PURPOSE: To describe smoking initiation, and to investigate factors that predict the early initiation of smoking in schoolchildren using a longitudinal approach. METHODS: A prospective study of smoking habits of children from the third and fourth grades through the eighth and ninth grades. The initial study population was 1970; 79.8% were white and 20.2% were African-American. Children were classified as "nonsmokers," "experimental smokers," or "current smokers" at five time points over 6 years. Multivariate regression models examined relationships of demographic and developmental factors with smoking initiation. RESULTS: Experimental smoking increased from 4% at Grades 3-4 to 42% at Grades 8-9, and current smoking prevalence rose from 0.4% to 9% over the same period. The mean age of initiation of smoking was 12.3 years. Smoking initiation (experimental smoking) was significantly different by racial group, socioeconomic status (SES), and pubertal development. White children and those of low SES were more likely to be experimental smokers, and also started earlier than African-American children and children of high SES. Once they started, white children advanced more rapidly to become current smokers. Boys had a higher prevalence of experimental smoking than girls at all time points. Children in rural areas were more likely than urban children to start smoking after age 12 years. Children who were at a higher pubertal stage than their peers were also more likely to experiment with smoking. CONCLUSIONS: Race, SES, and pubertal stage are important predictors of initiation of smoking in schoolchildren. This study indicates a need for smoking prevention classes in elementary and middle school, especially in areas with large numbers of white and low-SES youth. Also, smoking cessation programs, as well as smoking prevention classes, would be useful for middle school and high school students.
