X Close

What’s the real cost of sending kids back to school?

A member of staff wearing PPE takes a child's temperature at the school gates. Credit: Getty

December 30, 2020 - 3:32pm

Children are supposed to be going back to school soon. In our borough, primary-age children are expected to return on 4 January; there is to be a staggered return for secondary schools, but they should all be back by the 11th. Reports suggest it won’t happen, and that Tier 4 areas are going to learn their schools are to be shut, but Michael Gove says he is still “confident” they will remain open.

But should it? Just before Christmas, the London School of Hygiene and Tropical Medicine released modelling, taking into account the new B.1.1.7 variant of Covid that has been spreading in the South-East and elsewhere. Their central finding was that Tier 4 restrictions as they stand will not bring down the R value to below 1 — and, in fact, that even a national lockdown like that in November would not do it, “unless primary schools, secondary schools, and universities are also closed”, until at least 31 January.

It’s worth going into a bit more detail, though. The total forecasted deaths between December and June, under a “Tier 4, schools and universities open” scenario, is 107,000 (95% confidence interval: 98,300 – 118,000); under “Tier 4, schools and universities closed”, it’s 102,000 (90,800 – 115,000), a difference of 5,000.

Estimated transmissibility and severity of novel SARS-CoV-2 Variant of Concern in England, London School of Hygiene and Tropical Medicine

Let’s not belittle that. It looks like a small detail, because we’re so inured to awful news and because the numbers are so big already, but using those central estimates, we are talking about 5,000 lives, and probably more than 50,000 years of life, saved or lost. If it was 2,000 vs 7,000, we’d be much more concerned, but the actual number of lives saved would be the same. But it does suggest that schools are not one of the main drivers of transmission, if together with universities they cause less than 5% of the deaths.

I spoke briefly to Sarah Lewis, a professor of genetic epidemiology at Bristol, and Sunil Bhopal, a paediatric epidemiologist at Newcastle. Lewis felt that the estimates the model gave are very high — she points out that if we vaccinate 2 million a week, we’d have vaccinated the country’s 12 million over-65s by the middle of February, and since they account for 90% of the deaths it’s hard to see where 36,000 come from. And both pointed out that there are significant costs to keeping children out of school, in terms of learning loss, mental health risks, risks of exploitation and abuse. The children themselves are still at almost zero risk from the disease: the age-adjusted mortality data for the new variant, says Bhopal, looks much the same as for the old one.

Perhaps keeping children out of school until the end of January is not a disaster, for the price of saving 5,000 lives. But Bhopal makes the case that these tradeoffs need to be made explicit: what is our best estimate of the costs? How many years of schooling lost are worth a year of life saved? What will be our trigger points, in terms of cases per 100,000, for closing and reopening schools? What measures should we take to mitigate harms while schools are open? These questions need to be openly asked and answered.

What really makes a difference under this model, incidentally, is how quickly the vaccine is rolled out. The model adds either a 200,000-vaccinations-a-week scenario or a 2-million-vaccine-a-week scenario to its most stringent scenario above, and finds that under the 200k/week scenario, total deaths drop to a central estimate of 83,300 (73,800-93,800); under 2m/week, it’s 35,700. Schools, according to this modelling, have a non-negligible impact, but it’s vaccines, and how fast we can get them out to people, that will be the key difference.

Of course, it makes a difference to the calculations over whether to close schools, as well. The vaccine’s arrival means that we are no longer simply postponing Covid deaths but avoiding them; the costs of closing down are lower and the benefits are greater. It may well be that, given what we know and don’t know now, keeping schools closed until February is the right choice. But I would like to see the reasoning made explicit, and soon, because we’ve only got a few days before they go back, and parents need to plan. Hopefully this upcoming announcement will make things clear, because rumour and anonymous briefing is not the same as clear communication.


Tom Chivers is a science writer. His second book, How to Read Numbers, is out now.

TomChivers

Join the discussion


Join like minded readers that support our journalism by becoming a paid subscriber


To join the discussion in the comments, become a paid subscriber.

Join like minded readers that support our journalism, read unlimited articles and enjoy other subscriber-only benefits.

