November 6, 2020

expert reaction to the ONS infection survey for the UK

The Office for National Statistics (ONS) have released the latest data from their COVID-19 Infection Survey.

Prof Sheila Bird, Formerly Programme Leader, MRC Biostatistics Unit, University of Cambridge, said:

“My reading of the confidence intervals about ONS Infection Survey’s weekly reported estimates of SARS-CoV-2 incidence for England is, broadly, acceleration between the first and second fortnights. I suggest that it is too early to call it otherwise.

“My second table compares Scotland and England in terms of weighted percentage (95% confidence interval) testing positive by non-overlapping 14-day periods. England’s sample size is 29 times greater in the most recent fortnight and, consequently, England’s 95% confidence interval would be expected to be about a fifth the width of Scotland’s. Even so, we can be pretty confident, due to essentially non-overlapping confidence intervals, that Scotland’s SARS-CoV-2 prevalence was lower in the second fortnight of October than was England’s.

Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:

“Today’s report on the excellent ONS Infection Survey extends the data for another week, up to 31 October. As always, there’s much more detail about the situation in England than in the other UK countries – that’s because the survey has been running for much longer in England than elsewhere, and because the number of people tested in England is larger (essentially because the population of England is much larger than that of the other countries).

“There’s some encouraging news about the situation in the English community population, though the news isn’t all entirely good. The estimate of the percentage of people in the whole English community population that would test positive for the virus has continued to increase, and obviously it would be better if it has started to fall. But the latest week’s estimate is only about 9% higher than the estimate for the week before, so that’s some indication that the number of infected people is rising at a rate that might be quite a lot slower than it has been in recent weeks. ONS do urge a lot of caution on interpreting these figures, because there’s quite a lot of statistical uncertainty about some of them – and infections are still rising. But generally the prospects look better than they did last week.

“ONS also estimate the daily number of new infections – unlike some other sources of data, they can do that because they test people more than once. Their central estimate for the latest week is about 10% lower than the estimate for the previous week. There is a lot of uncertainty about that figure – we really can’t be sure that the daily number of new infections is really falling – but it’s certainly an indication that the rapid rise in new infections, that we saw in September and early October, is levelling off now. That’s pretty good news, though the news won’t be really good until it becomes clearer that new infections are falling rather than just levelling off. ONS say that there have been roughly 50,000 new infections a day for the past few weeks– that’s still quite a lot.

“Now some more detail. The estimate of the percentage of people in the whole English community population that would test positive for the virus has continued to increase, and for the latest week (to 31 October), ONS estimate that 1 in 90 of the English community population would test positive for SARS-CoV-2. (The previous week’s estimate was 1 in 100). Like all estimates based on surveys, there’s some statistical uncertainty about it. The data are consistent with a rate between 1 in 85 and 1 in 95. The central ONS estimate of the number of people who would test positive is 618,700, with a credible interval (plausible range) from 583,100 to 655,000. The central estimate is 50,600 higher than the estimate for the week before, which isn’t good, but that’s a considerably smaller figure than the increase between the previous week and the week before that (which was 135,000). So that’s the encouraging news on the number of people who would test positive – still going up, but going up quite a lot more slowly. These are all estimates for the English community population (excluding people who live in communal establishments like care homes, hospitals, or some university halls of residence) – but the great majority of the English population don’t live in communal establishments.

“At this level of infection, it’s pretty well certain that the number of false negatives (people who are actually infected but would test negative) considerably outnumber the number of false positives (people who are not infected but would test positive), so that the actual numbers of infected people will be higher than those estimates. This week, ONS have again done some estimation of the percentage of people who are actually infected, allowing for false positives and false negatives, and they indicate that the percentage of people infected, on average over the fortnight from 18 to 31 October, might well be somewhere between 1.30% and 1.57%. corresponding to somewhere between about 1 in 75 and 1 in 65 people being infected with the virus.

