key: cord-1017546-tviyo0gb authors: Miles, David K.; Heald, Adrian H.; Stedman, Michael title: How fast should social restrictions be eased in England as COVID‐19 vaccinations are rolled out? date: 2021-05-04 journal: Int J Clin Pract DOI: 10.1111/ijcp.14191 sha: 2964d9badfe530f18309e0c3f33dd06873f4ac32 doc_id: 1017546 cord_uid: tviyo0gb INTRODUCTION: Vaccination against the COVID‐19 virus began in December 2020 in the UK and into Spring 2021 has been running at 5% population/week. High levels of social restrictions were implemented for the third time in January 2021 to control the second wave and resulting increases in hospitalisations and deaths. Easing those restrictions must balance multiple challenging priorities, weighing the risk of more deaths and hospitalisations against damage done to mental health, incomes and standards of living, education and provision of non‐Covid‐19 healthcare. METHODS: Weekly and monthly officially published data for 2020/21 were used to estimate the influence of seasonality and social restrictions on the spread of COVID‐19 by age group, on the economy and on healthcare services. These factors were combined with the estimated impact of vaccinations and immunity from past infections into a model that retrospectively reflected the actual numbers of reported deaths closely both in 2020 and early 2021. The model was applied prospectively to the next 6 months to evaluate the impact of different speeds of easing social restrictions. RESULTS: The results show vaccinations as significantly reducing the number of hospitalisations and deaths. The central estimate is that relative to rapid easing, the avoided loss of 57 000 life‐years from a strategy of relatively slow easing over the next several months comes at a cost in terms of GDP reduction of around £0.4 million/life‐year loss avoided. This is over 10 times higher than the usual limit the NHS uses for spending against Quality Adjusted Life Years (QALYs) saved. Alternative assumptions for key factors affecting the spread of the virus give significantly different trade‐offs between costs and benefits of different speeds of easing. Disruption of non‐Covid‐19 Healthcare provision also increases in times of higher levels of social restrictions. CONCLUSION: In most cases, the results favour a somewhat faster easing of restrictions in England than current policy implies. The SARS-CoV-2 (Covid- 19) virus has impacted all aspects of life in the UK as elsewhere in the world. The number of reported cases of related hospitalisations and deaths rose rapidly once again at the end of 2020 across England and then fell rapidly through February 2021. At the same time vaccines against the virus have been administered at the time of writing to around 30 million people across the UK; the rollout of the vaccines looked likely to proceed rapidly into the Spring and beyond. It is likely that the extensive restrictions on travel and social interactions introduced at the end of 2020, and extended further early in 2021, have been the major factor behind the substantial decline in the rate at which the virus was spreading. As those who have recently been vaccinated gain some immunity, and as new doses are administered to more people, the rollout of the vaccine will play an increasing role in driving down the R numberthe key factor behind the spread of the virus. How fast restrictions should be eased as vaccinations bear down on the spread of Covid-19 is a critical policy issue. It should depend on how the risks of infections, hospitalisations and deaths are affected by easing restrictions-given the pace of vaccinations and their likely effectiveness. However, assessing how many more people might be infected and suffer serious illness if restrictions are eased sooner rather than later cannot be the only factor that is relevant to policy. The great benefits that severe restrictions bring is in reducing illness and deaths from the virus; but restrictions bring costs and it is clear they are very large: the disruption to education will negatively affect millions of young people for many years; the mental health and other distress that isolation brings for people forced to stay home is likely to be significant now and into the future; the number of businesses that will shed employees will rise the longer severe restrictions are kept in place and the unemployment this will cause will do significant harm to those who lose their jobs and struggle to find new ones. This study tries to bring an assessment of the costs and benefits of different strategies for easing restrictions together to inform that urgent policy question. It models the impact upon infection rates and deaths from different strategies for easing restrictions in England and sets those alongside an assessment of the costs of different strategies. We first consider the costs of different levels of restrictions. We then set out the model for the impact of varying restrictions on possible paths for infections and deaths. Finally, we bring together an assessment of the costs and benefits of different scenarios for easing restrictions. The costs of restrictions imposed to control the spread of the virus are not easy to measure accurately. Many of those costs are likely to arise in the future -the damage done to the life chances of young people whose education has been disrupted will be incurred over long periods stretching far into the future. The Institute for Fiscal Studies has estimated that this cost may already run to something like £350 billion 2 ; nearly all of that lies ahead of us. Businesses that will not be able to re-open after the pandemic has past will create