Household food insecurity (HFI) is an important indicator of material deprivation and a serious chronic public health issue that affects, from the 2021 Canadian Income Survey, 18·4 % of Canadian households(1). Household food security is monitored in Canada using the Household Food Security Survey module(2), whose questions are premised on a household’s financial ability to access adequate food. As such, food insecurity can be defined as the inadequate or insecure access to food due to financial constraints(Reference Loopstra3).
HFI has substantial adverse impacts on individuals’ health and the related healthcare costs in Canada(Reference Clemens, Le and Anderson4,Reference Anderson, Clemens and Le5) . People living in food insecure households have poorer self-rated mental, physical and oral health, greater stress and are more likely to suffer from chronic conditions such as diabetes, hypertension and mood or anxiety disorders(Reference Tarasuk, Mitchell and McLaren6–Reference Shankar, Chung and Frank8).
The persistently high prevalence and negative health implications of HFI have raised the spectre of the role of social protection programmes such as social assistance benefits, employment insurance benefits, universal childcare benefits and housing subsidies in mitigating households’ economic circumstances leading to HFI. Although tightly linked to income, HFI also reflects a household’s broader material circumstances including owning assets such as property, income stability and debt(Reference Tarasuk and Mitchell9). The measurement of HFI during the COVID-19 pandemic was hampered by the interruption of survey data collection, but as more comparable data have emerged, food insecurity rates in high-income countries appear to have remained relatively stable through the pandemic(1,Reference Pepetone, Frongillo and Dodd10) , in part perhaps because COVID monetary benefits mitigated the pandemic’s major income shock(Reference Reimold, Grummon and Taillie11). In Canada, the USA and Australia, pandemic recovery has been associated with increased food insecurity, possibly because inflation and food prices increases have pushed more economically vulnerable households into a food insecure state(1,Reference Kent, Murray and Penrose12,Reference Rabbitt, Hales and Burke13) . Thus, there is a need to identify interventions that might mitigate households’ economic vulnerability to food insecurity. This systematic review (SR) aims to synthesise the evidence on the impact of public policy interventions aimed at improving household financial circumstances on HFI in Canada. Public policy interventions refer to state-level sponsored programmes or activities at any level of government. Food-based interventions that seek to directly respond to households’ food needs were specifically excluded.
Methods
This SR was guided by the Cochrane Handbook for Systematic Reviews(Reference Higgins, Thomas and Chandler14) and follows the PRISMA reporting guidelines(Reference Page, McKenzie and Bossuyt15). The original research question was: ‘What interventions are effective in reducing household food insecurity in Canada?’ The protocol was created a priori and registered in Prospero (CRD42021254450).
During the SR process and in discussions with experts in the field of food insecurity (VT, LM), it became clear that the interventions should be grouped into two categories: public policy interventions (e.g. income support, housing assistance programmes) and food-based interventions (e.g. food banks, gardening programmes). These two types of interventions work at different levels. Food-based interventions endeavour to address food shortages at the household level directly, whereas public policy interventions target the underlying economic vulnerability of households to a range of basic needs including food but also housing and employment supports.
The analysis and reporting were conducted separately for the two types of interventions, resulting in two SR. This SR attempts to answer the question ‘What public policy interventions are effective in reducing household food insecurity in Canada’. The results of food-based interventions will be reported in a separate SR (Idzerda et al, Unpublished, 2024). The eligibility criteria, search strategy and study selection described below detail the SR process for the original research question, whereas the data extraction, risk of bias and Grading of Recommendations, Assessment, Development and Evaluation (GRADE) are specific to the SR on public policy interventions.
Eligibility criteria
Primary studies in English or French that assessed an intervention affecting households in Canada, had a comparison group and measured an outcome of HFI were included. The full lists of inclusion and exclusion criteria are included in Supplementary Material A.
Search strategy
The search strategy was developed by a Health Canada research librarian in collaboration with the authors (Supplementary Material B). It underwent a Peer Review of Electronic Search Strategies and was reviewed for quality by a second independent librarian(Reference McGowan, Sampson and Salzwedel16). The original search was implemented in April 2021, updated in November 2022 with a final update on 5 October 2023. Four electronic bibliographic databases were searched: Scopus, OVID Medline, Embase and EconLit. A complementary grey literature search was conducted in June 2021. Finally, the reference lists of seventeen related reviews were searched and experts were consulted to ensure that all eligible articles were included.
