WIXLIB

children in a household over the past twelve months (14). Questions are deliberately sequencedusing Rasch modelling and a skip pattern is used for the 18 questions to reduce respondent burden.Questions on more severe experiences of food insecurity are not asked if less severe experiencesare not affirmed (see Appendix A for all 18 questions on the HFSSM and skip pattern).Questions on the HFSSM relate to the presence and frequency of a variety of experiencesrelating to poor food access in the household. These include worrying that food would run outbefore obtaining money to buy more, not being able to consume balanced meals, cutting the sizeof meals, skipping meals, and at its most extreme, going a whole day without eating. All questionsare asked in the context of not having enough money to ensure experiences were due to a lack offinances as opposed to experiences of time restraints or voluntary dieting.Ten questions relate to the experiences of adults and the remaining eight questions relateto the experiences of children. The combination of affirmative responses can be used to determinefood insecurity for adults in a household, children in a household if present, and an overall statusfor the household. In cases where more than one adult or one child is present in the household, themodule is not able to determine the food insecurity status of one single individual.2.1.3. ClassificationHealth CanadaUsing responses from the HFSSM, Health Canada categorizes households into threecategories: i) food-secure, ii) moderately food-insecure and iii) severely food-insecure (15). Table2.1 shows the classification of food security status by the number of affirmative responses to the18-item module. If adult household food security differs from child household food security, themore severe of the two determines household food insecurity status.4

Table 2.1 Household food insecurity status as defined by Health Canada (15)StatusFood-SecureModerate foodinsecuritySevere foodinsecurityInterpretationNo or one indicationof difficulty withincome-related foodaccessCompromise inquality and/orquantity of foodconsumedIndication of reducedfood intake ordisrupted eatingpatterns10-item adult scale8-item child scale0 or 1 item on either scale2 to 5 positiveresponses2 to 4 positiveresponses6 or more positiveresponses5 or more positiveresponsesUnited States Department of Agriculture (USDA)The USDA classifies household food insecurity differently than Health Canada. Mostnotably, categorization of a household as food-insecure begins at a higher threshold, with 3 ormore responses to the HFSSM considered food-insecure. For ease of understanding, where USliterature is presented in this thesis, the number of affirmative responses to the HFSSM used todefine food insecurity is specified.Marginal Food InsecurityArguments for the inclusion of marginal food insecurity (defined by one affirmativeresponse in Canada or one to two affirmative responses in the US) as a separate category fromfully food-secure and moderate-severely food-insecure households have been discussedextensively in the literature due to their distinct vulnerability (5, 16, 17). Supporting evidenceincludes separate and discretely intermediate differences between individuals living in householdswith no affirmative responses and greater than one or two responses in measures of health,healthcare utilization, material hardships, spending patterns and household characteristics (18-21).5

Studies examining the health and well-being of individuals from infancy to adulthoodliving in marginally food-insecure households have identified poorer physical and mental healthdistinct from those in fully food-secure and moderate-severely food-insecure households (18, 2224). A greater description on the impacts of food insecurity on health occurs in Section 2.3.1. Inbrief, living in a marginally food-insecure household is associated distinctly with poorer physicalhealth such as number of chronic conditions, poorer mental health such as mood and anxietydisorders, and greater behavioural problems amongst children such as inattention and academicdifficulties (18, 22-24). Furthermore, an Ontario study demonstrated total healthcare costs over a12 month period per adult living in marginally food-insecure households to be significantly morethan those living in food-secure households and significantly less than those in moderately foodinsecure households in unadjusted and adjusted models (21).A study conducted amongst low-income families in an urban centre in Canadademonstrated that households with one or two affirmative responses to the HFSSM hadsignificantly greater odds of experiencing material hardships and constrained food purchasingcompared to households with no affirmative responses (20). This study also suggested that anincreasing number of affirmative responses was associated with an increasing burden of materialhardships, suggesting that the experience of marginal food insecurity is distinct from experiencesof moderate-severe food insecurity.In the US, an analysis of households with one or two affirmative responses on the HFSSMsuggests that their household characteristics are significantly distinct from households with noaffirmative responses and more similar to those with greater than two affirmative responses (16).In a study of pregnant women in the US, socioeconomic and demographic variables wereassociated with both marginal food insecurity (1-2 affirmative responses to HFSSM) and6

