COVID-19 Rate vs. Positivity Rate

The positivity rate is what percentage of tests come back positive. If you only test people who think they are sick or exposed, you get a high positivity rate. If you test lots of people, say all employees of a company before they come to work, you would expect a lower rate since many of those people are likely healthy. The COVID-19 rate is typically number of confirmed cases per 100,000 people in a community.

engineer-

Thank you.

You make a very good point. Each time a hear the positivity rate for a given locality cited, no mention is made of whether it was based on a random (or near-random) sample.

As to the COVID-19 rate is typically the number of confirmed cases per 100,000 people in a community, you need to say how the sample was selected.

The covid-19 rate (i.e. cases per 100,000) extremely sensitive to the sampling technique. There are lots of cases that are undetected because people with minor symptoms don't get test.

The positivity rate (i.e. positive cases per 100,000 covid tests given) is also sensitive to the sampling technique. Obviously if you give tests to everyone, the positivity rate will be lower than if you only give tests to people with symptoms.

Doubling the number of covid tests given will greatly increase the covid-19 rate.

Doubling the number of covid tests given will decrease the positivity rate. Given the prevelancy, I would expect this decrease to be less dramatic.

I think the positivity rate is a fairly good (not perfect) measure of the progression of the virus. If the positivity rate is going up, that is a bad thing (especially if the number of tests given is either equal or staying the same).

The experts, of course, look at a wide number of statistics to understand the progress of the disease.

The number of hospital beds available is a number this is obviously very relevant. When your community no longer has available hospital beds, you are all screwed.

maxdancona-

Thank you.

"Doubling the number of covid tests given will greatly increase the covid-19 rate." I would have guessed it could either increase or decrease the rate depending on the respective samples used prior to doubling and the sample used for the doubling.


"Doubling the number of covid tests given will decrease the positivity rate." Again I would have thought the results would depend on the respective samples used prior to doubling and the sample used for the doubling.

Technically you are correct, however you are ignoring how these tests actually work in real life. Generally people with symptoms are the first people to get tested... the people showing up in the hospital with symptoms are getting tested before the people who have no indication they are sick.

1. If you test the 10,000 sickest people in a city you will find a certain number of cases. To get the covid-19 rate you divide this number by the total population of the city. If you then test another 10,000... it is a mathematical impossibility that the covid-19 rate would go down (and the covid-19 rate will almost certainly go up as new cases are detected).

2. If you test the 10,000 sickest people in a city, the "positivity rate" is the number divided by 10,000. The population of the city has nothing to do with it. If you then test the 20,000 sickest people in the city (you are now test people that are slightly less symptomatic), the positivity rate will very likely go down.

Of course if you test a true randomized representative sample, then the case rate and the positivity rate should correlate. But that is not what we are doing in general.

I think there are some test populations that are interesting to experts, for example the positive rate on campuses (where people are forced to be tested).

The experts are looking at all the data.... and things aren't going well.

maxdancona-

Thank you. I think they should test either everyone or a random sampling of the population.

A random sample isn't possible unless you have the ability to force people to take the tests. It is often difficult to run experiments on human beings.

When you do experiments on humans, often we do a double blind experiment to compare the study group with a control group. In some cases, including this one... that doesn't work.

We are doing this in certain populations... the military and college campuses for example. These are populations that can't refuse a test.

Of course the experts look at this data too. And the clear fact is that the numbers are horrible.

Maybe we should change the law to require testing.

How do you propose we do that Gollum? We are having trouble getting people to wear a ******* mask.

maxdancona-

Yes, your point is well-taken.

I think that it would have to be done at the State level. I don't believe that the President has the authority.

I believe each State has a law granting the Governor the right to declare an emergency granting him emergency powers. The Governor should invoke that law to require universal testing, much as some governors have invoked universal masks.

I am aware that there will be compliance problems.

Let's be straight about what you are suggesting. You want a random sample.

First of all, there aren't nearly enough tests to test everyone. I assume you will want to use a big part of the tests for the people who are symptomatic. What you will do is commandeer some portion of the tests (maybe 5,000,000 or 100,000 per state) for your little project.

Then you will have some sort of a military draft. Each person for this study will have to be selected at random.

If someone's number comes up, they will be forced under penalty of law to complete the test. If they don't, they will be put in jail.

Is that what you are suggesting?

The facts are pretty clear.

1) The number of people with confirmed cases covid-19 is going up dramatically.

2) The percentage of people who take tests that are positive is going up dramatically.

3) The number of people showing up at hospitals with covid-19 is going up dramatically. The number of people who need respirators is going up.

4) The number of people dying of covid-19 is going up dramatically. And ,the number of people dying is also going up.

At some point the evidence becomes clear, you just have to accept it.

maxdancona-

OK, you win that round.

How about the United States Government enacts a lockdown where all persons in a certain geographic area must stay in their homes for 14 days.

At the conclusion of 14 days, a government representative would test each person. Those found to be positive would be moved to an isolated area, the rest could go free.

My understanding is that if you could do a military lockdown, it would be easy to end the pandemic in a couple of weeks.

That obviously isn't possible even in a country controlled by the military.

Our Supreme Court just said you can't take away constitutional rights even to stop a pandemic. Your idea is sound.... for a science fiction plot.