Update on Covid providing information on prevalence and hospital admissions for England and its regions. This post is best viewed using the browser or Substack app.
I've said it before and I'll say it again: your weekly reports are invaluable, Bob. It must involve a fair amount of work, but that work is hugely appreciated, especially as sufficient reliable information has rarely if ever been available from government or the health authorities.
After last weeks uptick Scotland's CARI positivity rate has continued its downward trend with todays dashboard showing a fall to 8.4% for w/e 25 Aug although hospital admissions look like they will increase for w/e 25 Aug when the numbers get confirmed next week.
Just to let you know that I present the test positivity data for England as it's published by UKHSA, which currently every two weeks. So it will be included in next week's report.
I also keep an eye on the test positivity rates for Scotland which did see a small uptick this week after 4 weeks of sustained falls. It will be interesting to see what happens to this rate next week as this weeks sample size was smaller than previous weeks which is reflected in the larger upper and lower confidence limits.
One of the challenges I have including the test positivity for other nations is that it has become increasingly difficult to interpret the positivity data, particularly as testing policies and calculation methods have changed over time.
For instance, the test positivity rate in Scotland differs from that of England because it is based solely on tests conducted at a selection of GP practices, while the positivity rate for England includes tests done in both hospitals and GP practices. Scotland employs this method to get a more immediate indication of community prevalence by avoiding the delay inherent in hospital testing. However, this reduces the number of tests used to calculate the positivity rate making it more sensitive to changes in testing behaviour at the GP practices.
Is that correct (re: test positivity data for England)? The dashboard currently has positivity data up to and including 17th August (last Saturday), so presumably must that have been updated earlier this week at the latest?
Rob -- You are right that the dashboard reports test positivity based on a rolling seven day average and is updated weekly.
However, although the dashboard documentation says that it is the same data as the test positivity in the UKHSA bi-weekly report they do differ. In the past year, the dashboard test positivity rate has been consistently higher than the bi-weekly report varying between 10 to 60% higher.
In previous years, the discrepancy was even larger and sometimes the two positivity rates moved in opposite directions, especially when we moved from mass PCR testing to LFD testing to targeted PCR testing. For consistency, this led me to only report the test positivity from the National influenza and COVID-19 surveillance report as it is based on hospital testing which is more standardised. The report also provides test positivity data for other respiratory viruses (including flu) which provide useful comparators. Although this report is bi-weekly over the summer months it will return to being published weekly in the autumn.
England has a GP sentinel surveillance which should be roughly comparable with the Scotland CARI program. The GP surveillance is distinct from the Respiratory Datamart which seems to be more based on hospital data. The GP sentinel data is located under "RCGP sentinel swabbing scheme in England" in the bi-weekly flu report.
Yes the National influenza and COVID-19 surveillance report does include GP sentinel data but again for a small number of tests. Also it is not included in the accompanying data file at present. The test positivity data that I use for my situation report is from the Respiratory DataMart system (Figure 1a) and you will see this in next weeks report.
Considering something like 77% of flu infections and 80% of cold infections never produce any symptoms, safe to say only a small portion of coronavirus infections ever get picked up by the Respiratory Datamart data. Only the ONS's general population infection study can pick up a clear picture of all the infections.
Clearly the ONS Covid Infection Survey (CIS) is the 'Gold Standard' way to assess prevalence in the general population. However, we don't have that data at the moment so have to use the metrics available to us today to assess the situation.
Whenever I present test positivity I make clear it is not the same as prevalence. I also show that it did follow the same trend as prevalence when the ONS Winter CIS was running and is a useful proxy for Covid levels.
Good point. A high test positivity does not necessarily mean a high prevalence. It could mean the virus is very virulent and / or the virus is very prevalent. Take flu for example. Flu has very high test positivity, but very low prevalence, because it is very virulent. On the other end of the spectrum, seasonal coronaviruses have very low test positivity, but very high prevalence, but they are very mild.
I've said it before and I'll say it again: your weekly reports are invaluable, Bob. It must involve a fair amount of work, but that work is hugely appreciated, especially as sufficient reliable information has rarely if ever been available from government or the health authorities.
🙏
Thank you so much for your hard work. The data you provide is so valuable and useful.
Very grateful that you keep these coming!
