The Homa Files

Finally, some C-19 “prevalence” estimates…
=============

Loyal readers know that I’ve been frustrated by the lack of practical, personally-relevant data that the public health “experts” have been gathering, analyzing and disseminating.

For example, I’d like to know how likely I am to bump into a person who is infected with C-19 if I leave my house.

Reasonable question, right?

In infectious disease speak, that number is called the “prevalence rate” … the percentage of people in an area who are infectious-likely at a specific point in time.

Current testing procedures don’t answer the question (see: Why is COVID testing still so haphazard?) … so, the “experts” fixate on metrics that are muddled and potentially misleading (see: Birx: “Key metric that I watch is the positivity rate”).

Finally, I’ve stumbled on some “machine learning” analyses that take a stab at answering my question: What’s the Covid prevalence rate?

OK, let’s get to work…

=============

First, an important disclaimer:

• Most of today’s data is drawn from a site called Covid19-projections.com that relies on an “intuitive model that builds machine learning techniques on top of a classic infectious disease model.”
• The model and site are the product of the University of Washington’s oft-maligned Institute for Health Metrics & Evaluation (IHME) which generated many of the early inaccurate Covid forecasts (see: What you need to know about the IHME Model)
• That said, I think that IHME is a credible source since its models are logical and very much data-driven.  And, IHME is very transparent. It even provides details of its approach and even links to analyses that are critical of it’s work.
• My conclusion: IHME’s early forecast errors were mostly a function of the sparse data set that they had to work with early-on.  The data set is still problematic … but it’s getting better, so IHME has more to to work with.

=============

Top-line numbers…

Note: All numbers should be considered approximate since they are estimates, not actuals … and, I’ve done some heavy rounding to make the numbers easier to internalize.

Primary source: IHME Covid Projections

OK, so what do the numbers say?

Already infected: IHME estimates that 40 million Americans have already been infected with the coronavirus … that’s about 12.5% of the total population (1 in every 8 people) …  or, about 17% of the 19 & over adult population (1 in every 6 adults).

Confirmed cases: To date, there have been about 5 million confirmed cases … that’s about 1.5% of the total population (2% of the adult population) … and, the number of confirmed cases is about 12.5% of total infections to date (based on the IHME estimate of those already infected) … said differently, only 1 in every 8 infections have been confirmed by Covid testing.

New infections & confirmations: IHME estimates that there are currently 400,000 new infections each day … of which, about 60,000 (15%) are confirmed via testing.   (Note: The 15% current rate is higher than the 12.5% total rate since testing has been ramped up).

=============

And, for the critical number …

Prevalence: IHME estimates that there are currently 6 million “active infections” … that’s a prevalence rate of 1.85% of the total population (1 in every 54 people) … or, 2.5% of over-19 adults (about 1 in every 40 adults).

Note: The “positivity rate” that Dr. Birx watches — the percentage of tests that return a positive diagnosis — is running about 8%.  Why? Because the vast majority of people getting tested have symptoms or have been exposed to someone who has tested positive.  They don’t reflect the profile of the overall population.

=============

What are my local odds?

OK, now we’re getting someplace … a national prevalence rate of 1.85%.

But, what really matters (to me) is the Covid prevalence in my local area.

IHME takes a stab at localizing the estimates.

For example, IHME estimates that my home state (Maryland) has a prevalence rate of 1.55% that is:

• Lower than the 1.85 national average
• Lower than the current hot spot states: FL 4.47%, TX 3.91%, CA 2.14%
• Higher than “cool” states, e.g. SD, ND, MT, OH that are running between 1.0% and 1.5%

Of course, I’d like a more localized number, e.g. the covid prevalence in the Annapolis area.

But, finally. I think I’ve got a ballpark number to work with.

=============

Now, if only public health officials would use the testing apparatus to gather, analyze and report Covid data that helps us make decisions re: how to live our lives.

I’m not holding my breath…

This entry was posted on August 7, 2020 at 10:00 am and is filed under Coronavirus - Covid 19, COVID-19 Tracking. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.