select search filters
briefings
roundups & rapid reactions
Fiona fox's blog
The Office for National Statistics (ONS) have released the latest data from their COVID-19 Infection Survey.
Prof James Naismith, Director of the Rosalind Franklin Institute, and University of Oxford, said:
“Today’s ONS figures (up to 22nd may) are consistent with other data.
“Scotland appears to have a higher prevalence of covid19 than the other parts of the UK.
“Case numbers of the so called Indian variant are increasing and seem to have halted the decline in overall numbers of infections. This variant has reached pretty much everywhere (as did the Kent and original strains did before it). It is difficult to apportion with certainty the rise in this variant to the effect of reduced social distance or to increased transmissibility. (Most likely a combination)
“It seems almost certain that we will face a third episode of rising covid19 infections. The questions are how large will the peak number of infections will be and how much harm will this third episode do. We can each make a difference to the answers.
“Factors which we can be certain will reduce both size and harm of this episode:
1)Effective isolation of people with illness or who test positive; testing is vital.
2)Increasing the portion of population doubly vaccinated
3)Wearing a face mask where possible indoors with non-household members, improving ventilation and washing hands
“Factors which we can sure will make this episode worse:
“We cannot alter the virus, the outcome will also depend on its properties. We should keep in our minds that large number of infections will allow the virus more chances to try out more variants.
“If overall cases are rising (as seems likely) then the following measures within the UK would appear to continue to be as ineffective as before at halting the virus spread:
“I appreciate they all sound good in theory and in press releases but in the real world of the UK, they have once again failed to halt viral spread.
“It seems likely that the Indian variant will mostly confine itself to the unvaccinated younger population. It is much less likely to cause serious disease in this group. However, less likely is not the same as zero. With large enough numbers of infections, appreciable numbers will get seriously ill.
“I do not believe we will see third large peak of deaths due to this variant, the successful vaccination campaign has and will prevent this. With one of the highest death tolls and thus the number of tragedies that befell families across the UK, this is a mercy that I am thankful for. However it remains too early to be sure what lies ahead, it will take time to know. We have to remember that an infection today, can result in hospital in 2 weeks time, with the most serious problems a week later. Another way to think about this, is today’s hospital figures are looking back two to three weeks ago where infections were still falling.”
Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:
“There have recently been some fairly clear signs of increases in the numbers of new confirmed Covid-19 cases on the dashboard at coronavirus.data.gov.uk, particularly in Scotland, to a lesser extent in England too. Some have suggested that these increases don’t really correspond to a real increase in the numbers of infected people, but are only, or largely, because of the surge testing that’s going on in a few areas where there is concern about rises in cases linked to the ‘Indian’ variant B.1.617.2. After all, many infections are asymptomatic or very mild, and without the surge testing, people with that kind of infection may simply not have been tested and so could not show up as confirmed cases on the dashboard. But there’s a good way to see whether that interpretation of the dashboard figures is true – to use instead data from the ONS Covid-19 Infection Survey (CIS). In that survey, a representative sample of the community population of the UK provides swabs to be tested for a current infection, simply in order to provide good national estimates of infection levels, and regardless of whether they have symptoms, are being tested for something to do with their job, are in an area with surge testing, or anything like that. So the CIS data are not subject to the biases that can arise with the dashboard confirmed cases because of changes in who is turning up to be tested, and why.
“The latest CIS data take the information up to the week of 16-22 May. That is unlikely to have clearly picked up any clear sign of changes because of stage 3 of the roadmap out of lockdown in England on 17 May, because it’s a bit early. But it provides reasonably clear signs of increasing numbers of infections in Scotland in particular. For the most recent week, ONS estimate that 1 in 630 people in Scotland would test positive, if they were tested for a current infection with the virus that can cause Covid-19. That’s about 8,300 people across Scotland. Last week, ONS were estimating that 1 in 1,960 people in Scotland would test positive, so 2,700 people across the country. While that looks like a very big increase in just one week, you do have to bear in mind, however, that the infection levels remain very low compared to what they were at the peak at the start of this year. That means that there’s a lot of statistical uncertainty in the estimates. That figure of 8,300 people testing positive in Scotland in the latest week has a margin of error going all the way from 4,400 to 13,700. For the week before, the margin of error went from 1,000 to 5,200 – so the margins of error for the two weeks overlap and it’s not even entirely certain that the numbers really increased. However, it’s likely that they did increase, and by quite a considerably number. (The reason for the wide error margins is that these results are based on only 16 people testing positive in Scotland in the survey in the last two weeks, out of about 13,000 people tested.) There are even wider margins of error in Wales and in Northern Ireland, because the numbers of people swabbed in the survey are smaller there than in Scotland, because the populations are smaller. ONS estimate that the rate of testing positive in Wales remains really low, with no real signs of a trend either way, and they estimate that about 1 in 3,850 people there would have tested positive in the most recent week. In Northern Ireland ONS say there are ‘early signs of a possible increase’, and estimate that 1 in 820 people would have tested positive in the latest week.
