Abstract
Objectives
The prevalence of household food insecurity in Newfoundland and Labrador (NL) fell sharply between 2007 and 2011, but it appears to have risen since then. Our objective was to compare the prevalence of food insecurity between 2011–2012 and 2017–2018 in relation to population socio-demographic characteristics.
Methods
Our analytic sample comprised all NL households in the Canadian Community Health Survey (CCHS) cycles 2011–2012 and 2017–2018, n = 6800. We stratified the prevalence of household food insecurity for each cycle by socio-demographic characteristics and applied multivariable logistic regression models to determine food insecurity odds in 2017–2018 compared with 2011–2012 after controlling for socio-demographic covariates. Interactions of covariates with survey cycle were tested and models stratified when warranted.
Results
The prevalence of food insecurity rose from 12.0% (95% CI 10.5, 13.6) in 2011–2012 to 14.7% (95% CI 13.1, 16.6) in 2017–2018. After adjusting for household socio-demographic variables, the odds of food insecurity was 1.49 (95% CI 1.27, 1.75) in 2017–2018. The prevalence of food insecurity had increased significantly among unattached individuals, renters, households with low educational attainment, and households with income above the Low Income Measure, with concomitant increases in the contribution of these groups to the total provincial prevalence of food insecurity in 2017–2018 compared with that in 2011–2012.
Conclusion
The odds of food insecurity among NL households increased significantly from 2011–2012 to 2017–2018. Given the serious negative health implications of household food insecurity, the province should draw on the lessons from its earlier success in food insecurity reduction to reverse the current trend.
Keywords: Food insecurity, Newfoundland and Labrador, Canadian Community Health Survey, Housing tenure
Résumé
Objectifs
La prévalence de l’insécurité alimentaire des ménages à Terre-Neuve-et-Labrador (T.-N.-L.) a connu une forte baisse entre 2007 et 2011, mais depuis, elle semble avoir augmenté. Nous avons cherché à comparer la prévalence de l’insécurité alimentaire entre 2011-2012 et 2017-2018 en lien avec le profil sociodémographique de la population.
Méthode
Notre échantillon d’analyse comprenait tous les ménages de T.-N.-L. inclus dans les cycles 2011-2012 et 2017-2018 de l’Enquête sur la santé dans les collectivités canadiennes (ESCC), soit n = 6 800. Nous avons stratifié la prévalence de l’insécurité alimentaire des ménages pour chaque cycle selon le profil sociodémographique et appliqué des modèles de régression logistique multivariée pour déterminer la probabilité d’insécurité alimentaire en 2017-2018 comparativement à 2011-2012 compte tenu des covariables sociodémographiques. Nous avons testé les interactions des covariables avec les cycles de l’enquête et stratifié les modèles au besoin.
Résultats
La prévalence de l’insécurité alimentaire a augmenté, passant de 12 % (IC de 95 % 10,5, 13,6) en 2011-2012 à 14,7 % (IC de 95 % 13,1, 16,6) en 2017-2018. Après l’apport d’ajustements pour tenir compte des variables sociodémographiques des ménages, la probabilité d’insécurité alimentaire était de 1,49 (IC de 95 % 1,27, 1,75) en 2017-2018. La prévalence de l’insécurité alimentaire avait augmenté de façon significative chez les personnes seules, les locataires, les ménages ayant un faible niveau d’instruction et les ménages ayant un revenu supérieur à la mesure de faible revenu, avec des augmentations concomitantes de l’apport de ces groupes à la prévalence provinciale totale de l’insécurité alimentaire en 2017-2018 par rapport à 2011-2012.
Conclusion
La probabilité d’insécurité alimentaire des ménages de T.-N.-L. a augmenté de façon significative entre 2011-2012 et 2017-2018. Étant donné les graves conséquences négatives de l’insécurité alimentaire des ménages pour la santé, la province devrait tirer des leçons de son succès antérieur dans la réduction de l’insécurité alimentaire pour inverser la tendance actuelle.
Mots-clés: Insécurité alimentaire, Terre-Neuve-et-Labrador, Enquête sur la santé dans les collectivités canadiennes, mode d’occupation du logement
Introduction
Household food insecurity, the inadequate access to food due to financial constraints (Health Canada, 2007), is a serious public health problem in Canada. In 2017–2018, food insecurity affected 12.7% of Canadian households, including 4.4 million people, the largest number to be recorded living in food insecure households since monitoring began in 2005 (Tarasuk & Mitchell, 2020). Among adults, food insecurity is associated with higher rates of chronic disease (Tarasuk et al., 2013), mental health problems (Jessiman-Perreault & McIntyre, 2017; Men et al., 2021a), premature mortality (Men et al., 2020a), and health care utilization (Men et al., 2020b). Among children, exposure to severe food insecurity has been linked to increased likelihood of asthma, depression, and suicidal ideation in adolescence and early adulthood (Kirkpatrick et al., 2010; McIntyre et al., 2017).
Both the probability and severity of household food insecurity are tightly linked to household socio-demographic characteristics, but risk also varies depending on the province or territory of residence (Tarasuk et al., 2019). Provincially, rates of food insecurity among households with children are inversely related to the provincial unemployment rate, minimum wage, and welfare incomes, and positively associated with the provincial income tax rate (Men et al., 2021b). The prevalence of household food insecurity has consistently been highest in Nunavut, with the rate steadily rising in that territory since 2005 when monitoring began. However, there have been significant decreases and increases in prevalence in some other jurisdictions over time. In 2011, Newfoundland and Labrador (NL) had the lowest rate of food insecurity in the country, at 10.6% (Loopstra et al., 2015). This is the lowest prevalence observed in any jurisdiction in Canada since the start of monitoring. Closer inspection of the available data for NL revealed an unprecedented five percentage point decline in food insecurity between 2007 and 2011 (Loopstra et al., 2015). Although this was a period of economic prosperity for the province, the drop in food insecurity was partly attributed to measures implemented under the province’s poverty reduction strategy launched in 2006 (Loopstra et al., 2015). The strategy included several changes to improve the financial circumstances of social assistance (‘Income Support’) recipients, including an increase and subsequent indexation of benefits. The rate of food insecurity among households reporting any income from social assistance fell from 59.9% in 2007 to 33.5% in 2012. Moreover, 44% of the drop in the provincial prevalence of food insecurity between 2007 and 2011 could be accounted for by the decrease in food insecurity among social assistance recipients. Unfortunately, this progress was not sustained.
