Independent report

Chapter 8: non-pharmaceutical interventions

Updated 10 January 2023

Introduction

Non-pharmaceutical interventions (NPIs), also known as ‘public health and social measures’, referred in COVID-19 to the measures to reduce transmission that did not depend on drugs, vaccines or other specific medical countermeasures. The aim throughout this pandemic, as with previous pandemics and major epidemics, is through science to get to medical countermeasures as soon as possible. Inevitably there was a period at the start of the pandemic when medical countermeasures were not available and almost all of the actions to blunt the effect of the pandemic had to be NPIs. This is hardly new – NPIs of some form have been used in almost all pandemics, from case isolation and contact quarantine during plagues in medieval Europe to public advice on safe sex in the HIV pandemic. NPIs had also been a standard part of pandemic planning since 2004, but they were not needed at scale in the 2009 H1N1 influenza pandemic. This pandemic was the first time in living memory that NPIs were used so extensively and at such scale in the UK. As medical countermeasures came on stream the relative contribution of NPIs decreased, but this was a gradual process.

Many NPIs, including hygiene measures, isolation and quarantine, can be a part of routine control of infectious disease outbreaks. A wide variety of measures sit under the term ‘NPIs’, and these can be grouped into:

• measures for individual protection from acquiring the infection from someone who unwittingly is infectious, including:

  • social (physical) distancing
  • surface cleaning
  • face coverings
  • encouraging meeting outdoors (for droplet or aerosol spread)
  • handwashing (for fomite spread)

• measures by an individual to reduce the probability that if they are unwittingly infectious they pass it on to someone else, including:

  • face coverings and masks
  • social (physical) distancing
  • handwashing
  • cough etiquette
• measures to identify (or self-identify) people who are infectious and get them to isolate until they are minimally infectious – the introduction of widespread testing was important in achieving this (see Chapter 6: testing), as was contact tracing (see Chapter 7: contact tracing and isolation)
• measures to limit the number of households that come into contact, and thereby reduce the chains of transmission. These include closing or limiting places where large numbers of people from different households come together in one place, especially if those places are crowded indoors or with poor ventilation. Examples in this pandemic have included restrictions on the hospitality sector, schools and public transport, and working from home. Full lockdown was the most extensive end of this but there was a spectrum
• measures to provide additional protection to the most vulnerable – which in this pandemic was initially termed ‘shielding’ – so that they are at lower risk of acquiring COVID-19 than the general population
• measures to reduce transmission in high-risk environments
• travel restrictions to prevent or slow the importation of cases, particularly in the early stages of the pandemic, and then to slow the importation of concerning new variants

These each varied in their effectiveness, difficulty, evidence base and negative social, public health and economic consequences. At all times in the pandemic to date, medical and science advice, as well as political decision-making, had to recognise the balance between the harms of not undertaking measures and the potential harms caused by these measures.

A central point that needed to be made repeatedly was that, in having to rely on NPIs in whole or part for the first 2 years of the pandemic, it was always a matter of the least bad option, not a ‘good’ one. In contrast to medical countermeasures where there are long-standing structures and processes to measure the benefits (protection from or treatment of disease) and risks (side effects) using clinical trials, many of the disbenefits of NPIs were in broader social, societal, educational or economic terms that were often harder to measure and required wider technical advice beyond the remit of the Scientific Advisory Group for Emergencies (SAGE) or the CMOs.

It was always the likelihood that medical countermeasures, the products of science, would increasingly come to bear, so NPIs were used to hold the line until that point. It was however very unclear at the start of the pandemic which medical countermeasures might be gained (vaccines, for example, had no role in controlling HIV – drugs did) or how long they would take. Conveying to policymakers and the public that reliance on NPIs was not indefinite, but that it was not possible to put a time on how long medical countermeasures would take to be deployed, was a challenge in the early stages of the pandemic.

The need for extensive use of NPIs was greatest early in this pandemic, when the population was all immunologically naïve (first wave) or mainly immunologically naïve (second wave), much was unknown about the virus, spread was rapid and mortality and morbidity were both high. The relative importance of NPIs in controlling the pandemic decreased as the availability of effective drugs and vaccines, alongside steadily increasing hybrid immunity from vaccination and infection, reduced both transmission and severe disease for the majority of people. We would expect a broadly similar pattern in a future pandemic, though delays in drugs or vaccines being available, or the emergence of a variant with greater transmissibility, vaccine escape or leading to more severe disease, could result in longer deployment of NPIs.

The next section sets out some of the key NPIs used in this pandemic, some considerations in their implementation and emerging evidence arising from their use – though that evidence has a number of limitations. The list of NPIs set out here is not exhaustive, nor are the references to emerging evidence. SAGE documents hold much greater detail on many of these NPIs, and further academic studies in the coming years will continue to increase our understanding of the effects of NPIs, both in controlling COVID-19 and their wider consequences – so we do not consider this the final point of knowledge.^{[footnote 1]}

There were some differences in the timing, extent and delivery of NPIs across the UK. However, here we offer a cross-UK view for clarity. There were also many further examples of NPIs used in other countries during this pandemic which will be of interest to future CMOs and GCSAs. The science of COVID-19 was a global science, and on many of these NPIs the scientific consensus shifted globally over the course of the pandemic as the epidemiology of the pandemic (such as seroprevalence or variants), its impact and responses to it varied across different countries over time. While changes to scientific consensus demonstrated growing scientific understanding of the pandemic, they could pose challenges in public communications when messages changed over time.

Finally, we should acknowledge in this section the remarkable response of society to the advice to use NPIs. The success of NPIs depended on people from all parts of society acting together to protect the most vulnerable, often at significant disadvantage to themselves. The fact they did, near universally, over prolonged periods – particularly in areas of enduring transmission – is one of the most important lessons of this pandemic, and an extraordinary tribute to a widespread sense of community across society.

Important considerations in deploying NPIs

A number of factors influenced choice of NPIs in this pandemic, including the predominant respiratory route of transmission, the mortality rate, and the distribution of mortality and morbidity across the population with the greatest risks in the elderly but much lower in children. The same NPIs used in other recent epidemics and pandemics, most obviously HIV (sexually transmitted), but also Ebola virus (touch transmitted) or Zika virus (vector transmitted) would have been largely ineffective.

In the early part of the pandemic, the force of transmission was such that extensive use of multiple NPIs used together was needed to get the reproduction number (R) below 1. In reviewing different combinations of NPIs to achieve this, there were some important considerations.

First, the ratio of harms and benefits looked different for individual NPIs. For example:

• hand washing has few downsides but was unlikely to be sufficient to bring community transmission of COVID-19 down significantly
• mass closure of settings may contribute substantially to bringing down community transmission but at a significant societal cost

Second, the appropriateness of NPIs was different for different groups – for example, ‘shielding’ advice for those with high clinical vulnerability. Similarly, differing local circumstances as well as varying levels of population movement across different areas were important when considering localised interventions – this is covered in more detail below under ‘Local tiers or levels’.

Third, the blend of NPIs chosen was important as different interventions can mitigate the risk of exposure via different routes (such as close-range droplet versus longer range aerosol transmission). Although modelling was used to help determine appropriate bundles of interventions, this was complicated by the fact that individual NPIs were not additive but interacting. For example:

• widespread working from home also impacted travel patterns, social contacts and hospitality use
• school closures reduced mixing of children, but also of parents

Some NPI packages may also have had impacts beyond the sum of their parts. Many NPIs used in this pandemic were also not mutually exclusive, such as regulations or guidance to work from home and closure of settings (most of which were workplaces). This had implications for communication and implementation of NPIs, public interpretation and acceptance, and also for generating and analysing evidence on the impacts of individual NPIs or NPI packages. Attempts to separate the effectiveness of individual NPIs were therefore both difficult and potentially misleading, as NPIs will likely always be implemented in packages and in a particular epidemiological context.

