## Abstract

Introduction:

Among cigarette smokers, lower levels of consumption, defined as smoking fewer cigarettes per day (CPD) or not smoking daily, are becoming more common. The relationship between cigarette consumption and smoking frequency (daily or nondaily) is not well characterized, and the natural history of light smoking (defined here as smoking ≤10 CPD) is poorly understood.

Methods:

We assessed changes in CPD and smoking frequency over time among light smokers (≤10 CPD) and very light smokers (≤5 CPD), using a population-based longitudinal survey of 3,083 adult smokers in Massachusetts who were interviewed three times over a 4-year follow-up period (in 2000–2001, 2002–2003, and 2005–2006). We used logistic regression to identify factors associated with light smokers’ progression to heavier smoking or smoking reduction/quitting.

Results:

Seventy percent of very light smokers were nondaily smokers. Very light nondaily smokers differed from very light daily smokers by younger age, higher socioeconomic status, a social smoking pattern, later smoking initiation, less evidence of nicotine addiction, and more recent and planned cessation efforts. Very light nondaily smokers and smokers consuming 6–10 CPD were more likely to remain in the same smoking category and were less likely to increase consumption than were very light daily smokers. Factors independently associated with increasing consumption among very light smokers were smoking daily, nicotine dependence, White ethnicity, social smoking, and having more friends who smoked; among smokers consuming 6–10 CPD, male gender and lack of quitting self-efficacy were associated with increasing consumption.

Conclusion:

Our findings indicate that most light smoking is not a gateway to heavier smoking.

## Introduction

Cigarette consumption in the United States has been declining steadily over several decades (Orzechowski & Walker, 2003). Lower levels of cigarette smoking, whether defined as smoking fewer cigarettes per day (CPD) or not smoking on a daily basis, are becoming more common among U.S. smokers (Porter, Jackson, Trosclair, & Pederson, 2003). As of 2006, approximately 20% of current smokers did not smoke every day (Rock et al., 2007) compared with just less than 17% in 1992 (Centers for Disease Control and Prevention, 1994).

Epidemiological studies consistently show a dose response for most adverse health outcomes related to smoking (U.S. Department of Health and Human Services [USDHHS], 2004). However, light smoking, generally defined as smoking no more than 10 CPD, still increases cardiovascular morbidity and mortality as well as all-cause mortality (Bjartveit & Tverdal, 2005; Kawachi et al., 1994; Luoto, Uutela, & Puska, 2000; Okuyemi et al., 2002; Prescott, Scharling, Osler, & Schnohr, 2002; Rosengren, Wilhelmsen, & Wedel, 1992), even at very low levels of cigarette smoking (e.g., ≤5 CPD or smoking less than daily; Bjartveit & Tverdal, 2005; Kawachi et al., 1994; Rosengren et al., 1992). These health implications underscore the importance of understanding the pattern and natural history of light smoking, particularly because light smoking is common among young smokers, whose smoking patterns may still be developing (Biener & Albers, 2004; Shiffman, Kassel, Paty, Gnys, & Zettler-Segal, 1994; USDHHS, 1994). In different individuals, light smoking may be a stable pattern of tobacco use, an intermediate step on the route to heavier and therefore riskier smoking, or a step leading to cessation. The smoker characteristics that are predictive of each of these outcomes are unclear. Identifying which light smokers are likely to increase consumption is of critical importance to both public health and public policy.

Information about patterns of light smoking is challenging to interpret because the phenomenon has been defined inconsistently in the literature. Various cutoffs have been used to define light smoking, and the relationship between smoking a small number of CPD and smoking on a “nondaily” basis (also called intermittent or occasional smoking) is often not clarified. Using a variety of smoking status definitions (typically nondaily or 1–5 CPD), existing longitudinal studies suggest that the lightest smokers are more likely than heavier smokers to either increase or decrease consumption (Hassmiller, Warner, Mendez, Levy, & Romano, 2003; Hennrikus, Jeffery, & Lando, 1996; Hyland, Rezaishiraz, Bauer, Giovino, & Cummings, 2005; Lindstrom & Isacsson, 2002; McDermott, Dobson, & Owen, 2007; Stanton, Papandonatos, Lloyd-Richardson, & Niaura, 2007; Wetter et al., 2004; Zhu, Sun, Hawkins, Pierce, & Cummins, 2003). However, these studies are of limited utility for identifying factors associated with transitions in smoking status in the U.S. context. They focus on non-U.S. populations (Lindstrom & Isacsson, 2002; McDermott et al., 2007) or narrowly defined U.S. populations (e.g., adolescents, college students, or employees; Hennrikus et al., 1996; Stanton et al., 2007; Wetter et al., 2004), lack a prospective design (e.g., using recall of prior year's smoking; Hassmiller et al., 2003), or do not focus on determinants of increased or decreased cigarette consumption (Hyland et al., 2005; Zhu et al., 2003).

We addressed these research gaps by analyzing survey data from a recent large population-based cohort of adult smokers in Massachusetts who were interviewed three times over a 4-year follow-up period. We disaggregated smokers into four groups by consumption (CPD) and frequency (daily vs. nondaily smoking), examined changes in smoking status over time in each of these groups, and identified factors associated with progression to heavier smoking and to quitting or reduction for different groups of light smokers (defined in this study as smoking ≤10 CPD).

## Methods

### Sample

A probability-sample random-digit–dialed survey was administered to residents of Massachusetts in 2001–2002 to study smoking practices, attitudes, and support for tobacco control policies in the state. The survey oversampled smokers, young adults, and recent quitters to ensure adequate power to study those subpopulations. Professional interviewers from the University of Massachusetts–Boston Center for Survey Research attempted to interview one systematically selected adult aged at least 18 years in each eligible household identified through an initial screening interview with an adult household resident. They successfully screened 66% of sampled households and interviewed 70% of selected adults for a final sample size at baseline of 6,739.

