On 30 Mar 2020, at 22:50, Rod Jackson <[log in to unmask]> wrote:
Hi James. You are correct that we need a gold standard to determine a false negative definitively. However comparing studies like the ones I quote is meaningful if they used similar criteria. I am sure the true false negative proportions in the studies I quoted will not be 0% and 80%, but they will be ball-park figures and highlight the key issue about their dependence on the pre-test probability. Moreover, they will change by the day depending on both testing proportions and disease prevalence. I didn’t bring up the other important issue which is the quality of the sample, which I understand can be quite variable.
Cheers Rod
On 31/03/2020, at 10:44 AM, McCormack, James <[log in to unmask]> wrote:
Hi Rod - thanks for the positive comments - really appreciate it. So many people are doing great work around the evidence - and primarily around the lack of evidence.
Also agree with your comments about the false positives.
The false negatives are interesting - I was just wondering how can we define a false negative - to do that we have to have a gold standard for the diagnosis - at present that is PCR - but if the virus should clear over time how do we differentiate between a false negative and the fact that it is negative because the virus has cleared. Maybe I’m missing something.
James
On Mar 30, 2020, at 2:13 PM, Rod Jackson <[log in to unmask]> wrote:
Thanks very much James for sharing this. In New Zealand groups of academic public health medicine docs are starting to write and peer review rapid reviews related to Covid-19 and I have shared your post with our local co-ordinator.
I assume that similar groups around the world are doing similar things, so could I please strongly encourage list members to follow the example set by James and share these and also peer review them.
I just did a rapid critique of your review James and thought it was excellent.
James, on a related topic, In your last post you mentioned that the specificity of PCR was pretty much 100%. I discussed this with colleagues and we agreed that even if there are possibly a few false positives, we should assume there are none.
A much more complicated issue is the so-called ‘false negative rate’ (i.e. the proportion of all negative tests that are false), which one of my colleagues conducted a rapid review on. This is actually 1 minus the negative predictive value (NPV) and for the pedantic is a proportion, not a rate. But the key point is that, like the positive predictive value (PPV), it is highly dependent on the pretest probability (disease prevalence) in those tested. A study in asymptomatic French nationals who had been in Wuhan found a false negative proportion of 0, while another study of older Chinese people with CT evidence of viral pneumonia, found a false negative proportion of 80%. I thought this was a good reminder that the PPV and NPV are highly context specific.
Cheers Rod JacksonUniversity of AucklandNew Zealand
On 31/03/2020, at 5:46 AM, McCormack, James <[log in to unmask]> wrote:
Hi Everyone
Just posted our first COVID-19 Rapid Review - Tools for Practice. All about hydroxychloroquine. It will take you just a few minute to read and you will be up-to date as of now, but things are changing hourly as we all know.
Will have one soon on assessing dyspnea virtually.James
To unsubscribe from the EVIDENCE-BASED-HEALTH list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=EVIDENCE-BASED-HEALTH&A=1
To unsubscribe from the EVIDENCE-BASED-HEALTH list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=EVIDENCE-BASED-HEALTH&A=1
To unsubscribe from the EVIDENCE-BASED-HEALTH list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=EVIDENCE-BASED-HEALTH&A=1