select search filters
briefings
roundups & rapid reactions
before the headlines
Fiona fox's blog
The Office for National Statistics (ONS), London School of Hygiene & Tropical Medicine (LSHTM), and Public Health England (PHE) have released the first early findings from a jointly lead study designed to better understand the spread of COVID-19 in schools in England.
This Roundup accompanied an SMC Briefing.
Dr Alasdair Munro (Clinical Research Fellow) and Prof Saul Faust (Professor of Paediatric Immunology and Infectious Diseases), NIHR Southampton Clinical Research Facility, University Hospital Southampton NHS Foundation Trust and University of Southampton, said:
“The results from the ONS school infection provide further evidence towards what is currently known on the prevalence of COVID-19 in school age children;
“This is consistent with findings from studies in English schools in the Summer term during the partial reopening (Ismail et al https://www.thelancet.com/journals/laninf/article/PIIS1473-3099(20)30882-3/fulltext).
“Reassuringly, in that study no cases were found in primary schools in low prevalence areas.
“New insights include the variation between schools in infection prevention measures. Whilst many measures are difficult to implement logistically, there does appear to be room to improve on the uptake of some less disruptive mitigation strategies to reduce risks.
“Other points to note are:
Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:
“It’s good that this work is being done, but it’s still at an early stage. The results in this first release can’t yet tell us anything very clear. I don’t find it at all surprising that, in broad terms, the results indicate that the pattern of infection in schools (in both pupils and staff) matches that in the community in the areas where the schools are, with higher estimated infection rates in schools with high community prevalence, and lower rates in schools with low community prevalence. After all, the people (staff and pupils) in schools are very much part of their local community, and it would be very strange if, on average, the pattern of infection in schools looked very different from the pattern in the broader community. Overall, I think this new survey is much more important in terms of what it promises in the way of good information in the future, than in terms of what it has found so far.
“There are several important limitations of the work so far, and generally these are made clear in the bulletin and its accompanying methodology document. The survey involved more participants in high prevalence areas of England than would be needed to represent the school populations in those areas – in the jargon, it ‘oversampled’ in those areas. There’s nothing wrong with that, because the progress of infections in these areas would generally be more important to public health, so it’s particularly important to get reasonable precise estimates there – and the oversampling does not interfere with comparisons between high and low prevalence areas. However, this (and other differences between the schools that were sampled and schools in England overall) means that the results so far can’t be used to give estimates of infection rates in schools across the whole country. A statistical process called weighting will, we’re told, be used in later bulletins, and that can allow for this kind of unrepresentativeness.
“The survey couldn’t use data from pupils or staff who were absent from school because they had a coronavirus infection, or had symptoms that might indicate an infection, so it can’t therefore give estimates of infection rates that include all the schools staff and pupils (present or absent). At this stage I can’t say how much effect that would have on the estimates. The survey results can’t yet say anything about transmission of the virus within and from schools – that needs there to be comparison of tests over time, so an exploration of transmission is promised in later rounds when more tests have been done. Though the survey design involved antibody tests, detecting evidence of past infection, as well as swab (antigen) tests that detect evidence of current infection, no antibody results have been published yet. The survey estimates the percentages of pupils and staff who would test positive on a swab test for current infection – this isn’t quite the same as estimating the percentages who are currently infected, because of the possibility of false positive and false negative test results. That does complicate the interpretation of the results slightly, but I don’t think it’s a major issue here, and it won’t interfere much with looking at trends over time when we have more data in later bulletins.
“There are also general limitations that would apply to estimates of just about anything from surveys. The participants (pupils and staff) who choose to take part in the survey may be different from those who choose not to, and that could mean that the results are not adequately representative of what would be found if everyone eligible took part. Since, obviously, we have no test results from the people who chose not to take part, it’s not really possible to know how much of an issue this really is. It might be a more important problem for pupils than staff, because the overall response rate (the percentage of eligible people who chose to take part) was considerably higher for staff (51%) than for pupils (19%), though that does depend on the reasons why people chose not to participate. There are related issues about how representative the eligible people are of all people (staff and pupils) in schools. I’ve already mentioned that staff and pupils who were absent couldn’t be tested, and if they were absent because of an infection, that might be an issue. Also, in secondary schools, students from year 11 were not eligible, because it was decided that taking part would be too disruptive to students taking public examinations such as GCSEs this year (and most of those would be in year 11). If infection and transmission patterns are very different in year 11 compared to other school year, that might be an issue.
“Finally, any results from a survey are subject to statistical uncertainty, simply because estimates are being given for groups of people who were not all included in the survey. The results include confidence intervals that indicate the plausible range for the true positive rates, around the central estimates. Since this is the first round of the survey, most of those confidence intervals are pretty wide, and this generally means that the differences between different areas and different groups of people (pupils, staff, primary and secondary schools) are, in the jargon, not statistically significant. It’s important to know that that does not mean there is no difference. It means that it remains plausible that there’s no difference (for example) between infection rates in pupils and staff, but that’s only one possibility – there could in fact be a difference as big as the difference between the central estimates, or even bigger. It’s simply that we don’t yet have enough data to be sure what the position is. Future rounds of the survey might clarify things.”
Prof Mark Woolhouse, Professor of Infectious Disease Epidemiology, University of Edinburgh, said:
“This report from the Schools Infection Survey is very welcome.
“The report gives estimates of the prevalence of COVID-19 infection in pupils and staff for 17 days in November. School-aged children often show no symptoms of infection so, because participants were tested regardless of symptoms, it gives a more accurate picture of levels of infection than routine case reporting.
“We already know that schools are safe for the vast majority of children and that school staff are at no more risk than any other profession. This survey helps us understand the likely contribution of schools to the wider epidemic.
“One in 88 participants were positive on the day of testing. There was very little difference between pupils and staff. Prevalence was higher in secondary than primary schools, confirming a pattern seen in other surveys. Prevalence was higher in high than low prevalence areas, which the authors interpret as schools ‘mirroring’ levels of infection in the wider community. We must remember that, of course, pupils and staff do not spend all their time in schools and may become infected elsewhere, not least within their own households, so this is a reasonable interpretation of the data.
“I agree with the authors’ interpretation. Though there have been many COVID-19 cases in schools there is little evidence that schools are driving the epidemic, as some have suggested they might. Schools stayed open during the recent lockdown in England but levels of infection fell quite sharply. This is consistent with experience in other countries and is further evidence that schools are not driving the epidemic.
“Surveys like the SIS are an important tool for monitoring and understanding the COVID-19 pandemic. Continued data collection will help us make evidence-based decisions about the best course of action in challenging times.”
https://www.ons.gov.uk/releases/covid19schoolsinfectionsurveyround1englandnovember2020
All our previous output on this subject can be seen at this weblink:
www.sciencemediacentre.org/tag/covid-19
Declared interests
Prof Kevin McConway: “I am a Trustee of the SMC and a member of the Advisory Committee, but my quote above is in my capacity as a professional statistician.”
Prof Mark Woolhouse: “NO COIs to declare.”
None others received.