Evidence use in the UK’s COVID-19 free school meals policy: a thematic content analysis

Abstract Free school meals (FSM) are a well-recognized intervention for tackling food insecurity among school children. National school closures during the COVID-19 pandemic meant that there was a need to rapidly adapt the delivery of FSM. A range of food-assistance policies were implemented, but it is not clear if they were evidence-based. This study aimed to determine the transparency of evidence use and identify other competing influences in the UK’s FSM policy decisions. Thematic content analysis was used to review 50 publicly available policy documents and debate transcripts on FSM policy published between March 2020 and 2021. This period covered the first national school closures (March 2020–July 2020), school holidays, and the second national school closures (January 2021–March 2021). The Evidence Transparency Framework (ETF) was used to assess the transparency of evidence use in policy documents. We found that overall transparency of evidence use was poor but was better for the Holiday Activities and Food (HAF) program. The Government showed preference for replacing FSM with food parcels, rather than more agentic modes of food assistance, such as cash-vouchers. This preference appeared to be closely aligned with ideological views on the welfare state. With an absence of evidence, value-based reasoning took precedent and was polarized by social media. This article highlights the need for a formal review into FSM, one which includes a comparison of low and high agentic food assistance policies. Such a review would address the evidence gap, improve food assistance policy, and aid policymakers in future periods of uncertainty.


Introduction
The COVID-19 pandemic both highlighted and widened health and social inequalities in the UK (Marmot et al. 2020). As such, rising levels of food insecurity became central to public discourse and a signal of wider societal issues (Koltai et al. 2021;Goudie and McIntyre 2021). The free school meal (FSM) policy provides a free lunch in school to low-income children of all ages whose parents receive state benefits. As FSM-eligible children represent some of the most disadvantaged in society, upon national school closures in March 2020, a fast policy response was required to protect these children from increased food insecurity. Over the pandemic, a range of food-assistance measures was used to replace FSM across regions of the UK. These included schools supplying food parcels, a low-agency approach, and giving supermarket vouchers or cash transfers, a higher agency approach. However, the schemes were implemented with varying degrees of success. For example, research identified that 50% of FSM-eligible children did not access FSM at the start of the pandemic (Parnham et al. 2020).
The complexity of the policy process is well established with a multifaceted range of competing influences on policymakers (Cairney, Oliver, and Wellstead 2016;Cairney 2021;van de Goor et al. 2017). Evidence use in policy formation is widely perceived to be beneficial (Oliver, Lorenc, and Innvaer 2014) and is a stated aim of the Government and civil service (Britchfield and Sasse 2020). The Government's general response to the COVID-19 pandemic has been criticized for its lack of transparency (O'Sullivan 2021). However, it is not clear to what extent evidence guided the FSM policy decisions.
The changes to food assistance during this period raise questions about the longterm responses to food insecurity. For example, the Holiday and Activities Food (HAF) program which provides educational activities and food to FSM-eligible children during school holidays was extended in response to the policy debate on COVID-19. This is a long-term policy and the arguments established in this debate will inform future discussion on food assistance. Through analyzing the UK's FSM policy response, we sought to examine the Government's priorities, consolidate lessons learnt and understand how to better influence the use of scientific evidence in future policymaking in times of uncertainty.
This study aimed to determine the transparency of evidence use behind FSM policy decisions in the UK during different phases (first and second school closures, school holidays) of the COVID-19 pandemic and identify other competing influences on decision making.

Study design
We used thematic content analysis to analyze policy documents and debates relating to FSM. The documents were retrieved from the UK Government website and were published in the first year of the COVID-19 pandemic. As the data were publicly available documents, this allowed us to analyze the transparency of decision making which was available to the public at the time and evaluate how this evolved throughout the study period. Thematic content analysis is well-suited to analyzing policy documents and debates (Theis and White 2021;Razavi, Adams, and White 2019;Pomeranz and Pertschuk 2019), permitting a theory-driven development of research questions and themes. However, we adopted a approach to the thematic content analysis, which enabled both the theory-driven development of research objectives and themes whilst iteratively capturing emergent themes (Mackieson, Shlonsky, and Connolly 2019).

