Article Text

Trends in cigarette consumption and time to first cigarette on awakening from 2002 to 2015 in the USA: new insights into the ongoing tobacco epidemic
1. Renee D Goodwin1,2,3,
2. Melanie M Wall4,5,6,
3. Misato Gbedemah1,
4. Mei-Chen Hu7,
5. Andrea H Weinberger8,9,
6. Sandro Galea10,
7. Michael J Zvolensky11,12,
8. Elizabeth Hangley13,
9. Deborah S Hasin3,4,5
1. 1 Department of Epidemiology and Biostatistics, Graduate School of Public Health and Health Policy, City University of New York, New York, USA
2. 2 Institute for Implementation Science in Population Health, City University of New York, New York, USA
3. 3 Department of Epidemiology, Mailman School of Public Health, Columbia University, New York, New York, USA
4. 4 Department of Psychiatry, College of Physicians and Surgeons Columbia University, New York, New York, USA
5. 5 New York State Psychiatric Institute, New York, New York, USA
6. 6 Department of Biostatistics, Mailman School of Public Health, Columbia University, New York, New York, USA
7. 7 Department of Sociomedical Science, Mailman School of Public Health, Columbia University, New York, New York, USA
8. 8 Ferkauf Graduate School of Psycholog, Yeshiva University, Bronx, New York, USA
9. 9 Department of Epidemiology & Population Health, Albert Einstein College of Medicine, Bronx, New York, USA
10. 10 Department of Epidemiology, Boston University School of Public Health, Boston, Massachusetts, USA
11. 11 Department of Psychology, University of Houston, Houston, Texas, USA
12. 12 Department of Behavioral Science, University of Texas MD Anderson Cancer Center, Houston, Texas, USA
13. 13 Department of Psychology, Queens College, City University of New York (CUNY), Flushing, New York, USA
1. Correspondence to Dr Renee D Goodwin, Department of Epidemiology and Biostatistics, Graduate School of Public Health and Health Policy, City University of New York, 55 West 125th Street, New York, New York 10027, USA; renee.goodwin{at}sph.cuny.edu

## Abstract

Objective The current study estimates trends in the number of cigarettes smoked per day (CPD) and percentage of smokers having their first CPD within 30 min of waking (time to first cigarette (TTFC)<30 min) among smokers from 2002 to 2015 in the USA overall, and adjusting for demographics. Trends in TTFC<30 min were also estimated by varying levels of cigarette consumption.

Methods Data were drawn from the National Household Survey on Drug Use, an annual nationally representative cross-sectional study of the US population aged 12 and older (n=54 079–58 397 per year). Linear time trend analyses of CPD and TTFC<30 min were conducted adjusting for age, gender and income; linear time trend analyses of TTFC among those at varying levels of CPD were then performed.

Results Estimates suggest that CPD declined overall from 2002 to 2015, and that the prevalence of TTFC<30 min declined overall among smokers (p<0.0001). The proportion of smokers consuming fewer (ie, 1–5, 6–15) CPD has increased while the number consuming 16+ CPD has decreased overall. Among those smoking 1–5 (p=0.0006) and 6–15 (p<0.0001) CPD, TTFC<30 min has increased significantly, but TTFC<30 min has remained unchanged among those smoking 16 or more CPD (p=0.5838).

Conclusions Findings suggest that smokers today are consuming fewer CPD, yet are increasingly likely to have their first cigarette earlier on awakening than they were a decade ago. Intervention and outreach efforts aimed at moving the prevalence lower may benefit from evaluating and addressing nicotine dependence even among lighter smokers.

• nicotine
• smoking caused disease

## Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

## Introduction

Globally, smoking is a leading preventable cause of premature death and disease resulting in more than 6 million annual deaths around the world.1–3 In the USA alone, smoking causes more than half a million deaths every year.4 5 Between 1965 and 1990, the prevalence of self-reported regular smoking among US adults in the past year dropped from 65% to 25.5%,6 and the 2015 prevalence of current (past 30 days) smoking was estimated as 22.2%.7 Identifying factors underlying the persistence of smoking is important towards developing new tobacco control programmes that will more effectively move the prevalence of smoking lower and reach public health targets.

