Introduction

The SARS-CoV-2 (COVID-19) pandemic significantly changed how food businesses operated. Before widespread vaccination, containment was the main strategy used to limit the spread of the virus. The introduction of virus containment measures from March 2020 led to an 89% global reduction in restaurant traffic (Aaron Allen & Associates 2020). In the UK at the start of the first national lockdown in March 2020, no food business was permitted to serve customers on their premises (The UK Government 2020a). However, food businesses, such as restaurants and pubs, were able to act as takeaways between March 2020 and March 2022 without requiring planning permission for a change of building use (Moore et al 2022). Previously, food businesses were required to get planning permission from their local authorities before they could provide takeaway services. In support of the pandemic lockdowns, the UK Government introduced a furlough scheme to cover the majority of employee wages for those businesses unable to operate as usual (Clark 2021). The proportion of wages covered by the Government changed as the COVID-19 pandemic progressed, with the scheme ending in September 2021. The restaurant sector was highly dependent on this scheme with 9% (approximately 159,400) of all employees furloughed at some point during the pandemic.

The general population was also restricted as to how often they could leave their homes and who they could interact with outside their household (Public Health England 2020). These restrictions on movement, coupled with fear of the virus, changed how people engaged with the out-of-home food sector. Evidence from Food Standards Scotland (2021) showed that the average number of visits to an out-of-home food outlet dropped from 4.2 times per week in 2019 to 2.6 times per week in 2020. Conversely, the market value of takeaways increased by 31% and the number of new customers using delivery services increased by 54% between 2019 and 2020. There is also evidence suggesting a significant rise in online food sales and the use of online food delivery platforms such as Deliveroo particularly during the first national lockdown in England (Panzone et al. 2021; The Guardian 2021).

Another important factor at play was the Government white paper on Planning for the Future which was released in August 2020 (The UK Government 2020b). Part of the changes proposed for the planning system included reduced restrictions on where restaurants and pubs could open, but greater restrictions on where takeaways could open nationally. This change in planning class for takeaways has been motivated by evidence finding an association between increased consumption of fast food and obesity (Fraser et al. 2012; Bahadoran et al. 2015). Since the responsibility for public health has moved to local authorities in 2012 (The UK Government 2012), more local authorities have taken an active role in managing and supporting the out-of-home food environment. For example, approximately 50% of local authorities have supplementary planning guidance in place restricting new takeaways within their jurisdiction (Keeble et al. 2019). Public Health England (now Office of Health Improvement and Disparities and UK Health Security Agency) developed a toolkit to support local authorities to monitor the out-of-home food environment, developed guidance (such as for planning), and worked with local food businesses to provide healthier food options (Public Health England 2017).

Given the current complexity in the out-of-home food sector due to Brexit (Financial Times 2022; The New York Times 2021), rising prices (The Lancet Public Health 2022), the changes in the planning policy (The UK Government 2020b), as well as the COVID-19 pandemic, it is important to understand how the out-of-home food environment has changed from pre-pandemic. To date, there have been no published national-level studies to analyze how the trends in the out-of-home food sector have changed since the start of the COVID-19 pandemic.

The aim of our research was to explore how the size and types of outlets in the out-of-home food environment has changed in England during the course of the COVID-19 pandemic. This study provides important evidence for the development of food, public health, and planning policy going forward both locally and nationally. The COVID-19 pandemic provided a shock to the food environment and it is important to understand the impact of this and what it means for the future of the type and numbers of outlets and how this may affect health. We took a two-stage approach, first, we looked at the changes between 2019 and 2021 in the six standardized food business types routinely collected in England. These are: (1) fast food/sandwich shop; (2) pub/night club/bar; (3) restaurant/café/canteen; (4) supermarket; (5) mobile caterer; and (6) other catering premise. Secondly, we compared the changes in the number of the chain (multiple outlets nationwide) fast food takeaway outlets against the non-chained independent takeaways.