There were also wider considerations in the deployment of NPIs. First, there is always a need for societal consent for NPIs, especially the most potentially damaging ones. There is also a major potential for NPIs to create or exacerbate inequalities and have widespread impacts across society in health, economic and social terms. Decisions on whether and how to implement such wide-ranging interventions go well beyond health and rightly sit with elected ministers on behalf of society. Evidence from observation and behavioural sciences shows that major interventions like NPIs must be felt by the public to be fair, and suspicions that some and not others were following rules was damaging to adherence. It was also important to work closely with local areas and with different communities to ensure NPIs were feasible and appropriately communicated, to understand how they were being interpreted in practice and to understand any barriers to adherence. This was not always in place at the outset of the pandemic and is an important consideration for future pandemics.

Different types of harms were considered from early in the pandemic by scientific advisers and CMOs and they were discussed publicly early in the pandemic.^{[footnote 2]} CMO advice from March 2020 was that excess mortality would come from a number of causes:

1. The most obvious is direct mortality from people dying of the virus despite best medical care.

2. A second major indirect cause of mortality is from the NHS emergency services being overwhelmed and therefore providing significantly less effective care both for those with coronavirus and for those with other medical emergencies.

3. A third cause of mortality and more commonly increased ill health will be the postponement of important but non-urgent medical care and public health programs such as screening while the NHS is diverting resources to manage the epidemic.

4. There is a strong correlation between economic disadvantage and ill health and in the long term any prolonged increase in poverty due to our countermeasures will feed through to poor physical and mental health outcomes.

Alongside this, the impact of the pandemic on disparities was an important concern. The Welsh Technical Advisory Cell for COVID-19 added this formally as a ‘fifth harm’.^{[footnote 3]} (For more detail see Chapter 2: disparities.)

A joint report by the Department of Health and Social Care (DHSC), the Office for National Statistics (ONS), Government Actuary’s Department and Home Office, commissioned for SAGE, has a more detailed working-through of COVID-19 direct and indirect impacts on excess deaths and morbidity.^{[footnote 4]} Political decision-makers had to consider all types of harm, including non-health harms and those that were harder to measure or assess.

Considerations when interpreting the evidence

In this chapter, we reference some emerging evidence on the effectiveness and wider impacts of NPIs in this pandemic, but there are several important caveats to this evidence that we would like to highlight here. There are also gaps in the evidence base on NPIs, which we expect will continue to evolve in the coming years.

As noted above, NPIs were implemented in packages and have complex combined effects, which complicates interpretation of the evidence base. It is not a simple additive calculation to understand their combined impact on transmission or indeed their wider impacts. It may never be possible fully to disentangle some of the effects of individual NPIs in this pandemic as many were used together.

Comparisons of NPI impacts over time and across the world were complicated by varying applications of NPIs, varying definitions, varying populations and changing epidemiology and immunity. The effectiveness of NPIs in reducing transmission was influenced by how interventions were implemented, communicated and interpreted, when and where they were used, how (and how far) they were adhered to, and changes in transmission dynamics due to pathogen or behavioural changes – all of which varied across populations and over time. Comparisons across studies is challenging due to this. It was important to have careful critical analysis of systematic reviews bringing together different studies, to interpret these with caution, and to have diverse scientific and professional experts reviewing the literature.

It was often not feasible to unpick the impacts of NPI policies from wider behavioural changes, and this demonstrates the importance of embedding behavioural studies when planning, implementing and evaluating NPIs. NPIs were usually introduced in response to escalating risk, and in this pandemic behaviours also changed in response to escalating risk and often prior to the formal implementation of the NPIs. For example, in a sample of UK adults on 17 to 18 March 2020, 45% reported they had stopped attending bars, pubs and restaurants, 27% had reported stopping seeing family and 30% stopping seeing friends entirely – all in advance of formal measures to limit such activities.^{[footnote 5]} This made measurement of the independent impact of government measures on transmission difficult to delineate.

For all NPIs there were potential gaps between the theoretical maximum reduction in transmission and that observed in practice due to, for example:

• the design and implementation of NPIs
• the effectiveness of communication around NPIs
• whether and how measures were enforced
• the level of support for behaviour change particularly for those with fewer physical and financial resources to adhere to NPIs

These factors can contribute to incomplete adherence, imperfect adherence (such as incorrect face covering wearing), and lower levels of people continuing to follow public health guidance even when not directed by government. Again, this underlines the importance of embedding behavioural and social sciences to inform modelling on the impacts of different NPI packages, which became more sophisticated as the pandemic moved on (for example, with the 2021 roadmap), as well as informing delivery of services.^{[footnote 6]}

Related to this, it was important to be clear where evidence and advice was about individual-level or societal-level NPI effects on transmission, and to consider the difficulties reading across from one level to the other. For example, although face coverings may have modest impact on risk of transmission for a given individual, at a societal level (and particularly with a highly transmissible pathogen) the benefits may be considerably greater.^{[footnote 7]}

Many research methods were not feasible for population-wide NPIs in an emergency – for example, randomised controlled trials. It was considered neither ethical nor feasible to, for example, randomise shielding and associated social support to only a selection of those deemed clinically vulnerable. It is noteworthy than even for face coverings, which may be the easiest NPI to assess, studies have not been done in a way that provides as clear an answer as medical countermeasures.

Observational studies on NPIs were often complicated by several potential confounders, such as changes in the availability and accessibility of testing alongside changes in behaviours and NPI implementation. Many NPIs were introduced and removed contemporaneously and alongside changes such as:

• variants influencing the force of transmission
• population immunity (vaccine-derived, through infection and hybrid) rising and waning
• methods for implementation and communication around NPIs adjusting
• individuals’ perception of risk and approach to risk management changing

This makes it extremely challenging to attribute causal impact to individual NPIs.

Some NPIs had a ‘critical mass’ of usage below which their impact on transmission may have been negligible but above which they could be an important tool in reducing transmission. The extent to which mobile phone contact tracing apps could alert potential contacts of cases, for example, was contingent on the proportion of the population downloading and using the app, testing, and reporting their result. Determining this critical mass was more straightforward for some NPIs than others.

When considering the impacts of NPIs, it was important continually to bear in mind the possible counterfactual they enabled us to avoid. For example, closure of certain settings had important economic, societal and indeed health impacts – but unmitigated transmission, too, would likely have had major significant harmful impacts across these domains.

Finally, the full effect of many NPIs cannot yet be investigated, because impacts may take many years to become apparent or because they have affected behaviour which has not yet returned to pre-pandemic levels (such as working from home patterns).

Summary of NPIs used in the pandemic with some emerging evidence and key principles

Many of the NPIs used in this pandemic were already well known to the general public when managing the spread of respiratory infections:

• hand washing
• covering coughs and sneezes
• cleaning surfaces

Others were deployed as a regular part of infection control and prevention practice (in health and social care settings in particular) but were perhaps less universally used by the UK public, such as face masks and physical distancing. Others were used for the first time at scale in the UK in living memory during this pandemic:

• limits on group sizes and activities to reduce contacts
• working from home orders
• closure of selected settings and events
• domestic and international travel restrictions
• shielding of the most clinically vulnerable
• lockdowns – perhaps the term that will come to be most associated with this pandemic

Some of these were centuries-old tools to manage epidemics and pandemics across the world, though in this pandemic there were adaptations to these based on the technology available – such as using a mobile phone app to alert individuals when they had been within 2 metres of a case.

In the UK the definition of a ‘contact’ was based on the combination of time and proximity that was judged to be highest risk for exposure and whether the contact was indoors or outdoors, and this was reviewed as the evidence base developed (for example, on transmission dynamics). The risk associated with contacts in the community changed as case rates rose and fell, population immunity increased, new variants arose and medical countermeasures became available.

NPIs in care homes, educational settings and healthcare settings are addressed in chapters 8.1 on educational settings, 8.2 on care homes and 10 on improvements in care respectively. For more details on testing see Chapter 6 and on contract tracing see Chapter 7.

Measures to reduce risk within interactions

Hand and environmental cleaning

Guidance on the frequency, duration and technique of hand and environmental cleaning to reduce fomite transmission was available in the UK from early in the pandemic using existing infection prevention and control guidance.^{[footnote 8]} As the pandemic progressed, and in particular as public settings reopened after the first lockdown, there were further measures to widen access to cleaning facilities such as hand sanitiser in public spaces and sprays to clean common touch surfaces.