The province elected not to measure food insecurity in the 2013–2014 and 2015–2016 cycles of the Canadian Community Health Survey (CCHS), but in 2017–2018, the prevalence of food insecurity in NL was 14.7% (Tarasuk & Mitchell, 2020). That was the second highest prevalence observed among the provinces in 2017–2018 (Nova Scotia was the highest at 15.3%). The prevalence of food insecurity in St. John’s was 17.3%, higher than any other census metropolitan area surveyed (Tarasuk & Mitchell, 2020). Over the period between 2011–2012 and 2017–2018, unemployment rose (Statistics Canada, 2021), but the provincial government reduced the low-income tax threshold (Government of Newfoundland & Labrador, 2014a) and introduced new tax benefits for low-income individuals (Government of Newfoundland & Labrador, 2016). The province also increased seniors’ benefits (Government of Newfoundland & Labrador, 2014b) and introduced a one-time 5% increase to the basic rate for social assistance in 2014 (Government of Newfoundland & Labrador, 2014b). However, social assistance was only indexed to inflation until 2012 (Government of Newfoundland and Labrador, n.d.). During this same period, there were policy reforms federally, including the introduction of Canada Child Benefit (Brown & Tarasuk, 2019). Against this backdrop of provincial and federal policy reforms and shifting macroeconomic conditions in NL, the risk of household food insecurity among specific population subgroups in NL may have also changed between 2011–2012 and 2017–2018.
Drawing on data for NL from the 2011–2012 and 2017–2018 cycles of CCHS, this study was undertaken to explore the marked increase in food insecurity prevalence over time. Our goal was to compare the prevalence of food insecurity in the province in 2011–2012 and 2017–2018 in relation to population socio-demographic characteristics.
Methods
Data and measures
We accessed the Master files of CCHS cycles 2011–2012 and 2017–2018 through the Toronto Research Data Centre. CCHS is a cross-sectional survey administered by Statistics Canada to collect health-related information on the Canadian population. The survey is designed to be representative of 98% of the Canadian population 12 years of age and older, residing in private dwellings (Statistics Canada, 2019). It omits individuals living on First Nations reserves and in institutions and members of the Armed Forces. Our analytical sample consisted of 6800 households (rounded sample size) in NL with complete data on household food insecurity.
Our outcome of interest was household food insecurity over the past 12 months, assessed using the 18-item Household Food Security Survey Module (HFSSM) (Health Canada, 2007). Households with one or more affirmative responses were considered food insecure, and households with no affirmative responses were considered food secure. We also classified households using the more conservative threshold of two or more affirmative responses to identify moderate or severe food insecurity (Health Canada, 2020).
The main independent variable was survey cycle. Co-variates included household income, main source of household income (employment/employment insurance/pension/social assistance/other income), highest level of education in household (less than secondary/secondary school complete/completed post-secondary below bachelors’/bachelors’ or higher complete), household composition (unattached individual/lone parent with children ≥ 18/couple without children < 18/ couple with children < 18/ lone parent with children < 18), housing tenure (homeowner/renter), Indigenous status of respondent (non-Indigenous/Indigenous), and immigration status of respondent (Canadian-born/immigrant). These characteristics have been identified as significant predictors of household food insecurity status in Canada in previous studies (Tarasuk et al., 2019). For household composition, an unattached individual is defined as someone who is either living alone or living with others to whom they are unrelated (Statistics Canada, 2015). Household income was included as a binary variable, differentiating households with incomes at or below the national before-tax Low Income Measure (LIM). This determination was based on income adjusted for household size by dividing by the square root of household size (Income Statistics Division, Statistics Canada, 2016) comparing adjusted household income to the LIM specific to the year of interview. Statistics Canada imputed income for households with missing values (Statistics Canada, 2013a). To minimize the loss of sample, households with missing data for other variables were assigned either a value consistent with the majority in the study population or the value of “Other” when this response was an option.
Statistical analyses
Recognizing the potential for CCHS design changes implemented in 2015 (Statistics Canada, 2016) to influence our results, study population characteristics were compared between survey cycles to identify compositional changes in the study population that could influence overall prevalence of food insecurity over this period.
The odds of any food insecurity and moderate/severe food insecurity by survey cycle were first determined using unadjusted logistic regression. Multivariable logistic regression models were then run to determine the odds in 2017–2018 compared with those in 2011–2012 after controlling for socio-demographic co-variates. Models were run separately on overall food insecurity and moderate/severe food insecurity. The category with the largest number of observations served as the reference group for all categorical variables. To investigate whether there were changes in odds of food insecurity over time among population subgroups, interactions between each of the socio-demographic covariates and survey cycle were tested separately in the fully adjusted logistic regression models. Stratified multivariable logistic regression models were run only for outcomes that were statistically significant. The significance level was set at p < 0.05.
To further explore potential changes in the vulnerability of specific population subgroups over time, the prevalence of household food insecurity for 2011–2012 and 2017–2018 was stratified by each socio-demographic characteristic, and the prevalence of food insecurity was determined among subgroups by each socio-demographic characteristic for each survey cycle. These estimates incorporated household weights. Overlapping 95% confidence intervals produced using bootstrap techniques were used to determine statistical significance.