Study selection
To verify potential eligibility, all titles and abstracts were screened by two independent reviewers using a standardised form developed a priori, which was piloted by all reviewers (LI, TC, CL, AC, EV, SK) in the software DistillerSR Version 2.37(17). Next, two reviewers independently screened the full text of each potentially eligible article. The reasons for excluding a study were recorded in both stages of screening. A list of excluded studies is available in Supplementary Material C.
All conflicts were resolved by consensus or a third reviewer. This was also done for data extraction, risk of bias and GRADE.
Data extraction
A data extraction form was developed a priori and was piloted by all reviewers. Two reviewers independently extracted data for each included study. Study information (objectives, study design, time of data collection, description of intervention and the method or tool to measure food insecurity) and participant characteristics (including any subgroups of interest) were extracted for all studies. The outcome of interest was change in level of food insecurity (food secure, marginal, moderate or severe) over time.
Data analysis
Where studies presented results using the same dataset, the study with the longest follow-up period was selected. All data points were utilised if there was no overlap in the data (i.e. all years reported in population level surveys). For studies with monetary interventions, dollar values reported were standardised to 2023 using the Bank of Canada Inflation Calculator (September 2023 Consumer Price Index)(18).
Data were synthesised narratively, as heterogeneity in the interventions meant that the data were not appropriate for pooling.
Risk of bias
The risk of bias was determined using the Effective Public Health Practice Project Quality Assessment Tool (EPHPP)(Reference Jackson and Waters19). This tool was deemed the most appropriate and encompassing due to the wide variety of study types. Two reviewers independently rated the risk of bias for each study.
Certainty of the evidence
The GRADE framework was used to rate the certainty and strength of the body of evidence(Reference Schünemann, Brożek and Guyatt20). Each outcome was assessed independently by two reviewers. The GRADE decision rules as they were applied to this study are presented in Supplementary Materials D. Randomised controlled trials and large population-based studies were started at high certainty of evidence, whereas observational studies were started at low certainty of evidence.
Results
Three categories of public policy interventions were found and assessed: income supplementation, housing assistance programmes and food retailer subsidies.
Descriptive summary of included studies
Of the 7,432 references screened for eligibility, seventeen reported on public policy interventions to reduce HFI (Fig. 1). Study characteristics are presented in Table 1. Sixteen (94 %) of the articles were published in the last 10 years (since 2013).
Fig. 1 PRISMA flow diagram of articles through the systematic review process
Table 1 Study characteristics of included studies
CCB, Canada Child Benefit; OAS, Old Age Security; GIS, Guaranteed Income Supplement; UCCB, Universal Child Care Benefit; OCB, Ontario Child Benefit; EI, Employment Insurance; CCHS, Canadian Community Health Survey; HFSSM, Household Food Security Survey Module; CIS, Canadian Income Survey; EI, Employment Insurance.
* Domain level risk of bias results for all studies can be found in Supplementary Material E.
† Text only indicates an up to 11.7 % increase in welfare income. Difference from reference year calculated from Tweddle, A., Battle, K., Torjman, S.: Welfare in Canada 2012. Values presented in standardised to 2012.
Risk of bias assessments
Overall, income supplementation studies were at low risk of bias, while housing and food retail subsidy intervention studies were at moderate to high risk of bias (Table 1). A summary of the detailed risk of bias results for all studies can be found in Supplementary Material E.
Summary of findings
Income supplementation
Ten studies on the impact of income supplementation on HFI were identified(Reference Brown and Tarasuk21–Reference Men and Tarasuk30). This included direct payments from a government body to an individual or household, such as child care benefits, guaranteed income supplementation for seniors, employment insurance and social assistance. Three of these reported on the same data for the sample population(Reference McIntyre, Dutton and Kwok23–Reference Emery, Fleisch and McIntyre25). Only the data from one of the three studies were used(Reference McIntyre, Dutton and Kwok23) as this study had the longest follow-up and encompassed the data from both other papers(Reference Emery, Fleisch and McIntyre24,Reference Emery, Fleisch and McIntyre25) .