One explanation that has been proposed to account for continued smoking in the population is that smokers who could quit easily, and therefore, ‘today’s smokers’ (in at least North America) have more difficult quitting due to any number of factors. If this is the case, we might expect to see a higher prevalence of obstacles to quitting and remaining abstinent among smokers today than in prior years. One of the hypothesised impediments most frequently studied is nicotine dependence (ND).2 8–10 ND is characterised by physical and/or psychological dependence on nicotine, and is present among some but not all people who smoke cigarettes. Smokers with ND generally have more difficulty quitting than those without ND.11–13 The amount/level of smoking (ie, number of cigarettes smoked per day (CPD)) is another strong predictor of difficulty quitting and is linked with genetic markers of addiction to nicotine.14 15 The amount of time between waking in the morning and smoking the first cigarette is commonly termed ‘time to first cigarette’ (TTFC), and a TTFC of less than 30 min (abbreviated as ‘TTFC <30 min’ for ease of language for the duration of the manuscript) has shown the strongest predictive value for ND in a single item. Specifically, prior work has shown that much of the predictive value of the Fagerström Test for Nicotine Dependence16 can be attributed to TTFC<30 min, and that this item has greater validity than any other single item in predicting ND. An increase in TTFC<30 min over time would suggest that smokers today may have more difficulty quitting than in the past.

Tobacco control policies, such as increases in price of cigarettes and taxation and smoke-free laws, which increasingly restrict zones of where smoking is allowed are likely to impact level of cigarette consumption. Relatedly, prior studies also suggest overall declines in cigarette consumption.17 Yet, dramatic increases in the overall quit rate have not been observed. It appears that the level of cigarette consumption is declining faster than the quit rate is increasing. As such, understanding trends in various impediments to cessation that are more (eg, CPD) and less (eg, TTFC<30 min) vulnerable to external forces (eg, smoke-free laws, cost) is needed to better appreciate current trends in smoking overall and potential forces that contribute to sustained smoking in the population.

It is also important to note that major shifts in the demographic characteristics of smokers have occurred over the past 10 years. For example, smoking is increasingly common among lower socioeconomic status (SES) segments of the population than those of higher SES.18 Therefore, adjustment for changes in demographic characteristics is needed to help disentangle trends in CPD and TTFC<30 min in the subgroups of the population who smoke.

The purpose of the current study is to use nationally representative data to estimate the 13-year trends (2002–2015) in the prevalence of CPD, TTFC and TTFC at varying levels of CPD among smokers thereby disentangling trends in TTFC<30 min from simultaneous changes in CPD over time. We investigated the following aims: (1) to estimate trends in the distribution of current, former and never smokers from 2002 to 2015; (2) to estimate trends in CPD among smokers from 2002 to 2015; (3) to estimate trends in TTFC among smokers from 2002 to 2015, adjusting for changes in CPD and demographics; and (4) to estimate trends in the prevalence of TTFC among smokers over time at varying levels of CPD (ie, 1–5. 6–15, 16+).

## Methods

The National Survey on Drug Use and Health (NSDUH) provides annual cross-sectional national data on the use of tobacco, other substance use and mental health in the USA.19–26

A multistage area probability sample for each of the 50 states and the District of Columbia has been conducted to represent the male and female civilian non-institutionalised population of the USA aged 12 and older. The annual public-use files from the years 2002 to 2015 were used in the analysis, consisting of approximately 55 000 participants per year (range: 54 079–58 397). Design-based weights are provided by NSDUH that adjust for non-response and poststratify to the known total US population aged 12 and older.

### Cigarette use

Current cigarette use was assessed by the question: ‘During the past 30 days, have you smoked part or all of a cigarette?’ Participants who responded ‘yes’ to this question were classified as current smokers.

Number of CPD in past 30 days was calculated for current smokers using this query: ’On the number of days you smoked cigarettes during the past 30 days, how many cigarettes did you smoke per day, on average?’ Responses were recoded into three categories; those who smoked 1–5 CPD (ie, a quarter of a pack or less), 6–15 CPD (ie, about half a pack) and 16 or more CPD (ie, a pack or more).

### Time to first cigarette

Participants reporting having smoked in the past 30 days were asked the TTFC (<30 min) after waking in the morning.12 TTFC of <30 min on awakening was used as an indicator of difficulty quitting in each annual survey. Those smoking within 30 min of waking were categorised as being ‘positive’ for TTFC<30 min and those not smoking within 30 min of waking were categorised as ‘negative’ for TTFC<30 min.

### Demographic characteristics

Demographics were obtained in the standard ways from the NSDUH: age (12–17 years old as reference group, 18–25 years old, 26 years old or older), gender (male as reference group, female), total annual family income (<$20 000 as reference group,$20 000 - $74 000,$75 000 or more).

### Statistical analysis

All analyses were performed incorporating the NSDUH sampling weights and controlling for the complex clustered sampling using SUDAAN V. 11.27 First, the prevalence of the smoking status (current smokers, former daily smokers, former non-daily smokers and never smokers) from 2002 to 2015 was calculated as were standard errors. We then calculated the proportions of number of CPD (1–5, 6–15, 16+), among current (past 30 day) smokers. Linear time trends of smoking status and CPD among current smokers were tested using logistic regression models with continuous year as the predictor. ORs and 95% CIs representing change in odds of the outcome across the 13-year span were presented (see tables 1 and 2).

