Archive for the ‘Covid-19 Testing’ Category

Birx: “Key metric that I watch is the positivity rate”

July 24, 2020

In yesterday’s post, I indicted that I was very disappointed with Dr. Brix’s answers  in a TV interview with Bret Baer.

One particular question & answer still has me scratching my head:

What is the key statistic that you track?

Birx’s answer: Test positivity (i.e. ratio of positive test results to total tests). It’s the most sensitive indicator and best early warning.

With all due respect, I think that Dr. Birx is confusing “positivity” with “prevalence”.

I’m way more interested in the latter prevalence: the percentage of the population that is currently infected with the virus.

Prevalence indicates how widespread the virus is at any point in time in a selected locale.

That gives me a sense of how safe it is to leave my house: How many people am I likely to run into who have the virus and may be contagious.


Determining prevalence requires periodic random sampling of the population.

That’s not what’s being done now.

A representative sample of the local populations is not being tested.

Why is that a problem?

The positivity rate (Birx’s key metric) is a function of who shows up to be tested.

If only people with covid-like symptoms are being tested, then of course, the positivity rate will be high.

If there’s a groundswell of asymptomatic people, the positivity rate will likely be relatively low.

My hunch: The testing “sample” is skewed to people with symptoms.

In a prior post, we reported that “Asymptomatics” are not rushing to get tested and provided some subjective reasons why that might be (e.g. no doctor referral, high hassle factor, privacy concerns).

So, positivity is, at best, a very crude measure of prevalence.


To that point, keep in mind…


FDA okays “pooled” testing for COVID-19…

July 20, 2020

A potentially big first step to efficient testing


Over the weekend, the FDA issued an emergency use authorization for pooled samples.

The Quest test is the first COVID-19 diagnostic test to be authorized for use with pooled samples.

Her are details from the FDA press release:


Why is COVID testing still so haphazard?

July 15, 2020

Test results come too late for therapeutic decisions … and “the science” still can’t answer basic questions.

Based on some back-of-the envelop arithmetic, I estimate that about 13 million Covid tests have been administered in the 3 weeks ending July 13

Note: The time period is strictly arbitrary.  And, since I don’t have all of the daily data series, I just derived rough estimates off the charts. I doubt conclusions would change much with a different time period or more precise numbers 


Now, let’s drill down on those numbers….


Of the roughly 13 million tests that were reported, over 12 million (94%) came back covid-negative’.

Note: Practically all headline reporting is “confirmed cases” — the test results that were covid-positive. Total test minus positive results estimates negative results.


My 1st question is who the heck are these 12 million people who are testing negative … and why, in the first place, are they even being covid-tested.

Some possibilities:

  • They’re exhibiting flu-like symptoms that might indicate covid.
  • They know (or think) that they’ve been exposed to the virus … via a super-spreader person or at a superspreading event
  • Their employer is making them take the test (e.g. frontline healthcare workers)
  • They’re participating in a medical research project (e.g. a vaccine trial)
  • They’ve been selected to be part of a random surveillance sample
  • They’re trying to be good citizens by participating in the “test, test, test” program
  • They’re just curious as to whether or not they’re infected.

I think the top 5 categories are pretty legit.

The last 2 strike me as a waste of constrained testing capacity.

Regardless, wouldn’t it be nice to know how the 12 million sorts out by those categories?

Apparently, that sort of classification data isn’t captured at the  point-of-testing … or, public health officials just aren’t making the data available.

That’s too bad


Moving on…

Less than 1 million of the 13 million tested Covid-positive … a “positivity rate” of 6%.

Who are these people and how are they treated when they’re tagged as infected?

I’d like to know:

  • How many present with severe, mild or no symptoms?
  • Do they have co-morbidities or not?  If yes, how many? Which ones?
  • What treatment plan is prescribed? Hospitalization? Quarantine? R&R?

Again, it appears that this sort of classification data isn’t captured at the  point-of-testing … or, public health officials just aren’t making the data available.


Bottom line: We need to focus constrained testing resources on therapeutic decision incidences … and, we need to gather and analyze more classification so that the testing helps pin down how this virus is acting.

Testing that simply increases the number of tests doesn’t seem to be getting us anywhere.

I made the same mistake that Gov. Cuomo is making …

May 29, 2020

… but, in my defense, I was only 9 years old at the time.

Flashback many years to my earliest little league baseball days.

1962 Ken & John - Little League

In our first game, the coach told us that — just like in the big leagues — he’d be  flashing secret signs to us when we were batting:

If he touched his cap, the sign meant “take the next pitch” … don’t swing at it.

If he brushed his chest, it meant “swing away”.

Simple enough…

Here’s what happened…


I tested negative, so I’m not infected, right?

May 28, 2020

Yesterday, we reached into our toolkit and pulled out behavioral economics and Bayesian Inference.

Our big conclusion in that post was that if C-19 tests are 90% accurate and 5% of the people in our reference group are walking around infected, then roughly 2/3’s of all people who get positive test results are not infected … they’re so-called false positives.

Now, let’s change one of our assumptions.

In the prior post, we assumed that we were asymptomatic, have been sheltering-in-place (i.e. minimal social contacts outside of our homes) and don’t work in a COVID-prevalent environment … and we used 5% as our base rate (of virus prevalence among our reference group).

Now, let’s assume that the reference group we’re working with is elderly, has a comorbid medical history of respiratory and heart problems and is experiencing COVID-like symptoms (high fever, persistent cough), have had contact with an infected person.  That’s essentially the only group that initially qualified for coronavirus testing.  Lets, assume that 75% of the people in that reference group are, in fact, infected with the virus.

