Read on Twitter
I've recently come across a disinformation around evidence relating to school closures and community transmission that's been platformed prominently. This arises from flawed understanding of the data that underlies this evidence, and the methodologies used in these studies.
Let's look at the paper being cited here. This study published in Nature Human Behaviour examined >50,000 interventions (at fine scale) across >200 countries. The number of countries examined allowed examination of these in depth. https://www.nature.com/articles/s41562-020-01009-0
As the authors state, the key strength of the study is that examination across so many different contexts allowed a disentanglement of interventions. The authors used four approaches, including a case-control analysis to specifically deal with this issue.
Reassuringly all four methods used reach the same conclusion with respect to ranking. As the authors themselves state, the richness of the data across many countries allows them to disentangle effects much better. The data are available here: https://github.com/amel-github/covid19-interventionmeasures
A quick look through the data shows that there is sufficient separation between different interventions in different countries over time- including school closures to be able to disentangle effects. Examination of multiple interventions also reduces the possibility of confounding
Of course, this is not the only study to reach these conclusions. Here is another study by the Lancet examining school closures & openings over 131 countries that reached the same conclusion. https://www.thelancet.com/journals/laninf/article/PIIS1473-3099(20)30785-4/fulltext#:~:text=On%20the%20basis%20of%20the,introduction%2C%20compared%20with%20the%20day
Such modelling studies have been consistently used to examine the impact of non-pharmaceutical interventions on R. This is valid methodology, and generally authors have been clear on where data were too limited to do this, and where data allowed effects to be disentangled.
The impact of school closures on transmission is now widely accepted, including by SAGE, as highlighted in this document.
Ironically, the same scientists who cast doubts on robust global evidence on school closures have often quoted flawed studies on symptom-based testing to claim that children & schools don't contribute substantially to or 'drive' transmission.
Much of the evidence cited to back-up claims around lack of schools 'driving' transmission is based on one or more of the following:
1. Symptom-based studies that underestimate transmission in schools & from children who are likely to be asymptomatic index cases & not identified
1. Symptom-based studies that underestimate transmission in schools & from children who are likely to be asymptomatic index cases & not identified
2. Studies where background transmission has been very low, or mitigation in schools has been in place or both
3. Studies in households at a time schools were closed
4. Studies in schools at a time school attendance was low (see PHE study)
3. Studies in households at a time schools were closed
4. Studies in schools at a time school attendance was low (see PHE study)
We need robust scientific discourse that is based on the evidence presented. It is valid to critique evidence of certain forms, but this needs to be done with sufficient scrutiny of detail and a deep understanding of methodology and data.
Just want to clarify that I've included a screenshot image rather than tagging the author of the tweet directly - not to avoid a discussion around this - but rather because I was blocked by her previously when I challenged her claims around children transmitting less than adults.
