Saturday, May 2, 2009

Laboratory Confirmed A/H1N1 Cases - update

Already out of date from earlier this afternoon, country count is now 18, with South Korea added to the reports 809 confirmed cases worldwide, with 197 in the USA.


Wisdom Speaker said...

"Thanks for the reference, that is an interesting read - which engenders the question, what is the estimated R naught for A/H1N1?"

We can estimate the answer from your chart of the cases versus date, but it's not a pretty picture.

From your chart, the time to go from 25 confirmed cases to 250 was 6 days. The time to go from 75 to 750 was 5 days. In the paper, the most extreme case had a 10-folding time of 9 or 10 days. This would suggest that R0 is greater than 2.4!

I sorely hope that the number of confirmed infections is rising faster than the number of ill persons ("institutional catch-up").

Unfortunately there are also reasons to suspect that not to be the case (incentives not to report confirmations in Europe to avoid "Phase 6", delays in testing due to limited supplies and throughput, under-testing documented in Mexico by Newsweek...).

One might also do another estimate based on the break-out date (early April?) and assuming that there are about 10 times as many un-confirmed cases as confirmed cases (based on the lag in patients coming in and getting tested). In that case it's been about a month to get to 10,000 ill persons, and we get R0 = 1.6. However, since some patients do not get such a bad case, many will not even see the doctor ("it's just a little virus"), and in that case there might be many more cases than that. If we assume that there are 9 or 10 sick people who didn't even get tested, for every one that did, then there might be 100,000 ill people now and we get back to R0 = 2.4.

In the latter scenario, and assuming that the interventions are not especially helpful in reducing the speed of the disease spreading, we hit 1,000,000 ill persons in mid-May and the peak of the epidemic (8,000,000 getting ill each day) in early June...

BTW, I think there is a lot of complacency/denial about the relatively low death toll so far.

Suppose we actually do have R0 = 2.4. And suppose that it takes 7 days for someone who gets sick to get a "confirmed" test, and another 7 days for the incidence of death (e.g. succumbing to pneumonia). In that case, the number of "real time" deaths will lag the number of eventual deaths by a factor of 10.

In other words, there are plenty of cases of people who are now sick, and who won't make it (due to whatever complications), but most of these "doomed" people just haven't died yet.

So, judging by the numbers, the virus will appear much less dangerous than it really is, up until the peak of the epidemic when the number of new cases finally stabilizes.

To get a better handle on the eventual mortality, the number of fatalities in Mexico should be compared to the number of actual illnesses from a few days ago, not the number of confirmed cases today.

Wisdom Speaker said...


Let's assume that this is a "normal" flu in terms of mortality of those who get ill, but that because very few have resistance, and it seems to be spreading very aggressively (high R0), it is likely to infect a higher proportion of the population than a normal seasonal flu.

In that case we must expect at least 30,000,000 infections in the U.S. (and possibly 100,000,000 or more -- note that high R0 correlates with a higher total number of cases in the PNAS paper by Germann.)

A normal winter's seasonal flu kills 30,000 people, i.e. about 1 out of every 1000 who gets sick.

So, this flu, even with "normal" flu fatality rates, can be expected to kill 30,000 to 100,000 people, by itself. If it maintains the high R0, then not only does the number of fatalities increase, but the duration of the epidemic also decreases.

Finally - one can multiply by 10-20 for the world as a whole, since the U.S. has only 5% of the world's population, and not all nations will have comparable medical care (but on the other hand the virus might not spread well in all climates).