Coronavirus SARS-Cov2 (COVID-19)

[jayakris]

You used tests per million as the basis of the analysis, but that is not a correct assumption at all, in my opinion.

So what IS the correct way to compare test penetration across countries? Cuz you were comparing India's testing regime to those of advanced countries. How were you doing that? (Just a hint would suffice, no need for a lecture )

Haha... Actually no real fully-justifiable reason, even if I sound like I know it and you don't when I write lecture pieces!

It was mostly just a judgmental call based on the fact that our TPR is good and that there are some multiple study reports of the statewide death numbers possibly being in the 4-to-5-times range. It is also from what I see in terms of death numbers in other countries and the test rates, and trying to see some pattern. TPR does correlate with deaths.

I think you were basically judging the cases we could have found, based on tests/million, and actually correcting what seemed like 4 or 5 times down to 3 because you also felt that the stat is a little less reliable. So I know you were also adjusting for the issues with tests/million stat that bother me.

The only difference in our thinking was that you were probably not trying to see what it would imply in terms of death numbers. Rightly so, as we have even more unreliable data on it (because of all the crazy things that happen in death-registration and reporting). But I felt that even at THRICE the cases, our true death numbers will have to go up to some 9 times what we have now. That is because I assume about 5% CFR for the unknown out-of-hospital deaths in village and poor urban populations, and about 1.75 to 2.00 for hospital CFR.

Ultimately it comes down to my putting a bit more trust in the general reports of some statewide death totals being 4 to 5 times. To match that, it will need to go down to TWICE in case numbers and conservatively 6 times in deaths.

But this is a case where multipliers of 2 versus 3 makes a big difference!! You may be right too, in the end. Who knows!!

[Atithee]

I’m just wondering how is that Maharashtra adds previous deaths EVERY day. Shouldn’t the reconciliation be complete for all last mismatches in one go? Adding it everyday means that there is some systemic issue, which is established by now, and there is a recon done every day for the prior day. What am I missing here? Is it information is coming district by district to correct past reports?

[jai_in_canada]

I was only projecting what the reported positive cases might have been if India had had even the test penetration of the average of the 10 worst affected countries - let alone the most tested countries. I figured that using the average is more reasonable than using the most tested countries as a basis.

I figured that using tests/million was reasonable as a comparison factor, instead of using the absolute numbers which don't factor in population. I don't know that there is any other data to make even a quasi objective comparison.

So I tested your hypothesis that India's reported positive cases was 50% of what it would have been if India had tested to the same extent as a European country or US. First, I was surprised to find a variation in the testing rates in the advanced countries. Second, I was surprised that India's testing was only about 25% of the AVERAGE of the 10 worst affected countries, and less than 10% of the UK. So I figured India is missing a lot more than 50% of the cases (which is what you had estimated). I figured it is missing more like 75% -- that might have been found with testing that matched the average of the 10 worst affected countries.

With regards to CFR, I was not even trying to estimate the real death count, for reasons that I mentioned. Recoveries and fatalities numbers almost everywhere are bogus anyway - some more or less than others. Almost everywhere, Case Fatality Rates underestimate the true Incident Fatality Rates.

Anyway, it's all hypothetical. Still it is interesting to try and estimate what has actually gone on in terms of morbidity and mortality. No question that India's reported cases and deaths are underestimates. To what extent? Jay & Jai have a bit of a difference of opinion.

P.s. Just occurred to me that we might be able to sue J&J for copyright infringement for using the Jay & Jai vaccine brand?

[depleter]

I’m just wondering how is that Maharashtra adds previous deaths EVERY day. Shouldn’t the reconciliation be complete for all last mismatches in one go? Adding it everyday means that there is some systemic issue, which is established by now, and there is a recon done every day for the prior day. What am I missing here? Is it information is coming district by district to correct past reports?

The thing I am thinking of is may be all these reconciliations are from post covid complications. may be...it's just a guess of mine.