Subscribe
Subscribe
Notify of
guest

25 Comments
Most Voted
Newest Oldest
Inline Feedbacks
View all comments
Johnny Sutherland
Johnny Sutherland
3 years ago

Is it just me or are others getting fed up of “computer model produces big number” = PANIC?

Tim Diggle
Tim Diggle
3 years ago

No, you are not alone.

I would be interested to see a rigorous comparison of the various forecasts with the subsequent outcomes. My overall feeling when looking at the ONS historic figures is that almost none of the outrageous “forecasts” presented to the public have ever proved to be remotely reliable. As they all seem to predict doom I am beginning to wonder about propaganda …..

Johnny Sutherland
Johnny Sutherland
3 years ago
Reply to  Tim Diggle

I don’t think its necessarily propaganda. When I was still working in manufacturing rather than recruiting for it I developed a number of computer models. Firstly people build their assumptions into the model in how things are calculated (I remember one blazing row with my boss over “when you do that it doesn’t mean this happens”; then there are the input assumptions eg the R number and IFR finally there’s the ability of the recipients to understand the output.

I’m not sure what the problem is with this lot.

LUKE LOZE
LUKE LOZE
3 years ago

Anyone who takes the long term modelling seriously needs to have their heads examined.

7882 fremic
7882 fremic
3 years ago

Making young unemployable is the worst crime a society can do. Letting people take the risk of a illness is chance, stopping schools is intentional. Studies always show poor students rarely catch up from missed school. A unemployable youth will sit around idle, and this always leads to two things when there is no constructive outlet for youthful energies (study and school being the proper way), they turn to anti-social or self destructive behaviors. Young men must burn off their energies constructively or they will do it destructively.

When you destroy the future of a youth you also wreck part of society which has to support and tolerate them. But you also wreck their life, and the life of those related to them.

Stopping school is criminal! That is is even considered is amazing, so destructive is it to the people, families, society.

Ben Scott
Ben Scott
3 years ago
Reply to  7882 fremic

Well said

Michael Dawson
Michael Dawson
3 years ago

My concern is that the absolute priority must be the vaccination programme, as Tom says. I don’t know what the potential constraints are on getting to 2m vaccinations a week very quickly, but if manpower is one then using the army to help with testing of pupils and students would be a criminal mis-use of their time.

Alex Camm
Alex Camm
3 years ago

The excess death rate for the year is @ 30000+
Deaths from covid is currently given @ 70000.
This means that miraculously @ 40000 people did not die from other illness or the number of ‘covid’ deaths is inaccurate.
If this number is compared to the past ten years or so then it is not out of the ordinary.
With an aging population we should expect the death rate to be higher irrespective of other csuses

Terence Riordan
Terence Riordan
3 years ago

Models are FINE IF they have all the variables and use well thought out and researched range of input values. This is a classic…a major variable for deaths is vaccination levels of likely people who would die. The modelling concept is FINE….most of the modelers are incompetent….cf Ferguson both in input and the appalling IT comprising his model.

Kelly Mitchell
Kelly Mitchell
3 years ago

Oh, no. The models tell us to lock down. Maybe we should go with the actual f*****g data and realize this is not even a real pandemic.

LUKE LOZE
LUKE LOZE
3 years ago

Spend 5 minutes browing coronavirus dot data dot gov dot uk, it’s a suprisingly good site.

One thing that really stands out it that the drop in cases in November is clearly visible, and upto 40%.

The interesting thing is the number of deaths which appears to have peaked in early November and has the slighest < 10% drop perhaps mirroring the drop in cases.

There appears to be a disconnect between case numbers and deaths, why is this? I don’t think there’s anything to cynical at play, but does show the confusion of stats that are driving decision making.

Sue L
Sue L
3 years ago
Reply to  LUKE LOZE

I think case numbers in this case are positive tests rather than actual medical cases.

LUKE LOZE
LUKE LOZE
3 years ago

I wouldn’t mind a relatively cautious approach of shutting
secondary schools and universities for a month. It would be relatively
simple to move courses back 4 weeks and make the time up in the Easter
and Summer holidays. If there was a genuine requirement the next academic year could be extended too, in order to help make up lost time.