All 3,083 respondents who indicated at baseline that they had smoked at least 100 cigarettes in their lives and currently smoked every day or some days were contacted again 2 and 4 years later. Of the baseline smokers, 1,726 (56.0%) completed follow-up interviews in 2003–2004 and 1,319 (42.8%) completed interviews in 2005–2006. The initial response rate and the follow-up response rates are roughly comparable with other major population-based surveys of smokers (Zhu et al., 2003). Respondents aged 30 years or younger, regardless of smoking status, and respondents indicating that they had quit smoking in the past 2 years also were contacted again. Those who initiated (n = 1) or resumed (n = 34) smoking at wave 2 were included in the 2-year longitudinal analyses.

### Measures

#### Smoking status.

At each interview, all adults aged 18 years or older who reported that they had smoked at least 100 cigarettes in their lives were asked (a) if they currently smoked every day or some days and (b) how many CPD they smoked, on average, over the past 30 days. Each smoker's level of smoking was categorized at each wave using these two dimensions. Cigarette consumption was categorized into three groups: ≤5, 6–10, and >10 CPD. Our focus on smokers consuming ≤10 CPD follows Okuyemi et al. (2002). The decision to further split smokers above and below 5 CPD is based on the “chipper” literature. Chippers are smokers who consume no more than 5 CPD over an extended period of time and who may have limited or no nicotine dependence (Shiffman, 1989). We use the term very light to refer to all smokers consuming no more than 5 CPD and light to refer to smokers consuming no more than 10 CPD. Smoking frequency was defined as smoking every day (daily smokers) or less than daily (nondaily smokers).

#### Covariates.

The survey assessed demographic factors, socioeconomic status, smoking and quitting behavior, social environment, and no-smoking policies at home and at work. Demographic factors included sex, age, race/ethnicity, and marital status. Race/ethnicity categories (84% White, 4% Hispanic, 6% Black, 1% Asian, and 4% other) were collapsed into the two categories of White and non-White because the numbers of respondents in individual non-White categories were insufficient to provide statistically stable estimates. Socioeconomic status was measured by years of education and income.

Measures of smoking behavior included the following: CPD, age at which the respondent began smoking, and the time to first cigarette after waking (a measure of nicotine dependence; Heatherton, Kozlowski, Frecker, Rickert, & Robinson, 1989). A pattern of social smoking was assessed with the question “Do you smoke mainly when you are with other people, mainly when you are alone, or as often by yourself as with others?” Quitting behaviors included having made an intentional quit attempt lasting 1 day or longer in the past 12 months, having a plan to quit smoking in the next 30 days, and self-efficacy for quitting (“How sure are you that you could refrain from smoking for at least 1 month if you wanted to?”; “very sure” = 1, else = 0).

Smoking in the social environment was assessed by asking how many friends were also smokers and the smoking policy of the respondent's home and workplace.

### Data analyses

Bivariate analyses examined differences in the distribution of covariates across smoking status categories defined both by consumption and by frequency. Statistically significant differences were assessed using chi-square tests.

Examinations of smoking status transitions were stratified by baseline smoking status and consisted of analyses with and without adjustments for covariates. Unadjusted analyses were conducted for 2- and 4-year intervals. Analyses of 2-year intervals included intervals beginning in both 2001–2002 and 2003–2004. Respondents whose smoking status was in the same stratum at the beginning of the 2001–2002 to 2003–2004 and 2003–2004 to 2005–2006 intervals were included twice for the analysis of that stratum. CIs were adjusted to account for repeated observations of individuals whose transitions were included for both intervals. For the 4-year intervals, transitions were defined as quitting (either quitting by wave 2 and staying quit at wave 3 or quitting between waves 2 and 3), reducing consumption, continuing in the same category of consumption (stable), increasing consumption, or a mixed trajectory (e.g., quitting or reducing and then relapsing to a level higher than baseline or increasing and then decreasing to baseline or lower levels of smoking). Only those respondents with smoking data for all three waves were included in the 4-year analyses.

We used logistic regression to conduct adjusted analyses determining which respondent characteristics were associated with either quitting/reducing smoking or increasing cigarette consumption over 2-year intervals. Because of attrition, we had an insufficient sample size to run multivariate analyses of 4-year transitions. Logistic models were calculated for both smokers consuming no more than 5 CPD and those consuming 6–10 CPD, controlling for nondaily/daily smoking. All covariates listed above were included in the regression models based on evidence that they are associated with light smoking: demographics (Hassmiller et al., 2003; Hennrikus et al., 1996; Husten, McCarty, Giovino, Chrismon, & Zhu, 1998; Hyland et al., 2005; Tong, Ong, Vittinghoff, & Perez-Stable, 2006; Wortley, Husten, Trosclair, Chrismon, & Pederson, 2003), socioeconomic status (Hassmiller et al., 2003; Hennrikus et al., 1996; Husten et al., 1998; Hyland et al., 2005; Wortley et al., 2003), quitting behavior and intentions (Hennrikus et al., 1996; Tong et al., 2006; Zhu et al., 2003), social environment (Hyland et al., 2005; Owen, Kent, Wakefield, & Roberts, 1995), and household or workplace smoking policies (Gilpin, Cavin, & Pierce, 1997; Hyland et al., 2005; Tong et al., 2006; Zhu et al., 2003). As was the case in the unadjusted analyses, respondents whose smoking status was constant from 2001–2002 to 2003–2004 were included twice in the longitudinal analyses, and SEs were corrected for repeated observations within individuals.