“This week, ONS report, for England, on infection rates for people working in education (teachers at school or university, education support workers, etc.). They found that, since the start of the school year, there’s no evidence for a difference in infection rate between any of the groups of education workers that they looked at and other key workers. Though there’s quite a lot of statistical uncertainty in these estimates, that’s reassuring both for the teachers and others involved, and for people concerned about virus outbreaks in schools, colleges and universities.

“ONS estimate that, in the week from 25 to 31 October, there were 45,700 new infections each day across the English community population (with a credible interval from 37,700 to 59,600). These figures, as always, are much higher than the daily reported ‘new cases’ on the Government dashboard, for which the daily rate (as an average over 7 days) was about 19,000 for England for the week in question. The difference is partly because people without symptoms would be unlikely to be tested under the system that provides these daily dashboard figures, so they would not show up as a confirmed case. People with no symptoms are still tested if they are in the ONS survey sample, and some of them will actually be infected and test positive. That’s part of the reason why the results of the ONS survey can tell us more, and more accurately, than the information from general swab testing that goes into the dashboard.

“As for the past several weeks, there are big differences in the rates of positive tests between different English regions. But the pattern has changed somewhat. Rates are the highest in two of the regions in the North (North West, Yorkshire & the Humber). But because increases in the North East region have levelled off for the past few weeks, rates there are now quite a lot lower than for the other two regions of the North. Infections in the North East, together with the East and West Midlands, are still probably slightly higher than the average for England, however. Infection rates in London, and particularly in the three other Southern regions (South, East, South West, East of England) are lower. The really rapid increases that we saw in some Northern regions a few weeks ago do seem to have levelled off, and there’s some indication of decrease in some places. There’s some indication of continuing increase in the Midlands, and also in the South East and South West. But all these trends are far from certain – the number of people tested in each region is, of course, smaller than the number for the whole country, so the regional trends are inevitably less clear-cut. ONS rightly encourage a lot of caution in interpreting trends in regional data.

“In terms of different age groups, ONS are estimating that the number of infections is growing in several, though no longer all, age groups. The rates are still highest in the younger age groups, particularly those aged about 17 (school year 12) to age 24. But the rates of infection seem to have levelled off in that age group, and also in the younger secondary school age groups, and ages 25-34, and they could well be falling. In older people, and indeed in children aged from 2 to about 10, the rate of infections might well still be increasing, but from a lower starting point.

The survey can’t be as definite about the position in the other three UK countries, because not so much data is available. Rates of positive tests  in Wales and in Scotland are rather lower than in England, though they may be higher in Northern Ireland than in England (though there is a lot of uncertainty because of smaller numbers of people being tested). Rates of infection are continuing to increase in Wales and in Northern Ireland. In Wales, ONS report that the increase in Wales seems to have slowed compared to previous weeks. They can’t say that about Northern Ireland yet, because not enough data is available. The survey has not yet been running for long enough in Scotland to give a clear picture of trends.”

Dr Jonathan Pearson-Stuttard, School of Public Health, Imperial College London, said:

“The ONS weekly surveillance data provides the most robust estimates we have as to the true prevalence of COVID-19 in communities because, unlike the Government’s daily testing data, it tests a random sample of the population irrespective of whether they have COVID-19 symptoms.

“Prior to this week’s surveillance report we have had several consecutive weeks of large increases in the estimate of daily new COVID-19 cases. This week however, for the first time since the summer, the ONS report similar levels of new infections to the previous week, approximately 46,000 each day. 

“While this is clearly welcome, there are several important points to note from the data. This is still a very high number of new cases each day – far higher than can be expected to be managed effectively by the current test and trace programme. The overall plateauing of case numbers masks the different patterns experienced by different age groups – teenagers and young adults saw a levelling off of infection rates but there were increases in all other age groups, including elderly groups who have the highest risk of hospitalisation and poor outcomes from infection. And the regions with the highest rates of COVID-19, such as the North West of England, have seen a decline in cases over this period, while other regions with lower infection rates, such as East and West Midlands, continue to see increases in infection rates.

