select search filters
briefings
roundups & rapid reactions
before the headlines
Fiona fox's blog
The Office for National Statistics (ONS) have released the latest data from their COVID-19 Infection Survey, and the government have released the latest estimates for the COVID-19 growth rate and R value.
Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:
Comment on the ONS Infection Survey bulletin:
“The latest release from ONS on their infection survey takes the data forward to last week, the week ending 13 March. Given that English schools reopened that week, there’s a temptation to look for some effect of that on infection rates, but actually it’s too early to say. The position in the other three UK countries would be even harder to detect, since they are having staggered dates for school reopenings, and also because trends aren’t so easy to observe there, because the numbers of people swabbed for the survey are smaller (because of the smaller populations) so that the statistical uncertainty is considerably greater than for England.
“Overall, it’s a mixed picture. The estimated proportion of people in the English community population who would test positive for the virus that causes Covid-19 fell again last week, compared to the week before, and indeed fell by a slightly greater percentage that the fall the previous week. So ONS are now estimating that 160,200 people in the English community population would test positive last week, or 1 in every 340 people, though the inevitable statistical uncertainty means that the data are compatible with anywhere between 1 in 305 and 1 in 385 testing positive. The estimate for the week before that was 1 in 270 testing positive, so quite a substantial decrease in the latest week. About a fifth fewer people would test positive, compared to the week before. The percentage testing positive has also gone down in Wales. ONS estimate that 1 in 430 in Wales would test positive, but there is a lot more uncertainty than for England, with an interval going from 1 in 290 right up to 1 in 715. The news from Northern Ireland and Scotland isn’t so good. In Northern Ireland, the central ONS estimate is hardly changed from the week before, with an estimated 1 in 315 people testing positive, but there’s a very wide margin of statistical error (1 in 195 to 1 in 560). These wide margins of error in Wales and Northern Ireland are partly because of the smaller populations there (so fewer people are included in the survey), but also because these margins of error inevitably get relatively wider as the number of infected people falls. ONS say that the infection rate in Northern Ireland looks roughly level, but that it’s hard to be sure because of the statistical uncertainty. In Scotland, though, ONS report that the percentage who would test positive looks to be increasing, with an estimated 1 in 275 testing positive (with a margin of error from 1 in 205 to 1 in 375). The week before it was 1 in 320. Because all these margins of error for the latest week overlap, we can’t be certain that there are really differences in infection rates between the four countries. But it’s concerning that the central ONS estimate for Scotland is now the highest of all four countries, given that it’s been considerably lower than the other three countries for some time until the end of February.
“Overall, the levels of infection in England, Wales and Northern Ireland are now roughly where they were in late September or early October last year. That’s a lot better than things were at the height of the latest wave, but still not very reassuring. Positivity rates in late September and early October in England and Wales were roughly five times as high as they were in July and August. (I can’t make that particular comparison for Northern Ireland because the ONS survey wasn’t yet running there in July and August. And in Scotland there are no ONS survey results before October, and current rates are below the October level, despite the possible increase in the latest week.) There’s still some way to go before we are at the low infection levels of last summer. I don’t like having to keep saying that we aren’t out of the woods yet, but, well, we aren’t. It’s true that we have vaccines now and we didn’t in September and October, and they will help with severe illness and death – but not everyone can be vaccinated.
“These survey results for the four countries are broadly in line with the data for new confirmed cases on the dashboard at coronavirus.data.gov.uk. Those case numbers also recently show some evidence of levelling off, or at least a slower decline, in Wales and in Northern Ireland, and some evidence of a small increase in cases in Scotland. They also show evidence of levelling off in England, unlike the ONS results. However, here I’d give more credence to the ONS survey results. The figures on the dashboard are for confirmed cases, so people have to be tested to count, and so the results can be affected by changes in people’s behaviour in coming forward to be tested. The English case figures on the dashboard (and not those from the other countries) include results from rapid lateral-flow testing, which has been increasing for some time, and these tests are more likely than the lab-based PCR tests to be carried out on people who have no symptoms. The ONS infection survey results come from testing a representative sample of people, for the sole purpose of estimating how infections are progressing, so they are not affected by changes in testing behaviour. So the apparent slowdown in decreasing cases for England on the dashboard could be due to changed testing behaviour, and therefore not reflect the actual position in the country – and the ONS survey results indicate that that may well be the position at this time. As always, we’ll know more next week.
“The infection survey results for the English regions again show a pattern that varies across the country. ONS report falling infections in four regions (West Midlands, East of England, South West, London), but they say that the trend is uncertain in the other regions, and that it’s possible that infections have recently begun to increase in the East Midlands. Broadly speaking, infection rates look lower in the Southern regions (East of England, South East, South West, London) than in the Midlands and North, but the smaller numbers of people swabbed in individual regions than in the whole country means that there’s a lot of statistical uncertainty. In fact the statistical margins of error for the regions all overlap, so we can’t actually be certain of any of these regional differences.
“In England, ONS report that rates of testing positive have decreased over the last two weeks in all the age groups that they look at, apart possibly from children of primary school age and below, where the trend is less certain. It’s much too early after the school reopening, which was in the middle of the last week analysed here, to conclude that this slowing of decrease in the youngest age group has anything to do with some of them being back at school. Again there’s rather a high level of statistical uncertainty for most of these age group estimates.
