I was talking about the entire world. Note my first sentence. “It is not just the gulf states.” Many of the Gulf states are still sitting on 1.1 daily increases. My example based on about 1.25 was illustrating how you could get huge disparities. Easily 100:1 when growth is massive. I was illustrating a general point about the numbers, one that applies everywhere.
I guess I assumed people could work out that a 10:1 disparity would occur with a lower growth number. 1.1 for three weeks will easily get you a 10:1 disparity. 1.1 is still rapid exponential growth for our purposes. If you have n ill people today, you will have over ten times that in three weeks, and 100 times that in six.
It isn’t hard to look at the numbers and work out the precise values for each country one is interested in. I leave that as an exercise for the reader.
Japan only tests on admission. There are likely others. The point is that it is very hard to compare countries. Japan is the sleeper in this mess.
I think your point about positivity rates is excellent,although I don’t think it accounts for the full difference. The WHO suggests a good benchmark is ten negatives per positive. Under that, (as New York is) and prevalence numbers for active cases are probably off. Worse, they may be skewed toward sicker patients, thus skewing your CFR.
Saudi Arabia, for example, tests about 10K per million and has 5 deaths/M with a CFR of about 0.6. For Kuwait, 42K tests/M; 8 deaths/M; CFR about 0.7.
New York: 47K tests/M; 1,200 deaths/M; CFR about 7.5%.
New York has tested about 5% of its population; Saudi Arabia about 7.5%. Although NY is certainly missing more active cases than is SA, 5% of a population is still a pretty good sampling. The higher your total sampling rate is, the less likely you need to approach the 10:1 target to get a reasonable (though still crude) idea for the prevalence of active cases.
But I want to thank you for making such a good point. I’ll go chew on it some more.
In addition to some of the reasons cited (higher percentage of elderly in New York, on a different part of the curve, …), here is another major possibility: Antibody testing across NYS comes up with an estimate that 12.3% of the population has antibodies for the virus. While false-positives might in principle be an issue, the fact that the rate varies so significantly between regions of the state suggests that the false positives can’t account for more than a couple percent of these at best.
If you use that number to compute a death rate, you get about 1%.
Elon Musk in the near past said that hospitals might be inflating COVID-19 numbers because they get more money for that. I don’t know how true this is. But this might be one of the reasons.
If there is a concern among some contributors here that countries and organisations are making up stats out of arse-protection, self-interest or even problems with counting big numbers reliably, then i suspect all of that applies doubly to private pundits, especially those who have a track record of shooting from the mouth and equating enlightened public policy with their own self-interest.
Not sure what ad hominem attack you read. And I have no grounds to question your veracity.
Meanwhile this NYT article asks the question of “Why Does the Virus Wallop Some Places and Spare Others?” inclusive of S.A. and may be of interest here. FWIW genetic differences among sub-populations is, as it should be, one possibility to investigate as a contributing factor.
Thanks. I read it, and have read almost everything I can find on SARS CoV 2. Like BCG, the “genetics” idea is pure speculation. Sure; maybe. But speculation.
I was curious with that article that they did not mention hydroxychloroquine wrt India’s approach. It is 180 degrees opposite of New York for early use of HCQ.
India and Bangladesh are using HCQ like crazy. The Indian Council for Medical Research recommends it for prophylaxis (not tx, interestingly enough) and the broad Indian medical community uses it very generously for positive cases, official recommendation notwithstanding.
By way of comparison, India’s reported CFR is something under 4%; Bangladesh under 2%; both nations so far have been doing pretty good keeping positives low per million. It will be kind of interesting to see how much that changes. Can they keep it up, and if so will it turn out HCQ is good for either prophylaxis or early treatment?
Iceland is another interesting cohort. They test like mad and are probably not missing any cases. Their CFR is 0.5%.
BTW, to the point of your quote from the article that “Turkey was fine…until it was not”…how is Turkey not fine?
Turkey has 1500 cases/M; NY 16K. Turkey has 41 deaths/M; NY 1250. Turkey’s CFR is 1/3 that of New York (Turkey attributes this to their early use of HCQ).
I haven’t looked at Russia.
On the ad hominem issue, I was wondering why you thought any other positions–highly unpopular here, LOL–that I take are relevant.
But maybe you think genes really are relevant for COVID. I remain unconvinced.
If increased testing frequency or pop density had anything to do with it, I don’t think India and Bangladesh would be doing as well.
Agree that age is certainly contributing, though not sufficient to drive differences. Florida seems to be doing much better than is New York, both in terms of overall prevalence and CFR that is half that of NY. In Fl, 20% are over 65 v NY’s 16%. And b/c NY is testing at twice the rate of FL, if anything NY should have more minimally symptomatic positives.
Why? That would be true only if the prevalence of COVID-19 is equal in New York and Florida, and there is no reason to believe that is the case.
Suppose, for a (contrived) example, that twelve percent of New Yorkers and six percent of Floridians are infected. In that case, if New York tested at twice the rate of Florida, they’d still find only about the same percentage of minimally symptomatic cases in each state (assuming similar testing protocols, test accuracy, etc.). Now, is Florida’s infection rate half of New York’s, more than half, or less than half? The evidence suggests it is much lower: per Worldometer, New York has more than nine times as many confirmed cases as Florida per million population (16,509 versus 1791). One of every three New York cases is positive, versus about 1 in 12 in the Sunshine State. That would suggest that it is Florida that is much more likely to be detecting minimally symptomatic positives.
If you test more frequently, the percentage of positives that are minimally asymptomatic is higher because you are coming closer to actual prevalence for all positives.
If you test less frequently, you tend to confine your testing to symptomatic patients, and if you test infrequently, you tend to confine your testing only to quite symptomatic patients.
Muslim men don’t wear burkhas. They tend to be very social, and in particular, social while praying, which they prefer to do in large groups where possible.
I think you are misunderstanding the point I am trying to make, which is that your testing results depend on BOTH the frequency of testing and the prevalence of the disease in the first place.
Consider, for example, two separate populations, each with 10,000 people. Both focus their testing first on the most symptomatic, and then widen their net.
Population A has 26 positive cases, of which 2 cases are serious and 24 are minimally symptomatic. They’ve done 250 tests in that population, so about 1 in 10 will be positive and 1 in 125 (less than one percent) will be a serious positive.
Population B has 260 positive cases, of which 20 are serious and 240 are fairly minimal. They’ve completed 500 tests, so about 1 in 2 will be positive and 1 in 25 will be a serious positive.
Although population B has been tested at twice the frequency of A, there are so many more cases in B that they would have needed to test at TEN TIMES the frequency to obtain about the same distribution of symptomatic/asymptomatic, because there are ten times as many cases to find. In A, less than one percent of tests will locate a serious case; there simply aren’t many serious cases to find. In B, even though testing is more prevalent, a much higher percentage of those tests are going to be focused on the quite symptomatic because there are so many more people with symptoms.
Testing in Population A, even at lower frequency, probably comes closer to revealing actual prevalence of the disease.
New York and Florida are close to equal in population (20 million versus 22 million), but one has had 330,000 cases and 25,000 deaths, while the other has found 37,000 cases and not even 2000 deaths. That suggests the disease to date has infected a much MUCH greater percentage of New York’s population.