The time from waking in the morning until the first cigarette of the day is strongly associated with nicotine use behaviors and has increasingly been used as a single-item measure of dependence in a range of smoking studies, including clinical trials, laboratory studies, and other investigations of cigarette use [
Our understanding of TTFC as a physiological indicator of nicotine addiction may be changed if, in fact, environmental factors play a large role in driving this behavior. Indeed, the social influences of tobacco use are considerable and are impacted by proximity and interaction with other smokers [
Regardless of the impetus for the creation of HSR, the presence of such rules may impact the measurement of TTFC and, consequently, TTFC’s predictive validity on tobacco dependence. When examining the relation between HSR and TTFC, it is important to consider the contextual factors which may help resolve whether if TTFC is an independent measure of nicotine dependence or simply a correlate of environmental or social factors. For example, it is possible that the physical limitations placed on smokers in households with restrictions (e.g., getting out of bed, getting dressed, and moving to a designated smoking location) may push back the TTFC to a later category (e.g., from having the first cigarette within 5 minutes to having the first cigarette within 10 minutes). However, this may not affect the urges or cravings to smoke and may not affect the number of cigarettes smoked per day. The nature of the HSR is another important element to consider, as not all HSR are full bans on smoking in the home. Some households have partial bans, where smoking is allowed in some areas of the home but not in others. Others may have bans on combustible cigarettes, but not electronic nicotine delivery systems (e.g., e-cigarettes and JUUL). These partial bans may not impact the TTFC, if such an association exists, in a manner similar to a full ban.
Few studies have examined the impact of HSR on nicotine dependence [
Data for this study come from the Pennsylvania Adult Smoking Study (PASS), a study of nicotine dependence and smoking behaviors conducted between 2012 and 2014 [
Participants provided information on age, gender, race/ethnicity, educational attainment, marital status, and household income.
Participants reported on their tobacco use history by responding to items from the Consensus Measures of Phenotypes and Exposures (PhenX) toolkit (version 5.1, March 23, 2012). Items include age started smoking, number of cigarettes per day, and nicotine dependence as measured by the Fagerstrom Test for Nicotine Dependence (FTND; [
Participants indicated the number of people living in their home and how many underage children (<18 years) lived in the home. Participants also indicated if they had rules regarding smoking in the household. Household smoking rules (HSR) were characterized as follows: (1) full ban, smoking not allowed anywhere in the home; (2) partial ban, smoking is allowed some places or sometimes within the home; and (3) no ban, smoking is allowed anywhere in the home/there are no “rules” regarding smoking in the home.
All analyses were completed using SPSS, Version 25 (IBM Corp, Armonk, NY). Initial bivariate analysis includes Chi-square tests comparing HSR categories with TTFC categories. ANOVA with post hoc comparisons were conducted to determine differences in demographic and smoking behavior measure by HSR category. Linear regression analysis was conducted to determine predictors of TTFC, followed by multiple mediation analyses to examine plausible relations between variables.
We used multiple mediation analyses and bootstrapping methods with bias-corrected confidence intervals for all pathway models [
Multiple mediation model.
The data used in the current analyses are cross-sectional, thereby limiting the interpretation of causality for the mediation analysis. It may be that implementing HSR affects smoking behaviors such as CPD and levels of addiction. Alternatively, smokers who smoke relatively fewer cigarettes or are less addicted may be more likely to implement HSR. Thus, we tested an alternate model to examine the model fit predicting HSR from TTFC, urges to smoke, and cigarettes per day, (Figure
Alternative multiple mediation model.
The models controlled for age (linear), gender (categorical), education (linear), total family income (linear), marital status (categorical), number of underage children in the home (linear), total number of people living in the home (linear), and age started smoking regularly (linear). Household smoking rules were dummy coded as (1) full ban vs. partial or no ban, (2) partial ban vs. full or no ban, and (3) no ban vs. partial or full ban. Given that multiple models were conducted, we controlled for potential Type II errors using false discover rate (FDR) corrections of
All data were initially screened to insure they met assumptions of normality of distribution and to examine for patterns of missingness. TTFC and total family income demonstrated significantly skewed values and were therefore normalized using log transformations. Log-transformed values were used in the analyses whereas actual, pretransformed values are reported in descriptive tables for interpretability. No variables were missing more than 5% of the data, and therefore, no further adjustments were made [
Sample descriptive statistics.
