So much in the media about #COVIDー19 testing, it seems important to correct some misconceptions. Thread

Two types of test:

Viral RNA tests - RT-PCR - swab tests - to detect current infection

Antibody test - serology - blood tests - to detect previous infection

No test gives perfect 100% accurate results. Tests need to be evaluated to determine their sensitivity and specificity. More evidence here is needed - hence this call http://www.finddx.org/Covid-19/dx-data

BUT - remember Bayes theorem?
Interpretation of a test result depends on 2 things -
1-Sensitivity/specificity of the test
2-Pre-test probability or risk of disease before testing
What does this mean in Covid-19?

Let’s assume 2% of the population is infected with Covid-19, and 50% of these have fever. If background prevalence of fever is 1% then anyone with fever has a 50% chance of having Covid-19 ie pre-test probability is 50%

If viral swab tests are around 70% sensitive 95% specific then a patient with fever (pre-test probability estimate 50%) who has a negative test still has a 24% chance of having the virus.

What if the pre-test probability is higher? eg patients with strongly suggestive symptoms of cough, fever and a known infectious contact. We might estimate a 90% pre-test probability.
If they have a negative test they still have a 74% chance of having Covid-19

You can use this calculator to play around with the numbers yourself and see what happens when you adjust the parameters https://calculator.testingwisely.com/playground 

The take home message is that if you have Coronavirus symptoms - then assume you have Coronavirus - even if your test is negative

This is especially important for NHS workers who are now getting prioritised for testing - they risk being super-spreaders if they are falsely reassured by a negative test

Yes we need to test, test, test in order to understand the spread of this pandemic and to plan the appropriate public health measures to reduced spread

But testing is not a panacea and without the correct interpretation of test results, false positives could do significant harm

False *negatives*
Why doesn’t twitter have an edit button!