Coronavirus Death Rates Over Time

Adam Larson aka Caustic Logic
June 7, 2020
edits 6-10
updates July 3, August 2

Background on this project
In my prior graphics exploring COVID19 death rates, I included some different tolls over time in a range of nations. Starting with a bar graph by UK Column back in late April meant to show lockdown countries had bad or no effect on death rates, I've made a couple if rebuttal graphics, starting in this blog post, and working up to this in a recent Twitter post, using the same proportional deaths-per-million UKC started with:

Part of the point there was to show China, and Iran, and how the supposed lockdown plot so many folks are alarmed about is probably not a political conspiracy - rather, it appears to be a real issue faced by all kinds of government, nearly all of whom rationally chose the effective response. Is that answer boring? Sorry.

But the variance over time is also something that starting image didn't consider, and didn't have as much time TO consider as I have now. The graph above compares levels from May 13 and May 21. That was interesting and seemed worth expanding on. So I started right off with a more systematic look at death rates over time, split into six spans of about 2 weeks each (aside from the first, covering whenever up to March 22) running up to the present - or a couple days ago now, June 4. And along the way I included far more nations, for a broader-yet reading over space as well as time.

As the morbid "massacre marketing" guy, let's say, I'm looking at deaths here. It might be more useful in general to look at infection rates, which also entails problems like overwhelmed hospitals (besides a lot of people who didn't even feel it). SARS2/COVID19 is not known for its extreme lethality - it seems to harm and kill far more readily than the common flu, but perhaps 99% of those infected so far survived. (It seems debatable to me, the tallies we have are incomplete and far from final, and it's not clear how many unconfirmed cases exist, and just how few of those (presumably very few) add to the fatalities rather than survivals.)

What really sets this virus apart is its rapid spread and ability to cause its effects on a mass scale in short order. But the scale is such that, even with transmission slowed by a global lockdown, and even though it's not super-deadly and that everyone tries to shield the elderly - it has already killed at least the 400,000 people just now confirmed. Nearly all of that has happened in the last three months despite lockdown, and that number is now growing a bit faster as the virus keeps being contagious but people can't keep sitting still. It will certainly be passing at least one million mark it's almost halfway to, and maybe several of them, depending. That, and the nature and speed of it is something I'll be watching.

Deaths from (or "with") the virus of course rely on its presence, so they tend to rise with its general spread. As I'll show, they rise less or even fall where the spread is generally suppressed. But fatality is variable even more so on a limited spread (one way or another) to those who would die. So lower rates will reflect a mix of:
a) general spread reduction and
b) specific reduction to the most vulnerable, and
c) other factors, including quality of medical care, etc.)

So for assessing the value of lockdown and other mitigation efforts, death toll is a bit limited. It reflects other factors, and excludes other benefits like hospital case load, besides the substantial costs of economic and social lockdown. But death seems to be the part that grabs the attention of mortals, that drives the most worry, and is perhaps the virus' most debated and misunderstood feature.

Or put another way, the relative lack of deaths so far is widely misinterpreted. I hope to provide some clarity on that.

The series given a fancy name CORONAVIRUS DEATH RATES OVER TIME (as of June 4, 2020)
My two sources tracking and compiling WHO data:
https://www.bing.com/covid

https://coronavirus.jhu.edu/map.html

Ongoing Lombardy updates: https://github.com/pcm-dpc/COVID-19/tree/master/schede-riepilogative/regioni

Source on Birmingham: https://www.thesun.co.uk/news/uknews/11753789/birmingham-coronavirus-worst-hit/

Populations used: quick internet search for each, looking fo the latest (and usually biggest) number, finding Worldometer's 2020 estimates seemed closer than even the latest actual counts, I used these for nations. For states and other places, Wikipedia, whatever popped up and seemed good. I take the number of millions, not the number, and round it off variously, sometimes to one or two decimal points, maybe three for tiny numbers. Ex: China is 1,438, UK is 67.5, Iran 84, Turkey 83.7, Mexico City is 8.8, and Iceland is 0.36.