All analyses were conducted using sampling weights. The weights accomplished two goals: accounting for oversampling in the survey design and correcting for differential attrition by gender, age, minority status, and education level. The attrition corrections were specific to the last wave of a longitudinal analysis. In the 2-year analyses, the weight assigned to an observation was specific to wave 2 for observations with a baseline of wave 1 and to wave 3 for observations with a baseline of wave 2. For the 4-year analyses, we used the weights specific to wave 3. Analyses were conducted using only those respondents for whom we had complete data on a given transition and for all relevant covariates. Losses to follow-up (after controlling for gender, age, minority status, and education) were associated with not being married, lower income, having more friends who smoke, and not having household or workplace smoking bans. Baseline smoking status was not associated with loss to follow-up. All analyses were conducted using Intercooled Stata 9.2 for Windows.

## Results

### Smoking frequency and consumption

Table 1 displays the prevalence of different smoking statuses at baseline along the dimensions of cigarette consumption and frequency of smoking. Some 70% (14.9%/21.3%) of the respondents who smoked on average no more than 5 CPD were nondaily smokers, and 88% of nondaily smokers (14.9%/16.9%) smoked no more than 5 CPD. In contrast, nearly all smokers (97%) consuming on average more than 5 CPD smoked on a daily basis. Based on this distribution, we collapsed smoking status categories into four consumption frequency groups for the subsequent analyses: (a) nondaily, ≤5 CPD; (b) daily, ≤5 CPD; (c) daily/nondaily, 6–10 CPD; and (d) daily/nondaily, >10 CPD.

Table 1.

Cigarette consumption and smoking frequency in the Massachusetts cohort at baseline (n = 2,980)

 Smoking frequency Cigarette consumption; % (number of subjects)a ≤5 CPD 6–10 CPD 11–20 CPD >20 CPD Total Daily 6.4 (222) 17.6 (527) 42.7 (1,253) 16.4 (464) 83.1 (2,466) Nondaily 14.9 (455) 1.8 (53) 0.2 (5) 0.0 (1) 16.9 (514) Total 21.3 (677) 19.4 (580) 42.8 (1,258) 16.4 (465) 100.0 (2,980)
 Smoking frequency Cigarette consumption; % (number of subjects)a ≤5 CPD 6–10 CPD 11–20 CPD >20 CPD Total Daily 6.4 (222) 17.6 (527) 42.7 (1,253) 16.4 (464) 83.1 (2,466) Nondaily 14.9 (455) 1.8 (53) 0.2 (5) 0.0 (1) 16.9 (514) Total 21.3 (677) 19.4 (580) 42.8 (1,258) 16.4 (465) 100.0 (2,980)

Note. CPD = cigarettes per day.

a

Table entries are weighted percentages.

### Characteristics of smokers by levels of cigarette consumption

Table 2 displays the baseline characteristics of the cohort of smokers by the four smoking status categories. Statistically significant differences across all four smoking categories were observed for all characteristics measured. Lighter smokers were generally younger and better educated, had higher incomes, and were more likely to be female and non-White. They started smoking at older ages, were less nicotine dependent, were more likely to have tried to quit in the past year, were more likely to plan on quitting in the next year, and were more likely to live and work in environments where smoking was banned. In most cases, the significant differences in characteristics across smoking groups appeared to be driven by the characteristics of those consuming more than 10 CPD.

Table 2.

Respondent characteristics at baseline by level of cigarette consumption (n = 2,980)

 Variable 1 2 3 4 p value; H0: 1 = 2 p value; H0: All categories equal ≤5 CPD, nondaily (n = 677) ≤5 CPD, daily (n = 580) 6–10 CPD (n = 1,258) >10 CPD (n = 465) Demographics (%) Female 59 67 61 48 .08 <.001 Age (years) .06 <.001 18–24 26 23 21 14 25–44 53 47 51 45 45–64 17 24 25 35 65+ 3 6 4 6 White 79 79 76 88 .96 <.001 Married 32 35 43 40 .59 .03 Socioeconomic status (%) Education ≤ high school 21 44 52 56 <.001 <.001 Income .001 <.001 <$20,000 12 14 10 14$20,000–$50,000 25 34 34 32 >$50,000 54 31 40 38 Unknown 9 20 16 15 Smoking behavior Mean number of CPD 1.7 3.8 9.2 23.9 <.001 <.001 Started smoking at age ≤16 years (%) 35 49 45 59 .004 <.001 Smoke first cigarette ≤30 min after waking (%) 7 18 37 74 <.001 <.001 Social smoking (smokes more often with friends) (%) 59 29 23 11 <.001 <.001 Quitting behavior and intentions (%) Made quit attempt in the past year 57 46 44 39 .03 <.001 Plans to quit smoking .003 <.001 In the next 30 days 41 33 31 26 In the next 6 months 25 32 34 34 After 6 months 27 19 22 19 No plan to quit entirely 8 16 12 22 Self-efficacy (very sure I can quit for ≥30 days) 69 49 41 27 <.001 <.001 Social environment (%) At least half of friends smoke 39 45 53 62 .17 <.001 No-smoking policies (%) Smoking banned in household 55 49 42 26 .23 <.001 Smoking banned in workplace 50 41 42 31 .05 <.001
 Variable 1 2 3 4 p value; H0: 1 = 2 p value; H0: All categories equal ≤5 CPD, nondaily (n = 677) ≤5 CPD, daily (n = 580) 6–10 CPD (n = 1,258) >10 CPD (n = 465) Demographics (%) Female 59 67 61 48 .08 <.001 Age (years) .06 <.001 18–24 26 23 21 14 25–44 53 47 51 45 45–64 17 24 25 35 65+ 3 6 4 6 White 79 79 76 88 .96 <.001 Married 32 35 43 40 .59 .03 Socioeconomic status (%) Education ≤ high school 21 44 52 56 <.001 <.001 Income .001 <.001 <$20,000 12 14 10 14$20,000–$50,000 25 34 34 32 >$50,000 54 31 40 38 Unknown 9 20 16 15 Smoking behavior Mean number of CPD 1.7 3.8 9.2 23.9 <.001 <.001 Started smoking at age ≤16 years (%) 35 49 45 59 .004 <.001 Smoke first cigarette ≤30 min after waking (%) 7 18 37 74 <.001 <.001 Social smoking (smokes more often with friends) (%) 59 29 23 11 <.001 <.001 Quitting behavior and intentions (%) Made quit attempt in the past year 57 46 44 39 .03 <.001 Plans to quit smoking .003 <.001 In the next 30 days 41 33 31 26 In the next 6 months 25 32 34 34 After 6 months 27 19 22 19 No plan to quit entirely 8 16 12 22 Self-efficacy (very sure I can quit for ≥30 days) 69 49 41 27 <.001 <.001 Social environment (%) At least half of friends smoke 39 45 53 62 .17 <.001 No-smoking policies (%) Smoking banned in household 55 49 42 26 .23 <.001 Smoking banned in workplace 50 41 42 31 .05 <.001