What about positivity rate? Scotland and Canada are seeing an uptick. Scotland up to 12.9%. Canada up to 15.6%.
https://scotland.shinyapps.io/phs-respiratory-covid-19/
https://health-infobase.canada.ca/respiratory-virus-detections/
After last weeks uptick Scotland's CARI positivity rate has continued its downward trend with todays dashboard showing a fall to 8.4% for w/e 25 Aug although hospital admissions look like they will increase for w/e 25 Aug when the numbers get confirmed next week.
Jeremy,
Just to let you know that I present the test positivity data for England as it's published by UKHSA, which currently every two weeks. So it will be included in next week's report.
I also keep an eye on the test positivity rates for Scotland which did see a small uptick this week after 4 weeks of sustained falls. It will be interesting to see what happens to this rate next week as this weeks sample size was smaller than previous weeks which is reflected in the larger upper and lower confidence limits.
One of the challenges I have including the test positivity for other nations is that it has become increasingly difficult to interpret the positivity data, particularly as testing policies and calculation methods have changed over time.
For instance, the test positivity rate in Scotland differs from that of England because it is based solely on tests conducted at a selection of GP practices, while the positivity rate for England includes tests done in both hospitals and GP practices. Scotland employs this method to get a more immediate indication of community prevalence by avoiding the delay inherent in hospital testing. However, this reduces the number of tests used to calculate the positivity rate making it more sensitive to changes in testing behaviour at the GP practices.
Is that correct (re: test positivity data for England)? The dashboard currently has positivity data up to and including 17th August (last Saturday), so presumably must that have been updated earlier this week at the latest?
Rob -- You are right that the dashboard reports test positivity based on a rolling seven day average and is updated weekly.
However, although the dashboard documentation says that it is the same data as the test positivity in the UKHSA bi-weekly report they do differ. In the past year, the dashboard test positivity rate has been consistently higher than the bi-weekly report varying between 10 to 60% higher.
In previous years, the discrepancy was even larger and sometimes the two positivity rates moved in opposite directions, especially when we moved from mass PCR testing to LFD testing to targeted PCR testing. For consistency, this led me to only report the test positivity from the National influenza and COVID-19 surveillance report as it is based on hospital testing which is more standardised. The report also provides test positivity data for other respiratory viruses (including flu) which provide useful comparators. Although this report is bi-weekly over the summer months it will return to being published weekly in the autumn.
England has a GP sentinel surveillance which should be roughly comparable with the Scotland CARI program. The GP surveillance is distinct from the Respiratory Datamart which seems to be more based on hospital data. The GP sentinel data is located under "RCGP sentinel swabbing scheme in England" in the bi-weekly flu report.
For example in https://assets.publishing.service.gov.uk/media/66bdf4b3c32366481ca4915c/Weekly_Flu_and_COVID-19_report_w33-1.pdf
Jeremy,
Yes the National influenza and COVID-19 surveillance report does include GP sentinel data but again for a small number of tests. Also it is not included in the accompanying data file at present. The test positivity data that I use for my situation report is from the Respiratory DataMart system (Figure 1a) and you will see this in next weeks report.
Considering something like 77% of flu infections and 80% of cold infections never produce any symptoms, safe to say only a small portion of coronavirus infections ever get picked up by the Respiratory Datamart data. Only the ONS's general population infection study can pick up a clear picture of all the infections.
https://www.cidrap.umn.edu/h1n1-2009-pandemic-influenza/uk-flu-study-many-are-infected-few-are-sick
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7127333/
Jeremy,
Clearly the ONS Covid Infection Survey (CIS) is the 'Gold Standard' way to assess prevalence in the general population. However, we don't have that data at the moment so have to use the metrics available to us today to assess the situation.
Whenever I present test positivity I make clear it is not the same as prevalence. I also show that it did follow the same trend as prevalence when the ONS Winter CIS was running and is a useful proxy for Covid levels.
Good point. A high test positivity does not necessarily mean a high prevalence. It could mean the virus is very virulent and / or the virus is very prevalent. Take flu for example. Flu has very high test positivity, but very low prevalence, because it is very virulent. On the other end of the spectrum, seasonal coronaviruses have very low test positivity, but very high prevalence, but they are very mild.
Here's an interesting study coming out of Kansas City which found very little flu in their prevalence study: https://www.cidrap.umn.edu/enterovirus-non-polio/fourth-kansas-city-school-nasal-swabs-test-positive-respiratory-viruses-study