Read on Twitter
1/ Grab a cup of coffee
We are now seeing striking correlations between (i) the # of COVID19 tests and (ii) % positivity within/across certain important individual PHUs in Ontario.
This has implications for local lockdown policy, and for the global PCR/false-positive debate...
We are now seeing striking correlations between (i) the # of COVID19 tests and (ii) % positivity within/across certain important individual PHUs in Ontario.
This has implications for local lockdown policy, and for the global PCR/false-positive debate...
2/
If you are not in Ontario, but involved with / following the PCR debate, this is still of interest to you. We are seeing similar testing/positivity correlation dynamics as seen in the UK. This thread borrows/builds substantially from @profnfenton here: https://twitter.com/profnfenton/status/1340032960284794881?s=20
If you are not in Ontario, but involved with / following the PCR debate, this is still of interest to you. We are seeing similar testing/positivity correlation dynamics as seen in the UK. This thread borrows/builds substantially from @profnfenton here: https://twitter.com/profnfenton/status/1340032960284794881?s=20
3/
Upfront caveat: correlation does not mean causation, and it is difficult to draw definitive conclusions from simple testing data. However, it is important to determine whether lockdown policy (for its unmeasurable collateral damage) is informed by false/flawed test results.
Upfront caveat: correlation does not mean causation, and it is difficult to draw definitive conclusions from simple testing data. However, it is important to determine whether lockdown policy (for its unmeasurable collateral damage) is informed by false/flawed test results.
4a/
First, the overall picture in Ontario. The daily # of tests and daily percent positivity (7-day rolling avgs) shown below.
Province-wide, we see no substantive correlation between tests & %pos, suggesting genuine fall increases in %pos, driving further lockdown policy.
First, the overall picture in Ontario. The daily # of tests and daily percent positivity (7-day rolling avgs) shown below.
Province-wide, we see no substantive correlation between tests & %pos, suggesting genuine fall increases in %pos, driving further lockdown policy.
4b/
…and here are the two largest Public Health Units (“PHU”s) and population centers, Toronto (3.1m people) and Peel (1.6m))
Again, no obvious testing/%pos correlations, especially recently…
(sidenote: %pos peaked & is on descent almost 2-weeks before Nov 23rd lockdowns…)
…and here are the two largest Public Health Units (“PHU”s) and population centers, Toronto (3.1m people) and Peel (1.6m))
Again, no obvious testing/%pos correlations, especially recently…
(sidenote: %pos peaked & is on descent almost 2-weeks before Nov 23rd lockdowns…)
5a/
But digging into individual PHU data (recently published by the province), we see striking underlying testing volume & %pos trends, esp. since Nov. 1, hiding in plain sight...
Here are 8 regional PHUs combined representing 2.5m people (shown individually in next tweet)….
But digging into individual PHU data (recently published by the province), we see striking underlying testing volume & %pos trends, esp. since Nov. 1, hiding in plain sight...
Here are 8 regional PHUs combined representing 2.5m people (shown individually in next tweet)….
5b/
Here in each of those 8 regional PHUs outside the GTA (including recently controversial PHUs w/dramatic case growth (Windsor/London/Hamilton)), we see tests and % pos are correlating tightly, corresponding with large testing growth, starting around Nov 1.
Here in each of those 8 regional PHUs outside the GTA (including recently controversial PHUs w/dramatic case growth (Windsor/London/Hamilton)), we see tests and % pos are correlating tightly, corresponding with large testing growth, starting around Nov 1.
5c/
Same charts again. Tests/%pos is increasing and decreasing at various times and different rates for these various PHUs, yet all exhibit strong correlations…
Same charts again. Tests/%pos is increasing and decreasing at various times and different rates for these various PHUs, yet all exhibit strong correlations…
5d/
This simple chart is effectively a ranking of all PHUs by the *strength of recent correlations* between testing and % pos. Note many have strong positive correlations, including near perfect correlations for Windsor/London/Hamilton (latter 2 facing lockdown prospects).