All descriptive analyses were conducted using household weights and bootstrap weights (n = 500 CCHS 2011–2012; n = 1000 CCHS 2017–2018) with STATA svy commands, but our regression models were run unweighted. Design changes in the CCHS between the cycles, including differences in how respondents under 18 years of age were weighted and the number of bootstrap weights (Statistics Canada, 2013a, 2019), precluded application of these weights to analyses of pooled data. Sensitivity analyses were performed on all regression models and descriptive analyses to assess the effect of excluding households with missing data on our results. All analyses were conducted in STATA version 16.1 (StataCorp, 2019).
Results
The distributions of several socio-demographic characteristics differed significantly between the 2011–2012 and 2017–2018 survey populations (Table 1). The proportion of households reporting their main source of income as employment decreased while the proportion reporting main source of income as pension/retirement savings increased. The proportion of unattached adults (living alone or with others) increased and the proportion of couples with children under 18 decreased. Additionally, the proportion of households with incomes below LIM decreased between 2011–2012 and 2017–2018.
Table 1.
Household characteristics by survey cycle for 6800 households in Newfoundland and Labrador, 2011–2012 and 2017–2018
| 2011–2012 survey cycle, % (95% CI) |
2017–2018 survey cycle, % (95% CI) |
|
|---|---|---|
| Main source of household income | ||
| Employment income | 64.3 (62.3, 66.3)e | 59.6 (57.6, 61.7)e |
| Employment insurance/workers’ compensation | 2.5 (1.9, 3.2) | 2.8 (2.1,3.6) |
| Pensions/retirement savingsa | 23.3 (21.5, 25.2)e | 27.6 (25.7, 29.5)e |
| Social assistance | 4.2 (3.1, 5.6) | 3.8 (2.9,5.0) |
| Other incomeb or none/not statedc | 5.8 (4.9, 7.0) | 6.3 (5.3,7.4) |
| Highest level of education in household | ||
| Less than secondary school | 14.0 (12.6,15.6) | 11.8 (10.6,13.1) |
| Secondary school completed | 15.1 (13.6,16.8) | 15.0 (13.5,16.7) |
| Post-secondary completed (less than Bachelors’) | 50.1 (47.9, 52.3) | 49.4 (46.8,52.0) |
| Bachelors’ degree or higher | 20.7 (18.8, 22.9) | 23.8 (21.5,26.3) |
| Household compositiond | ||
| Unattached individual/lone parent with children ≥ 18 | 28.5 (26.9, 30.1)e | 31.5 (30.6, 32.8)e |
| Couple without children < 18/Other household types | 42.8 (40.8, 44.8) | 43.3 (41.6, 45.0) |
| Couple with children < 18 | 22.8 (21.4, 24.2)e | 19.4 (18.1, 20.8)e |
| Lone parent with children < 18 | 6.0 (4.8,7.5) | 5.7 (4.7, 7.0) |
| Housing tenure | ||
| Dwelling owned/not stated | 78.4 (75.6, 81.0) | 77.3 (75.0, 79.4) |
| Dwelling rented | 21.6 (19.0, 24.4) | 22.7 (20.6, 25.0) |
| Respondent’s Indigenous status | ||
| Non-Indigenous/not stated | 90.6 (89.3, 91.7) | 90.3 (88.7, 91.7) |
| Indigenous | 9.4 (8.3, 10.7) | 9.7 (8.3, 11.3) |
| Respondent’s immigration status | ||
| Canadian-born/not stated | 97.8 (97.1, 98.3) | 96.6 (95.5, 97.4) |
| Immigrant | 2.2 (1.7, 2.9) | 3.5 (2.7, 4.5) |
| Before-tax household income | ||
| > Low Income Measure | 63.2 (60.9, 65.4)e | 71.4 (69.1, 73.5)e |
| ≤ Low Income Measure | 36.8 (34.7, 39.1)e | 28.6 (26.5, 30.9)e |
aPensions/retirement savings includes dividends/interest, Canada/Quebec Pension Plan, job-related retirement pensions, RRSP/RRIF, Old Age Security/Guaranteed Income Supplement
bOther income includes child tax benefits, child support, alimony, rental income, and scholarships
cThe proportions of missing observations (i.e., not stated) could not be presented separately due to disclosure risk
dAll categories include households with others (i.e., denoting relationships other than parent, child, or partner) present
eIndicates that 95% confidence intervals did not overlap and therefore the difference between survey cycles is statistically significant
The prevalence of food insecurity rose from 12.0% (95% CI 10.5, 13.6) in 2011–2012 to 14.7% (95% CI 13.1, 16.6) in 2017–2018, but neither the change in overall prevalence nor the prevalence of moderate/severe food insecurity was statistically significant (Fig. 1).
Fig. 1.
Prevalence household food insecurity in 2011–2012 and 2017–2018. Source: Author’s calculations from Canadian Community Health Survey cycles 2011–2012 and 2017–2018
The unadjusted odds of experiencing food insecurity was 46% higher in 2017–2018 compared with 2011–2012 (OR 1.46, 95% CI 1.26, 1.69), and the odds of moderate and severe food insecurity was 60% higher (OR 1.60, 95% CI 1.34, 1.91). After controlling for household socio-demographic characteristics, these differences increased to 49% and 64% respectively (Table 2).
Table 2.