Methods

Data

We used publicly available data from the Food Standards Agency’s Food Hygiene Rating Scheme (FSA FHRS). All data can be accessed at the following link: https://ratings.food.gov.uk/open-data/en-gb. The FSA FHRS is the national monitoring program of the hygiene of all food businesses to ensure that they are complying with food hygiene law in the UK. In each local authority, environmental health officers are required to inspect all food outlets within their local authority and upload the food hygiene rating within 28 days of an inspection (Food Standards Agency, N.A.). As part of the FHRS, they are also required to capture food business type inclusive of manufacturers, distributors, retailers, takeaways, restaurants, pubs, schools, hospitals, community centers, supermarkets, and mobile caters. For our research, we were interested in those businesses whose practices included out-of-home food provision. Therefore, from the FHRS, we selected data on six type of food outlets: (1) fast food/sandwich shop; (2) pub/night club/bar; (3) restaurant/café/canteen; (4) supermarket; (5) mobile caterer; and (6) other catering premise. Existing evidence has shown that the FSA FHRS dataset is more comprehensive in its coverage than other comparable commercial datasets (Kirkman et al. 2021) for the Northeast of England.

We restricted our data analysis to only local authorities (LA) in England between March 2019 and December 2021 (34 months). The FSA FHRS dataset includes information on food businesses in 311 English local authorities. Two local authorities: Three Rivers and East Staffordshire classified ‘mobile caterers’ as ‘other catering premises’ before August 2019 and May 2020 respectively but changed their classification to ‘mobile caterers’ after this period. Another local authority Lichfield classified ‘other catering premises’ as `restaurants’ after May 2020 but not before. Therefore, we excluded these local authorities with inconsistent food outlet classification from the analysis of ‘mobile caterers’, ‘other catering premises’, and ‘restaurants’.

Before conducting analysis as part of data preparation, we identified missingness in the dataset. During the COVID-19 pandemic, environmental health officers who normally inspect food businesses were seconded to other posts within the local authority. No face-to-face inspections on premises were being undertaken (Food Standards Agency 2021). Thus, although normally data is uploaded within 28 days, this did not happen during some months, so we dropped months where no new data was entered. Most of the missing months were in 2021. To account for data missingness, we used a simple linear interpolation method (Bayen and Siauw 2015). Research using the data pre-COVID (Brown et al. 2021, 2022) showed linear trends in the growth in all types of food outlets in the Northeast of England. Therefore, we assumed that the food environment (i.e., number of food outlets) is changing at a relatively constant rate between any two observations.

Outcome variables

The primary outcome variable is the monthly count for each of the 6 types of food outlets for each LA (1) fast food/sandwich shop; (2) pub/night club/bar; (3) restaurant/café/canteen; (4) supermarket; (5) mobile caterer; and (6) other catering premise. Our secondary outcome variable is a binary variable which equals one if a takeaway business is a chain and is equal to zero if a takeaway business is independent. To classify fast food chains, we used the following keywords to identify these businesses: McDonalds, KFC, Burger King, Dixy Chicken, Krispy Kreme, Taco Bell, Domino’s Pizza, Papa Johns, Pizza Hut, Pizza Express, Nando’s, Subway, Five Guys, Greggs, Chicken Cottage, Pret a Manger, Costa, Starbucks, and Caffe Nero. For non-chained hot-food outlets, we selected keywords that are related to the following seven cuisines/types of food outlets: fish and chips, chicken, pizza, Chinese, Thai, Indian, coffee and tea, kebab, and general takeaways. The full list of keywords that we used for this process can be found in online Appendix A.

The COVID-19 pandemic lockdown timeline in England

Figure 1 shows the timeline of the UK Government COVID-19 pandemic lockdowns and associated restrictions. To mitigate the impacts of lockdowns, food businesses were temporarily allowed to provide takeaway services without requiring permission from their local planning authority between March 2020 and March 2022 (Moore et al. 2022). Based upon the different pandemic restrictions in place, we defined six distinct time periods in our sample: (1) the pre-intervention period, March 2019–February 2020; (2) the first intervention period (the first national lockdown), March 2020–June 2020; (3) the second intervention period, Eat Out to Help Out Scheme,Footnote 1 July 2020–September 2020; (4) the third intervention period (second national lockdown), October 2020–December 2020; (5) the fourth intervention period (third national lockdown), January 2021–March 2021; and (6) the fifth intervention period is the easing of restrictions following the third national lockdown, April 2021–December 2021.