Full smoking ban | Partial smoking ban | No smoking ban | |||||||
---|---|---|---|---|---|---|---|---|---|
Male | Female | Total | Male | Female | Total | Male | Female | Total | |
Age | 33.7 (10.2) | 35.1 (10.4) | 34.5 | 38.1 (11.7) | 38.1 (11.1) | 38.1 (11.2) | 43.3 (11.3) | 44.5 (12.5) | 44.2 (12.1) |
Education | 15.4 (1.9) | 15.9 (2.1) | 15.6 (2.1) | 15.2 (2.1) | 15.3 (1.2) | 15.3 (1.2) | 14.7 (1.9) | 14.6 (2.1) | 14.6 |
Household income | 64k (38k) | 60k (43k) | 62k (41k) | 61k (42k) | 54k (29k) | 57k (36k) | 45k (28k) | 42k (30k) | 43k |
Underage children | 1.2 (1.3) | .98 (1.2) | 1.1 | .78 (.95) | 1.1 (.94) | .93 (.95) | .76 (1.1) | .60 (.84) | .70 (.94) |
Urges to smoke | 4.86 (2.53) | 5.60 (2.97) | 5.28 | 6.41 (2.57) | 6.53 (2.08) | 6.50 (2.32) | 6.77 (2.50) | 6.25 (2.58) | 6.44 (2.55) |
No. of people in household | 3.6 (1.5) | 3.2 (1.4) | 3.3 (1.5) | 3.2 (1.5) | 3.3 (1.1) | 3.3 (1.3) | 2.7 (1.6) | 3.1 (1.8) | 3.0 (1.7) |
Cigarettes per day | 14.7 (7.4) | 14.5 (7.9) | 14.6 | 18.8 (8.1) | 16.2 (7.1) | 17.5 (7.7) | 21.3 (9.0) | 18.0 (8.4) | 19.1 (8.7) |
Time to first cigarette | 32.8 (23.1) | 33.6 (32.3) | 33.2 | 17.5 (21.2) | 22.7 (27.1) | 20.2 (24.5) | 12.3 (14.2) | 17.0 (22.9) | 15.3 (20) |
HONC score | 6.6 (2.4) | 7.4 (2.1) | 7.1 (2.2) | 7.2 (2.2) | 7.7 (1.8) | 7.4 (2.0) | 7.0 (2.0) | 7.6 (1.9) | 7.4 (2.0) |
Age started smoking | 17.1 (4.1) | 16.9 (4.4) | 16.9 (4.3) | 16.9 (4.8) | 16.6 (3.7) | 16.8 (4.3) | 17.4 (6.5) | 16.1 (4.8) | 16.5 (5.5) |
Race | |||||||||
White | 87% | 91.1% | 89.3% | 88.7% | 88.1% | 88.4% | 82.8% | 78.4% | 80% |
Black | 8.7% | 5.6% | 6.9% | 7.5% | 6.8% | 7.1% | 13.8% | 17.6% | 16.3% |
Other | 4.3% | 3.3% | 3.8% | 3.8% | 5.1% | 4.5% | 3.4% | 3.9% | 3.8% |
Hispanic | 2.9% | 2.2% | 2.5% | 1.9% | 3.4% | 2.7% | 3.4% | 2.0% | 2.5% |
Note:
One-way ANOVA with Scheffe’s post hoc analyses examined differences on demographic and smoking behavior measure by HSR category (see Table
Overall, with regard to social, demographic, and environmental contexts, those with full smoking bans are younger, are more educated, have greater household incomes, and have more underage children in the home than those with no smoking restrictions. There were no differences between those with full and partial smoking bans on household income, education, and the number of children in the home; however, those with partial bans were older than those with full bans. With regard to smoking behaviors, those with full smoking bans smoked fewer cigarettes per day, had later TTFC, and had lower urges to smoke than both those with partial and no smoking bans.
For purposes of comparison to literature using categorical measures of TTFC, Chi-square analyses examined HSR category (full, partial, or no smoking ban) and TTFC category (e.g., smoking 0-5 minutes; 6-30 minutes; 31-60 minutes; >61 minutes after waking). Overall, the model highlighted differences between groups,
Chi-square of TTFC and HSR category.
>60 min | 31-60 min | 6-30 min | 0-5 min | Total | ||
---|---|---|---|---|---|---|
Full smoking ban | Count | 31 | 39 | 51 | 38 | 159 |
Percent | 19.5% | 24.5% | 32.1% | 23.9% | 100.0% | |
Partial smoking ban | Count | 12 | 12 | 40 | 48 | 112 |
Percent | 10.7% | 10.7% | 35.7% | 42.9% | 100.0% | |
No smoking ban | Count | 5 | 7 | 24 | 44 | 80 |
Percent | 6.3% | 8.8% | 30.0% | 55.0% | 100.0% | |
Total | Count | 48 | 58 | 115 | 130 | 351 |
Percent | 13.7% | 16.5% | 32.8% | 37.0% | 100.0% |
Chi-square differences between TTFC and household smoking restriction. Note: a, b, d: significantly different than both partial and no ban; c: not significantly different than partial or no ban; e: significantly different than no ban. All proportional differences at the.05 level.
