[Goose]

Until widespread testing begins (it's been a week away for about a month now), take no comfort in low numbers of positive tests.  They are not testing enough people to get any real sense of the actual numbers positive.

(My job largely involves analyzing survey data, and the lack of any real surveillance of this disease is making me crazy.)

The problems with doing widespread random testing are many. First, it uses up the limited PCR machine and technician time, and to a lesser expend "rare" (fixable with effort) testing materials. Second, and more important, the test has limited sensitivity. There will be a significant number of people who test negative that are in early stages of the disease and don't have high enough counts to test positive.  Therefore any number we measure regarding infection rate is going to be low by an unknown value. If we had lots of historical data about this, we MIGHT be able to infer how many people are actually infected, but we don't.  Third, the distribution of who is sick/will be sick is very non-random. There are localized "pockets" where the disease is epidemic and other much larger areas where apparently there is no disease. The location of these pockets is non-random but unknown until somebody shows symptoms.

These are some of the reasons that testing asymptomatic people is a questionable practice. A negative test doesn't necessarily mean they aren't sick. A negative test  following isolation for the "incubation period" (whatever that actually is/should be) is probably reliable. Anything less isn't. If we could test everybody and isolate all the positives, we could really slow down the infection rate, but it wouldn't go to zero. If we could test a large sample, we would know a lot more about the infection rate, especially if it is "significant". However, it is IMHO improbable this type of testing would tell us much with a very low overall infection rate with high density pockets. This later situation is where we currently find ourselves.

[Goose]

Until widespread testing begins (it's been a week away for about a month now), take no comfort in low numbers of positive tests.  They are not testing enough people to get any real sense of the actual numbers positive.

(My job largely involves analyzing survey data, and the lack of any real surveillance of this disease is making me crazy.)

The problems with doing widespread random testing are many. First, it uses up the limited PCR machine and technician time, and to a lesser expend "rare" (fixable with effort) testing materials. Second, and more important, the test has limited sensitivity. There will be a significant number of people who test negative that are in early stages of the disease and don't have high enough counts to test positive.  Therefore any number we measure regarding infection rate is going to be low by an unknown value. If we had lots of historical data about this, we MIGHT be able to infer how many people are actually infected, but we don't.  Third, the distribution of who is sick/will be sick is very non-random. There are localized "pockets" where the disease is epidemic and other much larger areas where apparently there is no disease. The location of these pockets is non-random but unknown until somebody shows symptoms.

These are some of the reasons that testing asymptomatic people is a questionable practice. A negative test doesn't necessarily mean they aren't sick. A negative test  following isolation for the "incubation period" (whatever that actually is/should be) is probably reliable. Anything less isn't. If we could test everybody and isolate all the positives, we could really slow down the infection rate, but it wouldn't go to zero. If we could test a large sample, we would know a lot more about the infection rate, especially if it is "significant". However, it is IMHO improbable this type of testing would tell us much with a very low overall infection rate with high density pockets. This later situation is where we currently find ourselves.