Among low-income populations, three studies demonstrated that income supplementation interventions had a positive effect on reducing moderate and severe HFI with a high level of certainty (Table 2). The odds of HFI were lowest in the intervention with the highest dollar value ($19 890/year) of income supplementation (OR 0·30, 95 % CI 0·27–0·33)(Reference McIntyre, Dutton and Kwok23) and increased (OR 0·85, 95 % CI 0·75, 0·96) as the dollar value ($8,368/year) decreased(Reference Brown and Tarasuk21). A similar but less pronounced trend was observed in five studies reporting on low-income households experiencing marginal, moderate or severe food insecurity. The lowest dollar value ($824) had no impact on reducing food insecurity(Reference Men, Fafard St-Germain and Ross29), however as the dollar amount increased so did the associated adjusted OR, becoming significant at higher levels of income supplementation. In addition, three studies also assessed the change over time against a matched control group (e.g. difference in difference analysis). In all three studies, the control group saw no change over time, while the intervention group had reductions in HFI over time(Reference Brown and Tarasuk21,Reference Li, Dachner and Tarasuk28,Reference Men and Tarasuk30) .
Table 2. Summary of findings Table for income supplementation interventions
* All dollar values are standardised to 2023.
† Values not reported by exposure group.
‡ Tarasuk et al. (2019) provides prevalence and odds ratios of food insecurity in relation to survey cycles. Only total number of participants was extracted as food insecurity prevalence changes per survey cycle.
List of abbreviations: OR: Odds Ratio, CI: Confidence Interval
Grade reasoning:
1 Inconsistency: differences in effect estimate among studies.
2 Indirectness: study population not representative of the whole population.
3 Imprecision: OIS value not met, or no effect/not significant effect with large confidence intervals.
Among the general population, there was a mostly positive effect of exposure to federal or provincial child benefit programmes on HFI as demonstrated across four studies with moderate certainty (Table 2). Assessing the impact of benefits on moderate and severe HFI among the general population in three studies revealed the same relationship, except for one study(Reference Men, Fafard St-Germain and Ross29). The reason for this may be associated with the low dollar value associated with the intervention.
Housing assistance programmes
The effect of housing assistance programmes on HFI was assessed in seven studies, Table 2 (Reference Li, Dachner and Tarasuk28,Reference Pankratz, Nelson and Morrison31–Reference Kirkpatrick and Tarasuk36) . Housing assistance programmes provide cash benefits designated for rental or other housing costs in approved commercial or public housing settings. This includes housing for precariously housed individuals, subsidised housing (reduced cost of public housing) and rental assistance programmes (money given to low-income households to use towards rental costs). Three studies assessed the impact of housing assistance programmes on HFI in homeless or precariously housed individuals and found no effect except for one subgroup where a larger proportion of those with high mental health needs achieved food security following the intervention v. those who did not receive the housing intervention(Reference Pankratz, Nelson and Morrison31–Reference O’Campo, Hwang and Gozdzik33).
The impact of exposure to subsidised housing programmes on HFI among low-income populations was evaluated in three studies, Table 3 (Reference Li, Dachner and Tarasuk28,Reference Loopstra and Tarasuk35,Reference Kirkpatrick and Tarasuk36) . Two studies found no association between low-income families that received housing subsidies and HFI in the large city of Toronto(Reference Loopstra and Tarasuk35,Reference Kirkpatrick and Tarasuk36) . However, the odds of HFI were lower among families with subsidised rent compared to households with market rent on a waitlist for subsidised housing (OR = 0·51; 95 % CI 0·30–0·86)(Reference Kirkpatrick and Tarasuk36). In the third study, there was no reduction in HFI following the introduction of a rental assistance programme ($550 per month, standardised to 2023 values) in the Province of British Columbia(Reference Li, Dachner and Tarasuk28). All post-intervention follow-up periods looking at HFI levels were at least 6 months in length.
Table 3. Summary of findings Table for housing interventions
Grade reasoning:
1 Study limitations: study had high or moderate risk of bias.
2 Inconsistency: differences in effect estimate among studies.
3 Indirectness: study population not representative of the whole population.
4 Imprecision: OIS value not met, or no effect/not significant effect with large confidence intervals.
Overall, these studies showed, with low to very low certainty, that housing assistance programmes for homeless or precariously housed individuals, and housing subsidies for low-income populations, may have little to no effect on HFI.
