Food-security status was assessed using the Household Food. Security Survey ...... Larsen K, Gilliland J. Mapping the evolution of 'food deserts' in a Canadian.
Research Recherche
Food Purchasing and Food Insecurity Among Low-income Families in Toronto
NAOMI DACHNER, MSc, LAURIE RICCIUTO, PhD, RD, Department of Nutritional Sciences, University of Toronto, Toronto, ON; SHARON I. KIRKPATRICK, PhD, RD, Department of Community Health Sciences, University of Calgary, Calgary, AB; VALERIE TARASUK, PhD, Department of Nutritional Sciences, University of Toronto, Toronto, ON
ABSTRACT
Résumé
Purpose: Factors underlying food-purchasing decisions were examined among a sample of low-income Toronto families.
Objectif. Les facteurs sous-jacents à la prise de décision lors de l’achat d’aliments ont été examinés dans un échantillon de familles à faible revenu de Toronto.
Methods: A cross-sectional survey was completed among 485 families residing in high-poverty Toronto neighbourhoods. Food-security status was assessed using the Household Food Security Survey Module. Open-ended questions were included to examine respondents’ food selection and management practices and their purchasing decisions for six indicator foods. Logistic regression was used to examine associations between factors influencing food-purchasing decisions, perceived food adequacy, and severity of food insecurity. Results: Twenty-two percent of families had been severely food insecure in the past 30 days. Respondents engaged in thrifty food shopping practices, such as frequenting discount supermarkets and budgeting carefully. Price was the most salient factor influencing food-purchasing decisions; the likelihood that families would report this factor increased with deteriorating food security. Preference, quality, and health considerations also guided food-purchasing decisions, but generally to a lesser extent as food insecurity increased. Household food supplies reflected constraints on food purchasing, and they diminished with increasing food insecurity. Conclusions: Despite their resourcefulness, low-income families struggle to feed their families. Dietitians have an important role to play as advocates for adequate income supports to promote food security and nutritional health. (Can J Diet Prac Res. 2010;71:127) (DOI: 10.3148/71.3.2010.127)
Méthodes. Une enquête transversale a été effectuée auprès de 485 familles résidant dans des quartiers défavorisés de Toronto. Le niveau de sécurité alimentaire a été évalué à l’aide du Module d’enquête sur la sécurité alimentaire des ménages. Des questions ouvertes ont été incluses afin d’examiner les pratiques de sélection alimentaire et de gestion des répondants ainsi que leurs décisions d’achat pour six produits alimentaires clés. La régression logistique a été utilisée pour examiner les associations entre les facteurs qui influencent la prise de décision lors de l’achat d’aliments, la perception du caractère adéquat des aliments et la gravité de l’insécurité alimentaire. Résultats. Vingt-deux pour cent des familles étaient caractérisées par une grave insécurité alimentaire au cours des 30 derniers jours. Les répondants utilisaient des pratiques d’achat consistant à se procurer des aliments économiques, par exemple faire les courses dans des supermarchés de rabais et effectuer une soigneuse planification budgétaire. Le facteur influençant le plus les décisions liées à l’achat d’aliments était le prix; la probabilité qu’une famille rapporte ce facteur augmentait avec la diminution de la sécurité alimentaire. Les préférences, la qualité et les considérations liées à la santé guidaient aussi les décisions d’achat, mais généralement dans une plus faible mesure plus l’insécurité alimentaire augmentait. La disponibilité alimentaire du ménage reflétait les contraintes liées à l’achat d’aliments et diminuait avec l’augmentation de l’insécurité alimentaire. Conclusions. Malgré leur ingéniosité, les familles à faible revenu luttent pour nourrir leur famille. Les diététistes ont donc un rôle important à jouer à titre de porte-parole pour un soutien du revenu adéquat qui favoriserait la sécurité alimentaire et la santé nutritionnelle. (Rev can prat rech diétét. 2010;71:127) (DOI: 10.3148/71.3.2010.127)
Introduction Several studies have documented associations between household income and the nutritional quality of food purchases and dietary intakes of Canadians, with low income linked to nutrie50
tionally inferior diets (1-5). Studies from Canada and the United States have also shown that women’s (6) and children’s (7,8) intakes in deprived circumstances fluctuate in relation to available
Revue canadienne de la pratique et de la recherche en diététique – Vol 71 nº 3, automne 2010
Research Recherche household financial resources. Further, household food insecurity—a more sensitive measure of food issues experienced by low-income families than income alone (9,10)—has been associated with dietary compromises of sufficient magnitude to have implications for health among Canadian adults and adolescents (11). While the dietary patterns associated with food insecurity must reflect the impact of severe financial constraints, we have little understanding of the factors underlying food-purchasing decisions among food-insecure households (12).