Search strategy
We searched for publicly available documents from the UK Government website and the UK Parliament website (Hansard). In addition, the National Archives were searched for previous versions of documents which were updated multiple times within the study period. The time-period of March 1 2020-March 31 2021 was used to capture the first year of the COVID-19 pandemic in the UK, including the first two national school closures, which corresponded with the first three waves of the COVID-19 pandemic. The following search terms were used "free school meals", "free school meals" "AND COVID", and "Holiday Activities and Food". In total, 152 documents were found, comprised 128 Government website documents and 24 documents from the National Archives. They included policy documents, impact assessments, guidance documents, and press releases. 519 references on the Hansard were found. Documents were initially scanned by their title and summary. References in Hansard were reviewed in their immediate context and references were consolidated as they often occurred within the same debate. Documents or debates were excluded if they referred to FSM as a marker of deprivation, referred to FSM before COVID-19 in its standard functioning, or where the mention of FSM was unsubstantial or uninformative (e.g. one sentence). Subsequently, 97 full-text documents and debates were reviewed for their relevance, of which 47 were excluded leaving a dataset of 50 articles. All documents analyzed are listed in Supplementary Table 1.

Theoretical frameworks
There are multiple frameworks which describe the policy process; these are frequently given with the caveat that in practice, policy cycles are complex. Therefore, several theoretical frameworks were used to develop our research questions and coding framework (depicted in Figure 1).
First, Policy Analysis theory (Cairney 2021) was used to guide our understanding of policy formation, including defining the problem, identifying the available policy options and the criteria used to compare policy options. This framework informed the development of our first three research objectives.
Second, we were informed by the Policy Skills Framework (Civil Service Learning 2014), which is used by the UK Civil Service to codify the essential skills of policy practice. The framework describes three components of equal importance for successful policymaking: (1) evidence (e.g. policy background and statistical analysis), (2) delivery (implementation and project management), and (3) politics (stakeholder management and Parliament engagement). We recognize that the three components are not mutually exclusive in practice, but this framework provided a useful way of categorizing policy influences which are recognized in UK policymaking. Therefore, we organized the influences on decision-making into three broad themes (evidence, delivery, and politics) which informed research questions four and five and our coding framework.
Third, to evaluate the use of evidence in policy making (research question four), we drew on two frameworks: the Evidence Transparency Framework (ETF) (Sense about Science 2016) and the Supporting Policy in Health with Research: an Intervention Trial (SPIRIT) framework (Redman et al. 2015). The ETF (Sense about Science 2016) assesses the use of evidence in four stages of the policy development: Diagnosis; Proposal; Implementation and Evaluation. The ETF was designed by the Institute for Government to rapidly and reliably rate UK government departments on their use of evidence in policy decisions (Sense about Science 2016). This tool suited our research project as it was designed using the same data sources as our study, fitted the UK context and has been used to evaluate the transparency of evidence use in a selection of government documents over time (Brown 2018). A benefit of the ETF is that it was designed to assess evidence transparency when "government first set out a policy publicly" and to enable public to can understand and interrogate a policy (Sense about Science 2016). The framework does not require a full "policy cycle" to have occurred since the policy introduction so is well-matched to the unique, short policy cycle during the pandemic. The SPIRIT framework (Redman et al. 2015) was also used to inform our thematic analysis. The SPIRIT framework identifies and categorizes factors Figure 1. Diagram of the relationship between the study's research objectives, themes, and theoretical frameworks. Note: dotted line describes where frameworks were associated with but did not directly inform the research objective.
likely to influence the use of research in policy and has informed multiple tools for evaluating policymakers' engagement with and use of evidence in policy (R. Makkar et al. 2016). The influences and pathways identified in this framework, such as political ideology, public opinion, and research capacity, informed the inductive codes in the coding framework.

