Addictive Behaviors 32 (2007) 1031 – 1042 The Heaviness of Smoking Index as a predictor of smoking cessation in Canada Michael O. Chaiton a,⁎, Joanna E. Cohen a,b , Paul W. McDonald b,c , Susan J. Bondy a,b a c Department of Public Health Sciences, University of Toronto, Canada b Ontario Tobacco Research Unit, Canada Population Health Research Group and the Department of Health Studies and Gerontology, University of Waterloo, Canada Abstract Nicotine addiction is believed to be a major impediment for many people in quitting smoking, but measures of nicotine dependence such as the Heaviness of Smoking Index (HSI) have had mixed success in predicting cessation. Using the National Population Health Survey, the relationship between HSI at baseline in cycle 2 (1996–1997) and successful smoking cessation at cycle 3 (1998–1999) and cycle 4 (2000–2001) was examined in 2938 Canadian adult smokers. A logistic regression model was developed for HSI as a predictor of smoking cessation, and then tested for interaction and confounding. The odds ratio of not smoking in cycle 3 was 2.08 (95% CI: 1.51, 2.86; p b 0.001) for low HSI (b2) compared to medium HSI. When the period of follow-up was extended, individuals with both high (N 4) HSI scores (OR 2.16; 95% CI: 1.11, 4.21; p = 0.02) and low scores (OR 2.22; 95% CI: 1.41, 3.49) had higher odds of not smoking at both cycle 3 and cycle 4 than those with medium HSI scores. The likelihood of reporting cessation was higher than expected in the Canadian population among highly dependent smokers, particularly among older smokers, those with middle or greater income adequacy, and those with no intention to quit smoking. There were no substantial changes to the results when those lost-to-follow-up were treated as continuing smokers. These findings indicate that nicotine dependence is only one factor in succeeding at a quit attempt. Individual and population strategies for smoking cessation may need to consider other influences such as cognitive, affective and environmental factors. © 2006 Elsevier Ltd. All rights reserved. Keywords: Dependence; Smoking cessation; Population study ⁎ Corresponding author. The Ontario Tobacco Research Unit, 33 Russell St., Toronto, Ontario, Canada M5S 2S1. E-mail address: [email protected] (M.O. Chaiton). 0306-4603/$ - see front matter © 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.addbeh.2006.07.008 1032 M.O. Chaiton et al. / Addictive Behaviors 32 (2007) 1031–1042 1. Introduction A variety of researchers and clinicians have speculated that the current smoking population in Canada and other developed countries may be becoming more and more resistant to quitting smoking (Fagerstrom et al., 1996; Jarvis, Wardle, Waller, & Owen, 2003). It is believed that smokers who are less addicted to nicotine quit smoking at greater rates, leaving a larger proportion of heavily addicted (so-called “hard core”) smokers in the population (Breslau, Johnson, Hiripi, & Kessler, 2001; Fagerstrom et al., 1996; Hughes, 2001; Irvin & Brandon, 2000; Niaura, Goldstein, & Abrams, 1994; Warner & Burns, 2003). If this hypothesis is true, it would have significant implications for the development of a future population strategy for smoking cessation. Therefore, understanding potential changes in the level of nicotine dependence in the population, and its relation to quitting, is crucial. Nicotine dependence is a distinct concept from the act of cigarette smoking and refers to the propensity of an individual's need for nicotine. Some conceptions of dependence suggest that it may reflect a physiological-based requirement implicit in an individual (Kemmeren, van Poppel, Verhoef & Jarvis, 1994). Studies within families have demonstrated a genetic influence on nicotine dependence (Henningfield, 1990; Kendler, 1999; Niu et al., 2000). The most common clinical measure of nicotine dependence is the Fagerstrom Tolerance Questionnaire (Fagerstrom, 1978; Fagerstrom & Schneider, 1989; Heatherton, Kozlowski, Frecker, & Fagerstrom, 1991; Heatherton, Kozlowski, Frecker, Rickert, & Robinson, 1989). While it is a relatively brief tool for clinical assessment, its length is still a problem for population-based surveys. Therefore, the Heaviness of Smoking Index (HSI) was developed in 1989 as a test to measure the same construct by using two questions from the Tolerance Questionnaire and the Fagerstrom Test for Nicotine Dependence: time to first smoking in the morning and number of cigarettes per day (Kozlowski, Porter, Orleans, Pope, & Heatherton, 1994). These questions were shown to have accounted for most of the predictive value of the Fagerstrom questionnaire and have been validated as providing similar results (Etter, Duc, & Perneger, 1999; Heatherton et al., 1989; Kozlowski