Background:
Smoking remains the leading risk for the total burden of disease in Western Europe, and smoking is more prevalent among adolescents from lower socioeconomic positions (SEP). Only eight percent of 15-year old Danish adolescents from higher social classes smoke daily or sometimes, whereas up to 37% of adolescents whose parents live from welfare benefits are smokers. The X:IT study has been the first large-scale smoking intervention to be effective in reducing uptake of smoking in adolescents in Denmark, but in order to develop effective interventions which work across social groups, it is important to study the social patterning of the effect.

Methods:
We used data from a Danish cluster randomized trial including 4041 year-7 students from 51 intervention and 43 control schools. Outcome measure 'current smoking' was dichotomized into smoking daily, weekly, monthly or more seldom vs do not smoke. We performed multilevel, logistic regression analyses of available cases and intention-to-treat (ITT) analyses, replacing missing outcome values by multiple imputation. The analyses were stratified by family socioeconomic position (SEP).

Results:
Analyses on imputed cases showed that overall, the OR for smoking among students at intervention schools compared with control schools was 0.67 (0.50-0.89). Among students from high SEP the OR was 0.60 (0.37-0.98), from medium SEP the OR was 0.65 (0.45-0.96), and from low SEP the OR was 0.78 (0.49-1.26).

Conclusions:
The X:IT smoking intervention showed significant effects among students from high and medium SEP, but not among students from the lowest SEP pointing at a social differential effect. Mechanisms behind inequalities in smoking interventions are understudied, and it seems important to examine which intervention strategies have the potential to improve health-related behavior among children and adolescents from lower socioeconomic groups.