Who experiences food security of differing severity?

Our first objective was to capitalise on the availability of contemporary, nationally representative data on food insecurity in EWNI to examine the demographic and geographical characteristics of people experiencing food insecurity of differing severity in EWNI. Our multivariate analyses revealed that food insecurity is concentrated in economically disadvantaged groups. Unsurprisingly, higher levels of food insecurity were more prevalent in lower-income households^{(Reference Loopstra, Reeves and Tarasuk2,Reference Bramley, Treanor and Sosenko11,Reference Pool and Dooris16)} , and higher incomes were particularly protective against severe food insecurity. In line with previous research, food insecurity was more prevalent among unemployed respondents^{(Reference Loopstra, Reeves and Tarasuk2,Reference Bramley, Treanor and Sosenko11)} , who are likely to be exposed to an often unreliable benefit system^{(Reference Power, Uphoff and Stewart-Knox28)}. Replicating existing research, food insecurity was more prevalent in households containing children^{(Reference Loopstra, Reeves and Tarasuk2,Reference Yau, White and Hammond6)} , especially lone-parent households^{(Reference Loopstra, Reeves and Tarasuk2,Reference Yau, White and Hammond6,Reference Bramley, Treanor and Sosenko11,Reference Brown, Mills and Albani17)} . As these latter associations held after controlling for banded income, we suggest that non-financial factors may be important; for example, time pressures that can translate into higher food costs may be especially acute in lone-parent households.

Higher levels of food insecurity was more prevalent among younger people^{(Reference Loopstra, Reeves and Tarasuk2,Reference Prayogo, Chater and Chapman14,Reference Armstrong, Reynolds and Martins29)} , where factors including precarious work and reduced benefit entitlements are known to create vulnerabilities^{(Reference McPherson30)}. Mixed-race respondents were more likely to experience food insecurity of differing severity, while Asian respondents had a lower prevalence of very low food security, which could reflect familial and community support^{(Reference Power, Doherty and Small31)}, and demonstrates the value of exploring food insecurity of differing severity. Worryingly, the concentration of food insecurity – particularly its more severe forms – among respondents with LTHC and food hypersensitivities corroborates evidence of vulnerability among those with disabilities^{(Reference Hadfield-Spoor, Avendano and Loopstra32)} and may reflect higher dietary costs for those with hypersensitivities⁽³³⁾. Compared with employed respondents, higher levels of food insecurity were less prevalent among retired people, consistent with Canadian research that identified a drop in food insecurity at pension age^{(Reference McIntyre, Dutton and Kwok34)}. Students were also less likely to experience food insecurity. This finding contrasts with recent evidence^{(Reference Armstrong, Gillespie and King35)} and demonstrates the importance of multivariate analyses that controls for income and other characteristics. Students may be protected by specific support or lower living costs elsewhere.

The lower prevalence of food insecurity in less deprived local areas suggests a protective role for the immediate local environment – even after controlling for income – while the lower prevalence of food insecurity in rural areas contradicts European evidence^{(Reference Garratt36)}. Contributing to mixed evidence on the topic^{(Reference Loopstra, Reeves and Tarasuk2,Reference Yau, White and Hammond6)} , gender was not associated with food insecurity.

Who uses food banks?

Our second objective was to identify who uses food banks in a nationally representative sample of EWNI. Overall, 3·6 % of respondents used a food bank in the previous 12 months. The strongest predictor of food bank use was, unsurprisingly, food security status, with people experiencing food insecurity being over four times more likely to use food banks. This finding replicates existing bivariate evidence⁽³⁷⁾ of severe food insecurity among people using food banks and reinforces evidence that food bank use is driven by need and not – as critics have suggested⁽³⁸⁾ – by supply-side factors or opportunism.

Predictably, food bank use was more prevalent among lower-income groups^{(Reference Loopstra and Lalor12)}. After controlling for income, food bank use was concentrated in those who were unemployed^{(Reference Loopstra and Lalor12)} or unable to work due to ill health^{(Reference MacLeod, Curl and Kearns13)}. Independent of their resources, these groups may have more ready access to referral agencies or be better able to make use of emergency food provision with restricted opening hours^{(Reference Loopstra, Lambie-Mumford and Fledderjohann39)}. As noted above, not working exposes people to an often unreliable benefit system, where food bank use is linked to receipt of state benefits^{(Reference Bramley, Treanor and Sosenko11,Reference Prayogo, Chater and Chapman14)} and benefit sanctions^{(Reference Loopstra, Fledderjohann and Reees40)}. A broad but inconsistent association between younger age and food bank use replicates previous evidence^{(Reference Bramley, Treanor and Sosenko11,Reference MacLeod, Curl and Kearns13,Reference Loh, Knight and Loopstra15)} , while the absence of associations by gender contradicts evidence for higher food bank use in men^{(Reference MacLeod, Curl and Kearns13)}. Our findings are important in light of past evidence linking food bank use with these characteristics using less rigorous analyses that did not mutually adjust for other characteristics. Taken together, our findings provide evidence that food bank use is most closely associated with financial disadvantage and experiences of food insecurity.