“The position is a bit clearer in England, because the number of people swabbed for the survey is larger. ONS estimate that about 1 in 1,120 people would have tested positive in the latest week, or about 48,500 people in all, with a margin of error from 38,400 to 60,200. That’s actually very similar to the estimate for the week before – but those figures were higher than the week before that, when ONS estimated that 40,800 would have tested positive, with a margin of error from 31,900 to 50,900. Those margins of error do overlap quite a bit, but ONS say there are “potential signs of an increase in the two weeks ending 22 May”.
“Overall, then, it seems that in Scotland the increase in confirmed cases on the dashboard at coronavirus.data.gov.uk does very probably correspond to reality, and has not arisen just because of any increased testing. On balance, it looks as if that’s the case in England too, though the evidence from the survey isn’t quite so clear that there definitely is a real increase. My own feeling is that there is probably a real increase in England, though possibly not yet a large one. (The dashboard anyway wasn’t showing increases in Wales, and not really in Northern Ireland either.)
“However, the rates of testing positive remain pretty low in all four UK countries, so I’m not duly concerned, not yet anyway. In Scotland, the rate of testing positive may have gone up quite a bit in the last two or three weeks, but it’s still only at the level it was in late April, and only about one seventh of the level at the January peak. Also in England, where the estimated rate of testing positive has changed less, it’s still at about the level where it was in late April. That’s less than a twentieth of where it was at the January peak. But it’s about double what it was last July and August. That’s a reason for some continuing caution but certainly not for major alarm. In Wales the latest estimated rate of testing positive is much lower, and is only around half of what it was last summer. I can’t compare results for Northern Ireland (or Scotland) with last summer, because the CIS was not running in those countries then.
“The CIS results can’t answer all the questions about why rates might be increasing in Scotland and England. They do provide some data on variants, and ONS conclude that positive tests compatible with the B.1.617.2 ‘Indian’ variant increased in England and in Scotland, but remained low in Wales (and the trend is uncertain in Northern Ireland). But it’s much harder to say from these data that the increases in positivity in Scotland and (possibly) in England are caused by the increase in this variant.
“Because of the low infection levels and the consequent relatively high level of statistical uncertainty in the estimates, it’s difficult to see clear trends in the data for English regions or for age groups in England. The problem is that the number of people tested for the survey in an individual region or age group is a lot lower than for the whole country, so the statistical margins of error are relatively much wider. ONS consider that rates of testing positive increased in the East of England in the latest week, and that there are some signs of a decrease in the South East, but trends are uncertainty in all the other regions. Despite the concerns about some places in the North West, the estimated rate of testing positive is not particularly high there (1 in 1,540) and is much higher in Yorkshire and the Humber (1 in 610) and in the East of England (1 in 630). This does perhaps demonstrate that all these regions are pretty large geographically, and that a regional average positivity rate can’t tell you what’s going on in smaller areas – but the CIS just can’t give precise estimates for smaller areas like Bolton or Bedford when infection rates are as low as they are now. For age groups, ONS report that there are signs of an increase in numbers testing positive in children of secondary school age, and possibly also in those of primary school age and younger, but again the patterns are not at all clear. The estimated rates of testing positive are really low in people aged 50-69 (1 in 2,500) and aged 70+ (1 in 2,000). That might well have something to do with the fact that so many people in those age groups have been vaccinated, often having had their second dose already.”
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 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.”
None others received.