2008 Adolescent Health Survey (AHS IV)
The Adolescent Health Survey is conducted every five years and examines a broad range of youth behaviours. Four of these surveys have been conducted since 1992. The latest survey was completed by 29,440 students in 1,760 classrooms between February and June of 2008. The AHS IV covered 50 of BC’s 59 school districts which contain 92% of all students enrolled in grades 7 through 12 in public schools across the province. Coverage rates were at or near 100% in all Health Service Delivery Areas (HSDA), except for Northeast HSDA and Fraser Valley HSDA.
McCreary Centre Society researchers conducted analyses on the 2003 AHS to increase the ability for results’ comparison across other surveys both provincially and nationally. Additional questions regarding students’ substance use ‘last Saturday’ were also incorporated into the 2008 survey to make it easier to explore students’ recent weekend use patterns. Another element of the ‘last Saturday’ question inquired specifically about the number of marijuana joints smoked at one time, as higher doses on a single occasion can be a strong predictor of acute harm.
Statistics Canada weighted the data to ensure it was representative of youth in grades 7 through 12 in every region of the province. All comparisons and associations reported in this study have been tested and are statistically significant at p < .05. However, the graphs and charts show frequencies that are not necessarily statistically significant at every point.
Questions about alcohol, tobacco, marijuana and other drug use are of special relevance to this bulletin. For students who indicated they have used these substances, ensuing questions inquired about frequency of use and related consequences. Students who reported that they have not used drugs, alcohol or tobacco were directed to skip these questions. A detailed methodology fact sheet discusses the source and rationale for the questions used in the AHS IV.
Adolescent Health Survey Validation Study
Survey results from the EKASS Adolescent Drug Use Survey, the McCreary Centre Street Youth Survey, and the 2002 McCreary Adolescent Health Survey were analyzed to examine inconsistencies in response rates between surveys on certain questions. Further, adolescent focus groups were used to get feedback on formatting and language issues associated with the surveys.
A mixed-methods approach informed our evaluation:
- Participating in a national teleconference working group of those who administer school-based provincial drug use surveys to develop consensus about core indicators;
- Secondary analysis of existing data, including comparisons of missing and inconsistent responses between two surveys with different administration methods;
- Incorporating new items on an original survey and evaluating possible changes in response; and
- Directly engaging youth in evaluating the meaning and formatting of existing and new items for their perceptions of clarity and ease of response.
Data Sources
East Kootenay Adolescent Drug Use Surveys, 2005 and 2007
The East Kootenay Adolescent Drug Use Survey was designed to be completed by all Grade 7 to 12 students in the East Kootenay area of southeastern British Columbia (see background). The intent of the survey was to assess substance use patterns, substance use behaviours, attitudes around substance use, and related risk behaviours amongst the target group. This region is a large rural area with a population of roughly 80,000 people, and is composed of three separate school districts. Thirty public schools and one private Christian school participated in the survey.
The three school districts which are in the East Kootenay area agreed to conduct the survey in their schools. Copies of the survey were provided for school boards to review prior to approval. The administrators of each of the target schools were advised of the survey and the time it was to take place. An explanatory letter home advised parents of the intent of the survey, that participation was voluntary, and encouraged them to contact the principal researcher if they had concerns.
The survey was administered in February 2007. Each school selected a given day within the week, in which all students would receive the survey. Students were not advised ahead of time that they would be completing a survey, but were allowed to decline participation at the time of the survey. In most schools, counselling staff arranged for distribution of surveys to the classrooms. Staff were provided with an instruction sheet describing how the surveys were to be handed out and collected. Student volunteers or school staff handed out the survey in each class. In lower grades, it was often the classroom teacher who handed out the surveys. In the middle and higher grades, schools either had the teachers or assigned students to hand out the surveys.
The survey took approximately 20 to 25 minutes to complete. Each classroom was provided with a set of instructions, which were read aloud, explaining the purpose of the survey and the proper way to complete it. The instructions explicitly told students not to put their names on the survey, and defined a number of terms such as 'use of alcohol', 'mushrooms', and 'prescription or over-the-counter drug use'. Students completed the surveys and placed them in manila envelopes placed at strategic points around the classroom. The envelopes were sealed, collected and returned to the EKASS office.
The first EKASS Drug Use Survey was conducted in 2002. Questions were developed based on a number of other surveys and were designed to get information on substance use history and frequency, and on the youth's sense of connection with family, friends, school, etc. The original survey also included a number of questions not found in other surveys that tried to assess reasons why youth used or did not use, and reasons why youth stopped or reduced their use, the type of people youth used with, as well as when and where they used. The 2005 survey expanded on the original survey. Changing to scanable forms and software required changes in formatting for questions. The 2005 survey also added questions about substance-use-related risk behaviours or consequences, such as impaired driving, conflict with family, friends and school, and unwanted or unplanned sexual activity.
The 2007 survey was similar to the 2005 survey, but included new questions and more formatting changes. In addition to past month, past year, and lifetime use, students were also asked about use on the Saturday prior to the survey and on the day before the survey. Two questions were added asking about past and current tobacco use, and two questions asking about the use of 'energy drinks'. Students were also asked about the volume of alcohol they consumed per drinking episode, and were given specific guidelines to define a 'drink'. Similarly, students were asked about the amount of marijuana they smoked during a given episode. An additional question was added asking about exchanging sexual activities for substances. Prior questions about who youth used with, and location and timing of use were expanded into three sets of questions, one related specifically to alcohol use, one to marijuana use, and one to other drug use. Similarly, the questions asking about being a passenger or operating a vehicle after substance use were expanded into three sets of questions, with a set each for alcohol, marijuana, and other drug use.