et al., 1994). The Fagerstrom tests are purported to proxy the physical addiction to nicotine and not to correspond to compulsive or habitual smoking (Dijkstra & Tromp, 2002; Fagerstrom, 1978). While these measures of nicotine dependence have been shown to be significantly associated with smoking cessation in many studies (de Leon et al., 2003; Fagerstrom & Schneider, 1989; Heatherton et al., 1991; Heatherton et al., 1989; Kozlowski et al., 1994; Payne, Smith, McCracken, McSherry, & Antony, 1994), a number of studies have found either no relationship or one that is less predictive of quitting than number of cigarettes per day (Dijkstra & Tromp, 2002; Frikart, Etienne, Cornuz, & Zellweger, 2003; McDonald, 2003; Salive et al., 1992; Wetter et al., 1994). All else being equal, it can be assumed that the connection between increasing levels of dependence and smoking cessation would be clear; however, dependence is only one factor in making, and succeeding, at a quit attempt. While the physical dependence model of nicotine dependence is widely accepted, many questions remain as to the nature of its definition, and importance with respect to other models; nevertheless, a common trait between competing theoretical models is that dependence entails a difficulty in achieving abstinence (Kenford et al., 2002). However, it is not clear how well measures of dependence adequately conceptualize dependence. Attempts to link measures of physical dependence with mechanisms that could explain the ability or the inability to quit smoking have met with limited success (Kenford et al., 2002). The purpose of this paper is to examine whether there is a relationship between selfreport nicotine dependency and the likelihood of subsequent smoking cessation in the general population. M.O. Chaiton et al. / Addictive Behaviors 32 (2007) 1031–1042 1033 2. Method 2.1. Design This study used the longitudinal panel data set from the population-based National Population Health Survey (NPHS), cycle 2 (1996–1997), cycle 3 (1998–1999), and cycle 4 (2000–2001). The original sample, cycle 1 (1994–1995), was not used as it did not measure both items needed to calculate HSI. Respondents from 19,600 households were recruited in 1994–1995 by random digit dialling, stratified to ensure adequate provincial coverage and sampled to reflect the general Canadian population (Tambay & Catlin, 1995). This survey is representative of the Canadian population. The complex, two-stage sampling design and longitudinal follow-up procedures have been described elsewhere (Swain, Catlin, & Beaudet, 1999; Tambay & Catlin, 1995). In all, 17,626 people completed the first cycle of the survey; the total national household response rate for completing at least one questionnaire was 88.7% of all in-scope households (Health Canada, 2000). The target population of the NPHS was household residents in all provinces, and excludes persons on Indian Reserves, Canadian Forces Bases, some remote areas in Quebec and Ontario, the Yukon, Nunavut, Northwest Territories, and long-term residents of hospitals and residential care facilities. 2.2. Subjects The study population included 3453 people identified as daily smokers, 18 years old and older in cycle 2 of the NPHS (1996–1997). Of this population, 426 daily smokers in cycle 2 were lost to follow-up in cycle 3 (1998–1999). Eighty-nine other people were excluded for incomplete information. Thus, data for 2938 daily adult smokers were available for the analysis, representing 85% of those eligible at cycle 2 (baseline for this study). 2.3. Measurement 2.3.1. Cessation Smoking cessation at follow-up was classified as a dichotomous outcome. People identified as daily smokers in cycle 2 who reported “Not at all” to the question, “At the present time, do you smoke cigarettes daily, occasionally, or not at all?” and who had not reported smoking in the past 30 days in cycle 3 were considered to have successfully achieved smoking cessation. For longer-term follow-up, smoking cessation was defined as self-report not smoking at both cycle 3 and cycle 4. 2.3.2. Nicotine dependence Nicotine dependence was assessed by the two questions from the HSI. The HSI has been shown to be a reasonably reliable and valid measure of nicotine dependence, associated with physiological measures of dependence such as carbon monoxide and cotinine levels (de Leon et al., 2003; Etter et al., 1999). The HSI has been found to identify a similar dependent population to the Fagerstrom Tolerance Questionnaire (FTQ) and the Fagerstrom Test for Nicotine Dependence (FTND) (Heatherton et al., 1989; Kozlowski et al., 1994). It is a six-point scale calculated from the number of cigarettes smoked per day (1–10, 11–20, 21–30, 31+ cigarettes) and the time to first cigarette after waking (≤5, 6–30, 1034 M.O. Chaiton et al. / Addictive Behaviors 32 (2007) 1031–1042 31–60, and 61 + min) (Heatherton et al., 1989). Nicotine dependence was then categorized into a threecategory variable: low (0–1), medium (2–4) and high (5–6). 