Note. CPD = cigarettes per day.

Within the group of very light smokers (≤5 CPD), we found substantial differences between daily and nondaily smokers (see Table 2). Very light nondaily smokers, compared with very light daily smokers, were better educated and had higher incomes but did not differ significantly in gender or by age as categorized in Table 2. However, an additional comparison (not shown) demonstrated that very light nondaily smokers were significantly more likely than very light daily smokers to be aged 30 years or under (65% vs. 46%, p < .001). Very light nondaily smokers smoked fewer CPD on average; started smoking at an older age; and were less nicotine dependent, more likely to have made a recent quit attempt, more likely to be planning to quit smoking soon, more likely to display self-efficacy regarding quitting, and more likely to smoke with friends.

### Longitudinal analyses

We analyzed 2,664 two-year transitions for 2,008 smokers and 1,106 four-year transitions for as many smokers. Table 3 shows the results by four levels of baseline smoking. Among light smokers, the patterns of those consuming no more than 5 CPD on a nondaily basis differed from the others. Very light nondaily smoking was a stable pattern for many smokers, with nearly half of such smokers maintaining the pattern for 2 years and nearly one-third maintaining it for 4 years. A substantial proportion of smokers consuming 6–10 CPD also exhibited a stable smoking pattern; 40% were stable for 2 years and 24% were stable for 4 years. Very light daily smokers had significantly less stable smoking patterns than did other light smokers. Only 21% maintained their smoking status for 2 years (p = .03 vs. ≤5 CPD, nondaily; p < .001 vs. 6–10 CPD), and only 7% maintained it for 4 years (p = .003 vs. ≤5 CPD, nondaily; p = .001 vs. 6–10 CPD). Both very light nondaily smokers and smokers consuming 6–10 CPD were significantly less likely to increase their cigarette consumption more than 2 years (20% and 19%, respectively) compared with very light daily smokers (32%; p < .001 and p = .03, respectively). Very light nondaily smokers also were more likely than very light daily smokers to quit during follow-up, although the difference was not statistically significant. Smokers consuming 6–10 CPD were significantly less likely to quit than very light nondaily smokers (p = .003 at 4 years), but they were not less likely to quit than very light daily smokers. Smokers consuming more than 10 CPD were a highly stable group, with 72% continuing to smoke at this minimum level after 2 years and 65% after 4 years. They were less likely to quit smoking than smokers at lower levels of consumption.

Table 3.

Two- and four-year trajectories by baseline smoking status

 Baseline smoking status 2-Year transitions, % (95% CI) 4-Year transitions, % (95% CI) ≤5 CPD, nondaily 328 smokers, 399 transitions 164 smokers, 164 transitions Quits 32 (26–38) 44 (34–54) Continues nondaily, ≤5 CPD 48 (41–55) 31 (22–42) Increases to >5 CPD or daily smoking 21 (16–26) 20 (14–29) Mixed trajectorya NA 5 (2–10) ≤5 CPD, daily 147 smokers, 160 transitions 65 smokers, 65 transitions Quits 27 (19–37) 27 (17–41) Reduces to nondaily smoking 18 (12–26) 21 (11–34) Continues daily, ≤5 CPD 21 (15–29) 7 (3–18) Increases to >5 CPD 35 (27–43) 32 (21–46) Mixed trajectorya NA 12 (6–24) 6–10 CPD 440 smokers, 524 transitions 219 smokers, 219 transitions Quits 21 (16–26) 28 (21–35) Reduces to ≤5 CPD 15 (12–19) 16 (10–23) Continues 6–10 CPD 40 (35–45) 24 (17–33) Increases to >10 CPD 24 (20–29) 19 (13–26) Mixed trajectorya NA 13 (9–19) >10 CPD 1,093 smokers, 1,581 transitions 658 smokers, 658 transitions Quits 14 (12–16) 21 (18–26) Reduces to ≤10 CPD 14 (12–17) 14 (11–18) Continues >10 CPD 72 (69–75) 65 (60–69)
 Baseline smoking status 2-Year transitions, % (95% CI) 4-Year transitions, % (95% CI) ≤5 CPD, nondaily 328 smokers, 399 transitions 164 smokers, 164 transitions Quits 32 (26–38) 44 (34–54) Continues nondaily, ≤5 CPD 48 (41–55) 31 (22–42) Increases to >5 CPD or daily smoking 21 (16–26) 20 (14–29) Mixed trajectorya NA 5 (2–10) ≤5 CPD, daily 147 smokers, 160 transitions 65 smokers, 65 transitions Quits 27 (19–37) 27 (17–41) Reduces to nondaily smoking 18 (12–26) 21 (11–34) Continues daily, ≤5 CPD 21 (15–29) 7 (3–18) Increases to >5 CPD 35 (27–43) 32 (21–46) Mixed trajectorya NA 12 (6–24) 6–10 CPD 440 smokers, 524 transitions 219 smokers, 219 transitions Quits 21 (16–26) 28 (21–35) Reduces to ≤5 CPD 15 (12–19) 16 (10–23) Continues 6–10 CPD 40 (35–45) 24 (17–33) Increases to >10 CPD 24 (20–29) 19 (13–26) Mixed trajectorya NA 13 (9–19) >10 CPD 1,093 smokers, 1,581 transitions 658 smokers, 658 transitions Quits 14 (12–16) 21 (18–26) Reduces to ≤10 CPD 14 (12–17) 14 (11–18) Continues >10 CPD 72 (69–75) 65 (60–69)

Note. CPD = cigarettes per day.

a

Mixed trajectory refers to quitting or reducing and then relapsing to more than baseline, or increasing and then decreasing to baseline or less.

