Chapter 3: Methods
3.2 Quantitative Data – 2006 Aboriginal Peoples Survey
3.2.1 Sample
the selection of individuals in the second phase. A total of 1,538 households were removed from the APS to avoid overlap with other surveys. Of the 61,041 individuals chosen for the APS, 13,238 were children (6 to 14 years old).
This study focused on only First Nations and Métis children aged 6 to 14 years. Inuit people were excluded because their unique BMI profiles and body fat distribution would need to be separately
accounted for in the analysis (Galloway et al., 2011; Hopping et al., 2010). Inuit people typically have shorter legs and high “trunk-to-stature proportions” which render the European BMI thresholds especially inapplicable to this population (Galloway et al., 2011). Moreover, the study focus is on Aboriginal peoples living off reserve, and the majority of Inuit people live in northern Canada where the factors impacting food security and access are dramatically different (Hopping et al., 2010; PHAC, 2011). Also, southwestern Ontario has much larger Métis and First Nations populations as compared to Inuit, thereby making the focus group and quantitative results more comparable (Statistics Canada, 2009b).
3.2.2 Data Collection
APS data were collected by Statistics Canada via telephone interviews with PMKs of Aboriginal children aged 6 to 14 years. Children 12 to 14 years old could complete the telephone survey themselves with parental permission. Personal interviews were conducted with PMKs in Labrador, the Northwest Territories (excluding Yellowknife), and Inuit regions, or if participants were unable to do the telephone interview (Statistics Canada, 2009b).
3.2.3 Measurements
Data for a wide range of social, economic, and health-related variables were collected by the APS. The outcome of interest in this study was overweight or obese BMI classification, while the predictors of interest were food insecurity, F&V intake, and junk food intake. Several other variables
were analyzed in this study as potential confounders of either weight status or food insecurity. These variables are listed in Section 3.2.3.3.
3.2.3.1 Dependent Variable
The dependent variable, BMI category, was determined using PMK-reported height and weight. The APS asked, “How tall is _____ without shoes on? (Best estimate)” and “How much does _____ weigh? (Best estimate),” in order to calculate BMI. The APS includes two measures of BMI: the Centers for Disease Control and Prevention (CDC) cut-offs and the IOTF’s age- and sex-specific cut-off points for obesity in children and adolescents. IOTF cut-offs are internationally recognized and based on BMI centile curve data from six countries, unlike the CDC cut-offs, which are based solely on American data (Cole et al., 2000). Hence this study used IOTF cut-offs, which classified children as being “underweight or normal,” “overweight,” or “obese.” In the statistical analysis, BMI was coded as either a dichotomous variable (overweight/obese versus normal/underweight) or as having three response levels (obese, obese or overweight, obese or overweight or normal/underweight) depending on the statistical procedure. Statistical analyses are discussed in Section 3.2.4.
3.2.3.2 Independent Variables
The key independent variables of interest were children’s food security status, F&V intake, and junk food intake. The APS measures food insecurity at the individual level rather than the household level. Food insecurity was determined by asking the question: “Has _____ ever experienced being hungry because the family has run out of food or money to buy food?” The limitations associated with the food insecurity measure are discussed in Section 6.2.
Studies indicate that there is not a clear consensus on what is considered a fruit, vegetable, or junk food, as these definitions vary depending on food preparation and processing, cultural perceptions,
study, fruits refers to “the seeds and surrounding tissues of a plant […] that have a sweet or tart taste,” and vegetables are the “edible plant parts including stems and stalks, tubers, bulbs, leaves, flowers, some fruits (cucumber, pumpkin, tomato), and seeds” (Pennington & Fisher, 2009). Junk food includes foods that are energy-dense and nutrient-poor (i.e., candy, soft drinks), as well as some F&V that are prepared in ways that significantly alter the food’s nutrient profile and energy density (i.e., potatoes to French fries) (Pennington & Fisher, 2009; Roark & Niederhauser, 2012; Thompson et al., 2011). Using these
definitions as a guide, responses for the following foods in the APS were analyzed: “Fruit (not fruit juices),” “Green salad,” “Potatoes,” “Other vegetables,” “French fries, potato chips and pretzels,” and “Candy, soft drinks, cakes, pies, etc.” Fruit, green salad, potatoes, and other vegetables formed the F&V category. French fries, potato chips and pretzels, and candy, soft drinks, cakes, and pies were put in the junk food category. Studies have associated F&V and junk food consumption with obesity outcomes; hence these food groups formed the focus of the analysis.
In order to determine the frequency with which children consumed particular foods, PMKs were asked: “Last week, on how many days did ______ consume the following foods and beverages?” The following response options were provided: Everyday, 5 or 6 days per week, 3 or 4 days per week, 1 or 2 days per week, and Never. These five response categories were also included in the statistical analyses to avoid losing information by creating arbitrary categories such as “high,” “medium,” and” low” intake. Overall diet quality could not be assessed given the limitations of the APS data, which are discussed further in Section 6.6.