Scientific consensus on the relative importance of hand and environmental hygiene shifted throughout the pandemic as evidence developed indicating more limited viability of virus in the environment than initially suspected, and strengthening evidence for the proportionately more significant role of airborne as opposed to droplet transmission.^{[footnote 9]}, ^{[footnote 10]}, ^{[footnote 11]} Of course, the likelihood of transfer is greatest the shortest amount of time since a surface has been touched, and so hand cleaning was generally more important than environmental cleaning. Hand and environmental cleaning no doubt played a role in reducing transmission risk across a range of settings, but it is important to remember that transmission in these settings also depended on proximity, types of contact and other mitigating measures (such as ventilation).

Nevertheless, hand and environmental hygiene advice has remained in place, not least because it has had the additional benefit of reducing transmission for some other infectious agents and, besides, is part of routine advice to many settings. Importantly, this measure also had almost no downsides except the impact of regular cleaning on operations (for example, in schools and businesses), some costs (such as installing basins or buying sanitiser), and on people for whom regular hand washing aggravated skin conditions. It was also a relatively straightforward intervention to implement, though the capacity to make facilities available for cleaning varied across different settings, areas and communities.

Social (physical) distancing

Physical distancing was identified as an important tool to reduce transmission early in the pandemic, and there is now substantial evidence in favour of its use to support control of SARS-CoV-2 transmission.^{[footnote 12]}, ^{[footnote 13]} The relative importance attributed to physical distancing shifted as evidence strengthened on the role of airborne as opposed to fomite transmission. However, widespread reduction in physical proximity of community contacts was important in interrupting both routes and in fact physical distancing had its greatest impact on reducing respiratory droplet transmission.

There was some global variation in who was advised to distance, when, where and to what degree. In the UK physical distancing was to 2 metres (or one metre with additional measures). Targeting of physical distancing to different groups or settings also adjusted throughout the pandemic as control strategies were updated in response to the changing situation. When the risk of severe disease from infection was higher (before vaccine and naturally derived immunity reached high levels) and there was an urgent need to reduce overall community transmission, physical distancing was deployed as a society-wide measure with all non-household physical encounters requiring distancing. When the aim was to manage risk in the community in a more targeted way as vaccines reduced the risk of severe disease from infection, physical distancing was focused on higher risk settings, situations and individuals, and became advisory.

There were some wider considerations when deploying physical distancing as an NPI that our experience in this pandemic has underlined:

• the variable feasibility of distancing in different settings and situations – for example, in personal care services, and its economic impacts
• the variable impacts of physical distancing – for example, on young children in childcare for whom physical distancing may have an important developmental impact or on those with dementia
• the potential need for widespread detailed guidance and support to implement distancing – for example, in workspaces using one-way routes or in hospitality with adjustments to seating
• the importance of environment and surrounding behaviours in the effectiveness and role of physical distancing – for example, whether contact is indoor or outdoor, in a crowded setting or taking place for a long time
• the interactions between distancing and other NPIs^{[footnote 14]}

Face coverings in the community

This section refers to cloth masks as ‘face coverings’ to distinguish them from medical grade face masks. Personal protective equipment (PPE) in health and social care is covered in more detail in Chapter 10: improvements in care of COVID-19.

Although some countries, especially in East Asia, promoted widespread use of face coverings or masks from an early stage, the global and UK scientific consensus on the appropriateness of face coverings or masks for preventing transmission evolved during the early stages of the pandemic. In April 2020 SAGE advised that on balance there were benefits in widespread use of face coverings, though as the country was under a national lockdown at the time this was unlikely to be instrumental in reducing community transmission.^{[footnote 15]} In the same month, interim World Health Organization (WHO) guidance advised against the use of face masks for healthy (uninfected) people in community settings. However, as evidence on the routes of transmission and the effectiveness of face masks evolved, this was updated in June 2020 to recommend their use in the community.^{[footnote 16]}, ^{[footnote 17]} In late July 2020, as the national lockdown in the UK gradually lifted, face coverings became mandatory in a range of public settings across the UK, such as on public transport or in shops (though this differed slightly across the UK nations).^{[footnote 18]} There was variation in enforcement across the UK’s 4 nations and across different settings that needed to adapt rapidly to a number of new requirements.

Evolving recommendations on face covering or mask use in the community – from the WHO, the UK government and other governments worldwide – were at times difficult to communicate. They were, however, a reflection of a developing evidence base and also of operational realities at different stages of the pandemic and the need to continually balance multiple risks. At the outset of the pandemic, for example, demand for face masks globally was extremely high and there was concern that widespread use of medical-grade face masks in settings where they were thought to have marginal or no effect would impact supply lines for health and social care professionals who were in close contact with infectious and vulnerable people.^{[footnote 19]}

The type of face covering was not mandated in the UK outside healthcare settings, and there was widespread use of cloth face coverings by the public. There was some evidence outlining differences in effectiveness across different types of face covering (which we do not cover here). Alongside this, face covering quality and correct wearing were both important. However, in the context of high case rates and a proportion of asymptomatic and pre-symptomatic transmission, logic follows that it is more important to have more people wearing some form of effective face covering correctly rather than fewer wearing high-grade respirators. Feasibility of implementation was important – face coverings were relatively cheap and widely available (once global shortages had cleared) and relatively straightforward to implement with public guidance.

In general, face coverings were advised or mandated in the UK during periods and areas of high transmission and in higher risk settings or situations where distancing and sufficient ventilation were not feasible. In contrast to some countries, they were never recommended outdoors except in very crowded environments. Their purpose has primarily been as source control, with some protection to uninfected wearers – however, in reality adherence varies across different settings, situations and individuals, and studies still give widely varying estimates of their impact on transmission.^{[footnote 20]} Widespread face covering use had some potential impacts on social and educational interactions, such as for younger children, those with dementia and those who rely on facial expression or lip reading for communication. Some groups were exempted from guidance to use face coverings.^{[footnote 21]} There were strong, and opposing, views on the best approach to face coverings in public alongside scientific discourse on the topic.

Ventilation

As evidence accumulated on the importance of airborne transmission for SARS-CoV-2 (see Chapter 1: understanding the pathogen), ventilation was increasingly advised as an important measure to reduce risk of transmission.^{[footnote 22]} In the UK, ventilation was encouraged with guidance for people to go outside or open windows, which was relatively simple and lower cost than many other interventions, although it was not feasible in all situations (for example, in large buildings without individual control over ventilation). This was clearly harder to adhere to in the winter months. The primary purpose was to dilute any airborne viable SARS-CoV-2, and by doing so effectively reduce the range of potential transmission from a given individual, so types of activities within settings also had to be considered.^{[footnote 23]}

More extensive interventions were also deployed, such as funding to put ventilating measures like high efficiency particulate air filters or carbon dioxide monitors in place. Buildings can also be fitted with passive ventilation systems, but this requires expert input, quality standards and training as well as capital investment which would be considerable if implementing across society. There was also a need to balance ventilation needs with energy efficiency and heat retention in buildings and the impact of noise, air pollution and security on ventilation behaviours were important.^{[footnote 24]}

Measures to restrict personal contacts

Case isolation

As is the case for many infectious diseases, case isolation was one of the first and most important tools for controlling SARS-CoV-2 transmission and was implemented in the UK from the beginning of the pandemic when only a handful of cases were in the UK and in advance of population-wide NPIs. It has remained in place throughout, with varying levels of enforcement and guidance.

The effectiveness of case isolation is reliant on the speed and completeness of case identification, the speed and effectiveness of guidance to cases instructing isolation, and finally the extent of adherence to isolation by cases. The speed, accuracy and completeness of case identification in this pandemic was reliant on available testing, due to the relatively generic symptoms of COVID-19 and the existence of asymptomatic and pauci-symptomatic infection (see Chapter 6: testing). It was also reliant on contact tracing to identify further potential cases (see Chapter 7: contact tracing). Case identification was less complete at points when testing demand outstripped supply or when contact tracing was not performed swiftly enough to identify further cases before they began to transmit SARS-CoV-2 to others. It was important throughout to have accurate case identification so that only true cases and contacts were asked to isolate to avoid unnecessary disruption, and for the public to trust that this was the case.