Table 4 presents results from multivariate longitudinal analyses used to identify factors independently associated with 2-year transitions in smoking status (either quitting/reducing or increasing smoking) among two groups of light smokers categorized by cigarette consumption (≤5 CPD, 6–10 CPD) at the start of the follow-up period. The analyses identified different factors for these two groups. Plans to quit in the next 30 days and household smoking bans were associated with an increased likelihood of quitting over the 2-year follow-up period in the very light smoker group (≤5 CPD). Consistent with the analysis in Table 3, very light smokers who smoked daily had a greater risk of progressing to higher consumption than did very light nondaily smokers. Very light smokers who were White, who smoked soon after waking (suggesting greater nicotine dependence), and who had a majority of friends who were smokers were at greater risk of increasing smoking. Very light smokers who smoked mostly with friends were less likely to increase smoking over time.

Table 4.

Adjusted odds of changing smoking status for low-rate smokers

 ≤5 CPD (436 smokers, 532 transitions) 6–10 CPD (421 smokers, 499 transitions) Quit Increase Reduce or quit Increase AOR (95% CI) AOR (95% CI) AOR (95% CI) AOR (95% CI) Demographics/SES Female 1.39 (0.83–2.32) 0.73 (0.43–1.24) 1.55 (0.84–2.87) 0.55 (0.32–0.96)* Age >30 years 0.95 (0.55–1.65) 1.19 (0.66–2.16) 0.69 (0.38–1.27) 1.60 (0.89–2.86) White 0.94 (0.44–2.04) 3.06 (1.38–6.80)** 0.28 (0.14–0.56)*** 1.34 (0.69–2.60) Married 1.03 (0.90–1.18) 0.99 (0.85–1.16) 0.99 (0.85–1.15) 1.04 (0.90–1.21) Education ≤ high school 0.83 (0.47–1.45) 1.31 (0.75–2.28) 0.75 (0.41–1.40) 1.08 (0.65–1.82) Income >$20,000 1.08 (0.43–2.73) 1.66 (0.72–3.83) 2.93 (0.95–9.09) 1.18 (0.48–2.90) Income unknown 1.68 (0.49–5.81) 1.52 (0.50–4.63) 2.43 (0.52–11.35) 0.74 (0.22–2.44) Smoking characteristics Daily smoker 1.02 (0.58–1.81) 2.04 (1.15–3.61)* 0.38 (0.18–0.81)* 1.68 (0.70–3.99) Began smoking ≤16 years old 1.02 (0.63–1.64) 1.01 (0.60–1.70) 1.07 (0.60–1.88) 1.33 (0.81–2.19) First cigarette within 30 min of waking 0.50 (0.22–1.13) 3.35 (1.51–7.43)** 0.32 (0.16–0.64)*** 1.43 (0.85–2.39) Smokes mostly with friends 1.13 (0.67–1.90) 0.55 (0.31–0.98)* 1.07 (0.52–2.19) 0.65 (0.34–1.25) Cessation history and intentions Sure can quit for ≥30 days 1.35 (0.77–2.36) 0.72 (0.42–1.21) 0.95 (0.53–1.71) 0.50 (0.29–0.88)* Made a 24-hr quit attempt in past year 0.97 (0.54–1.74) 0.97 (0.58–1.62) 2.07 (1.14–3.74)* 1.06 (0.62–1.81) Plans to quit in next 30 days 1.78 (1.02–3.14)* 0.88 (0.51–1.50) 1.22 (0.65–2.28) 1.34 (0.76–2.36) Social environment More than half of friends smoke 0.85 (0.49–1.49) 1.88 (1.08–3.28)* 1.26 (0.68–2.35) 1.26 (0.76–2.07) Household smoking ban 1.74 (1.02–2.95)* 0.79 (0.45–1.37) 1.03 (0.58–1.83) 0.76 (0.47–1.24) Workplace smoking ban 0.83 (0.52–1.31) 0.77 (0.46–1.28) 1.10 (0.63–1.93) 0.73 (0.41–1.29)  ≤5 CPD (436 smokers, 532 transitions) 6–10 CPD (421 smokers, 499 transitions) Quit Increase Reduce or quit Increase AOR (95% CI) AOR (95% CI) AOR (95% CI) AOR (95% CI) Demographics/SES Female 1.39 (0.83–2.32) 0.73 (0.43–1.24) 1.55 (0.84–2.87) 0.55 (0.32–0.96)* Age >30 years 0.95 (0.55–1.65) 1.19 (0.66–2.16) 0.69 (0.38–1.27) 1.60 (0.89–2.86) White 0.94 (0.44–2.04) 3.06 (1.38–6.80)** 0.28 (0.14–0.56)*** 1.34 (0.69–2.60) Married 1.03 (0.90–1.18) 0.99 (0.85–1.16) 0.99 (0.85–1.15) 1.04 (0.90–1.21) Education ≤ high school 0.83 (0.47–1.45) 1.31 (0.75–2.28) 0.75 (0.41–1.40) 1.08 (0.65–1.82) Income >$20,000 1.08 (0.43–2.73) 1.66 (0.72–3.83) 2.93 (0.95–9.09) 1.18 (0.48–2.90) Income unknown 1.68 (0.49–5.81) 1.52 (0.50–4.63) 2.43 (0.52–11.35) 0.74 (0.22–2.44) Smoking characteristics Daily smoker 1.02 (0.58–1.81) 2.04 (1.15–3.61)* 0.38 (0.18–0.81)* 1.68 (0.70–3.99) Began smoking ≤16 years old 1.02 (0.63–1.64) 1.01 (0.60–1.70) 1.07 (0.60–1.88) 1.33 (0.81–2.19) First cigarette within 30 min of waking 0.50 (0.22–1.13) 3.35 (1.51–7.43)** 0.32 (0.16–0.64)*** 1.43 (0.85–2.39) Smokes mostly with friends 1.13 (0.67–1.90) 0.55 (0.31–0.98)* 1.07 (0.52–2.19) 0.65 (0.34–1.25) Cessation history and intentions Sure can quit for ≥30 days 1.35 (0.77–2.36) 0.72 (0.42–1.21) 0.95 (0.53–1.71) 0.50 (0.29–0.88)* Made a 24-hr quit attempt in past year 0.97 (0.54–1.74) 0.97 (0.58–1.62) 2.07 (1.14–3.74)* 1.06 (0.62–1.81) Plans to quit in next 30 days 1.78 (1.02–3.14)* 0.88 (0.51–1.50) 1.22 (0.65–2.28) 1.34 (0.76–2.36) Social environment More than half of friends smoke 0.85 (0.49–1.49) 1.88 (1.08–3.28)* 1.26 (0.68–2.35) 1.26 (0.76–2.07) Household smoking ban 1.74 (1.02–2.95)* 0.79 (0.45–1.37) 1.03 (0.58–1.83) 0.76 (0.47–1.24) Workplace smoking ban 0.83 (0.52–1.31) 0.77 (0.46–1.28) 1.10 (0.63–1.93) 0.73 (0.41–1.29)

