Gender differences in relationships between sociodemographic factors and e-cigarette use with smoking cessation: 2014–15 current population survey tobacco use supplement

Abstract

Background

There is conflicting evidence regarding whether men and women are equally likely to quit smoking. We assessed whether gender differences in smoking cessation varied between different sociodemographic groups and across e-cigarette use.

Methods

The 2014–15 cross-section of the Current Population Survey Tobacco Use Supplement was weighted to represent the US adult population of current/former smokers (N = 16 040). Log binomial models tested whether gender modified the relationships between race/ethnicity, education, income or e-cigarette use and 90-day smoking cessation in the past year.

Results

Gender was not associated with cessation in adjusted models (RR = 0.97, CI: 0.85, 1.11). There were no statistically significant interactions between gender and sociodemographic covariates. Current e-cigarette use was associated with higher cessation (RR = 1.53, CI: 1.30, 1.81), and the association varied by gender (Interaction P = 0.013). While male e-cigarette users had a 15% predicted cessation in the past year (CI: 12, 18%), female users had a 9% predicted cessation (95% CI: 7, 11%). Probability of cessation for female e-cigarette users was not different from non-users.

Conclusions

These findings suggest that there are no gender differences in smoking cessation in the USA overall, or by sociodemographic groups. Current e-cigarette use is associated with higher likelihood of recent successful smoking cessation, particularly for men.

Introduction

Tobacco use remains the leading cause of preventable death in the United States,¹ and thus encouraging smoking cessation is key to improving population health outcomes. With smoking prevalence falling faster in men than in women,² it is important to understand the role of gender in smoking behavior and related health outcomes.^3–7 Men and women perceive different risks and benefits to smoking cessation, which can affect motivation to quit.⁸ For example, there is evidence that women expect cessation to bring withdrawal and weight gain more than do men.⁹ In addition, with emerging interest in whether or not electronic cigarette use is related to smoking cessation, it is important to know whether this association differs by gender.

Previous studies have focused on gender differences in short-term cessation due to smoking cessation counseling, nicotine replacement therapies or cessation medications.^10–14 Several studies found women less likely to quit smoking from these interventions,^10–12 while one found no gender difference,¹⁴ and still another found heterogeneity in the treatment-cessation relationship by gender and cigarettes per day.¹³ It is difficult to make generalizations from these inconsistent findings, especially because they use samples not designed to reflect the general population.¹⁵ For example, only sampling those with a desire to quit will not capture differences in motivation that affect overall cessation success. In addition, short follow-up periods may underestimate relapses.

There is contradictory epidemiologic evidence regarding gender differences in cessation at the population level. In their 2016 review, Smith et al.¹⁵ broke down the evidence by study design. The majority of prospective observational studies (31 out of 46) revealed no gender difference, especially with longer follow-up.¹⁵ Cross-sectional evidence was mixed—18 studies found no gender differences, eleven studies found women have lower cessation, and nine studies found women have higher cessation.¹⁵ Notably, the majority of these samples were not nationally representative. The review concluded that women may be slightly less likely to achieve long-term smoking abstinence, and that this pattern could be driven by gender differences in relapse.¹⁵

Demographic, social and behavioral factors may modify the relationship between gender and cessation to explain the conflicting findings. Although both male and female smokers in the United States tend to be disadvantaged with respect to education, income and other measures of social exclusion, there may be important gender differences in how specific sociodemographic characteristics relate to cessation.¹⁶ For example, one study found that financial strain was associated with lower odds of successful cessation among men but not women.¹⁷ Another study noted an interaction between gender and employment: employment was associated with higher cessation for men but lower cessation for women.¹² Regarding the intersection of gender and race, a sample from Minneapolis-St. Paul revealed that more white men than black men had sustained cessation for a week or more, whereas more black women had done so than white women.¹⁸ There has not yet been a nationally representative study in the United States examining how the relationship between gender and cessation may differ by sociodemographic factors.