“ONS have added a new interactive feature to do with age groups – charts for the four UK countries (Figure 5) that show the estimated percentage testing positive in single-year age groups, together with bars to indicate the statistical margin of error, and you can press a button to show how these have changed over time since the end of January. It’s interesting to remind yourself just how much infection rates have fallen, across all ages, since then, and indeed how the rather large differences in infection rates between different ages have tended to flatten out. That flattening out is less clear in Scotland recently – but what becomes very obvious is the very large amount of statistical uncertainty around the estimates in Wales, Northern Ireland and, particularly, Scotland, once the animation moves into March. Really the margins of error in Scotland are so wide that the pattern of infection across the ages is extremely uncertain. The high levels of uncertainty for the countries apart from England is largely because of the smaller numbers being swabbed for the survey outside England, but the particularly wide margins for Scotland appear to be because of the way the statistical model tends to behave when overall trends are uncertain and possibly changing direction, as they currently are in Scotland.
Comment on the R number and growth rate estimates:
“The latest Government estimated ranges for the R number and the growth rate of new infections have been published. The range for the R number for the whole of the UK has changed slightly. Last week it was 0.6 to 0.8, and this week SAGE have increased the top end of that range slightly, so that it runs from 0.6 to 0.9. This does not mean that the actual R number has definitely increased – for instance, it could have actually been 0.7 both last week and this. The widened range does mean, however, that the level of uncertainty about the R value is slightly greater, and the increased upper limit does mean that R could be higher. The top end of the range is still below 1.0, though, and that indicates that infections are still estimated to be falling. And of course vaccination will mean that serious illness, hospitalisation and deaths should be falling faster.
“The range for the growth rate of new infections is -6% to -3%, indicating that the number of new infections tomorrow will be somewhere between 6% and 3% lower than the number today. (So for every 100 new infections today, the estimate is that there will be between 94 and 97 tomorrow.) Both ends of the range are slightly higher than last week, when it was -7% to -4%. A growth rate of -6%, if it continued at that level over time, would mean that new infections halved every 11 days, which is a pretty fast decline. A growth rate of -3% would mean a halving of new infections roughly every three weeks.
“The range for the R number for the whole of England this week is 0.7 to 0.9, which is higher at both ends than last week’s range of 0.6 to 0.8. The growth rate range for England now runs from -6% to -2%; last week it was -6% to -4%. The ranges for this week still indicate that SAGE believes infections are decreasing across England as a whole, but maybe a little slower than they were. Given all the uncertainties, that’s not really out of line with the latest estimates from the ONS Infection Survey. The ranges for R and the growth rate for the English regions have nearly all changed a little, and where they have changed, it is in the direction of a small slowing in the rate of decrease of infections. The upper ends of the R value ranges are 0.9 for every region – so the ranges are all clearly below 1.0, indicating decreasing cases. Likewise, all the regional growth rate ranges are all placed clearly below 0, again indicating falling numbers of new cases.
“The webpage reporting the revisions to the R and growth rate ranges provides, as usual, links to the reports from the devolved administrations giving their latest estimates of R and/or the growth rate there. Broadly, these match the latest results from the ONS Infection Survey. For Wales, R is below 1 but there is evidence that infections are falling less slowly than they have been recently. For Northern Ireland, the range for R based on positive test results is now 0.9 to 1.1, indicating that the decrease in infections could have levelled off or even, possibly, reversed. For Scotland, the range for R is 0.7 to 1.0, though the growth rate range is below 0 at -6% to -2%. These are slightly out of line with the ONS estimate that infection rates might be starting to increase there – but there is quite a lot of statistical uncertainty around those ONS estimates.”
Prof James Naismith, Director of the Rosalind Franklin Institute, and University of Oxford, said:
“Today’s ONS numbers bring reassurance that for the UK as a whole the prevalence of the virus has continued to decline (to around 3 per 1000 people). A similar story is told by the national R numbers. The lockdown continues to be effective at reducing infection.
“Broken down, the ONS numbers by different parts of the UK tell slightly different stories. Of note is that prevalence in Scotland may have increased but has certainly not decreased. Within England, East Midlands may also have shown a small uptick. The so called Kent variant is now dominant in UK.
“What can be concluded from these data? The UK has suffered enormously with the covid19, we were hit hard by the new variant but the lockdown has now saved thousands of lives.
“The strictness of social distancing measures required to limit the spread of the more infectious variants are very hard to sustain, this may underly the uptick in some regions of the UK and rises in the EU.
“Uncontrolled the new variants will overwhelm health care systems.
“Vaccination offers the only realistic hope to end the pandemic, the UK’s advanced vaccination program will save tens of thousands of lives.
“Since the new variant is much harder to control than the original strain, countries without advanced vaccine programs are going to face a very difficult road ahead. I strongly support UK decision to ship vaccines to developing countries and urge rich countries to support vaccination efforts in these countries.
“I would strongly urge the world to continue to develop treatments against coronaviruses. The vaccines which will end the pandemic are a triumph of science. Provided we do not stop work now, new treatments will become available over the next two years that will prevent the large number of deaths and economic disruption between a new coronavirus emerging and vaccines being deployed.”
ONS Infection Survey:
R number and growth rates:
https://www.gov.uk/guidance/the-r-number-in-the-uk
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.