Initial linear regression models examined if having full household smoking bans, versus partial bans and no restrictions on smoking, was predictive of TTFC controlling for age, gender, educational attainment, household income, average cigarettes per day, and loss of autonomy as measured by the HONC. For these models, TTFC was used as a continuous variable of actual minutes between waking and the first cigarette of the day. The full model demonstrated good fit,
Next, models examined if having only partial smoking bans, versus full bans and no smoking restrictions, predicted TTFC. The overall model fit was adequate,
Final models examined if having no smoking bans, versus full or partial bans, predicted TTFC. The overall model fit was adequate,
Alternative multinomial regression models were conducted reversing the order of the predictor and criterion (HSR and YYFC); these models examined if TTFC could predict HSR group (full, partial, or no ban) controlling for age, gender, educational attainment, household income, underage children in the home, number of people living in the home, average cigarettes per day, and HONC scores. Results demonstrate adequate model fit,
Overall, as expected, CPD and nicotine addiction as measured by the HONC predicted TTFC in regression models. Additionally, education emerged as a significant predictor or TTFC for those in all HSR categories. Alternate models predicting HSR categories found that age and income differentiated those with no bans from those with full or partial bans whereas an earlier TTFC differentiated between those with full bans from those with no or partial bans.
Multiple mediation models were guided by six hypotheses (see Figure
Findings of the first model examining the effect of a full household ban suggest an overall appropriate model fit,
Multiple mediation results.
Full smoking ban | Partial smoking ban | No smoking ban | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
SE | SE | SE | ||||||||||
HSR => CPD | -2.41 | .89 | -2.7 | .01 | 1.25 | .99 | 1.23 | .21 | 2.40 | 1.18 | 2.05 | .04 |
HSR => urges | -.87 | .29 | -3.01 | .002 | .91 | .33 | 2.78 | .005 | -.003 | .39 | -.009 | .99 |
HSR => HONC | -.25 | .23 | -1.05 | .30 | .42 | .26 | 1.66 | .10 | -.36 | .30 | -1.21 | .22 |
CPD => TTFC | -.81 | .14 | -5.68 | <.001 | -.84 | .14 | -5.89 | <.001 | -.01 | .003 | -5.75 | <.001 |
Urges => TTFC | -4.46 | .45 | -9.88 | <.001 | -.10 | .01 | -10.5 | <.001 | -.11 | .01 | -10.7 | <.001 |
HONC => TTFC | -.99 | .56 | -1.78 | .08 | -.97 | .56 | -1.72 | .08 | -.01 | .01 | -.89 | .36 |
HSR => TTFC (direct effect) | .13 | .05 | 2.04 | .04 | -1.45 | 2.29 | -.63 | .52 | -.14 | .06 | -2.33 | .02 |
Note: => indicates pathway between variables (see Figure
The next model examined the effect of a partial smoking ban and demonstrated an acceptable model fit,
The final multiple mediation models examined the effect of having no household smoking ban on TTFC. The overall model fit was acceptable,
Overall, full household bans have both a direct effect on TTFC and indirect effects mediated through cigarettes per day and urges to smoke. No household bans had a direct effect and an indirect effect, mediated through cigarettes per day, on TTFC. Having a partial ban had no direct effect on TTFC but did have an indirect effect, mediated through urges to smoke.
In this alternative model with the predictor and outcome variables (HSR and TTFC) reversed, the hypotheses are (1) later TTFC is associated with fewer cigarettes per day (path D in Figure
Findings of the alternative analyses found later TTFC are significantly associated with (1) fewer cigarettes per day,
The findings of the present study demonstrate a complex relation between household smoking restrictions and the time to the first cigarette of the day. We found that a full smoking ban was more likely to be categorized in the 31-60-minute and more than 60-minute categories of TTFC than those with partial or no bans, and less likely to be in the 0-5-minute category. Having a full smoking ban and having no smoking ban were both directly related to TTFC, even when considering smoking-related behaviors such as cigarettes per day and other social and demographic factors such as education, income, and age; however, there was no direct effect of having only a partial ban on TTFC. There were also indirect effects and mediated effects of HSR on TTFC, and these mediated effects varied by HSR category: for those with a full smoking ban, both cigarettes per day and urges to smoke were mediators, for those with no smoking bans only cigarettes per day mediated the effect, and for those with partial bans, only urges to smoke mediated the relation between HSR and TTFC. Additionally, findings demonstrate that those with partial smoking bans tend to be similar to those with full smoking bans with regard to social and demographic factors such as education and income but were more similar to those with no bans on smoking-related factors such as cigarettes per day, TTFC, and urges to smoke.