Analysis
We defined three phases to reflect different stages of the policy debate. These were guided by school closure dates only and not wider COVID-19 restrictions on businesses and social gatherings. These were: Phase 1: the first national school closure (March 20 2020-July 15 2020). This corresponded with the first wave of COVID-19 infections in the UK. Phase 2: school holidays. Including Easter 2020, Summer half-term 2020, Summer holidays 2020, October half-term 2020, Christmas 2020, and February half term 2021. Phase 3: the second national school closure (January 5 2021-March 31 2021). This corresponded to the third wave of COVID-19 infections in the UK, linked to the Delta variant.
For each study phase, the following research questions were asked of the data: 1. What was the policy problem? 2. What were the options? 3. What action was taken? 4. Was it transparent that scientific evidence was used to guide the policy decisions? 5. What were the other factors which influenced the policy decisions?
The research questions were each informed by theory, as shown in Figure 1 and described in the previous section.
Thematic content analysis was conducted to qualitatively review the data and answer these five research questions. To assess the transparency of the policy document evidence, we used the ETF to guide our thematic content analysis. The transparency of evidence use was assessed in four areas: diagnosis (what policymakers know about the issue); proposal (why the Government has chosen this intervention); implementation (why this method for delivering the intervention has been chosen); and evaluation (what evaluation has been conducted or is planned). Types of evidence considered were those set out by the ETF and included peer-reviewed literature but also grey literature, such as Government documents, external reviews, external interest groups, case studies, and reports from scrutiny bodies such as the National Audit Office (NAO) (Rutter and Gold 2015).
Analysis began with reading and familiarization (JCP). A coding framework was created to organize theory-driven a priori codes into the five overarching research questions. The data were coded in a second reading (JCP) using the coding framework, to which emergent codes were added. All researchers agreed upon the finalized coding framework that included both deductive (theory-driven a priori codes) and inductive codes (emergent codes from the data). The final reading (JCP) ensured the data were coded in line with the finalized coding framework, which is presented in Supplementary Table 2. There were numerous themes identified in the data under the fifth research question (non-evidenced based factors which influenced the policy decisions). It was beyond the scope of the article to present them all, therefore we focused on three themes: "delivery", "politics" and "policy actors and public opinion". These themes were chosen as they were the most prominent in the data and closely aligned with our theoretical frameworks. A collection of salient extracts from the data are presented in Supplementary Table 3 and organized by theme. Quotes presented in the text were chosen as illustrative examples of main themes, as well as to reflect the range of speakers and document types in the data (Eldh, Årestedt, and Berter€ o 2020).
The ETF also includes a scoring component; however, this is not included in the main article. Analysis in which policy documents were scored using the ETF is presented in Supplemental Table 4.
The frequency that key terms ("Free School Meals", "Marcus Rashford", and "Holiday Activities and Food") were mentioned in Parliament per day over the study period were taken from the Hansard website and displayed graphically.

Descriptive analysis
In total 50 documents and debates were analyzed in this study (Table 1). The greatest proportion were debates from the House of Commons (n ¼ 17, 34%) followed by guidance documents (n ¼ 15, 30%).
Discussion of "Free school meals" in Parliament increased in frequency during the period leading to the first school closure in March 2020 ( Figure 2). However, the most active periods surrounded key policy debates in June 2020, October 2020, and January 2021. The influence of a single policy actor, Footballer Marcus Rashford, is observed from June 2020 when he wrote an open letter to Members of Parliament (MPs), from which point his name was often referenced alongside FSM. Mention of the HAF program was more apparent in October 2020, during the debate on continuing FSM in school holidays, preceding the announcement of the expansion of the program in November 2020. On March 20 2020 schools closed nationally for all but vulnerable children and children of key workers during term-time to reduce the transmission of COVID-19. There was no established system for continuing FSM provision for children at home. Local Authorities and state-funded schools have a legal responsibility to provide FSM in term-time to eligible low-income children, so it was necessary for the Government ensured this provision continued. However, the policy problem was different for school holidays. There was no legal requirement or existing precedent to continue FSM provision during holidays before the pandemic. The Government recognized the increased need of low-income families and that the boundaries between home and school were "blurred" by the national school closures. Therefore, the policy problem for FSM delivery centered around two main questions. During term-time, the question centered over which mode of FSM provision was the best to ensure continuity for children. Whereas during school holidays, the question centered around whether it was necessary to continue FSM provision at all.
"Under normal circumstances, schools are not expected to provide free school meals to disadvantaged children who are not attending due to illness or if the school is closed." Department for Education. "COVID-19: free school meals guidance for schools". March 19 2020 3.2.2. Theme 2: What were the policy options? A summary of policy options available to and considered by the Government is presented below. The policy options were identified through thematic content analysis of publicly available documents and are ordered by the degree of the agency they provide to beneficiaries (see exemplary quotes 2.1-2.5 in Supplementary Table 3). Low agency policies, such as providing meals, limit the beneficiary's capacity to make choices on the food they consume whereas high agency policies, such as cash transfers, give beneficiaries more choice to decide how to use the benefit (McLaren, McIntyre, and Kirkpatrick 2010).
Do nothing: Although schools were legally required to provide FSM in term-time, the Coronavirus Bill introduced on March 23 2020 gave schools the "Power to disapply or modify provisions in relation to education and childcare" (Department of Health and Social Care 2020). In school holidays, not providing FSM was the standard practice. Give meals/food parcels: Schools provide prepared food or ingredients through their usual caterers. Supermarket vouchers: Vouchers are given to beneficiaries to buy food at the supermarket. The vouchers could be arranged by the school or through the national scheme arranged centrally by the Government. Cash transfers: Direct payment of the equivalent money into beneficiaries' bank accounts. Could be arranged by the school or local authorities. Increased benefits: The benefit system could be used to transfer the equivalent money to eligible families in replacement of FSM.