3.2.3.3 Control Variables
Potential confounders of either food insecurity or obesity were identified based on a review of previous literature. Age, gender, region, lone parent status, number of people living in the household, household income, PMK education, birth weight, breastfeeding, sports activities or lessons attended per week, and number of hours per day spent watching TV, playing on the computer, or playing video games
were controlled for in the analysis. Census Metropolitan Area (CMA) is defined as an “area consisting of one or more neighbouring municipalities situated around a core” (Statistics Canada, 2012). The CMA must have a population of at least 100,000, and 50,000 or more of these residents living in the core (Statistics Canada, 2012). All APS response categories were for these variables were kept intact for the analysis, with the exception of physical activity and sedentary behaviour variables. A new variable called “sports” was created to represent the frequency with which children played sports every week.The variables TV watching, computer time, and video gaming were combined to create the variable
“sedentary.” Reading was not included as a sedentary variable for several reasons. Shields & Tremblay (2008) found that reading was not associated with obesity for adults, partially because most people did not spend large amounts of time reading per week (Shields & Tremblay, 2008). While there are currently no child-specific studies on reading and sedentary behaviour as a predictor of obesity, other behaviours such as screen time are a greater concern since children spend much more time engaging these activities. Also, the way the reading question was structured in the APS would not have allowed it to be
incorporated into the sedentary behaviour variable. The frequency response options were categorized as times a child read per week or month, as opposed to number of hours per day. Table 1 shows all of the response and predictor variables used in the statistical analysis, as well as the respective reference categories used when entering these dummy variables into the regression models. The majority of the reference categories were chosen based on ease of interpretation, with the exception of the references for region, household income, and PMK education, were the highest frequency categories in the sample were chosen. Different reference categories were used for the diet variables for First Nations and Métis
Table 1: Response and Predictor Variables
Variable Response Categories
BMI category Binary Logistic Regression:
Overweight/Obese Underweight/Normal* Proportional Odds Model:
Obese
Obese or overweight*
Obese or overweight or normal/underweight* Food insecurity
Fruit and vegetable intakea
Food insecure Not food insecure Everyday*
5 or 6 days per week 3 or 4 days per week 1 or 2 days per week Never*
Junk food intakeb Everyday*
5 or 6 days per week 3 or 4 days per week 1 or 2 days per week Never* Gender Male* Female Age 6 to 8 years* 9 to 11 years 12 to 14 years
Region Census Metropolitan Area*
Other urban Other rural
Arctic (deleted from analysis)
Lone parent status Lone parent household
Not lone parent household* Number of people living in the household 2 people*
3 people 4 people 5 people 6 people
Table 1 continued.
Variable Response Categories
Household income PMK education Less than $20,000 $20,000 to $39,999* $40,000 to $59,999 $60,000 to $79,999 $80,000 to $99,999 $100,000 and over Lower than high school High school diploma*
Certificate or diploma lower than university
University certificate or diploma below a Bachelor’s University completed, at least a Bachelor’s
Other
Birth weight Less than 2267 grams*
Between 2267 and 3174 grams Between 3174 and 4081 grams 4081 grams and over
Breastfeeding Never*
6 months or less 7 to 12 months More than 13 months
Breastfed, but length unknown Unknown
Sports Never*
Less than once a week 1 to 3 times per week 4 or more times per week Sedentary activity
(Video gaming, TV or computer time) None* 1 hour per day 2 hours per day 3 hours per day 4 hours per day 5 hours per day
*The asterisk refers to the categories which were used as reference groups.
a The reference categories for fruit and vegetable intake were different for First Nations and Métis children. For
First Nations children, the reference category was “Never,” and for Métis children the reference was “Everyday.”
bSimilarly, the reference categories for junk food intake were “Never” and “Everyday” for First Nations and Métis
3.2.3.4 Additional Variables of Interest
Frequency of experience with food insecurity as well as coping strategies were not used in the regression analyses, however the sample distribution was reviewed in order to better understand the severity of food insecurity in the populations studied. PMKs who answered “Yes” to the food insecurity question were further asked, “How often?” with the following response options: “More often than the end of each month,” “Regularly, end of the month,” “Every few months,” “Occasionally, not a regular
occurrence,” “Don’t know,” and “Refused.” Additionally, those who answered, “Yes,” to their children experiencing hunger were also asked: “How do you cope with feeding ______ when this happens?” PMKs were provided a list of options and asked to check all that applied: “Parent/guardian skips meals or eats less,” “Children skip meals or eat less,” “Cut down on variety of food family usually eats,” “Seek help from relatives,” “Seek help from friends,” “Seek help from social worker/government office,” “Seek help from food bank (emergency food program),” “Use school meal program,” “Other,” “Don’t know,” and “Refused.”