Article Text
PDF
Abstract
Background Empirical data suggest that children with infrequent tobacco use have difficulty quitting smoking.
Methods Data were obtained from the nationally representative Global Youth Tobacco Survey of middle-school students in Cyprus and Greece. Regression analyses examined associations between smoking frequency (smoking days per month or cigarettes smoked per day) and loss of autonomy (difficulty refraining from smoking).
Results The prevalence of lost autonomy was 40% among subjects who smoked 1 or 2 days/month and 41% among subjects who averaged less than one cigarette/day and increased in a dose-response pattern. Regression models derived from the Cyprus data were replicated by the Greek data.
Conclusions Two national surveys confirm previous reports of difficulty with smoking cessation with infrequent smoking. Since loss of autonomy is universally recognised as a core feature of addiction, our data indicate that young adolescents experience symptoms of nicotine addiction with infrequent tobacco use.
- Nicotine addiction
- tobacco
- dependence
- children
- GYTS
- smoking
- addiction theories
- Cyprus
- Greece
- Addiction
Statistics from Altmetric.com
Through the end of the last century, authorities believed that addiction to nicotine did not start until a person smoked a minimum of five cigarettes/day.1 2 Under this threshold theory, smoking at subthreshold levels is attributed to social factors and smoking pleasure, but not addiction.2–5 Under the competing sensitisation-homeostasis theory, smoking is motivated primarily by symptoms of addiction that develop with infrequent smoking, sometimes after the first cigarette.6 The first longitudinal study of the onset of nicotine addiction contradicted the threshold theory: symptoms of addiction, such as failed attempts at cessation, were documented in children who smoked only a few cigarettes per month,7 a result that has been replicated repeatedly.8–11 Case histories demonstrate that one cigarette satisfies an early addiction for a week or more, but as tolerance develops the frequency of smoking must be increased to maintain comfort.12 13 Thus, while the threshold theory stipulates that there is no association between addiction symptoms and smoking frequency below a level of consumption of five cigarettes/day, the sensitisation-homeostasis theory predicts a dose-response relation.
A key feature of nicotine addiction is the loss of autonomy,14 15 which is recognised when symptoms present an obstacle to cessation either by making abstinence unpleasant or difficult to maintain.16 The loss of autonomy is a critical milestone in the trajectory of smoking in adolescents.17 The Hooked on Nicotine Checklist is a validated and widely used 10-item measure of the loss of autonomy.18–23 Three items from this measure assess difficulty in quitting or refraining from smoking. The Global Youth Tobacco Survey (GYTS), which was initiated by WHO, the Centers for Disease Control and Prevention (CDC) and the Canadian Public Health Association24 assessed a loss of autonomy through three items concerning difficulty in quitting and refraining from smoking. We used GYTS data from Cyprus and Greece to test the hypothesis that young smokers experience a loss of autonomy with infrequent smoking.
Methods
The GYTS follows a standardised methodology for constructing sampling frames, selecting schools and classes, preparing questionnaires, carrying out field procedures and processing data.24–26 The GYTS was administered to nationally representative samples in Cyprus and Greece during the school years 2004–2005 and 2005–2006 by trained field workers. At the first stage, the probability of school selection was proportional to student enrolment in grades 1–3 of all middle schools. At the second stage, classes within the selected schools were randomly selected. All students in attendance in the selected classes were invited to participate. The anonymous self-administered questionnaire included items on cigarette and other tobacco use and covered a variety of topics in addition to those pertaining to our analyses.24 Responses were documented on standardised answer sheets which were sent to the CDC for data entry.
Ever use of cigarettes was assessed with the following item, “Have you ever smoked cigarettes?” Predictor variables included the number of smoking days per month (none; 1–2; 3–5; 6–9; 10–19; 20–29; every day) and the number of cigarettes smoked per day on the days when they smoked (<1, 1; 2–5; 6–10; 11–20; >20), asked of all current smokers.
As defined, smokers enjoy full autonomy over their smoking when they can freely abandon smoking without suffering or difficulty. Individuals who “want to stop smoking now” are smoking despite their desire not to do so, a clear indication that they do not enjoy full autonomy. Loss of autonomy was assessed with three questions. Q1: “How easy or difficult would you find it to go without smoking as long as one week?” Q2: “How easy or difficult would you find it to give up smoking altogether if you wanted to?” Q3: “Do you want to stop smoking now?” Q1 and Q2 had the following response options: very difficult, fairly difficult, fairly easy and very easy, while Q3 had yes/no options. An answer of very difficult or fairly difficult to the first two questions or a “yes” answer to the third question were each given a score of “1” while the other responses were scored as zero. The scores for the three questions were summed to provide a dichotomous measure (a zero score indicates no loss of autonomy and a score of 1 or more indicates a loss of autonomy).
