Testing a School-Based E-cigarette, Tobacco, and Betel (Areca) Nut Use Prevention Curriculum for Guam Youths

The people of the U.S. Affiliated Pacific Islands (USAPI) face higher cancer incidence, especially lung/bronchia and head-and-neck cancer, and poorer cancer outcomes, compared with the U.S. nationally. This may partly be driven by the high rates of cigarette smoking and betel (areca) nut use in the USAPI. Previous data suggest that that adolescents on Guam, as young as middle school students report markedly higher e-cigarette and tobacco product use prevalence in the USAPI compared with the USAPI nationally. Guam youths are also at risk for the use of betel nuts. Yet, currently there are no tobacco product/areca nut use prevention programs that have been developed for and tested specifically USAPI adolescents. The proposed study will develop a school-based substance use prevention curriculum for e-cigarette, tobacco product, and areca nut use prevention among Guam youths. The curriculum will use lessons incorporating innovative videos and culturally grounded activities. The study's specific aims are: 1. To develop a school-based curriculum for e-cigarette, tobacco product (i.e., cigarette, smokeless tobacco), and betel nut use prevention among middle school students in Guam. 2. Test the efficacy of the school-based curriculum in a randomized controlled trial.

BACKGROUND AND SIGNIFICANCE The study focuses on e-cigarette, tobacco and areca nut prevention among adolescents in Guam. By developing and testing a school-based e-cigarette, tobacco and areca nut prevention program, the project will target psychosocial, lifestyle, and environmental/physical factors contributing to cancer disparities in the US Affiliated Pacific Islands (USAPI). Thus far, nationally, there has been limited effort in terms of developing a theoretical framework that would attempt to explain tobacco and areca nut use etiology among USAPI youth. Previous research indicates that social and environmental influences play key roles in shaping tobacco and areca nut use among USAPI youths. In addition, skills needed to resist social influence appear to be protective against tobacco and areca nut use. Particularly, past research has found peer influence and environmental risk factors such as parental permissiveness, ease of access to tobacco/areca nut at home and school and through adults and friends, to be strongly associated with current tobacco/areca nut use. On the other hand, competence in navigating through social situations conducive to tobacco/areca nut use and refusal self-efficacy (i.e., confidence in being able to say "no" at the face of peer pressure) have been found to be strongly associated with lower likelihood of current tobacco/areca nut use. The curriculum to be developed and tested will focus on informing adolescents on the risks of tobacco product/areca nut use as well as training adolescents on social influence resistance skills in a culturally appropriate way, using innovative use of video vignettes. The intervention will seek to influence behavior (i.e., prevention and reduction of tobacco product/areca nut use) by countering the cognitive predispositions to tobacco product/areca nut use (e.g., by increasing risk perceptions, reducing positive expectancies) and by providing youths with skills to resist specific social influence to use tobacco products/areca nut on specific occasions. RESEARCH DESIGN This is an experimental study. Of the eight public middle schools in Guam, four schools will be randomly assigned to the "Treatment" and four to the "Control" condition. Students in the Treatment schools will receive the classroom-based curriculum whereas students in the Control schools will receive the standard health education curriculum (treatment-as-usual). Participants will be assessed at pre-test, immediate post-test, and six months after the post-test. Note that the decision to provide an intervention, albeit weaker, to the control group was guided by the ethics of good community practice. In preliminary discussions with schools, principals unanimously expected an intervention. Participants Participants will be 600 Guam middle school students. Based on a pilot study, the mean age of the participants is likely to be 12.7 years (SD = .89); 49% of the participants are likely to be girls; and 47% of the participants are likely to be Chamorus, 27% other Pacific Islanders, and 25% non-Pacific Islanders (i.e., Filipinos, Chinese, Japanese, white, and others). Treatment and Control conditions will represent roughly equal proportion of participants. Recruitment All eight public middle schools will be involved in the study. The Guam Department of Education (GDOE) has been very supportive of this study. At each school, seven classes will be randomly selected to participate in the study. Intervention delivery The classroom curriculum will begin in the Treatment schools within two weeks of the pre-test. The social media campaign will begin in all schools around the same time. Below we provide details on the delivery of each intervention component. Classroom-based curriculum: Curriculum implementation. The curriculum will be delivered by teachers of the Treatment schools who normally teach physical health education classes. The curriculum will be implemented once a week for 45-60 minutes over a 10-week academic quarter in a required physical/health education course. All sessions of the curriculum will follow the same basic format of Ho'ouna Pono. Each lesson will be aligned with the Guam Content and Performance Standards for Health (6-8 grade) and/or the National Common Core Standards in public education. The teachers will be trained to implement the curriculum through a hybrid (i.e., in-person and virtual), credit-granting course sanctioned through the GDOE. Data collection Data will be collected at three timepoints from students in both Treatment and Control conditions: pre-test (baseline), immediate post-test (i.e., approximately a week after the last classroom-based curriculum session is delivered), and six months after the post-test. Data will be collected using self-report questionnaires, although self-report of cigarette use and areca nut use will also be validated using biochemical verification methods in random subsample of participants. At each timepoint the same questionnaire will be administered across Treatment and Control groups. The self-report questionnaire will include the following assessment measures. Measures Demographic measures. Age, gender, socio-economic status, and ethnicity will be assessed. Socio-economic status (SES) is measured based on participants' report on their parents' level of education. However, past research shows that a good proportion of adolescents do not know about parental education level. Hence, a measure of household crowdedness (i.e., the number of usual residents in a household divided by the number of rooms in the house) will also be included to assess SES. Ethnicity will be assessed with a single item previously used by us in Guam. A measure of acculturation will also be included. Intrapersonal/Dispositional variables. These will include self-control, assessed in terms of behavioral and emotional self-control; sensation seeking; social competence in resisting social influence; general self-efficacy; refusal self-efficacy; and self-esteem. The items specific to substance use will be adapted for cigarette smoking, e-cigarette use, chewing tobacco use, and areca nut use. Social/Interpersonal variables. These will include family and friend areca nut/tobacco product use, perceived norms, and social normative beliefs. Perceived norms and social normative beliefs will be adapted for cigarette smoking, e-cigarette use, chewing tobacco use, and areca nut use. Cultural/Environmental variables. These will include availability of areca nut/tobacco at home and school, parental permissiveness, knowledge about areca nut/tobacco products, areca nut/tobacco product use expectancies, and areca nut/tobacco product use harm/risk perceptions. Exposure to content relevant to tobacco products and areca nut on social media. Exposure will be assessed using 12 items previously validated by us for tobacco products and areca nut. The items assess exposure to user-generated content, traditional marketing, and prevention messages. Areca nut and tobacco product use susceptibility. Areca nut use and tobacco product use susceptibility will be separately assessed for cigarette smoking, e-cigarette use, chewing tobacco use, and areca nut use with a measure of adolescent substance use intentions (13), adapted for each substance. Areca nut and tobacco product use. The study will assess areca nut only use, combined areca nut and tobacco use, cigarette smoking, e-cigarette use, and chewing tobacco use separately. For each of these substance use outcomes, the study will assess lifetime (ever) use (e.g., "Have you ever used a areca nut in your lifetime?") and past-30-day use (e.g., "In the past 30 days, on how many days did you use areca nut?). Self-report measures of adolescent substance use have been extensively studied for their validity, by using multiple methods, and they have consistently been found to be valid. Data Analysis Data analysis will be performed with the help of the PIPCHE Biostatistics Core. The primary goal of the analysis is to examine the efficacy of this group randomized trial (16) for substance use prevention among middle schoolers. Aim 1 will compare the change in tobacco and areca nut beliefs, attitudes, initiation and use over time between the two randomization conditions, accounting for the clustering of classes within schools and the randomization unit of schools. In particular, the following mixed regression will be performed: where Z is a vector of covariates for each student at each time point, T is time, G is randomization group, TxG is the interaction of time and group, and η is a random effect for clustering of classes within schools and δ is a random effect for repeated measures within student. The link function ƒ is linear for continuous outcomes, such as a belief score, and logistic for dichotomous outcomes, such as initiation; outcomes will be transformed as needed to meet model assumptions. Potential covariates would include age, sex of student, grade, ethnicity, parent's education level, sensation seeking and teacher sex. Time T will be parameterized as indicator variables for baseline, post-intervention, and six month follow-up, or as continuous months if warranted by a monotonic change over time. The intervention effect will be assessed by the F-test in linear models and chi-square test in logistic models for the βTG parameter, with the degrees of freedom adjusted to account for the eight randomization units. Covariate-adjusted means and 95% confidence intervals will be estimated from the models for each level of randomization group and time point. The SAS procedure GLIMMIX will be used to fit these mixed models, which can accommodate data from students with missing data at some time points and allow for the results to be more generalizable to the student population. However, the study will assess the effect of missing data by comparing characteristics of students with complete and incomplete data and will perform multiple imputation if needed. As exploratory analyses, the study will test for differential intervention effects in subgroups, such as defined by student ethnicity, by adding interaction terms between subgroup and treatment/time variables and assessing the significance of the three way interaction. The study will also explore the effect of dose of the intervention, by studying change in outcomes by number of session attended among students in the intervention schools, and the effect of fidelity assessments of the intervention per classroom. Statistical Power The sample size of eight schools, with four randomized to each condition, each with seven classes of 20 students, is sufficient to ensure adequate statistical power to detect meaningful differences between intervention arms in all outcomes. For Aim 1, the effect size, Cohen's d was calculated based on a z-test comparing means and a chi-square test comparing proportions, with n=8 schools, with four schools per randomization condition, each with 20 students. The variance of the differences was adjusted for the Design Effect of the present study, accounting for the clustering in a group-randomized trial (GRT) by adjusting the test degrees of freedom to the number of communities. The intraclass correlation coefficient (ICC) of outcomes within schools and within class in schools was set to 0.01-0.02, based on past school-based studies, for an overall ICC varying between 0.02 to 0.04. The study assumed a critical level α of 0.05 (two-sided) or 0.005 (Bonferroni-corrected for 10 tests), a power of 80%, and up to a 20% attrition in sample size at post-intervention. The effect size for a test of means is medium at 0.29-0.38 for α=0.05 and at 0.37-0.50 for α=0.005. The following differences between conditions in changes in proportions will be detectable with α=0.05: 0% versus 4.0% and 3% versus 10.5%; and with α=0.005: 0% versus 6.5% and 3% versus 15.5%.