2.3.3. Other variables There is reason to suggest that the HSI may be sensitive to a number of characteristics that may confound or mediate the relationship between nicotine dependence and smoking cessation, such as demographic factors (age, gender), differences in access, resources and knowledge (education, income), related physiological factors (depression, alcohol dependence) or environmental factors (smoking restrictions at home and at work). These potential confounders and mediators were obtained from cycle 2 data using standard measures available in the NPHS. Income adequacy was defined using standards developed by Statistics Canada based on annual total household income and the number of persons in the household and categorized into low, low middle, middle, high middle, and high groups which reflect different standards of living rather than income quintiles (Health Canada, 2000). Education was set as the highest level of education achieved at the time of cycle 2. To describe the subject's state of depression, results from the short form of the Composite International Diagnostic Interview (CIDI) to measure major depressive episode were used (Health Canada, 2000). Having current symptoms of depression and negative affect has associated with continued smoking (Covey, 2004). The NPHS measure, which examined recent history of depression over the previous 6 months, was found to be very sensitive (97%) but had modest specificity (72%) (Patten, 1997). The NPHS measure of alcohol dependence was adapted from the Kessler and Mroczek series on the CIDI 12-month Short Screening Scales (Health Canada, 2000). Self-reported restrictions on smoking at home (smoking in the home vs. no smoking in the home) and at work (complete restrictions vs. some or no restrictions on smoking) were also included. 2.4. Statistical analysis 2.4.1. Multivariable analyses Logistic regression was used to predict reported not smoking in cycle 3 with the HSI score at cycle 2, initially unadjusted and then with the covariates included. Odds ratios for HSI, 95% confidence intervals and p-values were reported. Confounding was considered to be present if potential confounders resulted in at least a 10% change in either beta estimate of HSI, but all potential confounders were left in the model regardless of the p-value or change in beta. Interaction effects involving HSI and other covariates were assessed using cross-product interaction terms and interaction was deemed to be present if the term significantly improved the fit of the model (5% alpha). Potential interactions were also examined using stratified models. All regression models were repeated using sustained non-smoking status at both cycles 3 and 4 as the outcome variable. 2.4.2. Missing respondents Missing observations are not explicitly noted where 5% or fewer responses were missing for a particular item. For those items, missing observations were excluded from the analysis. For most variables, the number of missing values was small. For items with greater than 5% of respondent answers missing or coded not applicable (income, smoking restrictions at work), a dummy variable was assigned to the missing data and the observations were included in the analysis. The frequency of missing item responses for each variable was studied to determine if those lost to follow-up were different from those who completed data at all time points, by assessing the effect of coding all lost to follow-up as continuing smokers or quitters. M.O. Chaiton et al. / Addictive Behaviors 32 (2007) 1031–1042 1035 2.4.3. Variance estimation To account for the complex multistage sampling design, the Bootstrap Replicating Reweighting utility was used to calculate all estimates of variance and p-values. This utility uses a set of resampling weights provided by Statistics Canada specifically for the longitudinal NPHS data set and routines to obtain bootstrapped variance estimates (Health Canada, 2000). All analyses were weighted. 3. Results Of the survey participants included in this study, 47% were older than 40 years of age, 53% were males, 19% reported below middle income adequacy, 54% had formal education beyond a high school degree (e.g. at least some technical college or university), and 46% reported an intention to quit within the next Table 1 Characteristics of adult (≥ 18 years) daily smokers in cycle 2 (1996–1997), NPHS Percent Age 18–40 years 41 + Gender Female Male Income adequacy Lowest Low middle Middle High middle High Missing Education Less than high school High school diploma Some college/university College or university degree Intention to quit in next 6 months Yes No Number of cigarettes smoked per day 0–10 11–20 21–30 31 + Time to first cigarette 0–5 min 6–30 min 31–60 min N1 h n = 2938. Analytic weights were applied. 