England is preparing to enter a second national lockdown after scientific advisers warned that coronavirus infections and the numbers of people in hospital are rising steeply in many areas of the country. But what is the science behind the decision? What data prompted the lockdown? The prime minister and two leading advisers, Sir Patrick Vallance, the government’s chief scientist, and Prof Chris Whitty, England’s chief medical officer, made the case for a second national lockdown with a haul of data. They began with the rate of new infections, around 50,000 cases per day, according to the Office for National Statistics, but noted that an infection survey run by Imperial College, London, points to nearer 100,000 cases per day. They showed daily hospital admissions and deaths rising since the end of August. Medium-term projections from the modelling subgroup of the government’s scientific advisers, Sage, found that admissions could surpass those in the first wave by the end of November, with the death rate hitting 800 per day a week later. With no further action, hospital bed capacity would be overrun in December. How robust is the data? The data on infections and deaths is by no means perfect, but it was other data that caused most alarm. The advisers presented new modelling from several academic groups on the number of daily deaths England might expect from now until next summer. Many of the groups, such as Imperial College and the London School of Hygiene and Tropical Medicine, predicted deaths rise to around 2,000 a day by the end of the year. Most striking was modelling from Cambridge University and Public Health England, which suggested 4,000 deaths per day in December. The figure has come under intense scrutiny which has highlighted the provisional nature of many modelling studies. The Cambridge model differed from the others in important ways. While there are claims that the model looked at deaths 60 days after a positive test rather than the more common 28 days, this is incorrect. There is another reason the model was an outlier. The modelling, performed three weeks ago, used data on coronavirus deaths to infer how rife infections were when the people who died became infected some weeks earlier. This infection rate, and the consequent deaths, were then projected forwards to February. The approach meant that the model showed how the epidemic six or more weeks ago might look if it continued to rage unabated. This led to a substantial overestimation of daily deaths. Fed with recent data, the model is far more in line with the others Sage uses, with all estimating roughly 1,000 deaths per day by mid-December if no more is done to contain the virus. The models all point to the number of people in hospital and deaths exceeding those in the first wave without further action. What needs to happen in the next four weeks? Government science advisers called for an immediate “circuit breaker” in September to stop the epidemic spiralling out of control. The three-tier system that was brought in instead was known, even at the most stringent tier 3 level, to be insufficient to bring the R value below 1 and push infections down. The result is a far larger epidemic that will take longer to bring back under control. In the next four weeks, it will be crucial for people to follow the new guidelines, which aim to reduce the amount of contact people have with one another. Scientists will be watching the R value, estimated each week, to look for signs that it is falling, but also infection rates and hospital admissions. How will scientists know if R is below 1? Researchers on the modelling subgroup of Sage use several data streams to estimate the national and regional R values. The number of new infections picked up by hospital and community tests, but also surveys run by the ONS and Imperial College, should be among the first to reveal whether infections are slowing. Other data also feed in though, from phone surveys that quiz people about the number of people they are in contact with, to mobility data compiled by Google and Facebook. Death rates will be the last to show any improvement: there may well be people infected this week who go on to die from Covid after the four weeks have passed. Pushing R below 1 means the epidemic is shrinking, but it is not the only consideration. Scientists will also be watching for the prevalence, or total number of cases, to fall. If the number of people infected drops substantially over lockdown, and if the remaining infections are in those who rarely need hospital treatment, the lockdown could be eased before R is brought below 1. What role will Sage have in the decision to ease lockdown? The Sage modelling subgroup reviews coronavirus data constantly and reaches a consensus each week about the R value and the growth rate of the epidemic. These are passed on to Sage for broader discussion. As with all policymaking, the decision on how and when to ease the lockdown will be for ministers rather than Sage, but the committee will submit proposals for the next steps. Primarily, the experts will want to protect the NHS from being overwhelmed and keep deaths down as winter comes. But other factors will feed into their advice, such as the state of NHS test and trace, and planning for people, including students, moving around the country in the run-up to Christmas.