On false positives in COVID19 testing again: we are being misled over confirmatory testing

 

It has long been claimed that many of the COVID19 'cases' reported (where a 'case' is simply a positive test result) are false positives. This short article addresses an aspect of the problem not widely discussed before and suggests that many of the very large number of new 'cases' reported amid great hysteria each day could be people who do not have COVID19.

One of the biggest misunderstandings about the notion of accuracy for a type of test (such as a PCR test) is the idea that there is a (single) false positive rate, namely the probability that the test will be positive for a person who does not have the virus. The false positive rate at any particular time - and in any particular testing lab - will depend on multiple factors. Some of these factors relate to the lab doing the testing, which includes: experience of staff, pressure staff are working under, probability of contamination; the particular equipment and chemicals used; and, of course, the much discussed Ct-value threshold used. Other factors relate to the source of the samples being tested: for example, whether they were primarily from people with symptoms, whether they were primarily from people in hospitals, whether they were primarily from people who had been in recent contact with a COVID19 positive person, etc.  If any of these factors change, the false positive rate is likely to change. Without information about all these factors there is nothing we can conclude with any kind of certainty from counts of tests and positives alone.

However, as I explained in this article, by far the biggest factor affecting the false positive rate is whether and how a positive test is repeated before actually declaring that a person has tested positive (this is called confirmatory testing). Let us suppose that on any single test the current false positive rate is 0.01 (i.e. 1%). Then, if a lab does no confirmatory testing, 0.01 can be crudely assumed to be the false positive rate (meaning 1 in every 100 without the virus would wrongly test positive). However, if the lab does confirmatory testing whereby a new sample is taken from the person who tested positive, then it reasonable to assume that the two tests are 'independent' and in that case the probability a person who does not have the virus tests positive both times is 0.01 times 0.01 which is 0.0001, i.e. 0.01% (so only 1 in every 10,000 without the virus would wrongly be classified as testing positive). 

Hence, for a Lab in which confirmatory testing is carried out rigorously, the false positive rate can be close to 0 even if 'the false positive rate' for individual tests is relatively high.

When I posted this article suggesting that - due to pressure of massively increased numbers of tests - it might be possible that confirmatory testing was not always carried out - a number of indignant respondents on Twitter ridiculed the idea, insisting that confirmatory testing was always done. But, in the absence of any clear information about this everything points to the likelihood that rigorous confirmatory testing may not have been widely conducted at all since mass testing began in late summer. 

The official NHS Guidance document on PCR testing says the following (page 7):

Positive results that are sent by NHS England and NHS Improvement pathology network laboratories for confirmation to a PHE laboratory will be considered presumptive positives until confirmed. (see list –Appendix 2). Confirmation is not required if network laboratories are confident in the test they have adopted and assured of an accurate result. If in any doubt, samples can be referred to a PHE regional laboratory local to the NHS testing laboratory for confirmatory testing, for an initial period until the NHS network laboratory is assured their testing is robust, accurate and safe. After this time confirmation by local PHE laboratories will no longer be required. Presumptive positive/positive results will be notified to the co-ordination center for contact tracing, which will start immediately.

So we know that confirmation was never required providing the labs were sufficiently confident that they could 'accurately test' first time round. Even if confirmatory testing is done there is also no definitive requirement to test a new sample, meaning that the second test may not be independent of the first; in that case there will not be such a great reduction in the false positive rate.  

The simple anecdotal evidence that independent confirmatory testing is now unlikely to be done routinely is that we know many people who have got a positive test result (outside of hospitals) and they have never been asked to do a repeat test.

The Government website that produces the daily number of  'new cases' is supposed to only count 'confirmed' cases. In other words, it should not include any positive test where that was the first positive test for the person sampled. As we now know that this is no longer true it means that the published 'number of cases' are not all 'confirmed' cases. Hence, it will include a greater percentage of false positives than previous periods when more confirmatory testing was done.

The massive increase in testing since September means that it is very likely an increasingly smaller proportion of confirmatory testing has been done, which in turn means the false positive rate would have increased.  Similarly, when testing is scaled back the false positive rate is likely to decrease. While there was also likely to have been a small resurgence of COVID19 in September, the changes in false positive rate are likely to be the most significant causal explanation for the close correlation between number of tests and test positivity (% of test that are positive) that has been observed in all areas of the UK since September.  

