This feature is the main article in the 15 May issue of Tes magazine. Because of the huge importance of the issue to our readership and the wider public, we have decided to put it in front of our paywall and make it free to all readers. 

“Guided by the science”: it’s a phrase designed to cut through political tribalism, to settle nerves and, perhaps, to shield ministers from blame. It’s a phrase we have heard at every briefing, in every interview with our politicians and in every guidance document published since the coronavirus outbreak was labelled a pandemic. It’s a phrase that has been wielded to justify schools staying open and closing, and, increasingly, to justify their return.

The UK’s prime minister reached for it on 18 March, when announcing that schools in England would be closed indefinitely, except to children of key workers and the most vulnerable pupils. The government would “take the right steps at the right time, guided by the science”, said Boris Johnson.

Yet he had also insulated himself with the same phrase six days earlier when announcing that the government, at that point, was not closing schools. The government had been “guided by the science” and “the scientific advice is that [school closures] could do more harm than good at this time”, Johnson said.

“The science” changes fast, apparently.

And when Johnson announced last week that the reopening of primary schools in year group stages could begin by 1 June, depending on national progress on transmission rates and infection numbers, he took things up a notch: the government was “going to be driven by the science”.

Where the research reaches a consensus is that children are unlikely to become ill as a result of the novel coronavirus, unless they have a pre-existing health condition (as confirmed by the government in its latest guidance to schools published 11 May). This is one reason why the timing of school closures across the world - more than 100 countries had opted for closure by mid-March - has been so variable: other factors, not just pupil welfare, have driven the decision.

Those who work in schools have repeatedly asked whether their own health has been one of those deliberated factors: are school staff - teachers, teaching assistants, office staff, business managers, lunchtime supervisors, caretakers and more - safe? And what about the wider school community - the parents, grandparents, carers and others?

There have been no clear answers. So, as ministers continue to pledge to be “guided by the science” with schools around the world opening their doors once more, what can the scientific evidence gathered so far tell us about the likely impact of school closures and reopenings on transmission of the novel coronavirus, and about the potential risks for those who work in schools?

Every pandemic is different - that is one important thing to acknowledge. But research on the 1918 influenza pandemic - the “Spanish Flu” - for example, can still shine a light on the policy being formed in response to the novel coronavirus pandemic. What we find looking back at this historical research is a reason why school closures have become an established tactic in pandemic-response plans - and it’s not because of fears for staff safety.

The evidence base on the effectiveness of school closures comes from the impact this has had on countering the spread of influenza. One key study in the field looked at the varying responses in 43 US cities during the 1918 influenza pandemic. The paper, published in the Journal of the American Medical Association in 2007, found that school closure and public gathering bans were the most common combination of “non-pharmaceutical intervention” (NPI) deployed, ranging between one and 10 weeks in duration. This combination was “significantly associated” with reductions in weekly excess death rates.

J Alexander Navarro, assistant director of the University of Michigan’s Centre for the History of Medicine, one of the authors of that paper, emphasises that the research did not examine the effects of school closures as an individual NPI. But, he adds, it “concluded that the early, layered and sustained use of NPIs in 1918”, including school closures, “was associated with lower peak and overall morbidity and mortality”.

Meanwhile, a review of scientific evidence on school closures during influenza pandemics for the UK’s Department of Health in 2014 judged that “school closure may form a useful component of a mitigation strategy during pandemic influenza, but the timing and duration of closure needed to produce an effect is unclear”.

Influenced by this research, most countries affected by the novel coronavirus have opted to close schools as part of a range of social distancing measures aimed at reducing wider community transmission of the virus and, thus, avoiding breaching capacity in critical care units.

But does the evidence base about the effectiveness of school closures from influenza outbreaks hold true for the Covid-19 pandemic?

At the time of writing, the most detailed published summary of scientific advice to the government on school closures and the coronavirus remains the four-page “consensus view”, dated 19 February, from the Scientific Pandemic Influenza Group on Modelling (SPI-M-O). The group advised the government that the likely impact of closures in mitigating a coronavirus outbreak in the UK would be “very uncertain”.

