Introduction

During 2020, COVID-19 mitigation measures were put in place in Great Britain (England, Scotland and Wales) in an attempt to control the spread of the virus and limit its impact on public services. The level and geographic scope of restrictions varied considerably over the year, from the initial national ‘lockdown’ implemented at the end of March, with some relaxation during May and June, to ongoing restrictions for the rest of the year that differed across and within England, Scotland and Wales, in terms of non-essential travel, social gatherings, and openings and controls on a large range of retail, hospitality and leisure outlets [1–3], Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. COVID-19 lockdown regulations affecting the on-trade in England, Scotland, and Wales during 2020.

https://doi.org/10.1371/journal.pone.0261609.t001

With respect to alcohol, on-trade outlets (pubs, bars and restaurants) were fully closed from 21^st March to 4^th July (England), 13^th July (Wales) and 15^th July (Scotland). From July onwards, restrictions varied according to local authority and devolved status (England, Scotland or Wales), affecting the size and type of social gatherings, and whether, where and when on-trade alcohol sales could take place, i.e., outside or inside the premises, at the bar or only at tables, and in terms of licencing hours. Some repeated shorter term full closures of on-trade premises were introduced during October and November, and at the end of December [1–3].

Many studies have suggested that COVID-19 confinements and other mitigation measures have led to reporting of increased mental health problems, including anxiety and depression [4, 5]. For alcohol consumption, however, a mixed picture has emerged, at least in Great Britain. On the one hand, population survey data has suggested that there have been increases in alcohol consumption during this period [6–8], and mortality registration data has shown an increase in alcohol-related deaths in England and Wales [9]. On the other hand, analyses of off-trade purchase and sales data [10, 11] found no increase in purchases or sales, when taking into account foregone purchases due to pub closures.

One explanation of these differences could be that, whereas overall levels of alcohol sales and consumption have not increased, the distribution of changes within the population during lockdown has varied. During complete or partial confinement restrictions, people at higher risk of alcohol related harm, such as those already drinking heavily, or in lower socioeconomic groups, may have been over-represented in surveys as opposed to lighter drinkers [12].

Using British household data for the six years 2015 to 2020, we study the impact of COVID-19 on alcohol purchases and consider the extent to which any changes varied over time and by household characteristics, such as age of main shopper, household income, social grade, area of residential deprivation and how much alcohol households normally purchased. Specifically, we ask if households that normally bought more alcohol, and households with indices of deprivation increased their purchases disproportionally.

Methods

Results and discussion

Data for just over five million separate alcohol purchases were provided by 79,417 households over the six years from 2015 to 2020. Of the households, 70,903 provided purchase data for 2015 to 2019, and 29,890 for 2020. The distribution of socio-demographic factors of the households for the two time periods is illustrated in Table 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Distributions of socio-demographic characteristics of households for the two time periods, 2015 to 2019 and 2020.

https://doi.org/10.1371/journal.pone.0261609.t002

Households with older residents (age of main shopper) were more likely to be heavier purchasers of alcohol, be classified as social grade E and have a lower income, but not be situated in a more deprived residential area, than households with younger residents, Table 3. Households in the north of England were slightly more likely to be heavier purchasers of alcohol, to have older residents (age of main shopper), be classified as social grade E, and to have a lower income, and considerably more likely to be situated in a more deprived residential area.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Differences in socio-demographic factors by age of main shopper and geographic location of household.

https://doi.org/10.1371/journal.pone.0261609.t003

Plots of purchases by calendar day of year for the average of 2015–2019 and for 2020 demonstrates parallel trends between the two time periods, confirming the appropriateness of 2015–2019 as a control period, Fig 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Plots of purchases of grams of alcohol by day of year for 2015–2019 (averaged) and 2020.