This simple chart is effectively a ranking of all PHUs by the *strength of recent correlations* between testing and % pos. Note many have strong positive correlations, including near perfect correlations for Windsor/London/Hamilton (latter 2 facing lockdown prospects).
5e/
Lastly on specific Ontario PHUs: the peculiar case of Thunder-Bay. A mid-sized but somewhat remote Northern city of ~150k
…almost entirely escaped the fall “wave”, until it exploded from <0.5% pos to >2.0% in <3 weeks, and peaked
(is this seasonality or something else?)
Lastly on specific Ontario PHUs: the peculiar case of Thunder-Bay. A mid-sized but somewhat remote Northern city of ~150k
…almost entirely escaped the fall “wave”, until it exploded from <0.5% pos to >2.0% in <3 weeks, and peaked
(is this seasonality or something else?)
6a/
So what can we conclude?
First note: in next tweets I take substantial liberty borrowing from @profnfenton from his short paper linked here that illuminates similar dynamics happening across UK PHUs since September. I hope you don’t mind professor…
https://probabilityandlaw.blogspot.com/2020/12/uk-covid-testing-data-remarkable.html
So what can we conclude?
First note: in next tweets I take substantial liberty borrowing from @profnfenton from his short paper linked here that illuminates similar dynamics happening across UK PHUs since September. I hope you don’t mind professor…
https://probabilityandlaw.blogspot.com/2020/12/uk-covid-testing-data-remarkable.html
6b/
The correlations *potentially* suggest some “systematic non-natural factor independent of a virus” - @profnfenton, who poses two hypotheses…
The correlations *potentially* suggest some “systematic non-natural factor independent of a virus” - @profnfenton, who poses two hypotheses…
6c/ …
1. Testing is very accurate; as more (less) people think they have the virus, more (less) get tested (both tests & positivity increase/decrease)
2. As testing increases (decreases); false-positivity increases (decreases) due to possibility of human error.
1. Testing is very accurate; as more (less) people think they have the virus, more (less) get tested (both tests & positivity increase/decrease)
2. As testing increases (decreases); false-positivity increases (decreases) due to possibility of human error.
6d/
Here the professor outlines limitations of the 2 hypotheses, but makes an argument for #2 (false positivity/lab errors +/- with test volume). The arguments seem universal, and applicable to the dynamics I outlined for Ontario… see the highlight for the TL;DR…
Here the professor outlines limitations of the 2 hypotheses, but makes an argument for #2 (false positivity/lab errors +/- with test volume). The arguments seem universal, and applicable to the dynamics I outlined for Ontario… see the highlight for the TL;DR…
7/
In support of hypothesis #2…
…an anecdote from Windsor/London, who are seeing large cases rises. According to the below, Windsor sends most tests to London, which has ~3,000 tests/day capacity…
# of recent daily tests taken in Windsor/London alone? More than 3,000…
In support of hypothesis #2…
…an anecdote from Windsor/London, who are seeing large cases rises. According to the below, Windsor sends most tests to London, which has ~3,000 tests/day capacity…
# of recent daily tests taken in Windsor/London alone? More than 3,000…
8/
Important to note I am not making any assertions/ conclusions of actual lab error. I am only providing data and information that is publicly available, and asking reasonable, common sense questions. If anyone in public health Ontario has insights, we’re all ears.
Important to note I am not making any assertions/ conclusions of actual lab error. I am only providing data and information that is publicly available, and asking reasonable, common sense questions. If anyone in public health Ontario has insights, we’re all ears.
9/
As @profnfention notes, we need more data; not just the number of people tested, but also those who are asymptomatic being tested, and those who tested positive, how many developed symptoms.
Well…in Canada, we actually have SOME interesting data on this…
As @profnfention notes, we need more data; not just the number of people tested, but also those who are asymptomatic being tested, and those who tested positive, how many developed symptoms.