Considering interactions between gender and sociodemographic characteristics may help explain disparities in cessation and reveal barriers that can be removed to promote social and economic equity.¹⁵ The few previous studies on the sociodemographic variation in gender differences in smoking cessation have not examined the role of e-cigarettes, which may or may not operate like cessation aids. Currently, it is unclear whether e-cigarettes (i) lead to dual use by replacing some but not all cigarettes per day to reduce risk and circumvent smoke-free air laws, or (ii) wean smokers off combustible cigarettes to help them quit all together. One systematic review of 38 studies showed that e-cigarette use was associated with significantly less smoking cessation.¹⁹ However, recent population-based studies have shown that smoking cessation is higher among those using e-cigarettes compared to those who do not,²⁰ and higher frequency e-cigarette use is positively associated with smoking cessation.²¹ Longitudinal evidence is also mixed. Some studies found no difference in cessation after following smokers who did and did not use e-cigarettes.^22,23 Others found cessation was predicted by intensive e-cigarette use²⁴ and long-term e-cigarette use.²⁵ None of these analyses examined whether the association between e-cigarette use and cessation was consistent for men and women.

The purpose of this study is to (i) determine whether smoking cessation differs by gender using a nationally representative sample, (ii) examine whether gender differences in cessation are modified by key sociodemographic factors and (iii) explore if e-cigarette use is associated with cessation differently for male and female smokers. We hypothesize that women will have lower smoking cessation than men. Because female smokers tend to be more socially disadvantaged, we expect that gender differences will be stronger among more disadvantaged groups. We expect e-cigarette use to have a stronger association with cessation for men than women, resembling findings for cessation treatments. This study tests these hypotheses using Current Population Study-Tobacco Use Supplement data from 2014 to 2015. This work will provide baseline national cessation patterns by sociodemographic groups to inform and help evaluate interventions and policies that target long-term cessation.^7,26,27

Methods

Data

This study used Current Population Study-Tobacco Use Supplement (CPS-TUS) data from 2014 to 2015. The US Census Bureau administers the CPS using multiple-stage and multiple-frame probability sampling independently in each state to produce national and state estimates of non-institutionalized individuals.²⁸ Every 3–4 years, the TUS is administered through CPS to all adults (18 or older) within a sampled household.²⁸ This analysis appended the data from the most recent three waves of the TUS: May 2014, August 2015 and January 2015. If respondents were sampled more than once in those three waves, we used their first response. Respondents were included in our sample if they self-reported their survey responses, were age 25 or older (in attempt to capture respondents’ true highest level of education), and smoked this time last year. After excluding 691 respondents who were missing data on key variables, a total of 16 040 respondents remained for analysis. CPS-TUS data were downloaded from International Public Use Microdata Series.²⁹ This dataset did not include the CPS-TUS variables about e-cigarette use, which were downloaded from The National Bureau of Economic Research.³⁰ The two datasets were merged separately for each wave using three variables that together uniquely identify respondents, and then the waves were appended into one dataset. This research uses de-identified publicly available data, and as such, is considered not regulated by the University of Michigan Institutional Review Board.

Variables

The outcome was smoking cessation. An individual was considered to have successfully recently quit smoking if he/she reported smoking at least 100 cigarettes in his/her lifetime, reported smoking 12 months prior to the current survey, reported not smoking at the time of current survey, and had not smoked in the past ninety days.³¹ For a sensitivity analysis, we also examined smoking abstinence for 6 months, rather than 3.

The main independent variable was gender (male/female). We considered sociodemographic variables including age as a continuous variable (top-coded at 85), age-squared, four education categories (less than high school, high school graduate, some college, college graduate or more), mutually exclusive race/ethnicity categories (white Non-Hispanic [hereafter white], black Non-Hispanic [hereafter black], Hispanic, Other Non-Hispanic [hereafter other]), and five categories of household income ($14 999 or less, $15 000–$29 999, $30 000–$49 999, $50 000–$74 999, $75 000 or more). To adjust for nicotine dependency, we included number of years that respondents smoked and average cigarettes smoked per day last year. We have two variables indicating e-cigarette use. The first asks all respondents about current use daily/some days. A subset of respondents with a quit attempt in the past year were additionally asked whether they switched to e-cigarettes when trying to quit smoking cigarettes.

Statistical analysis

For descriptive statistics, we calculated unweighted counts and weighted percentages and means for the complete sample and separately for men and women. We tested differences by gender using adjusted Wald tests.