PURPOSE In 2005, we undertook a study of low-income Toronto families to examine the association between food insecurity and household and neighbourhood characteristics. Survey methods, neighbourhood mapping, and in-depth qualitative interviewing (13) were employed. Two-thirds of families had been food insecure in the past 12 months (13). This is more than seven times the national household prevalence of 8.8% in 2004 (14). Further, more severe food insecurity was found among those with lower incomes and educational levels, reliance on social assistance, and lone motherhood (13). Elsewhere we have reported on the relationship between families’ food-security status and participation in community food programs (13) and local food retail access (15). Here we examine the relationship among factors influencing families’ food-purchasing decisions, their perceived food adequacy, and food security.
METHODS
Recruitment and data collection From November 2005 to January 2007, 501 respondents were recruited from 12 randomly selected high-poverty Toronto neighbourhoods. Quota sampling was used to achieve an approximately balanced sample of families living in subsidized housing and market rental accommodations. Potential respondents were approached at the door by trained interviewers and screened for inclusion. Household members were eligible to participate in the study if they had at least one child aged 18 years or younger, lived in a rental accommodation and had resided in their current dwelling for at least one month, were sufficiently fluent in English to complete an oral interview, and had a gross household income at or below the middle level of a five-level income adequacy scale developed by Statistics Canada (16). Interviewer-administered surveys were conducted with the family member who had primary responsibility for food shopping. Interviews took place in the respondent’s home and were approximately 60 minutes. The study protocol was approved by the Human Subjects Research Ethics Board at the University of Toronto. The response rate was 62%. Sixteen families were excluded because they had incomes that exceeded the income threshold for participation. The final analyzed sample includes 485 families. Household food-security status over the past 12 months and past 30 days was assessed using the Household Food Security Survey Module (HFSSM), which includes questions related to
qualitative and quantitative compromises in foods consumed by adults and children (17). Because of our interest in the immediate circumstances related to food-purchasing behaviour and the adequacy of food on hand, in our analysis we drew upon the subset of questions capturing household experiences over the previous 30 days. These questions focus on indicators of reduced food intake and disrupted eating patterns among adults and children but do not assess manifestations of less severe food insecurity; the 30-day measure therefore assesses severity of household food insecurity but cannot be used to differentiate food-secure from marginally food-insecure households (18). The 30-day measure provides an index of acute food shortages in the household, which has been shown to be strongly correlated with dietary compromise (19-21). To examine food selection in the context of food insecurity, we asked respondents about factors influencing their purchasing decisions for six indicator foods commonly found in Canadian households (3). Milk, fruit, vegetables, meat, and breakfast cereals were selected because of previous research suggesting that their purchase is sensitive to household income (3,5) and foodsecurity status (20,22). In addition, margarine was examined because of the strong association between price and nutritional quality for this product (23). Perceived food adequacy was assessed through questions asking respondents whether they were usually able to purchase the milk, fruit, vegetables, and meat needed for the household, and, if not, whether this was because of lack of affordability, inability to get to the store, or other reasons. Interviewers also recorded the number and types of fruits and vegetables and the brand of any margarine on hand; participants were encouraged to check their pantries and refrigerators to ensure accuracy and to note which, if any, of these foods had been obtained from a food bank. Finally, to enable a more complete exploration of food selection and management practices in the context of economic restraint, we asked participants open-ended questions about the management of household expenses (i.e., food, rent, bills), their decision-making in terms of purchasing indicator foods, and the strategies they used when running low on food or money for food. Interviewers recorded respondents’ answers verbatim. Analysis Households were categorized according to food-security status, using the eight HFSSM questions designed to capture household experiences over the past 30 days. Families with no affirmative responses were considered food secure or marginally food insecure, families with one or two affirmative responses were considered moderately food insecure, and those with three or more affirmative responses were considered severely food insecure over the past 30 days (18). Answers to questions about decision-making in terms of purchasing indicator foods were grouped into six categories (i.e., price, health, preference, purpose, quality, and other) encapsulating the range of factors that participants reported influenced their decisions. Households were then assigned binary Canadian Journal of Dietetic Practice and Research – Vol 71 No 3, Fall 2010
e51
Research Recherche Table 1 Income by food-security status (n=485), by mean +/- standard error Income
All families (n=485)
Food secure/marginally insecure (n=330)
30-day income ($) Income as % LICOa
1986 ± 30 61.9 ± 0.9
2093 ± 30 64.8 ± 1.1
Moderately food insecure Severely food insecure (n=49) (n=106) 1894 ± 173 56.4 ± 2.8
1698 ± 60 55.6 ± 1.3
LICO = low income cut-off a Annual income expressed as a percentage of the Statistics Canada 2006 (before tax) LICO for a family of equivalent size
variables indicating whether or not each factor was relevant to purchasing decisions for a particular indicator food. The most common factors for each food were then examined in relation to families’ food-security status. Statistical analyses were performed using SAS, version 9.1.3 (SAS Institute, Cary, NC), and SAS Survey Procedures were used to account for the clustering in the sampling design. Logistic regression was used to examine the association of severity of food insecurity with factors influencing food-purchasing decisions and with perceived food adequacy. A binary variable for housing type was included in the logistic model to account for the stratification of the sample by families living in subsidized and market rental accommodations. Household composition (number of adults and number of children aged up to five, six to 11, and 12 to 18 years), the respondent’s education level and ethnicity, and season were included as covariates in regression models to account for the potential confounding influence of these variables. Although income is also related to food-purchasing behaviours and food-security status, it was not included as a covariate because its inclusion would have limited statistical power and thwarted our ability to examine differences related to foodsecurity status. Further, income is an insensitive measure of purchasing power because it does not capture additional resources families may have (e.g., savings, credit, gifts, loans). Open-ended questions were analyzed using qualitative methods (24). Responses were carefully reviewed and grouped into concepts or themes that would elucidate the particular area of inquiry. The open-ended questions related to managing household expenses and strategies used when running low on food were analyzed to provide context for the quantitative results; the responses to the questions used to derive the six decision-making factors were analyzed to provide further insight into families’ food-purchasing behaviours.
