Reasons why coronavirus mortality of young adults may be underestimated.
post by avturchin · 2020-03-15T16:34:29.641Z · LW · GW · 13 commentsTL;DR Our estimation of young adult mortality could be biased because of longer survival time before death in ICU and triage.
The data from Asia is that the mortality of young adults between 20-39 is like 0.2 per cent. This sometimes contradicts claims from the ground in Italy, like:
"There are many young people in intensive care, so we dispelled a 'myth', for which Coronavirus is a disease that affects only the elderly: stay at home, who can, don't go out, maybe just so we could come out! There will be a happy ending only if we respect the rules. So stay at home, respect the safety distances ”
However, Italy still almost didn’t report deaths of people below 50. The main question is: could it be taken as evidence of the low young adult's mortality in the situation after the medicine collapse?
The lack of reports of the death of young adults could have 3 possible explanations which need additional verification:
1) Better survival on the ventilator. Young adults survive better if they put on the ventilator. I don’t know how many young adults are currently on the ventilator. However, if there will be a shortage of ventilators, the mortality of young adults will jump.
2) Better triage. Young adults are getting better triage. Because young adults seem to survive better on ventilators and also are a more valuable part of society, they are given ventilators, and older people are turned off if there is a shortage of ventilators. It was reported that in Italy they had to turn off 40 years old people to save 30 years old ones (I lost the link, but will add when will find).
3) A longer time before death. Some young adults could survive for a much longer time on the ventilator and will die anyway. Maybe they can survive 6-8 weeks on the ventilator but will die eventually. In that case, we are just not starting to see these deaths, as the jump of infections started around mid-January, or around 8 weeks ago. In China, there are still 3000 people in serious condition, and as the peak of infections was at the beginning of February, most of them were diagnosed more than a month ago.
The main question here is the age of people in ICU, not age of deaths, as younger people may live longer in ICU. Some data points: Half of the Dutch patients in ICU are below 50, and 16 years boy is in a coma.
13 comments
Comments sorted by top scores.
comment by Steven Byrnes (steve2152) · 2020-03-15T19:46:39.791Z · LW(p) · GW(p)
Another possible confounder (sorta in the other direction) is the "inoculum", i.e. how many virus particles you get from the outside. It's big if you're a nurse and patients sneeze in your face all day. It's small if you touch a glove that touched a door that touched a hand that touched a sick person's tissue. It seems to be common knowledge (AFAICT) that bigger inoculum leads to worse symptoms for any disease. (Why? I guess there's kinda a race between the virus reproducing itself and the immune system figuring out how to fight it, and a big inoculum gives the virus a head start? Just guessing.)
Anyway, the anecdotes I've seen about healthy young adults dying of COVID-19 tend to be stories of doctors and nurses ... thus with presumably very big inocula. (I haven't systematically searched, that's just what I've come across.) Maybe this wouldn't generalize to low-inocula contexts, which are probably more common in the general public...
comment by mononote · 2020-03-15T18:57:20.280Z · LW(p) · GW(p)
A remark on data in Italy.
In the report from ISS [3] it is said that, as of 2020-03-13, in Italy there have been several deaths below the age of 50, in particular
- 4 cases in the age range 40-49
- 2 cases in the age range 30-39.
On the last two cases, it is known that they are
- one male person of age 39 with "psychiatric conditions, diabetes and obesity" who died at home,
- one female person of age 39 with "pre-existing cancer pathology" who died in hospital.
There are also other informations in that document (most common symptoms, most common patologies, statistics on the number of patologies). I'm not aware of data on the age distribution of people in ICU, but there is a well known case, "patient 1 from Codogno", of age 38 who was in critical conditions and was on ventilator for quite long.