Do the same groups experience food insecurity and use food banks?

Noting that the prevalence of food insecurity far outstrips food bank use, our third objective was to determine in a nationally representative sample whether the same groups experience food insecurity and use food banks. Looking first at similarities, food insecurity was the strongest predictor of food bank use. Both food insecurity and food bank use were more prevalent at lower incomes and among unemployed and younger people. After accounting for food insecurity status, we did not find an independent association between food bank use and being a lone parent or living in a rural or deprived area. Our findings suggest that food bank use is predicted primarily by the concentration of food insecurity in these groups but that other unmeasured factors are also likely to play a role. Having an LTHC was associated with a greater prevalence of food insecurity – particularly its more severe forms – however, having an LTHC was not associated with food bank use after employment status was controlled. Based on these findings, we suggest that higher food bank use among people with LTHC is driven by a greater prevalence of food insecurity and lower workforce participation in this group^{(Reference Hadfield-Spoor, Avendano and Loopstra32)}.

We also identified some key differences in the prevalence of food insecurity and food bank use for different groups. Food insecurity was more prevalent in waves 5 and 6, in households containing children, in respondents with food hypersensitivity and in Wales, without an accompanying rise in food bank use. These patterns may be understood through reference to a recently proposed conceptual framework in which the likelihood of food insecurity leading to food bank use is impacted by a combination of people’s thoughts and feelings about food bank use and the operational and landscape features of community support available^{(Reference Lambie-Mumford and Loopstra24)}. For example, food bank provision may be unsuited to people with hypersensitivities^{(Reference Garthwaite, Collins and Bambra41)}, while more broadly, the cost-of-living crisis has left some food banks struggling to source adequate supplies to meet growing demand⁽⁴²⁾, which could explain the rise in food insecurity over time without a parallel rise in food bank use.

Conversely, some groups had a lower prevalence of food insecurity but did not report lower levels of food bank use. Despite rural and less deprived areas having a lower prevalence of food insecurity, they did not have comparatively lower levels of food bank use, which contrasts with past evidence for the highly uneven provision of food banks in rural areas^{(Reference May, Williams and Cloke43)}. Retired people and students were similarly less likely to be food insecure but no less likely to use food banks, which may again reflect good access to referral agencies and availability of food banks among these groups. Finally, some characteristics were associated with food bank use only. Respondents unable to work due to ill health did not experience a higher prevalence of food insecurity but were more likely to use food banks, which may reflect support-seeking behaviours that connect these respondents with referral agencies or greater potential to access food banks during opening hours. Discrepancies between food insecurity and food bank use by different ethnic groups potentially reflect the methodological consequences of aggregating diverse groups during analyses and highlight the importance of detailed research by ethnicity^{(Reference Power44)}.

Geographical variations in food insecurity and food bank use

Our fourth objective was to explore country- and region-level variation in food insecurity and food bank use. Respondents in Wales were more likely to experience food insecurity, but food bank use did not vary by country. As above, this pattern could reflect unmet needs in the form of limited access to referral agencies or food banks or the availability of other forms of community food provision that people may access instead of food banks. Region was not associated with either food insecurity or food bank use. The absence of interactions by country demonstrates that food insecurity and food bank use are associated with the same demographic characteristics across countries.

Policy implications

The latest data show that one in five adults in EWNI are food insecure, unquestionably identifying this phenomenon as a critical and intensifying nutrition and health issue. A robust policy response to UK food insecurity is long overdue, yet the current approach of allowing the third sector to take responsibility is demonstrably inadequate, as evidenced by uneven food bank use^{(Reference MacLeod, Curl and Kearns13,Reference Pybus, Power and Pickett45)} , which in any case does not alleviate food insecurity^{(Reference Loopstra and Tarasuk46)}. Our findings contribute to these discussions by providing further robust evidence that certain groups – households containing children, respondents with food hypersensitivities and in Wales – experience an elevated likelihood of food insecurity but did not report correspondingly greater food bank use. Thus, while the commitment of emergency food providers is laudable, they should not be relied upon as central responses to food insecurity.