RESULTS
Sample description The mean annual income of households in this study was 62% of the Statistics Canada low income cut-off (LICO), a measure of low income indicating straitened circumstances (25) (Table 1). Fifty-eight percent of households were headed by a lone parent, most of these being lone mothers, and most respondents had completed high school (Table 2). Our sample was ethnically diverse: 82.5% of respondents were born outside Canada. e52
This finding is consistent with the demographic composition of Toronto’s low-income neighbourhoods (26). Twenty-two percent of households had been severely food insecure and 10% had been moderately food insecure over the past 30 days. Responses to the individual questions used to determine food-security status are presented in Table 3. Nearly all those categorized as severely food insecure reported that adults cut or skipped meals in the 30 days preceding their interview because they lacked the money to buy food; almost one-third reported doing this for more than 14 days in the previous month (data not shown). In severely food-insecure families, about 50% of respondents reported hunger and 25% reported going an entire day without food. While food deprivation among children was less common in these families, 41% said they cut the size of their children’s meals and almost 25% reported that their children went hungry. The resourcefulness of families living under extreme financial constraint was illustrated by responses to open-ended questions about the management of household expenses and strategies when money for food was scarce. Paying rent was generally the priority, while food costs and bill payments were seen as more flexible. Many families routinely received two or more cheques each month, including earnings or welfare payments plus other supplemental income (e.g., child tax benefit). The timing and amount of these cheques often appeared to structure families’ monthly expenditures, with one disbursement typically allotted to rent. Additional cheques were used to make minimum bill payments to prevent the termination of services, and to meet other household needs including food. To stretch their food dollars, respondents reported using a number of strategies at home: making food from scratch, getting creative with leftovers, using fillers to stretch meals (e.g., adding potatoes and carrots to chicken or flour to stew), and using canned foods. At the grocery store, economizing strategies included sticking to a fixed budget, using shopping lists, coupons, and flyers, purchasing sale items, comparison shopping, and buying in bulk. Food-purchasing decisions Among respondents who reported usually purchasing a given indicator food, price was the most salient factor influencing decisions (Table 4). The greater the severity of food insecurity, the more likely families were to report price (Table 4) and to report price as the sole factor in their purchasing decisions for each of
Revue canadienne de la pratique et de la recherche en diététique – Vol 71 nº 3, automne 2010
Research Recherche Table 2 Study participants’ demographic characteristics (n=485) All families Characteristic (n=485)
Moderately food insecure (n=49)
Severely food insecure (n=106)
Main source of income (past 30 days) Employment Social assistance Other government transfersa Otherb
241 (49.7%) 139 (28.7%) 89 (18.4%) 16 (3.3%)
190 (57.6%) 68 (20.6%) 61 (18.5%) 11 (3.3%)
20 (40.8%) 18 (36.7%) 11 (22.5%) 0
31 (29.3%) 53 (50.0%) 17 (16.0%) 5 (4.7%)
Household type Two-parent Lone mother Lone father
202 (41.7%) 265 (54.6%) 18 (3.7%)
155 (47.0%) 159 (48.2%) 16 (4.9%)
20 (40.8%) 29 (59.2%) 0
27 (25.5%) 77 (72.6%) 2 (1.9%)
Number of children 1 2 3 ≥4
177 (36.5%) 166 (34.2%) 92 (19.0%) 50 (10.3%)
124 (37.6%) 116 (35.2%) 54 (16.4%) 36 (10.9%)
16 (32.7%) 17 (34.7%) 13 (26.5%) 3 (6.1%)
37 (34.9%) 33 (31.1%) 25 (23.6%) 11 (10.4%)
Living in subsidized housing
253 (52.2%)
157 (47.6%)
30 (61.2%)
66 (62.3%)