unemployment that is being masked by temporary government schemes that have until now stopped unemployment from rising significantly. Research by Peter Lambert and John Van Reenan of the London School of Economics 3 suggests that around nine hundred thousand firms-mainly smaller businesses that employ around 2.5 million people-were at significant risk of not being viable by mid-2021. The extent to which that risk crystallises will depend on how • Vaccination against the COVID-19 virus began at the start of December 2020 in the UK and into Spring 2021 was rolling out at the rate of 5% population/wk. • High Levels of social restrictions were implemented for the third time at the start of January 2021. • The easing of those restrictions needs to balance multiple challenging priorities. • Our central estimate is that the current strategy of relatively slow easing comes at a cost in terms of GDP reduction of £0.4million/life-year loss avoided-over 10 times higher than the normal public health expenditure limit used to evaluate health resources spent against Quality Adjusted Life Years saved. • This should be taken into account in any policy decisions. • The return of NHS activity to the previous 'normal' levels will also be slower if restrictions are eased more gradually. and when restrictions are eased. The damage that unemployment and disrupted education brings goes beyond the lower incomes (lost GDP) now and into the future-the damage to the physical and mental health of reduced employment chances is imperfectly reflected by a calculation of lost aggregate output. Work by Carol Propper of Imperial College suggests that the rise in unemployment in the UK that came in the wake of the global financial crisis of 2007-08 increased the number of people in the UK with chronic health problems by around 900 000. 4 It is plausible that the rise in unemployment that comes about as a result of Covid-19, and the restrictions imposed to deal with it, may be of the same order of magnitude. Much of this damage to the welfare of people would have occurred even without any government restrictions. It is only that part of the costs of the pandemic that came from restrictions that should be weighed up against the benefits that such restrictions bring. Thus, there are two formidable measurement issues that arise in assessing what costs come from restrictions-first, what are the scale of the many different costs, which spread far into the future, from the virus (over and above the direct costs to the health of those infected) and, second, how much of those costs are caused by government restrictions and how many would have arisen even without such restrictions because of voluntary changes in behaviour from a cautious public. Our strategy is to take a very narrow definition of costs, simply focusing on the value of total current incomes (ie Gross Domestic Product (GDP)) lost as a result of government restrictions during the months when they were in place. This will generate an estimate of the damage that is likely to underestimate substantially long-run costs. It would only be a central estimate of economic costs if the impact of restrictions is only felt while they are in place and that once they are lifted economic activity returns to its prepandemic path. Recent Office for Budget Responsibility (OBR) estimates suggest that the economic damage from Covid-19, and restrictions taken to counter its effects, will last for many years; they suggest incomes will be 3% lower even in the long term. UK GDP fell dramatically below its prepandemic levels in the Restrictions were eased significantly between June and August. Over those 3 months, GDP recovered so that on average it was around 10% below the levels of the corresponding months of 2019. By early autumn of 2020, at which point restrictions had been relaxed to a level as low as they were to reach in 2020, GDP was around 6% lower than for the same months of 2019. The tightening in restrictions that came at the end of 2020 had yet to have its full impact on UK GDP by December 2020 and it seems likely that output in the first quarter of 2021 was further below the prepandemic level than in the autumn of 2020. It would be a very substantial overestimate of the immediate The overall impact on GDP UK, we use the following assumptions for the effect of restrictions on immediate levels of GDP (Table 1) . We stress again that in only taking the immediate hit to GDP from restrictions we are taking a narrow measure of their costs and erring on the side of underestimating their harm. We make no allowance for the indirect harm to health and welfare that restrictions may have created. It is clear that the response to the pandemic has disrupted the provision of non-Covid general healthcare and those costs may last for many years. For example: • The number of patients awaiting elective treatment was at an alltime high of 4.46 million in November 2020. 5 • There have been large reductions in Primary care contacts for many conditions including diabetic emergencies, depression, selfharm, stroke, anxiety, transient ischaemic attack, heart failure, myocardial infarction, unstable angina and asthma exacerbation. 6 • A total of 40 000 fewer patients started cancer treatment across the UK in 2020, driven largely by a reduction in cancer diagnoses during the year. 7 • As noted above, the impact of job losses goes beyond the immediate loss of output from those not working. Higher unemployment has consistently been found to have a negative impact on a range of physical health and mental health outcomes through poverty, stress, unhealthy behaviours and risk of future unemployment perpetuating negative cycles. Health consequences of unemployment are likely to increase with its duration. 