Fig. 1
figure 1

Timeline of COVID-19 lockdowns and measures in England

Full size image

Statistical analysis

To statistically examine the impact of lockdown policy on the food environment, we used a panel interrupted time series analysis (PITSA) approach (Linden 2017, 2021, 2022). The PITSA is commonly used to evaluate the effect of an intervention or a series of interventions on an outcome variable across a number of cross-sectional units. The PITSA allows to control for pre-existing trends. There is evidence showing that all types of food outlets have an increasing trend in the past decade (d'Angelo 2020; Brown et al. 2022). The PITSA is capable of examining the treatment effects of multiple interventions that occur within a specific time frame. In our setting, we have identified five interventions as shown in Fig. 1, which gives us six segments. Therefore, we are able to compare the changes in the food environment before and after each intervention and estimate the effects of multiple intervention periods using the PITSA.

Specifically, we estimated the following model:

$${Y}_{ti}={\beta }_{0}+{\beta }_{1}{\mathrm{First}}_{ti}+{\beta }_{2}{\mathrm{Second}}_{ti}+{\beta }_{3}{\mathrm{Third}}_{ti}+{\beta }_{4}{\mathrm{Fourth}}_{ti}+{\beta }_{5}{\mathrm{Fifth}}_{ti}+{\beta }_{6}{T}_{ti}+{\beta }_{7}{T1}_{ti}{\mathrm{First}}_{ti}+{\beta }_{8}{T2}_{ti}{\mathrm{Second}}_{ti}+{\beta }_{9}{T3}_{ti}{\mathrm{Third}}_{ti}+{\beta }_{10}{T4}_{ti}{\mathrm{Fourth}}_{ti}+{\beta }_{11}{T5}_{ti}{\mathrm{Fifth}}_{ti}+{\varepsilon }_{ti}$$
(1)

where Yti represents the count of one of the six types of food outlets (1) fast food/sandwich shop; (2) pub/night club/bar; (3) restaurant/café/canteen; (4) supermarket; (5) mobile caterer; and (6) other catering premise) in a local authority, i, in month, t. So Eq. (1) is estimated 6 times for each of the six different types of outlets. First is a dummy variable indicating the first intervention period. T is the number of months since the start of our analysis. T1 is the number of months since the first intervention (the value of pre-first-intervention period is set to 0). Accordingly, Second to Fifth indicate the second to the fifth intervention periods, and T2–T5 represent the number of months since the corresponding intervention. We also included five interaction terms, T1*First to T5*Fifth, for the five intervention periods. ε is the clustered error term at the local authority level.

β0 is the intercept that estimates the starting number of a type of food outlets. β1–β11 are the parameters of coefficients to be estimated. β1–β5 estimate the immediate effect of the five interventions on the number of food outlets. β6 estimates the trend of the outcome variable until the first intervention (i.e., the first national lockdown).

β7–β11 estimate the change in the trend (growth rate) of the outcome variable after each of the five interventions. β7 estimates the change in the trend of the outcome variable after the first intervention. β8–β11 estimate the difference between the current intervention trend and the previous intervention trend rather than the difference to the pre-intervention trend (Linden 2017, 2022).

Results

Table 1 summarizes the average number of food outlets across the local authorities by type of food outlets every 3 months over the study period. We found that all types of food outlets, except for pubs, had an increasing trend over the whole study period. The number of pubs remained constant over time.