Adjusted odds ratios of overall food insecurity and moderate or severe food insecurity in relation to survey cycle and socio-demographic characteristics (n = 6800)
| Overall food insecurity adjusteda OR (95% CI) |
Moderate or severe food insecurity adjusted OR (95% CI) |
|
|---|---|---|
| Survey cycle | ||
| 2011–2012 | 1 | 1 |
| 2017–2018 | 1.49 (1.27, 1.75)* | 1.64 (1.35, 2.00)* |
| Main source of household income | ||
| Employment | 1 | 1 |
| Employment insurance/workers’ compensation | 2.56 (1.79, 3.65)* | 2.92 (1.97, 4.33)* |
| Pension/retirement savings | 0.57 (0.45, 0.71)* | 0.48 (0.36, 0.65)* |
| Social assistance | 3.33 (2.39, 4.63)* | 3.16 (2.23, 4.49)* |
| Other income or none/unstated | 0.87 (0.64, 1.19) | 0.79 (0.54, 1.16) |
| Highest level of education in household | ||
| Less than secondary | 0.92 (0.73, 1.15) | 0.90 (0.69, 1.18) |
| Secondary school graduate | 0.98 (0.78, 1.22) | 1.06 (0.81,1.37) |
| Completed post-secondary, below Bachelors’ degree/unstated | 1 | 1 |
| Bachelors’ or higher complete | 0.60 (0.46, 0.78)* | 0.42 (0.28, 0.62)* |
| Household composition | ||
| Unattached individual/lone parent with children ≥ 18 | 1.51 (1.24, 1.85)* | 1.78 (1.38, 2.3)* |
| Couple without children < 18/other household types | 1 | 1 |
| Couple with children < 18 | 1.61 (1.27, 2.02)* | 1.81 (1.34, 2.44)* |
| Lone parent with children < 18 | 2.00 (1.48, 2.72)* | 2.09 (1.46, 2.98)* |
| Housing tenure | ||
| Dwelling owned by member of household | 1 | 1 |
| Dwelling rented | 2.33 (1.94, 2.79)* | 2.61 (2.1, 3.25)* |
| Indigenous status of respondent | ||
| Non-Indigenous/unstated | 1 | 1 |
| Indigenous | 1.26 (1.02, 1.57)* | 1.34 (1.03,1.73)* |
| Immigrant status of respondent | ||
| Non-immigrant/unstated | 1 | 1 |
| Immigrant | 1.48 (0.93, 2.38) | 1.75 (0.99, 3.10) |
| Income | ||
| > Low Income Measure | 1 | 1 |
| ≤ Low Income Measure | 2.95 (2.43, 3.57)* | 3.59 (2.82, 4.57)* |
*p < 0.05
aAdjusted for main source of household income, highest level of education in household, household composition, housing tenure, Indigenous status, immigration status, and income
Testing for interactions between survey cycle and socio-demographic characteristics revealed significant interactions for both any food insecurity and moderate/severe food insecurity with household composition and for any food insecurity with housing tenure. (The p-values for all the interaction terms are presented in Appendix 1, Table 4) Sample sizes were too small to run models stratified by household composition. Rerunning our multivariable logistic regression model stratified by housing tenure yielded an adjusted odds of food insecurity of 2.01 (95% CI 1.53, 2.65) for renters and 1.29 (95% CI 1.06, 1.57) for homeowners in 2017–2018 compared with 2011–2012.
Stratifying the overall prevalence of food insecurity by socio-demographic characteristics revealed significant differences by education, household composition, and housing tenure and income, but not main source of income, Indigenous status, or immigration status (Table 3). In both 2011–2012 and 2017–2018, the prevalence of food insecurity was highest among households reliant on social assistance. This prevalence rose from 46.2% in 2011–2012 to 65.0% in 2017–2018, but the change was not statistically significant. When looking at those households who experienced food insecurity, a majority were reliant on employment incomes in both cycles. In both 2011–2012 and 2017–2018, most food-insecure households had incomes at or below LIM, but the rate of food insecurity among those with incomes above LIM rose significantly over time and they contributed more to the overall prevalence of food insecurity in 2017–2018 than in 2011–2012.
Table 3.
Contribution to total food insecurity prevalence and prevalence among subgroups by socio-demographic characteristics among 6800 households in Newfoundland and Labrador, 2011–2012 and 2017–2018
| 2011–2012 survey cycle % (95% CI) | 2017–2018 survey cycle % (95% CI) | |||
|---|---|---|---|---|
| Total prevalence of food insecurity | 12.0 (10.5, 13.6) | 14.7 (13.1, 16.6) | ||
| Breakdown of total prevalenceb | Within-group prevalence | Breakdown of total prevalenceb | Within-group prevalence | |
| Main source of household income | ||||
| Employment income | 6.1 (5.1, 7.3) | 9.5 (8.0, 11.4) | 7.6 (6.3, 9.9) | 12.7 (10.6, 15.1) |
| EI/WC | 0.8 (.5, 1.4) | 34.5 (21.7, 50.1) | 0.9 (.6, 1.3) | 32.4 (22.2, 44.5) |
| Pensions/retirement savings | 1.9 (1.5, 2.5) | 8.2 (6.3, 10.5) | 2.5 (2.0, 3.2) | 9.1 (7.2, 11.5) |
| Social assistance | 1.9 (1.3, 2.8) | 46.2 (34.4, 58.3) | 2.5 (1.8, 3.4) | 65.0 (53.4, 75.1) |
| Other/unstated | 1.2 (.8, 1.8) | 20.2 (13.7, 28.7) | 1.3 (0.9, 1.9) | 20.6 (14.2, 28.9) |
| Highest level of education in household | ||||
| Less than secondary | 2.4 (1.8, 3.2) | 17.1 (13.0, 22.1) | 2.2 (1.7, 2.9) | 18.9 (14.8, 23.8) |
| Secondary school completion | 1.8 (1.3, 2.6)a | 12.2 (9.0, 16.3)a | 3.7 (2.9, 4.7)a | 24.6 (19.6, 30.3)a |
| Post-secondary school completion (less than Bachelors’) | 6.4 (5.3, 7.7) | 12.8 (10.7, 15.2) | 7.2 (6.0, 8.6) | 14.5 (12.2, 17.2) |
