Study design
The Youth Smoking Survey (YSS) is a biennial, nationally generalizable school-based, paper-and-pencil survey that measures determinants of tobacco use among youth [20]. The target population was students in grades six through 12 (aged 11–18) at public and private schools (n = 450) in nine provinces. Those residing in the province of Manitoba, or territories of Yukon, Nunavut, and Northwest Territories and those living in institutions or on First Nations reserves were excluded (representing about 4 % of the Canadian population) from the 2012/2013 cycle. Surveys were pilot tested to assess the logic and student understanding of the questions. Approximately 73 % of respondents participated with passive parental permission, and 27 % participated with active parental permission. The YSS survey was administered during class time, and participants were not remunerated. Survey development, design, weights, response rates, and data collection protocol for the 2008 YSS have been published [20]. Overall, the school response rate (the percent of schools that participated in the study once approached) was 64 % (range 38 % in Ontario to 96 % in Newfoundland). The overall student response rate (the percent of eligible students within participating schools) was 72 %. The 2012/2013 YSS was administered to 47,203 youths in grades six through 12 attending schools (in Quebec, secondary school ends at grade 11). Given the low prevalence of hookah use among grades 6–8 students (2.0 % had ever tried hookah; 0.9 % reported using hookah in the last 30 days, and 0.3 % had reported using flavored hookah in the last 30 days), this study restricts analyses to grades 9–12 students (n = 27,404). Data from the 2012/2013 cycle were collected between November 2012 and June 2013. Data from the 2010/2011 YSS were also used to examine hookah use over time. Recruitment, participation, data entry, and cleaning procedures from the 2010/2011 cycle were identical to those in the 2012/2013 cycle. In 2010/2011, however, data collection excluded youth residing in New Brunswick rather than Manitoba. Data were analyzed in 2014. This study was approved by the University of Waterloo Human Research Ethics Committee, the Health Canada Research Ethics Board, and appropriate School Board and Public Health Ethics committees.
Measures and data sources
Relevant, dichotomous outcomes from the 2012/2013 YSS dataset included ever use and last 30 day use of hookah, and students’ reporting that they believe using hookah is less harmful than smoking cigarettes. Students were asked, “Have you ever tried any of the following?,” with a variety of response options including “Using a water-pipe (hookah) to smoke shisha (herbal or tobacco),” and “I have not tried any of these things.” Ever users of hookah were defined as those who indicated they had tried the water-pipe (hookah) option. Students were also asked, “In the last 30 days, did you use any of the following?” Response options included a variety of tobacco products, including “a water-pipe (hookah) to smoke shisha (herbal or tobacco),” and “I have not used any of these things in the last 30 days.” Last-30-day hookah users were defined as those who indicated they used a water-pipe (hookah) in the last 30 days. Last-30-day use is a commonly used standard of current use, originating with the Centers for Disease Control. Finally, students were asked, “Do you believe that using a water-pipe (hookah) to smoke shisha (herbal or tobacco) is:” with response options, “More harmful than smoking cigarettes?;” “Less harmful than smoking cigarettes?;” and “Neither more harmful nor less harmful than smoking cigarettes?” Those who responded “Less harmful than smoking cigarettes” were defined as perceiving hookah to be less harmful than cigarettes.
Independent variables included the respondent’s sex, grade [9–12], province (where the four Atlantic provinces were grouped together based on their cultural similarity and the relatively small n), self-reported race/ethnicity, (white, black, Asian, Aboriginal, Latin American, or “other”), cigarette smoking status, and the amount of weekly spending money (in Canadian dollars) received. Respondents who had smoked 100 or more cigarettes in their lifetime and smoked cigarettes in the past 30 days were considered current smokers and were compared to all others: Former smokers were those who reported smoking 100 or more cigarettes in their lifetime but did not smoke in the last 30 days; non-smokers were those who reported smoking fewer than 100 cigarettes in their lifetime. Weekly spending money was categorized as none, $1–$20, $21–$100, and more than $100.
Differences in ever and last-30-day use of hookah between the 2010/2011 and the 2012/2013 YSS cycles were also assessed. In 2010/2011, ever use and last-30-day use were assessed via the same question format with a slight difference in wording. Ever users of hookah were those who reported ever trying “a water-pipe to smoke tobacco (also known as hookah, shisha, narghile, hubble-bubble, or gouza).” Last-30-day hookah users responded yes to the question, “In the last 30 days, did you use a “water-pipe to smoke tobacco (also known as a hookah, shisha, narghile, hubble bubble, or gouza)?”
Statistical analysis
Survey weights were used to adjust for sample selection (school and class levels), non-response (school, class, and student levels), and post-stratification of the sample population relative to grade and sex distribution in the total population. Bootstrap weights were used for all regression analyses so that the variances take account of the sample design.
The first objective was to examine the prevalence and correlates of hookah smoking. Descriptive statistics were used to show the prevalence of use of plain and flavored hookah by sex, grade, geographic region, self-reported race/ethnicity, and weekly spending money. Proportions were calculated as percentages to show flavored hookah as a percent of overall hookah use and to show the percent of students who reported believing that hookah use is less harmful than smoking cigarettes. To examine correlates of hookah use among respondents in grades 9–12 with complete data for the variables of interest, two logistic regression models were created to examine independent variables related to the odds of ever and current (last-30-day) use of hookah.
The second objective was to examine students’ perceptions of the harm of hookah smoking and to test whether these perceptions were significantly associated with hookah use. To examine students’ perceptions of hookah smoking, a logistic regression model was created to examine independent variables related to the odds of a student believing hookah is less harmful than smoking cigarettes. Independent variables included sex, grade, geographic region, self-reported race/ethnicity, cigarette smoking status, and weekly spending money. Finally, to examine whether beliefs about hookah use were associated with hookah use, a fourth logistic regression model was created to examine whether beliefs about harms associated with hookah use were associated with odds of current hookah use among all respondents. Covariates included sex, grade, geographic region, self-reported race/ethnicity, cigarette smoking status, and weekly spending money. Finally, a fifth logistic regression model was fitted with all covariates to examine whether survey year (2010/2011 or 2012/2013) was significantly associated with last-30-day hookah use.
Full models initially consisted of main exposure, outcome, and covariates. Assumptions of logistic regressions (e.g., sufficient sample size for single cell counts) were checked, and Akaike’s information criterion (AIC) goodness-of-fit tests were used to check model fit. The final models showed the measure of association between an independent variable of interest and outcome. Logistic regressions were conducted by using PROC SURVEYLOGISTIC in SAS 9.3 (SAS Institute Inc, Cary, North Carolina). To aid interpretation of increases or decreases in probability of hookah use, conditional marginal effects were calculated using the LSMEANS (least squares means) statement in the PROC SURVEYLOGISTIC procedure in SAS, using the “OM” (observed margins) along with the “ILINK” (inverse link) option.