Time spans used: basically bi-monthly, starting from some dates I already covered, running 14-16 days, then simple 14-day spans for the last two (and any further spans I add), so they compare directly, while the earlier ones are a bit different. It's used enough it's worth putting the key right here.

Note that deaths follow infection by a few days to about two weeks or longer (?? details...), so the end of one span basically reflects realities at the end of the previous one.

Scale used: moderately magnified scale for smallest numbers, moderate reduction of scale from 800 up - not the best set-up, but I'm not re-doing it all. The scale is pretty steady in the middle, light gray, darker for magnified bottom, lighter for rolling-away upper end. So understand all bars have fatter bottoms we can read a bit better, and the tallest ones have their tops a bit truncated. The middle still has more space than it needs, I now see.

At right is a basic illustration of exponential growth, as it would look on my charts (including the lower and upper distortion, somewhat). Let's say the shaded spans are varying but semi-uniform, unspecified periods of doubling, as the pixels double in each span (in the middle anyway). In a calendar-based system, the growth will vary with local dynamics, but with stupid but crowded animals, each period should grow lager and larger at a steady rate until around the threshold of herd immunity is reached - which, as I'll explain, we are nowhere near. Only then would it slow and finally stop. (a full bar showing that might have 6 exponential spans as at right, then a 7th span just bigger than 6, a smaller 8th and 9th, a tiny 10th one, and then nothing.)

Even herd animals will employ "social distancing" along the way to group immunity (look it up), and with humans, we can massively change the course with science and communication, slow the spread greatly and even stop it completely - not that that's always easy. And you might notice in the plates below, no country has results like the bar at right (though some cities come close). Many start out that way, with the second span far deadlier than the fist, with one span sometimes doubling the last. But they have a tendency to then stabilize and even shrink over later periods, sometimes even getting cases and deaths so low they have zero for an entire span, or maybe forever. That is, wherever effective measures are taken.

The InfoGraphics here:
Plate A: the deadliest fronts, so far

This looks at the early hotspots I know of, aside from the Diamond Princess cruise ship (China, Iran, Italy), and the places hit even harder, but slower:  USA (states of NY and WA as examples of internal variance), lockdown Europe's highest death rates (Ireland, France, Spain, tiny Andorra, Belgium), the herd immunity crowd (Sweden, Brazil, Belarus), and between the last two classes, those in Europe now on lockdown who started with HI (NL, UK)

Plate B: How most of the world has fared so far

Starting with the three famous for doing poorly (China, Iran, Italy), they anchor the majority of countries and regions that have done no worse, with a vast majority of the world's people and lower death tolls than seen in plate A. The emphasis here is on showing higher rates as I noted them, but also a sampling of sizes and patterns. Seven harder-hit localities are also shown along with 47 nations under some kind of "lockdown" stay-at-home order (as far as I know). 11 others who've managed varyingly without such a policy, each for their reasons, are also shown: "no lockdown" (NLD), managed with early containment (EC), some closures and local lockdown (PLD), and no evident embrace of "herd immunity" (HI), as with those on plate A.

So looking across these, we can see where it spread more early on (lighter pink shades) vs. in later phases (red-brown). You have to zoom in to see some the spans, and some aren't there - occasionally a place has zero deaths (or it rounds down to it), so it doesn't show. I show fractions down to about 0.05, rounded up to 0.1, shown as a small 1/8 bar 1px wide (ideally - see Egypt, others).  Especially in bigger countries, tiny early returns are sometimes done in partial lines or little blocks - that hit me halfway through, as others have full lines I exaggerated a bit to even be visible, or were unfairly rounded to zero.

Plate C: Spread over time in the United States of America, so far

Lockdown orders: Not all are noted - in fact jut a few; unless otherwise specified, all states appear to have issued some such orders. Each order that is shown here is marked roughly in its time span, as dated. (the marker has nothing to do with the background scale, except to suggest why subsequent periods experienced less fatalities.) Any relaxation of said orders is even less-noted - just two cases.
NLD = no lockdown, as I've read - president Trump's 'Coronavirus guidelines for America' applied to everyone, but five states declined to add their own specific orders: Iowa, North Dakota, South Dakota, Arkansas, Nebraska (4 shown here - I picked the worst case between the Dakotas.)

