Hacker News Reader

Comment by zuminator | Hacker News Reader

Is there anywhere in the US where we are doing random testing of residents to determine what the real infection rate is here? From all I can see we're currently just adding up the number of people who have come in with symptoms and tested positive instead of getting a real handle on what percentage of the population is infected. The CDC website right now says that there are 938 total cases.[1] That's laughable. At this point I'd sooner believe 93,800 or 938,000 than 938. Oh wait: "Now that states are testing and reporting their own results, CDC’s numbers are not representative of all testing being done nationwide." OK, so nobody's keeping up with the totals? Is that what I'm reading?

[1] https://www.cdc.gov/coronavirus/2019-ncov/cases-in-us.html

View on HN

The CDC gave local labs the go-ahead to develop their own tests, but that was very recently and it seems very likely that the CDC doesn't have a centralized reporting system in place.

I really doubt there's capacity to do random tests for scientific field study, considering there's a large backlog of people with symptoms or suspected contact w/ infected that are waiting to be tested.

For example, here's aerial footage of the lines in Denver for a drive-thru clinic restricted to patients with "a doctor's order confirming they meet requirements": https://twitter.com/ABC/status/1237869000169644032

It's tragic that a lack of test kits is costing us the ability to accurately track the spread of this disease and its true mortality rate. If right now there are 100,000 (actual) positive and 50 deaths, that's a whole different situation than 1000 (reported) positive and 50 deaths.

Hmm, I'm not so sure...

Even if the mortality rate is much lower than what we suspect today, we know that the health system is going to be overloaded -- just as is the case in Italy.

Johns Hopkins is keeping up with the totals.

https://coronavirus.jhu.edu

Thanks for that link, very helpful.

I've been using this Live Science web page which appears to be fairly regularly updated and sources States data and news reports: https://www.livescience.com/coronavirus-updates-united-state...

It's been pretty much on the ball.

I've been maintaining a spreadsheet with States-level department of health Covid-19 situation page data and the totals roughly tally.

As of 13:00 GMT March 12: 1,300+ positive cases, 38 deaths. You could multiply the positive number by factor of 10 say to get a rough estimate of the total number of persons tested to this point (~13,000 let's say) that individual states, private labs, and the CDC have checked. (Hard to say though as only about 1/3 to 1/2 of states release full data and some are under-reporting (Nebraska and Texas afaict)) Another niggle is that it can (does?) take more than one "test" to perform a test.

I would think we be able to estimate fairly correctly. I assume pretty much everyone who dies from it would end up in a hospital and counted. And since we know the death rate, we know for every 1 death there were x who have it and will/have recovered.

Except the death rate is determined by deaths/cases, which relies on counting the cases.

There have already been missed deaths that were later attributed to Covid-19 in the US. Since the fatality profile of Covid-19 closely matches the fatality profile you expect anyway (people are more likely to die when they are old or have preconditions), it is actually a problem to get reliable death counts. Fatalities must have been tested, or tested retroactively, for correct attribution, and since the US currently isn't even close to doing proper testing of living patients, you can probably add two and two together by yourself.

The death rate is likely to vary depending on the strain and on the ability to treat effectively, both of which may change over time.

Reply on news.ycombinator.com