Levels of adherence to isolation no doubt varied across individuals and over time, though it was difficult to read clear signals from surveys of the public on this issue due to obvious social pressures to answer positively and resulting possible response biases. Ability to isolate also varied across individuals, and so financial and practical support for isolation was provided during this pandemic. In some countries, accommodation away from the home was also provided for cases to isolate, but in the UK the scale of case numbers in the community from early spring 2020 onwards rendered this challenging to implement for all case isolation. Accommodation was however provided for quarantining inbound travellers early in the pandemic, and for homeless people – in the ‘Everybody In’ initiative – an unprecedented and important measure to protect this highly vulnerable population.

Timely and complete adherence to case isolation can go some way to avoiding or delaying the need for population-wide measures like lockdown, but realistically adherence will never be 100%. There may come a point in any pandemic where case rates have reached such a level that population-wide measures to limit contacts are needed to reduce community transmission. Some of those used in this pandemic are outlined below.

Working from home, closure of specific settings and closure of public events

These 3 interventions are addressed together here because they overlapped considerably: events and settings were also workplaces, and events often took place within settings. Transmission within households was a considerable risk that was difficult to mitigate, and therefore limiting out-of-household contacts through these measures was important not only for the individual taking part in an activity, but also in protecting their household contacts. In addition, restrictions on these activities reduced a series of related contacts, such as travelling to settings or events, or social contacts linked to these activities (such as visiting a restaurant before an event). Closure of outdoor events, for example, was probably more important for restricting associated activities such as travel or congregating in indoor hospitality venues than for limiting outdoor contacts within the event itself.

The impacts of large parts of society working from home can be considerable, reducing the number of contacts not only within the workplace but also in other public settings such as public transport and hospitality. In late 2020 SAGE noted that around a third of contacts were linked to work and its associated activities, and in late 2021 it assessed that reintroduction of working from home guidance may have the largest impact on transmission out of proposed measures (such as certification and face coverings).^{[footnote 25]}, ^{[footnote 26]} One global systematic review of NPI studies found that workplace closure was associated with a reduction in transmission in 12 out of 14 studies – though it was difficult to distinguish within these studies (and generally when reviewing evidence on this measure) where working from home was distinct from settings closures.^{[footnote 27]}

Working from home measures differed slightly across the UK over the course of the pandemic, and there were changes over time in whether working from home was mandated or advised. For large parts of the pandemic, guidance for employers and individuals was that those able to work from home should do so – and this had implications for disparities.

First, because patterns of workplace attendance varied across the country and also linked to existing disparities. Analysis of the DHSC tracker survey between January and February 2021 found that non-essential workplace attendance was significantly independently associated with a range of socio-demographic variables and personal circumstances. Financial hardship, lower socio-economic status, having a dependent child at home and working in certain key sectors were associated with higher likelihood of workplace attendance.^{[footnote 28]}

The second reason was that only around half of the working population were likely to be able to work from home at any given point.^{[footnote 29]} Those unable to work from home continued to face risk of infection when attending work, though the measure itself acted to reduce overall transmission so there remained overall benefit to this group from the intervention in terms of SARS-CoV-2 transmission. Those working from home for extended periods, meanwhile, may have faced mental and physical health impacts such as musculoskeletal issues from altered working patterns. It was, however, challenging to assess the impacts of working from home as many have not returned to pre-pandemic working patterns and there were several potential confounders to observational approaches, such as differing practices across different workforce groups or demographic differences across different professions (for example, the hospitality sector tending to have a relatively young workforce).^{[footnote 30]}

Closure of specific settings (such as hospitality, non-essential retail, personal care, leisure settings or places of worship) was also implemented slightly differently across the UK throughout the pandemic. Some of this was a national question and some differed by locality (see: ‘Local tiers or levels’ below); regardless, local complexities in communication and implementation needed to be considered. The combined impact of settings closures has been easier to determine than the contribution of individual setting types: a SAGE paper in September 2020 concluded that the closure of bars, pubs, cafés and restaurants together had a moderate impact on COVID-19 transmission overall (medium confidence), while closure of close-contact personal services (such as hairdressing) may have had a lower impact on transmission (low confidence).^{[footnote 31]} The effects of all the closures on transmission was, however, cumulative and it made logical sense to start with settings where large numbers of households met together indoors, often with limited ventilation in close proximity. The relative impact of settings closures on transmission also depended on mitigations in place while open (such as ventilation and distancing). Essential shops and schools were likely to become more important relatively as sites for transmission as other places where households mixed indoors closed. It was important to consider a range of factors in determining risks associated with different settings including the layout of the setting and physical features (such as ventilation), activities taking place within settings, who would be mixing within settings, at what proximity and how often (see Chapter 1, section 9: What were higher risk settings of transmission for SARS-CoV-2?).

The evidence base was complicated, however, by changes in behaviours throughout the pandemic in response to changing policies and restrictions, and to wider factors such as seasons and holidays. There was therefore a need to consider these NPIs as complex behavioural interventions with many barriers and enablers. A range of studies and information was needed to unpick this, such as qualitative studies to understand the drivers of changing behaviours, local intelligence on likely higher risk settings to manage individual outbreaks, and monitoring of contact patterns – for example, through the CoMix and COVID-19 Rapid Survey of Adherence to Interventions and Responses (CORSAIR) studies.^{[footnote 32]}, ^{[footnote 33]}

As with working from home, there were important considerations for disparities associated with settings closures. Some groups had relied on particular settings for social or economic support more than others and settings closures and their associated loss of work can disproportionately impact those in insecure work, with little financial security, or working in sectors more directly impacted such as hospitality or retail. Again, continual monitoring of impacts alongside links to local communities helped signal where some disparities were arising.

Finally, there were limits on the numbers and settings for gatherings and public events across the UK to reduce mixing in the community and drive down transmission. There was some variation in this across the 4 nations throughout the pandemic, though importantly some related policies such as business support were set by the UK government. Greater understanding of the range and modes of transmission and the dispersion parameter, alongside epidemiological studies on outbreaks, has supported a growing understanding of the potential role of gatherings and events in outbreaks of SARS-CoV-2.^{[footnote 34]}, ^{[footnote 35]} In June 2020, for example, SAGE noted the strong evidence for super-spreader events, and in August 2020 the SAGE Environmental Modelling Group and Public Health England (PHE, subsequently the UK Health Security Agency (UKHSA)) produced a joint review of evidence that singing and shouting were associated with transmission of SARS-CoV-2. ^{[footnote 36]}, ^{[footnote 37]} This type of evidence generation gave greater clarity on some potential risks for activity types, settings and participant numbers. However, transmission dynamics vary by event, setting, attendees and their relationships, behaviours within an event or setting and background epidemiology (such as dominant variants and community case rates), and so assessing the risk of given events, settings or activities is highly complex. Proximity of relationship with other attendees at an event, for example, may be as important as setting or size of the event – one meta-analysis found that contacts at social events with family and friends were higher than those for casual contacts.^{[footnote 38]}

The type and setting of gatherings and events were important factors in the extent to which they enabled transmission, as well as possible mitigations in place such as testing before and after events, ventilation and face coverings. The UK Events Research Programme, which examined the risk of COVID-19 transmission from attendance at events and interventions to reduce that risk, pointed out the importance of these measures to limit transmission. Low testing adherence following its pilot events, however, limited its ability to reach firm conclusions on how far particular mitigations at events impacted transmission.^{[footnote 39]}

There was a particular concern about the impact of limiting social mixing at one-off life events where the timing was not movable, in particular end-of-life meetings and funerals. The family and social importance of these is considerable but they also often involved elderly or medically vulnerable people mixing. Getting the balance right here was extremely difficult for policymakers.