| Bachelors’ degree or higher | 1.4 (.9, 2.0) | 6.5 (4.4, 9.6) | 1.6 (1.1, 2.4) | 6.9 (4.8, 9.8) |
| Household composition | ||||
| Unattached/lone parent with children > 18 years | 4.3 (3.5, 5.4)a | 15.1 (12.3, 18.6)a | 7.3 (6.2, 8.7)a | 23.3 (19.6, 27.4)a |
| Couple without children under 18/other household types | 3.8 (3.0, 4.8) | 8.9 (7.1, 11.1) | 2.7 (2.1, 3.6) | 6.3 (4.9, 8.2) |
| Couple with children under 18 | 2.3 (1.7, 3.0) | 9.9 (7.4, 13.0) | 2.6 (1.9, 3.5) | 13.2 (9.7, 17.8) |
| Lone parent with children under 18 | 1.6 (1.1, 2.4) | 27.3 (20.2, 35.7) | 2.1 (1.5, 2.9) | 36.3 (27.0, 46.9) |
| Dwelling ownership status | ||||
| Homeowners | 6.7 (5.6, 7.9) | 8.5 (7.2, 10.0) | 6.4 (5.4, 7.6) | 8.3 (7.1, 9.8) |
| Renter households | 5.3 (4.2, 6.6)a | 24.6 (20.5, 29.3)a | 8.3 (6.9, 9.9)a | 36.5 (31.2, 42.2)a |
| Respondent Indigenous status | ||||
| Non-Indigenous/not stated | 10.6 (9.1, 12.2) | 11.7 (10.1, 13.4) | 12.8 (11.2, 14.7) | 14.2 (12.4, 16.2) |
| Indigenous | 1.4 (1.1, 1.9) | 15.2 (11.5, 19.8) | 1.9 (1.5, 2.5) | 19.6 (15.8, 24.0) |
| Respondent immigration status | ||||
| Canadian-born/not stated | 11.8 (10.3, 13.4) | 12.0 (10.5, 13.7) | 14.0 (12.4, 15.8) | 14.5 (12.8, 16.4) |
| Immigrant | 0.2 (0.1, 0.5) F | 10.5 (4.7, 22.0) F | 0.7 (0.4, 1.3) | 20.6 (11.7, 33.5) |
| Income | ||||
| > Low Income Measure | 3.5 (2.8, 4.3)a | 5.5 (4.4, 6.8)a | 6.6 (5.5, 7.8)a | 9.2 (7.7, 10.9)a |
| ≤ Low Income Measure | 8.5 (7.2, 10.1) | 23.2 (19.7, 27.0) | 8.2 (7.0, 9.6) | 28.5 (24.8, 32.6) |
a Indicates that 95% confidence intervals did not overlap and therefore difference between survey cycles is statistically significant
bThe categories within each variable sum to the total prevalence
F indicates estimate with high uncertainty as per Statistics Canada’s standard (coefficient of variation > 33.3%)
Households in which the highest level of educational attainment was secondary school contributed more to the overall prevalence of food insecurity in 2017–2018 than in 2011–2012, a finding consistent with the increased rate of food insecurity among households with this level of education (i.e., 12.2% in 2011–2012; 24.6% in 2017–2018). Significant increases were also observed in the prevalence of food insecurity among unattached individuals, renters, and households with income above LIM, with concomitant increases in the contribution of these groups to the provincial prevalence of food insecurity in 2017–2018 compared with 2011–2012.
The results from sensitivity analyses conducted by excluding households with missing co-variates were virtually the same as the results from the analyses including these households (results available upon request).
Discussion
This study was undertaken to explore the marked increase in food insecurity prevalence charted in NL between 2011–2012 and 2017–2018. Our comparison of sample characteristics across these two survey cycles revealed several differences in the socio-demographic composition of the population, but adjusting for these variables only heightened the odds of households in NL experiencing any food insecurity or moderate/severe food insecurity in 2017–2018 compared with 2011–2012.
Part of the explanation for the increased odds of food insecurity among NL households in 2017–2018 compared with 2011–2012 may be the increased vulnerability of renter households. They made up a greater proportion of the overall prevalence of food insecurity in 2017–2018 compared with 2011–2012, and their odds of food insecurity in 2017–2018 was double that in 2011–2012, after adjusting for other socio-demographic characteristics. Homeowners also experienced an increase in vulnerability to food insecurity during this period, but it was much smaller. While homeownership typically confers protection against food insecurity, this depends on the value of the home and whether or not there is a mortgage (Fafard St-Germain & Tarasuk, 2020). However, CCHS does not include data on housing costs or mortgage levels among homeowners. A comparison of data from the 2011 National Household Survey and the 2016 Census shows that average monthly shelter costs rose from $682 to $836 for renters and from $779 to $984 for owners in the province, potentially heightening risk for food insecurity over the study time period (Statistics Canada, 2017; Statistics Canada, 2013b). The increases in shelter costs may also have contributed to the increased rate of food insecurity observed among households with incomes above LIM in 2017–2018.
In response to concerns about housing affordability, in 2011 and 2012, the provincial government increased investments in its Rent Supplement Program which subsidizes units for eligible low-income households (Government of Newfoundland & Labrador, 2011). Additional investments in this program were included in the 2017–2018 provincial budget (Government of Newfoundland and Labrador, n.d.). Our findings not only highlight the importance of interventions to support renter households but also suggest a need for evaluation of the Rent Supplement Program in relation to household food insecurity.
