What is a good test/population rate to be able to "trust" the positivity rate?
Meaning, when NYC was only testing people with symptoms the rate was through the roof, so we sort of know not to use it as a basis for anything.
My zip code is currently 5.2% positivity in the last 2 weeks (https://www.phila.gov/programs/coronavirus-disease-2019-covid-19/testing-and-data/#/ - I am 19147). There have been 582 tests / 10,000 people, or 5% of people have had a test. Is that enough tests to consider that number really high? Or, is it inflated because of a lack of tests?
10025 is testing 700/100,000 per day based on the NYC dashboard. Over 2 weeks that would be 980/10,000, so nearly 10% of residents are getting tested in a 2 week period - and the positivity rate is much lower (even if you halved my zipcode, the UWS is still better, but its not as stark a difference).
So, should I be totally freaked out by a 5.2% positive rate? Or is that inflated because of not enough testing? (And yes, I know we should be freaked out because relative numbers are skyrocketing.)
Also, not sure if you noticed, but NYT now has # hospitalized nationally as one of their metrics. Given the lag with deaths, and the fluctuation of #s based on testing, I know Cuomo liked to talk about hospitalizations during his briefings. I assume this is NYT finally getting national numbers to start to use this as a metric to counter the "it's because we are testing more" rhetoric, which I think its great.
I think you're touching on something that is a fundamental issue with testing programs. How do we contextualize them?
One of the problems is that there tends to be considerable selection bias in testing. People get tested much more often if they have a suspicion of having COVID--either due to symptoms or suspected exposure. So it's rather difficult to assess how representative these samples are.
New York State has come to a decision about the thresholds that they consider serious, and I think this is based on testing availability within the state. I don't know what PA's assessment is, but I will admit that a 5% rate sounds concerning to me personally if the overall trend is also positive.
I think it's a good resource because it uses the effective reproduction coefficient (an assessment of how many cases are generated for each case identified) as well as other factors to assess the risk of a resurgence in a given location. Currently, Philadelphia is in dangerous territory according to this measurement. If I were you I'd be very careful, based on that. New York isn't in much better shape, either, by this measure, for what it's worth, and I'm starting to act a lot more carefully as a result.
Couple of questions for you -
What is a good test/population rate to be able to "trust" the positivity rate?
Meaning, when NYC was only testing people with symptoms the rate was through the roof, so we sort of know not to use it as a basis for anything.
My zip code is currently 5.2% positivity in the last 2 weeks (https://www.phila.gov/programs/coronavirus-disease-2019-covid-19/testing-and-data/#/ - I am 19147). There have been 582 tests / 10,000 people, or 5% of people have had a test. Is that enough tests to consider that number really high? Or, is it inflated because of a lack of tests?
10025 is testing 700/100,000 per day based on the NYC dashboard. Over 2 weeks that would be 980/10,000, so nearly 10% of residents are getting tested in a 2 week period - and the positivity rate is much lower (even if you halved my zipcode, the UWS is still better, but its not as stark a difference).
So, should I be totally freaked out by a 5.2% positive rate? Or is that inflated because of not enough testing? (And yes, I know we should be freaked out because relative numbers are skyrocketing.)
Also, not sure if you noticed, but NYT now has # hospitalized nationally as one of their metrics. Given the lag with deaths, and the fluctuation of #s based on testing, I know Cuomo liked to talk about hospitalizations during his briefings. I assume this is NYT finally getting national numbers to start to use this as a metric to counter the "it's because we are testing more" rhetoric, which I think its great.
Thanks!
-david
I think you're touching on something that is a fundamental issue with testing programs. How do we contextualize them?
One of the problems is that there tends to be considerable selection bias in testing. People get tested much more often if they have a suspicion of having COVID--either due to symptoms or suspected exposure. So it's rather difficult to assess how representative these samples are.
New York State has come to a decision about the thresholds that they consider serious, and I think this is based on testing availability within the state. I don't know what PA's assessment is, but I will admit that a 5% rate sounds concerning to me personally if the overall trend is also positive.
One resource I like to use is the Cerner Resurgence Risk Index, which you can find for certain individual localities here: https://www.cerner.com/covid-19/predictive-models/reopening-projections
I think it's a good resource because it uses the effective reproduction coefficient (an assessment of how many cases are generated for each case identified) as well as other factors to assess the risk of a resurgence in a given location. Currently, Philadelphia is in dangerous territory according to this measurement. If I were you I'd be very careful, based on that. New York isn't in much better shape, either, by this measure, for what it's worth, and I'm starting to act a lot more carefully as a result.