During the past ten years, the sale of oral moist snuff has increased in Norway, while the sale of tobacco for smoking has decreased. Snuff use and occasional smoking have to some degree replaced daily smoking among adolescents and young adults. The snuff marketed in Norway and Sweden (snus) is a non-fermented, moist and smokeless tobacco product [1, 2]. The sale of snuff is illegal in the European Union (EU), except in Sweden where the legal use is claimed to reduce the smoking rates [3–5]. Smokeless tobacco is used on a relatively wide scale in Norway, a country which is not a member of the EU.

The ban on cigarette smoking in restaurants and bars, which was introduced in Norway in June 2004, may have influenced changes in choice of tobacco type. A Norwegian national survey among pupils in lower secondary school (13–16 years) showed the prevalence of daily smoking to be 5% in 2005, which was half the rate found in the survey five years earlier. Occasional smoking decreased from 18 to 9% in the same period. Snuff use among boys did not change, showing 4% daily and 12% occasional users in 2005. An increase in occasional snuff use from 2% to 5% from 2000 to 2005 was found among the girls [6].

The use of snuff is considered to be less harmful than cigarette smoking, but the evidence of health risks is by no means consistent [7–10]. Two recent reviews on possible health effects of snuff produced conflicting results; one concluded that there is limited epidemiological evidence about the health effects, whereas the other indicated increased risk of myocardial infarct and cancer, assessing experimental evidence from animal studies in addition to research in humans. Both reports concluded, however, that snuff use causes nicotine dependence [11, 12]. Combined use of snuff and cigarettes among male adolescents has been associated with higher levels of nicotine dependence than cigarettes alone [13]. Most users of snuff combine it with smoking cigarettes [14]. The International Agency for Research on Cancer stated in 1985 that there was a carcinogenetic effect of snuff, which was confirmed in 2005 [15, 16].

In Western countries, daily smoking is known to be negatively associated with socio-economic status (SES) [17–22]. The association of snuff with SES is less clear. A Swedish study pointed out an increase in snuff use among well educated urban young people [12]. A similar trend has been shown for occasional smoking [23–26]. In a Swedish city, snuff use was more common among 18 years old pupils attending vocational schools than academic schools and among boys whose parents had no more than compulsory education [27]. In Sweden, regional differences have been found for snuff use, with the highest prevalence in northern rural areas [28]. In the 1980's, prevalence of snuff use was 10% daily and 23% occasional among Norwegian army conscripts, also among athletes and highly educated people [29]. Compared with smoking, the use of snuff seems to differ less by SES and more by region [11, 17].

The aim of this study is to describe the use of tobacco in 15–16 year old pupils by gender, educational ambitions, family background factors, and urbanization. In particular, this study aims to improve knowledge of socio-economic differences in snuff use and combination use of snuff and smoking. Considering that Nordic countries are in the late stages of the smoking epidemic, we expected to find marked SES differences in the prevalence of daily smoking in our study [18, 20]. Little is known, however, about the extent of SES differences in adolescents' occasional smoking and snuff use, which may both represent tobacco use epidemics that differ from daily smoking. Based on existing literature in older age groups, we would expect less SES difference for occasional smoking and snuff use than for daily smoking, or even a positive association between SES and occasional smoking.

The results from binominal regression models of daily smoking, occasional smoking, and snuff use (daily or occasional) are shown in figure 2 and table 4. The interaction term of gender with parents' country of birth being a country with majority of Muslims was statistically significant, and this interaction term was included in all the models.