I presume the education unions would be fully in favour of such a move, I’m suprised they’ve not mentioned it.

Neil John
Neil John
3 years ago
Reply to  LUKE LOZE

Those of us on the ‘front line’, I’m a Trades Union H&S rep in a Russell Group Uni, would support setting back by several months. It might affect the Chinese who should be starting their courses next week, but how many will be coming now is open to question.

Jonathan Bagley
Jonathan Bagley
3 years ago
Reply to  Neil John

I lecture in a Russell group university. Half our dept intake is Chinese. They will all be continuing their courses in 2 weeks’ time. For my subject, all teaching has been online this academic year. We have too many students (due to the Gov caving in to teacher assessed Alevel grades) rather than too few.

Neil John
Neil John
3 years ago

Your fortunate, we have a large number due to fly in to be on campus next week for their hands-on practical course work, IF they come. Back in September those that arrived from China came fully suited in tyvek hooded overalls, wearing respirators and googles, such was the fear of the then unmutated Wuhan virus.

7882 fremic
7882 fremic
3 years ago
Reply to  LUKE LOZE

The education unions would jump at anything which harms the students, if this last year is an indication of their wishes.

Jim McNeillie
Jim McNeillie
3 years ago

The vaccinating over-65s factor makes the assumption that all over-65s will readily and voluntarily submit to being vaccinated.

Andrew Thompson
Andrew Thompson
3 years ago
Reply to  Jim McNeillie

And if they don’t and they die who will get the blame?

Dave Tagge
Dave Tagge
3 years ago

Lewis and Bhopal over a cogent rebuttal, assuming their estimate of ~2 million vaccinations per week is broadly accurate. (I’m not familiar enough with the UK’s COVID vaccination supply expectations to know if that’s the case.)

More broadly, however, the experience to date of COVID modelling makes me suspicious that the 95% confidence intervals from LSHTM are far too narrow. If reasonable 95% CI’s around those numbers for 6 months of reported COVID deaths, for a given set of assumptions, are more like 50k to 225k – i.e., roughly half to double the central estimate – then it highlights just how much uncertainty is entailed in this modelling. That leads to a reasonable conclusion that focusing on a change of ~5% in the central point estimate is applying false precision to the output of such models.

Neil John
Neil John
3 years ago

I wonder if section 44 employment rights act 1996 will be the way out of this issue?

Neil John
Neil John
3 years ago

As a T.U H&S rep in the Uni sector I get some insight to students and their behaviour. Whilst some have been good and not partying etc a large number have been s-TOO-DENSE too. Freshers in Hall’s had some parties, going out to buy booze whilst infected in several cases, we offered testing on a weekly basis, ~60% uptake in Hall’s and averaged 9 new cases weekly. Then with the travel home window pre-testing the uptake went up, and as we’d been informed by those that live near the s-TOO-DENSE in Hovels of Maximum Occupation were a huge problem, with a large number of positive tests, which IF they’d stayed put might not have been too bad, but they travelled ‘home’, some transported by parents, others on public transport no doubt spreading as they went. The effect of commuter students as infection spreaders is also an issue thats been ignored, the rapid spread from London out into the commuter dormitory areas too. How many parents & grandparents will die as result thus far, how many more locals if the Unis reopen and the entitled s-TOO-DENSE carry on as before…

Hardee Hodges
Hardee Hodges
3 years ago
Reply to  Neil John

The best one can do with those “s-TOO-DENSE” is to throw a huge party in an isolated facility with suitable accommodation. Let them all get ill at once and get over it (screening out those with certain issues). After a week or so no more “s-TOO-DENSE” to worry about.

Neil John
Neil John
3 years ago
Reply to  Hardee Hodges

Indeed, it almost seemed to be what some Universities were trying to achieve, in the hope of ‘herd immunity’.

Colin Haller
Colin Haller
3 years ago

Then there are those who don’t die but merely suffer “long Covid.” Less of that would also be helpful for those individuals AND the burden upon the NHS for the rest.