Read on Twitter
What’s going on with COVID testing in the UK? Here are the figures for the last two weeks https://coronavirus.data.gov.uk/cases 1/24
They fell off a cliff because @PHE_UK introduced more stringent testing criteria. Their aim was to reduce false positive results that have been reported. 2/24 https://www.gov.uk/government/publications/sars-cov-2-rna-testing-assurance-of-positive-results-during-periods-of-low-prevalence
The jury is out on how big a problem this is but, having analysed the data I think it accounts for almost all results through the summer 3/24 https://logicinthetimeofcovid.com/2020/09/07/waiting-for-zero/
Since the end of June we have seen positive tests account for 1% or fewer of the tests performed. The increased numbers of tests makes for more positive tests but the percentage is still negligible. 4/24 https://twitter.com/TechClive/status/1303332998197252096
This has allowed us to establish what the false positive rate is for the different populations being tested. Community testing having a higher proportion of symptomatic patients, results in a higher false positive. 5/24 https://logicinthetimeofcovid.com/2020/09/08/variation-in-the-false-positive-rate-2/
This is what a real outbreak looks like. The numbers are percentages of positive tests. Until that first 6% none of those results were real. They were false positives and should have been disregarded. 6/24 https://twitter.com/vincentglad/status/1303243869933404161?s=20
Testing remains important to find the 6% when it appears. We need two systems. Cases that were PCR screening positive (current testing). Results for confirmed cases. 7/24
If there are only 13 convincingly positive cases in England then how big is the outbreak in the North West or among 20-29 yr olds? 8/24
All the data is meaningless without knowing the denominator - how many 20-29 yr old and North Westerners were tested? Has testing focused on these groups? 9/24
False positives occur at a set rate for a population. If you test more of one group within that you will find more positives. i.e. ~1% of all those tested will be positive regardless. 10/24
Testing has to strike a balance between not missing cases and calling cases positive when they are not. There are times when we really do not want to miss any. During a major outbreak picking up every possible case must be the priority. 11/24
Those times passed at the end of June. Since then the priority should have been not calling cases positive when they are not. 12/24
PHE have taken a step in the right direction to tackle the false positive problem. There would have been 2800+ positive cases (we do not know how many were tested yet). This has been revised down to 13. 13/24
The remaining cases are being assessed further. If this means repeating a PCR test then for many the cause of the false positive in the first round will still exist for the second test. 14/24
False positives are not a mistake that happens in a lab (though they can be). They are a function of the population being tested. 15/24
We can expect to see case numbers rise as these second PCR assessments filter through. This will raise the R value considerably for a time. 16/24
Are the 13 cases real? Without seeing extra evidence of these cases all being contacts or infecting others or characteristic chest CT findings then I would still be sceptical. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7185831/ 17/24
With the number of tests being carried out detecting a few people for whom the result looks convincing is inevitable. 13 makes a rate of <0.01%. 18/24
Does that mean it is all over? Absolutely not. Much of the country remains vulnerable and antibody levels are low everywhere compared with where they need to be. 19/24 https://www.thelancet.com/pdfs/journals/lancet/PIIS0140-6736(20)31484-7.pdf
Moving forward, testing should continue. Results from regions with positives under 1.5% of testing numbers should be subject to scrutiny before reporting. 20/24
For any region that has more than 1.5% we still need to be careful. Claims of an outbreak are extraordinary claims and need extraordinary evidence. Proof of transmission or of CT chest findings should be required. 21/24
This is standard practice in New Zealand where PCR is used as a screening test only. 22/24
Having established an outbreak then every positive test should be reported as such. If we start to get outbreaks then finding every case becomes more important than occasional false positive tests. Having a timely result in those circumstances is also critical. 23/24
Coronaviruses are seasonal. If you have vulnerable relatives I recommend putting your sprouts on for Christmas in November - but make sure you eat them before they go soggy. 24/24