	I. Smoke daily no snuff N = 11351	II. Smoke occasionally no snuff N = 11539	III. Snuff (daily or occasionally) no smoke N = 10473	IV. Smoke and snuff. Combination users (daily or occasionally) N = 10932
Constant	1.3	11.2	11.9	13.1
Gender
Boys	0.0	0.0	0.0	0.0
Girls (parents born in Norway)	11.9 (10.6, 13.2)	11.1 (9.7, 12.4)	-10.7 (-11.6, -9.7)	-11.2 (-12.4, -10.1)
Girls (parents born in country w. major. of Muslims)	-4.5 (-9.6, 0.6)	-1.8 (-7.2, 3.6)	-1.8 (-4.5, 1.0)	-5.0 (-8.9, -1.1)
Age, years
Under 15.6	0.0	0.0	0.0	0.0
15.6–15.9	1.1 (-0.6, 2.8)	-0.8 (-2.6, 1.0)	0.4 (-0.9, 1.7)	0.1 (-1.5, 1.7)
15.9–16.1	1.5 (-0.2, 3.2)	-0.5 (-2.3, 1.4)	0.4 (-1.0, 1.7)	0.4 (-1.2, 1.9)
16.1–18.4	2.5 (0.8, 4.3)	-0.3 (-2.1, 1.6)	0.4 (-0.9, 1.7)	0.9 (-0.7, 2.5)
Parents' marital status
Married/cohabiting	0.0	0.0	0.0	0.0
Divorced, separated, etc.	10.0 (8.5, 11.5)	3.3 (1.8, 4.8)	1.4 (0.3, 2.5)	4.5 (3.2, 5.9)
Parents' country of birth
Norway	0.0	0.0	0.0	0.0
Country w. major. of Muslims (boys)	3.8 (0.7, 6.8)	-0.01 (-3.1, 3.0)	-9.7 (-11.4, -7.9)	-9.1 (-11.7, -6.5)
Country w. major. of Muslims (girls)	-12.7 (-19.6, -5.8)	-12.9 (-20.0, -5.8)	-0.8 (-4.4, 2.9)	-2.9 (-8.3, 2.5)
Other foreign countries	-2.4 (-5.2, 0.4)	-2.2 (-5.2, 0.8)	-4.5 (-5.8, -3.1)	-4.0 (-6.1, -1.9)
Education ambitions
Academic studies	0.0	0.0	0.0	0.0
Upper secondary school, general studies	5.3 (2.5, 8.1)	2.6 (-0.4, 0.6)	2.0 (-0.2, 4.2)	2.5 (-0.01, 5.0)
Upper secondary school, vocat. studies	12.7 (11.1, 14.4)	1.7 (0.01, 3.3)	2.5 (1.4, 3.8)	9.1 (7.5, 10.7)
One year of upper sec. school/other plans	11.1 (7.4, 14.8)	4.0 (0.4, 7.6)	5.1 (2.1, 8.1)	6.9 (3.5, 10.3)
Undecided	1.0 (-0.7, 2.6)	-2.3 (-4.1, -0.5)	0.8 (-0.4, 2.0)	1.1 (0.4, 2.5)
Family economy
Very well off	0.0	0.0	0.0	0.0
Well off	-2.3 (-4.4, -0.2)	-2.2 (-4.5, 0.04)	- 1.2 (-3.0, 0.6)	-2.2 (-4.2, -0.1)
In between	-0.5 (-2.8, 1.8)	-0.4 (-2.8, 2.1)	-2.7 (-4.6, -0.9)	-1.9 (-4.1, 0.3)
Short of money	5.8 (0.9, 10.6)	4.8 (-0.1, 9.7)	-2.9 (-6.0, 0.3)	1.4 (-2.9, 5.7)
Urban-rural
Urban	0.0	0.0	0.0	0.0
Rural	-1.7 (-3.0, -0.3)	-2.4 (-3.8, -1.0)	1.0 (-0.04, 2.1)	1.7 (0.5, 2.9)

* The first line shows the constant term which equals expected tobacco use when all covariates are zero. The other lines show risk differences × 100 (with 95% confidence interval) for tobacco use. Zero values are the reference categories

The first line in table 4 shows the constant terms from the model, which is the expected prevalence of tobacco use when all covariates are at their reference values. In other words, a boy in the youngest age quartile, with parents from Norway and living together in an urban area, with academic educational plans and considering his family economy to be very good. The other lines show the risk differences, which are to be added to the constant term when the covariates are not at their reference values. The constant and all model coefficients are multiplied by 100 to increase readability. To calculate the expected prevalence of daily smoking for any covariate pattern, simply add the risk differences in table 4.