Purchases are the sum of purchases in grams of alcohol per adult per household per day across all households for each day of the year, converted to a per cent scale, where 100% is the mean of the sum of purchases per day for the average of the days of 2015–2019 up to 22^nd March, covering the days of the year before confinement.

https://doi.org/10.1371/journal.pone.0261609.g001

Increase in purchases by socio-demographic characteristics of households

Table 4 lists the incidence rate ratio for increased purchases of grams of alcohol by household characteristics. The greater the amount of alcohol normally purchased by a household, the greater the ratio of increase of alcohol purchases following confinement. The highest purchasing households increased their purchases of alcohol following confinement more than 17 times compared to the lowest purchasing households. Households where the age of the main shopper was 65 years or older increased their purchases more than households where the main shopper was aged younger. Households from social grade E (which includes those with state pensions only, casual workers and unemployed with state benefits only) increased their purchases more than households in grade AB (which includes those with higher and intermediate managerial, administrative and professional occupations); however, those in grade D, which includes manual workers, had a lower increase in purchases than households in grades AB. Higher income households increased their purchases more than lower income households. For residential deprivation, those in the most deprived group increased their purchases more than those in the least deprived group. In general, households in the north of England (except for the north-west) increased their purchases by more than households in the south of England. Households in Scotland and Wales increased their purchase by a much smaller amount compared to the regions of England.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Incidence rate ratio of increased purchases of alcohol (95% confidence intervals) by household characteristics.

https://doi.org/10.1371/journal.pone.0261609.t004

Discussion

We show that the introduction of COVID-19 confinement in Great Britain towards the end of March 2020 was associated with households buying 29% more off-trade alcohol (expressed as grams of alcohol) for the rest of the year than would have been expected based on average purchases throughout 2015–2019. The increase was greater (38%) during the times of full pub closure (end of March to early to mid-July 2020) than for the rest of the year (25%).

However, when taking into account that households could not buy on-trade alcohol from the end of March to early to mid-July 2020 (as outlets were closed) and that there were continuing restrictions for the rest of the year in purchasing on-trade alcohol, a different picture emerges. In this hypothetical scenario, during the times that pubs were closed, there was no change in the overall amount of alcohol that households purchased (combining off-trade and on-trade inferred from previous years’ purchasing profiles), and a decrease in purchases thereafter.

Increases in purchases were predominantly driven by households that were usually the highest purchasers of alcohol. The top one fifth of purchasing households (by how much they normally purchased) increased their purchases 17 times more than the bottom one fifth of purchasing households. There was some evidence to suggest that the most disadvantaged households increased their purchases more than the least disadvantaged households, based on social grade and deprivation index, and, to some extent, on household income. In general, households in the north of England increased their purchases more than households in the south of England and the rest of Great Britain, probably because households in the north of England tended to be more disadvantaged and to some extent heavier purchasers of alcohol in general than households elsewhere.

That households with older residents (as measured by age of main shopper) increased their purchases more than households with younger residents may be partly explained by the fact that such households tended to be heavier purchasers of alcohol in general, and to be more disadvantaged than households with younger residents.

There is a paradox when comparing our results with other analyses. On the one hand, our results are consistent with customs and excise tax data [21, 22] and other sales data that showed no overall increase in alcohol purchases following COVID-19 confinement [11]. On the other hand, our results do not confirm the self-reported increases in alcohol consumption due to lockdown found in surveys [8, 23–25]. However, that heavy buying households increased their purchases and that, to some extent, the most disadvantaged households increased their purchases more than the least disadvantaged households, might explain the increases in alcohol consumption reported in some surveys and the increase in alcohol-related deaths reported in England & Wales [9].

Data from the behavioural risk factor surveys of Public Health England [8] show changes in patterns of drinking following confinement, with more non-drinkers, less lighter drinkers and more heavier drinkers following lockdown. In the surveys, the changes occurred across all age groups, but increases in the proportion of heavy drinkers were less in the younger and older age groups, and more in the middle aged groups (35–64 years), dissimilar to our results where increases in purchases were greater in those aged 65 plus years than the younger age groups. In the surveys, there was a steep social grade gradient with greater increases in the proportion of heavy drinkers moving from social grade group E to group A, whereas we found a greater increase in purchases in social grade group E, compare with AB, but a much lower increases in purchases in social grade group D compared with group AB. In the surveys, increases in the proportion of heavy drinkers occurred across all regions of England, with the increases being higher in the north, rather than in the south of England, similar to our findings. Increases in purchases were much lower in Scotland and Wales than in England.