Why would it be uncertain, considering the evidence from past pandemics? With influenza, children can become ill and there is also clear evidence to show they are key transmitters of infection. There is less clarity about the role of children in coronavirus transmission.

A study from Iceland, which tested around 13,000 people in the country’s wider population, has drawn much attention. The study, published in the New England Journal of Medicine, found that none of the 848 children under the age of 10 who were part of the study tested positive for the coronavirus, compared with 100 of the 12,232 individuals aged over 10 who tested positive (equating to 0.8 per cent of the total sample).

However, many feel the evidence is not yet conclusive. “Unfortunately, I don’t think we are any clearer on the role that children play in the community transmission of the virus - not enough of those studies have been done,” says Jonathan Ball, professor of molecular virology at the University of Nottingham.

While the evidence suggests that children usually suffer more mild symptoms than adults, “we also know that people with mild symptoms can shed lots of virus from their throat and their nose”, adds Ball.

He continues: “Until proven otherwise, I think it is sensible to consider that children have the potential to be major spreaders of the infection.”

Backing up that hypothesis, German researchers, led by virologist Christian Drosten - described as the German federal government’s “go-to expert” on the coronavirus - recently analysed the amount of virus harboured by children and adults testing positive, finding no significant differences between age groups.

“Based on these results, we have to caution against an unlimited reopening of schools and kindergartens in the present situation,” Drosten’s team said in their paper. “Children may be as infectious as adults.”

The chief medical officer for England, Chris Whitty, took a more cautious view in the government’s daily briefing on 27 April. In answer to a question, he responded that, while scientists are “still really learning” how children influence “the actual spread of this virus”, pupils being at school “certainly contributes” to an increased rate of infection.

Meanwhile, a review of existing research on children and the coronavirus, such as the Iceland study, published in the BMJ’s Archives of Disease in Childhood, has suggested that “children do not appear to be super-spreaders” and thus that “governments worldwide should allow all children back to school”.

So there is little agreement. But where there is less doubt is that adults in a school can transmit the virus to each other. So if schools were to remain open, then social distancing between staff, between parents and carers, and between anyone coming in and out of the school would be essential. Many teachers will say that would be near impossible.

But rather than looking at that particular issue, the scientific evidence gathered to date has focused on the role of school closures in reducing wider community transmission rates - the key aim of such policies.

Yet even on that, there is some uncertainty. In an attempt to collate and evaluate research on the effectiveness of school closures during coronavirus outbreaks present and past (the 2002-04 Sars outbreak was an example of an earlier epidemic), a group of UK academics conducted a “rapid systematic review” of existing studies, published in The Lancet Child and Adolescent Health on 6 April.

The review highlighted modelling on the potential spread of the coronavirus by researchers at Imperial College London, led by Professor Neil Ferguson, which was influential in persuading the UK government to opt for a strict lockdown, including school closures, in March.

Such modelling studies are predictions based on evidence: looking at the likely impact of NPIs on mortality rates and demand for healthcare services, using existing data about numbers of social contacts across different age groups and on coronavirus transmission rates. The Imperial modelling assumed that a third of transmission would take place in schools.

But the Lancet review said the Imperial modelling predicted that “school closures alone would prevent only 2-4 per cent of deaths, much less than other social distancing interventions”, and that “other less disruptive social distancing interventions in schools” should be considered “if restrictive social distancing policies are implemented for long periods”. Ferguson, director of the MRC Centre for Global Infectious Disease Analysis at Imperial, has since clarified that, while school closure “as a measure on its own is predicted to have a limited effectiveness in controlling Covid-19 transmission, when combined with intense social distancing it plays an important role in severing remaining contacts between households and thus ensuring transmission declines”.

The Lancet review generated plenty of media coverage around the world, with The Sun offering a version of the review’s findings that was typical of many publications: “Closing schools has a tiny impact on the spread of coronavirus but a harmful effect on kids and society.”

That interpretation is something of a stretch, given that the review found there was “no data on the relative contribution of school closures to [coronavirus] transmission control” from China and Hong Kong - at that time the main places in a position to potentially provide such evidence (as distinct from modelling studies making predictions).

