select search filters
briefings
roundups & rapid reactions
Fiona fox's blog
The Office for National Statistics (ONS) have released the latest results of their COVID-19 infection survey for England.
Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:
“The Office for National Statistics (ONS) are continuing their admirable practice of giving weekly reports on the results from their COVID-19 infection survey. The weekly reports do always add a little more to the picture of infections across the community population of England (so not including people in hospitals, care homes, or other institutional settings). Again these results are provisional – ONS and their partners are still working to find the best ways to present the data, and not all the results from swabs sent for testing in the most recent week had been received, so the most recent data may change slightly as these test results are received.
“However, another week has passed, and many more tests have been done. So what has changed? In many ways, not very much. There is evidence that the proportion of people testing positive for the virus is continuing to fall as time passes, and that falling pattern is statistically significant, so that it is unlikely to be some kind of chance effect of the way people were chosen to be in the survey. The proportion of people testing positive for the virus in the most recent 14-day period, 25 May to 7 June, is estimated at 0.06% – that’s 6 people in every 10,000, or about 33,000 in the whole community population of England. There’s quite a lot of uncertainty about this, because the estimates come from a sample survey of about 20,000 people, and the number of people who would test positive could plausibly be somewhere between 14,000 and 68,000, rather than exactly 33,000. But these figures are well down on the previous 14 days (11-24 May), when the estimate of the number of people testing positive was 22 people in every 10,000, or 124,000 in the whole community population (with a range from 54,000 to 243,000). The central estimate of the number infected is therefore down by 74% in the most recent 14 days compared to the previous 14 days. A more detailed analysis by ONS’s partners at the Universities of Oxford and Manchester also shows a clear, and statistically significant, downward trend. The rate of decline towards the end of the period covered is rather slow, but that is not really surprising now that the number of people testing positive is getting small.
“There is no clear sign in the data of an increase in the proportion of people testing positive since the various loosenings of lockdown rules have occurred in the past few weeks, though I should add that it would not be easy to detect a small increase from a sample survey of people, even one with a reasonably large and increasing sample like this one. (But certainly there is no sign of a large increase.)
“The broad conclusions from looking at subgroups of the population, by age groups, gender, type of work, and so on, are not changed much from last week. In a few cases, the changes might look substantial, but the margins of error are very substantial so these changes may well not reflect reality. You have to remember that, over the whole study period from 26 April to 7 June, only 97 individuals have tested positive for COVID-19. This is enough to give a reasonably accurate indication of the number infected in the whole population, but not necessarily of the numbers infected in subgroups divided by age, gender, and so on. The evidence is still clear that there is a higher percentage of infections in people working in patient-facing health and care roles than in other working age people. As in last week’s report, there is no further data on antibody tests (that indicate whether someone was infected with the virus at an earlier time).”
Prof Sheila Bird, former Programme Leader, MRC Biostatistics Unit, University of Cambridge, said:
“The ONS Infection Survey, which has begun recruiting its June cohort of household-residents, has good news to report: the weekly trend has been a solid decrease in the proportion of participants testing positive for COVID-19 so that the ONS-team estimates that, at any time in the past fortnight (25 May to 7 June) an average of 33,000 people in the community in England would have been swab-test positive (95% CI: 14,000 to 68,000).
“This good news may appear to conflict with the consensus regional estimates for R that were also released this afternoon and which include 1.0 for all-England and for each English region – East of England excepted. However many models contributed regional assessments (not disclosed), most of them will have relied – in part – on back-calculation from hospital admissions for and deaths from COVID-disease and so the resultant estimates for R potentially relate to incidence some two to three weeks previously. Unlike NICE appraisals, the basis on which the consensus judgement was arrived at on the basis of the regional-R assessments provided is not wholly transparent. However, readers are steered away from over-reliance on R when infection-incidence is what really matters.
“Hence, back to what matters, the other key estimate that the ONS Infection Survey provides, based on the repeat-swab-samples over the past 6-weeks, is the daily number of persons who are newly infected, which is currently estimated as 4,500 (95% CI: 3,200 to 6,200).”
Prof Keith Neal, Emeritus Professor of the Epidemiology of Infectious Diseases, University of Nottingham, said:
“There is no doubt that the epidemic is decreasing. We currently have more restrictions than other European countries have and for some countries more than they ever had. There are currently only minor lifting of restrictions that have had time to show up in an increased rate of infection and this has not happened. Most of these related to outside activities where social distancing is easier and also at significantly much lower risk of infection.
“If the outbreak is not slowing as fast as the rest of Europe who have far fewer restrictions then there must be another reason. Cases are higher in highly urbanised environments and the UK is much more urbanised with a higher population density than other countries. If this is a factor then it might explain the slow drop in infections and also why the UK was more severely affected as this factor would contribute to both. The other explanation is that people are meeting more than the current guidelines allow. These figures will need to be kept under review as mass gatherings are thought to be a key factor in COVID-19 transmission.”
All our previous output on this subject can be seen at this weblink: www.sciencemediacentre.org/tag/covid-19
Declarations of interest:
Prof McConway: “Prof McConway is a member of the SMC Advisory Committee, but his quote above is in his capacity as a professional statistician.”
None others received