The time to first cigarette is often considered the best single marker of nicotine addiction, in part because it is also the best behavioral indicator of nicotine intake and nicotine biomarkers, and because it is relatively easy to assess. The utility of a highly sensitive and specific single indicator has advanced the understanding of nicotine addiction; however, there remains a lack of understanding as to how social environmental factors and physiologic measures of addiction help predict TTFC, and the interrelationships of these behaviors to TTFC. One important physiological measure in particular is smoking urges or cravings, which are considered an essential underlying characteristic of nicotine dependence, and which may, in turn, be affected by environmental factors that prompt smoking urges29. In the current study, we show that a full household smoking ban is associated with a reduction in CPD and urges to smoke in the morning; both of which are associated with an increased TTFC. However, partial smoking bans were not associated with the number of cigarettes per day and were associated with increased urges to smoke.
These findings help to understand why TTFC is a strong predictor of dependence; namely, that it is partially driven by cravings/urges, a defining characteristic of dependence and may be further influenced by HSR. For example, the relation between HSR and TTFC may come through the effect a smoking ban has on removing or altering smoking cues, which trigger an urge to smoke30. For example, when a smoker awakes, if a pack of cigarettes is not adjacent to the bed, the immediate urge to smoke may be reduced. Nevertheless, the specific aspects of home restrictions that may affect smoking behaviors needs further investigation; it may be that the relation between home restrictions and TTFC may reflect efforts to reduce overall smoking by lighter, less-addicted, smokers who are more motivated to quit. Whereas conventional wisdom might suggest that household restrictions may reflect motivation to reduce secondhand smoke for others in the household (e.g., children and other individuals in the household); however, findings of the current study show that, whereas individuals with full smoking bans had more underage children in the home, the presence of underage children or the number of other individuals in the household was not a significant covariate in any of the models examining the relation between HSR, cigarettes per day, nicotine addiction, and TTFC.
In addition to the above factors, the current study allowed us to determine the association of other salient behaviors with nicotine dependence that may be associated with TTFC, as indicated by alternative measures of dependence. For example, full or partial bans had no effect on the HONC score in mediation analysis. The HONC is a measure of the loss of autonomy, a theory of nicotine dependence that considers different behavioral measure of nicotine dependence than physiological nicotine cravings that precede the desire to smoke. Thus, whereas home smoking restrictions may have some effect on key behaviors associated with nicotine dependence, such as the number of cigarettes per day, it did not have an effect on this dimension of dependence.
Importantly, the findings of the current study demonstrated that the effects of household smoking restrictions on TTFC depended on whether the household restriction was a full or partial ban. Overall, it was found that individuals with partial smoking bans appear to be more similar to those who implement full smoking bans with regard to education and income yet are more similar to those who have no household smoking rules at all when it comes to nicotine dependence as measured by TTFC and CPD. Not only did partial household smoking bans not have an impact on TTFC or CPD, partial bans were associated with an increase in cravings. Whereas the study was not able to assess if partial bans are effective at decreasing secondhand smoke in the home for others living in the home, it does demonstrate that household rules that do not include a full ban on smoking in the home are not associated with changes in nicotine dependence, CPD, and may increase urges to smoke.
This study should be evaluated in light of its limitations. The nature of the cross-sectional does not allow for casual inference; thus, all models in the current study are presented as plausible associations between these variables which may inform future studies. We presented an alternative model in which the independent and dependent variables were switched in the sequence of relations. In the alternative model, TTFC had a direct effect on HSR, but there were no significant mediational pathways. However, regardless of the model, there is a clear association between TTFC and HSR. The present study offers potential mechanisms for this relation; however, further research is warranted to understand how HSR may impact the measurement of TTFC, whether or not TTFC should be “adjusted” or weighted based on the presence of HSR, and specifically, how HSR may be the result or cause of altered smoking behaviors. Additionally, the definition of “partial” smoking ban was somewhat ambiguous, and we could not quantify how much, when, and where smokers could smoke in the home—or if these bans were enforced, by either the participant or others in the home. It is possible that some participants who said they lived under a partial smoking ban may have had no effective restrictions on their in-home smoking.
Data is available upon request of the authors.
Authors have no financial interests or benefits related to this research to disclose.
We acknowledge the National Institute on Drug Abuse, National Institutes of Health (R01 DA026815), and National Institutes on Minority Health and Health Disparities, National Institutes of Health (R01 MD013338).