Number of daily tests and test positivity rate for England

In contrast to the 'number of cases' which has risen steadily as more tests are conducted (and which cannot be considered as a reliable measure of the infection rate), the data on NHS triages for COVID (see https://digital.nhs.uk/dashboards/nhs-pathways) - which includes all 999, 111 calls and 111-online reports also supports the hypothesis that the big increases in number  of 'cases' since September is due in part to increase in false positives; while the greater availability of tests could explain some of the decrease in people using the NHS triage path, the data below still provides evidence that, after a small resurgence in September, the number of people actually ill with COVID has remained quite steady and much lower than it was in the spring.

 

NHS daily COVID triage reports (999, 111 calls and 111-online)

To give some idea of the impact of a changing false positive rate, let us suppose for illustration that (as above) the individual test false positive is 0.01 and that therefore the independent confirmatory false positive rate is 0.0001. Then, with confirmatory testing if the population infection rate is 0.5% (1 in 200) then, assuming an 80% true positive test rate, it follows from Bayes Theorem that 97.6% of those testing positive really are positive. However, without confirmatory testing only 28.7% of those testing positive really are positive.

There is no getting away from the fact that the data on testing and cases provided by the Government is insufficient to make any definitive conclusions about the true status of COVID19. I have already listed the many additional pieces of information needed to make anything other than massively uncertain conclusions. As this article has explained we also need to know how many of the new 'cases' were subject to confirmatory testing. Given the incredible resources being spent on testing, if the Government cannot provide the additional information I have listed (which all must be available somewhere) there really is no point in carrying out these tests. 

 

Update: There is also the issue of the difference between PCR test results and LFT test results

See also previously on this blog:

 

      

• Pooled COVID19 testing makes the data on 'cases' even more dubious  

  Remarkable relationship between number of tests and positivity rate when we drill down into regions
  

  COVID-19 in the UK: the remarkable divergence between number of 'cases' and number of people reporting symptoms
  

• How to explain an increase in proportion testing positive if there is no increase in infection rate
• We are still not getting the basic testing data we need
  
• Why we know so little about COVID19 from the data provided
• Impact of false positives in Covid testing
• Covid19 hospital admissions data: evidence of exponential increase?
  
• Don't panic: limits to what we know about Covid-19 PC testing, inferred infection rates and alse positive rates
  
• A privacy-preserving Bayesian network model for personalised COVID19 risk assessment and contact tracing
  Covid-19: Infection rates are higher, fatality rates lower than widely reported 
• Coronavirus: country comparisons are pointless unless we account for these biases in testing 
• Why most studies into COVID19 risk factors may be producing flawed conclusions - and how to fix the problem
• Causal explanations, error rates, and human judgment biases missing from the COVID-19 narrative and statistics
• Covid-19 risk for the black and minority ethnic community: why reports are misleading and create unjustified fear and anxiety
• UK Covid19 death rates by religion: Jews by far the highest and atheists by far the lowest 'overall' - but what does it mean?
• All COVID articles on this blog 

 

 

COVID-19 in the UK: the remarkable divergence between number of 'cases' and number of people reporting symptoms

6 Jan 2020: There is an important update to this article here

On 23 December over half a million people (507,384)  - the highest daily total yet - had a COVID-19 test in the UK (see https://coronavirus.data.gov.uk). That was an increase of 39.2% over the total tested (364,388) just one week earlier on 16 Dec. 

Yet, during exactly the same period there was an average daily total of just 8,500 total NHS triages (see https://digital.nhs.uk/dashboards/nhs-pathways) - and that number includes all 999, 111 calls and 111-online reports. The increase from 16 Dec (when it was 8239) to 23 Dec (when it was 9185) was just 11%. And compare how low those numbers are to when the virus was at its peak in March; for example, on 19 March there were 152,088 NHS triages. 