53 47 47 53 6 13 29 37 10 6 28 18 29 25 46 54 24 45 27 4 24 36 19 21 1036 M.O. Chaiton et al. / Addictive Behaviors 32 (2007) 1031–1042 Fig. 1. Distribution of HSI score among 2938 adult (older than 18 years) daily smokers in cycle 2 (1996–97) of the NPHS. 6 months (Table 1). Almost a majority (45%) of daily smokers reported smoking between 11 and 20 cigarettes per day, while the most commonly reported (36%) time to first cigarette was between 6 and 30 min. The proportion of less dependent smokers (HSI b 2) was 25% while the proportion of heavy dependent smokers (HSI N 4) was 13.5% (Fig. 1). Individuals with a low HSI score (HSI b 2) were more likely to report not smoking at follow-up; however, there did not appear to be a linear relationship between HSI scores and cessation (Fig. 2). Of the sample, a total of 13.3% (n = 341) reported not smoking in cycle 3 and 9.6% of the cycle 2 sample reported not smoking in both cycle 3 and cycle 4 (n = 197). For the regression model predicting not smoking at cycle 3, those with low HSI scores were more likely to quit than those with medium HSI scores (OR: 2.26; 95% CI: 1.65, 3.05; p b 0.001) only (Table 2). After including covariates, the odds ratio of not smoking was 2.08 (95% CI: 1.51, 2.86; p b 0.001) for the low HSI group compared to the medium group and 1.54 (95% CI: 0.91, 2.57; p = 0.102) for the high HSI group compared to the medium HSI group (Table 2). Only “household restrictions on smoking” met the criterion for a confounder, decreasing the estimate of HSI by 12%. None of the interaction terms were significant. Time to first cigarette, number of cigarettes per day and the cross-product interaction term for those variables were then used in the regression instead of HSI. All terms–time to first (OR: 0.75; 95% CI: 0.60, 0.93; p = 0.010), number of cigarettes per day (OR: 0.60; 95% CI: 0.41, 0.88; p = 0.008) and the Fig. 2. Percent of adult (older than 18 years) daily smokers in cycle 2 (1996–97) reporting not smoking at follow-up at cycle 3 (1998–99) by HSI score at baseline of the NPHS. M.O. Chaiton et al. / Addictive Behaviors 32 (2007) 1031–1042 1037 Table 2 Predictors of reported not smoking in cycle 3 (1998–1999) using baseline characteristics of adult (≥ 18 years) daily smokers at cycle 2 (1996–1997), NPHS Model 1 High HSI (N4) Medium HSI Low HSI (b 2) Age (in decades) Male gender Low/low middle income adequacy Completed high school or less Intention to quit Probability of alcohol dependence Probability of depression No smoking in the house Complete smoking restrictions at work Model 2 OR 95% CI p-value 1.60 Referent 2.26 0.95, 2.70 0.075 1.65, 3.05 – – – – – – – – – b 0.001 OR 95% CI p-value 1.54 Referent 2.08 1.17 1.07 1.75 0.87 1.21 0.76 1.63 1.86 0.63 0.91, 2.57 0.102 1.51, 2.86 1.05, 1.30 0.77,1.48 1.25, 2.46 0.57, 1.31 0.89, 1.65 0.27, 2.17 0.97, 2.73 1.12, 3.11 0.40, 0.99 0.001 0.004 0.682 0.001 0.493 0.231 0.612 0.063 0.017 0.043 n = 2938. Analytic weights were applied. interaction term of the two variables (OR: 1.24; 95% CI: 1.03, 1.50; p = 0.026)–were significant. This suggested, when calculated, that among those smoking 20 or fewer cigarettes per day, the longer the period after waking until having the first cigarette, the greater the probability of not smoking at follow-up, as would be expected. However, among smokers of more than 20 cigarettes per day, the longer the period after waking, the lower the probability of reporting not smoking. To examine long-term cessation, the HSI score was used to predict not smoking at both cycle 3 and cycle 4 (Table 3). The magnitude of the odds ratio for reported not smoking for people with low HSI scores was roughly equivalent for predicting not smoking at cycle 3 and both cycle 3 and cycle 4 (2.08 and 2.22, respectively). People with high HSI scores were significantly more likely to report not smoking at both cycle 3 and cycle 4 than people with medium scores (OR 2.16; 95% CI: 1.11, 4.21; p = 0.02), a relationship that was not significant when not smoking at cycle 3 only was the outcome. Table 3 HSI at cycle 2 (1996–1997) as a predictor of reported not smoking at cycle 3 (1998–1999) and cycle 4 (2000–2001) among adult (≥ 18 years) daily smokers, NPHS Lost to follow-up excluded, n = 2579 High HSI (N4) Medium HSI Low HSI (b 2) OR 95% CI 2.16 Referent 2.22 1.11, 4.21 0.023 1.41, 3.49 0.001 0.99, 3.68 0.054 1.35, 2.14 b 0.001 Lost to follow-up included as continued smokers, n = 3337 High HSI (N4) 