Table 1 Average number of food outlets by type across the local authorities over time
Full size table

Table 2 presents the results of our interrupted time series analysis for the number of food outlets categorized by FSA FHRS Individual plots illustrating the temporal change for each type of outlet can be seen in online Appendix Figs. A1–A6. β0 shows the estimated average number of each type of food outlet across the local authorities in March 2019 the start of our study. β6 shows that, before March 2020, the first national lockdown, the number of all types of food outlets, except for pubs, had statistically significantly increased.

Table 2 Interrupted time series analysis of the number of food outlets by the FSA FHRS business type
Full size table

β1–β5 show changes in the count of food outlets. Lockdowns and associated COVID-19 restrictions had immediate impacts on the number of food outlets. After England entered the first national lockdown in March 2020 (β1), the number of all types of food outlets statistically significantly decreased. The number of restaurants declined by 2.43, the number of other caters decreased by 2.57, there were smaller declines in fast food of 1.04 and mobile food outlets 1.60, and the smallest decline in supermarkets 0.22. There was not a significant decline in the number of pubs. However, the number of all types of food outlets except for pubs and supermarkets had statistically significantly increased immediately following the Eat Out to Help Out Scheme (β2) ranging from 4.29 for other catering to 1.50 for restaurants, 1.75 for fast food, and 0.87 for mobile caters. After the second national lockdown (β3), there was a significant increase for all types of food outlets with the largest rise again for other catering (4.86) and an increase of 1.27 for fast food, 1.54 for restaurants, a smaller increase for pubs (0.56), mobile caterers (0.64), and supermarkets (0.27). The third national lockdown (β4) had no significant change on the number of all types of food outlets. After the UK government announced the gradual easing of the final national lockdown (β5), the number of mobile caterers (1.58), restaurants (1.03), pubs (0.43), and supermarkets (0.13) increased, but the number of fast food outlets and other catering premises did not have statistically significant change.

β7–β11 show the changes in trends for growth of different types of outlets. Compared with the pre-lockdown period (β7), there were no significant changes in the growth of fast food outlets, mobile caterers, and pubs during the first lockdown. The growth rate of restaurants (– 0.38) and supermarkets (– 0.08) had statistically significantly decreased, but the growth rate of other catering premises (0.95) had statistically significantly increased compared with the pre-lockdown period. Compared with the first national lockdown period (β8), growth in fast food (0.54), other catering (2.06), mobile caters (0.49), and restaurants (0.59) increased. There was no change in trend for pubs and supermarkets. During the second national lockdown (β9), there was a further increase in the growth rate of fast food outlets (0.80), other catering premises (3.15), and restaurants (0.97). There was no change in trends for the other three types of food outlets. During the third national lockdown (β10), there was a significant decrease in the growth of all types of out-of-home food outlets, 0.36 for fast food, – 1.40 for other catering, – 0.31 for mobile caters, – 0.40 for restaurants, – 0.17 for pubs, and – 0.06 for supermarkets. With the gradual easing of the final national lockdown (β11), we found no significant changes in the growth of mobile caterers, restaurants, and pubs compared to the previous period. However, there was a significant increase in the growth of supermarkets (0.06) and decrease in the growth of fast food outlets (– 0.98) and other catering premises (4.56) compared with the third national lockdown period.

Comparing the magnitude of coefficients between different types of food outlets, we found that pubs and supermarkets were the least sensitive to the lockdowns and COVID-19 restrictions. Other types of food outlets were more sensitive to external changes over this period. Overall, other catering had the largest growth over the lockdown period which reversed after restrictions eased.

Fast food chains versus non-chained hot-food outlets

Table 3 details the results for interrupted time series analysis for the number of fast food chains and non-chained fast food outlets. Respective plots can be found in online Appendix Figs. A7 and A8. Online Appendix Fig. A9 compares the growth rate of fast food chains and non-chained hot-food outlets since the beginning of our study. On average, each local authority had an estimated mean of 37 fast food chains and 234 non-chained hot-food outlets in March 2019. During the pre-intervention period (β6), both fast food chains and non-chained hot-food outlets grew by 0.17 and 0.79 per month.