Table 1

Prevalence of smoking status from 2002 to 2015 (The National Survey on Drug Use and Health, aged 12+)

Table 2

Prevalence of CPD among current smokers from 2002 to 2015 (The National Survey on Drug Use and Health, aged 12+)

The prevalence and associated standard errors of TTFC<30 min were calculated each year among all current smokers and separately by smokers using the same number of CPD (1–5, 6–15, 16+). Linear time trends were tested in each group using logistic regression models with continuous year as the predictor. Multiple logistic regression was then used to adjust for demographics (ie, age, gender, total annual family income) in the model of all current smokers. ORs (from unadjusted models) and adjusted ORs from multiple logistic regression models and 95% CIs representing change in odds of TTFC<30 min across the 13-year span were presented (see table 3). Interactions between year and each demographic characteristic were tested among all smokers and none were significant, so analyses were not further stratified by demographics. Visual inspection of residual plots did not suggest any deviation from linear trends in any outcomes over the 13-year period.

Table 3

Prevalence of TTFC nicotine dependence stratified by a number of CPD among current smokers (The National Survey on Drug Use and Health, 2002–15, aged 12+)

## Results

### Interaction between year and each demographic characteristic

No interactions were found to be significant between year and each demographic characteristic among current smokers (interaction year × demographic characteristic, p value=0.2055 for age; p value=0.2923 for gender; p value=0.2588 for family income).

### Prevalence and distribution of smoking from 2002 to 2015

Among US adults ages 12 and older, the prevalence of current and former smoking declined whereas the prevalence of never smoking increased from 2002 to 2015 (see table 1). There were no significant changes in the prevalence of former smoking (see table 1).

### Changes in CPD among current smokers

Among current smokers, the proportion of those smoking 1–5 CPD and 6–15 CPD increased significantly from 2002 to 2015, while the proportion of current smokers smoking 16 or more CPD decreased (p<0.0001; see table 2). Specifically, the proportion of those smoking heavily (ie, 16 or more CPD) decreased from 37.8% in 2002 to 27.6% in 2015, whereas the proportion smoking 1–5 CPD and 6–15 CPD increased respectively from 37.0% to 43.6% in 2002 and from 25.3% to 28.8% in 2015 (see table 2).

### Changes in the prevalence of TTFC<30 min among current smokers

Results of unadjusted analyses suggest that the prevalence of TTFC<30 min declined significantly among current smokers from 2002 to 2015 (OR=0.89 (0.84, 0.94), p<0.0001; see table 3). After adjusting for CPD and demographics, results showed an increasing trend in TTFC among smokers (AOR=1.18 (1.10, 1.26; p<0.0001)).

### Changes in prevalence of TTFC<30 min among current smokers stratified by CPD

Stratifying by level of CPD, a significant increasing trend in the prevalence of TTFC<30 min among those smoking 1–5 CPD (OR=1.19 (1.08, 1.32)) and 6–15 CPD (OR=1.26 (1.14, 1.39); see table 3) was observed. Among those smoking 16 or more CPD, there was no significant change in the prevalence of TTFC<30 min from 2002 to 2015 (OR=1.03 (0.92, 1.16)). These results remained significant after controlling for demographics.

## Discussion

The purpose of this study was to examine whether ‘today’s smokers’ in the USA may have more difficulty quitting than smokers in the past by separately examining trends over time in two indicators of difficulty quitting from 2002 to 2015: TTFC<30 min and CPD. There were four key findings. First, there was a large decline in CPD with 37.8% of current smokers reporting smoking 16+ CPD in 2002 and only 27.6% of current smokers reporting smoking 16+ CPD in 2015. Second, when considered overall, a modest 3.3% decline in the prevalence of TTFC<30 min was observed among smokers in the US population over time (50.4% in 2002 to 47.1% in 2015). Third, when this simultaneous decrease in CPD was taken into account, an interesting finding emerged such that, rather than a decrease, a significant increase in TTFC among smokers was observed. Fourth, there was an increase of 6.6% in use of 1–5 CPD and 3.5% in use of 6–15 CPD from 2002 to 2015 and a significant increase in TTFC<30 min was observed among those smoking 1–5 CPD and 6–15 CPD. Finally, adjusting for demographics had little impact on these relationships.