moderate-severe food insecurity, suggesting that differences exist between marginally foodinsecure and fully food-secure households with pregnant women (5). Although distinctionsbetween marginally food-insecure and fully food-secure households have been made amongstpregnant women in the US, no Canadian studies have highlighted differences between the twoclassifications amongst households with pregnant women in Canada.Weekly household food expenditures amongst marginally food-insecure households (1-2affirmative responses to HFSSM) in the US are significantly less than food-secure households andsignificantly more than food-insecure households, suggesting marginally food-insecurehouseholds experience a distinct level of material deprivation (16). Canadian research has sincevalidated similar patterns in household spending amongst marginally food-insecure households (1affirmative response to HFSSM) with spending patterns more similar to moderate-severely foodinsecure households compared to food-secure in the prioritization of basic needs such as food overdiscretionary spending and investments/assets (19).Considering marginal food-insecurity as part of food-secure households mayunderestimate insecure access to food and mask the household characteristics and health impactsof this experience. Measurements that include marginally food-insecure households may betterassist in improved accounting of food-insecure households requiring assistance and the fullspectrum of associations between food insecurity and health.2.2. Epidemiology of Household Food Insecurity in Canada2.2.1. PrevalenceGeneral PopulationThe prevalence of household food insecurity in Canada is determined using the CanadianCommunity Health Survey (CCHS), a national cross-sectional survey representative of 97% of the7

Canadian population. The CCHS excludes individuals living on First Nations reserves, those livingin institutions full-time and certain remote Quebec health regions (see Section 4.1.1. for detailedsurvey design). According to the most recent national estimate from CCHS 2012, 12.6% ofhouseholds or approximately 4 million Canadians experience a degree of food insecurity in Canada(1). It is also estimated that 16% of households with children under the age of 18 experience foodinsecurity (1).Populations Related to Pregnant WomenIn addition to measuring food insecurity on the household level, the prevalence of individualsliving in households experiencing food insecurity is commonly measured due to the relationshipbetween food insecurity and poorer health. No studies have estimated the prevalence of foodinsecurity amongst pregnant women in Canada. However, using data from Statistics Canada fromCCHS, an elevated prevalence of food-insecure individuals amongst related populations have beenidentified. For example, the prevalence of moderate-severe food insecurity was 34-42% higheramongst women of childbearing age (20-44 years) compared to men in 2011-2012 (Figure 2.1)(25). During the life stage chronologically closest to pregnancy for which Canadian data areavailable, an analysis of a pooled sample of women from 2005-2014 who had given birth withinthe past 2 years identified a 17.6% prevalence of food insecurity (26). Households led by lonefemale parents also experienced the highest prevalence of food insecurity by householdcomposition at 34.3% of households compared to male lone parents at 17.2% in 2012, furtherhighlighting gender disparities amongst food-insecure households.8

Figure 2.1 Percent of Canadians in moderate and severe food-insecure households by ageand gender (CCHS 2011/2012) (25)9.1% 9.7%8.9%11.9%11.1%7.8%6.5% 7.7%2.1% 2.9%12 to 19 years20 to 34 years*35 to 44 years*Males45 to 64 years65 years and overFemales* significant difference in prevalence between females and males at p 0.052.2.2. PredictorsIn addition to the prevalence of food insecurity in the general Canadian population,multivariable analyses have been conducted to identify associations between householdcharacteristics and the odds of experiencing food insecurity when accounting for confoundingcharacteristics (27-30). Beyond identifying simple correlations, recognizing predictors of insecurefood access allows for a better understanding of which characteristics are associated with increasedvulnerability in Canada. This can better facilitate the development of public policies by targetingpredictors associated with this problem.Household Income, Homeownership, Source of IncomeAs measurement of food insecurity in Canada is defined by experiences due to financialconstraints, low before-tax annual household income is a strong predictor of experiencing foodinsecurity in adjusted and unadjusted models (27-29). Income accounts for a large component ofwealth within a household and thus its ability to secure access to food. However, there arelimitations to before-tax annual household income as a predictor, as this measure does not captureafter-tax income or other components of wealth such as assets and savings. Household income also9

