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

expert reaction to latest data from the ONS COVID-19 Infection Survey

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 latest results from the ONS Covid-19 Infection Survey (CIS) take the data forward to the week ending 17 July, so just under a week ago, though for some of their estimates, the latest available figures are for an earlier data. On the face of it, the bulletin seems not to tell us much that we didn’t already know from the daily figures for new confirmed cases on the dashboard at Infections are rising across nearly all the country. But the dashboard figures can be biased, because they depend on the number of people who decide to be tested, or have to be tested because their work requires it or they need a test result in order to get into a location or venue. The types and numbers of people who are tested for those purposes can change over time, and it’s possible that some changes in the numbers of cases on the dashboard come from those changes rather than truly reflecting the progress of the pandemic. The CIS gets its data from a survey or a representative sample of households across the UK, who are tested only to provide information on the spread of the virus, so the results won’t be affected by that kind of bias. But, against that, the CIS obviously doesn’t survey the entire population, so there will be some level of statistical uncertainty in its results. That level of uncertainty is low for a whole large country like England, but relatively larger for a smaller area or group of people, like a single age group or a smaller English region or even a whole smaller country such as Wales or Northern Ireland.

“Though the trends in the CIS results do, in broad terms, match the results on the dashboard at, there’s on important aspect where they provide a different picture. That is to do with what they both measure. The main CIS results, which I’ll describe below, are estimates of the numbers who would test positive, on a swab test for an infection with the virus that can cause Covid-19. People testing positive might have a brand new infection that just became detectable on the day they were swabbed for the survey, or they might have an infection that started a few days or even a week or more earlier. The dashboard results are for new cases – that is, cases that tested positive for the first time on a particular day. ONS also produce estimates of new infections, which they can do (unlike the REACT-1 survey) because in the CIS, people are tested on more than one occasion. In principle at least, it’s those so-called ‘incidence’ estimates from the CIS that should look like the dashboard results for new confirmed cases. Because the CIS incidence results are based on fewer swab results, and because of various technical complications, they are available later than the other (‘prevalence’) results, so the latest available incidence estimates are for the week ending 3 July, and the statistical uncertainty is relatively large. ONS stopped publishing incidence estimates between mid-April and mid-June because at that time, infection levels were so low that the estimates weren’t reliable. But they started again a couple of weeks ago.

“The trends in the incidence figures do, broadly, match the trends on the dashboard figures for new cases, in the sense that they are rising, in roughly the same way. But there’s one important difference. For the latest week available, 27 June to 3 July, ONS estimate that there were about 52,000 new infections every day. There’s a fair bit of statistical uncertainty about that – the true figure could plausibly be anywhere between about 46,000 and about 58,000 each day. But the number of UK-wide new confirmed cases on the dashboard for that week works out at a bit under 26,000 a day – about half as much. I don’t think this difference means there is some big problem with the ONS estimates. Instead I think it probably arises because something like half of people who are infected with the virus don’t have symptoms, or at least don’t have symptoms that cause them to get tested, so they just don’t show up in the dashboard new cases counts. Previous analyses of CIS and other data have confirmed that that’s broadly the position. The latest 7-day average count of daily new cases in the UK for the dashboard, which is the average for 10-16 July, is about 45,000. That’s already a worryingly high number, but these CIS results indicate that the true daily number of new infections (as opposed to new ‘cases’ found through routine testing) might well be nearer 90,000 for that week.

“Turning back to the ‘prevalence’ data, which gives counts of people testing positive regardless of when their infection began, the broad trends in the new CIS data do also match what we’ve seen in the dashboard data. Infections are increasing, across each of the UK countries taken separately. Just in the most recent week, ONS say that the increase isn’t so clear in Scotland, and in England, the increase is rather slower than it has been in recent weeks, but it’s a considerable increase nevertheless. So not too much comfort there – we’ll have to hope that the increases do slow up and level off. But that might well not happen for a long time, particularly given the removal of so many restrictions in England on 19 July (just after the period covered by this bulletin). In England, the numbers of people who would test positive for an infection, in the most recent week, is about where it was in late January this year. In Wales and in Northern Ireland, it’s relatively quite a bit lower, but still roughly at the level from mid-February. In Scotland, despite the fact that the trend isn’t very clear in the most recent week, the estimated number of infected people is at the highest level it has been since the CIS began there in October last year. In the English regions, the percentage testing positive has continued to increase, except in the North West and the North East, where ONS report that the trends just in the most recent week are uncertain. That’s a little bit of good news, given the fairly rapid rates of increase in those two regions in recent weeks. Rates of testing positive, however, remain lowest in the regions of Southern England outside London (South East, South West, East of England).

“In England, the rate of testing positive increased in all the age groups that ONS use for analysis, except for the group covering those of secondary school age (school years 7 to 11), where again the trend in the most recent week is uncertain though rates were rising up till then. However, the proportion who would test positive does very hugely between age groups. In those aged from school year 12 (age 16 or 17 years) to age 24 years, ONS estimate that 1 in every 30 would test positive. For those aged 70 and over, only 1 in 290 would test positive. That must have a lot to do with vaccination.”



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



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.”

in this section

filter RoundUps by year

search by tag