Study
The 2002 Perry and colleagues study reports the long-term effects of Project Northland implemented in rural Minnesota. The study randomized 24 school districts into a treatment and a delayed program comparison condition in 1991, where they remained until 1998. This study sample is the class of 1998 and examines the long-term outcomes of the intervention. The treatment group consisted of 1,401 students, and the comparison group consisted of 1,549 students, with data collected at baseline in 1991 in the sixth grade and each spring of each year, from 1992 to 1998. The program was implemented in two phases. The first was delivered during grades 6 to 8 from 1991 to 1999. It included a classroom social-behavioral component as well as parent, community, and peer involvement. A brief five-session intervention that was applied in ninth grade concentrated on issues of alcohol use and road safety, around the age when the cohort would be eligible to drive. The second phase was implemented in grades 11 and 12, which made use of classroom curricula on the consequences and legality of teen alcohol use, material for parents to use in communicating with children about alcohol, print media circulated to commercial alcohol vendors in the school district, and community organizing campaigns promoting alcohol-free activities and reducing access to alcohol for high school students. The comparison schools were offered treatment on a delayed basis, with the first phase of the intervention in 1994 to 1997 and the second phase in 1998 to 2000.
The sample was 53 percent male, 93 percent white, and 5 percent American Indian. While the number of surveys completed was associated with baseline drinking (higher drinking linked to fewer completed surveys), this did not differ by treatment condition. Additionally, survey completion numbers were higher for whites than for nonwhites, though this association also did not differ by treatment condition. The surveys used a Tendency to Use Alcohol Scale, assessment of binge drinking based on the Monitoring the Future Study, and measures for peer influence, self-efficacy, and perceived access to alcohol. To assess the community action efforts to reduce commercial availability of alcohol to high school students, measurements were taken of the success rate of alcohol purchase attempts by young-appearing buyers. Telephone surveys were conducted with parents at grade 10 (n = 2,048) and grade 12 (n = 1,793) covering their acceptance of underage drinking, permissive norms, and opposition to alcohol-control policies and parental monitoring.
Data from the study was assessed using growth-curve analysis, a technique that identifies the patterns of changes over time and tests whether these patterns differ among treatment groups (or other subgroups). The study employed mixed-effects linear regression for repeated measures, using multilevel modeling of individual, school district, and treatment condition trajectories of alcohol use. Mixed-model regression analyses were used to analyze alcohol purchase attempts and the parent surveys. The authors did not conduct subgroup analyses.
Study
Komro and colleagues (2008) reviewed the outcome evidence of the impact of Project Northland implemented in an urban multiethnic setting in Chicago, Ill. (PNC), over 3 years. The study randomized neighborhoods as study units to either the intervention or the comparison ‘prevention as usual’ condition. The PNC intervention was designed to be implemented consecutively in the autumn of sixth to eighth grades. Ten study units received the PNC intervention with 32 schools, and the 12 comparison units had 29 schools. Assessments were made at baseline in fall 2002 in the sixth grade, and at follow-up points at the end of every school year of treatment: in spring 2003 of sixth grade, in spring 2004 of seventh grade, and in spring 2005 of eighth grade. Because of missing data at some follow-up points, the number of participants in the PNC group (n = 2,501–2,538) and the comparison group (n = 3,079–3,147) varied. While 5,812 students completed at least one survey, 2,373 students completed all four. The study sample was 50 percent male, 43 percent African American, 23 percent Hispanic, 13 percent white, and 75 percent received free or reduced-price lunches, indicating a low socioeconomic status. Forty-seven percent lived with both their parents, and 74 percent reported English as the primary language in the home. There were no statistically significant demographic differences between the treatment and comparison groups.
The study was interested in the alcohol and drug use outcomes of adolescents, but it also assessed parents, community leaders and commercial access to alcohol by underage drinkers. The study administered a student questionnaire assessing youth alcohol and drug use, behaviors and attitudes, parent and community leader surveys, and alcohol purchase attempts (using women at least 21 years old whom a panel judged to be younger appearing—i.e., underage).
Data was analyzed using mixed-effects regression models for the repeated measures testing for differences between intervention groups. Mean trajectories were modeled for each student, unit, and study condition (taking into consideration their internal variations—that is, across time, across students, and across units respectively) to perform the growth curve analysis. Analysis of parent and community leader surveys used mixed-model analyses of variance (ANOVA). The authors conducted subgroup analyses on level of participation in the program.