Lockdown

This was the most intensive measure taken to reduce spread of COVID-19 and was highly effective even in the face of more transmissible variants. This pandemic was the first time in living memory that lockdowns were implemented across so many countries worldwide, so extensively and for such a long period of time. Variations of them were, however, well documented throughout history, and the principles behind lockdowns follow the same epidemiological logic as settings or events closures, working from home and limits to contacts. Definitions and implementation varied worldwide and throughout the pandemic, but broadly lockdowns consisted of:

• travel restrictions
• closure of all non-essential settings
• stay-at-home orders

Lockdowns were highly effective in reducing transmission of SARS-CoV-2. SAGE concluded that the lockdown introduced in March 2020 was associated with a reduction in the reproduction number (R) from an estimated range of 2.5 to 3.0 to an estimated range of 0.5 to 0.7 – though with an initial period of continued high case rates due to ongoing household transmission.^{[footnote 40]} This was due to high adherence and significant sacrifices by the public who went to great efforts to follow guidance and protect one another from exposure.^{[footnote 41]}

There was a range of possible wider impacts arising from lockdown, such as on mental health, levels of physical activity and levels of domestic abuse and safeguarding concerns. These are explored below, but this overview is neither complete nor exhaustive, nor is it possible to say whether these associations were certainly with lockdown as opposed to with the wider conditions of the pandemic. There will also be further impacts that were not measured or have not yet been fully realised.

On mental health, an analysis of the UK Households Longitudinal Study during the first lockdown found that general psychological distress increased substantially from 19% (95% confidence interval 18% to 20%) to 30% (95% confidence interval 29% to 32%). This could be partly the influence of the pandemic itself. However, the most significant decline was in ‘enjoyment of day-to-day activities’. Symptoms of poor concentration, poor sleep and loss of purpose were also cited, and loneliness in young people in particular increased during the first lockdown.^{[footnote 42]}

On physical activity, there were likely variable effects, with some forms of activity (such as walking and cycling) potentially increasing while team sports and activities taking place at (closed) venues reduced sharply during lockdowns. Official road traffic statistics, for example, show a marked increase in cycling in 2020 compared to 2019.^{[footnote 43]} A PHE study showed a decrease in the average duration of strength and balance activity for older people from 126 to 77 minutes per week in March to May 2020 compared to the corresponding period in 2019.^{[footnote 44]} This may have been a result of lockdown, shielding or voluntary precautionary behaviour. Reductions in activity will have had a number of associated health impacts, particularly for those with existing health needs such as musculoskeletal conditions.^{[footnote 45]} Importantly, physical activity outdoors was permitted during the UK’s lockdowns, which was not the case in many comparable countries (in which a lockdown meant staying indoors without leaving the home at all).

On domestic abuse and children’s safeguarding, concerning increases were seen in presentations after the first lockdown, such as suspected abusive head trauma in children, which may suggest delayed reporting of concerns due to settings closures and confinement to the home.^{[footnote 46]} Authors of several studies have cautioned, however, that attaining comparable data and attributing causes for changes in domestic violence patterns is highly complex.^{[footnote 47]}

The health and wellbeing impacts of economic changes during lockdown also bear consideration, and the literature linking health to macroeconomic changes is well established.^{[footnote 48]} However, it was difficult to unpick these from the wider impacts of the pandemic itself, or indeed to say whether these impacts would have been similar or worse had unmitigated spread taken place and people had proactively adjusted behaviours. There was evidence that people changed their behaviour in response to the pandemic even when not legally required to do so, as outlined above under ‘Considerations when interpreting the evidence’.

These impacts – whether from the pandemic or lockdown itself – were not evenly felt across society, and it is particularly difficult to quantify this due to variable data capture of key demographic characteristics such as ethnicity (see Chapter 2: disparities). Lockdown requirements were adapted to lessen these impacts as evidence on routes of transmission and relative risks of different activities evolved and as the introduction of vaccines reduced the risk of both transmission and severe disease. One example was a ‘bubbling’ policy which allowed contact between 2 households in specific circumstances. Another was loosening restrictions on access to green spaces and playgrounds to enable lower-risk social contact and encourage physical activity – an important protective factor for physical and mental health.^{[footnote 49]} There is a much wider discussion of these in SAGE minutes and papers.^{[footnote 50]}

It is important to note that we are using the term ‘lockdown’ to mean extensive social and economic closure across society, by law, usually including some degree of stay-at-home orders. In the later part of the pandemic there was some use in the media and elsewhere of the term ‘lockdown’ to mean a more limited set of NPIs.

Local ‘tiers’ or ‘levels’

Any NPI (from face covering guidance to lockdown) can theoretically be implemented at any geographical level. Graded NPI packages (known in England as ‘tiers’ and in Scotland as ‘levels’) were implemented at local or regional (rather than national) level during this pandemic, to adapt control measures to local circumstances.

It was difficult to quantify the impacts of local tiers or levels, both on transmission and beyond this on health and wellbeing, in part due to significant confounding where areas with similar populations and epidemiology entered the same level or tier at the same time, and in part due to difficulties in pinpointing the effects of localised NPIs where neighbouring areas with extensive travel links were in different tiers or levels. A scientific pandemic influenza modelling group (SPI-M) analysis on the impact of a 3-tier locally specified set of interventions in England, presented to SAGE in late 2020, found that:

• local authorities in the lowest tier continued to see epidemics growing
• some local authorities in the middle tier saw a reduction in their epidemics
• all local authorities in the highest tier saw decreased growth rates after introduction^{[footnote 51]}

There was, however, significant heterogeneity at this point, both across UK nations and between local authorities implementing measures – SAGE noted that most areas in the highest tier in England had additional restrictions above the minimum set for that tier.

There was an ongoing balance between minimising wider disruption by implementing targeted local policies, being effective enough to be meaningful, and avoiding ‘border’ problems between local tiers. Localisation of measures was complex, particularly where there was extensive travel between localities. Although routine local travel and contact patterns across different areas were recognised in public health advice on local tiers or levels, it was generally not possible to ‘lock down’ an area outside a pre-existing administrative boundary such as a county or district border so it was often difficult to specify tiers or levels to be relevant to local travel or contact patterns. Localised interventions may be pragmatic where there is considerable variation across local areas and minimal travel between them, but once transmission is widespread regional and national policies are often preferable.

There was also a need for decision-makers to consider political issues such as public perceptions of fairness when highly localised policies were implemented, and this influenced policies around tiers and levels. Some people in areas under heavier restrictions may have felt the impact on local economies was unfair, for example, while some of those under lighter restrictions may have felt they were not being properly protected. Of course, public and political support for any measure is an important element of its success or failure, and this can be particularly difficult at smaller geographies where people moved regularly between different tiers and changing rules and communications could rapidly become confusing. The introduction of multiple different levels or tiers nationally also meant increased demands for clarification and advice for local areas and stakeholders. These issues were important – but they also had to be balanced with epidemiological need. In some cases local concerns about increasing tiers or levels delayed decisions to such a point that the intervention was less effective by the time it was enacted.

Finally, there was a risk of widening disparities across different areas by localising restrictions. For example, areas with high levels of low-income manual employment were more likely to see people having to attend the workplace and therefore see higher case rates, which in turn triggered restrictions that further impacted the local economy. Of course, these considerations of fairness and disparities were equally important in deciding on national policies, particularly at times when prevalence in different areas varied widely.

Other measures

Travel restrictions

International travel restrictions were also introduced during this pandemic, and they included bans on travel to and from certain places, testing requirements (including the type of test required and timing of tests relative to travel) and quarantine (both directly supervised and advisory). There was some variation in travel restrictions across the UK, but as a broad summary:

• in January and February 2020, travel restrictions focused on specific countries (initially East Asian) judged to be high risk for case importations, with advice against travel for UK nationals and quarantine of arrivals from these countries^{[footnote 52]}
• in March 2020, as the UK entered a national lockdown, this shifted to advice against all non-essential travel worldwide as well as domestically
• as lockdowns lifted across the UK in summer 2020 and tests were available, advice adjusted to enable international travel in specified circumstances, and testing requirements were introduced for international travel, whether pre-departure or on arrival
• restrictions were updated throughout the pandemic (both across the UK and globally) as variants of concern were detected in different areas

Border issues combine epidemiological information with wider travel, trade and geopolitical considerations, which are rightly the preserve of elected political leaders. Here we focus on the epidemiology. The important epidemiological principles underlying advice on travel restrictions were:

• imported cases generally matter most when there is a very low level of domestic infection but higher rates elsewhere – for example, in delaying establishment at the very outset of the pandemic, which gives time to assess potential further countermeasures. Importantly, the extent of this delay is often days or at most weeks, rather than months
• when domestic transmission was very high, imported cases were such a small proportion of total infection burden that they made little significant difference to the epidemic
• when local incidence and prevalence reduced and imported cases were a higher proportion of total cases, preventing imported cases became more important again. This question was reconsidered for variants of concern, where the risk associated with importing an individual case changed according to which variant it was. The changing ratios of domestic to imported cases (including of variants) was a gradual process and there was not a set threshold below or above which travel restrictions were or were not important

Judging the appropriateness of travel restrictions in supporting pandemic control depended on the respective rates of infection (including for specific variants) across different territories. There were also considerations beyond the remit of CMO or GCSA advice, such as wider public confidence in the government’s response, or the impact of restrictions on travel and trade. In epidemiological terms, the following 2 considerations were important.