Note.AOR, adjusted odds ratio; CPD, cigarettes per day; and SES, socioeconomic status.

*

p < .05;

**

p < .01;

***

p < .001.

Among smokers consuming 6–10 CPD, three factors were associated with a decreased likelihood of quitting or reducing smoking more than 2 years: being White, smoking daily, and smoking the first cigarette of the day within 30 min of waking. Smokers consuming 6–10 CPD who made a 24-hr quit attempt in the past year were more likely to reduce cigarette consumption or quit. Smokers consuming 6–10 CPD who were women or who expressed confidence that they could quit smoking for 30 consecutive days were less likely to increase their cigarette consumption over a 2-year period.

## Discussion

With low levels of cigarette smoking on the rise, it is essential to characterize light smokers (≤10 CPD) and understand their smoking trajectories. The appropriate public health and policy response depends on knowing the extent to which light smokers are increasing cigarette consumption on the way to established smoking, decreasing consumption on the way to quitting, or in a stable pattern of tobacco use. To target interventions appropriately, it is important to identify which light smokers are at risk of increasing their tobacco consumption and which light smokers are likely to reduce or quit.

This longitudinal study addressed these questions. We found that the natural history of smoking was much more fluid among those consuming no more than 10 CPD than it was among heavier smokers, but relatively few light smokers increased their consumption over the course of 4 years. Reducing cigarette consumption and quitting smoking together constituted the most common outcomes for light smokers. These findings are consistent with longitudinal studies of light smokers that allow a comparable analysis (Hassmiller et al., 2003; Hennrikus et al., 1996; Hyland et al., 2005; Lindstrom & Isacsson, 2002; McDermott et al., 2007; Stanton et al., 2007; Wetter et al., 2004; Zhu et al., 2003).

We focused special attention on smokers at the lowest levels of tobacco consumption by separately analyzing very light daily and nondaily smokers. The characteristics and natural history of very light smokers differed markedly according to whether or not they smoked daily. Very light nondaily smokers had a more stable smoking trajectory than did very light daily smokers, who were more likely to progress to higher and therefore riskier levels of cigarette consumption.

Our most important contribution is identifying factors associated with progression to higher levels of smoking. Among the very light smoker group, these factors included White race/ethnicity, having the first cigarette of the day within 30 min of waking (a well-accepted measure of nicotine dependence; Heatherton, Kozlowski, Frecker, & Fagerström, 1991; Heatherton et al., 1989), smoking more often alone than with others, and having a majority of friends who smoke. These data suggest that even very light smokers are at risk of progressing to heavier smoking if they have regular nicotine exposure and if they already have some evidence of nicotine dependence.

Individuals who most often smoke with friends may be less likely to increase their consumption because they have fewer opportunities to smoke (Moran, Wechsler, & Rigotti, 2004). However, very light smokers with a majority of friends who smoke may have a greater likelihood of increasing cigarette consumption because smoking is more visible in their social environment. Trotter, Wakefield, and Borland (2002) reported that social cues are instrumental in perpetuating smoking. It may be harder for smokers with many friends who smoke to avoid smoking cues, and they likely to face greater social reinforcement of their smoking behavior.

Among smokers consuming 6–10 CPD, there are fewer clear predictors of transitions to heavier smoking. Those who are not confident that they can quit smoking are likely expressing evidence of nicotine dependence (Siahpush, McNeill, Borland, & Fong, 2006). Thus, indirect evidence suggests that nicotine dependence may play a role in increasing cigarette consumption over time for those smoking no more than 5 CPD and for those smoking 6–10 CPD.