One of the Lancet review’s authors, Chris Bonell, professor of public health sociology at the London School of Hygiene and Tropical Medicine (LSHTM), observes that when it comes to the impact of school closures, “the key finding from the review, I would say, is the lack of evidence”.

He adds that there are “plausible reasons why the impact of school closures would be less than they would be with influenza. But that’s not to say there’s no impact”.

Further research, beyond that examined in the Lancet review, has suggested similar. One of the key UK modelling studies looking at the predicted impact of a range of NPIs, compiled by researchers at the LSHTM, was presented to SPI-M-O in February and published in April (not until after the Lancet review appeared). That modelling “judged that, based on the prevailing evidence, school closures were likely to have a relatively small impact on reducing the burdens of Covid-19”, given that children appear to have a relatively small role in transmitting the virus, says Nick Davies, research fellow in mathematical modelling at the LSHTM, who led the research. “However, school closures are still predicted to have some impact, and closing schools was the right move,” he adds.

Have we actually got any evidence from the current pandemic that steers us towards whether all this modelling is correct or that gives us firmer data? There is a tentative indication from Australia, where the National Centre for Immunisation Research and Surveillance published findings from an investigation of all Covid-19 cases in New South Wales schools on 26 April. The New South Wales government has kept schools open but encouraged parents to keep their children at home.

This report, described as “preliminary” and not yet subjected to peer review (and that has come under some criticism), said that between March and mid-April, 18 individuals (nine students and nine staff) from 15 schools were confirmed as Covid-19 cases.

The investigation traced 735 students and 128 staff who were “close contacts” of these initial cases. Among those, “no teacher or staff member contracted Covid-19 from any of the initial school cases”, while “one child from a primary school and one child from a high school may have contracted Covid-19 from the initial cases at their schools”.

The findings “suggest that spread of Covid-19 within NSW schools has been very limited”, the researchers said.

The usefulness of this data is debatable and there has been no comparable published study in England, where education secretary Gavin Williamson has convened a subgroup of the government’s Scientific Advisory Group for Emergencies to ensure that “when schools are open it’s done in the best possible way with the best scientific and medical advice”.

But with that science obviously unclear, many staff remain fearful about the prospect of schools reopening more fully. Though the Imperial modelling suggested that school closures alone would prevent “only” 2-4 per cent of deaths, in the words of the Lancet review, teachers argue that for them the impact would be huge: would they potentially be at risk of being in the 2-4 per cent?

“It is of paramount importance that staff, pupils and parents feel confident that it is safe to be in school,” stresses Geoff Barton, general secretary of the Association of School and College Leaders. Key questions that must be addressed are around how to maintain social distancing in schools, whether personal protective equipment will be provided, and how potential phased reopening with different groups of children in school at any one time could be introduced, he adds.

Paul Whiteman, general secretary of school leaders’ union the NAHT, agrees: “Schools should only reopen when the scientific evidence is clear that it is safe to do so. Safe for pupils, safe for staff, safe for parents.”

The safety of the schools’ workforce needs to be a much more detailed consideration - not just for the protection of individuals and their families, but because with around 700,000 staff employed in UK schools, that workforce is of a size where its health in a pandemic matters for the wider population.

The latest government guidance says “limiting the numbers of children going back to school and college initially then gradually increasing numbers … reduces risk of increasing the rate of transmission.” But the truth is that, without detailed knowledge about the role of children in transmission of the virus, we may not know for some time what the true level of risk to teachers is. And even when schools go back, we will still struggle to judge the impact in any changes to wider transmission rates.

Benjamin Cowling, professor and head of the University of Hong Kong’s Division of Epidemiology and Biostatistics, lead author on a paper looking at the impact of social distancing measures against coronavirus in Hong Kong, says that with so many NPIs implemented at the same time, “it’s very difficult to say whether school closures have been important, or whether it’s the other things that have been more important”.

