It is generally considered really good to test as many people as possible. If nothing else, it helps make the statistics far more accurate. It can also reveal who should be isolated for longer.
It is generally considered really good to test as many people as possible.
If nothing else, it helps make the statistics far more accurate. It can also reveal who should be isolated for longer.
Take a look at the results of the Santa Clara study for the problems with doing a study using a test of unknown accuracy. They found 50 positive results from 3330 people but we don't really know...
Take a look at the results of the Santa Clara study for the problems with doing a study using a test of unknown accuracy. They found 50 positive results from 3330 people but we don't really know what that means.
More testing will probably help, but if it's non-random and there is no pre-screen, we might not be able to tell what's going on due to an unknown number of false positives.
I think the only way to gauge accuracy is to test very large quantities of people, and test them 2-3 times at once, and follow up like 1-3 weeks later. The larger the sample size, the easier to...
I think the only way to gauge accuracy is to test very large quantities of people, and test them 2-3 times at once, and follow up like 1-3 weeks later. The larger the sample size, the easier to determine statistical significance. 10,000,000 will provide a much better estimate than 10,000.
Either way, false negatives are far more harmful than false positives when trying to contain the spread ASAP. All negative tests should remain highly isolated for 2-3 weeks for a followup (in case of false negatives), and all positive tests get very rigid quarantines for 2-4 weeks.
That doesn't really get you around whether you have systematic errors. Perhaps some people have slightly different antibodies that the test doesn't pick up ever. Perhaps some people have some...
That doesn't really get you around whether you have systematic errors. Perhaps some people have slightly different antibodies that the test doesn't pick up ever. Perhaps some people have some other antibody for another virus that the test picks up. Same with direct tests (for the virus). Systematic errors are plausible and you can't catch those by retesting. But you can compare with a known test. So do a study of 10000 samples that you test with both lab equipment and field tests. If the results disagree, you have a much better idea of what is going on.
If your test produces enough false negatives that you have to worry even if you receive a negative result, you might as well not test. Even if a retest would get you clarity. With covid-19, a false negative should be tolerable, we're just going to have to catch it down the line - i.e. test and quarantine the people your false-negative-patient infected. As long as a test enables us to lower R, it's a good one. No measure is perfect, so we need defense in depth anyway.
Nothing is said about whether testing everyone would be a good idea or what would actually be done with the data; it's just assumed it's good.
It is generally considered really good to test as many people as possible.
If nothing else, it helps make the statistics far more accurate. It can also reveal who should be isolated for longer.
Take a look at the results of the Santa Clara study for the problems with doing a study using a test of unknown accuracy. They found 50 positive results from 3330 people but we don't really know what that means.
More testing will probably help, but if it's non-random and there is no pre-screen, we might not be able to tell what's going on due to an unknown number of false positives.
I think the only way to gauge accuracy is to test very large quantities of people, and test them 2-3 times at once, and follow up like 1-3 weeks later. The larger the sample size, the easier to determine statistical significance. 10,000,000 will provide a much better estimate than 10,000.
Either way, false negatives are far more harmful than false positives when trying to contain the spread ASAP. All negative tests should remain highly isolated for 2-3 weeks for a followup (in case of false negatives), and all positive tests get very rigid quarantines for 2-4 weeks.
That doesn't really get you around whether you have systematic errors. Perhaps some people have slightly different antibodies that the test doesn't pick up ever. Perhaps some people have some other antibody for another virus that the test picks up. Same with direct tests (for the virus). Systematic errors are plausible and you can't catch those by retesting. But you can compare with a known test. So do a study of 10000 samples that you test with both lab equipment and field tests. If the results disagree, you have a much better idea of what is going on.
If your test produces enough false negatives that you have to worry even if you receive a negative result, you might as well not test. Even if a retest would get you clarity. With covid-19, a false negative should be tolerable, we're just going to have to catch it down the line - i.e. test and quarantine the people your false-negative-patient infected. As long as a test enables us to lower R, it's a good one. No measure is perfect, so we need defense in depth anyway.