Data from the Office for National Statistics for England and Wales finds an increase in wholly attributable alcohol-related deaths during 2020 compared to previous years [9]. Compared with 2019, whilst deaths were 8.5% higher during the first quarter of 2020 (prior to confinement), deaths were 17.4% higher during the second quarter, 21.9% higher during the third quarter, and 28.3% during the fourth quarter. There were variations by region, with the increase in deaths being higher in the north as opposed to the south of England, mirroring the changes in purchases that we found. Further, deaths in Wales decreased during the last three quarters of 2020. In England, there was no huge difference in changes in deaths by deprivation quintile. For men, during 2020, the increases in the most deprived quintile were 4.2 times greater than in the least deprived quintile, compared to a 3.8-fold difference in 2019. For women, the respective differences were 3.0 in 2020 and 3.2 in 2019.

Putting all of this together would suggest that heavier and more socially disadvantaged drinkers were disproportionately affected by drinking more alcohol following COVID-19 lockdown than lighter and less socially disadvantaged drinkers. This would imply that during such times, there is a need to strengthen implementation of alcohol policy measures to reduce the harm done by alcohol, such as those put forward by the WHO SAFER initiative [26], as well as strengthen structural policies that help to improve all people’s socioeconomic prospects, aligned with approaches that address the social determinants of health [27, 28].

Our analyses have several important strengths. We obtained product bar code data from a large number of households, with a large number of daily data points before and after the analyzed event (the introduction of COVID-19 lockdown). We undertook controlled interrupted time series analyses, using purchases over the years 2015 to 2019 as time controls for 2020, subtracting the differences between the respective time periods for our analyses. The use of time controls helps to control for any confounding events that would affect both time periods, such as limitations related to data collection.

As we have noted in previous publications, analyses of such household purchase data have some limitations [10, 29–32]. A key limitation of our study is that, except for the purchases during the period of COVID-19 confinement (between 23^rd March and 4^th July 2020, when on-licensed premises were closed, with, in principle, all legal alcohol purchases captured), we only measure off-trade alcohol purchases and not on-trade purchases. The data also has limitations, with heavy drinkers tending to be under-represented in household panel data [33, 34], with alcohol purchases tending to be under-reported in these datasets [35, 36]. It may also be the case that such under-recording of alcohol is higher among households purchasing the highest levels of alcohol. Additionally, we are only able to assess changes in off-trade alcohol purchases as opposed to actual levels of alcohol consumption for these time periods. Adults in a household may not have an equal share of the alcohol purchased, and not all adults in a household may be drinkers of alcohol.

Conclusions

There is a long-standing and well characterized alcohol harm paradox where alcohol-related harms are disproportionally experienced by people living in lower socio-economic status groups despite reported heavy drinking across all socio-economic status groups. Here we see a new paradox in which British data suggest no increases in overall sales and purchase of alcohol following COVID-19 lockdowns, yet survey and mortality data suggesting otherwise is due to differential changes by population sub-groups. The top one fifth of purchasing households (by how much they normally purchased) increased their purchases 17 times more than the bottom one fifth of purchasing households. The most disadvantaged households increased their purchases more than the least disadvantaged. Further, households in the north of England increased their purchases more than households in the south of England, mirroring changes in alcohol-specific death rates during 2020 compared with previous years. This suggests that alcohol policy to reduce high consumption of alcohol becomes more important during extraordinary times, such as COVID lockdowns. That the increase in purchases was much less pronounced in Scotland and Wales, compared with England, could be attributed to the minimum unit price of alcohol introduced in both jurisdictions, such a policy shown to reduce alcohol purchases, particularly amongst the heaviest purchasing households during times of lockdown [30].

Acknowledgments

We thank Kantar Worldpanel for providing the raw data and reviewing the method description as it describes the purchase data. Professor Kaner is a National institute for Health Research (NIHR) Senior Investigator, and Director of the NIHR Applied Research Collaboration, North East and North Cumbria. Dr. O’Donnell is a National Institute for Health Research (NIHR) Advanced Fellow. The views expressed in this article are those of the authors and not necessarily those of NIHR, or the Department for Health and Social Care.