Comment by flukus | Hacker News Reader

Comment by flukus | Hacker News Reader

Parent Full thread

The WHO has been reporting ~1% of cases are asymptomatic, this has it's own caveats because it comes after China's lock down but there's little reason to believe there a huge numbers of unknown cases out there, at least in countries testing well.

> With a virus which can be extremely mild in the vast majority of cases

I haven't seen a breakdown of how many get what the "mild" symptoms, but anything not requiring hospitalization including pneumonia are considered mild. Many of these mild cases would at least be going to their doctor to get a week off work.

Edit - found a break down: https://www.who.int/docs/default-source/coronaviruse/who-chi...

> As of 20 February 2020 and 12based on 55924 laboratory confirmed cases, typical signs and symptomsinclude:fever (87.9%), dry cough (67.7%), fatigue (38.1%), sputum production (33.4%), shortness of breath (18.6%), sore throat (13.9%), headache (13.6%), myalgiaor arthralgia (14.8%), chills(11.4%), nausea or vomiting (5.0%), nasal congestion (4.8%), diarrhea (3.7%), and hemoptysis (0.9%), and conjunctival congestiion (0.8%).

Those symptoms are worthy of a doctor visit and skipping work.

> The WHO has been reporting ~1% of cases are asymptomatic, this has it's own caveats because it comes after China's lock down but there's little reason to believe there a huge numbers of unknown cases out there, at least in countries testing well.

Unfortunately, the United States is currently a country that is not testing well.

Knowing 1% of reported cases are asymptomatic doesn’t tell us anything about what percent of asymptomatic cases are unreported. Only way to know what is testing with random sampling from the asymptomatic general population.

> Only way to know what is testing with random sampling from the asymptomatic general population.

They did. They tested a lot of people in contact with infected people (so even more likely to be an asymptomatic carrier than the general population) and their findings from the link above is:

> Asymptomatic infection has been reported, but the majority of the relatively rare cases who are asymptomatic on the date of identification/report went on to develop disease. The proportion of truly asymptomatic infections is unclear but appears to be relatively rare and does not appear to be a major driver of transmission.

Given the evidence the onus is proof is now on anyone claiming there are large amounts (of course there will always be some) of undetected cases.

I am not talking about entirely asymptomatic cases, which I believe are fairly rare. I am talking about mild cases which could be easily mistaken for a common cold, which is the most common presentation.

I think what you may be missing is that 100,000 people in WA could have gone to their doctor last week with “a cold” and they wouldn’t have been tested for COVID, because the FDA would not allow state labs to run the tests and the CDC refused to run it on mild cases.

I use the example of WA because it’s glaringly obvious that a lack of testing has resulted in a significant number of cases going undetected over the last couple weeks in that area.

The old CDC guidelines only allowed severe cases to be tested, so the only confirmed COVID cases were severe. Resulting in a “fatality rate” that was pushing 10%.

So it’s just an example of how this number is inflated by lack of testing. I believe a similar situation is playing out in Italy.

By comparison, South Korea has an extremely robust testing regime (free drive through testing stations, over 100,000 tested, capacity of over 10,000 per day) [1] and has identified 6,284 cases and has seen 40 deaths with currently 52 in serious condition. That puts their “known case fatality rate” at 0.6%-1.5%.

Simply put, the “known case fatality rate” can only possibly overstate the actual fatality rate, unless you believe there are deaths due to COVID happening that are being misattributed. This is obviously an entirely different category of error than not testing someone with a mild cold.

It’s useful to take what we know about “known case fatality rate” in heavily-tested populations and we can use this to estimate how many undetected cases are present in the population in under-testing locales. E.g. 10 COVID deaths would point to the likelihood of over 1,000 actual cases of COVID in the community. This of course assumes similar demographics and standards of care, which may not be the case.

[1] - https://www.ctvnews.ca/health/south-korea-takes-fast-food-ap...

~90% of cases get a fever which is uncommon for a cold so if there were large amounts of people with the disease it would at least be showing as a huge spike of people with the flu, plus we'd likely see a spike in deaths. It won't be mistaken for just a cold. The way it's playing out in Italy is that there were a lot of diagnosed cases early on but once the testing caught up they seem to be on the same exponential curve as everywhere else, but this was only high in percentage terms and low in raw numbers.

There is definitely under reporting of cases (and some misdiagnosed deaths), but so far no reason to think it's a tip of an iceberg type of scenario and there are large masses of undiagnosed people. For calculating mortaility rates there's also the lag time between getting the virus and death to take into consideration.

Going back to your original post about the fatality rate in WA, the numbers are simply too low to indicate anything at all. There are only 4 countries on Earth we can get much statistical significance from and of those Iran is the only one that lacks testing at the moment.

You make a very good point that a high fever is uncommon with a cold, and so at that point people would be thinking they perhaps have the flu and, depending on the culture and health care system, could decide to go get a flu test.

Once that’s negative you would expect maybe a COVID test would be done, again, depending on the health care system, availability of testing, and perhaps the financial impact.

Reply on news.ycombinator.com