A study on the impact of home ownership on HFI in Canada before and after the 2008–2009 recession demonstrated that home ownership likely reduced the risk of HFI during this time(Reference McIntyre, Wu and Kwok34). Specifically, among renters, the risk of HFI increased significantly post-recession (OR = 1·16, 95 % CI 1·05–1·29), whereas homeowners had a non-significant slight increase in HFI over the same period(Reference McIntyre, Wu and Kwok34).
Food retailer subsidy interventions
Food retailer subsidy programmes include direct payments by government bodies to food retailers to reduce the price of foods sold to the public prior to the point of purchase. One study assessed a federal food retailer subsidy programme, Nutrition North Canada, in Nunavut Territory, Table 4 (Reference St-Germain, Galloway and Tarasuk37). After controlling for several covariates, the rate of HFI increased by 13·2 percentage points (95 % CI 1·7–24·7) after implementation of the subsidy programme(Reference St-Germain, Galloway and Tarasuk37). The implementation of Nutrition North Canada may have increased rates of HFI, but the evidence is very uncertain.
Table 4. Summary of findings Table for food retail interventions
Grade reasoning:
1 Study limitations: study had high or moderate risk of bias.
2 Indirectness: study population not representative of the whole population.
Discussion
The objective of this SR was to synthesise the evidence on public policy interventions to mitigate HFI in Canada. Three categories of interventions were found and assessed: income supplementation, housing support and food retailer subsidies.
Income supplementation studies
This SR found that income supplementation (in the range of $824–$19 890 CDN standardised to 2023) for low-income Canadians likely leads to fewer households being food insecure, with the size of effect possibly increasing as the monetary value of the intervention increases. This is aligned with another recent SR conducted in Canada and the USA, which found moderate-certainty evidence of an association between offering monetary assistance and reduced food insecurity (ten studies; pooled random effects; adjusted OR, 0·64; 95 % CI 0·49–0·84)(Reference Oronce, Miake-Lye and Begashaw38). Although the studies were generally well executed, income supplementation has been inferred from an exposure and was never actually observed. In the absence of experimental data, such as data from a basic income experiment, it will be difficult to determine the ‘dose’ of income necessary to mitigate HFI in vulnerable households. Future research on a possible dose–response curve should be undertaken to set the threshold for which income supplementation has a meaningful impact on HFI in Canada.
A limitation of the income supplementation studies is that most utilised the income variable of the Canadian Community Health Survey (CCHS). This was self-reported, before-tax and imputed by Statistics Canada for 30 % of respondents. In some, but not all, cases the imputation was considered in the analysis. It is likely that there is measurement error on this variable resulting in misclassification of low-income individuals. Additionally, there are marked differences between studies in the operational definitions of what has here been referred to as ‘low income’. This heterogeneity matters because the sensitivity of HFI to income interventions is likely greatest among the most resource-constrained households, but it has not been feasible to take baseline incomes into account in this analysis. Comparison of these studies is further limited as the adjusted OR drawn from studies differed in their identification of and adjustments for potentially confounding factors and observed the effects of the increments in income over different periods of time. Whether the initial effects of increases in benefits on HFI are sustained over time depends on several factors including changes in macroeconomic conditions and policy context (e.g. whether new benefits are indexed to inflation or how their introduction affects other relevant programmes and policies). Since the CCHS and Canadian Income Survey exclude people living in remote locations and on First Nation reserves, as well as those within institutions, the results of this SR cannot be applied to those populations.
An additional limitation has been the inability to determine whether the effects observed here represent reductions in the likelihood or severity of HFI among already-affected households, or the prevention of HFI (or more severe food insecurity). The studies reviewed all used cross-sectional survey data that included measures of HFI over the prior 12 months. Within-household changes were not observed over time; rather, inferences about the effectiveness of specific interventions were drawn from comparisons of HFI status among comparable groups before and after the introduction of policy changes. Although some studies employed robust cross-sectional designs that utilise econometric methods, which capitalise on natural variations in policies to estimate an intent-to-treat(Reference Brown and Tarasuk21,Reference Ionescu-Ittu, Kaufman and Glymour22,Reference St-Germain, Galloway and Tarasuk37) , longitudinal studies may be useful to distinguish interventions that prevent food insecurity from those that reduce its prevalence or severity among already-affected households.
