select search filters
briefings
roundups & rapid reactions
before the headlines
Fiona fox's blog
The Office for National Statistics (ONS) have released the latest data from their COVID-19 Infection Survey.
Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:
“The prevalence estimates in this week’s bulletin on the ONS Covid-19 Infection Survey (CIS) continue the general picture that’s been around for several weeks now. The latest figures cover the week 19-25 September – last week. Again the short-term trends in the numbers estimated to test positive for the virus that can cause Covid-19 are different in different UK countries, though they remain pretty high in all four countries. In three of the four countries, there also isn’t a clear pattern of increase or decrease running over several recent weeks. (The exception is Wales, where the trend has been pretty well unremittingly upwards since at least June.)
“Short-term trends are different in different regions of England too. In the CIS estimates for smaller areas of the UK, there is evidence of quite substantial differences in the percentages who would test positive between areas that are geographically close to one another, though those small-area estimates have wide statistical margins of error.
“This all fits an overall picture that, while the overall average infection level across the whole UK is high and is not changing fast from one week to the next, the local position in smaller areas (separate UK countries, regions, and smaller areas) can differ quite a lot from place to place and indeed, sometimes, from week to week. What’s important, I’d say, is not so much the overall position in the UK or its four nations, as the way those overall positions are affected by what’s going on in smaller areas.
“For England, ONS estimate that about 1 in every 85 people in the community population would have tested positive for the virus in the most recent week. That looks like a slightly higher rate of infection than the week before (ending 18 September), when the estimate was that 1 in 90 would test positive. But, taking the margins of error into account, it’s not entirely clear that the level definitely changed. (The margin of error in the latest week goes from 1 in 80 to 1 in 90, and the week before it went from 1 in 80 to 1 in 95.) ONS say that the rate in England “continued to fluctuate”. This fluctuation, often in different directions from week to week, does broadly match the pattern in counts of new confirmed cases on the dashboard at coronavirus.data.gov.uk
“In Wales, ONS estimate that the number of people who would test positive increased in the most recent week. The increase is not large – last week’s estimate was that about 1 in 60 would test positive, this week’s figure is 1 in 55, and the margin of error is large because the number of people swabbed for the survey in Wales is relatively small. But this upward trend in Wales has been going on for many weeks now. About a month previously (week ending 28 August), the ONS estimate was 1 in 110 testing positive in Wales, and another month earlier (week ending 24 July) it was 1 in 160. Again, this trend of increase does broadly match the trends in dashboard new confirmed case figures, though they have been rather more variable.
“The trend in Scotland looks better. ONS estimate that the numbers testing positive decreased in the most recent week, to about 1 in 55 testing positive, so a fall from the peak (1 in 45) that was reached two weeks earlier. Last week, ONS reported that infections were levelling off in Scotland, which was good news after earlier rapid increases there, and this week’s figure adds hope that infections there are really coming down now. That also fits in with the trend in Scottish new confirmed cases on the dashboard. But, just because the numbers have come down from the peak over the most recent couple of weeks, the variability that we’ve seen in recent times and in other parts of the UK does indicate that we can’t be certain that numbers in Scotland will continue to fall. I very much hope they do.
“In Northern Ireland, ONS say that the short-term trend is ‘uncertain’. The statistical margin of error for the Northern Ireland estimates is relatively wide, because the number of people tested in the survey is relatively small there, so trends in the survey results there are often rather unclear. The ONS estimate for the latest week is that 1 in 65 people would test positive for the virus in the community population in Northern Ireland, which is slightly better than last week’s estimate of 1 in 60 – but this week’s margin of error goes from 1 in 50 to 1 in 90. You can see why ONS can’t be definite about the trend there. However, there is some rather uncertain evidence from the survey, looking back longer than just one week, that infection rates in Northern Ireland are falling slowly, and that would broadly match the pattern of confirmed cases on the dashboard.