1.91 Medium HSI Referent Low HSI (b 2) 2.06 p-value Analytic weights were applied. Adjusted for age, gender, income adequacy, education, intention to quit, alcohol dependence, depression, and restrictions in the home and at work. 1038 M.O. Chaiton et al. / Addictive Behaviors 32 (2007) 1031–1042 Table 4 HSI at cycle 2 (1996–1997) as a predictor of reported not smoking at cycle 3 (1998–1999) and cycle 4 (2000–2001), stratified for age, gender, income adequacy and intention to quit among adult (≥18 years) daily smokers, NPHS Age b 41 years, n = 1476 ≥ 41 years, n = 1462 Gender Female, n = 1495 Male, n = 1443 Income adequacy Less than middle, n = 685 Middle or greater, n = 2093 Intention to quit No intention to quit, n = 1601 Intention to quit, n = 1337 High HSI (N4) Medium HSI Low HSI (b 2) High HSI (N4) Medium HSI Low HSI (b 2) High HSI (N4) Medium HSI Low HSI (b 2) High HSI (N4) Medium HSI Low HSI (b 2) High HSI (N4) Medium HSI Low HSI (b 2) High HSI (N4) Medium HSI Low HSI (b 2) High HSI (N4) Medium HSI Low HSI (b 2) High HSI (N4) Medium HSI Low HSI (b 2) OR 95% CI p-value 0.81 Referent 1.73 3.46 Referent 2.69 1.59 Referent 1.65 2.55 Referent 2.89 0.23 Referent 3.70 2.77 Referent 2.05 3.25 Referent 4.16 1.16 Referent 1.35 0.01, 65.01 0.925 0.83, 3.62 1.65, 7.25 0.146 0.001 1.42, 5.11 0.52, 4.83 0.002 0.411 0.78, 3.49 0.90, 7.22 0.193 0.079 1.43, 5.83 0.00, 312.77 0.003 0.688 1.29, 10.66 1.38, 5.55 0.015 0.004 1.25, 3.37 1.13, 9.41 0.005 0.029 2.04, 8.45 0.34, 3.94 0.000 0.809 0.75, 2.42 0.316 n = 2579. Analytic weights were applied. Adjusted for age, gender, income adequacy, education, intention to quit, alcohol dependence, depression, and restrictions in the home and at work. The finding that people with high HSI scores had a greater odds of quitting than those with medium scores was also evident for some subgroups in the stratified analysis, particularly for smokers greater than 40 years of age, those with middle or greater income adequacy, and those with no intention to quit smoking (Table 4). There was a suggestion of a similar relationship among males ( p b 0.10). There was some indication of a linear relationship between not smoking at follow-up and HSI score among people less than 41 years of age and those with less than middle-income adequacy. However, there were very small numbers of younger and low-income high HSI smokers who reported not smoking at follow-up; estimates were unstable with very large confidence intervals. For gender and intention to quit, there was little difference in the odds of quitting between the high HSI and low HSI groups, compared to the medium HSI group, although the magnitude of the difference between the groups varied. The magnitude of the association of HSI and smoking cessation was essentially unchanged when those lost to follow-up were included as continued smokers. The odds of reporting not smoking dropped from 2.26 to 1.93 (0.86, 2.39) for high HSI with a sample size of 3337 when those lost to follow-up were reassigned. For not smoking at cycles 3 and 4, the estimate for high HSI smokers dropped from 2.16 to 1.91 (0.99, 3.68) (see Table 3). M.O. Chaiton et al. / Addictive Behaviors 32 (2007) 1031–1042 1039 4. Discussion This study found that HSI is associated with reported changes in smoking status in Canada. Smokers who had low HSI scores were more likely to have reported quitting at follow-up. However, unexpectedly, the data did not indicate a constant negative linear relationship between increasing HSI scores and a decreasing probability of cessation at follow-up. Smokers with high HSI scores were more likely to report not smoking at both cycle 3 and cycle 4 than smokers with medium scores. While unexpected, the difficulty in distinguishing the cessation patterns between medium and high dependent smokers in this data set is consistent with findings from earlier studies that used HSI as a linear measure of dependence. In a population recruited into a stop smoking program, Kozlowski et al. (1994) found that the scale poorly predicted smoking cessation above an HSI score of 4. In their study, as in the current analysis, the length of follow-up seemed to have an effect on the relationship between HSI and smoking cessation. Smokers with lower HSI scores quit earlier than those with higher scores, but the long term cessation rates were similar (Kozlowski et al., 1994). Coambs, Li and Kozlowski (1992) were also able to demonstrate a constant negative linear relationship between HSI and quitting only when examining age subgroups. Similarly, in this study, while people reporting low levels of nicotine dependence as measured by HSI were most consistently likely to report quitting, highly nicotine dependent smokers, particularly those older and wealthier, also