8 The rate at which new infections occur and the health costs associ- Those values are plausible levels for R at the beginning of 2020 when the virus arrived in the UK but before anyone was aware of risks. It seems, however, unlikely that even if government restrictions were eased quickly, the R value (excluding immunity) would return to such levels. People are unlikely to believe all risk is gone when they would be reminded every day of new infections and deaths. We use a somewhat lower estimate of what R will become when government restrictions are eased. We also factor in the impact of vaccinations and seasonality which already have played some role in the path of the virus. We allow for both factors to influence the R number in the scenarios for easing restrictions. The UK was the first country to approve vaccines for use in the humidity and sunlight. 13 The ultraviolet (UV) sensitivity of coronaviruses and SARS-CoV-2 indicates that a proportion of the SARS-CoV-2 virus might be inactivated after being exposed to sunlight during summer in most world cities. We take account of the likely level of existing immunity, seasonal factors and government restrictions in affecting the effective R number. It is the R number that we use in a simple SIR model that drives the level of new infections. We assume that risks of hospitalisation and death for the newly infected depend on age, using fairly crude compartmentalisation into those at high risk of serious illness and death (age over 65) and those at relatively low risk (age under 65). We make a number of assumptions for the impact of a range of factors on R and assess their plausibility by how well they account for the past evidence on the course of the pandemic in the UK. We here describe how we parameterised the impact of factors and then consider the ability of the model to track data on the course of the virus. We make use of government estimates of current effective R values, published weekly from 27th May 2020. 14 We used these estimates and scaled up to an R0 equivalent-that is the R number were all people to have been susceptible. The transition between R0 and effective R values is made by adjusting for the percentage of the population remaining susceptible; the effective R number is R0 multiplied by the proportion of the population that is susceptible, that is, the total population minus those vaccinated (adjusted by the assumed effectiveness of vaccination) and minus those with an assumed immunity from past infections. The estimated level of immunity we use is based on government estimates reported by the ONS 15 We divided the year into four 3-month seasons that corre- 4.0 days for the symptomatic; this seems to imply a period of being infectious of between 6.7 and 8.6 days after infection. We take the 7-day infectiousness period as our base case but also consider the possibility of a 14-day infectiousness period. We assume that some proportion of those who are infected are hospitalised, and that happens 1 week after they show symptoms. We assume that those who die with the virus do so 2 weeks after they are infectious and become ill. SAGE papers suggest an average lag between the onset of serious illness and death of 5 days (for those not hospitalised) and 10 days (for those hospitalised). Our 7- The total annual number of patients being treated (ie contacts) in To establish the impact of social restrictions on the non-Covid healthcare provision, we grouped months according to the levels of social restrictions, as shown in Table A1 , and then related the total monthly activity in healthcare services with different levels of urgency. We compared the numbers of accident and emergency attendances, emergency admissions, 19 GP attendances 20 and new outpatient referrals 21 to the level of provision in the same months in previous years. Based on the data in Figure 1 , we assume that the way in which the R0 value has varied with restrictions and with the season is as shown in Table 2 . Figure 2 illustrates the large variation in mortality risk level associated with age; the oldest 18% of the population accounted for 14% of the cases but 90% of the deaths, a 50-fold higher case fatality risk compared with the other group with 82% of the population and 10% of the deaths. Table A2 shows all the model parameters we use for simulations. Using the parameters set out in Table 2 , we ran the model through The model was then run forward to estimate the number of deaths There is significant uncertainty about several factors that affect the spread of the virus. No strong policy conclusions should be made based on a single best guess for these factors. So we considered several variations and sensitivities to our base case setting of key parameters. 1. Seasonality: We considered the impact of assuming that March may be closer to winter than summer in terms of the seasonal effect on R We allowed for all R values to be increased by 10% and simulations with the infectious period to be twice as long, at 14 days. We allowed for vaccine effectiveness to be lower at 50% and 15% (for doses 1 and 2) against the base case of 65% and 20% and a slower rollout reducing to 2 000 000/wk. 4 . The worst-case scenario is included using a combination of the above variants being simultaneously applied. 