Good news- coronavirus test count

Subject: Good news on the coronavirus test count

Please forgive me, but after too many hours suffering through statistical analysis courses at Georgia Tech, I just can't blindly swallow the conclusions by our public officials. The Hillsborough Emergency Planning Group no doubt overwhelmed their audience with their detailed briefing slides -- but let's look behind the wizard's curtain....

The increase in raw number of positive tests (slide 1 attached) tells us nothing. Without a contextual reference.

Gaping holes in their analysis:

1. The key piece of information missing is how many tests were conducted the week prior, compared to this last week? What if there were 10 times as many tests done this last week as the week prior? Of course you would expect more positives.

2. What exactly was tested? Viral tests or antibody tests? There are in fact two types of tests - the viral test and the antibody test - but no information Is available on the website to differentiate as to what type of tests are included in the counts.

If those positive test counts include antibody tests, that tells you absolutely nothing other than the individual may have been exposed at some time - could have been yesterday, could have been six months ago. It is not an indisputable indicator that social distancing or wearing masks has or has not impacted anything. And knowing if you have antibodies, or not, also tells you nothing, since no one knows for sure if you can be re-infected.

If the tests are viral tests —that really tells you nothing unless you look at the actual treatment required — were they hospitalized? Are they in intensive care? Were they treated and released? The second slide does show numbers of ICU vs non-ICU patients -- but when you dig into this question -- it looks like all the percentages are either improving or stable:

The number of hospitalizations relative to the number of tests conducted actually went down by one percent — check the numbers — the week prior was 189 hospitalizations out of 4273; this last week it was 257 hospitalizations out of 6892 or a decrease in percentage from 4.42% to 3.73%. And what about the people hospitalized — were they in there for intensive care or non-intensive care? if you check out those numbers you’ll see the ICU percentage rate the week prior was 21% and the percentage rate for this last week was 21% —- no change.

Simply looking at the increase in the raw number of positive test results with no contextual reference is very misleading.