Four days after those 152,088 people reported symptoms through NHS triages there were, according to the Government website, 1,379 confirmed 'cases' (where, of course, 'cases' simply means a positive test result). On 18 Dec just 7,969 people reported symptoms through NHS triages. Yet, on 22 Dec there were 44,903 'cases'.  So, whereas in March less than 1% of the number reporting symptoms through the NHS tested positive, we now have a situation where, for every person reporting symptoms through the NHS we have nearly six testing positive.  If that March ratio applied today then the 7,969 people with symptoms would equate to about 72 'cases', not the 44,903 reported. 

The remarkably diverging difference between 'cases' and people reporting symptoms provides yet more evidence that the vast majority of those currently testing positive do not have the virus. 

There are causal explanations for part of the divergence: in March - unlike now -  it was primarily patients hospitalized with the virus who were being tested, so people who may have had the virus but had minor (or no symptoms) may not not have been recorded as 'cases'. It is also possible that more people with symptoms now are choosing to go straight for a test and not report their symptoms through the NHS. However, these causal explanations go nowhere near to explaining the scale of the divergence observed (625 times as many cases per people reporting symptoms since March).

We have, of course, been reporting on the problem with testing (especially PCR testing) for a long time now. Those who insist that the UK faces a continued and increasingly massive health crisis point to the rise in both number of 'COVID-19' deaths and hospitalisations. But those are simply the number of deaths and hospitalisations of people who have tested positive for COVID-19. Given the massive increase in testing, the official COVID-19 figures for deaths and hospitalisations are - like the number of 'cases' - completely uninformative.

The argument that an increase in test positivity rate (the percentage of tests that are positive) is - along with the increase in deaths and hospitalisations - definitive 'proof' of how 'bad things are' can also be easily challenged, for reasons discussed extensively here and here. 

It is also important to note that those who point to the Government-backed ZOE COVID symptom app tracker for estimates of number of cases are also being misled, because the website clearly states that the estimate is based on 'the app + swab tests'.

Finally, here is an update of the COVID 'case' data using the Government figures (but taking account of number of tests). All the usual caveats discussed here apply.

And it all point to the need for a reminder of this:

See also: Previous posts on COVID data

 

Pooled COVID19 testing makes the data on 'cases' even more dubious

This article by Clare Craig, Jonathan Engler, Mike Yeadon, and Christian McNeill lays out a convincing case why PCR testing for COVID19 has failed.

While I was aware of many of the issues they raise, until speaking with Clare, Jonathan and also Joel Smalley, Keith Johnson, Martin Neil and Scott McLachlan yesterday I was unaware of another major problem with PCR testing that is not discussed in the article. Specifically, when mass PCR testing for COVID19 started in the UK in late summer, so too did the idea to implement pooled testing as described in this official NHS document. It is not clear how extensively the idea has been implemented, but it could have created more problems.

The motivation for pooled testing for any virus is that you can do much more efficient testing when population infection rates are low. Instead of separately testing each sample you combine, say, 10 samples and test the pooled sample. When prevalence is low then the pooled sample will most likely test negative and so we can conclude that all of the individual samples in it were negative. On the relatively rare occasions when the pooled sample tests positive we conclude that at least one of the individual samples would be positive and so (only when this happens) we retest each of the individual samples to confirm which ones were positive.

A proposed implementation of the NHS Guidelines for pooled testing by McNally and Ball  is described here.

However, it is clear that there are the following fundamental problems with using pooled testing for COVID19. 

1. For a start, pooled testing - according to the NHS document - should be used only when the population infection rate is in the range 1% to 2% and when the infection rate among those being tested (the test positivity rate) is around 8%. But many areas currently have much higher test positivity rates (for example, as we can see below, London is now at 15% and the number tested there in the week ending 18 Dec was 410,463 - an increase of over 57% from the 261,075 just one week earlier). So, this should certainly rule out pooled testing. Moreover, the title of the NHS document suggests it should be used only on samples from asymptomatic people, which makes sense since those with symptoms will have a much higher probability of being infected. However, it is not clear if this has been the case. The NHS document says: 

 "It is also expected that as winter approaches the demand for COVID-19 testing will increase, putting more pressure on an already stretched system. Sample pooling strategies increase testing capacity when resources are limited.” 