As discussed a bit below, localities can be hit harder than states or nations that average together areas of many kinds. City data is limited on my end end and may be extreme, but in the US, I looked at some counties. In the Bing tracker, these don't have date breakdowns, just total fatalities. At least the top 4 here, all in New York and New Jersey, had death tolls well above Lombardy's rate of 1600/million:  two were just above and two just below the mark of 2,000/m. Or something like 1 out of every 500 people alive in these 4 counties ... now isn't.

This was done a few days ahead of other graphics on May 30. Some updates one week later, just on 2 of those contests:
* Queens NY (2172.4) vs. Essex NJ (2127.5)
* Cass ND (329.7) vs. King WA (258.3). King added about 2 points. Cass added about 90. In a week.
* The whole bunch of smaller counties at or near zero will be at least a bit smaller. All US numbers are growing a bit faster than most numbers in the world.

Discussion on what this all shows

1) lockdown saves lives - not singlehandedly, but ... This was a highly contagious pathogen with vigorous growth everywhere it appeared, until people stopped moving around so much in some areas. The virus does not teleport through walls, on 5G waves or otherwise. But the more important change regarding death toll is keeping it from spreading specifically to certain people who would die from it (hereafter "those who'd die"). We're learning it's hard to say just who all these are, but they benefit from a slowed general spread (mostly attributable to lockdown) and from their own protective measures.

Still, deaths continue at a moderate average rate that tends to be smaller where "lockdown" measures are smart and widely adhered to. Take China: they've effectively eliminated the virus for 1/6 of the world's people, and they did it months ago. Unless you think coronavirus would naturally claim just TWO lives in China in the last 1.5 months ... a lot of lives have been saved just there. At least a few thousand have been saved in the other 5/6 of the world, maybe even hundreds of thousands so far.  In the end we might prevent tens of millions of deaths (if the end is far enough out).

Even that, in a world of more than 7,000 million people, is arguably not such a big deal. The costs of all this lockdown are considerable, and the whole trade-off remains debatable. But let's insist the debate take honest consideration of the basic fact that lockdown is helping greatly to mitigate a genuine disaster. The costs may not be worth it, but they're not for nothing. Sorry, but the math can't be that easy.

2) Those who've really dealt with the virus deal with it for real: See China as the example most extremely manifested - a hard blow in one area triggers a sort of immune response that chokes the virus into virtual annihilation. I re-examine that here with a new timescale for its details, new colors, and a 9-span scale where the last didn't need a color, and the two before needed to be exaggerated to even appear.

Softer but effective containment can be seen with a similar pattern that hits zero in Iceland, New Zealand, a few others like Hawai'i - all with just 10-22 dead on the way, and a lot of ocean around them. Bigger problems squeezed to more than zero and that persist include Iran and Italy (plate B, the other gold boxes), and in Denmark with the biggest early death toll here outside a gold box (note span 1 goes back to whenever, and is huge - they had 125 dead by the end of February.) Good collapsing of expected growth can also be seen in Switzerland, Austria, Germany, Washington State, most of those in the second row. The Dominican Republic suppressed it well after a late start and bad rate in span 3. Greece shows clearly the start of exponential growth in spans 1 and 2, almost perfectly inverted into exponential decrease over the later spans, and all nice and stubby (low deaths per populace).

3) why "no lockdown" doesn't seem to fail as badly as you'd think: A few podunk places with sparse populations and limited contact with the outside world think they can get by without lockdown or early containment, and for some reason. So it's no surprise they tend to do okay without, so far: North Dakota has 86.8/m dead (well below national average), with Pres. Trump's lockdown advice, but none added by the state. Belarus with 26.9/m (Plate A, right side) seems grossly undercounted compared to their exploding cases, and to Pres. Lukashenko's promise Coronavirus would kill no one there (deaths were happening, but he declared the first 23 were racked up to any comorbitity that could be found, and suggested the rest would or should be). Here's a current list of middling-top countries by cases, with Belarus now ahead of the Netherlands, but their strangely low death rate compared (note that Qatar is apparently undercounting even worse, reporting less than 1/10 of 1% - unless I'm missing something, they're engaged in absurd falsification). Global average is currently 5.797%. So undercounting might help some of these numbers stay low.