Predictors of cessation among those consuming no more than 5 CPD tell a slightly different story. Intention to quit in the next 30 days was the strongest predictor of quitting in this group, suggesting that little may be preventing these light smokers from quitting once they have decided to do so. Our findings also confirm that household bans are associated with smoking cessation among very light smokers just as they are in more general populations of smokers studied elsewhere (Farkas, Gilpin, Distefan, & Pierce, 1999).

Many factors associated with increased cigarette consumption among very light smokers—White race, daily smoking, first cigarette within 30 min of waking—were negatively associated with reducing cigarette consumption (including quitting) among smokers consuming 6–10 CPD, further highlighting the importance of these predictors in the establishment of heavier smoking patterns.

Our cross-sectional and longitudinal analyses found associations between White race/ethnicity and baseline smoking status as well as the likelihood of increasing or quitting/reducing cigarette consumption. This is consistent with published findings indicating that Black, Hispanic, Asian, and Native American smokers are more likely than White smokers to be occasional smokers (Evans et al., 1992; Hassmiller et al., 2003; Husten et al., 1998; Tong et al., 2006; Wortley et al., 2003). To our knowledge, no published studies have looked at race/ethnicity as a predictor of smoking status change among light smokers. We do not have sufficient sample size or proper covariates to ascertain the extent to which the White/non-White difference is social and cultural versus possibly biological.

The bivariate analyses confirmed prior work showing that non-White race/ethnicity, younger age, higher socioeconomic status (education and income), and certain smoking characteristics (smoking mostly with friends, starting to smoke later in life, having evidence of more nicotine dependence, and more recent and planned quitting activity) are defining characteristics of light smokers (Evans et al., 1992; Gilpin et al., 1997; Hassmiller et al., 2003; Husten et al., 1998; Lindstrom & Isacsson, 2002; McDermott et al., 2007; Tong et al., 2006; Wetter et al., 2004; Wortley et al., 2003).

Our findings need to be viewed in the context of certain limitations. Smoking status and cigarette consumption were obtained by self-report only. However, biochemical validation is not required for population surveys (Benowitz et al., 2002). Unobserved changes in smoking status may have occurred between survey waves. Our sample is drawn from one state and was comprised largely of Whites. Therefore, the results may not generalize to other geographic areas or to non-White smokers. The losses to follow-up at waves 2 and 3 provide a potential source of bias, especially for the 4-year follow-up findings. We attempted to reduce this potential bias by using survey weights for waves 2 and 3 designed to adjust for differential attrition across several dimensions, but this approach may not eliminate all biases. The consistency of our findings regarding smoking status stability and quit rates with other longitudinal studies provides some reassurance that the bias is not large.

This longitudinal study found that a large proportion of light smokers (smoking ≤10 CPD) maintained a static level of consumption more than 2 years, but over 4 years an increasing number further reduced their smoking or quit. Daily smokers consuming no more than 5 CPD are the group in the greatest flux; they are as likely as other light smokers to reduce or quit smoking but are more likely to increase consumption. Light smokers who do increase consumption tend to be those exhibiting signs of nicotine addiction and those in a social environment conducive to smoking. These findings suggest that although light smokers should be encouraged to quit or be offered treatment to aid in quitting, we can be reassured that only a minority of light smokers go on to heavier smoking.

## Funding

National Heart, Lung, and Blood Institute (HL04440); National Cancer Institute's Tobacco Research Initiative for State and Community Interventions (CA86257).

## Declaration of Interests

The authors have no competing interests to report in relation to the present study.

## References

Benowitz
N
Jacob
P
Ahijevych
K
Jarvis
M
Hall
S
LeHouezec
J
, et al.  .
Biochemical verification of tobacco use and cessation
Nicotine & Tobacco Research
,
2002
, vol.
4
(pg.
149
-
159
)
Biener
L
Albers
AB
Young adults: Vulnerable new targets of tobacco marketing
American Journal of Public Health
,
2004
, vol.
94
(pg.
326
-
330
)
Bjartveit
K
Tverdal
A
Health consequences of smoking 1-4 cigarettes per day
Tobacco Control
,
2005
, vol.
14
(pg.
315
-
320
)
Centers for Disease Control and Prevention.
Cigarette smoking among adults—United States, 1992, and changes in the definition of current cigarette smoking
MMWR Morbidity and Mortality Weekly Report
,
1994
, vol.
43
(pg.
342
-
346
)
Evans
J
Gilpin
E
Pierce
JP
Burns
D
Borland
R
Johnson
M
, et al.  .
Occasional smoking among adults: Evidence from the California Tobacco Survey
Tobacco Control
,
1992
, vol.
1
(pg.
169
-
175
)
Farkas
AJ
Gilpin
EA
Distefan
JM
Pierce
JP
The effects of household and workplace smoking restrictions on quitting behaviours
Tobacco Control
,
1999
, vol.
8
(pg.
261
-
265
)
Gilpin
E
Cavin
SW
Pierce
JP
Adult smokers who do not smoke daily
,
1997
, vol.
92
(pg.
473
-
480
)
Hassmiller
KM
Warner
KE
Mendez
D
Levy
DT
Romano
E
Nondaily smokers: Who are they?
American Journal of Public Health
,
2003
, vol.
93
(pg.
1321
-
1327
)
Heatherton
TF
Kozlowski
LT
Frecker
RC
Fagerström
KO
The Fagerström Test for Nicotine Dependence: A revision of the Fagerström Tolerance Questionnaire
,
1991
, vol.
86
(pg.
1119
-
1127
)
Heatherton
TF
Kozlowski
LT
Frecker
RC
Rickert
W
Robinson
J
Measuring the heaviness of smoking: Using self-reported time to the first cigarette of the day and number of cigarettes smoked per day
,
1989
, vol.
84
(pg.
791
-
799
)
Hennrikus
DJ
Jeffery
RW
Lando
HA
Occasional smoking in a Minnesota working population
American Journal of Public Health
,
1996
, vol.
86
(pg.
1260
-
1266
)
Husten
CG
McCarty
MC
Giovino
GA
Chrismon
JH
Zhu
B
Intermittent smokers: A descriptive analysis of persons who have never smoked daily
American Journal of Public Health,
,
1998
, vol.
88
(pg.
86
-
89
)
Hyland
A
Rezaishiraz
H
Bauer
J
Giovino
GA
Cummings
KM
Characteristics of low-level smokers
Nicotine & Tobacco Research
,
2005
, vol.
7
(pg.
461
-
468
)
Kawachi
I
Colditz
GA
Stampfer
MJ
Willett
WC
Manson
JE
Rosner
B
, et al.  .
Smoking cessation and time course of decreased risks of coronary heart disease in middle-aged women
Archives of Internal Medicine
,
1994
, vol.
154
(pg.
169
-
175
)
Lindstrom
M
Isacsson
SO
Long term and transitional intermittent smokers: A longitudinal study
Tobacco Control
,
2002
, vol.
11
(pg.
61
-
67
)
Luoto
R
Uutela
A
Puska
P
Occasional smoking increases total and cardiovascular mortality among men
Nicotine & Tobacco Research
,
2000
, vol.
2
(pg.
133
-
139
)
McDermott
L
Dobson
A
Owen
N
Occasional tobacco use among young adult women: A longitudinal analysis of smoking transitions
Tobacco Control
,
2007
, vol.
16
(pg.
248
-
254
)
Moran
S
Wechsler
H
Rigotti
NA
Social smoking among US college students
Pediatrics
,
2004
, vol.
114
(pg.
1028
-
1034
)
Okuyemi
KS
Harris
KJ
Scheibmeir
M
Choi
WS
Powell
J
Ahluwalia
JS
Light smokers: Issues and recommendations
Nicotine & Tobacco Research
,
2002
, vol.
4