If pooled testing was only used on asymptomatics, then testing data that differentiates between symptomatics and asymptomatics (data which I have argued is critical for determining the true infection rate) would be available, but we are constantly told it is not.

2. As explained here, calculating the optimum number of samples (and process) for pooled testing requires prior knowledge of: 

• The current true population infection rate
  
• The false positive test rate for asymptomatics (i.e. the probability an asymptomatic person without the virus tests positive)
• The false negative test rate for asymptomatics (i.e. the probability an asymptomatic person with the virus tests negative) 
  

Moreover, all of these values need to be low for pooled testing to be efficient and accurate. But, as I have made clear many times on this blog, NONE of these are known; indeed, one of the main purposes of testing asymptomatics is to find out what these really are.

So, while theoretically the concept of pooled testing (especially if you are testing asymptomatic people) for COVID19 is sound, practically it seems flawed before we start. 

3. Asymptomatics with the virus have low viral load, so even for a single sample it requires high Ct values to detect it. Because pooled samples are mixed from the original samples, even higher Ct values are needed to detect the virus if it is indeed there. But, while high Ct values decrease the false negative rate, they increase the false positive rate. This is tolerable when the population infection rate is very low because there will be sufficiently few positive samples so that they can all be retested individually. Suppose, for example, that the false positive rate for a pooled sample is 5% and the false positive rate for a single sample is 2%. Then the probability that an individual sample will wrongly test positive is the probability it will wrongly test positive both in a pooled sample and then individually. This is 0.05 times 0.02 which is 0.001, i.e. just 0.1%.  Similarly, if there is no pooled testing, but there is confirmatory testing for each individual sample that tests positive, then the probability that a sample will wrongly test positive twice is 0.02 times 0.02 which is 0.0004, i.e. 0.04%.

The false positive rates for a PCR test are illustrative. They could be higher or lower

 This explains why – in the summer when the infection rate was close to zero – the number of ‘positives’ (and necessarily the number of false positives) was also very low. All those people who have ridiculed the notion of false positive rates above 1% for PCR testing simply do not understand the role of confirmatory testing (whether in pooled or individual testing). But, while the conditions for pooled and/or confirmatory testing were OK for the summer (low population infection rate, lots of testing but mostly asymptomatic people) that changed as soon as the infection rate (as well as illnesses which have similar symptoms) increased in the autumn. 

4. Even if the population infection rate is stable any big increase in the number of tests conducted is likely to lead to a greater proportion of false positives whether or not pooled testing is used. This is because, as testing increases not only is there greater potential for various types of handling and contamination errors due to limited resources and time, but inevitably it seems fewer confirmatory tests can be carried out. If pooled testing is performed without confirmatory testing then the false positive rate (using the above example figures - see table) increases from 0.1% to 5%. That is a massive increase. If pooled testing is not applied then the false positive rate rises from 0.04% when confirmatory testing is used to 2% when it is not. 

It is difficult to find out just how much pooled testing has been going on and whether and when there has been a drop in confirmatory testing. It is likely that there will have been major differences between labs. But, given the massive increase in testing since September, if for whatever reason it is felt that confirmatory testing cannot be carried out, then this would explain  why we are seeing a very close correlation between number of tests and test positivity.

p.s. Jonathan Engler raises the following additional relevant points: 

"The PCR test has been worked up from bench-top to industrialised scale, in assay development terms, overnight. In particular, there is only quite limited testing of cross-reactivity with other viruses. What testing there was – as outlined in the Drosten paper – revealed some cross-reactivity.

Also from the Drosten paper they do report on 4 positive cases out of 310 negative controls containing other viruses: These examples of “weak initial reactivity” are somewhat brushed off, justified by them testing negative 2nd time around, but they do account for >1% of tests. (What if the 306 initial negatives had been tested again – might some of those then have then been positive on their 2nd test?). The point of this in relation to pooling is that there are clearly uncharacterised sources of false positivity, likely including but possibly not limited to unknown or undetected coronaviruses; even a single such instance could contaminate the entire pool, and then testing at high cycle rates could surely magnify such issues greatly."