1 ISS page
2 Update on 2020-03-12
3 Report on deaths
comment by avturchin · 2020-03-17T00:42:12.789Z · LW(p) · GW(p)
New data: "If we assume the age distribution of severity is the same in Korea vs. China, but that the Korean rate of severe illness is the correct one, this gives us 3/19*10 = 1.6% probability of a person under 40 having a severe case, given that they have been infected." (COVID-19 Risks For Young People)
Replies from: Bucky↑ comment by Bucky · 2020-03-17T13:05:02.422Z · LW(p) · GW(p)
Note that severe =/= critical (I think that post confuses severe with critical in the conclusions)
In the three Chinese studies which are quoted, severe includes e.g. cases of shortness of breath and high fever which, whilst possibly hopsitalisable under normal circumstances, not obviously fatal if hospitals are full. Severe doesn't imply "requiring mechanical ventilation or other intensive care."
A better estimate for ICU cases based on that evidence would be ~0.4% as (severe + critical) / severe = 4 in the 44k person China data.
Of course some severe but not critical cases might become critical if not treated, so death rate without any treatment would be between the two.
comment by Sherrinford · 2020-03-16T13:59:56.516Z · LW(p) · GW(p)
Based on the Diamond Princess data, the case fatality ratio was estimated be 1.9-2.3% for all age groups, and 14-18% for people aged 70 or older, and seemingly around 0.2% for people up to age 39 (https://cmmid.github.io/topics/covid19/severity/diamond_cruise_cfr_estimates.html?fbclid=IwAR12StTC_CnjEWf2nGkw0b9tqEuh6kZ_qeEMXetM_cAHoZW3E0-yAchk66A). The infection fatality rate was estimated to b about 50% of the case fatality rate. I assume this isolated-people data may be helpful.
Replies from: dmitry-vaintrob, avturchin, Sherrinford↑ comment by Dmitry Vaintrob (dmitry-vaintrob) · 2020-03-16T14:44:01.968Z · LW(p) · GW(p)
DP Cruise didn't have any fatalities under age 70, so not sure where you're getting the under-29 number. Also since the population is older the case fatality was over-estimated. This study https://cmmid.github.io/topics/covid19/severity/diamond_cruise_cfr_estimates.html?fbclid=IwAR2jCOZcBGHYBWC_dqSzwvX7T7-DOpwm8L84qqW8k6QtKa05Inv35Pk3Ezs estimates adjusted CFR form DP cruise ship data (assuming treatment!) to be .5%, largely in agreement with other numbers I'd heard. Though the sample size is ridiculously small, so the error bounds are terrible.
Replies from: Sherrinford↑ comment by Sherrinford · 2020-03-16T19:19:03.161Z · LW(p) · GW(p)
Ok, thank you. I added the word "estimated" above, but now I guess I still misunderstood what they were reporting in table 2 then.
↑ comment by avturchin · 2020-03-16T15:07:09.770Z · LW(p) · GW(p)
Yes, but everybody on the ship got first class health care, and 233 people are still ill (1.5 months later form most infections), including 15 in critical condition. We don't know the age of those who are still ill or how many younger people will die if there will be no health care. Also, there will be disproprtionally small number of young people on that ship.
Replies from: Sherrinford↑ comment by Sherrinford · 2020-03-16T18:59:01.810Z · LW(p) · GW(p)
The study is relevant because it aims to control for the age distribution, but of course if data is biased you never know completely. That everybody got perfect health care should also apply to all age groups, shouldn't it?
Replies from: avturchin↑ comment by avturchin · 2020-03-16T21:29:13.628Z · LW(p) · GW(p)
Yes, older people will also have higher mortality when medicine fails.
And: "We see people in their thirties with no medical history, images of lungs are terrifying"
Replies from: Sherrinford↑ comment by Sherrinford · 2020-03-16T21:54:37.486Z · LW(p) · GW(p)
Sorry but I do not see where older people having higher fatality when medicine fails leads to bias in the diamond princess data that would fit what you asked for in the original post. Please explain.
Replies from: avturchin↑ comment by Sherrinford · 2020-03-16T19:20:34.375Z · LW(p) · GW(p)
I would edit this to reflect the comments below, but somehow my words get deleted after typing.