[Atithee]

@depleter—every day? Isn’t the definition fixed now? And, why Maharashtra alone in such a high number?

[jayakris]

Suresh guessed it would be 60-61K today, and Depelter guessed it would be under 60K... I was unsure who would be right, and didn't know even late at night, as it depended on whether Ladakh would show more than 42 cases or not!

Well, Suresh won!! ... 50 cases in Ladakh, so we reached 60008 for Monday ... But, man, you guys are becoming very good at this stuff.

So +60.0K from about 1847K tests at 3.25% positivity. The seven-day average fell by 3.7K to 78.5K... It was 1341 deaths (excluding the 1392 reconciliation in MAH)

-----------------
ICMR update: 381,375,984 total tests... Monday tests: 1,751,358

[suresh]

Ha ha! This is a pyrrhic victory. It looks like fall in the seven day average will be around 3-4K next week. Let us hope that it goes down to 60K by next Monday.

[jayakris]

And we did 4.2M doses (3.7M new people) on Monday, which is pretty much the highest totals we have done so far (matching a Friday from 10 weeks ago). More importantly, it is a 20% rise from last Monday, and so the rising trend continues. We should be at 5M per day soon, and should quite easily reach 300M people with at least one dose in about three weeks time by early July.

[depleter]

@depleter—every day? Isn’t the definition fixed now? And, why Maharashtra alone in such a high number?

(This is purely my guess). If they are adding up deaths due to post covid complications they have to check up with all the people who got covid and it will be a daily and long process. It's a more plausible answer I think. And definition was never fixed. The definition of a covid death varies from state to state and even doctor to doctor. And why only Maharashtra? Because no one else counts those deaths. Deaths due to post covid complications is a sizable number and states just don't add them with covid deaths. Forget about post covid complications, some states don't even add deaths which have covid to covid deaths because they say the person died due to some other complications and not due to covid. I think this is how Telangana operates actually. My Dad has some relatives there and he heard these stories from them.

[suresh]

Estimates of excess deaths in four states:

[depleter]

And we did 4.2M doses (3.7M new people) on Monday, which is pretty much the highest totals we have done so far (matching a Friday from 10 weeks ago). More importantly, it is a 20% rise from last Monday, and so the rising trend continues. We should be at 5M per day soon, and should quite easily reach 300M people with at least one dose in about three weeks time by early July.

Glad that the doses are high but the doses are not being replenished fast enough as the count of doses with the states is falling rapidly. The center said the states would be getting 120M doses this month which would 4M/day. But even if our avg daily inoculation count is less than 4M/day, the doses with the states is decreasing. May be during the last 2 weeks of this month the states will be provided with higher avg of doses. And Niti Ayog said on June 1st that in July we will get a supply of 200M doses. So, the supply will continue to go up continuously. This actually makes me question the timing of central govt taking over of vaccination from states. They perfectly knew in April that the supply will go up in late June or July, so they have pulled in states to have some cover. And they are taking over the vaccination completely just when the supply side constraints are over.

[suresh]

Glad that the doses are high but the doses are not being replenished fast enough as the count of doses with the states is falling rapidly. The center said the states would be getting 120M doses this month which would 4M/day. But even if our avg daily inoculation count is less than 4M/day, the doses with the states is decreasing. May be during the last 2 weeks of this month the states will be provided with higher avg of doses. And Niti Ayog said on June 1st that in July we will get a supply of 200M doses. So, the supply will continue to go up continuously. This actually makes me question the timing of central govt taking over of vaccination from states. They perfectly knew in April that the supply will go up in late June or July, so they have pulled in states to have some cover. And they are taking over the vaccination completely just when the supply side constraints are over.

I have a feeling the estimate of the supply numbers given out by the government is overly optimistic.

[depleter]

I have a feeling the estimate of the supply numbers given out by the government is overly optimistic.