Theme 3: What action was taken?
The policy action over the study period is displayed in Figure 2. 3.2.3.1 Phase 1first national school closure. Between March 2020 and July 2020 FSM were replaced using food parcels or supermarket vouchers. There was a consistent emphasis on flexibility and choice for schools in statements from the Education Secretary, although Government guidance displayed a preference for schools to deliver food parcels or meals through existing caterers (quotes 3.1-3.4).

Phase 2 -School holidays.
Policy action regarding FSM in school holidays varied drastically during the study period, ranging from full continuation of FSM using local or national voucher schemes and food parcels (March 2020-July 2020, January 2021-March 2021) to no FSM replacement (October 2020). Continuation of FSM in school holidays was closely linked to the national closure of schools. Additionally, the decision was debated in Parliament on two occasions; in June 2020 parliament voted to continue FSM but in October 2020 parliament voted against this action (quotes 3.5-3.14).

3.2.3.3
Phase 3second national school closure. Schools returned from Christmas break for one day before national school closures were announced for January 5 2020. Initially, the Government "strongly encourage[d]" (quote 3.15) schools to provide FSM through food parcels. However, this advice was later moderated to "schools may consider" food parcels (quote 3.16). The national voucher scheme was reinstated on January 18 2021 (quotes 3.15-3.17).
3.2.4. Theme 4: Was it apparent that scientific evidence was used to guide the policy decisions? Analysis of the policy documents revealed little transparent evidence use. Only one document, a policy document mentioning the HAF program (Department of Health and Social Care 2021), cited the scientific evidence used. All other documents indicated evidence use but did not include a citation. The transparency of evidence use depended on the policy, there was less transparent evidence use for FSM policies compared to the HAF program. For example, in HAF guidance documents there was a brief background section giving the reasoning behind the policy decision.
"For some children that [reduced incomes and increased costs in school holidays] can lead to a holiday experience gap, with children from disadvantaged families: less likely to access organised out-of-school activities; more likely to experience 'unhealthy holidays' in terms of nutrition and physical health; and more likely to experience social isolation." Department for Education. "Holiday activities and food programme 2021". December 16 2020 However, in FSM guidance documents there was no background section. Only two documents provided a justification for the FSM policy during COVID-19, yet these justifications did not appear to be evidence-based.
Analysis of parliamentary debates also did not indicate transparent evidence use. The most consistent use of evidence was in the diagnosis stage. Politicians regularly quoted scientific evidence to support their argument of whether there was unmet need in the target population, arguing for (quotes 4.1-4.2) and against (quote 4.3).
However, indication of evidence use in the proposal or implementation stage of policy development was lacking (quotes 4.4-4.9). This was especially apparent for justifying the mode of FSM provision, of which there was no scientific evidence use. Further to this, one of the Government's reasons to not choose cash-first approaches was that they may be misspent by beneficiaries, evidence was used by opposition politicians to refute this argument.
"Extensive research by the World Bank in all world economies, not just the poorest, proves that cash transfers work and that concerns around their use on "temptation goods" are "unfounded."

Sharon Hodgson MP (Lab). House of Commons. January 18 2021
In similarity with the policy documents, there was more transparent evidence use for the HAF program in parliamentary debate, with an indication that the recommendations of the National Food Strategy (Dimbleby 2020) were closely followed (quotes 4.7-4.8). The National Food Strategy was a government-commissioned independent report on the UK's food system. As part of recommendations to reduce food insecurity in children, Part 1 of the report suggested expanding the HAF program, amongst other suggestions, such as expanding FSM eligibility. In the evaluation stage, opposition politicians used evidence to highlight flaws in the National Voucher scheme in May 2020 (Phase 1). The audits of the National Voucher scheme by the Commons Public Account Committee (House of Commons Public Accounts Committee 2021) and the NAO (National Audit Office, Davies G 2020) may have factored into the Government's initial decision not to continue the National Voucher scheme in January 2021.
"Data from the Food Foundation this month shows that approximately 500,000 children entitled to free school meals have received no substitutes since March. This is 31% of entitled children." Baroness Bakewell of Hardington Mandeville (Lib Dem). House of Lords. May 14 2020.