Vulnerability to food insecurity also appears to have changed over time in relation to household composition. Unattached individuals and lone parents with children over 18 experienced an increase in prevalence of food insecurity between 2011–2012 and 2017–2018 and comprised half of all food-insecure households in the province in 2017–2018. The apparent lack of change in vulnerability among households with children under 18 may in part reflect the mitigating effect of the Canada Child Benefit (CCB), introduced in 2016. This benefit has been associated with a decrease in severe food insecurity among low-income households with children in provinces that collected data on food insecurity immediately before and after implementation of the CCB (Brown & Tarasuk, 2019).
Increases in unemployment have been associated with increases in food insecurity prevalence provincially (Men et al., 2021a) and in census metropolitan areas in Canada (Sriram & Tarasuk, 2015), but it is not possible from our analyses to relate the observed change in food insecurity to the rise in unemployment in NL over this same period. Over half of food-insecure households were reliant on employment incomes in 2017–2018 and we found no significant changes in vulnerability in relation to households’ main source of income. Nonetheless, losses in employment income potentially related to the increased unemployment in the province may have contributed to households’ financial hardships.
The decreased vulnerability of social assistance recipients emerged as a key contributor to the drop in food insecurity prevalence in NL between 2007 and 2011 (Loopstra et al., 2015), but households reliant on social assistance comprised a relatively small fraction of food insecure households in both 2011–2012 and 2017–2018. Although not statistically significant, the prevalence of food insecurity among social assistance recipients was considerably higher in 2017–2018 than in 2011–2012, suggesting that people reliant on this income support may be facing greater vulnerability, potentially related to their benefits no longer being indexed to inflation (Government of Newfoundland and Labrador, n.d.).
The rate of food insecurity among households with before-tax incomes above LIM increased between 2011–2012 and 2017–2018, but LIM status did not have a significantly different impact on odds for food insecurity between survey cycles. However, changes in the collection of data on the amounts and sources of household income between the cycles precluded a fuller investigation of the relationship between income and food insecurity in this study. Prior to 2016, information on household income was based entirely on self-report while after 2016 it was linked to tax records (Statistics Canada, 2013a, 2019). This change may have introduced differential misclassification error with respect to LIM.
Our comparison of data from CCHS 2011–2012 and 2017–2018 is limited by the fact that this survey underwent a redesign in 2015, entailing changes to the collection strategy, sampling frame, and weighting procedures (Statistics Canada, 2016). No analysis of the effects of this redesign on comparisons of CCHS data collected before and after it has been published. The changes may have contributed to the observed differences in the socio-demographic profile of survey respondents between survey cycles. While the decrease in the proportion of households reliant on employment income and increase in the proportion reporting pensions/retirement savings as their main source of income are consistent with the rise in unemployment over this period and the aging population in the province, we cannot rule out design effects. Additionally, changes in the collection of data on household income between survey cycles may have contributed to the relative increase in households with incomes above LIM in 2017–2018.
Recognizing that we could not differentiate true changes in the population composition from those related to changes in the sampling design and data collection method for income, we controlled for differences in the population composition between survey cycles by adjusting for socio-demographic characteristics in our multivariable models. We still found a highly significant increase in the odds of food insecurity in NL in 2017–2018. However, we cannot rule out the possibility that our stratification and comparison of the provincial food insecurity prevalence by socio-demographic characteristics across cycles may have been influenced by the survey redesign. Our results may also reflect the effect of additional (unobserved) factors that were not controlled for in the regression.
Our use of cross-sectional data and the descriptive nature of our analyses preclude any inference about causal relationships. We are further limited by the province’s decision to not measure food insecurity in the 2013–2014 and 2015–2016 cycles of CCHS. Additionally, we were limited in our analyses of population subgroups, including those defined by household composition, because of small cell sizes. This also precluded any examination of changes in the vulnerability of rural and remote communities and racial/cultural groups other than Indigenous respondents. Given evidence of substantial Black-white racial disparities in food insecurity in Canada (Dhunna & Tarasuk, 2021), this is an important limitation to address in future research. Finally, the small sample size precluded analyses of changes in the severity of household food insecurity over time. The higher adjusted odds of moderate/severe food insecurity compared with any food insecurity suggests that this problem may be increasing in both magnitude and severity, but more data are required to confirm this.
Conclusion
The increased odds of food insecurity among NL households between 2011–2012 and 2017–2018 indicates a sharp reversal of the trend identified from 2007 to 2011 in that province. This change has occurred over a period when the province elected not to monitor food insecurity. More work is needed to identify the reasons for the rise in food insecurity, but our study highlights the increased vulnerability of renters and unattached individuals. Given the serious negative health implications of household food insecurity, the province should draw on the lessons from its earlier success in reducing the prevalence and severity of household food insecurity through poverty reduction measures and move swiftly now to reverse the current trend.
Contributions to knowledge
What does this study add to existing knowledge?
The province of Newfoundland and Labrador had been Canada’s success story in food insecurity reduction, achieving an unprecedented 5 percentage point drop in prevalence following the 2006 introduction of the province’s poverty reduction strategy. This tremendous gain now appears to have been reversed, with the odds of food insecurity among households in Newfoundland and Labrador rising significantly between 2011–2012 and 2017–2018. The greatest increases in vulnerability were observed among renters and unattached individuals and lone parents with children over 18.
What are the key implications for public health interventions, practice or policy?
Given the serious negative health implications of household food insecurity and the increased health care costs associated with this condition, the province needs to draw on the lessons from its earlier success in food insecurity reduction to implement policy interventions that reverse the current trend.
Our findings highlight the importance of interventions to better support renter households and halt the escalating vulnerability of unattached individuals and lone parents with older children in Newfoundland and Labrador.
Acknowledgements
The research was conducted at RDC Toronto, a part of the Canadian Research Data Centre Network (CRDCN). This service is provided through the support of the University of Toronto, the Canadian Foundation for Innovation, the Canadian Institutes of Health Research, the Social Science and Humanity Research Council, and Statistics Canada.
