“IT’S JUST THE TESTING, BRO.”

Are we seeing more #COVID19 cases in #Canada’s large provinces now compared to July just because we are testing more?

No. A thread. (1/13)

All 4 of the largest provinces have seen a resurgence in cases. In just the past week, the change in 7-day rolling avg: (2/13)
QC: +55.6% (115.9/day 180.3/day)
ON: +29.5% (124.6/day 161.3/day)
BC: +22.2% (86.7/day 106.0/day)
AB: +4.5% (140.4/day 146.7/day)

Some provinces are undoubtedly testing more, particularly BC. In early July, the province was performing remarkably few tests, fewer than 2,000/day on average. They’re doing closer to 5,000/day now. (3/13)

Alberta has gone from just over 6,000/day to closer to over 10,000/day. Ontario and Quebec have been more or less flat since August at ~26,000/day and ~/9,000/day, respectively. How about cases? (4/13)

All four provinces have seen significant growth in daily case numbers since mid-August. Cases in BC have been growing continuously since July, from about 10 cases/day on July 1 to 100 cases/day now, a ten-fold increase. (5/13)

In the same time span, Alberta has gone from 50 cases/day to about 150/day now. Ontario has about same as at the beginning of July, around 160–170 cases/day. However, Ontario was at 85–90 cases/day in mid-August. (6/13)

Quebec’s trajectory has been like a roller-coast: ~75 cases/day at the beginning of July, increasing throughout the month, then down to ~75 cases/day again by mid-August. Today the province is at 180 cases/day. (7/13)

The rapid recent climb in cases in Ontario and Quebec cannot be explained by increases in testing, as testing has not changed much over this period. Alberta and BC have increased testing since the beginning of July, by about 2.5 times and just under 2 times, respectively. (8/13)

In all cases, growth in cases has far outstripped growth in testing. There is a simple reason for this: exponential growth. (9/13)

Besides, we don’t expect detected cases to scale linearly with the number of tests performed. When tests are scarce, those with the highest probability of testing positive are prioritized (e.g., the sickest). (10/13)

When testing is common, those with a lower probability of testing positive also receive tests (e.g., asymptomatic people). Thus there are diminishing returns for detecting positive cases as the # of tests goes up, assuming the underlying outbreak remains the same size. (11/13)

But the underlying outbreak is not remaining the same size. It’s growing. And it ain’t just because of the testing. (12/13)

To end with some international context: we’re not the only country seeing a resurgence, and ours is far from the most dramatic. Check out the recent epidemic curves of Spain, France, and the UK. (13/13)

#COVID19 #Canada #OpenData available: https://github.com/ishaberry/Covid19Canada

More maps and trends: https://art-bd.shinyapps.io/covid19canada/ 

Courtesy of the COVID-19 Open Data Working Group: https://opencovid.ca/ 

Re: Quebec testing numbers, see this thread. The flat trend since mid-July still stands whether you're looking at people tested or screening tests performed. https://twitter.com/perreaux/status/1303701354221899776