Housing assistance studies
This SR showed that housing assistance programmes for homeless and precariously housed populations as well as housing subsidies for low-income populations may have little to no effect on HFI. The main limitation is the small number and limited scope of the included studies. More high-quality experimental studies among different population groups (low-income, precariously housed and homeless) across the full spectrum of existing policy interventions that potentially impact households’ housing circumstances (e.g. rent supplements, rent controls and rent-geared-to-income housing) are required before one can determine whether there is an impact of these policies on HFI. In studies of effectiveness, it is also important to consider jurisdictional differences in housing policies. Home ownership, compared to renting, seemed to be protective against HFI in one study. Therefore, high quality experimental studies focused on affordable home ownership should also be explored, particularly since both provincial and federal governments in Canada incentivise home ownership(39).
Food retailer subsidy study
Assessment of a single study on exposure to the Nutrition North Canada food retailer subsidy revealed that HFI increased following the introduction of the programme among remote northern populations. Nunavut, the territory studied, has long been characterised by much higher rates of food insecurity than the rest of Canada(Reference St-Germain, Galloway and Tarasuk37). Whether the observed increase was directly related to the introduction of the food retailer subsidy programme cannot be established, and there has been no research to determine whether food insecurity similarly increased in other areas covered by the programme. Nonetheless, the persistently high rates of food insecurity in northern Canada have brought this programme under review by deferral authorities(40).
There is also a clear need for effective tailored interventions to mitigate food insecurity among Indigenous communities and in northern Canada. As Nutrition North Canada’s implementation possibly led to worsening food insecurity in at least one target area, a place to start is to rethink what changed with the introduction of this programme, in consultation with Indigenous communities, recognising that income and Indigenous food sovereignty are important considerations(Reference Leblanc-Laurendeau41).
This SR found studies concerning three categories of public policy interventions. It is important to note that no literature was found for some categories of public policy interventions, such as studies evaluating the effects of other market subsidy interventions (e.g. programmes that subsidise out-of-pocket costs for essentials such as utilities, prescription drugs and dental care). Further original studies on the other types of public policy interventions should be conducted.
Although Canada monitors HFI annually and food insecurity rates are a component of Canada’s Official Poverty Dashboard of Indicators(42), governments have yet to set a public policy goal of HFI reduction with a target rate. The lack of a specific public policy objective related to HFI may be impeding deliberate public policy work to reduce rates through the interventions reviewed here or other means.
Conclusion
This SR examined the existing body of research on public policy interventions to reduce HFI and placed moderate to high certainty on the evidence showing that income supplementation reduces HFI. Many questions remain in terms of how to optimise this intervention, such as the amount, frequency and delivery mechanism of the income supplementation. In addition, no studies have been designed to clearly differentiate interventions that mitigate households’ experiences of food insecurity from those that prevent HFI in the first place.
Acknowledgements
The authors wish to thank Bernard Choi, Janet Potvin and Genevieve Gariépy for their help with the screening process as well as Kate Morissette for her input into the GRADE process. We would like to acknowledge the Health Canada Library for their assistance and support in designing the search strategy.
Financial support
This research received no specific grant from any funding agency, commercial or not-for-profit sectors.
Conflict of interest
There are no conflicts of interest.
Authorship
L.I., first author, responsible for major conception of the project, leading project at all stages; T.C., second author, contributed in the planning and execution of the project in all stages, conducting analysis; C.L., A.C., E.V. and S.K., contributing authors, contributed during screening, data extraction, evidence analysis, GRADE, writing and review process; L.M. and V.T., contributing authors, acted as subject matter experts on the topic, assisted with scope and development of project, analysis of results, writing of discussion section; A.J.G., last author, responsible for major development and guidance on the project, contributed to all parts of the project.
Ethics of human subject participation
Not applicable.
Supplementary material
For supplementary material accompanying this paper visit https://doi.org/10.1017/S1368980024000120
Open access