London COVID testing data (from Govt website) 24 Dec 2020

 

 

 

See also:

• PCR based COVID testing has failed
  
• Pooling of asymptomatic SARS COV-2 COVID-19 samples for (PCR/or other) testing 
• Pooled testing for SARS-CoV-2 could provide the solution to UK’s testing strategy
• UK Covid Testing data: Remarkable relationship between number of tests and positivity rate when we drill down to regions

As London goes into Tier 4 COVID lockdown here is proof that the government data for London is flawed

With the sudden announcement of the Tier 4 lockdown for London yesterday I decided to look at the London hospital admissions and 'case' data (where, by definition, 'case' = positive test). Using the data at https://coronavirus.data.gov.uk and filtering on London I did a plot of hospital admissions as a percentage of cases (this is a crude measure of 'severity' of the virus at any time). I used 7-day rolling averages for both cases and admissions and assumed a 4-day lag from positive test to admission (none of these assumptions makes a lot of difference to the results). The results I got below are very interesting for two reasons.

 The first reason is that you can see that the percentage of 'cases' leading to hospitalisation has been steady at around 7% for over 3 months, so no increase in 'severity'. In fact, as recently as mid-September it was much higher at 12%, and in June it was really much higher at ....120%. Which brings me on the second reason why the results are, to say the least, interesting. It should, of course, be impossible for the number of COVID hospitalizations to be greater than the number of COVID 'cases'. Yet it is not my analysis that is in error here. Here are the raw data, for example, from June:

I have highlighted some examples of admission numbers which do not appear to be feasible given the previous days number of new cases. Now, of course, for any new admission we do not know date of the associated positive specimen. For my chart above I have assumed on average a hospital admission follows 4-day after a specimen was take that ended up positive. But we know that a hospital 'admission' is also classified as COVID when there is a (first) positive test after admission. However, whichever way you look at this data it cannot be correct. The data for COVID hospital admissions (in June at least) must be exaggerated (it cannot be that 'cases' are underestimated because a hospitalisation cannot be classified as COVID without a positive test, i.e. a recorded 'case').

In case anybody thinks the error must be in the way I have extracted the data, here are screen shots for relevant part of June from the government website:

Now, there they may be understandable reasons for errors in the Government data. For example, perhaps only a subset of hospital admissions were recorded in June, or perhaps there are different definitions of what 'London' is defined as for 'cases' compared to 'admissions'.  But, whatever, if the Government cannot get such basic data correct, it hardly inspires trust in their Draconian decisions. And how do we know that the number of COVID hospital 'admissions' is still not being exaggerated?

Update (21 Dec): Here is the London case fatality rate. Again, not much here to suggest the need for panic.

See also:

Remarkable relationship between number of tests and positivity rate when we drill down into regions

UK Covid Testing data: Remarkable relationship between number of tests and positivity rate when we drill down to regions

23 Dec update: This article provides further explanation for these observed results

Following on from this post where I expressed concern at the unavailability of testing data broken down into regions, the data are now available at https://coronavirus.data.gov.uk/details/testing. And the results reveal something startling that you simply cannot see by looking at just the UK data overall. 

First of all here is the overall (not very informative) plot of the number of daily tests versus percentage of positive tests (also called the positivity rate) for the whole UK:

Now even if there was no genuine increase in the population infection rate then, when you massively increase the number of people being tested (as happened in early September), you would expect to see a corresponding increase in number of 'cases' (i.e. positive test results) simply because of the probability of false positives. Hence, many argued that the increase in cases observed when testing was ramped up in September was explained largely by the false positive rate. However, if this hypothesis were correct, we would not see an increase in the positivity rate - unless the false positive rate increased (e.g. due to increased human error) as more people were tested. The fact there was an increase in positivity rate in September therefore seems to confirm a major increase in infection rate in that period.

Since early November there is no obvious correlation between number of tests and positivity rate. Hence, is seems justifiable to conclude that the number of tests performed has no impact on the positivity rate (or vice versa) and hence that the recent (slight) increase in positivity rate since early Dec represents a genuine increase in spread of the virus that may justify the renewed lockdowns.

