NEW: Tue 21 April update of coronavirus trajectories

Daily deaths:
• Still too early to say if US has peaked
• Beginning to look like UK has
• But descents look much slower than ascents
• Successes in dark blue: Australia, Norway, Austria

Live charts
Why do I say UK daily deaths may have peaked?

Here’s week-on-week change in daily deaths.

This gets rid of weekly reporting patterns and asks, are more people dying than at same point last week?

In UK, blue bars mean we’re now seeing *fewer* deaths than same day last week.
Now back to cumulative deaths:
• US death is highest worldwide and still rising fast 📈
• UK curve still matching Italy’s
• Australia still looks promising

All charts:
Daily new cases:
• Feels increasingly safe to say daily confirmed infections in US have peaked, though descent from peak is slow
• New cases falling in four countries that acted early: New Zealand, Australia, Norway, Austria (also 👋 Greece)

All charts:
Cases in cumulative form:
• US curve beginning to taper?
• Turkey still battling a severe outbreak
• Japan has passed Korea’s total, Singapore has passed Japan’s curve: both show the danger of thinking a country has dealt with covid

All charts:
Subnational region daily deaths:
• NY daily confirmed covid deaths now descending (we’re excluding nursing homes for consistency)
• Daily London deaths also appear to have peaked
• Most Western cities/regions now in plateau or decline phase

All charts:
Subnational death tolls cumulatively:
• NY curve tapering, but has passed Lombardy for world’s highest subnational death toll

All charts:
Covid outbreaks better understood at regional level than national; here are >100 regions:
• Rio de Janeiro 📈
• 31 US states now shown
• Stockholm yet to peak?
• Sicily, Sardinia, Balearics, Canaries all low curves: do islands fare better?

All charts:
Small multiples for daily new deaths in countries:
• Norway locked down while Sweden didn’t; Norway’s daily death toll rising much more slowly than Sweden’s
• Australia faring well
• In Europe: Austria, Denmark, Greece, Norway faring well

All charts:
Small multiples for daily cases in 77 countries:
• 10 African countries
• Early action in Australia & New Zealand may have turned corner 🇦🇺🇳🇿📉
• Austria & Norway locked down early; new cases falling
• Watch as countries relax lockdowns 👀

All charts:
I talk a lot about whether or not peaks have been reached.

A good metric for this is hospitalisations:
• More reliable than confirmed cases (not influenced by testing regimes)
• Shorter lag than deaths

So here are some charts showing hospitalisations in various countries:
First, Italy:

Colour = phase of outbreak
• Red: more new hospitalisations every day than day before
• Orange: total hospitalisations rising, but rate of increase slowing
• Blue: fewer people in hospital than before

Almost all Italian regions now in "reduction" phase :-)
Next, Spain:
• Madrid now firmly in "reduction" phase, Catalonia fighting to stay in reduction
• (Other Spanish regions not shown due to problems with their hospitalisation data)
• All regions beginning to see total occupancy fall, including former epicentres Ile de France & Grand Est
• Hospitalisation data patchy from state to state
• NY & Louisiana both in the "reduction" phase, hospital bed occupancy dropping 👍
• Rate of acceleration falling in NJ & Colorado, hopefully soon net reduction
• Connecticut also on the path towards falling occupancy
Great Britain:
• Hospitals in almost every region now have fewer covid patients than same time last week 👍
• Suggests UK is at or near peak for new confirmed infections, though UK testing still lagging, and care homes absent from this view
• Remember Sweden has not locked down like most places. Social gatherings still common.
• But data show steady deceleration in new hospitalisations
• Daily hospital beds occupied now unchanged vs last week, and on course for net week-on-week reduction very soon
Wrapping up: our focus has shifted to tracking excess all-cause mortality (numbers of people dying for any reason at all) as we think this avoids pitfalls of different countries’ covid-death reporting methodologies.

So my call-out tonight is:
Final bits:

Here’s a video where I explain why we’re using log scales, showing absolute numbers instead of per capita, and much more:
And a chart showing why we're using absolute numbers rather than population-adjusted rates:
Please email [email protected] with feedback, requests & subnational data.

All of these are invaluable, and we incorporate your suggestions and data every day.

We’ll keep getting back to as many people as possible.

Have a good night, folks :-)
Shout-out to @caprosser, whose suggestion I’ve incorporated into the daily deaths y-axes

y-axis now log(n+1) instead of log(n) so we can include zero, meaning China etc don’t drop off the bottom

Doesn’t change anything in terms of how to read them, just makes them cleaner 🤓😀
You can follow @jburnmurdoch.
Tip: mention @twtextapp on a Twitter thread with the keyword “unroll” to get a link to it.

Latest Threads Unrolled: