THREAD: Many analyses try to estimate what the “right” capacity is for weekly #COVID19 testing. One touchstone is positivity rates. Unless diagnostic tests are being administered with unusual precision only to those infected, high positivity rates suggest under testing. 1/n
Right now positivity for #COVID9 tests is very high; and in past week it’s risen. This suggests still lot of infection across America and under testing. But if testing was widespread (a level able to detect new outbreaks before they become large) how low should positivity be? 2/n 
For example, let's look at positivity rates for flu, where testing is widely available. We're oddly complacent about flu spread, allowing it to become epidemic. With #COVID19, goal will be to test at level to prevent epidemic spread, not simply to track trajectory of the epidemic
When flu infection is low, at the beginning of the epidemic, positivity rates on tests are around 1-4%. This data is from early October of 2017, before the flu epidemic began in earnest that year, and when infection first started to spread more widely 4/n cdc.gov/flu/weekly/wee…
In height of flu season, positivity reaches 30%, but that’s a full blown epidemic. It is too high in the context of covid. Before epidemic builds, rates much lower. The goal of #COVID19 surveillance: Detect outbreaks while small; addressable with case-based interventions. 5/n
Surveillance for other infectious diseases, where the goal is to use widespread testing of suspected cases as a way to control disease outbreaks, could also provide some clues as to the right level of testing to have in place. 5/n
Analyses of screening campaigns for tuberculosis in high risk populations have found positivity rates of around 1-2%. This study is of TB surveillance system that was established among 106,000 people in rural Bangladesh, with a positivity rate of 1.4% 7/n cambridge.org/core/journals/…
Another recent, albeit very different touchstone: screening of blood donations for Zika in Puerto Rico during 2016 outbreak. Analyses from 4/3/16 - 6/11/16 found 68 (0.5%) presumptive viremic donors from 12,777 tested. Highest weekly incidence was 1.1% 8/n cdc.gov/mmwr/volumes/6…
HIV testing offers additional insights. Overall positivity rate for HIV tests conducted at CDC funded test sites was 0.4% in 2017 for those newly diagnosed (higher in certain high risk populations) kff.org/hivaids/fact-s… Other series mirror findings annals.org/aim/fullarticl… 9/n
These are distinct diseases, with different spread dynamics, incidence, testing criteria. But for all of these dangerous pathogens, like covid testing, goal is epidemic control. This data can help inform thinking on right amount of covid testing based on positivity rates. 10/n
The high positivity rates on #COVID19 testing would suggest our testing capacity is still too low. At the current levels of infection, to achieve a positivity rate of 1%, we would need to have conducted 70+ million covid tests. To date, we've conducted close to 4 million. 11/n
Put another way, If we use positivity rates in South Korea, Germany as guide, then with our 1M/week test capacity in the U.S., we'd want to have only about 20k-70k confirmed cases/wk in U.S. to have comparable positivity rate. This week, in the U.S. we added about 200K cases 12/n
For reference, about 2% of tests come back positive in South Korea, and about 7% in Germany. Here’s how positivity rates break down in the different states in the U.S. 13/n
In the coming weeks capacity for testing will continue to build. We are making much progress. Our goal should be screening that allows anyone with #covid symptoms to receive a timely and accurate test; as well as those who may have been exposed to an infected person. 14/n