Example: Boys in the upper quartile of age, with parents living together and born in a country with majority of Muslims, vocational study plans, the family considered short of money, and living in a rural area, have an expected prevalence of daily smoking of 1.3 (constant) + 2.5 (effect of age) + 0 (effect of parents marital status) + 3.8 (effect of Muslim influence for boys) + 12.7 (effect of voc. study plans) + 5.8 (effect of economy) – 1.7 (effect of rural area) = 24.5%.

Smoking was more prevalent among adolescents with vocational rather than academic ambitions, single parents, and poor self-reported family economy. Having parents from Muslim counties conferred an increased risk for boys and a decreased risk for girls for daily smoking, compared to adolescents with Norwegian parents. Snuff use and occasional smoking had weaker associations with educational ambitions, family economy and single parenthood than daily smoking. Combination use was associated with single parenthood and vocational study plans. Gender differences are generally found in Scandinavian countries, with higher prevalence of smoking among the girls and higher prevalence of snuff use among the boys [13, 27, 33].

The strengths of this study are the large and representative study population (nearly 16000 adolescents), high response rate (87%), and a standardized data collection with trained field personnel in all counties.

The main weakness of our study is that all information is self-reported and collected at one point in time [34]. Some pupils may over report their ambitions to attend academic studies and underreport their smoking habits for social desirability reasons, leading to stronger associations in the direction found in our study. Answers, however, were confidential and anonymous, which has been shown to lead to valid self-reported information on adolescent smoking [35–37]. Ethnicity divided only into three groups is a crude measure and was chosen because Muslim cultural influence is a factor known to affect the use of tobacco [30, 31]. In the light of the low smoking rates for Muslim women, girls with parents from these countries may underreport their smoking habits due to social desirability [30].

The amount of tobacco used was not asked, which may lead to misclassification. A study from New Zealand showed that 30% of the adolescents reporting to be occasional smokers turned out to be daily smokers when they were asked about the frequency of smoking [37].

We did not have access to parental socioeconomic data in our study. Instead the pupils were asked to give a subjective assessment of the family economy. It is of increasing acceptance to use adolescents' own reports of social status instead of their often inaccurate reports of the SES of their parents [38]. One weakness with the binomial regression model used is that some covariate combinations may give negative smoking prevalence. These combinations are rare or non-existing in the data.

A positive relationship with age was found for smoking, but not for use of snuff or combination use. Worldwide, 19% of 13–15 year old non-smokers reported in 2000–2007 that they might start smoking during the next year [39]. Our analyses showed a higher prevalence of smoking and lower prevalence of combination use in urban than in rural areas. Little is known about the relationship between adolescent smoking and urbanization. Previous studies show the pattern among adults to differ between countries [40–43].

Our study supports previous findings that Muslim identification is associated with high smoking prevalence among men and low prevalence among women [30, 31]. Adolescents with different cultural backgrounds have been found to influence each other's health behaviour. For example, in the Oslo part of our study, students with a Norwegian background drank alcohol less frequently when attending schools with a larger proportion of students with a Muslim background [44]. This cross-cultural effect on prevalence of smoking and snuff use seems, however, relatively small compared with overall differences in prevalence of smoking and snuff use between groups of adolescents with different country backgrounds. Further investigation into the cross-cultural effects of tobacco and snuff use is warranted.

Our study showed a negative association between smoking and adolescents' own judgement of family economy, in line with other studies finding a higher risk of tobacco use among adolescents in non-affluent families [45, 46]. Our study is also in accordance with other studies showing a higher risk of tobacco use for adolescents with single parents compared with adolescents living with both parents [47–50]. One in four children in Norway are living with only one parent [51], which often implies low income. As the mean age of the pupils in our study was 15.9 years and the initiation age for daily smoking 13.2 years, the probability is relatively high that establishment of the family economy and parents' divorce came before smoking initiation. This may give grounds for a cautious interpretation of these SES-variables as predictors.