Well…in Canada, we actually have SOME interesting data on this…
10/
…in Canada we have a program called “FluWatchers” which prior to #COVID19 attempted to surveil influenza through voluntary participants for that responded to a weekly email whether they had a cough and fever in the week.
https://www.canada.ca/en/public-health/services/diseases/flu-influenza/fluwatcher.html
…in Canada we have a program called “FluWatchers” which prior to #COVID19 attempted to surveil influenza through voluntary participants for that responded to a weekly email whether they had a cough and fever in the week.
https://www.canada.ca/en/public-health/services/diseases/flu-influenza/fluwatcher.html
11a/
This data is now included in Canada’s weekly COVID19 Epidemiological Report, and here is a screenshot of the FluWatchers data (those reporting cough OR fever) in the most recent report, which shows trends since March…
This data is now included in Canada’s weekly COVID19 Epidemiological Report, and here is a screenshot of the FluWatchers data (those reporting cough OR fever) in the most recent report, which shows trends since March…
11b/
…look closely at the green bar (% of respondents reporting cough OR fever)…this # has remained in the 2.3%-2.7% range all fall/winter…and is inconsistent with positivity rises we have seen post Oct 1 in many areas…
…& is below seasonal expectations vs. prior years…
…look closely at the green bar (% of respondents reporting cough OR fever)…this # has remained in the 2.3%-2.7% range all fall/winter…and is inconsistent with positivity rises we have seen post Oct 1 in many areas…
…& is below seasonal expectations vs. prior years…
11c/
…again, it’s very hard to draw conclusions from this (only ~12,000 people nationwide are surveyed weekly), but it is very interesting… especially when noting the late September bump, consistent with the late September bump in Ontario (after which correlations appeared)
…again, it’s very hard to draw conclusions from this (only ~12,000 people nationwide are surveyed weekly), but it is very interesting… especially when noting the late September bump, consistent with the late September bump in Ontario (after which correlations appeared)
12/
Lastly, despite the recent sharp rises in % positivity in many Ontario PHUs outside of the GTA, according to the ACES Pandemic Tracker, admissions in Ontario ex-GTA are normal/below-normal for this time of year…
Lastly, despite the recent sharp rises in % positivity in many Ontario PHUs outside of the GTA, according to the ACES Pandemic Tracker, admissions in Ontario ex-GTA are normal/below-normal for this time of year…
13a/
Wrapping up, I want to break down the reported % positivity by PHU (& groups of PHUs) in Ontario. The province is mulling sweeping lockdowns this weekend, and its important to stay informed of regional trends, which in my view do not support damaging lockdowns…
Wrapping up, I want to break down the reported % positivity by PHU (& groups of PHUs) in Ontario. The province is mulling sweeping lockdowns this weekend, and its important to stay informed of regional trends, which in my view do not support damaging lockdowns…
13b/
This is the largely useless chart put forth by the Ontario Science Table in their most recent “Science Brief”, which I will clean up and clarify in the next tweet…
This is the largely useless chart put forth by the Ontario Science Table in their most recent “Science Brief”, which I will clean up and clarify in the next tweet…
13c/
Here is Ontario broken down into five areas… the only major PHU grouping still experiencing % pos increases is the Windsor/London/Hamilton group (and we have asked questions above regarding these increases)…
Here is Ontario broken down into five areas… the only major PHU grouping still experiencing % pos increases is the Windsor/London/Hamilton group (and we have asked questions above regarding these increases)…
13d/
Same chart. Note that October 10th restaurant/gym restrictions had no appreciable effect on % positivity rises, and Toronto and Peel *peaked* (so far) well before the November 23rd lockdown!
Same chart. Note that October 10th restaurant/gym restrictions had no appreciable effect on % positivity rises, and Toronto and Peel *peaked* (so far) well before the November 23rd lockdown!
14/
If you made it to the end, thank you for reading this weighty thread.
The more we debate and get answers to common sense questions, perhaps we can put a quicker end to abusive, damaging, and non-sensical lockdowns.
Please share widely.
If you made it to the end, thank you for reading this weighty thread.
The more we debate and get answers to common sense questions, perhaps we can put a quicker end to abusive, damaging, and non-sensical lockdowns.
Please share widely.
