The nature and scale of Covid transmission in schools is still poorly understood, despite considerable research efforts worldwide. But key decisions, such as closing (or rather moving to online learning)  or opening schools, have to be made despite the scarcity of evidence. So we must make the most of every relevant dataset that does become available, to try and advance our understanding.

Most recently, the Department for Education has released data on teacher absence, for reasons related to Covid. This data spans the period since the October half term and covers all state-funded schools in England, split by education phase and date, and by location.

School leaders and administrators are asked to count separately the number of teachers on roll who are unavailable because:

• they are a suspected case of Covid;
• they are a confirmed case of Covid;
• they are requested to remain at home by the school because of contact within the school;
• they are isolating because of contact outside the school; and
• any other reasons for absence.

We must bear in mind that this is workplace absence data, not (apart from one) medically certified health data. The data and our analysis must be treated with caution.

Trends in teacher absence

The chart below simply shows Covid-related teacher absence week-by-week, by type of school. The rate averages around 4% for secondary schools, 4.7% for primary schools and higher again for special schools.

By contrast, and as we show here, pupil absence over the same period was higher in secondary schools: For example, absence in secondary schools was 28% compared to 15% in primary schools. So here is the first puzzle: teacher absence is higher in primary schools while pupil absence higher in secondary schools.

The next two charts present more detail on the nature of Covid-related teacher absence, separately by primary and secondary schools, and split into the categories noted above.

There are a number of interesting features in this pair of graphs.

First, rates of teacher absence due to suspected Covid are very similar in the two school phases, and the same is true of teacher absence due to actual Covid. This is the next puzzle: we might have expected higher rates in secondary schools as younger pupils are less likely to transmit infection.

Second, these rates vary very little over the half-term, somehow bucking the national trend in overall infection rates for working age people, which changed quite considerably over this period – to see this spelled out, see the chart below.^[1]

Third, the rates of teachers self-isolating following contacts outside school are also about the same between phases. Both rates follow the same pattern over time, with the rate being consistently about 0.5% points higher in secondary school.

But the most interesting comparison, and clearly the one driving the overall difference, is the very different rates of teachers sent home by their school because of close contact with a confirmed case in school. This reaches over 3.5% of teachers in primary schools, twice the level of the peak in secondary schools. This is the final puzzle.

Deciphering the data

How can we explain these facts? We certainly do not have definitive answers, and all contributions are welcome.

Our current best story is this. Starting from well-established facts, we hypothesise that there was a much lower rate of pupil infection in primary schools (see figure 7 here).

If this was all there was to it, there would have been substantial differences in teacher absence deriving from in-school contacts, with the rate much higher in secondary schools.

But in fact we see the reverse: higher teacher absence in primaries. So the (lower) pupil infection rate in primary must have a much greater effect on the number of teachers being sent home than the same rate would in a secondary school.

This in turn must have disproportionately raised pupil absence in primary schools, as classes found themselves without a teacher, meaning that the difference in pupil absence between phases was less than it might otherwise have been.

Why the greater effect of pupil infection on teacher absence? Primary schools are physically much smaller than secondaries, so social distancing strategies for teachers (and pupils) from pupils must be harder. Second, it may be harder for teachers in primary (and special) schools to maintain social distance from pupils. Third, primaries have far fewer staff, and are organised differently, so this must reduce scope for coping and redeploying staff.

This seems to us to be the key new fact emerging from this data: the much greater impact that an initial infection in a primary school has on teachers isolating than in secondary schools.

Relationship with national infection rates

Finally, what might explain the relationship between teacher absence data for confirmed Covid cases and national infection rates, as seen in the fourth chart above?

The time series patterns are rather different, with the rate for teachers being much flatter over these weeks. (Note that we have not tried to align the scales for the two series. These two measures are sufficiently different in a number of dimensions that a direct comparison is not reliable.)

The best assessment we have to date of infection rates by occupation is from the ONS, in a single integrated study, which concludes for the period 2 September to 16 October 2020 that “there is no evidence of difference in the positivity rate between teachers and other key workers” (see figure 12). It would be interesting to see this work repeated for the November to December period we look at here.

[UPDATE: The Office for Statistics Regulation has written to ONS to elaborate on the strengths and limitations of the analysis undertaken to arrive at this conclusion].

Message for policymakers

We take it that the main goals of interest for policy are pupils’ and teachers’ health, and pupils’ time in school. It seems to be the case that home learning does not work well for many, and pupils do best in school with actual teachers. Hence the government’s and the chief medical officer’s statement of the prime aim of keeping schools open.

We find two points to emphasise for policy from the teacher absence data.

First, the difference by phase between pupil absence and teachers sent home to isolate suggests that quite a substantial fraction of pupil absence in primary school may be due to the greater effect of pupil infection on teacher absence. The lack of investment in spatial and staffing flexibility may have led to significantly more lost learning.

Second, in secondary schools, the potentially infectious contact that teachers have splits roughly half and half between those from the teachers’ life outside school, and those arising in school. This re-emphasises the degree to which infection in schools reflects infection in the broader community, and that the dangers to teachers’ health are not only, perhaps not even mainly, to be found within schools. The debates between closing or opening schools must keep this fact central.

Thanks for comments to Ines Hassan, Anna Vignoles, Richard Wilkinson.

Want to stay up-to-date with the latest research from FFT Education Datalab? Sign up to Datalab’s mailing list to get notifications about new blogposts, or to receive the team’s half-termly newsletter.