“The ONS weekly surveillance data will be extremely important to assess the impact of the nationwide restrictions implemented on 5^th November in the weeks to come.”

Prof Paul Hunter, Professor in Medicine at the University of East Anglia, said:

“The most recent ONS COVID-19 infection survey data has been released. This is a prevalence survey and not an incidence survey so it does not directly measure the number of new cases although it allows these to be estimated.  The most recent data covers the period 25 to 31 October so it is already a week or so out of date. In particular this data cannot be used to show any impact of the current lockdown which only stated yesterday.

“Nevertheless this data does suggest that the epidemic is not increasing as rapidly and may even have started to decline. The peak estimated incidence per 10,000 people per day was 9.52 in the period 17 to 23 October and this was a 47% increase on the value of 6.46 during the period 10 to 16 October. By contrast in the most recent period the estimated incidence was 8.38, a 12% decline. However this credible interval in the estimates does not allow us to be confident that the epidemic is indeed now declining and that it started to decline before the current lockdown.

“It is important that we do not make too many strong assumptions based on a single data source. However, the daily published UK data also support the suggestion that the autumn epidemic may be waning as the cases reported  in the 7 days ending 5^th November was 157,857 compared to 154,873 in the previous 7 days and 136,860 in the 7 days before that. Daily hospitalisation rates may also be starting to level out. Sadly death rates are continuing to increase, but we would not expect to see any reduction for another week or so. Finally, the ZOE COVID Symptom Study UK Infection Survey is also suggesting a decline in the epidemic over the past week or so.

“Taken together these sources do suggest that the has been a decline in the spread of the epidemic over the past week or so. Whether this turns out to be a temporary decline or a longer term trend, possibly as a result of the imposition of the three tier system, it is too early to say. Nevertheless these observations are very welcome and hopefully when the current lockdown ends we will continue to see a continuing decline throughout the rest of the year and into 2021.”

Prof James Naismith FRS FRSE FMedSci, Director of the Rosalind Franklin Institute, and University of Oxford, said:

“Today’s ONS data release for the week ending 31st Oct brings welcome news. Although the virus is still growing, it does appear to have stabilised with 46, 000 (38, 000 to 60, 000) new cases per day; the previous week was 52, 000 (38, 000 to 78, 000).

“The ONS survey is a reliable guide to virus spread although this week there is larger uncertainty than normal which the ONS explains. Importantly these data present a picture consistent with the ZOE KCL data, the virus is spreading at a constant rather than an increasing rate. This is evidence that the social restrictions prior to lockdown have had a real impact.  

“Should we be at or past the peak of infections, then I would not expect the death rate to exceed 1000 a day for any prolonged period; however, we will are still very likely to face daily death tolls of 500 a day for a period in November. Each death represents a human tragedy.  It can be misleading to read too much into a single report, however the match with ZOE is encouraging. Should next week’s data show a similar stabilisation or reduction, then we can be confident that the second wave has for now stabilised.  The national lockdown will not begin to show up in ONS figures for another two weeks, but we would expect it to bring a rapid decrease in the number of new infections

“These numbers from ONS would suggest track and trace is reaching around 21 % of the contacts of infected people in the UK.

“In a democracy politicians must make the very difficult decisions to restrict our lives, and science cannot say this is the ‘right’ decision. I believe that SAGE provides UK government with high quality evidence and analysis on the spread of infections, the capacity of the NHS and the health, financial cost of social restrictions. This is best that science can do and we are fortunate to have this expertise. We do not have perfect information and the future is always uncertain. The clear risk is that the virus is not controlled quickly enough to stop hospitals over filling. The Prime Minister and parliament have decided that on balance their judgement is that the costs of a further ‘lockdown’ are preferable to running the risk of overfilling hospitals. 

“If we can contain the virus until the new year, mass testing, vaccines and new medicines will transform our outlook.”

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 the Advisory Committee, but my quote above is in my capacity as a professional statistician.”

None others received.