reported a substantial likelihood of cessation. Heavier smokers have been found to be more likely to use pharmacotherapies or otherwise seek help in quitting smoking, actions that may improve the likelihood of quitting (McDonald, 2003; Pierce & Gilpin, 2002). While these help-seeking behaviors were not evaluated here, they would warrant future investigation as a mechanism in linking dependence to quitting behaviors. However, in California, over this time period, less than 18% of quitters used a nicotine replacement product to quit (Pierce and Gilpin, 2002). There may be other factors such as access to care, which may be proxied by socioeconomic status, that are potentially mediating factors in the relationship between dependence and cessation; however, further research is required to delineate specifically the effects of these other factors. In other research, it has been found that for the majority of smokers, dependence is only one factor in making, and succeeding at, a quit attempt with cognitive, affective, and environmental influences being other important factors (McDonald, 2003; Kenford et al., 2002). Constructs of dependence as merely a physiological effect, such as those measures derived from self-administration patterns are inherently limited in their ability to capture ability of the individual to stop smoking. Heatherton et al. (1991) suggest that the question examining whom has greater difficulty in quitting is different from examining whom is less likely to quit. While the heaviness of smoking can be considered the degree to which an individual is attached to cigarette smoking, this conception of dependence may be only partially related to actual chances of quitting. Other models may better explain the likelihood of cessation. For instance, the affective model of nicotine dependence, where nicotine dependence is related to individual capacity to experience negative affect and an individual's expectations that quitting nicotine would improve this affect, is a better predictor of smoking cessation than a physical model alone (Kenford et al., 2002). Kenford et al. (2002) directly compared the ability of a physical model of nicotine addiction, including the FTQ and blood cotinine levels, against an affective model of dependence. While both models were predictive of cessation independently, the affective model, specifically post-quit negative affect, was the most powerful predictor of success, and accounted for much of the predictive validity of the physical dependence measures. This suggests that HSI may not be a good scale of dependence, particularly in general populations. While the 1040 M.O. Chaiton et al. / Addictive Behaviors 32 (2007) 1031–1042 HSI does appear to be a proxy for a certain set of physical nicotine dependence measures and has been shown to be useful within a clinical setting, the meaningfulness of the scale to predict smoking cessation in a general population may be limited. As such, these data do not support the suggestion that falling prevalence will necessarily entail a greater proportion of smokers with high nicotine dependence, at least if dependence is measured by HSI. There are other indications that dependence has not been increasing in Canada, despite a drop in smoking prevalence. Daily cigarette consumption levels are dropping and increasing numbers of heavy and medium smokers report intentions to quit in the near future (Health Canada, 2002). In 2001, only 10% of current smokers could be classified as highly dependent, whose continued smoking has been suggested to be primarily nicotine-driven (Fagerstrom et al., 1996). However, there are a number of limitations of this study. The NPHS does not collect detailed smoking histories, so smokers may have quit and relapsed between interviews. Since the survey only confirms smoking cessation within the past 30 days, individuals, potentially lighter smokers, who quit for significant periods of time and relapse more often may be more likely to be erroneously identified as having quit. However, errors associated with self-report smoking cessation were found to be small in absolute terms in a large cohort, particularly in a cohort not specifically selected for tobacco use such as this general health survey (Clark, Gautam, Hlaing, & Gerson, 1996; Murray, Connett, Istvan, Nides, & Rempel-Rossum, 2002; Pierce & Gilpin, 2003). While there was considerable attrition of completed surveys by cycle 4 (21.9%), treating those lost to follow-up as continued smokers did not affect the interpretation of the results. HSI as a measure for assessing nicotine dependence may have significant floor effects among lighter smokers seen in non-clinical populations and the validity of the measure for assessing dependence in a general