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

expert reaction to latest data from the ONS 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 regular weekly report from ONS about their infection survey is out a day earlier than usual this week.  In terms of prevalence (the numbers of people who would test positive for a current infection with the virus that causes Covid-19, whether the infection just happened or happened some days of even weeks before), it takes the data up to the week ending 27 March.  In terms of incidence (just looking at new infections), the data go up to the week ending 20 March.  (For technical reasons the incidence data can’t be produced so quickly.)

“The overall picture is rather mixed, and could be summarised rather crudely by saying that some measures are a bit higher than the week before, some are bit lower, and so at a UK-wide level there’s not much change.  That’s a bit disappointing, given the huge declines in infections that we saw from the peak at the turn of the year until just a week or two ago.  But it’s not surprising, I’d say.  Schools have reopened (and though they’ve closed again now for the Easter holiday, that won’t show up in these data), and some other loosenings of lockdown had already occurred in the period covered.  More contact between people means more possibilities for infection to spread.  A levelling off in infections, or a small increase, should be much less of a problem, now that many people in the most vulnerable groups have been vaccinated and should be protected to a considerable extent from serious illness or death, if they do become infected.  But overall, infection levels are roughly where they were in late September last year, and that’s roughly four times the level of last July and August.  I’m not immensely concerned about the current trends in infections, but we’ve still got to keep a careful eye on what happens in the next weeks and months.  Once infections start rising, we’ve seen before how quickly the rises can sometimes take off.

“To the details.  First, on prevalence (people who would test positive, regardless of whether the infection is new).  ONS estimate that in England for the most recent week (ending 27 March), about 1 in 370 people in the community population would test positive.  That’s a slightly lower prevalence than the previous week, when the estimate was 1 in 340, though there’s some statistical uncertainty about that.  In Wales, ONS estimate that 1 in 570 people would test positive.  That’s also a lower prevalence than the week before, but ONS conclude that the trend in Wales is uncertain – partly because the level of statistical uncertainty is greater in the other three UK countries than in England, because fewer people are swabbed for the survey.  In Northern Ireland, despite this uncertainty, ONS conclude that there are ‘early signs of an increase in the percentage of people testing positive’, and their estimate for the most recent week is 1 in 220, the highest of all four UK countries.  In Scotland, after concluding last week that the positivity rate had gone up compared to the previous week, ONS now feel that it has probably decreased again (to 1 in 320).  If you add up the ONS estimates of the actual numbers of people infected, to get a UK-wide figure, the central ONS estimate is that the number who would test positive in the latest week has fallen by somewhere round 10% compared to the week before – but this trend is far from certain because of the statistical uncertainty.

“There does seem to be increasing evidence of differences between the UK countries and also the regions in England.  That seems usual when infection rates overall are relatively low, and it’s not surprising because a relatively small outbreak in a relatively small area can shift the national or regional figures noticeably when infections are low.  In the English regions, ONS report an increase in people testing positive in the East of England, a decrease in the South East and South West, but say that the trends are uncertain elsewhere.  The pattern of changes has varied quite a lot across the English regions in recent weeks, but the upshot of these various changes is that the prevalence of positive tests is quite a bit lower in the South East and South West (with 1 in 680 estimated to test positive in the South East and 1 in 770 in the South West) than in the north (1 in 230 in the North East and in Yorkshire and the Humber, 1 in 280 in the North West), with the other regions in various places between.  I must emphasise that there are wide margins of statistical error on these regional estimates, because the numbers of people swabbed for the survey in a single region is small compared to the number for the whole country.  But I think there is a risk that the North and maybe the Midlands might end up with higher infection rates than the South, as happened last summer when infection rates fell overall, and that could possibly increase the impact of outbreaks there.

“The pattern of trends on testing positive by age is also not very clear, and for England ONS report that the rate of testing positive fell in those aged 50-69 but that the trends are uncertain in all the other age groups that they look at (including the 70+ group as well as all the younger groups).  Rates of testing positive in the oldest age groups (50-69 and 70+) are, however, still a lot lower than in the youngest age groups – the rates in school-age (and younger) children in England are between two and three times as high as in those aged 50 and over.  Again, though, the margins of error are wide, particularly for the school-age groups.

“The picture on trends in incidence is slightly more concerning, though not unexpected.  ONS just began last week to publish estimates of the rates of new infections, based on their survey, after a break to improve the estimation model that they use.  The latest incidence estimates are for a week before the latest prevalence estimates, for technical reasons, so the latest prevalence estimates we have are for the week ending 20 March.  But they can be compared with the estimates from the week before that, on a consistent basis.  In England, Wales, and Northern Ireland, ONS estimate that the rate of new infections increased over that period.  In Scotland, they estimate that the rate fell.  However, they are rightly very cautious about the directions of these trends, because numbers of new infections are low and the margins of error are wide.  ONS do conclude that there are ‘early signs of an increase of new PCR-positive cases in England’ over those two weeks, but they feel that they cannot be sure of the trend in the other UK countries.  I think that caution is very wise.  None of the estimated differences between the weeks is enormous, in either direction – for instance, for England, ONS estimate that the daily number of new infections in the week ending 20 March was between 9,600 and 14,600, and the week before it was between 8,000 and 13,300, so probably an increase but not too large and of uncertain size.  Overall, though, adding up the estimates for all four countries, the central estimate shows daily infections going up by about 10% over these two weeks, but with a very wide margin of error.  A small increase is not surprising – schools have reopened.  Also the numbers of new confirmed cases on the dashboard at remained pretty level over that period at around 5,500 a day, after falling consistently and pretty rapidly for many weeks.  (That 5,500 a day figure is a lot smaller than the ONS estimates, because the dashboard figures count up only people who have a new positive test result.  People who are infected but have no symptoms are unlikely to be tested and so won’t show up in the dashboard figures, and we’ve heard today from a study1 that found that many people who do have symptoms may also not be tested.  But the key point is that the trends in the two sets of numbers are pretty similar – little change up or down, and certainly no big falls as we saw a few weeks earlier.)  After 20 March, the new confirmed cases on the dashboard seem to be falling again, albeit very slowly.  We’ll just have to wait to see whether that is reflected in next week’s infection survey bulletin.”




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