with the Black menthol users smoking more cigarettes per day. The number of cigarettes smoked per
day is positively associated with tobacco dependence. Thus, results would suggest that Black menthol users were more dependent on tobacco than Black non-menthol users. However, it is important to note that cigarettes smoked per day is only one indicator of tobacco dependence, and that we failed to find significant differences between the groups on several other indicators. Results of post-hoc cessation analyses indicated that White menthol users were less likely to maintain continuous smoking abstinence than White non-menthol users. However, it should be noted that some cell sizes were small in these post-hoc analyses (i.e., n = White menthol users), so these results should be replicated with larger samples. It was also interesting to note that, unlike the White and Black women, Latina menthol users had higher rates of continuous abstinence than Latina non-menthol users. This pattern, although non-significant and in need of replication, suggests that future studies might examine racial/ethnic interactions and/or conduct racial/ethnic subgroup analyses when examining the effects of menthol use on cessation. Also, we should note that continuous smoking abstinence is a relatively conservative measure of post-delivery smoking status, which does not forgive lapses following the delivery date. Results using other measures of abstinence may yield different results. Limitations of this study include that participants were treatment-seeking and self-referred. Therefore, results may not generalize to non-treatment seeking pregnant/postpartum women, or to pregnant/postpartum women who do not volunteer for relapse prevention studies. In Project MOM, smokers had quit by the time of enrollment ( -33 rd week of pregnancy) in the study. Results may not generalize to women who smoke later into their pregnancies. Finally, Project MOM was conducted in the Houston, Texas metropolitan area and results may not generalize to pregnant/postpartum women from other locations, including rural areas. Funding Sources for the Collection of Primary Data Used for Secondary Data Analysis Project MOM was supported by a grant from the National Cancer Institute (R1CA89; PI; David W. Wetter). The title of this grant is Smoking Relapse Prevention Among Postpartum Women. References Agresti, A. (2). Categorical Data Analysis, 2nd edition. Hoboken, New Jersey: John Wiley & Sons, Inc. Bender, R., & Benner, A. (). Calculating ordinal regression models in SAS and S-Plus. Biometrical Journal, 42, 677-699. Hajek, P., West, R., Lee, A., Foulds, J., Owen, L., Eiser, J. R., et al. (1). Randomized controlled trial of a midwife-delivered brief smoking cessation intervention in pregnancy. Addiction, 96, 48-494. McBride, C. M., Curry, S. J., Lando, H. A., Pirie, P. L., Grothaus, L. C., & Nelson, J. C. (1999). Prevention of relapse in women who quit smoking during pregnancy. American Journal of Public Health, 89, 76-711. McGowan, M. J. (, September). Ordinal outcomes with the continuation ratio model. Paper presented at the Proceedings of the Northeast SAS Users Group Conference. Philadelphia, PA. Piper, M. E., Piasecki, T. M., Federman, E. B., Bolt, D. M., Smith, S. S., Fiore, M. C., et al. (4). A multiple motives approach to tobacco dependence: the Wisconsin Inventory of Smoking Dependence Motives (WISDM-68). Journal of Consulting and Clinical Psychology, 72, 139-4. Reitzel, L. R., Vidrine, J. I., Businelle, M. S., Kendzor, D. E., Costello, T. J., Li, Y., Daza, P., Mullen, P. D., Velasquez, M. M., Cinciripini, P. M., Cofta-Woerpel, L., & Wetter, D. W. (). Preventing postpartum smoking relapse among diverse, low income women: A randomized clinical trial. Nicotine & Tobacco Research, 12, 326-3.