Analyses were performed using SPSS 16.0 and R, an open-source language and environment for statistical computing (R development core team), freely available at http://cran.r-project.org. T-tests and χ2 tests were used for group comparisons. A series of logistic regression models evaluated the association between smoking frequency and the risk of loss of autonomy. Cigarettes smoked per day and days smoked per month were used as categorical predictor variables with the lowest category serving as the reference category. We also examined dose-response models. All models were adjusted for a standard set of potential confounders, age, gender, age of smoking initiation, parental smoking, friends' smoking and adolescents' disposable income. To determine if the models could be replicated in another population, the regression models derived from the Cyprus data were applied to the data from Greece. All tests were two-sided, and a p <0.05 was considered significant. To account for the dependencies of the clustered data, the multilevel models were estimated with random effects for the five geographical regions for Cyprus and four for Greece.
Results
A total of 7263 and 6378 students completed the GYTS questionnaire in Cyprus and Greece, respectively. The mean age of students was 14.4 (SD 1.0) and 13.9 years (SD 1.2), while the percentage of males was 50.5% and 50.4% in Cyprus and Greece, respectively. In Cyprus, 29.8% and 32.0% of fathers and mothers had earned a university degree compared to 31.0% and 32.7% in Greece. The school response rate was 91% and 90%, the student response rate was 90% and 89%, and the overall response rate (ie, the school rate multiplied by student rate) was 81% and 80% in Cyprus and Greece, respectively. The percentage of students who had ever tried smoking and were current smokers (reported smoking in the past 30 days) in Cyprus were 27.1% and 9.5%, and in Greece, 34.7% and 13%. The mean age of smoking initiation among middle-school students was 11.5 years (SD 1.8) in Cyprus and 9.7 years (SD 0.8) in Greece. Among current smokers in Cyprus, 6.8% were smoking one or more cigarettes per day, compared to 10.0% in Greece.
In table 1, population characteristics in relation to smoking status for Cyprus are presented (n=7263). Smoking was more common among boys and when parents smoked. Smoking prevalence increased with age, number of friends who smoked, and adolescents' disposable income. In table 2 we present the association between tobacco use measures and the three indicators of loss of autonomy for 1932 students that had tried smoking). A loss of autonomy was reported by 40.2% of students who smoked only 1 to 2 days per month and by 40.9% of students who smoked less than one cigarette on the days they reported smoking.
Smoking in relation to population characteristics—Cyprus (n=7263)
Loss of autonomy among students who had tried smoking—Cyprus (n=1932)
In table 3, the regression models for Cyprus are presented for 676 current smokers. In the adjusted models, both days smoked per month and cigarettes smoked per day were associated with loss of autonomy in the dichotomous, trend and dose-response models. A caveat applies to the interpretation of these models. There is no level of tobacco use that has been shown to be without risk of a loss of autonomy.11 The lowest exposure groups that served as the referent in our models (smoking 1–2 days/ month and smoking less than 1 cigarette/day) had a prevalence of lost autonomy of 40–41%. Therefore, the lack of a significant difference in risk for smoking 3–5 days/month compared to 1–2 days/month does not signify an absence of risk; indeed 52% of this group reported a loss of autonomy.
Derivation of logistic regression models for the association of smoking with loss of autonomy—Cyprus (n=676) (multilevel modelling)
An important finding with respect to the derived regression models was the χ2 values of the Omnibus test for the fitting of models ranging from 25.4 to 73.8 for days smoked per month and 26.1 to 60.8 for cigarettes smoked per day, with 7 degrees of freedom. All χ2 values were significant. The Nagelkerke R2 for the models ranged from 12% to 24% for days smoked/month and 15% to 35% for cigarettes smoked/day.
In table 4 we summarise the validation of our derived regression models using the corresponding data from 826 current smokers from Greece. In the adjusted models, both days smoked per month and cigarettes smoked per day were associated with loss of autonomy in the dichotomous, trend, and dose-response models. The magnitude of these associations in the adjusted dichotomous models were similar for Cyprus and Greece (2.92 vs 2.96 for days smoked per month, and 2.75 vs 2.55 for cigarettes smoked per day). The trend and dose-response models also showed similar results. The disparities at the highest level of consumption (>20 cigarettes/day) probably reflect instability in the estimates owing to the rarity of heavy smoking among middle-school students.