Table 3 The number of fast food chains and non-chained hot-food outlets
Full size table

β1–β5 show the immediate impact of interventions on the number of chained and non-chained food outlets. After England entered the first national lockdown in March 2020 (β1), the number of chained and non-chained food outlets decreased by 0.17 and 2.02. Following the Eat Out to Help Out Scheme (β2), the number of non-chained hot food outlets had a statistically significantly increase of 2.15. However, fast food chains were less sensitive to the Eat Out to Help Out Scheme. After the second national lockdown (β3), chained and non-chained food outlets had significant increases of 0.26 and 2.50 respectively. The third national lockdown (β4) led to an increase in the fast food chains (0.22) but a decrease in the non-chained hot food outlets (– 0.73). After the UK government announced the gradual easing of the final national lockdown (β5), there was an increase in both chained (0.12) and non-chained food outlets (1.31).

β7–β11 show the changes in the growth rate of chained and non-chained food outlets. Compared with the pre-lockdown period (β7), the growth rate of fast food chains significantly decreased by 0.15, but the growth of non-chained hot food outlets had no significant change. Compared with the first national lockdown period (β8), we found no significant change in the growth of fast food chains but a significant increase in the non-chained hot food outlets (0.77). During the second national lockdown (β9), the growth of fast food chains kept the same but the non-chained hot food outlets had a further increase of 1.40. During the third national lockdown (β10), the growth of both chained and non-chained food outlets experienced significant decreases of 0.05 and 0.80. With the gradual easing of the final national lockdown (β11), although both chain and non-food chains continue to grow, the growth of fast food chains started to rebound (0.13), but the rate of growth of non-chained hot food outlets decreased by – 1.27 compared with the growth in the previous intervention period.

Discussion

Using national data at the local authority level, our study provided the first statistically robust analysis of how the food environment has been changing in England since the start of the COVID-19 pandemic. Our results showed that during the lockdowns associated with containing the COVID-19 pandemic, there was a decrease in the growth in all types of food outlets. In between lockdowns with government programs such as the Eat Out to Help Out scheme, there was an increase in growth in all types of food outlets except supermarkets and pubs. After the easing of all restrictions, there was an increase in growth for mobile caterers, restaurants, pubs, and supermarkets but not fast food outlets and other catering. When comparing chained and non-chained hot-food outlets, non-chained food outlets were more responsive to government policy such as the Eat Out to Help Out scheme. After the easing of restrictions, both chained and non-chained food outlets continued to grow, but the former experienced a significantly higher rate of growth and the latter experienced a significantly lower rate of growth compared to the lockdown periods. Differences in growth and growth rates between the different types of food outlets during the pandemic may stem from how people used them with some being more sensitive to changes in policy restricting movement.

These findings have implications on how to manage the food environment going forward. Evidence from both the US and UK exploring changes in the food environment over time has uniformly tended to show an increase in all types of food outlets (James et al. 2017; d'Angelo et al. 2020). The COVID-19 pandemic led to a sudden unexpected change how people engaged with their food environment (Moore et al. 2022). Particularly at the start of the pandemic, there was an increase in people eating less food from the out-of-home sector and more food prepared at home purchased from supermarkets (The Obesity Health Alliance 2020). As a result, the supermarket sector in the UK had an estimated £4 billion growth in sales during the first national lockdown period (Panzone et al. 2021). As COVID-19 restrictions ended, expenditure at out-of-home food businesses returned to pre-pandemic levels (Office for National Statistics 2022a). However, our findings suggest that not all types of outlets have benefitted from this increased expenditure. Only supermarkets (β11 in Table 2) and chained fast food outlets (β11 in Table 3) had significant growth following the ease of COVID-19 restrictions.

The situation following the end of restrictions is more challenging because of the cost-of-living crisis and Brexit. Particularly, the rise in the cost of food was one of the largest upward contributors of inflation between June and July 2022 (Office for National Statistics 2022b). This may lead to longer term changes in how people engage with the out-of-home food sector and wider health inequality as many families will likely suffer from increased food insecurity (The Lancet Public Health 2022). With rising energy bills and post-Brexit immigration rules implemented on 1st Jan 2021, many food outlets were closed because of the high operating costs and staff shortage (BBC 2022; Financial Times 2022; The New York Times 2021). This may have negatively impacted on local food variety and availability, which may then have implications for dietary intake from the out-of-home food sector and its impact on health.