We estimated a series of regression models predicting 90-day smoking cessation. Relative risks were calculated from generalized linear models using a log link and binomial family. The statistical significance of overall interactions for categorical variables were assessed with adjusted Wald tests. The first set of models examined the interactions between gender and other sociodemographic variables (Table 2). Model 1 examined the unadjusted bivariate relationships between each covariate and cessation, adjusting for survey month. Model 2 mutually adjusted for all variables in the model. Next, we tested for interactions between gender and the sociodemographic covariates: Model 3 tested gender interacting with race/ethnicity, Model 4 tested gender interacting with education, and Model 5 tested gender interacting with income. We adjusted for potential confounding based on sociodemographic patterns documented in the past literature. For example, Model 4 adjusted for race/ethnicity, given racial/ethnic disparities in educational attainment in the USA.³² Model 5 adjusted for race/ethnicity and education, given the relationship between these variables and income.³³

Table 2.

Model results from sociodemographic variables—relative risks of recent cessation (95% confidence interval), CPS-TUS 2014–15

N = 16 040	Model 1	Model 2	Model 3	Model 4	Model 5
Gender
Male	1.00	1.00	1.00	1.00	1.00
Female	0.94 (0.83, 1.08)	0.93 (0.82, 1.06)	0.88 (0.76, 1.02)	0.79 (0.60, 1.02)	1.77 (0.59, 0.99)
Age	0.92 (0.90, 0.95)	0.92 (0.90, 0.96)	0.94 (0.91, 0.97)	0.93 (0.90, 0.96)	0.93 (0.90, 0.96)
Age2	1.00 (1.00, 1.00)	1.00 (1.00, 1.00)	1.00 (1.00, 1.00)	1.00 (1.00, 1.00)	1.00 (1.00, 1.00)
Race/ethnicity
White NH	1.00	1.00	1.00	1.00	1.00
Black NH	0.65 (0.51, 0.83)	0.69 (0.53, 0.89)	0.48 (0.33, 0.71)	0.64 (0.50, 0.83)	0.69 (0.53, 0.89)
Hispanic	0.92 (0.71, 1.19)	0.87 (0.67, 1.13)	0.70 (0.49, 0.99)	0.84 (0.65, 1.09)	0.87 (0.67, 1.13)
Other NH	0.86 (0.61, 1.21)	0.72 (0.51, 1.00)	0.82 (0.53, 1.23)	0.71 (0.51, 0.99)	0.71 (0.51, 1.00)
Female*Race			P = 0.5077
Female*White NH			1.00
Female*Black NH			1.37 (0.83, 2.27)
Female*Hispanic			1.08 (0.64, 1.80)
Female*Other NH			0.75 (0.37, 1.53)
Educ
College or more	1.00	1.00		1.00	1.00
Some College	0.65 (0.55, 0.78)	0.78 (0.65, 0.93)		0.68 (0.53, 0.86)	0.77 (0.65, 0.93)
HS Grad	0.47 (0.39, 0.56)	0.60 (0.50, 0.73)		0.47 (0.37, 0.61)	0.60 (0.50, 0.73)
Less than HS Grad	0.32 (0.24, 0.42)	0.45 (0.33, 0.61)		0.37 (0.25, 0.55)	0.45 (0.33, 0.61)
Female*Educ				P = 0.4321
Female*≥College				1.00
Female*Some College				1.15 (0.82, 1.61)
Female*HS Grad				1.35 (0.94, 1.92)
Female*<HS Grad				1.11 (0.65, 1.90)
Income
$75k or More	1.00	1.00			1.00
$50k–$74 999	0.76 (0.62, 0.93)	084 (0.68, 1.03)			0.81 (0.61, 1.05)
$30k–$49 999	0.68 (0.56, 0.83)	0.80 (0.65, 0.98)			0.71 (0.54, 0.94)
$15k–$29 999	0.50 (0.41, 0.62)	0.64 (0.51, 0.80)			0.57 (0.42, 0.78)
Less than $15 000	0.49 (0.40, 0.60)	0.70 (0.55, 0.88)			0.55 (0.39, 0.78)
Female*Income					P = 0.2541
Female*≥$75k					1.10 (0.74, 1.65)
Female*$50k–$74 999					1.31 (0.90, 1.92)
Female*$30k–$49 999					1.31 (0.86, 2.00)
Female*$15k–$29 999					1.57 (1.03, 2.41)
Female*<$15 000					1.00
Years smoked	0.99 (0.98, 0.99)	1.01 (0.99, 1.02)	1.00 (0.99, 1.01)	1.01 (1.00, 1.02)	1.01 (1.00, 1.02)
Cigs. per day last year	0.97 (0.96, 0.97)	0.97 (0.96, 0.98)	0.97 (0.96, 0.97)	0.97 (0.96, 0.98)	0.97 (0.96, 0.98)
Intercept	–	0.73 (0.28, 1.87)	0.68 (0.34, 1.37)	1.08 (0.52, 2.27)	1.47 (0.69, 3.10)

