Behavioral problems and the occurrence of tobacco, cannabis, and coca paste smoking in Chile: evidence based on multivariate response models for school survey data.

Abstract

In this study we estimate suspected links between youthful behavioral problems and smoking of tobacco, cannabis, and coca paste. In the Republic of Chile, school-attending youths were sampled from all 13 regions of the country, with sample size of 46,907 youths from 8th to 12th grades. A Generalized Estimating Equations (GEE) approach to multiple logistic regression was used to address three interdependent response variables, tobacco smoking, cannabis smoking, and coca paste smoking, and to estimate associations. Drug-specific adjusted slope estimates indicate that youths at the highest levels of behavioral problems are an estimated 1.1 times more likely to have started smoking tobacco, an estimated 1.6 times more likely to have started cannabis smoking, and an estimated 2.0 times more likely to have started coca paste smoking, as compared to youths at the lowest level of behavioral problems (p<0.001). In Chile, there is an association linking behavioral problems with onsets of smoking tobacco and cannabis, as well as coca paste; strength of association is modestly greater for coca paste smoking.