select search filters
roundups & rapid reactions
before the headlines
Fiona fox's blog

expert reaction to reanalysis of model used for Imperial Report 9 and impact of school closures

An analysis, published in The BMJ, looked at the Imperial College London Report 9 and the impact of school closures.


Prof Mark Woolhouse, Professor of Infectious Disease Epidemiology, University of Edinburgh, said:

“Rice et al. use an epidemiological model developed by a team at Imperial College, London to explore the impact of school closures on the dynamics of COVID-19 in the UK.

“The headline result – that school closures in March could lead to a larger death toll over time – is intriguing and may cause some concern. In fact, this result applies to a specific, and probably unrealistic, scenario and should not be interpreted as a prediction. Nonetheless, this counterintuitive result does shed some light on the current debate about whether to allow herd immunity to build up in age groups that rarely experience severe consequences of infection.

“The key feature of the modelling exercise is that having flattened the first wave using social distancing measures (“lockdown”) the interventions are lifted and a large second wave results. In reality, measures are being reimposed in response to the second wave, so the scenario is unrealistic.

“In the scenario modelled the reason school closures lead to a net increase in deaths is because during the period when the most vulnerable were advised to protect themselves there is no build-up of immunity among children and the wider community. If measures are then released in all groups at the same time then the virus spreads faster, and more of the vulnerable group die, than if some level of immunity had been allowed to build up in the young.

“Rice et al. go on to make two further points, both important. First, they illustrate that the death rate is particularly sensitive to measures in place to protect the elderly and vulnerable. Second, strategies to minimise the number of cases (which the UK government is currently focussed on) are different from strategies to minimise the number of deaths. Achieving the former can, for some scenarios, compromise the latter.

“Though the modelled scenario may be unrealistic, it does illustrate the general principle that, of itself, lockdown solves an immediate crisis without providing a long-term solution. In trying to keep the epidemic suppressed we risk a recurrent cycle of restrictions of one form or another.

“Rice et al. also make the point that these results were available to policy makers when this model was first published on March 16th. The implication is that it was known then that school closures could have adverse impacts on the course of the epidemic beyond the first wave.

“Perhaps the key lesson that should be drawn from this study is the warning that if we allow short-term thinking to dictate our response to COVID-19 then we may not make the best decisions for minimising the public health burden over the longer term.”


Prof Paul McKeigue, Professor of Genetic Epidemiology and Statistical Genetics, University of Edinburgh, said:

“ The conclusion of this paper — that closing schools would slow the epidemic but increase the number of deaths among older adults in a second wave — is consistent with the general theme of “focused protection” advocated in the Great Barrington Declaration released this week.“

“ The authors bring out a key point that was not widely appreciated in March — the Imperial College modellers and SAGE reports explicitly warned that the lockdown would only slow the epidemic so as to prevent ICU capacity from being overwhelmed as they predicted, and that on-off lockdowns would have to continue for up to 2 years to ensure this.   Unless a vaccine became available, the only deaths that would be prevented by lockdowns would be the extra deaths resulting from the predicted overload of the health service.

“The methods and conclusions of this study are sound, given that they used the Imperial College model.  However there have been criticisms of the Imperial College model that it gives predictions of epidemic size that are too high.”


A spokesperson for Imperial’s COVID-19 Response Team, which produced Report 9, said:

“This provides further independent confirmation that Imperial’s modelling in March was robust, reproducible and sound in its conclusions. We welcome this independent analysis of Report 9 as we continue to advance our understanding of the early epidemic.”


Prof Matt Keeling, Professor of Populations and Disease, University of Warwick, said:

“This is a solid and well written paper, and I have no problems with the basic science – but I feel the press release often takes elements of the paper out of context, in a situation where context is all important.

“1) The paper perpetuates the myth that it was the Imperial model alone that was responsible for the lock-down. There were many modelling groups that were involved with the analysis of available data in the UK and extrapolation of the growing epidemic, all of whom were feeding into SPI-M and SAGE. To suggest that the lockdown was due to one group in particular is to under-value the work done by SPI-M and SAGE to achieve a scientific consensus.

“2) It needs to be stressed that all the simulations assume that interventions are only in place for 3 months (18th April – 18th July) and then completely relaxed. This gives rise to a strange set of scenarios where a second wave is allowed to progress in an uncontrolled manner.

“It is this that leads to the counter-intuitive headline finding “that school closures would result in more overall covid-19 deaths than no school closures” – actually what the authors find is that a short period of intense lock-down (including the closure of schools) leads to a large second wave if it is allowed to run with no controls.  To be fair the authors do highlight this in the paper, but it is not in the reported press release.

“This also leads to the strange conclusion that “while general social distancing would reduce the number of covid-19 cases, it would increase the total number of deaths compared with social distancing of over 70s only”. Again this is all about the magnitude of the uncontrolled second wave. I’d also point out that social distancing of the over 70s (a much talked about policy) is extremely difficult, many of the most vulnerable require the care of other members of the population; while it is easy to switch off this component in a model achieving this is practice is far more difficult.

“3) While it’s reassuring to have the results of a complex simulation model confirmed (although it is unclear if the authors have simply re-run the code and analysed the results or whether they have sought to validate the details and assumptions within the model framework), I’m unclear what the paper adds to the current public health crisis facing the UK.  We are now at a time where most measures of the epidemic are increasing exponentially; urgent and drastic national-scale action is required if we are to re-establish control – we should be looking forwards not backwards. I would like to see predictions of how to control the unfolding second wave, rather than an analysis of historical options.

“In this context, I’m concerned that some of the statements such as “adding … university closures to other measures would increase the overall total number of covid-19 deaths compared with no closures” will be taken out of context, and applied to the current wave, where universities appear to be hot-spots of infection that can spill into the surrounding community.”


‘Effect of school closures on mortality from coronavirus disease 2019: old and new predictions’ by Ken Rice et al. was published in the BMJ at 23:30 UK time on Wednesday 7 October.

All our previous output on this subject can be seen at this weblink:


Declared interests

Prof Mark Woolhouse: No conflicts of interest. Prof Woolhouse is a member of SPI-M

Imperial’s COVID-19 Response Team: The Imperial COVID-19 Response team produced Report 9 which is reanalysed in this paper.

Prof Matt Keeling: Prof Keeling is a member of SPI-M

None others received.

in this section

filter RoundUps by year

search by tag