Table 1

Characteristics of nine longitudinal studies investigating the determinants of self-initiated smoking cessation of a duration of at least 6 months in population-based samples of adolescents and young adults

First author, date of publication (reference)Study location (timeframe)Duration of follow-up; surveys after baselineSampling frame, sampling methodSample size, participant age/grade range at baselineNumber of quitters (%)Participant smoking status at baselineDefinition of cessationAdjustmentAnalytical method
Bricker, 200918Washington State, USA (1984–2004)11 years; onceSchool based (convenience), from the combined control and intervention cohorts of the HSPP*991 students whose parents were ever regular smokers; 17 years233 daily smokers (23.5%) plus 257 weekly smokers (25.9%)At least weekly and daily smokersAt least 6 months of abstinenceGender, parental educationMultivariate logistic regression
Mittelmark, 198819Minneapolis/St. Paul metropolitan area, USA (not mentioned)2 years; three times (each spring and autumn over 2 academic years)School based (convenience)72; 7th–11th grade32 (44.4%)Identified on the first and/or second surveys to be current cigarette smokersReport of non-smoking on the third and fourth surveysDiscriminant analysis to develop a multivariate model
Tucker, 200517California and Oregon, USA (1985–2001)Calendar time period used for analysis: 1995–2001: 6 years; onceSchool based (representative), from the project ALERT360; 23 years103 (26%)In 1995, smoked in the past month and on at least 40 occasions in the past yearMost recent quit attempt lasted 6 months or longerControlling for gender, race/ethnicity, income, and education+for all predictors+for age at first use and quantity/frequency of smoking (average number of cigarettes smoked per day during the past 30 days)Multivariate logistic regression analysis
Ellickson, 200116California and Oregon, USA (1985–1995)Calendar time period used for analysis 1985–1995: 10 years; twiceSchool based (representative), from the project ALERT1093; 18 yearsDefined as smokers (cigarette use during the last year) in 1990Not smoked in last yearLogistic regression analysis (univariate or multivariate analysis undefined)
Ellickson, 200113California and Oregon, USA (1985–1990)Calendar time period used for analysis 1988–1990: 2 years; onceSchool based (representative), from the project ALERT827; Grade 10100 (12.1%)Smoking at least 11–20 times during the past year at grade 10Not smoking during the past year at grade 12Six separate models tested§+controlling for age at first use and for smoking quantityMultivariate logistic regression analysis
Tucker, 200212California and Oregon, USA (1985–1995)Calendar time period used for analysis 1990–1995: 5 years; onceSchool based (representative), from the project ALERT711; Grade 12106 (14.9%)Smoking at least 10–19 times during the past yearNot smoked at all during the past yearSix separate models tested§+controlling for age at first use and for smoking quantityMultivariate logistic regression analysis
Hansen, 198520USA (1981–1982)15 months; twiceSchool based (convenience)392; 15–16.5 years44 (11.2%)SmokersMaintaining quitting: smoker at wave 1 and non-smoker at wave 2 and 3Discriminant and multivariate analysis
Chang, 200621Taiwan (2000–2002)2 years; once or twiceSchool based (convenience)494; Grade 1076 (15.4%)Smoked in the past yearSmoked in the past 12 months in the 10th grade survey but not in the 11th±12th grades surveysAll psychosocial factors in 10th gradeGeneralised estimating equation in univariate and multivariate analysis
Rohde, 200414Western Oregon, USA (1987–1998/2001)11–14 years; four timesSchool based (representative)242; 14–18 years53 (22%)Daily smokers at survey T3No smoking during the 12 months prior to turning age 25 yearsVariables demonstrating significant effects (MDD, antisocial personality disorder scores, family history of drug and alcohol use disorders, history of ND, gender×ND, male gender)Univariate and multivariate logistic regression analysis
  • * Bricker et al18 used a study sample which came from the combined control and intervention cohorts of the Hutchinson Smoking Prevention Project (HSSP), a group-randomised trial of school-based smoking prevention in third grade. The authors had previously reported that there was no intervention impact relative to the prevalence of daily smoking or for other smoking outcomes either at grade 12 or at 2 years after high school.22

  • Current cigarette smokers: report of smoking monthly or more, carbon monoxide level of 8 ppm or greater and saliva thiocyanate level of 85 μg/ml or greater.

  • Project ALERT was implemented in seventh and eighth grade, but this multiyear drug use prevention programme had no impact once the prevention lessons stopped. At grade 10 and 12, there was no significant effect on smoking behaviour.15

  • § (1) Demographic variables (age, ethnicity and parental education), (2) smoking environment, (3) smoking beliefs, (4) rebelliousness, substance use, and problem behaviours, (5) social bonds and (6) health status (for (2)–(6), demographic variables were included).

  • Psychosocial factors in six domains: (1) smoking and substance use behaviours: frequency and quantity of tobacco and other substance use, (2) individual factors: demographics, working status, academic achievement, antismoking attitude, beliefs, refusal self-efficacy, deviance, problem behaviours, (3) family factors: family structure, socioeconomic status, parental communication, supervision, attachment, parents' smoking disapproval attitude, (4) peer factors: peer tobacco and other substance use, peer smoking attitudes, peer deviance behaviours, (5) school factors: proportion of student smoking, teachers' smoking, teachers' smoking disapproval attitude, teacher supervision and (6) community factors: community identity, the availability of tobacco and other substances and community smoking disapproval attitude.