5. An upside scenario was evaluated with the vaccine being more effective and the vaccination rate increasing. Table 3 for easing restrictions and using the base case assumptions over seasonality, vaccine rollout and its effectiveness. Table 2 showed how the assumptions for R varies with restrictions and season. These simulations adjust the susceptible population week by week based on vaccine rollout and immunity assumed to be acquired from the history of infections. The average monthly level of healthcare provision grouped by level of social restriction is shown in Results from using more adverse assumptions are shown in Table 6 . The final alternative scenario (and the only one that is less adverse than the base case) is where vaccine effectiveness is higher. There is considerable upside potential if vaccine effectiveness is higher and if delivery is speeded up. This suggests that at those higher levels of effectiveness, the immediate removal of social restrictions would lead to less than half as many deaths. Easing restrictions comes at a cost-more people will be infected The rule used in the National Health Service for the effectiveness of medical treatments is that cost per life-year saved from treatment should be no higher than £30 000. Based on that rule, the expected costs of a very slow easing of restrictions over the next 6 months seems higher than its likely benefits. Some government departments use an acceptable resource cost per extra healthy year of life saved at a higher level of £60 000. Even on that basis, a somewhat faster easing of restrictions than current government plans seem to imply looks warranted. Easing restrictions faster generates risks and it may need to be reversed. But there are inevitably risks with any strategy. If one considered that any reversal of an easing of restrictions was itself hugely costly, then clearly a more cautious pace of easing is warranted. But to attach a huge weight to any reversal has some unwelcome implications. It is certainly possible that later this year, new mutations of the virus might come to the UK 23 and if they are especially virulent/life-threatening, tighter restrictions than those that have been in place in February could become warranted. Yet to tighten restrictions to the level that could become warranted simply to avoid ever having to tighten restrictions again would be a bizarre strategy that itself entails huge inherent risks. The current strategy of relatively slow easing comes at a cost in terms of GDP reduction of up to £400 000/life-year-over 10 times higher than the normal public health expenditure limit used to evaluate health resources spent against quality adjusted life years saved. This should be taken into account in any policy decisions going forward. The return of NHS activity to the previous 'normal' levels will also be slower the more gradually restrictions are eased and has a yet unclear impact on future population health outcomes and mortality in all strata of society. None of the authors has any conflict of interest. The data that support the findings of this study are publically available. David K. Miles https://orcid.org/0000-0002-6387-0041 Adrian H. Heald https://orcid.org/0000-0002-9537-4050 Unlocking" Roadmap Scenarios for England v2 The crisis in lost learning calls for a massive national policy response A major wave of UK business closures by April 2021? The scale of the problem and what can be done" Covid-19 Analysis Series No.016; 2021. The Centre for Economic Performance Macroeconomic Conditions and Health in Britain: Aggregation, Dynamics and Local Area Heterogeneity Growing backlog of planned surgery due to covid-19 Indirect acute effects of the COVID-19 pandemic on physical and mental health in the UK: a population-based study Cancer services during COVID-19: 40,000 fewer people starting treatment; 2021. K Roberts Cancer Research UK Employment and unemployment -How does work affect our health. The Health Foundation SAGE 80 minutes: Coronavirus (COVID-19) response; 2021 Summary of modelling on scenarios for easing restrictions Oxford coronavirus vaccine shows sustained protection of 76% during the 3-month interval until the second dose -Oxford University News UK COVID-19 vaccine uptake plan Virus survival as a seasonal factor in influenza and poliomyelitis The R-value and growth rate in the UK The latest reproduction number (R) and growth rate of coronavirus (COVID-19) in the UK OBNS) Coronavirus (COVID-19) Infection Survey, UK COVID-19) in the UK Healthcare in England COVID-19) in the UK Deaths in England nandc ommun ity/birth sdeat hsand marri ages/death s/bulle tins/ death sregi stere dweek lyine nglan dandw alesp rovis ional/ weeke nding 5febr uary2021 NHS England A&E Attendances and Emergency Admissions 2020-21 Living with Covid-19: Balancing Costs Against Benefits in the Face of the Virus Mutated COVID-19 may foretell a great risk for mankind in the future How fast should social restrictions be eased in England as COVID-19 vaccinations are rolled out? Prime Minister asked those who could not work from home to go back to work, avoid public transport if possible; and encouraged "unlimited amounts" of outdoor exercise, and allow driving to outdoor destinations within England. "Stay at Home" slogan changed to "Stay Alert".HI SPRING Outdoor sports amenities and outdoor non-food markets reopen. Prohibitions on leaving home replaced by a prohibition on staying overnight away. Gatherings from more than one household limited to six people outdoors and are prohibited entirely indoors, with exceptions including education. Primary schools reopened and secondary followed on 15th Most lockdown restrictions lifted. Hospitality businesses permitted to reopen. New health and safety guidance on operating businesses "COVID securely" published. Gatherings up to 30 people legally permitted, still recommending people avoid gatherings larger than six.