population may be questioned. Nevertheless, HSI and similar measures of dependence such as the Fagerstron Tolerance Questionnaire (FTQ) or the Fagerstrom Test for Nicotine Dependence (FTND) are commonly used in both clinical and population surveys to assess dependence (Etter et al., 1999; Etter, Vu, & Perneger, 2000). The significant interaction between cigarettes per day and time to first cigarette also suggests that caution should be used when interpreting the effect of these measures when used independently. The reliability of the measure for assessing dependence may be questionable and a short valid tool suitable for accurately measuring dependence in the general population is urgently needed. This study adds to existing knowledge of the relationship of nicotine dependence and smoking using a commonly used and validated measure of dependence, HSI, with clinical relevance. The large, population-based sample was drawn to be generalizable to the Canadian population and includes high follow-up rates over a long-term period. Few other studies examine smoking cessation with a follow-up period of greater than 4 years. Under some circumstances, heavily dependent smokers, as measured by HSI, quit at rates similar to those of other smokers. The HSI score, on its own, does not adequately identify smokers who are not able to quit. When age or income was also considered, this study does identify subgroups with very low reported odds of not smoking, but the HSI may be less effective for explaining quitting behavior for a general population. Medium dependent smokers are the largest group of smokers in Canada and this study found that they have the lowest overall likelihood of quitting. Quit rates vary substantially between demographic groups and by level of dependence and understanding the mediating and moderating factors that influence the relationship would be key to developing effective interventions to increase quitting in populations where prevalence has seen substantial declines. A different conceptualization of dependence relying on more than self-administration patterns may better capture the likelihood of quitting in a population. M.O. Chaiton et al. / Addictive Behaviors 32 (2007) 1031–1042 1041 Acknowledgements Support for Michael O. Chaiton was provided by the Canadian Institutes of Health Research Strategic Training Program for Tobacco Research. Funding for Paul W. McDonald came, in part, from the Heart and Stroke Foundation of Ontario, Grant Number HBR 4858. References Breslau, N., Johnson, E. O., Hiripi, E. O., & Kessler, R. (2001). Nicotine dependence in the United States: Prevalence, trends and smoking persistence. Archives of General Psychiatry, 58(9), 810−816. Clark, P. I., Gautam, S. P., Hlaing, W. M., & Gerson, L. W. (1996). Response error in self-reported current smoking frequency by black and white established smokers. Annals of Epidemiology, 6(6), 483−489. Coambs, R. B., Li, S., & Kozlowski, L. T. (1992). Age interacts with heaviness of smoking in predicting success in cessation of smoking. American Journal of Epidemiology, 135(3), 240−246. Covey, L. (2004). Comments on “History of depression and smoking cessation outcome: A meta-analysis”. Nicotine and Tobacco Research, 6(4), 743–745; author reply 747–749; discussion 751–742. de Leon, J., Diaz, F. J., Becona, E., Gurpegui, M., Jurado, D., & Gonzalez-Pinto, A. (2003). Exploring brief measures of nicotine dependence for epidemiological surveys. Addictive Behaviors, 28(8), 1481−1486. Dijkstra, A., & Tromp, D. (2002). Is the FTND a measure of physical as well as psychological tobacco dependence? Journal of Substance Abuse Treatment, 23(4), 367−374. Etter, J. F., Duc, T. V., & Perneger, T. V. (1999). Validity of the Fagerstrom test for nicotine dependence and of the heaviness of smoking index among relatively light smokers. Addiction, 94(2), 269−281. Etter, J. F., Vu, D. T., & Perneger, T. V. (2000). Saliva cotinine levels in smokers and nonsmokers. American Journal of Epidemiology, 151(3), 251−258. Fagerstrom, K. O. (1978). Measuring degree of physical dependence to tobacco smoking with reference to individualization of treatment. Addictive Behaviors, 3(3–4), 235−241. Fagerstrom, K. O., Kunze, M., Schoberberger, R., Breslau, N., Hughes, J. R., Hurt, R. D., et al. (1996). Nicotine dependence versus smoking prevalence: Comparisons among countries and categories of smokers. Tobacco Control, 5(1), 52−56. Fagerstrom, K. O., & Schneider, N. G. (1989). Measuring nicotine dependence: A review of the Fagerstrom tolerance questionnaire. Journal of Behavioral Medicine, 12(2), 159−182. Frikart, M., Etienne, S., Cornuz, J., & Zellweger, J. P. (2003). Five-day plan for smoking cessation using group behaviour therapy. Swiss Medical Weekly, 