Would love to be put right by a pro.
But from simple perspective, to see this many deaths to date, either the disease has to be very deadly or many more people are infected than the government has even hinted at.
~18.5 day lag from infection to death?
@GidMK
-
-
Replying to @JamieWoodhouse @devisridhar
Median lag from infection until death is somewhere between 10-18 days, but not everyone is diagnosed on day 1 of infection. Many people are diagnosed when they get seriously ill and present to hospital, and die shortly after
2 replies 0 retweets 1 like -
Replying to @GidMK @devisridhar
Thanks Gideon. That's why I've tried to skip "confirmed/diagnosed cases" and instead focus on the relationship between actual infections - detected or not, and deaths. Actual infections critical for public comms re: lockdown complacency. "Confirmed cases" are iceberg tip?
1 reply 0 retweets 0 likes -
Replying to @JamieWoodhouse @devisridhar
Yes and no. Much better to stick to confirmed cases as this is the data you have. Extrapolating to 'actual' infections is extremely challenging and really relies on your assumptions more than it does the data
1 reply 0 retweets 0 likes -
If you are going to extrapolate, best evidence after the fact from Wuhan seems to indicate that 'confirmed' cases are somewhere between 20-50% of all cases, so a sense-check would be if you are extrapolating beyond a multiplier of 5x it's unlikely to be true
1 reply 0 retweets 0 likes -
Replying to @GidMK @devisridhar
OK - but if we assume confirmed cases (9.5k) are 20% of the total infected number, then we have ~47k total infections now per your article. If you use 14 days infected to death, that implies 14 days ago we had ~1300 total infections (4 doublings). So CV kills 1 in 3 infected!?!
2 replies 0 retweets 0 likes -
Replying to @JamieWoodhouse @devisridhar
Nope. You can't just extrapolate back like that. 14 days ago the UK had ~800 confirmed cases. Moreoever, see my first point in terms of using current death numbers and a median death time to extrapolate anything - you can't really
1 reply 0 retweets 0 likes -
Replying to @GidMK @devisridhar
We had 456 on 11/3. And your 456 x 5 implies 2280 real infections then. Yet we've seen 463 deaths 14 days later. Implying CV kills 20% or 1 in 5 of everyone infected? I just don't understand how we can get this many deaths without many more real infections 2 weeks back.
2 replies 0 retweets 0 likes -
Replying to @JamieWoodhouse @devisridhar
Because you can't just extrapolate from the median length of time. It's an average. That's why we use a case-fatality ratio, because this kind of extrapolation can be extremely misleading
2 replies 0 retweets 0 likes -
Replying to @GidMK @devisridhar
I understand this isn't precise, and it may just be my brain... But I still don't understand how an estimated 2,280 total real CV19 infected (from your 5x factor applied to "confirmed cases"), 14 days ago... Can cause 463 deaths to date.
2 replies 0 retweets 0 likes
If I have an 10,000 boxes, with a median of 14 apples inside, I could easily have 500 boxes with only 1 apple and another 500 with 30 apples Now replace boxes with cases of COVID-19, and apples with time until death
-
-
Replying to @GidMK @devisridhar
I get the apples, + I can understand how deaths/confirmed cases will be skewed by demographic. But we're talking about the total number infected, tested or not. That seems to have a much more normal distribution through the pop. So mean/median reasonable for a top down est?
0 replies 0 retweets 0 likesThanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.