Note. All models adjusting for study wave; NH = non-Hispanic, HS = high school; P-value for interaction from Adjusted Wald test.

In a second set of models, we assessed the relationships between e-cigarette use, gender and smoking cessation (Table 3). In the first of these models, we examined current e-cigarette use and cessation (Model 1) and tested an interaction between gender and current use (Model 2), adjusting for other sociodemographic covariates. Then, in a subset of the sample, we examined the relationship between successful recent cessation and switching to e-cigarettes when trying to quit (Model 3), then testing the interaction between switching and gender (Model 4). We graphed predicted probabilities for interaction models.

Table 3.

Model results from e-cigarette variables—relative risks of recent cessation (95% confidence interval), CPS-TUS 2014–15, N = 16 040 and 8268

	Model 1	Model 2	Model 3	Model 4
N	16 040	16 040	8268	8268
Gender
Male	1.00	1.00	1.00	1.00
Female	0.93 (0.81, 1.06)	1.01 (0.87, 1.16)	0.90 (0.79, 1.01)	0.95 (0.82, 1.11)
Current e-cig use	1.59 (1.34, 1.87)	1.91 (1.53, 2.39)
Female*current e-cig use		0.66 (0.47, 0.91) P = 0.0126
Switched to e-cigs			0.84 (0.73, 0.96)	0.93 (0.77, 1.12)
Female*Switched to e-cigs				0.81 (0.62, 1.06) P = 0.1267
Intercept	1.21 (0.57, 2.56)	1.11 (0.52, 2.36)	1.94 (0.95, 3.97)	1.84 (0.90, 3.76)

All models adjusting for study wave, age, age-squared, race/ethnicity, education, household income, years smoked and cigarettes per day last year; P-value for interaction from Adjusted Wald test.

All analyses were conducted in 2018 using Stata version 15,³⁴ and survey weights were applied to account for the complex survey design.

Results

Characteristics of sample

Our final sample was 16 040 current and former smokers 25 years or older (Table 1). After applying survey weights, 48.5% of the sample were women, the mean age was around 48 years old, and 72% of respondents were white. There were gender differences in some sociodemographic and behavioral variables in Table 1, but not in successful recent cessation (7.5% for men and 7.1% for women, P = 0.383). In the sensitivity analysis, 3.8% of the sample had recently quit smoking for 6 months. Compared to men, women were more likely to be using e-cigarettes currently (13.1 versus 11.2%, P = 0.003), and to switch to e-cigarettes when trying to quit (36.0 versus 32.5%, P = 0.005).

Table 1.

Sample characteristics by gender, raw counts (weighted percent) and weighted means (SD), CPS-TUS 2014–2015, N = 16 040