Appendix 1
Table 4.
p-values for all interactions tested in fully adjusted logistic regression models
| Interaction terms | p-values for overall food insecurity | p-values for moderate or severe food insecurity |
|---|---|---|
| Main source of household income | ||
| Survey cycle × employment insurance/workers’ compensation | 0.519 | 0.145 |
| Survey cycle × pensions/retirement savings | 0.143 | 0.127 |
| Survey cycle × social assistance | 0.870 | 0.162 |
| Survey cycle × other/unstated | 0.424 | 0.627 |
| Highest level of education in household | ||
| Survey cycle × less than secondary | 0.056 | 0.138 |
| Survey cycle × secondary complete | 0.115 | 0.724 |
| Survey cycle × Bachelors’ degree or higher complete | 0.072 | 0.508 |
| Household composition | ||
| Survey cycle × unattached/lone parents with children > 18 years | 0.037* | 0.266 |
| Survey cycle × couple with children under 18 | 0.048* | 0.065 |
| Survey cycle × lone parents with children under 18 | 0.001* | 0.004* |
| Housing tenure | ||
| Survey cycle × dwelling rented | 0.007* | 0.391 |
| Respondent Indigenous status | ||
| Survey cycle × Indigenous | 0.48 | 0.555 |
| Respondent immigration status | ||
| Survey cycle × immigrant | 0.195 | 0.233 |
| Income | ||
| Survey cycle × ≤ Low Income Measure | 0.62 | 0.136 |
*p < 0.05
Author contributions
Both authors contributed to the study conception and design. The data analyses were conducted by Zahra Hussain, and she wrote the first draft of the manuscript. Both authors contributed to subsequent versions of the manuscript, and both authors read and approved the final manuscript.
Funding
This study was supported by Canadian Institutes of Health Research grant PJT 153260.
Data availability
Data may be obtained from a third party and are not publicly available. The data that support the findings of this study are accessible through Statistics Canada with restrictions. For the present study, the data were accessed under contract through the Statistics Canada Research Data Centre at the University of Toronto.
Code availability
The Stata code used to analyze these data can be made available upon request.
Declarations
Ethics approval
Ethics approval for the present study came from the Health Sciences Research Ethics Board, University of Toronto.
Consent to participate
NA
Consent for publication
NA
Conflict of interest
The authors declare no competing interests.
Footnotes
Publisher's note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- Brown EM, Tarasuk V. Money speaks: Reductions in severe food insecurity follow the Canada Child Benefit. Preventive Medicine. 2019;129:105876. doi: 10.1016/j.ypmed.2019.105876. [DOI] [PubMed] [Google Scholar]
- Dhunna, S., & Tarasuk, V. (2021). Black-white racial disparities in household food insecurity from 2005–2014, Canada Canadian Journal of Public Health:1–15. 10.17269/s41997-021-00539-y [DOI] [PMC free article] [PubMed]
- Fafard St-Germain AA, Tarasuk V. Homeownership status and risk of food insecurity: Examining the role of housing debt, housing expenditure and housing asset using a cross-sectional population-based survey of Canadian households. International Journal for Equity in Health. 2020;19(1):5. doi: 10.1186/s12939-019-1114-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Government of Newfoundland and Labrador. (2011). A clear path to opportunity, self-reliance and a prosperous future – budget 2011 stands strong against poverty. Retrieved December 9, 2020 (https://www.releases.gov.nl.ca/releases/2011/hrle/0419n12.html)
- Government of Newfoundland and Labrador. (2014a). Enhancements to tax programs to benefit seniors, low income and small business. Retrieved December 9, 2020 (https://www.releases.gov.nl.ca/releases/2014/fin/0327n04.htm)
- Government of Newfoundland and Labrador. (2014b). Poverty reduction strategy total investments surpass $1 billion. Retrieved December 9, 2020 (https://www.releases.gov.nl.ca/releases/2014/aes/0327n22.htm)
- Government of Newfoundland and Labrador. (2016). Budget 2016 protecting our vulnerable residents. Retrieved December 9, 2020 (https://www.budget.gov.nl.ca/budget2016/releases/budget_release2.pdf)
- Government of Newfoundland and Labrador. (n.d.). Poverty reduction initiatives - budget 2017/18. Retrieved December 9, 2020 (https://www.gov.nl.ca/cssd/files/publications-pdf-prs-prs-initiative-budget2017-18.pdf)
- Health Canada . Canadian Community Health Survey, Cycle 2.2, Nutrition (2004) - Income-Related Household Food Security in Canada. Government of Canada. Ottawa: Office of Nutrition Policy and Promotion, Health Products and Food Branch. Health Canada; 2007. [Google Scholar]
- Health Canada. (2020). Determining food security status. Government of Canada. Retrieved February 2, 2021 (https://www.canada.ca/en/health-canada/services/food-nutrition/food-nutrition-surveillance/health-nutrition-surveys/canadian-community-health-survey-cchs/household-food-insecurity-canada-overview/determining-food-security-status-food-nutrition-surveillance-health-canada.html)
- Income Statistics Division, Statistics Canada. (2016). Low income lines: What they are and how they are created. Retrieved January 10, 2021 (https://www150.statcan.gc.ca/n1/en/pub/75f0002m/75f0002m2016002-eng.pdf?st=s-2l9aEw)