BC Adolescent Health Survey 2003
The BC Adolescent Health Survey is a province-wide stratified cluster sample of students in classrooms from grades 7 through 12. The pencil and paper survey has approximately 140 items focused on a variety of health and risk issues for youth, not just substance use, and has been conducted in 1992, 1998, and 2003. The sampling frame is based on enrolled students, and the sampling strategy was developed in consultation with Statistics Canada to be representative at the provincial level, within the five Health Authorities and within the 15 Health Service Delivery Areas (HSDA) throughout BC. School districts are invited to participate, and a sample of classrooms stratified by grade is randomly drawn from across each HSDA. Each district determines whether they will require parental notification and student consent or written parental consent and student assent, or some combination. The sampling level is increased when a school district requires active parental consent, as that tends to significantly reduce the participation rate. Where a classroom has been selected within a non-participating school district, it is reallocated to participating districts based on equal probability of selection methods.
In 2003, 45 of the 59 school districts participated, and about 25% of them chose active parental consent procedures. More than 1550 classrooms were surveyed, with more than 30,500 students, resulting in an overall 76% participation rate. Public health nurses and nursing students who received standardized training administered the surveys in the classrooms, with teachers out of the room, and provided consistent assurances of confidentiality and anonymity. In addition, the survey has a self-sealing cover that does not contain questions, and on each page there is a reminder at the bottom that the results are confidential. The survey was completed in one class period, or about 45 minutes, by most students who speak and read English. The questionnaire incorporates several items about the use of tobacco, alcohol, cannabis and other substances, including ever use, frequency and amount of alcohol use, age of first use, harms from substance use, and risky behaviours related to substance use, all located near the middle of the survey.
After data collection, Statistics Canada further assists in weighting the data to adjust for the differential probability of participation, the response rates, and the clustering design effects. They also weight the data to develop estimates based on actual enrolment (N=~300,000).
Methods to Evaluate Item Comprehensibility and Formatting
Four focus groups, with five to ten youth of both genders between the ages of 14 to 20, were conducted to explore youths’ perceptions about the current questions on the BC AHS, as well as some of the items from the EKASS, items from a recent street youth survey and an alternate school survey from McCreary Centre Society, and items from other in-school drug use surveys across Canada. One of the focus groups included exclusively Aboriginal youth; two focus groups included diverse youth with high substance use experience, and the fourth included five youth with diverse ethnic backgrounds but low substance use exposure, including three whose families spoke a language other than English at home, all from a high-SES area of Vancouver. In addition, three youth (two males, one female) have been interviewed to date by completing a pilot instrument followed by cognitive processing interviews that focus in-depth on the perceptions and meanings students saw in the item wording. Focus groups lasted approximately 90 minutes and included pizza and beverages.
Analyses
Secondary analyses of existing data explored missing response rates for key demographic variables and substance use, consistency between related items on different pages of the survey, and relationships between demographic characteristics and response patterns. The new 2007 EKASS survey was in the field during the project, and so only the first 60% or so of surveys were included in this analysis. Because the data scanning software has limited capacity for recoding variables or conducting statistical analyses, it should be noted that for the 2005 survey (N=4,709), the EKASS research team originally harmonized the responses within the survey by imputing missing responses where possible, based on existing responses elsewhere in the survey, e.g., a student did not respond to ever used in lifetime on the front page but indicated an age of first use on a subsequent page, or where a student reported having used in the past month but was missing or inconsistent in also reporting having used in the past year, or lifetime use. In 2007, data from the first three schools returning surveys (N=1,384) were similarly harmonized, and 881 from the next schools were not, to provide an estimate of how much alteration had potentially been made in 2005 for reports using the original software. However, both EKASS 2005 and 2007 were converted to datasets in SPSS for the analyses conducted in this current project. Similarly, the BC AHS 2003 was evaluated using SPSS, both for analyses that included the entire dataset (unweighted N=30,588) and then separately for the East Kootenay HSDA (unweighted N=1,772).
Analyses included descriptive frequencies of missing, linear or logistic regressions to explore demographic and other predictors of missing data, and McNemars’ test of paired data to examine inconsistent responses within individual surveys between pages. Analyses of group and individual interview data focused on specific recommendations by youth, common areas of preference in wording, as well as noting areas of confusion or varying interpretation of items.
Developing Consensus on Core Indicators Across Provincial School-Based Surveys
As part of this effort, Dr. Saewyc participated in regular teleconference and email meetings of the principal investigators leading the provincial school drug surveys for the four Maritime provinces, Ontario, Manitoba, Alberta and British Columbia, in a process facilitated by CCSA and funded by Health Canada. The group shared their items and experiences with survey administration, sampling issues, and discussed both core and ancillary indicators. The group also met in Victoria in May 2007 during the International Research Symposium in order to continue the dialogue. The consensus built within the group on key indicators was also shared with the other portions of the BC Pilot, to assess the feasibility of creating common indicators across different population groups (i.e., adults, high-risk populations, etc.). As well, the suggestions for new measures put forth by the BC Pilot group regarding recent use measures was communicated to the national working group, who decided to wait until BC had completed the pilot work and a first round of provincial-level implementation and testing before incorporating such measures into national guidelines on core indicators.
The group also discussed the preparation of a national report during the next cycles of all the surveys, using the agreed-upon core indicators and analytical and reporting strategies that will render provincial estimates comparable to each other, albeit with the usual caveats when different sampling and administration methods are used.