133(3–4), 39−43. Health Canada (2000). National population health survey documentation. Health Canada (2002). Canadian tobacco use monitoring survey. Heatherton, T. F., Kozlowski, L. T., Frecker, R. C., & Fagerstrom, K. O. (1991). The Fagerstrom test for nicotine dependence: A revision of the Fagerstrom tolerance questionnaire. British Journal of Addiction, 86(9), 1119−1127. Heatherton, T. F., Kozlowski, L. T., Frecker, R. C., Rickert, W., & Robinson, J. (1989). Measuring the heaviness of smoking: Using self-reported time to the first cigarette of the day and number of cigarettes smoked per day. British Journal of Addiction, 84(7), 791−799. Henningfield, J. E. (1990). Can genetic constitution affect the ‘objective’ diagnosis of nicotine dependence? American Journal of Public Health, 80(9), 1040−1041. Hughes, J. R. (2001). Distinguishing nicotine dependence from smoking: Why it matters to tobacco control and psychiatry. Archives of General Psychiatry, 58(9), 817−818. Irvin, J. E., & Brandon, T. H. (2000). The increasing recalcitrance of smokers in clinical trials. Nicotine and Tobacco Research, 2 (1), 79−84. Jarvis, M. J., Wardle, J., Waller, J., & Owen, L. (2003). Prevalence of hardcore smoking in England, and associated attitudes and beliefs: Cross sectional study. BMJ, 326(7398), 1061. Kemmeren, J. M., van Poppel, G., Verhoef, P., & Jarvis, M. (1994). Plasma cotinine: Stability in smokers and validation of selfreported smoke exposure in nonsmokers. Environmental Research, 66(2), 235−243. Kendler, K. S., Neale, M. C., Sullivan, P., Corey, L. A., Gardner, C. O., & Prescott, C. A. (1999). A population-based twin study in women of smoking initiation and nicotine dependence. Psychology and Medicine, 29(2), 299−308. 1042 M.O. Chaiton et al. / Addictive Behaviors 32 (2007) 1031–1042 Kenford, S. L., Smith, S. S., Wetter, D. W., Jorenby, D. E., Fiore, M. C., & Baker, T. B. (2002). Predicting relapse back to smoking: Contrasting affective and physical models of dependence. Journal of Consulting and Clinical Psychology, 70(1), 216−227. Kozlowski, L. T., Porter, C. Q., Orleans, C. T., Pope, M. A., & Heatherton, T. (1994). Predicting smoking cessation with selfreported measures of nicotine dependence: FTQ, FTND, and HSI. Drug and Alcohol Dependence, 34(3), 211−216. McDonald, P. (2003). Consideration and rationale for a national action plan to help Canadian tobacco users: Ontario Tobacco Research Unit. Murray, R. P., Connett, J. E., Istvan, J. A., Nides, M. A., & Rempel-Rossum, S. (2002). Relations of cotinine and carbon monoxide to self-reported smoking in a cohort of smokers and ex-smokers followed over 5 years. Nicotine and Tobacco Research, 4(3), 287−294. Niaura, R., Goldstein, M. G., & Abrams, D. B. (1994). Matching high- and low-dependence smokers to self-help treatment with or without nicotine replacement. Preventive Medicine, 23(1), 70−77. Niu, T., Chen, C., Ni, J., et al. (2000). Nicotine dependence and its familial aggregation in Chinese. International Journal of Epidemiology, 29(2), 248−252. Patten, S. (1997). Performance of the composite international diagnostic interview short form for major depression in community and clinical samples. Chronic Diseases in Canada, 18(3), 109−112. Payne, T. J., Smith, P. O., McCracken, L. M., McSherry, W. C., & Antony, M. M. (1994). Assessing nicotine dependence: A comparison of the Fagerstrom tolerance questionnaire (FTQ) with the Fagerstrom test for nicotine dependence (FTND) in a clinical sample. Addictive Behaviors, 19(3), 307−317. Pierce, J., & Gilpin, E. (2003). A minimum 6-month prolonged abstinence should be required for evaluating smoking cessation trials. Nicotine and Tobacco Research, 5(2), 151−153. Pierce, J. P., & Gilpin, E. A. (2002). Impact of over-the-counter sales on effectiveness of pharmaceutical aids for smoking cessation. JAMA, 288(10), 1260−1264. Salive, M. E., Cornoni-Huntley, J., LaCroix, A. Z., Ostfeld, A. M., Wallace, R. B., & Hennekens, C. H. (1992). Predictors of smoking cessation and relapse in older adults. American Journal of Public Health, 82(9), 1268−1271. Swain, L., Catlin, G., & Beaudet, M. (1999). The national population health survey—its longitudinal nature. Health Reports, 10 (4), 69−82. Tambay, J., & Catlin, G. (1995). Sample design of the national population health survey. Health Reports, 7(1), 29−38. Warner, K. E., & Burns, D. M. (2003). Hardening and the hard-core smoker: Concepts, evidence, and implications. Nicotine and Tobacco Research, 5(1), 37−48. Wetter, D. W., Smith, S. S., Kenford, S. L., Jorenby, D. E., Fiore, M. C., Hurt, R. D., et al. (1994). Smoking outcome expectancies: Factor structure, predictive validity, and discriminant validity. Journal of Abnormal Psychology, 103(4), 801−811.
© Copyright 2026 Paperzz