Some might count Brazil, having not passed the mark of 200/m dead (5.33% rate). Local lockdowns have kept it from being worse, but that advantage is being squandered by the Bolsonaro government on its third Health Minister of the crisis. Cases are rising sharply; deaths in Rio De Janeiro are now well past Belgium's average, headed for 1,000 and then Lombardy levels quite soon (plate A).

No-lockdown policies that work tend to rely on successful early containment - not on letting it rip towards herd immunity - and to play out in island nations, where infected people can't just swim there unchecked (see lower box on plate B). If they can get it contained and watch arrivals closely, they can keep it contained, and many do. Taiwan did well, stays well at less than one death per million - which was just seven dead, and it still hasn't gone up, and new cases remain at or near zero. But they did prepare a plan B centered on lockdown, not on an "oh well" herd immunity some now promote. Japan did less well, has slowly climbing deaths, just past 900 now (that's just over 7 dead so far for every one million citizens). NLD states of the USA, connected as they are to other states, do far worse than island nations, ranging between 75-200/m so far, and rising pretty on-par with the rest. Iowa, pop. 3.16 million, was at 183.2/m (579 dead) for plate C. Three days later it's 189.2 (598 dead).

Malawi has (or had) lockdown orders under Supreme Court review. They also seem to have an implausible frozen death toll of 4 that's held for a couple of months. It's possible they have it that in-check, with no lockdown policy. But I doubt that. I didn't look into the story there any further, nor into Indonesia, or fuller details on the others.

But I noted no-lockdown Nicaragua had rising numbers, then a pause in updates; On May 25, a paltry 25 confirmed cases suddenly jumped to 3,425 and deaths leapt (w/post-mortem confirmation, I presume) from 8 to 58, then no update for a week at the Bing counter. Johns Hopkins tally froze at 35, not 58. Both have by now been updated and settled on an revision in the middle at 46 dead, 1,446 cases. None have been added. A few days later it was revised to 46 dead, 1,118 cases. Can they wiggle it back to zero? (note June 10: No - Nicaragua's numbers keep rising slightly from that fudgy re-start point - 55 dead, 1464 cases.)

Optional: more on incomplete counts
Zambia (lockdown as far as I know) has a similar issue; the counts rises on May 14 from a few cases and a few dead to 654 cases and to 0, then 7, then 42 dead, on the graph. Then no graph updates for three weeks. But the number of cases listed went up to 1,057, and the deaths listed reverted to 7, agreed to on the Johns Hopkins list. Now it's 1,089 on both lists, and both are stuck at saying 7 dead, the virus seemingly contained, that 42 then silence being just a typo? (note June 10: just now something was jiggled and they report 10 deaths, not 7, or 42, or anything added since, but cases up to 1,200.)

Other similar cases in Africa:
Tanazania: jump in data 4-18, slow rise to 5-8, no news
Benin: jump in data 5-10, last news 5-14
Eritrea: last news 4-18.
Mayotte: last news 4-21.

Zimbabwe, Uganda, Togo, all others I checked remain in touch, even with recent leaps and similar, small death tolls so far. Sudan's graph stopped, but numbers came in showing moderate rises (359 dead now). But a significant part of the deaths in Africa these last weeks has gone unreported. And I don't have a screen capture for this one, but I saw South Africa's tally jump by some 3,000 deaths, then quickly revert to the 300+ they had, climbing only slowly from there, then faster (952 as of writing - 16.4/million). Are they hiding the bulk of a much larger death toll? If so, the only evidence I've seen suggests they admitted to it briefly, but maybe retracted for fear the bad news would make people give up on lockdown (?). And maybe that was just a data entry error.

Aside from Nicaragua, I didn't notice any such cuts in news outside Africa. But North Korea has never appeared as reporting at all. Neither has Turkmenistan, that I've seen. Do they both really have nothing to report?