Theme 5: What other factors influenced these decisions?
In this section, we will focus on the themes "delivery", "politics", and "policy actors and public opinion". We acknowledge that numerous other factors influence policy decisions, which are not evidence based; however, it is beyond the scope of this article to comprehensively discuss them all, please see Supplementary Table 3 (quotes 5.1-5.44) for more information.
3.2.5.1. Delivery. The FSM policy response was initially guided by feasibility and delivery of implementation (quotes 5.1-5.4). Quickly establishing a system which would function for as many schools as possible was prioritized. A lack of capacity and need for producing policy at-pace was referenced as a reason for forgoing the usual policy process (quote 4.11). The pace of delivery was frequently referenced as a reason for the initial poor functioning of the National Voucher scheme. Additionally, The Government's advice that schools should try to deliver food parcels first may have partly been driven by concerns over reducing costs. This was most apparent in phase two; schools had returned to school for one day before the national closures, resulting in a large waste of food and money.
"One key reason why the Government gave schools the choice was that they were aware of the operation of their own school catering staff-but also, certain suppliers had already purchased food and they had already paid for it so, obviously, moving to a voucher system immediately could have resulted in food waste." The Parliamentary Under-Secretary of State, Department for Education and Department for International Trade (Baroness Berridge [Con]). House of Lords. January 14 2021.

Politics.
Politicians' views on the welfare state appeared to be strongly associated with the Government's FSM policy decisions and their opinion on the purpose of FSM. Preference toward less agentic replacements of FSM during national school closures, such as food parcels, was often justified by the ideological view that more agentic policy options, such as cash or vouchers, would be misspent (quotes 5.8-5.10). Opposition politicians disagreed with this stance (quotes 5.11-5.15). Small state ideology was also central in the Government's view that FSM should not be continued outside of term-time. It was argued that continuing temporary measures would set a precedent and increase welfare spending (quote 5.16). In this way, politicians disagreed over the purpose of FSM (quotes 5. 17-5.22) and the topic of FSM grew increasingly emotional and divisive over the study period (quotes 5.37-5.44).
"I am unashamedly committed to the Conservative ideas of a small state [ … ] some parents just do not or cannot prioritise their children's needs over their own. We must [ … not allow] the Government to step in and do the easiest thing-throw money at the problem." Sally-Ann Hart MP (Con). House of Commons. June 16 2020 3.2.5.3. External policy actors and public opinion. It is important to recognize the influence of external actors, public attention, and social media in the policy decisions that were made and on developing divisive debate (quotes 5.23-5.33). Marcus Rashford ran an influential and ultimately successful Twitter campaign for the Government to "#maketheUturn" on their decision not to continue FSM in the Summer 2020 holidays. However, Rashford's intervention was viewed less favorably by some Conservative MPs (quote 5.30), further polarizing political views on FSM.