Our findings have policy implications at both the local, national, and international level. For local authorities to successfully manage their food environment, it is important to be able to monitor how the food environment is changing over time and how shocks to the system, such as the COVID-19 pandemic and the associated containment measures and impact on the supply chain, affect the rate of growth of different types of food outlets. This can help them prioritize what types of policies they should focus on (e.g., restriction of new food outlets, working with existing food outlets to reformulate their menu to provide healthier options, or even creating incentives for certain types of food businesses to locate in the area).

At the national level, these findings can be used to monitor the food environment and develop national policy to feed into the Government’s ambition to reduce obesity in adults and children by half in 2030 (Department of Health & Social Care 2020). The success of the Eat Out to Help Out scheme has shown that central government-led interventions are significantly more impactful than the voluntary favored scheme of the past (Theis and White 2021).

Internationally, these findings are relevant to governments from other high-income countries which are also grappling with a ‘bruised’ hospitality sector where trends may be similar for the out-of-home food sector. In the UK, as is the case in many other high-income countries, takeaways tend to cluster in areas of high deprivation; contributing to health inequalities (Green et al. 2021; Hobbs et al. 2019; Public Health England 2016). Thus, it is also important to consider trends in out-of-home food consumption and what this may mean for the contribution of this sector to rising obesity rates globally.

It is also worth noting how the various interventions impacted upon different businesses during the pandemic and what learning can be applied to future policies that may target an improvement in the healthiness of our out-of-home food environments. We showed heterogeneity in the response by different types of food businesses toward opening and closing outlets to the various policy interventions by the UK government. There may have been various factors behind this such as the accessibility to furlough funding depending upon business type (e.g., it may have been administratively more challenging for small food businesses to apply for and access furlough funding) (Office for National Statistics 2022c). Therefore, it is important for policy-makers to consider these differences when developing future food policy.

Strengths and limitations

Our study has several strengths. Our dataset is at local authority level and covers all food outlets in England over the pre-pandemic, pandemic, and post-pandemic periods. We have used a comprehensive and validated data source and applied robust statistical estimates. There are some limitations. For example, our dataset contains a number of missing observations, which may lead to a bias our estimations. In addition, changes in how food outlets were inspected during the pandemic, which included the use of remote rather than in person inspections (Food Standards Agency 2021). Plus, additional responsibilities related to the pandemic for environmental health teams which have reduced their capacity to enter data to the FSA system in a timely manner (Moore et al. 2022). This may mean that we have underestimated the subsequent growth and decline in out-of-home food outlets.

Conclusion

The COVID-19 pandemic was a significant shock to the out-of-home food environment in the UK. In order to understand what lasting impact this may have had we explored trends in growth for 6 different types of food outlets: (1) fast food/sandwich shop; (2) pub/night club/bar; (3) restaurant/café/canteen; (4) supermarket; (5) mobile caterer; and (6) other catering premise during and after COVID-19 pandemic restrictions. During the restriction of movement as a result of the lockdowns, there was a decrease in the number of all types of out-of-home food outlets. The Eat out to Help Out scheme had the Government’s desired effect of stimulating growth in the out-of-home catering sector. However, following the end of COVID-19 restrictions, with the population of England adjusting to a new normal, only supermarkets experienced significant growth. Our results also highlight the greater adaptability of non-chained hot-food outlets during the pandemic compared with chained fast food outlets. However, the growth of fast food chains started to rebound when COVID-19 restrictions were lifted which was not seen for non-chained hot-food outlets. Policy-makers should take into account the differential impact of government interventions on various types of food businesses evidenced from our results when developing future policies to ensure equal opportunities for economic growth and the potential impact on population health.