Could smoking affect educational ambitions, as well as the opposite being the case? Academic ambitions may be influenced by tobacco use via mediating variables such as attachment to peers with higher or lower academic ambitions. Interestingly, a follow-up study of 16 and 18 year old pupils in Finland found smoking to predict attained educational level. Adolescents' health related lifestyle, rather than health status, with smoking as the strongest predictor, had impact on later educational level. Smoking was considered to be a marker of a broader lifestyle, combined with a rejection of an achievement ideology and the adoption of an anti-school orientation. The number of cigarettes smoked was found to be negatively associated with later educational level [52]. As occasional smokers consume fewer cigarettes than daily smokers, this finding is in line with our finding that occasional smokers had higher educational ambitions than daily smokers, but not as high as non-smokers.

Adolescents' educational ambition has been used as a social indicator by others and is found to correlate with school marks and parents' education level [53, 54]. Our results support earlier findings that academic orientation as well as school performance is shown to be closely associated with adolescents' health and health-related behaviour, including smoking [27, 55, 56]. These associations may be due to parental influence or other factors in the social environment. Peer, teacher and environmental influence may also differ between vocational and academic school-classes [57].

The negative association found between SES and daily smoking was expected. Several other studies confirm these findings among adolescents [19] and it is consistent with Norway being in the late stage of the tobacco epidemic, where the prevalence of smoking continues to decline and gradually reaches a stable minimum level. The decline in prevalence of smoking among lower SES groups lags behind the decline in higher SES groups [20, 58].

We expected a positive association between SES and occasional smoking. We found, however, a negative association, although weaker than for daily smoking. A study among 16–18 year old students from Norway found occasional smokers to be in higher academic courses than daily smokers, in line with the differences in educational ambitions in our study [59].

In studies on adults, occasional smokers had higher education levels than daily smokers [24, 26]. Our study of a younger age group may indicate a shift to lower SES for occasional smokers, as the tobacco epidemic in general is on the decline. In a Norwegian study from 2006, adolescents rated the "smoker prototype" as less attractive than the "non-smoker-prototype," even amongst regular smokers [60]. Being a non-smoker was associated with being independent, smart and self-confident, indicating that the attitudes towards any type of smoking are slowly changing to be more negative. The spread of attitudes about tobacco use from higher to lower SES levels has been described [18, 20, 58]. Young people today may be some of the first to adopt a wave of negative attitudes towards occasional smoking, with young people in higher socio-economic groups leading on with tobacco-free practice, and others adopting the negative attitude while still using tobacco.

We had expected less SES difference for snuff use than for daily smoking. This expectation was met regarding educational ambitions and parents' marital status. In a Swedish city, 18 year old students in vocational courses were nearly twice as likely to use snuff as students in academic programmes [27]. Adolescents' own educational orientation was used as a measure, with the results corresponding to our findings using educational ambitions as a measure. Subjective family economy in our study was positively associated with daily snuff use among boys. Our results indicate that snuff use is associated with a higher SES than daily smoking, although snuff use may undergo a similar shift as smoking, starting with decreasing prevalence of use in higher socio-economic groups, and young people being the first to change their habits.

In a time of rapid changes in tobacco use, in particular among adolescents, it is important to recognize subgroups at high risk. Our study has clearly indicated high-risk for tobacco use among those with ambitions for a vocational rather than academic career, and from less affluent or single parent families. The social and family background differences were largest for daily smoking and less pronounced for occasional smoking and snuff use. There may be an ongoing shift towards lower SES among all groups of tobacco users, including occasional smokers and snuff users. The trends for smoke and smokeless tobacco should be followed, as well as factors contributing to the start and cessation of tobacco use.