However, it turns out the overall UK graph is hiding what is happening at the regional level. The following regional graphs of number of tests versus positivity rate are taken straight from https://coronavirus.data.gov.uk/details/testing (Note that the website only has the regional breakdown for England; for Scotland Wales and N Ireland we do not even have the overall number of tests):

 

The trend plots in these main regions are different, with increases and decreases in positivity rates occurring at different times. Yet, in each region since mid-September, the positivity rate very closely correlates with the number of tests irrespective of the direction of movement.

But it doesn't stop there. The website also now provides the same data at a lower level of granularity - namely health authorities. And, when you drill down to those, you can see from these examples chosen at random (despite some very different trend plots at these local levels), the correlation between number of tests and positivity rates is almost perfect:

For example, both the number of tests and the positivity rate for Ashford have recently been steadily increasing while in North Tyneside both the number of tests and the positivity rate have been steadily decreasing.

So, what can we conclude from these remarkable local correlations and why they started in September? [UPDATE: see this article for explanation] It suggests some systematic non natural factor independent of a virus. Otherwise two possible (causal) hypotheses are:

1. That the testing is highly accurate and since September people have only got tested when they think they might have the virus. As the infection rate increases (resp. decreases) the number of people choosing to get tested increases (resp. decreases).
2. That as the number of people tested increases (resp. decreases) the false positive rate increases (resp. decreases) due to increased (resp. decreased) possibility of human error.

One definite problem with hypothesis 1 is that - if it were true - we would expect to see similarly identical (but delayed) trend plots in hospital admissions and deaths, which does not seem to be the case. Another definite problem with hypothesis 1 is that it relies on the assumption that people only get tested when they either have symptoms or have been in recent contact with a person recently tested positive. But we know this is not the case. A very large proportion of people tested since September (including tens of thousands of students) have had neither symptoms nor a recent positive contact. 

One possible problem with hypothesis 2 is that it it fails to explain what happened between July and September when testing increased but there was no real increase in the positivity rate. Well, on the one hand testing levels were still sufficiently low for there to be plenty of checks in place (including confirmatory tests for positive results); that may not have been possible with such a big increase in testing in the autumn; also nobody doubts that, while the virus was almost gone in the summer there was an increase in infections in the autumn, so it is inevitable that the positivity rate would start rising then.  But, in the absence of some systematic non natural factor independent of a virus, the fact we see remarkable correlations involving rises and falls since September suggest that this must at least in part be explained by hypothesis 2.  In other words, if a local authority decides to increase testing then they would see an increase not just in number of 'cases' but also in the positivity rate. In that case a decision simply to increase testing could lead to a region being moved to a higher lockdown 'tier'. 

Of course, because of the massive limitations of what can be concluded from simple data like number of tests and number of positive (as explained here) there is massive uncertainty about any of the above conclusions. The analysis certainly supports the need for the kind of additional data for which I have been I have been arguing for a long time; in particular, we need to know not just number of people being tested but also number of people tested who are asymptomatic and – of those  testing positive – the number who subsequently developed real symptoms. Only then might we be able to accurately determine whether the infection rate is really increasing or decreasing. 

See also

• How to explain an increase in proportion testing positive if there is no increase in infection rate
• We are still not getting the basic testing data we need
  
• Why we know so little about COVID19 from the data provided
• Impact of false positives in Covid testing
• Covid19 hospital admissions data: evidence of exponenial increase?
  
• Don't panic: limits to what we know about Covid-19 PC testing, inferred infection rates and alse positive rates
  
• A privacy-preserving Bayesian network model for personalised COVID19 risk assessment and contact tracing
  Covid-19: Infection rates are higher, fatality rates lower than widely reported 
• Coronavirus: country comparisons are pointless unless we account for these biases in testing 
• Why most studies into COVID19 risk factors may be producing flawed conclusions - and how to fix the problem
• Causal explanations, error rates, and human judgment biases missing from the COVID-19 narrative and statistics
• Covid-19 risk for the black and minority ethnic community: why reports are misleading and create unjustified fear and anxiety
• UK Covid19 death rates by religion: Jews by far the highest and atheists by far the lowest 'overall' - but what does it mean?
• All COVID articles on this blog 

We still are not getting the most basic data needed about COVID-19 testing

UPDATE: This follow up article addresses some of the issues raised here.