First, there was significant global variation in the quality of data informing risk assessments for travel to different areas. In the initial stages of the pandemic, the combination of limited testing, pre-symptomatic and pauci-symptomatic spread and syndromic surveillance relying on non-specific symptoms meant that the extent of transmission in a country was often not visible until hospitalisations began to rise. Data sharing globally was also challenging, with a lack of data on risks for many areas. Retrospective genomic analysis showed that in February 2020 the UK saw several hundred incursions from countries not known to be a high transmission risk at the time, mostly in Europe.^{[footnote 53]} This improved during the pandemic with the expansion of genomic and other surveillance and with early sharing of data which was critical in judging potential risks, though it remained an ongoing challenge worldwide with many countries unable to provide timely, accurate and representational data.

Second, as with all other NPIs, travel restrictions were not implemented in a policy or epidemiological fixed state and their impact on transmission depended on a number of factors including adherence to measures and behaviours surrounding travel. One international study on border policies, for example, noted that the impact of quarantine on imported cases depended on adherence and other factors such as testing policies in different countries.^{[footnote 54]}

Interpretation of testing data from travel-related testing required careful review and caveats – for example, to flag that positive test rates for those quarantining from a certain country did not translate to case rates in that country. Besides this, in some cases those travelling internationally may have particular behavioural patterns or health profiles that are not representative of the wider population.

Shielding

There were 3 potential approaches to supporting those with heightened risk of severe disease during this pandemic:

1. To identify those at higher risk (who in this pandemic at the outset were thought to number in the millions) and inform them so they would be able to better manage their own risk.

2. To put a programme in place with guidance on managing risk, and support to do so, alongside a wider package of NPIs to reduce transmission in the community.

3. To put measures in place only for those at higher risk, without a wider package of NPIs to reduce community transmission.

In this pandemic, the first 2 options were adopted, and are described here alongside our reflections. Some people (most well known as part of the Great Barrington Declaration) promoted targeting NPIs to the vulnerable group alone or implementing only shielding as a viable option to reduce overall severe disease and deaths, allowing the infection to spread in all others. There were serious questions about the practicalities, ethics and indeed effectiveness of such an approach. For a highly transmissible infection with often minimal symptoms it was extremely difficult to target specific people or groups successfully. Identifying the vulnerable is also an inexact science and the level of vulnerability and associated numbers of those affected changed through the pandemic. Ultimately the most effective way to reduce risk for the vulnerable was to reduce overall community transmission. Many of those shielding lived in households or settings with others who could be at risk of introducing infection when community rates were high, and those requiring care and support services also had regular contacts from outside the home.^{[footnote 55]}

A core element of the programme to support those at higher risk was advice not to leave their home unless essential (‘shielding’), while a wider package of NPIs sought to reduce community transmission.^{[footnote 56]} This advice was accompanied by supportive measures such as eligibility for free food and medicines delivery, differential GP follow-up, access to virtual services, statutory sick pay, and various other forms of support which local authorities and the voluntary and community sector played a major part in delivering. The programme was paused after the first wave but continued in some areas with high transmission (such as Leicester). It formally ended in September 2021 with a letter to those remaining on the clinically vulnerable list.^{[footnote 57]}

The first iteration of a list of clinically vulnerable people across the country was supported by expert panel review of current epidemiological and clinical data, alongside routine data sets (a particular strength of the UK NHS) which were used to identify those considered most vulnerable.^{[footnote 58]} This was updated throughout the pandemic in order to establish and maintain as accurate a ‘Shielded Patient List’ as possible given current knowledge, which was then flagged in GP records and used to support shielding policy and associated initiatives across the system as well as direct communication with patients. GPs themselves also supported this process by flagging which of their patients had specified health conditions (though people could also opt out of being on the list by contacting their GP – shielding was not compulsory).

This list evolved as understanding of the disease and data on vulnerabilities grew, and in October 2020 a risk prediction model called QCovid® was released that estimated a person’s combined risk of catching coronavirus and being admitted to hospital, as well as their combined risk of catching coronavirus and dying (see also Chapter 2: disparities; Chapter 1: understanding the pathogen; Chapter 1, section 5: How severe was this disease, and were there longer-term sequelae?).^{[footnote 59]}, ^{[footnote 60]} Using the anonymised health records of more than 8 million people from GP records, hospital records and mortality data during the first wave, the tool supported population risk assessment, clinical support, vaccine rollout prioritisation and patients themselves in understanding potential vulnerabilities to severe COVID-19. Importantly, QCovid® also included a measure of socio-economic deprivation.

The approach to development of a list varied slightly across the UK. Scotland, for example, did not apply QCovid® to population records, though findings in England from use of QCovid® such as the identification of vulnerability in adults with Downs Syndrome and people with chronic kidney disease stage 5 resulted in these groups being added to Scotland’s clinically vulnerable list. Scotland did not use QCovid® for clinical decision-making, for clinically vulnerable group vaccine prioritisation in January 2021 (when Scotland’s Shielding List itself was used as a proxy), or for the Shielding List update in February 2021. This was partly due to QCovid® not being compatible with Scottish data structures (such as CancerCare records, or measures of deprivation), and the requirement for separate validation of the model for the Scottish population. It was also partly due to data gaps in earlier iterations of the QCovid® model (which are explored below), with early data used to develop the model not accounting for protection from vaccination or for newly established variants.

The tool adapted to new evidence as it became available and did not initially incorporate background infection rates, seropositivity (whether through vaccine or previous infection), the possible mitigating impacts of other interventions (such as lockdowns) or behaviours that could heighten or reduce risk. It was continually validated by a 4-nations panel with expert support, particularly clinical, and was updated accordingly:

• 1.5 million people were added to the list in February 2021 after further evidence emerged through the pandemic on relative risk for either both single or multiple conditions, around 800,000 of whom were then prioritised for vaccine rollout
• children and young people were removed in August 2021

It is currently difficult to quantify the impact of shielding on either SARS-CoV-2 transmission, COVID-19 outcomes or wider impacts, because its early and universal application for relevant groups left no control groups – nor would it have been considered ethical to do so. However, in summer 2021 QCovid® was validated to be performing well in predicting COVID-19 mortality.^{[footnote 61]}

The quality, breadth and completeness of data and evidence available on clinical vulnerabilities impacted the accuracy of the list, and these improved throughout the pandemic. This was partly why the UK was able to enact a formalised shielding policy, with existing datasets alongside local and GP intelligence enabling targeted support to those advised to shield. The evolution of more accurate risk predictions was in part a consequence of amassing further data from the pandemic itself, and in part a result of major efforts to lift barriers in data access, coding read-across between datasets, record linkage and of course bringing together the technical skills to analyse that data.