Suppl. 2
(pg.
S103
-
S112
)
Orzechowski & Walker
The tax burden on tobacco: Historical compilation, 2002, Vol. 37
,
2003
Arlington, VA
Orzechowski & Walker
Owen
N
Kent
P
Wakefield
M
Roberts
L
Low-rate smokers
Preventive Medicine
,
1995
, vol.
24
(pg.
80
-
84
)
Porter
S
Jackson
K
Trosclair
A
Pederson
L
Prevalence of current cigarette smoking among adults and changes in prevalence of current and some day smoking—United States, 1996–2001
MMWR Morbidity and Mortality Weekly Report
,
2003
, vol.
52
(pg.
303
-
304
306–307
Prescott
E
Scharling
H
Osler
M
Schnohr
P
Importance of light smoking and inhalation habits on risk of myocardial infarction and all cause mortality. A 22 year follow up of 12 149 men and women in the Copenhagen City Heart Study
Journal of Epidemiology and Community Health
,
2002
, vol.
56
(pg.
702
-
706
)
Rock
V
Malarcher
A
Kahende
J
Asman
K
Husten
C
Caraballo
R
Cigarette smoking among adults—United States, 2006
MMWR Morbidity and Mortality Weekly Report
,
2007
, vol.
56
(pg.
1157
-
1161
)
Rosengren
A
Wilhelmsen
L
Wedel
H
Coronary heart disease, cancer and mortality in male middle-aged light smokers
Journal of Internal Medicine
,
1992
, vol.
231
(pg.
357
-
362
)
Shiffman
S
Tobacco “chippers”—Individual differences in tobacco dependence
Psychopharmacology
,
1989
, vol.
97
(pg.
539
-
547
)
Shiffman
S
Kassel
JD
Paty
J
Gnys
M
Zettler-Segal
M
Smoking typology profiles of chippers and regular smokers
Journal of Substance Abuse
,
1994
, vol.
6
(pg.
21
-
35
)
Siahpush
M
McNeill
A
Borland
R
Fong
GT
Socioeconomic variations in nicotine dependence, self-efficacy, and intention to quit across four countries: Findings from the International Tobacco Control (ITC) Four Country Survey
Tobacco Control
,
2006
, vol.
15

Suppl 3
(pg.
iii71
-
iii75
)
Stanton
CA
Papandonatos
G
Lloyd-Richardson
EE
Niaura
R
Consistency of self-reported smoking over a 6-year interval from adolescence to young adulthood
,
2007
, vol.
102
(pg.
1831
-
1839
)
Tong
EK
Ong
MK
Vittinghoff
E
Perez-Stable
EJ
American Journal of Preventive Medicine
,
2006
, vol.
30
(pg.
23
-
30
)
Trotter
L
Wakefield
M
Borland
R
Socially cued smoking in bars, nightclubs, and gaming venues: A case for introducing smoke-free policies
Tobacco Control
,
2002
, vol.
11
(pg.
300
-
304
)
U.S. Department of Health and Human Services
Preventing tobacco use among young people: A report of the surgeon general
,
1994
Atlanta, GA
U.S. Department of Health and Human Services, Public Health Service, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health
U.S. Department of Health and Human Services
The health consequences of smoking: A report of the surgeon general
,
2004
Washington, DC
U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health
Wetter
DW
Kenford
SL
Welsch
SK
Smith
SS
RT
Fiore
MC
, et al.  .
Prevalence and predictors of transitions in smoking behavior among college students
Health Psychology
,
2004
, vol.
23
(pg.
168
-
177
)
Wortley
PM
Husten
CG
Trosclair
A
Chrismon
J
Pederson
LL
Nondaily smokers: A descriptive analysis
Nicotine & Tobacco Research
,
2003
, vol.
5
(pg.
755
-
759
)
Zhu
SH
Sun
J
Hawkins
S
Pierce
J
Cummins
S
A population study of low-rate smokers: Quitting history and instability over time
Health Psychology
,
2003
, vol.
22
(pg.
245
-
252
)