4) Insights on "herd immunity": This is a concept misunderstood by many, perhaps by me. Some think collective or "herd immunity" happens at a random point in the spread, usually the point that helps them argue lockdown was unneeded; the community at large, or the virus in that area, changes so that it spreads or at least kills less, just about the same time it reaches crisis level that lets the government impose restrictions. They're certain the restrictions have nothing to do with the mysterious natural change that occurs just then.

But as I gather, "herd immunity" happens not by magic and not totally, but when the virus finds far more immune people than susceptible ones. A few infections and deaths may still occur, but outbreaks mellow and fizzle out, and the virus might fall through the cracks. The problem is, this happens after a virus has mainly run its natural course - barring a vaccine we don't have yet, a solid majority become immune only after they've been infected - and survived. Most will live, but only after spreading it to others, some of whom won't, and they might add to the hospital burden along the way. I don't believe it happens by any lucky mutation or telepathic immunity transfer prior to this, not even to make some anti-lockdown argument; viruses have no political point to make - they just replicate.

The usual threshold for herd immunity, as I gather, is something like 80%, ranging to maybe below 70% or up to 90% of the population who might be exposed (basically everyone). The extremely contagious nature of this virus suggests the higher end will apply - 90% may need to be infected before we can stop worrying about people being infected. This is an absurd solution for such a dangerous pandemic - invite almost the entire problem, then the last scraps of it will go away? Not the right answer.

What we see is nowhere near the HI threshold anywhere on Earth. San Marino, a micostate within Italy, population ~34,000 (0.034m), might be close: they had a startling 1,236 confirmed infected - about 1/3 of them, surely others unconfirmed - possible herd immunity level, o likely to see a mild touch of it. 42 died (a very high per-capita rate of 1,236/m) but a while back - the last two were on 4-26 then a straggler on 5-23. They still get 0-4 new cases a day in such a tiny populace, so full herd immunity seems to elude them.

It may be halfway there in Stockholm, where they bragged of 30% infection some time ago, and it's likely true by now, perhaps past 45%. Herd Immunity is farther off in Sweden at large - maybe 25% there. It's maybe like 10-20% along in other hard-hit places, and yet to really begin or resume the creep in other places, like India and China, where 1/3 of the world's people live (it may be starting in India, but still not resuming in China). Most of Africa, Southeast Asia, scattered other nations have kept it contained from low to no levels, and lots of areas within countries already hit have all of their progress towards herd immunity ahead of them.

So to be clear, there is no basis on earth for infection or death rates to fall due to natural collective or "herd" immunity. Falling rates at this time can only result from artificially suppressed transmission. (as far as I know - again, I'm no expert)

Now consider the damage caused so far is from incurring, on average, just a small portion of the infections needed to achieve that goal. 400,000 dead might be just 5-10% of what we'd need for global herd immunity (which would be 4-8 million dead). It might be lower, or higher - I don't really know. Do you?

And consider early herd immunity proponents Netherlands and United Kingdom (plate A). Both started with a let-it-rip policy along with Sweden, but quickly changed course to more of a lockdown plan. See how Netherlands had the worse initial lesson (higher per-capita death toll in the early spans), and then apparently learned it better, achieving far better suppression, with the UK's terrible death rate going on double theirs. This further underlines point 2.

Even ahead-of-the-curve Sweden admits some of its regret; their chief epidemiologist Anders Tegnell publicly defends his overall herd immunity plan, but admits it's imperfect, having killed more than they'd like, just so far, and that they would do it differently if given another chance: "If we were to encounter the same disease, given exactly what we know about it today, I think we would reach a conclusion to act somewhere in between what Sweden did and what the rest of the world did," The nation's leading epidemiologist just didn't realize how contagious it was, or did but chose to barrel ahead and see a lot of costly old folks die. And others shared his total ignorance; Tegnell adds that they understood the elderly were at high risk of death, but "we didn't know that the disease would enter (care homes) so easily and for the spread to be so big."