There are some important principles that bear consideration for similar interventions in the future:

• shielding should normally be an addition to rather than an alternative to other NPIs to reduce community transmission, for the reasons outlined above
• at the outset of a pandemic in an immunologically naïve population with potentially high-risk comorbidities for a novel disease, it was thought important to act swiftly and advise people on their potential risk based on the understanding of the disease at the time. An iterative approach was then needed when forming a list of clinically vulnerable, particularly in the early stages of the pandemic when the disease was moving through populations with different health profiles to the UK’s
• communication about clinical vulnerability is complex and can have long-term impacts. Early messages on clinical vulnerability may ‘stick’ even as the evidence base evolves or may become confused with other vulnerability issues such as economic vulnerability – and often they overlap. There can therefore be a wider group of people who are not formally clinically vulnerable but may be particularly concerned for their health, and they may as a consequence follow shielding advice. In addition, some groups previously thought to be high risk were then found to no longer be deemed ‘clinically vulnerable’. Early interventions to protect the vulnerable, regardless of whether they are formally lifted, may effectively stay in place for many over a much longer period due to ongoing concerns about risks. Communication about the intervention itself therefore needed to be clear as to who was vulnerable and why this was changing, as well as what was being asked and why. Communications also needed to be accessible to different groups – accessible versions of shielding letters were in time hosted online, with translations into 13 different languages, easy read and audio described versions – but this took time. Resource is needed from early on to make it happen more speedily in the future
• shielding had a major impact on many people’s lives – ONS data describing a self-delineated group of those at higher risk (not necessarily the same as those advised to shield for clinical vulnerability reasons) highlighted that shielding was likely to have saved lives but with considerable associated psychological morbidity in some^{[footnote 62]}
• the risk of experiencing these negative wider impacts was high for clinically vulnerable populations, and for some this was compounded by having limited space at home to exercise or by a lack of digital tools or skills. ONS surveys of those considered to be ‘clinically extremely vulnerable’ (though noting this list changed throughout the pandemic) have been run to help to understand, among other things, behaviours, mental and physical health and support requirements during the pandemic.^{[footnote 62]} These have helped to inform guidance and communications, as well as council funding allocations in relation to shielding
• there can also be wider impacts on services, such as GPs and clinicians needing to offer advice on clinical vulnerability, and this needed to be incorporated into plans
• having a tool in the public domain to understand risk in the community has been helpful (in this pandemic QCovid®), and existing data sets have enabled coordination to support those advised to stay at home such as food or medicine delivery

Reflections and advice for a future CMO or GCSA

Point 1

In the absence of pharmaceutical interventions, NPIs are the only option for pandemic control.

Adherence was generally very high across a range of NPIs in all 4 nations of the UK and in all groups with the public proving willing to take extraordinary measures in order to protect one another in a public health emergency. This included, for example, the efforts made by young people even though they correctly perceived limited personal risk.

Point 2

NPIs have complex impacts and involved balancing multiple known, potential and unknown harms and benefits.

On several issues evidence and scientific consensus in the UK and globally evolved over this pandemic, such as on the relative contribution of asymptomatic and aerosol transmission, and this had to be continually reviewed and clearly communicated to decision-makers and the public.

There were challenges developing the evidence base on the impact and effectiveness of individual NPIs, especially in real-world settings. We initially lacked off-the-shelf study designs, evaluation protocols, protocols for funders’ rapid review and prioritisation and a pre-agreed framework for NPIs.

Point 3

The effectiveness of NPIs depends largely on how far individuals are able and willing to adhere.

It was therefore important to understand behavioural responses to changing policies (including enforcement and supportive measures), changing social norms around adherence and changing risk perceptions as natural and vaccine-acquired immunity increased. Behaviours may not always match intention, and understanding the gap between the two was important. Local authorities managing outbreaks fed back that this gap between intention and practice was linked to:

• financial considerations (such as reimbursement for time off work)
• practical considerations (such as support with getting shopping or caring responsibilities)
• information needs (such as providing clear detail on why individuals should isolate and how they can do this practically and safely)
• emotional and mental health support

The effectiveness and impacts of NPIs were difficult to predict and could vary significantly between groups, locations and stages of the epidemic.

Point 4

Trust was important in public communications around NPIs so that people knew what to do and, as importantly, why.

Using the right communicators, different voices, and the best methods of communication to build trust and communicate clearly and consistently to a range of people throughout the pandemic was a difficult judgement call.

In this pandemic, a range of professional voices played a part, as did community champions and other local and community leaders. Volunteering initiatives and co-production can also develop relationships and support dialogue between service providers and communities that do not start with a high level of trust in authorities.

Local directors of public health amplified and clarified national messaging, which was the main source of information for a large proportion of the population, as well as giving local messages and providing leadership.

Point 5

The risks of entrenching or exacerbating inequalities in the deployment of NPIs needs to be considered.

Those living in the most deprived areas were:

• least likely to be able to work from home
• more likely to use family or neighbour care-givers
• more likely to use public transport
• more likely to live in high-density accommodation
• more likely to have insecure employment and minimal or no financial resilience

All of these increased the risk of exposure in a population which was also more likely to have the co-morbidities which increased the chance of hospitalisation or death. Although support packages such as self-isolation payments can be designed, there still remain important epidemiological issues such as living in high-density multi-generation homes when isolating, or work conditions that involve living together, so it is important to understand local circumstances and community support networks when designing and implementing NPIs.

References

1. SAGE papers are accessible via the SAGE homepage on GOV.UK ↩

2. See, for example, UK COVID-19 press conference, 16th March 2020 ↩

3. Welsh Government. 16 July 2021. Technical Advisory Group: 5 harms arising from COVID-19 ↩

4. SAGE. 17 December 2020. Research and analysis: DHSC/ONS/GAD/HO: Direct and indirect impacts of COVID-19 on excess deaths and morbidity - December 2020 update ↩

5. SAGE. YouGov and Cabinet Office COVID-19 Public Attitude Research, 18 March 2020 ↩

6. Cabinet Office. Executive summary of UK government spring 2021 COVID-19 Response (roadmap) ↩

7. SAGE-Environmental Modelling Group. Application of physical distancing and fabric face coverings in mitigating the B117 variant SARS-CoV-2 virus in public, workplace and community settings, 13 January 2021 ↩

8. NHS Guidance: How to wash your hands ↩

9. SAGE. 26 May 2020. Research and analysis: Evidence of environmental dispersion of COVID-19 for different mechanisms, 14 April 2020 ↩

10. SAGE. 21 June 2020. Research and analysis: Transmission of SARS-CoV-2 and Mitigating Measures – update, 4 June 2020 ↩

11. SAGE. 26 January 2022. Research and analysis: NERVTAG and EMG: Role of aerosol transmission in COVID-19 – 22 July 2020 ↩

12. SAGE. Minutes: Early evidence on distancing discussed at SAGE 9th meeting, 20 February 2020 ↩

13. UK government Social Distancing Review: Report. July 2021. Available from: COVID-19 Response – Spring 2021: roadmap reviews ↩

14. SAGE. SAGE return for COVID-19 strategy: sequencing of social distancing behavioural and social interventions, 6 May 2020 ↩

15. SAGE. 29 May 2020. SAGE 27 minutes: Coronavirus (COVID-19) response, 21 April 2020 ↩

16. WHO. 6 April 2020. Advice on the use of masks in the context of COVID-19: Interim guidance ↩

17. WHO. 5 June 2020. Advice on the use of masks in the context of COVID-19: Interim guidance ↩

18. UK Parliament. 24 July 2020. Summary of Scientific advice and policy timeline on community use of face masks and coverings in the UK. As part of the COVID-19: July update on face masks and face coverings for the general public ↩

19. WHO. 6 April 2020. Advice on the use of masks in the context of COVID-19: Interim guidance ↩

20. SAGE-Environmental Modelling Group. Application of physical distancing and fabric face coverings in mitigating the B117 variant SARS-CoV-2 virus in public, workplace and community settings, 13 January 2021 ↩

21. For England, for example: Cabinet Office and DHSC Guidance. Updated 27 January 2022. [Withdrawn] Face coverings: when to wear one, exemptions and what makes a good one ↩

22. SAGE Environmental Modelling Group (EMG). Research and analysis: Role of ventilation in controlling SARS-CoV-2 transmission, 30 September 2020 ↩

23. SAGE Environmental Modelling Group (EMG). Research and analysis: Role of ventilation in controlling SARS-CoV-2 transmission, 30 September 2020 ↩

24. SAGE EMG. Research and analysis: EMG: Simple summary of ventilation actions to mitigate the risk of COVID-19, 1 October 2020 ↩

25. SAGE. 12 October 2020. Research and analysis: Non-pharmaceutical interventions (NPIs) table, 21 September 2020 ↩

26. SAGE. 22 October 2021. Research and analysis: SPI-B, SPI-M and EMG: Considerations for potential impact of Plan B measures, 13 October 2021 ↩

27. Mendez-Brito et al. Systematic review of empirical studies comparing the effectiveness of non-pharmaceutical interventions against COVID-19 Journal of Infection, 2021. Vol 83, Iss 3, Pp 281-293 ↩

28. S. Michie, H.W.W. Potts, R. West, R. Amlȏt, L.E. Smith, N.T. Fear, G.J. Rubin. Factors associated with non-essential workplace attendance during the COVID-19 pandemic in the UK in early 2021: evidence from cross-sectional surveys Public Health, 2021; 198; 106-113 ↩

29. SAGE. 22 October 2021. Research and analysis: SPI-B, SPI-M and EMG: Considerations for potential impact of Plan B measures, 13 October 2021 ↩

30. SAGE. 17 December 2020. Research and analysis: DHSC/ONS/GAD/HO: Direct and indirect impacts of COVID-19 on excess deaths and morbidity - December 2020 update ↩

31. SAGE. 12 October 2020. Research and analysis: Non-pharmaceutical interventions (NPIs) table, 21 September 2020 ↩

32. CMMID Repository. Summary page and list of reports for CoMix study - Social contact survey in the UK ↩

33. NIHR. Summary page and list of reports for the CORSAIR study ↩

34. See, for example, Brandal Lin T, MacDonald Emily, Veneti Lamprini, Ravlo Tine, Lange Heidi, Naseer Umaer, Feruglio Siri, Bragstad Karoline, Hungnes Olav, Ødeskaug Liz E., Hagen Frode, Hanch-Hansen Kristian E et al. Outbreak caused by the SARS-CoV-2 Omicron variant in Norway, November to December 2021. ↩

35. Dasha Majraa, Jayme Benson, Jennifer Pitts, Justin Stebbing. SARS-CoV-2 (COVID-19) superspreader events Journal of Infection, 2021. Vol 82, Pp 36-40 ↩

36. SAGE. SAGE 42 minutes: Coronavirus (COVID-19) response, 18 June 2020 ↩

37. SAGE. 4 September 2020. Research and analysis collection: PHE and EMG: Aerosol and droplet generation from singing, wind instruments and performance activities, 13 August 2020 ↩

38. Thompson H, et al. Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) Setting-specific Transmission Rates: A Systematic Review and Meta-analysis Clinical Infectious Diseases, Volume 73, Issue 3, 1 August 2021, Pages e754–e764 ↩

39. DCMS. Updated 1 July 2021. Policy paper: Events Research Programme: Phase I findings ↩

40. SAGE. Research and analysis: Summary of the effectiveness and harms of different non-pharmaceutical interventions, 21 September 2020 ↩

41. Jeffrey B, Walters CE, Ainslie KEC et al. Anonymised and aggregated crowd level mobility data from mobile phones suggests that initial compliance with COVID-19 social distancing interventions was high and geographically consistent across the UK [version 1; peer review: 2 approved] Wellcome Open Res 2020, 5:170 ↩

42. Niedzwiedz CL, Green MJ, Benzeval M, et al. Mental health and health behaviours before and during the initial phase of the COVID-19 lockdown: longitudinal analyses of the UK Household Longitudinal Study J Epidemiol Community Health 2021;75:224-231 ↩

43. Department for Transport. Statistical data set: Road traffic statistics (TRA) ↩

44. PHE. August 2021. Wider impacts of COVID-19 on physical activity, deconditioning and falls in older adults ↩

45. Smith TO, Belderson P, Dainty JR, et alImpact of COVID-19 pandemic social restriction measures on people with rheumatic and musculoskeletal diseases in the UK: a mixed-methods studyBMJ Open 2021;11:e048772. doi: 10.1136/bmjopen-2021-048772 ↩

46. Sidpra J, Abomeli D, Hameed B, et al. Rise in the incidence of abusive head trauma during the COVID-19 pandemic Archives of Disease in Childhood 2021;106:e14 ↩

47. See, for example, ONS: Domestic abuse during the coronavirus (COVID-19) pandemic, England and Wales: November 2020 ↩

48. See, for example: Parmar D, Stavropoulou C, Ioannidis J P A. Health outcomes during the 2008 financial crisis in Europe: systematic literature review BMJ 2016; 354 :i4588 doi:10.1136/bmj.i4588 ↩

49. See, for example: Vanaken GJ, Danckaerts M. Impact of Green Space Exposure on Children’s and Adolescents’ Mental Health: A Systematic Review. Int J Environ Res Public Health. 2018 Nov 27;15(12):2668. doi: 10.3390/ijerph15122668. PMID: 30486416; PMCID: PMC6313536. ↩

50. SAGE. Scientific evidence supporting the government response to coronavirus (COVID-19) ↩

51. SAGE. Research and analysis: Impact of Interventions TFG: The UK’s 4 nations’ autumn interventions (update), 26 November 2020 ↩

52. PHE and DHSC. 10 January 2020. News story: Novel coronavirus and avian flu: advice for travel to China ↩

53. L De-Plessis, J McCrone, A Zarebski et al. Establishment and lineage dynamics of the SARS-CoV-2 epidemic in the UK. Science, Jan 2021. Vol 371, Issue 6530, pp. 708-712. DOI: 10.1126/science.abf2946 ↩

54. Bou-Karroum L, Khabsa J, Jabbour M, Hilal N, Haidar Z, Abi Khalil P, Khalek RA, Assaf J, Honein-AbouHaidar G, Samra CA, Hneiny L, Al-Awlaqi S, Hanefeld J, El-Jardali F, Akl EA, El Bcheraoui C. Public health effects of travel-related policies on the COVID-19 pandemic: A mixed-methods systematic review. J Infect. 2021 Oct;83(4):413-423. doi: 10.1016/j.jinf.2021.07.017. Epub 2021 Jul 24. PMID: 34314737; PMCID: PMC8310423 ↩

55. McKeigue, P.M., McAllister, D.A., Caldwell, D. et al. Relation of severe COVID-19 in Scotland to transmission-related factors and risk conditions eligible for shielding support: REACT-SCOT case-control study BMC Med 19, 149 (2021) ↩

56. UKHSA and DHSC. 21 March 2020. COVID-19: guidance on protecting people defined on medical grounds as extremely vulnerable ↩

57. UKHSA and DHSC. 17 September 2021. Guidance: COVID-19: letter to patients on end of shielding programme ↩

58. Docherty et al BMJ 2020, Docherty A B, Harrison E M, Green C A, Hardwick H E, Pius R, Norman L et al. Features of 20 133 UK patients in hospital with covid-19 using the ISARIC WHO Clinical Characterisation Protocol: prospective observational cohort study BMJ 2020; 369 :m1985 doi:10.1136/bmj.m1985 ↩

59. NHS Digital. Summary page on QCovid ↩

60. Clift A K, Coupland C A C, Keogh R H, Diaz-Ordaz K, Williamson E, Harrison E M et al. Living risk prediction algorithm (QCOVID) for risk of hospital admission and mortality from coronavirus 19 in adults: national derivation and validation cohort study BMJ 2020; 371 :m3731 doi:10.1136/bmj.m3731 ↩

61. Nafilyan V, Humberstone B, Mehta N, Diamond I, Coupland C, Lorenzi L, Pawelek P, Schofield R, Morgan J, Brown P, Lyons R, Sheikh A, Hippisley-Cox J. An external validation of the QCovid risk prediction algorithm for risk of mortality from COVID-19 in adults: a national validation cohort study in England. Lancet Digit Health. 2021 Jul;3(7):e425-e433. doi: 10.1016/S2589-7500(21)00080-7. Epub 2021 May 25. PMID: 34049834; PMCID: PMC8148652 ↩

62. ONS. 13 May 2022. Coronavirus and clinically extremely vulnerable (CEV) people in England: 4 April to 23 April 2022. Analysis of people previously considered to be clinically extremely vulnerable (CEV) in England during the coronavirus (COVID-19) pandemic, including their behaviours and mental and physical well-being. ↩ ↩²