“Overall, allowing for the margins of error, the ONS estimates indicate that infection levels are pretty similar in Wales, Scotland and Northern Ireland, with 1 in 55 or 1 in 65 testing positive, though the trends are in different directions in Wales (increasing) and Scotland (decreasing). In England, for the latest week, the level of infection is somewhat lower at an estimated 1 in 85. Of course, 1 in 85 testing positive is quite a high level of infection, similar to the rate in early February this year when we were in lockdown, but vaccinations have changed the importance of such a high level of infection. Also (see below) infections are becoming more and more concentrated in the youngest age groups, particularly secondary school pupils, and most of those will not have been vaccinated, while most of those in older groups will have been vaccinated.
“The ONS estimates for the English regions emphasise that the overall England figure is an average of really quite different positions in different regions. There is a relatively high level of statistical uncertainty about the regional estimates, because the number of people swabbed for the survey in a single region is bound to be considerably less than the number samples for the whole country. ONS report an increasing short-term trend in four regions (the North West, Yorkshire and The Humber, the East Midlands and the South West), a decreasing short-term trend in two regions (the North East and London), and uncertain trends in the other three regions (West Midlands, East of England and the South East). I definitely agree with ONS who say the overall pattern is one of continuing fluctuation, often with increases in one week followed by decreases in another, or vice versa. For the most recent week, the region with the lowest estimated proportion testing positive is London, about 1 in 130. Rates are also relatively low in the East of England and the South East, though they are somewhat higher in the other South of England region (the South West) and in the West Midlands and the North East, and a bit higher still in the East Midlands, the North West and particularly Yorkshire and the Humber (where the rate is estimated at 1 in 55 testing positive in the most recent week).
“This fluctuation and the differences between regions is even clearer in the ONS estimates for smaller areas (shown in the map in Figure 3 of the CIS bulletin), though those estimates are usually subject to wide margins of statistical error. The areas with high levels of infection do change quite considerably over time, and the level of testing positive can be very different between areas that are geographically close to one another. As just one example, in Salford for the most recent week, ONS estimate that about 1 in every 30 people in the community population would test positive. That’s a very high rate, even if we take into account the wide margin of error (from 1 in 20 to 1 in 55). But in the adjoining area, Wigan and St Helens, the estimated rate of testing positive is very much lower at 1 in 110 (margin of error from 1 in 65 to 1 in 210). There are large differences between areas that are adjoining or very close to one another in many other places on the map too, which does add to the evidence that the current state of the pandemic is very far from consistent across the whole country.
“On individual age groups in England, ONS again refer to fluctuation, and not all the trends are at all clear. But ONS report that the percentage testing positive increased for those from age 2 up to school year 11 (age 16 or 17). The increase is in fact much more marked for the older children in his range, from school year 7 (age 11 or 12) to year 11 – that is, the main secondary school ages. In that age group, ONS estimate that about 1 in 20 children would test positive. That’s an extraordinarily high rate. The rate in that age group didn’t start increasing really fast, according to the ONS figures, until roughly mid-September, so a little time after schools had reopened. The 1 in 20 figure is actually the estimate for the middle of the most recent week, 22 September – their estimate for the last day of that week, 25 September, is nearer 1 in every 15. The midweek estimate for the younger group (age 2 to school year 6) is only about half as big, 1 in 40, and not showing anywhere near such a clear increasing trend. It’s true that children of these ages are very unlikely to get seriously ill if they are infected. But they can pass on the infection to others.
“It’s pleasing, however, to see a continuing fall in the rate of testing positive in those aged between school year 12 and age 24 – who, of course, have now been vaccinated in large numbers. The ONS estimate for that group is that 1 in 90 would test positive, and that rate has been falling pretty steadily at least from mid-August, when it was about 1 in 30. Age groups older than that have much lower rates of testing, between 1 in 130 for those aged 35-49 to 1 in 190 for the over 70s – though a small warning sign for that oldest group is that ONS say that “there were early signs of a possible increase [in the percentage testing positive] for those aged 70 years and over.”
Declared interests
Prof Kevin McConway: “I am a Trustee of the SMC and a member of its Advisory Committee. I am also a member of the Public Data Advisory Group, which provides expert advice to the Cabinet Office on aspects of public understanding of data during the pandemic. My quote above is in my capacity as an independent professional statistician.”