Derivation of logistic regression models for the association of smoking with loss of autonomy—Greece (n=826) (multilevel modelling)
Discussion
This is the first study to demonstrate a dose-response relation between frequency of tobacco use and the loss of autonomy over tobacco use among middle-school students. It is also the first study to examine the loss of autonomy using representative national surveys. Symptoms of lost autonomy were quite prevalent (40%) among youths who reported smoking on only 1–2 days/month, and among those averaging less than one cigarette on days when they smoked (41%). These data are consistent with prior reports from older adolescents.11 In two studies from the USA, the mean frequency of tobacco use when autonomy was lost was 7–8 cigarettes/month.7 8 In the current study, a loss of autonomy was reported by half of those who smoked on 3–5 days/month. Using dichotomous, trend and dose-response regression models, we found that the loss of autonomy increased with increasing frequency of cigarette consumption in a consistent pattern across both national surveys. This confirms the reliability of adolescent self-reported data on nicotine addiction, consistent with previous research.18–23
Current theory postulates that nicotine addiction develops through a physiological process that would be expected to transcend social and cultural influences.6 We have previously demonstrated that youths' reports of symptoms of nicotine dependence cannot be attributed to socially based expectations about smoking.27 The case for addiction at low levels of consumption is strengthened by the consistency of the self-report data obtained from youths living in countries that might be considered to be at two extremes in terms of cultural tolerance for smoking at the time of the survey (the USA and Greece).
The threshold theory postulates that the risk of lost autonomy remains virtually zero until consumption approaches five cigarettes per day. We are aware of no study documenting the absence of nicotine addiction in a representative sample of infrequent smokers. By documenting the loss of autonomy among a representative sample of infrequent, low-consumption smokers, our data contradict the threshold theory and add to a large body of research demonstrating symptoms of lost autonomy among individuals who smoke fewer than five cigarettes per day.7–12 18 22 28–31
The sensitisation-homeostasis theory asserts that symptoms of lost autonomy can appear after the first few cigarettes and increase proportionally with tobacco use, a prediction that is supported by empirical data.11 32 Providing strong support for this model, a prospective study demonstrated that the frequency of tobacco use predicts the progression of symptom development, while the number of addiction symptoms predicts the escalation in smoking frequency in a vicious positive feedback cycle.33 In further support of the sensitisation-homeostasis theory, the current study is the first to show a strong dose-response relation between frequency of smoking and loss of autonomy using multivariable models.
Study strengths include the use of large nationally representative samples with the results from Cyprus being confirmed by the data from Greece. A study limitation is that only three questions were used to assess the loss of autonomy. Since the Hooked on Nicotine Checklist includes 10 symptoms of loss of autonomy, our study probably underestimates the true prevalence of lost autonomy in these populations. As in all prior studies on this topic, we relied on self-reported symptoms. Self-reported loss of autonomy has been the strongest predictor of continued smoking in longitudinal studies, far surpassing any psychosocial risk factor in this regard.8 34
Conclusion
By deriving our models with national data from Cyprus and replicating them with national data from Greece, we provide strong evidence to support the sensitisation-homeostasis theory by showing that loss of autonomy over tobacco use develops very commonly with intermittent tobacco use. Other countries with GYTS data may perform similar analyses to contribute to our understanding of nicotine dependence. Future versions of the GYTS may be improved by including a more robust measure of the loss of autonomy. Our findings support targeting interventions at the earliest stages of smoking initiation.
What this paper adds
The common assumption that addiction does not begin until a person smokes at least five cigarettes daily is reflected in the proposition that cigarettes could be rendered non-addictive by reducing nicotine levels to the point where a smoker would be unable, as a practical matter, to obtain as much nicotine as is delivered by five ordinary cigarettes. This premise has been challenged by reports from English-speaking countries that youths who smoke only a few days per month have difficulty quitting.
Here we report national survey data from Cyprus and Greece showing youth who smoke only 1–2 days per month commonly face difficulty with quitting.
The close agreement of these data with reports from the USA, Canada and New Zealand demonstrates that the onset of addiction with infrequent use transcends national and cultural boundaries. Strategies to reduce the addictiveness of cigarettes should be based on experiential data not presumptions.
Acknowledgments
The authors would like to thank the officers at the Ministry of Health in Cyprus and the Ministry of Health and Social Solidarity in Greece for their support of the GYTS project. The study was supported by a grant from the Cyprus Ministry of Health and the Greek Ministry of Health and Social Solidarity. Considerable support was also received from the Office for Smoking and Health at the Centers for Disease Control and Prevention.
References
Footnotes
Funding Centers for Disease Control.
Ethics approval This study was conducted with the approval of the Centers for Disease Control.
Provenance and peer review Not commissioned; externally peer reviewed.