does not account for variability in expenses between households such as costs for food, medication,debt, and housing.In addition to household income, other economic circumstances such as homeownershipand a household’s main source of income are consistently cited as among the strongest predictorsof experiencing food insecurity in Canada (27-29). These characteristics reflect the capacity forhouseholds to weather negative income shocks such as unexpected decreases in income orincreases in expenditures (31).For example, homeownership represents household wealth and assets beyond income aswell as access to credit, which assists in buffering a household from economic constraints that mayresult in food insecurity (32). However, research has not yet explored potential differences betweenhomeowners with and without a mortgage which may further demonstrate differences in foodinsecurity risk due to variation in debts and assets. In contrast, renters are more vulnerable toinflationary pressures that could contribute to negative income shocks resulting in experiences offood insecurity (32).Notable variation in the odds of experiencing food insecurity has been observed by ahousehold’s main source of income. Most vulnerable relative to households relying onemployment for income are those reliant on public programs such as social assistance (welfare anddisability), employment insurance or worker’s compensation (27-29). Although these policies varyby province and territory, households reliant on social assistance consistently have the highestprevalence of food insecurity of any sociodemographic group with 7 in 10 of these households inCanada reporting some level of food insecurity in 2012 (1). This is in part due to the low levels ofincome support provided by these programs.10

When observed in a multivariable model adjusting for household income, jurisdiction ofresidence, education, and household composition; vulnerability amongst households reliant onsocial assistance remains. Beyond relatively low income benefit payments provided by socialassistance programs, recipients face strict limits on assets and earnings, preventing thesehouseholds from accruing a financial buffer to weather income shocks. Furthermore, householdsreliant on social assistance may have other characteristics unaccounted for in multivariableanalysis in the literature such as lower levels of family/community support and physical/mentaldisabilities.In contrast, households reliant on senior’s incomes (including pensions, dividends andinterest) are at lower odds of experiencing food insecurity, likely in part due to economic protectionprovided by public programs such as Old Age Security and the Guaranteed Income Supplementprograms (27) (see Section 2.2.3. Public Policy and Food Insecurity).Province/TerritoryIn addition to economic and socio-demographic variation between provinces andterritories, residency by jurisdiction is associated with differences in the adjusted odds ofexperiencing food insecurity (Table 2.2) (27, 29). This is likely due to variation in economicpolicies impacting vulnerable populations such as differences in minimum wage, child benefits,childcare policies and taxation. Cost of living also differs by jurisdiction such as food costs andheating costs – both impacting the economic circumstances that determine food insecurity risk (27,33).11

Table 2.2. Relative adjusted* odds of experiencing food insecurity by severity, compared toOntario in 2011-2012 (21)NLPENSNBQCMBSKABBCYKNTNUMarginal FoodInsecurity– –– – ––Moderate FoodInsecurity– –– –– Severe FoodInsecurity – – – –– - increased odds; decreased odds; – no difference;*Adjusted for household income, main source of household income, highest education in household,household structure, housing tenure, Indigeneity, immigration status and urban residence.Adapted from Tarasuk et al. 2019. BMC Pub Health.Household StructureMultiple studies have observed differences in household structure as being associated withthe probability of food insecurity in Canada when controlling for confounding factors. A studyusing the HFSSM with CCHS data from 2011-2012 demonstrated couples with no children hav

Table 5.7 Weighted prevalence of poor health and well-being amongst pregnant food-insecure women compared to food-secure women, CCHS 2005-2014 (n 4,754) Table 5.8 Adjusted odds of poor health and well-being amongst pregnant food-insecure women compared to food-secure women, CCHS 2005-2014