Govt's June projection given in middle of May - Covishield - 65M, Covaxin - 25M and SPutnik V - 10M adding upto a total of 100M
They actually increased their projection to 120M after SII said they can provide 100M doses in June at the end of may. Does the govt really doesn't know about SII's production capabilities? If we combine both the projections the total actually comes upto 135M. But they kept it at 120M. May be the govt knows the expectations and are actually keeping the projections lower than the count they are getting from the suppliers.

[jai_in_canada]

OK, I'm going to go out on a very flimsy limb and provide a prediction for daily new cases this week and the next 2 weeks, and then a longer term prediction.

This week we will end up averaging 56K new cases per day -- yesterday was about 62K new cases, so to get to the weekly avg. of 56K daily new cases, we should end this week at about 50K new cases (by Fri/Sat).
Next week (of Jun 21) we will end up averaging 40-45K new cases per day.
And the following week (of Jun 28) we will go to about 30-35K new cases per day.
You can put your money on this if there is a betting website, and thank me later (yeah, I know, pretty bold and probably dead wrong, but that's what is beautiful about gambling! There is actually a method behind this madness, though!).

I think that we will drop for 4 more weeks after that (until mid-August) to get to about 15-20K cases per day on average.
The for the following 6-7 weeks, I expect us to hold the level at about 15-20K new cases per day (until late Sep).

Then.... will we see a new wave start to build in late September (there is already talk of a Delta Plus variant)?
Will vaccinations have ramped up by then to flatten Wave 3.0?
Or will the Delta Plus variant escape the vaccine and run ashore like another tsunami?
Don't bet on the last part just yet.

[jayakris]

Estimates of excess deaths in four states:

Actually a lot more info is there in the twitter thread where Chinmaya Tumbe (and IIM Ahmedabad professor)... Worth looking closely at. See this thread unroll page connected to it - https://threadreaderapp.com/thread/1397 ... 61441.html (Tumbe has looked at the death reports of Mumbai from 1916-18 and has a screenshot of the graphs from then!! Wow).

Basically, his compilation of numbers from multiple sources for the 4 states are pointing to something closer to what JIC was thinking (around 8 to 14 times the deaths) than what I was thinking (4 to 7 times). And some populous states like MP may have 30 times the deaths and all that. It is hard to believe, but as more and more such data come out, we have to believe it.

But there is also one issue in what Prof. Tumbe is doing - which is a projection estimate of normal deaths for 2020 itself that is made from the 2017-2019 rates. That is the right thing to do. I believe he did it for Gujarat, but I am not sure if he did it for all the states. The officially reported deaths for march-May in 2020 are unusable because of the lockdown and all the issues last year, so normal deaths for both 2020 and 2021 are estimates. But there is again a problem. As India has a very large number of normal deaths and Covid deaths are only a very small part of it anyway, even a small error in the 2020 and 2021 projection for normal deaths could cause serious multiplication factors to all these excess death numbers. (On top of it, to complicate it even further, there is a serious curve inflection going on with the normal deaths in recent years from a dropping trend to a rising trend too, even pre-Covid. See this - https://www.macrotrends.net/countries/I ... death-rate )... So the 2020 and 2021 normal death estimates could be quite wrong too. Argh.

So the bottom line is that it is VERY difficult to say anything. I am going to stick to a reasonable death-multiplier in the range between the range of 4 to 14 where I am on the lower end with a multiplier of 6 and JIC is at the upper end with a multiplier of 9. Also, both JIC were talking of overall data from both the waves. Chinmay Tumbe is focusing only on the second wave, and he says the multiplier for the first wave might have been only 2 or 3. So, maybe our estimates are biased by that. May be it is 3 for the first wave and 8 for the second wave, if we go by my judgment, and it is 5 for the first wave and 12 for the second wave if we go by JIC's estimates. And maybe neither of us should use national numbers and go state by state and use different numbers, because there are huge disparities from state to state. I may try working some spreadsheet numbers later when I get time and see what we will find.

It all mean that we know nothing!!! But it is clear that there is some serious multiplier for deaths in the second wave. Not less than 5 for sure. Probably 8 or 10.