Respondent’s education Did not complete high school Completed high school Some or completed post-secondary education
109 (22.5%) 163 (33.6%) 213 (43.9%)
63 (19.1%) 104 (31.5%) 163 (49.4%)
9 (18.4%) 21 (42.9%) 19 (38.8%)
37 (34.9%) 38 (35.9%) 31 (29.3%)
Respondent’s ethnicity Black Caucasian/European East Asian, South Asian or Southeast Asian Otherc
204 (42.1%) 88 (18.1%) 133 (27.4%) 60 (12.4)
136 (41.2%) 52 (15.8%) 102 (30.9%) 40 (12.1%)
22 (44.9%) 8 (16.3%) 16 (32.7%) 3 (6.1%)
46 (43.4%) 28 (26.4%) 15 (14.2%) 17 (16.0%)
a b c
Other government transfers include Employment Insurance, Workers’ Compensation, Child Tax Benefits, GST credits, and seniors’ benefits. Other income includes child support, rent received from someone living in the household, or informal sources of income. Other ethnicity includes West Asian, Central Asian, Middle Eastern, Central American, South American, and First Nations/Inuit/Métis.
Table 3 Frequency of indicators of disrupted eating patterns and food deprivation among food-insecure families over the previous 30 days (n=155) Indicatora Adult(s) cut size of meals or skipped meals Respondent ate less than felt he or she should Respondent hungry but did not eat Respondent lost weight Adult(s) did not eat for a whole day Cut size of child(ren)’s meals Child(ren) skipped meals Child(ren) hungry
Moderately Severely food insecure food insecure (n=49) (n=106) 26 (53.1%)
99 (93.4%)
22 (44.9%) 4 (8.2%) 9 (18.4%) 2 (4.1%) 4 (8.2%) 1 (2.0%) 2 (4.1%)
92 (86.8%) 51 (48.1%) 41 (38.7%) 28 (26.4%) 43 (40.6%) 27 (25.5%) 24 (22.6%)
The wording of each item included an explicit reference to financial limitation (i.e., “...because there wasn’t enough money for food”).
a
Food secure/ marginally insecure (n=330)
the foods examined (Table 5). Economizing practices appeared to depend on the nature of the food being purchased. For margarine, this simply meant purchasing “the cheapest,” “whatever is on sale,” “the best value,” or whatever provided “the most bang for my buck.” But for perishable items like fruit, vegetables, and meat, the durability of an item and other considerations such as whether the product would get eaten or go to waste were factored into the cost of an item. Some respondents reported buying frozen vegetables and meat because they were cheaper and lasted longer than fresh items. Although the specific kinds of meat that families bought appeared to be influenced by cultural background, many reported buying cheap cuts, especially chicken legs and wings and ground beef. Many also stocked up on sale items or bought large family/economy packages of meat that they apportioned and froze at home. To keep costs down and avoid waste, some respondents reported purchasing small amounts, buying only what was needed after they considered the meal/ eating plans for the week. Cereal was considered expensive, and many reported regularly monitoring grocery store flyers for sales. Preference also guided food-purchasing decisions, particularly in the cases of cereal, fruit, and meat (Table 4). For these Canadian Journal of Dietetic Practice and Research – Vol 71 No 3, Fall 2010
e53
Research Recherche Table 4 Factors influencing participants’ decisions to purchase indicator foods, by food-security status Total Fooda Margarine Price Health Preference Quality Purpose Other Cereal Price Health Preference Quality Purpose Other Fruit Price Health Preference Quality Purpose Other Vegetables Price Health Preference Quality Purpose Other Meat Price Health Preference Quality Purpose Other
Food secure/ marginally insecure
Moderately insecure
Severeley insecure
p valueb
236 (56.3%) 144 (51.4%) 147 (35.1%) 108 (38.6%) 93 (22.2%) 68 (24.3%) 0 — 3 (0.7%) — 2 (0.4%) —
30 (69.8%) 12 (27.9%) 14 (32.6%) — — —
62 (64.6%) 27 (28.1%) 11 (11.5%) — — —
0.0047 0.0006 0.1900 — — —
225 (50.2%) 94 (21.0%) 297 (66.3%) 3 (0.7%) 0 16 (3.5%)
134 (43.8%) 69 (22.6%) 220 (71.9%) — — —
31 (70.5%) 6 (13.6%) 27 (61.4%) — — —
60 (61.2%) 19 (19.4%) 50 (51.0%) — — —
0.0105 0.3556