Well, knowing or carefully guessing such things was their job. They know it now, based on a lot of trial and fatal error. Whether it could have been guessed seems like a philosophical question; Tegnell now wonders "if there had been any way to prevent that," like what the rest of the world did? Could they have noted their early partners in the HI camp - the UK and Netherlands - saw cause to change course, and follow suit? Yes, they could have and should have, but chose to keep the faith, which has caused far more deaths than Sweden should have had; see charts, plate A: Stockholm, had 884/m killed by June 4, higher than Belgium's rate, and rising to be far above - heading for Lombardy rates, but a bit slower. Faster now - 3 days on, Stockholm has 2,137 dead - for its 2.38 million people, that's 897.5/m.
https://www.msn.com/en-us/news/world/sweden-coronavirus-expert-defends-strategy-but-admits-failing-to-protect-the-elderly/ar-BB14YG1v?ocid=msedgntp
See also: https://consortiumnews.com/2020/05/25/covid-19-how-trustworthy-and-humane-is-swedens-pandemic-strategy/

And besides fatalities, those who live sometime pay a heavy price - cardiac and circulatory damage has been widely documented in survivors, already proving fatal in at least a few cases. However much of that has happened, we'll need maybe 20 times more of it before we can enjoy the fruits of herd immunity. And so, I still suggest every person and every community that has avoided the virus so far to try and keep it that way, forever if possible.

5) Locality variances: Hubei vs. China, Lombardy vs. Italy, Queens vs. NY vs. USA, Stockholm vs, Sweden, Rio De Janeiro vs. Brazil - the former wind up with more far more infections and deaths by far than their national averages. This only makes sense, considering how viruses spread.

Diamond Princess - extreme example:
3,711 passengers and crew - 712 cases - 13 deaths = 19% infected, 1.8% of infected died (note: middling-older people take cruises the most - not the young, nor they very old). The ship had .00371 million people. 13 dead makes for 3,504/million, far above any land-based toll. To hit herd immunity, it would have to hit probably 4x as many or more (prob. above 80%), killing about 4x as many. But I believe the captain ordered some kind of cabin lockdown to limit spread.

Many lockdown critics would support that call - after all, that was on a boat, not a nation spanning maybe hundreds of miles. Indeed, geography adds cushioning, but only so much. In the end, it may be entirely eroded by time - those 13 were infected during the span of a cruise turned extended quaratine - however long, it was measured in days. The rates it's far higher than ae current, and have taken a couple of months to build up. And they'll have x months to keep going too.

The highest national average so far might be San Marino - a tiny enclave in Italy - that had deaths even higher than Lombardy's, but not shown on plate A (is it a nation? A region? Do I want to re-scale this so it fits?). Population ~34,000 (0.034m), just 42 died to hit that level. I don't have the stats for New Yok City itself, but the buroughs as counties have death tolls a bit past 2,000/m in a few cases, and New Jersey has about the same - so more than 1 in 500 has died. The USA averaging great cities and great plains is at 330 (New Jersey: 1335/m. New York: 1235/m, California just over 100/m Texas: just 63.1/m, and half the states around that or lower, so far). Italy and UK are around 550-600/m, with Italy's Lombardy region just over ~1,600 - nearly half what that ship saw. Birmingham in the UK was at 896/m last I heard: 1,076 dead in "the West Midlands city" (The Sun) where the population seems to be ~1.2 million. Rio De Janiero, pop. 6.5m, had 6,010 dead on June 4 (924.6/m) and 3 days later it's 1,021/m (a further 629 dead - 210/day in one city). Belgium is a bit over 828 and barely rising at all, while the smarter or luckier half of Europe is mainly between 60 and 100/m or even lower and stable, and it ranges down to a suppressed 3.22 in China, in the 3-10-50 range over other sizeable nations.

In that light, a lot of captains of state order people to stay in their cabins as possible. Is that really surprising?

Okay, I'm done for now.

Update, July 3:
The ensuing weeks saw increased growth, mainly in newly-opened battlefronts all over the planet. Back on June 18, I tallied the next -week span and issued a single plate focused on bad growth rates. A first version was hastily posted on Twitter. Then I adjusted some numbers that hadn't been finalized (several nations called early, US states called too high), the global total, and the right date, and made this improved version. I meant to post it here then, but never did.