Indeed, our results reinforce existing evidence for the protective role of higher incomes, especially in relation to severe food insecurity. This finding is especially relevant in the current financial climate, where shocks such as the cost-of-living crisis are disproportionately disruptive to low-income households. In the UK National Food Strategy, Henry Dimbleby recommended that ‘ideally, of course, the true cost of eating healthily should be calculated into benefits payments’^{(Reference Dimbleby20)}. Yet, the final strategy contains no measures to directly protect incomes from either work or benefits, instead endorsing peripheral measures such as community healthy eating programmes and a limited extension of free school meals that excludes some households experiencing low or very low food security. The strategy has consequently been criticised for failing to meaningfully address food poverty^{(Reference Goodwin47)}. We urge ministers to align wages and benefits to meet the costs of a healthy diet, noting that food insecurity is not associated with cooking confidence or food management strategies^{(Reference Armstrong, Reynolds and Martins29)}, so individually focused interventions are unlikely to be effective. Instead, structural interventions to enhance incomes are needed. Additional support is needed for people with LTHC and lone parents, groups who are disproportionately likely to experience food insecurity and whose potential to work is commonly constrained. Further targeted support is also needed for people with LTHC and food hypersensitivities, who are most at risk of severe food insecurity.

Limitations

This study is inevitably affected by certain limitations posed by the F&Y2 dataset. As a private household survey, the current analyses will have underestimated the overall prevalence and depth of food insecurity and the scale of food bank use, which is concentrated among vulnerably housed and roofless people^{(Reference Loopstra48)} who are unlikely to have been surveyed in F&Y2. Our analyses are nonetheless valuable in demonstrating the recent growth of food insecurity and in identifying the current characteristics of housed people experiencing food insecurity and food bank use. By asking respondents about their receipt of food parcels, the F&Y2 data offer an incomplete picture of alternative low- and no-cost food provision, such as social supermarkets and community cafes, and informal strategies such as bin diving. Further data are needed to contextualise the current evidence on food bank use within the wider alternative food landscape. We do however caution against allowing a more inclusive understanding of alternative food provision to detract from the overarching goal to build financial resilience to allow people to source food in mainstream, socially acceptable ways.

Future research directions

It would be fruitful to explore the persistence of food insecurity in response to psychological, economic, social and relational factors. Such analyses would be especially valuable in light of the high prevalence of food insecurity among lone parents and respondents with LTHC. Panel models using longitudinal data such as the UK’s Understanding Society would enable a closer approximation to identifying causality and enable an exploration of the frequency of food bank use, where approximately half of the recipients access food banks once^{(Reference Prayogo, Chater and Chapman14,Reference Garratt49)} , but repeat visits are unevenly distributed across different groups^{(Reference Garratt49)}. While F&Y2 asks about the frequency of food bank use, its sample size does not enable detailed analyses on the topic. Sophisticated multivariate analyses of repeat food bank use have the potential to distil groups of short- and longer-term food bank recipients who may benefit from differentiated policy interventions.

By using ordinal logistic regression models, our analyses offer insights not only into the demographic characteristics associated with food insecurity but also the relevance of these characteristics to the severity of food insecurity. Noting the elevated odds of severe food insecurity among respondents with LTHC and food hypersensitivities, future research could valuably explore whether there exist specific demographic profiles of people experiencing food insecurity of differing severities. Such insights could inform bespoke policy interventions.

Conclusions

Our descriptive analyses found that 20·8 % of respondents experienced food insecurity in the past 12 months and 3·6 % had used a food bank over the past 12 months, rising to 17·9 % among those experiencing food insecurity. Our multivariate statistical analyses revealed that food insecurity in EWNI was concentrated among economically disadvantaged groups (those with low incomes and unemployed respondents), younger respondents, households containing children (especially lone-parent households), mixed-race respondents, in Wales and respondents with LTHC or food hypersensitivities. Worryingly, having an LTHC or food hypersensitivity was associated with more severe experiences of food insecurity, a consideration that has received very little research attention to date and that demonstrates enhanced vulnerability among people with existing poor health, where good nutrition may be particularly important.

Food bank use was similarly more prevalent among respondents experiencing food insecurity and unemployed and low-income respondents. However, we observed some differences, where certain respondents – households containing children, with food hypersensitivities and in Wales – experienced an elevated likelihood of food insecurity but did not report an elevated likelihood of food bank use. These discrepancies suggest a divergence between need and crisis support that merits further research attention. No clear geographical variation was evident across outcomes.

Worryingly, the rise in food insecurity, especially severe food insecurity, was not accompanied by an increase in food bank use over the survey period, demonstrating a divergence between need and crisis support. To reduce food insecurity and its negative nutrition and health consequences, policy measures are urgently needed to meaningfully strengthen people’s financial resources and provide enhanced support to vulnerable groups.