Characteristic	Total	Men	Women
Gender	16 040 (100)	7580 (51.55)	8460 (48.45)
Age (25–85)	47.55 (12.22)	47.18 (11.82)	47.94 (12.63)
Race/Ethnicity
White NH	12 379 (71.97)	5754 (69.07)	6625 (75.05)
Black NH	1783 (12.94)	803 (12.92)	980 (12.95)
Hispanic	10 003 (9.58)	552 (11.52)	451 (7.51)
Other NH	875 (5.52)	471 (6.49)	404 (4.49)
Education
<High school	2436 (15.73)	1183 (16.09)	1253 (15.35)
High school grad	5973 (36.16)	2900 (37.04)	3073 (35.23)
Some college	5469 (34.09)	2434 (32.15)	3035 (36.15)
College grad+	2162 (14.01)	1063 (14.72)	1099 (13.27)
Family Income
<$14 999	3757 (23.30)	1564 (20.18)	2193 (26.62)
$15 000–$29 999	3483 (21.15)	1536 (19.83)	1947 (22.56)
$30 000–$49 999	3587 (22.29)	1738 (23.14)	1849 (21.40)
$50 000–$74 999	2563 (16.20)	1337 (18.18)	1226 (14.09)
$75 000+	2650 (17.06)	1405 (18.67)	1245 (15.34)
Cigs. per day last year (1–40)	14.17 (9.42)	14.84 (9.41)	13.46 (9.34)
Years smoked (<1–81)	29.33 (12.70)	29.23 (12.61)	29.42 (12.75)
Recent quit
No	14 900 (92.72)	7024 (92.51)	7876 (92.94)
Yes	1140 (7.28)	556 (7.49)	584 (7.06)
Current e-cig use
No	14 106 (87.87)	6745 (88.75)	7361 (86.92)
Yes	1934 (12.13)	835 (11.25)	1099 (13.08)
Switched to e-cigs
No	5492 (65.71)	2573 (67.45)	2919 (63.98)
Yes	2776 (34.29)	1178 (32.55)	1598 (36.02)

Gender and sociodemographic factors related to cessation

Table 2 shows regression results for the relationship between sociodemographic variables and cessation, with the first models testing unadjusted bivariate relationships and the second model mutually adjusted for all variables. There were no gender differences in cessation in the model adjusting for age, age-squared, race/ethnicity, education, income, years smoked and cigarettes per day last year (Model 2: RR = 0.93; CI: 0.82, 1.06). In this adjusted model, black smokers had a lower likelihood of cessation than whites (Model 2: RR = 0.69; CI: 0.53, 0.89). Cessation was higher with each increasing level of education (test for trend F = 38.05, P < 0.0001) and with each increasing level of income (test for trend F = 13.67, P = 0.0002).

Models 3, 4 and 5 (Table 2) reveal no statistically significant interactions between gender and race/ethnicity, education, or household income in the relationship with cessation. In sensitivity analysis, gender was not associated with 6-month cessation, and none of the interaction terms between gender and sociodemographic variables were statistically significant.

Gender and e-cigarette use related to cessation

In adjusted models, currently using e-cigarettes was associated with higher likelihood of cessation in adjusted models (Table 3, Model 1: RR = 1.59; CI: 1.34, 1.87). There was a gender difference in the relationship between current e-cigarette use and cessation (Model 2: interaction P = 0.0126). Among those who were currently using e-cigarettes, the predicted probability of cessation was significantly higher among men than women (Fig. 1: predicted percent cessation = 15.27; CI: 12.18, 18.38 for men, predicted percent cessation = 9.31; CI: 7.18, 11.43 for women), whereas there were no differences in cessation by gender among those who were not current e-cigarette users. Interestingly, switching to e-cigarettes when trying to quit smoking was associated with lower likelihood of recent cessation (Model 3: RR = 0.84, CI: 0.73, 0.96). While not statistically significant, there was the same gender pattern as current use, with women who switched to e-cigarettes less likely to successfully quit smoking than men who switched and no gender difference between those who did not switch to e-cigarettes (Fig. 1, P-interaction = 0.127).

Fig. 1.

Predicted successful recent cessation for men and women by current e-cigarette use and by switching to e-cigarettes when trying to quit smoking.

When looking at 6-month cessation, current e-cigarette use was associated with higher cessation (RR = 1.46, CI: 1.15, 1.85) and switching to e-cigarettes was marginally statistically significantly associated with lower cessation (RR = 0.82, CI: 0.68, 1.00). The interactions of gender with current and switched e-cigarette use were not statistically significant (P = 0.421 and P = 0.881, respectively).

Discussion

Main findings of this study

This study examined the relationship between gender and smoking cessation, and the extent to which that relationship varied by other sociodemographic characteristics. Examining how sociodemographic characteristics relate to cessation moves beyond individual responsibility for behavior modification by recognizing that smoking may relate to social and economic inequality.²⁶ We found that gender was not associated with cessation, and this relationship was not modified by other sociodemographic factors. However, there were gender differences in 3-month cessation related to current e-cigarette use, such that men currently using e-cigarettes had higher probability of having recently quit than women currently using e-cigarettes.