Now another two weeks has passed ... . The global death rate is accelerating a bit now, but not by much - it holds pretty steady on average, including the many containment success stories like Germany, Austria, Switzerland and Denmark here (done on the side for someone who didn't want to learn after all (blocked and likely muted me) - that's common, BTW - covidiocy requires maintenance). But the trend in many areas shown above continued, in some of those places and others I noticed later. As the global death toll passed 500,000, cases were soaring worldwide, and while the deaths that caused is lower than it has been (and/or pending), a lot more people died in the next 14 days, with a growing number waiting to join them. Here's the (final?) plate for July 2, at 521,298 confirmed deaths (Already as of writing, it's at 525,491).

Latin America has done poorly. Death rates over 250/m in Peru, Chile, and Brazil, with Mexico coming near that. The worst growth there is in El Salvador, Guatemala, and Bolivia, all with moderate prior tolls roughly doubled in the last 14 days. Some localities of course do worse; Santiago, Chile, just reported some 3,000 new deaths from earlier, bringing their rate close to that or Rio DeJaneiro. Brazil (along with Russia) just at the end stopped providing regional death tolls, or a plotting of them over time (or at least they disappeared from the Bing tracker I use.)

India has increased growth, now close to 13/m dead (17,834 and rising faster). That's not massive on average, and slower now in Maharashta/Mumbai, but recently increased cases and death in Delhi map as shown, now at 96/million dead, (currently 7-2 shows 2,803 - a repeat of the prior tally. 0 deaths being unlikely, I took the next rise to 2923 and used that). Pakistan is a bit worse off at about 20/m dead, Bangladesh a bit less so as 11, and all three had more die in the last 2 weeks than in the previous 2 weeks.

In the Middle East, Iraq has done worst among those I follow, more than doubling their previous death toll in these 2 weeks. Saudi Arabia has about the same 50/m dead as Iraq. Egypt is not much better off, and Saudi Arabia is worsening , Afghanistan and Yemen saw continuance of a death expansion. Iran had its improved rates worsen for a second span. E. Europe and the Caucasus has bad growth in Azerbaijan, Armenia. N. Macedonia, Albania, and Moldova fared poorly. Russia had enough new deaths to sit now at 65/m, just a hair under the global average of 67.18. Kazakhstan's small populace wasn't spread out enough to prevent deaths doubling to 188, and nearby Kyrgyzstan fared worse, suddenly shooting to a death toll on par with India's. Syria and Venezuela saw limited fatality situations fall apart further. Sub-Saharan Africa is seeing inceased fatalities all over, especially in Cabo Verde, Equatorial Guinea, C.A.R., S. Africa, Eswatini, Angola, Malawi, Ghana, Kenya, and mighty Ethiopia. Nigeria saw moderate growth, enough to include here. Some who seemed to not be reporting now appear to be reporting sporadically.

The UK, Sweden, Canada, others in "the West" did well enough at keeping deaths low they didn't merit inclusion here. Especially places like Italy, Spain, Belgium, New York, and New Jersey that were hit hard early on, they continue to keep transmission and deaths low, whatever the hassle it continues to take. The USA average is not much worse than the UK, but includes enough bulk and potential, and some bad sub-patterns it's included. At just over 130,000 dead, we're at close to 375/m average now, close to 6x the global average. Alabama has a bad trend that's likely to get worse - some morons there take it on themselves to host coronavirus parties, where people compete to contact it and win a cash prize. ("several" such parties known of just in Tuscaloosa). That's bound to play into the death toll as it passes the 1,000 mark soon. A few states included just for not holding it as steady as others - there might be worse patterns I missed, but generally the US is doing a lot better than the worst places otherwise shown here.

And still, China ... the eternal 4,634 dead listed at the Bing counter is challenged by the tracker run  by Johns Hopkins University 4,641 - and I've seen it rise recently when the other didn't. It seems they've missed some seven deaths, adding to the two they list, that would be nine people dead, out of 1,438,000 people, over the last 2.5 months. They still sit at 3.227 dead for every one million people, currently 1/20 of the global average and falling.

Update August 2: As of July 30, many nations have revised or should revise their death tolls - a look at the scale of this suggests at least 25% more people have died than is currently admitted, and it could be nearly twice, or as high as 1.3 million dead already, despite the extreme measures. See here for the details.