As I have been arguing regularly on this blog, there is no point in citing numbers of COVID-19 "cases" (which are not necessarily people 'ill' or even with symptoms, but simply the number of tests that are positive) without also citing the number of tests performed. 

The (Government) decision about whether to move a region into a higher (resp. lower) tier is based on whether the number of "cases" goes above (resp. below) a certain threshold number per 100,000 residents. But, as there are wide variations in the number of people per 100,000 who are tested, this strategy is ludicrous. Because (since July) many (most?) people tested do not have any symptoms and (because of the various reasons why there may be false positives), the more people you test the more "cases" you will find. Hence, a region which has a  genuinely low infection rate, but a disproportionately high number of tests, may find itself in Tier 3 lockdown; conversely, a district with a genuinely high infection rate, but a disproportionately low number of tests, may find itself in the relative 'freedom' of Tier 1. [Update: as David Paton notes, the ZOE symptom app data also suggests the decisions about Tiers is irrational]. https://twitter.com/cricketwyvern/status/1338843277232115713?s=20

The Government website does provide the national figures for daily number of tests, which means that I can produce plots such as these which show why it is so important to consider the number of tests conducted instead of simply citing number of "cases".

 

Number of cases increase when numbers tested increase (but note the axes have different scales, but I have clarified this with different colours)

With the increase in more random testing, the blue line is more informative of the national COVID19 status than the red line, but it is the latter not the former, driving policy decisions

 

As soon as we produce such plots it becomes clear that the scale of the 'second wave crisis' has been massively exaggerated (see yesterday's post for more upated plots that factor in numbers tested) and that there is merit to the claim that what we have is a 'casedemic'.

Yesterday I decided to try to do similar analyses per region (i.e. taking account of the number of daily tests per 1000 residents) to see whether it was the numbers tested that was driving, for example, the decision to move certain regions (like London) to Tier 3. However, curiously the data are not available even though it is supposed to be. [18/12/20 UPDATE: the regional testing data is now available] While the Government website gives the overall national testing figures, the same website fails to give the figures for individual regions, even though that it appears to offer exactly this option at https://coronavirus.data.gov.uk/details/download. 

That page invites you to choose and download a whole range of testing data for any region. But no matter which daily testing data you select those particular fields  (but not, e.g. "daily cases" and "daily deaths") always come up empty when you download the file [UPDATE: the page is now no longer even allowing you to select to download by different regions].  

I also tried searching the websites/dashboards provided by different regions/local authorities themselves. Some of those I found, such as Buckinghamshire and London provide a lot of very useful data (including the daily number of people reporting symptoms - see screenshots below), but curiously they also do not have the daily testing numbers.

So, given that millions of people's lives are adversely affected by the decisions made based on these numbers, the fact that they are being hidden is becoming increasingly disturbing. If the numbers tested per region are consistent (i.e. all regions have a similar number of tests per 1000 residents) then there is not a problem when it comes to regional decisions. In that case, let's see the data showing this consistency. If they are not consistent (as I suspect) then we are all being conned. 

One final point: people have been arguing with me that my concerns are 'wrong' because what matters is that, for example, in London "COVID19 hospital admissions" and "deaths" are rising. But notwithstanding the fact that overall autumn/winter rises are inevitable, the problem is that the data on number of hospital admissions and deaths attributed to COVID19 suffer from the same problems as COVID19 "case" data: we actually have no idea at all how many of those classified as "having COVID19" really do have a virus that causes hospitalization and death. What we do know is that a very large proportion of those classified as COVID19 hospital admissions were admitted for something other than COVID19 but tested positive for it after admission. Similarly, the death numbers are based on anybody who had a positive test within 28 days of death irrespective of the actual cause. All this means is that as testing numbers increase, so inevitably will the number of hospital admissions and deaths even if some (or even most) have nothing to with COVID19.

London COVID19 dashboard, 15 Dec 2020. Note that there is little here to explain the reason for the decision to move London to Tier 3.  https://data.london.gov.uk/dataset/coronavirus--covid-19--cases

Buckinghamshire dasboard, 15 Dec 2020. https://covid-dashboard.buckinghamshire.gov.uk/

See:  All COVID articles on this blog