Samir Bhatt, a senior lecturer in Imperial College London’s Department of Infectious Disease Epidemiology, was one of the authors, alongside Ferguson, of a paper estimating the impact of NPIs on Covid-19 infection rates across 11 European nations. Similarly to Cowling, he says that the interventions were “all implemented so rapidly that it’s hard to distinguish which had the most impact. Therefore, from our model, we cannot say if school closure had a big impact individually, just that all interventions had an impact”.

However, another team of UK researchers have published a more recent paper that tries to determine which lockdown measures have been most effective in reducing the virus’ spread, by looking at case and death numbers across 30 European countries and comparing these to the start dates of different NPIs in those countries.

They argue that a range of lockdown measures have been introduced in stages by different nations, making it possible to judge the impact of individual measures - thus enabling them to go beyond the forecasts of mathematical modellers by looking at results.

The research suggests that closure of educational institutions had “by far the biggest impact” of any individual lockdown measure across those European nations, says Paul Hunter, professor in medicine at the University of East Anglia’s Norwich Medical School, lead author on the paper.

But there are caveats: the results are preliminary, plus association between school closures and reduced death or case rates does not necessarily mean school closures caused those effects, cautions Hunter. The study also judges closure of primary schools, secondary schools and universities as a combined measure, without being able to determine which had most impact.

There is nothing in the research that can offer any explanation as to why closing educational institutions could be effective - whether the key is the resulting reduced transmission between children, between children and teachers, or among teachers and parents.

But the paper’s conclusions are echoed by another recent study on the varying impact of NPIs, published in the Canadian Medical Association Journal, which looked at case numbers across 144 nations and regions and found a “strong association” between school closures and reduced growth in the pandemic.

“I think the weight of evidence is pointing towards schools being important” in reducing the spread of the virus, says Hunter.

The study he led, like so much other research on the novel coronavirus, has been published as a preprint, meaning it has not yet been subjected to peer review by other scientists. Researchers are rightly rushing to get their results out given the urgency of the pandemic and the need to inform policy responses. But as a consequence, results are being published without the usual quality checks and before they have been evaluated by the scientific community.

As we have seen, many of these preprint studies on the impact of school closures and potential virus transmission among children present conflicting results.

Scientific consensus often takes years to form on an issue and this will be no different.

So when UK government ministers talk about being “guided by” or “driven by” the science on lockdown measures, what they are talking about is the government’s information from its scientific advisers - which is often reliant on mathematical modelling making predictions, and which the government has been criticised for failing to release in full.

There is no “the science” in any broader sense, no single point of scientific consensus on the issue of the impact of school closures, because the novel coronavirus is still too novel for sufficient peer-reviewed evidence to have been gathered.

“That’s the problem,” says Julii Brainard, senior research associate at Norwich Medical School, another author on the NPI impact paper alongside Hunter. “All these decisions [on lockdown policies] have to be made with incomplete, rather imperfect information.”

Despite uncertainty over the risks to staff, schools are reopening across the world. Primary school teachers in Denmark were some of the first to return, but the way they operate is now very different. Pupils are kept in small groups and do not stay in school for the whole day, teaching takes place outside wherever possible, and children wash their hands at least once every hour and a half, according to Dorte Lange, vice-president of the Danish Union of Teachers.

The Danish labour market regulation model, based on collective agreements between unions and employers, ensured a “trustful reopening” with the aim of ensuring that “teachers are safe”, she adds.

For example, Lange continues, negotiations between unions and the authorities have ensured that teachers with existing health conditions, such as asthma and diabetes, can seek a consultation with their doctor - and the doctor’s advice to the individual teacher on returning to work must be respected by school employers.

In the UK, too, a “collective” approach to reopening schools will be essential, says Bonell. He stresses that any mechanisms for school reopening must be “cautious” and developed with teachers: “It can’t be imposed on teachers - that just won’t work.”

He argues that there has not been enough gratitude to teachers, who are stepping into schools every day despite all the unknowns laid out here - unknowns that will remain when eventually they do return to schools.

“We have the NHS clap,” Bonell says. “I think we also have to recognise the fact that teachers are doing a great thing here by keeping these schools open for the most vulnerable kids.”

John Morgan is a freelance journalist

This article originally appeared in the 15 May 2020 issue under the headline “The science behind school reopenings”