Discussion
This analysis of the UK's COVID-19 FSM policy decisions found that overall transparency of evidence use was low but was more evident for the HAF program than for FSM. In absence of evidence use, the Government's preference toward less agentic forms of food assistance, such as food parcels appeared to be driven by ideological views on the role of the welfare state. Policy delivery and responding to increased public attention were other rationale present in the decision making.
There was greater availability of readily accessible scientific evidence on the HAF program compared to FSM, which may explain the difference in evidence use. The Government commissioned a literature review on HAF in 2018 (Evans 2020) and the National Food Strategy released recommendations to extend the program in July 2020 (Dimbleby 2020). There is no comparable review for FSM or modes of food assistance. This finding is common in the evidence-based policy making literature . It has been highlighted that an accessible reservoir of evidence can aid evidence use in policymaking (van de Goor et al. 2017;Redman et al. 2015). For example, interviews with MPs about their approach to evidence use revealed that the perceived credibility of evidence was an important factor in determining evidence use (Britchfield and Sasse 2020;Kenny et al. 2017) and government sources of data were found to be most useful by MPs (Oliver and de Vocht 2017). This is a view found to be shared by policymakers from multiple EU countries (van de Goor et al. 2017). This highlights the success of the National Food Strategy (Dimbleby 2020) in instigating policy change. The report was well-researched, made evidence accessible, and was well-regarded by a range of policy actors including academics, non-governmental organizations, and the public. Furthermore, while this study has one country and policy context, this conclusion has relevance to future periods of uncertainty worldwide. Evidence-based policy recommendations need to be well-synthesized and highly accessible to aid evidence use in times of uncertainty.
However, despite the availability of credible evidence for the HAF program, the transparency of evidence use was still poor. The unprecedented nature of the pandemic necessitated policy changes on a shorter timescale to normal. Yet, these were expensive, long-term policies and the decisions made during this period will likely extend beyond the reach of the pandemic. Therefore, evidence-based reasoning remains essential. It has been suggested that it is precisely because these decisions were made outside of the standard policy process that additional scrutiny should be applied, not less (Airoldi and Vecchi 2021). Furthermore, evaluation of Government policy prior to COVID which used the ETF similarly showed that the Department for Education's policies scored low for evidence transparency (Brown 2018), indicating that this Department in particular may need to improve the quality of their communication with the public.
The lack of a concise summary of FSM policies compared to HAF may have acted as a barrier to time-pressed MPs (Britchfield and Sasse 2020;Kenny et al. 2017) and allowed value-based decisions to take a greater role in decision making. Policymakers make decisions with "bounded rationality" and in fast-paced or uncertain policy environments irrational or emotional shortcuts are used to make decisions quickly (Cairney, Oliver, and Wellstead 2016). Furthermore, the usual pressures on policymakers were intensified by the pandemic (Vickery et al. 2022). Indeed, we saw that in the absence of evidence, individual political ideological perspectives strongly influenced policymakers' view on modes of FSM, which were closely aligned with wider debates on the welfare state. Furthermore, our research indicated that social media played a role in polarizing these value-based views. MPs are known to cluster into value-based networks on Twitter (Weaver et al. 2018), further entrenching and validating their opinion. The political debate was divisive and steeped in emotional language; it is therefore unsurprising that this political environment did not produce a strategic, organized, or efficient approach to FSM during COVID-19.

Strengths and limitations
This is the first study to define and examine the FSM policy decisions during the COVID-19 crisis in the UK, which was a rare and unprecedented policy space. We used the ETF which was developed for evaluating transparency of evidence use in UK policy documents. UK policymakers have recognized the need to benchmark their transparency of evidence use to highlight areas of best practice (Britchfield and Sasse 2020), so this work serves to highlight areas of improvement. Furthermore, we used multiple types of evidence and sources offering a diverse and comprehensive picture of the rationale behind policy decisions.
Some limitations should be noted. The main type of documents reviewed, guidance documents and debates, are not typically expected to include detailed referencing. The fast-paced nature of policy during the pandemic meant the Government may not have had capacity to produce in-depth evidenced work. However, the study period was over a year and long-term policy decisions were made. Therefore, we determined that it was reasonable to evaluate the evidence use within this timeperiod. Furthermore, the ETF was designed to be applied to policy documents when the "government first set out a policy publicly" (Sense about Science 2016). The framework emphasizes the necessity for transparency in decision making from the outset. Finally, only one reviewer read the data, however, multiple researchers confirmed the final coding framework.

Policy implications
There is still uncertainty regarding when the pandemic will end, and the future COVID-19 restrictions required. Therefore, it is concerning that despite such debate, time, and resources invested in the area, there may still not be a resolute and welljustified approach to continuing FSM if another national school closure was required. This research highlights the need for a formal review into FSM, including comparison of the merits and issues with both low and high agentic food assistance policies. The policy options deployed over the pandemic were heterogeneous and encompassed a wide range of food assistance modalities. This could provide an opportunity for evaluating food-assistance approaches and addressing the existing evidence-gap. A formal review would not only ensure that FSM children are better supported in future pandemic restrictions but would also have important implications for other food assistance policies in the UK. Understanding the most effective approach to food assistance is critical to gathering future political support and ensuring value for money. Furthermore, this article also has important implications for understanding future discussions on food assistance policy. The political tensions generated in this period will go on to inform future policy discussions, which will likely continue to be highly divisive.

Conclusion
Our evaluation of the UK's COVID-19 FSM policy decisions found that evidence use was not transparent, although was improved for the HAF program. In absence of evidence, value-based rationale, such as ideology may take precedent and be polarized by social media. A credible and well-synthesized review of the FSM policy in this period could ensure lessons are learnt and increase the likelihood that future food assistance policy is evidence-based. The lessons must be incorporated into a future pandemic plan which prioritizes protecting low-income families from food insecurity.

Ethical approval
Ethical approval was not required for this study as the data used were publicly available from the UK Government website.