Our results offer novel insight into the nature of the smoking population in the USA from 2002 to 2015. While cigarette consumption and TTFC<30 min declined in the overall population of smokers over time, the prevalence of TTFC<30 min, a key indicator of physical dependence on nicotine, increased significantly among those who are smoking at lower levels of daily consumption, which make up a growing percentage of the smoking population. One possible interpretation of this pattern of results is that tobacco control has been effective in encouraging even people who are dependent on nicotine to smoke fewer CPD, though not necessarily to quit entirely. So, it is conceivable that formerly heavy smokers are moving into lower CPD categories, and therefore we are seeing higher levels of dependence among those with lower CPD over time. This may be one reason why we see an increase in TTFC<30 min exclusively among those at lower levels of CPD. Furthermore, a recent study suggests that while CPD decreased over the past 20 years in the US population, nicotine intake has not changed (per cotinine levels)28; these data suggest that declines in reported cigarette consumption have not lead to a similar decline in nicotine intake. Accordingly, compensatory smoking behaviours may be at work towards maintaining the same nicotine dose with fewer cigarettes. In considering TTFC<30 min and CPD both as indicators of difficulty quitting and remaining abstinent, a focus on TTFC<30 min in the implementation of smoking cessation efforts, especially for those at lower levels of CPD consumption, may be a next step towards moving the overall prevalence of smoking lower.

There are several limitations of this study. First, we only have detailed smoking information (ie, CPD) for past 30-day smokers. Therefore, the analyses are restricted to current smokers only. However, the timeframe for TTFC<30 min is also over the past 30 days so this may be the best fit to examine both CPD and TTFC<30 min prevalence. Second, our data only extend back to 2002. There have been more significant tobacco control events prior to 2002, the effects of which we are unable to capture in this analysis. Future studies that can examine the changes in prevalence of barriers to quitting among smokers over a longer time period are needed to understand time trends over the past several decades. Third, in this study we examined current smokers by CPD level, combining daily and non-daily smokers. Therefore, the range of smoking amount and frequency is quite wide. We chose this approach because trends when stratified by daily versus non-daily smoking were fairly similar. However, given the changes over time in CPD, ND and prevalence of daily and non-daily smoking, future work examining the interplay of these factors on a more fine-grained level may be informative and worthwhile in gaining a deeper understanding of the relations between these factors and how they may be changing over time. Fourth, replication of these findings with other indicators of barriers to successful quitting would be useful. Research that can tease out the role of various impediments to sustained abstinence in persistent smoking may be useful in terms of informing intervention strategies. The study has also several strengths. First, the use of the NSDUH provides a large sample size for 13 consecutive years. Second, NSDUH annually collects detailed measures of cigarette use frequency and quantity for current smokers. Third, barriers to cessation (eg, CPD and TTFC<30 min), which are often grouped together, rather than disentangled, provide a unique perspective into the dynamic and juxtaposed trends in various barriers to quitting among smokers in the population. Fourth, the recency of the information increases the generalisability of these results to the current smoking population.

Strong tobacco control policies clearly have an impact on quit rates both in the USA and internationally.29 Yet, our results suggest that failure to address key barriers to quitting—at least among smokers at lower levels of consumption in the USA—may be one key barrier towards bringing smoking rates even lower. While a decline in CPD overall is a positive sign in terms of physical health and can be considered a success from a harm reduction perspective, increased TTFC<30 min among remaining smokers at lower levels of consumption may signal that remaining light smokers continue to be less likely to quit. Understanding whether and to what degree today’s smokers have more trouble quitting than in the past even as the number of smokers has declined—including among those smoking fewer CPD—is important in the future of tobacco control planning.

• Despite declines over the past several decades, smoking remains a major public health problem in the USA.

• Short time to first cigarette upon awakening in the morning (TTFC<30 min) and number of cigarettes smoked per day (CPD) are two key indicators of difficulty quitting smoking.

• Prior work suggests a steady decline in daily cigarette consumption overall.

• No prior study has examined whether there have been changes in TTFC (<30 min) among those at various levels of CPD.

• Results suggest that while the proportion of smokers consuming 16+ CPD has decreased, the proportion consuming 1-5 CPD and 6-15 CPD has increased over the past decade.

• TTFC (<30 min) has increased among those smoking fewer CPD (i.e., 1-5, 6-15) over time, but appears unchanged among those smoking 16+ CPD.

• Intervention and outreach efforts aimed at moving the prevalence lower may benefit from evaluating and addressing nicotine dependence among lighter smokers.

## Footnotes

• Contributors All authors made substantial contributions to the conception or design of the study, as well as the analysis and interpretation of the work. All authors contributed to drafting the manuscript and/or revising it for important intellectual content. All authors give final approval of the manuscript. All authors agree that they are accountable for all aspects of the work. RDG conceived of the original and wrote the manuscript; DSH, SG, AHW and MJZ provided critical feedback in the study design, interpretation of results and writing of the manuscript; MMW designed the analytic plan and oversaw data analysis; MCH and MG conducted the statistical analyses; EH contributed to literature review and drafting and revising key content of the manuscript.

• Funding This work was supported by grant #2R01 DA20892 (Goodwin) from NIDA.

• Competing interests None declared.

• Provenance and peer review Not commissioned; externally peer reviewed.