- Jessiman-Perreault G, McIntyre L. The household food insecurity gradient and potential reductions in adverse population mental health outcomes in Canadian adults. SSM - Population Health. 2017;3:464–472. doi: 10.1016/j.ssmph.2017.05.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kirkpatrick SI, McIntyre L, Potestio ML. Child hunger and long-term adverse consequences for health. Archives of Pediatrics & Adolescent Medicine. 2010;164(8):754–762. doi: 10.1001/archpediatrics.2010.117. [DOI] [PubMed] [Google Scholar]
- Laidley J, Aldridge H. Welfare in Canada. Toronto, ON: Maytree; 2020. [Google Scholar]
- Loopstra R, Dachner N, Tarasuk V. An exploration of the unprecedented decline in the prevalence of household food insecurity in Newfoundland and Labrador, 2007–2012. Canadian Public Policy. 2015;41(3):191–206. doi: 10.3138/cpp.2014-080. [DOI] [Google Scholar]
- McIntyre L, Wu X, Kwok C, Patten SB. The pervasive effect of youth self-report of hunger on depression over 6 years of follow up. Social Psychiatry and Psychiatric Epidemiology. 2017;52(5):537–547. doi: 10.1007/s00127-017-1361-5. [DOI] [PubMed] [Google Scholar]
- Men F, Elgar FJ, Tarasuk V. Food insecurity is associated with mental health problems among Canadian youth. Journal of Epidemiology and Community Health. 2021 doi: 10.1136/jech-2020-216149. [DOI] [PubMed] [Google Scholar]
- Men, F., Urquia, M. L., Tarasuk, V. (2021b). The role of provincial social policies and economic environment in shaping household food insecurity among families with children in Canada. Preventive Medicine 148, 106558. 10.1016/j.ypmed.2021.106558 [DOI] [PubMed]
- Men F, Gundersen C, Urquia ML, Tarasuk V. Association between household food insecurity and mortality in Canada: A population-based retrospective cohort study. CMAJ. 2020;192(3):E53–60. doi: 10.1503/cmaj.190385. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Men F, Gundersen C, Urquia ML, Tarasuk V. Food insecurity is associated with higher health care use and costs among Canadian adults. Health Affairs. 2020;39(8):1377–1385. doi: 10.1377/hlthaff.2019.01637. [DOI] [PubMed] [Google Scholar]
- Nord M, Coleman-Jensen A, Gregory C. Prevalence of U.S. food insecurity is related to changes in unemployment, inflation, and the price of food. 167. Washington, DC: Economic Research Service, United States Department of Agriculture; 2014. [Google Scholar]
- Sriram U, Tarasuk V. Changes in household food insecurity rates in Canadian metropolitan areas from 2007 to 2012. Canadian Journal of Public Health Revue Canadienne de Santé Publique. 2015;106(5):e322–27. doi: 10.17269/cjph.106.4893. [DOI] [PMC free article] [PubMed] [Google Scholar]
- StataCorp. (2019). Stata Statistical Software: Release 16. StataCorp LLC.
- Statistics Canada. (2013a). Canadian Community Health Survey (CCHS) Annual Component: User Guide 2012 and 2011–12 Microdata File.
- Statistics Canada. (2013b). Newfoundland and Labrador (Code 10) (Table). National Household Survey (NHS) Profile. 2011 National Household Survey. Statistics Canada Catalogue No. 99–004-XWE. Retrieved January 9, 2021 (https://www12.statcan.gc.ca/nhs-enm/2011/dp-pd/prof/details/Page.cfm?Lang=E&Geo1=PR&Code1=10&Data=Count&SearchText=Newfoundland%20and%20Labrador&SearchType=Begins&SearchPR=01&A1=All&B1=All&GeoLevel=PR&GeoCode=10)
- Statistics Canada. (2015). Appendix A Glossary. Retrieved July 19, 2021 https://www150.statcan.gc.ca/n1/pub/75f0002m/2010004/app-ann-eng.htm)
- Statistics Canada. (2016). Canadian Community Health Survey (CCHS): Summary of Changes. Retrieved April 5, 2021 (https://www23.statcan.gc.ca/imdb/p2SV.pl?Function=getMainChange&Id=4995)
- Statistics Canada. (2017). 2016 Census of Population, Statistics Canada Catalogue No. 98–400-X2016228. Retrieved April 5, 2021. https://www12.statcan.gc.ca/census-recensement/2016/dp-pd/dt-td/Rp-eng.cfm?LANG=E&APATH=3&DETAIL=0&DIM=0&FL=A&FREE=0&GC=0&GID=0&GK=0&GRP=1&PID=110571&PRID=10&PTYPE=109445&S=0&SHOWALL=0&SUB=0&Temporal=2017&THEME=1
- Statistics Canada. (2019). Canadian Community Health Survey (CCHS) Annual Component: User Guide 2018 and 2017–2018 Microdata File.
- Statistics Canada. (2021). Table 14–10–0090–01 Labour force characteristics by province, territory and economic region, annual. 10.25318/1410009001-eng. Accessed 9 Jan 2021.
- Tarasuk V, Fafard St-Germain AA, Mitchell A. Geographic and socio-demographic predictors of household food insecurity in Canada, 2011–12. BMC Public Health. 2019;19(1):12. doi: 10.1186/s12889-018-6344-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Tarasuk, V., & Mitchell, A. (2020). Household food insecurity in Canada, 2017–18. In Research to identify policy options to reduce food insecurity (PROOF). Toronto, ON. Retrieved October 5, 2021 (https://proof.utoronto.ca/wp-content/uploads/2020/03/Household-Food-Insecurity-in-Canada-2017-2018-Full-Reportpdf.pdf)
- Tarasuk V, Mitchell A, McLaren L, McIntyre L. Chronic physical and mental health conditions among adults may increase vulnerability to household food insecurity. The Journal of Nutrition. 2013;143(11):1785–1793. doi: 10.3945/jn.113.178483. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
Data may be obtained from a third party and are not publicly available. The data that support the findings of this study are accessible through Statistics Canada with restrictions. For the present study, the data were accessed under contract through the Statistics Canada Research Data Centre at the University of Toronto.
The Stata code used to analyze these data can be made available upon request.