What is already known on this topic

These results clarify the relationship between gender and smoking cessation in the general US adult population of current and former smokers. The lack of gender differences in cessation supports the conclusion of Smith and colleagues’ review, where the majority of prospective and cross-sectional studies in the general population found no gender difference in quitting smoking, in contrast to trials of cessation treatments.¹⁵ Our results might differ from previous studies because we used 90 days to define cessation, while other studies used shorter or longer time-windows. However, sensitivity analyses using 6-month cessation still showed no relationship between gender and cessation, and no effect modification by sociodemographic variables.

In our study, more women than men reported current e-cigarette use or switching to e-cigarettes when trying to quit smoking. Women’s higher e-cigarette use is consistent with one national probability sample,³⁵ but differs from some other national studies.^20,36 Our gender composition of e-cigarette use is aligned with these latter studies when we include never smokers, who exhibit different e-cigarette use patterns than smokers.²⁰

Current e-cigarette use was associated with increased likelihood of cessation, consistent with some prior national studies,^20,21,36 as well as some longitudinal analyses.^24,25,37 Current use was also associated with cessation in the sensitivity analysis looking at 6-month cessation. Interestingly, the sub-sample analysis of those with recent quit attempts revealed that switching to e-cigarettes when trying to quit smoking was associated with lower likelihood of recent cessation. It could be that people successfully quit without using e-cigarettes and then start using them afterwards, at the time of the survey. Future research should explore the role of intentions and timing of e-cigarette use to clarify its effects on smoking cessation.

What this study adds

We found gender differences in the relationship between current e-cigarette use and cessation: men currently using e-cigarettes were more likely to have recently quit smoking than female current users. It may be that women tend to be dual users while men use e-cigarettes as replacements after recently quitting smoking. Research has shown that many smokers use e-cigarettes to help them quit or reduce cigarette smoking.³⁸ Our results show that women are more likely to switch to e-cigarettes when trying to quit smoking. This highlights a paradox. Women are using e-cigarettes more but e-cigarettes are only associated with cessation for men. It is possible that men receive more nicotine from e-cigarettes than do women based on gender differences in the type of e-cigarette and the way it is smoked, contributing to e-cigarettes’ differential usefulness for cessation.³⁹ Alternatively, it might be that men and women exhibit differences in how accurately they report e-cigarette use and cessation.

Sensitivity analysis examining 6-month cessation did not find a statistically significant interaction between gender and current e-cigarette use. This gender difference may not remain at 6-month cessation, or the small sub-sample that quit for 6 months ( < 4%) lacked the statistical power to detect gender differences in the association. Further research should consider gender differences in cessation for current e-cigarette users, especially before policies operate under assumptions of e-cigarette’s universal usefulness for smoking cessation.

Overall, exploring intersections of gender with other social and behavioral factors can help identify when to tailor tobacco interventions for the specific and diverse needs of subgroups of smokers.⁴⁰ Finding no gender interactions with sociodemographic factors in a nationally representative study is an important contribution to the debate on gender and cessation. Our finding that men current e-cigarette users were more likely to be recent quitters than women current e-cigarette users motivates further examination of the role of gender in how e-cigarettes relate to cessation.

Limitations of this study

Excluding those missing on variables in our analysis resulted in a slightly younger sample and more female respondents. In addition, 3- and 6-month cessation do not guarantee longer-term success in smoking abstinence. The gender relationship with cessation and gender/sociodemographic variable interactions with cessation remained null when examining 12-month cessation, which included a larger sample of all current and former smokers. This sample is not ideal for investigating the relationship between e-cigarettes and 12+ month cessation because e-cigarettes were less common before 2014. Future longitudinal research can help better understand the timing of e-cigarette use and cessation, and can follow up to see who maintained abstinence long-term.

Despite these limitations, using the latest sample of TUS with population weights offered the opportunity to generalize to US adult current and former smokers. Our finding of no gender difference in cessation can be a baseline from which future tobacco control research can examine whether specific treatments, policies and programs lead to gender disparities in cessation. Policymakers should consider the potential for e-cigarettes to be used as cessation aides, while bearing in mind that their use may relate to smoking cessation more in men than in women. Building upon our results, future work in gender and smoking cessation can highlight which policies have the potential to encourage successful smoking cessation in both men and women.

Funding

Dr Fleischer is supported by the National Institutes of Health under Award Number R37CA214787. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.