Childhood predictors of smoking in adolescence: a follow-up study of Montréal schoolchildren

Methods

For our first survey (1990–1992), we made a random selection of 18 Montréal schools after a stratification of all school boards in central Montréal. They were ranked within school boards by socioeconomic status, as described in that report.6 In each school, all the children in 1 class in each of grades 1, 3 and 5 (corresponding to ages 5–7, 8–9 and 10–12 years, respectively) were invited to take part through a questionnaire addressed to and answered by their parents. Each child in grade 5 who participated was also asked, in private, if he or she smoked. With parental permission, we visited the homes of a random sample of children (31%) to gather additional information on housing conditions and to collect a sample of the children's saliva to analyze for cotinine concentration as a measure of their individual uptake of environmental tobacco smoke.7^,8^,9^,10 Satisfactory cotinine measurements were available for 191 of 309 children, who composed the present study population. At the time of the second survey (1994–1996), all 191 children (80% of whom had since entered high school) answered a questionnaire about their health and smoking habits and performed spirometry.

The outcome of interest was having become a smoker by the second survey, defined as an answer of “Yes” to the question, “Have you ever smoked as much as 1 cigarette a week for a month?” Social and cultural factors that may affect levels of environmental tobacco smoke in the home and physiologic data from the first survey are shown in Table 1. During the second survey, we recorded pubertal status (defined as a “yes” answer to the survey question “Have you started your menstruations?” for girls or “Has your voice changed?” for boys), the high school currently attended (a composite marker of peer pressure to smoke and the scholastic and sporting expectations attributed to each school) and changes in family smoking behaviours since the first survey, which we interpreted as indications of parental attitudes toward smoking (negative, if an increase in the number of adult smokers in the home had been allowed; and positive, if smoking cessation in the home appeared to have been encouraged). Univariate analysis was used to select potential explanatory factors, based on 2-tailed p values of 0.10 or less. Independent predictors were determined by multivariate logistic regression analyses.

Results

Of the 191 study subjects with data on cotinine levels, 103 (54%) were girls and 88, boys. Their average age at the time of the first survey was 9.2 years, and at the second, 13.0 years. None of the grade 5 children reported being a smoker, although 14 children (7.3%) did not answer the question. Four children at the first survey had concentrations of salivary cotinine that were markedly elevated, but in only 1 child did the level approach 20 μg/L, the usual cutoff for defining a smoker.8 At the time of the second survey, 84 children (44.0%) had become smokers; of these, 10 boys and 10 girls smoked more than 10 cigarettes per week. Attaining puberty between surveys was the most significant determinant of becoming a teenage smoker (56.4% of postpubertal children compared with 17.9% of those still prepubertal when the follow-up survey was administered, p = 0.001). Subsequent analyses were therefore stratified by the children's pubertal status at the second survey (Table 2).

The only factor associated with becoming a teenage smoker in both pre- and postpubertal children was salivary cotinine concentration (adjusted in the analysis for sex, number of siblings, socioeconomic status of the parents, crowding index, number of smokers at home and total number of cigarettes smoked in the home). In prepubertal children, a marginally significant childhood factor was female sex; in postpubertal children, forced vital (lung) capacity and the presence of adult smokers in the home, whether single (i.e., 1 smoker) or multiple. The specific high school currently attended, which was recorded at the second survey, also showed marginal significance (p = 0.09; data not shown).

Interpretation

Our study population was small, and multiple comparisons were made; our results must therefore be interpreted with caution. Moreover, salivary cotinine levels may reflect current smoking by the child as well as environmental tobacco smoke; given the low rates in the first survey both of salivary cotinine and of reported smoking, however, this effect is likely to have been small.

An unexpected finding in the multivariate analysis was the lack of any significant association between becoming a teenage smoker and the number of smokers in the home in childhood (p = 0.99). Also unexpected was the association between a decrease in the number of adult smokers in the home between surveys (thought to indicate a positive parental attitude toward smoking cessation) and a decrease in the number of adolescents becoming smokers (the reduction in risk is shown in Table 2). This could have been caused not only by parental disapproval but alternatively by nonsmoking teenagers pressuring their parents to quit.8^,9^,10 It nonetheless shows the importance of a dynamic attitude at home against smoking.

As for the anatomic and physiologic findings, the most important observation was that salivary cotinine in childhood (a measurement of the nicotine actually entering the bloodstream) was an independent predictor of adolescent smoking, after adjustment in the analysis for amount smoked at home and other relevant factors (see final footnote, Table 1). Given also the association (admittedly not strong) between forced vital capacity and adolescent smoking, we suggest that lung size (or some associated characteristic) increases the uptake of environmental tobacco smoke, maximizes the influence of passive smoking in childhood and induces smoking in adolescence. In this context, the recently reported reduced density of dopamine D1 receptors in the ventral striatum of the brains of adult smokers of cigarettes led those investigators11 to speculate that a hypodopaminergic state may play an important role in sustaining “nicotine-seeking” behaviour and future addiction.

In conclusion, we hypothesize that enhanced susceptibility to environmental tobacco smoke in childhood increases the risk of nicotine-seeking behaviour in adolescence. If proved correct, this would be valuable information to use in the ongoing campaign against smoking in teenagers.

𝛃 See related article page 382