Here’s the Bayesian results chart would look like:


The question: what is the likelihood that the people who fit this profile are correctly diagnosed as having the virus (or not)?


If I test positive for COVID, am I infected?

May 27, 2020

The answer may surprise you, and it has big implications for test & trace.

In a prior post, we reported that “Asymptomatics” are not rushing to get tested and provided some subjective reasons why that might be (e.g. no doctor referral, high hassle factor, privacy concerns).

OK, let’s up our game a notch or two and throw some math & economics at the problem.


I’m a fan of “Freakonomics” … the popular call sign for a discipline called Behavioral Economics … the study of the rationality that underlies many seemingly irrational decisions that people sometimes make.

And, in my strategic business analytics course, I used to teach something called Bayesian Inference … a way to calculate probabilities by combining contextual information (called “base rates” or “priors”) with case-specific observations (think: testing or witnessing).

Today, we’ll connect Freakonomics and Bayesian Inference and apply them to the COVID testing situation…


MUST READ: How will we know when we’ve turned the COVID-19 corner?

May 26, 2020

Stay focused on the number of Daily New Deaths!

Cutting to the chase, I’ve concluded that the most reliable number being reported is the number of COVID-19 related “Daily New Deaths”.

According to Worldometers – the best data aggregation site that I’ve found so far – there have been almost 100,000 COVID-19  related deaths in the U.S. so far.


Keep in mind that “COVID-related” means “COVID present”, not necessarily “COVID caused” … and that, along the way, “present” was redefined from “confirmed” to “presumed”


From an analytical perspective, the chart of total deaths will, by definition, never crest and turn down. It’s rate of growth will eventually slow down, though, but that’s hard to read that from a chart.

So, I think it’s more useful to look at “Daily New Deaths” …. if that number keeps going up then, by definition, we haven’t turned the corner.

When Daily New Deaths start trending down then, by definition, we have turned the corner.

Here’s our charting of what Worldometers has reported since the first coronavirus cases were identified.


The dotted line is the 7-day moving average which smooths some of the day-to-day “noise” in the data.

Based on the 7-day moving average, it appears that the rate of growth of COVID-19 deaths trended downward since about April 21.

Bottom line: If you want to know if we’re starting to turn the corner, keep your eye on the number of COVID-19 related “Daily New Deaths”.

Choose the level of aggregation based on your specific interest … world, nation or state.

Note: I’ll be focusing on the U.S. national number … and the national number less the 3 state hot spots: NY, NJ, CT


More specifically, why “Daily New Deaths”?


Shocker: “Asymptomatics” not rushing to get tested.

May 21, 2020

Quick quiz: If you wanted to get tested for COVID-19, how would go about getting a test? What are your odds of actually getting tested?

The Washington Post ran an obvious-became-evident exposé:  “As coronavirus testing expands, a new problem arises: Not enough people to test.”

These days, COVID testing capacity is said to be exceeding demand in some (many?) locales.


So, why aren’t people rushing to get tested?


COVID: How about squeezing the data and doing some old-fashioned profiling?

May 12, 2020

Hint: Go back and ask people who have been tested or hospitalized.


Finally, Gov. Cuomo has directed hospitals to ask new coronavirus patients for some demographic and behavioral information such as their occupation, usual transportation mode and neighborhood.

Cuomo says the early results from this info-seeking initiative and the state’s antibody testing have provided  some confirming data and some “shocking” revelations, including:

  • 96% had an underlying health condition (a.k.a. comorbidity factors); new admissions were predominantly minority, predominantly older; 22% came from nursing homes.
  • 66% of NY’s new coronavirus hospitalizations are people who are either retired or unemployed and not commuting to work on a regular basis … only 17% were employed.
  • The majority of recently hospitalized coronavirus patients are people who say that they have followed the precaution of staying home.
  • Only 4% in New York City said they had been taking public transportation.
  • A low percentage of new hospitalizations were essential employees — nurses, doctors, transit workers, grocery store employees — who were getting sick at work.
  • Sources: WSJ   NY Daily News

Of course, these sample sizes are small and the results may or may not be projectably true.

The point is that “they” should have been recording this sort of information from the get-go.

The plan is to start asking a battery of questions when people are tested for the coronavirus (both diagnostic and antibody testing, I assume) and when they’re admitted to the hospital.

That’s fine, but I’ve got a better idea…


Do Americans really care about nursing homes?

May 7, 2020

Maybe it’s time for a national gut-check

Earlier this week, there was a heart-wrenching story on TV.

A woman was telling the story of the Massachusetts state-run Holyoke Soldiers’ Home that had suffered over 80 C-19 fatalities. Her elderly father was one of the casualties.

She had been trying for weeks, to no avail, to speak with her father, or at least get a status report on his condition. Her first contact was when he was being wheeled to the coroner’s van.

click to see details

Of course, the daughter was heart-broken and observed “nobody seemed to care … they’re just old people”.

That struck a chord with me

Of course, people who have loved ones in nursing homes are concerned about their level of care.

At a minimum, they want their loved ones kept safe and comfortable.

But, what do we as a nation really think?



May 5, 2020

Over the weekend, Gov. Cuomo reported final results from the NY antibody test program..

Key Data:

12.3% of the NY state sample tested positive for coronavirus antibodies.

Note: The rate in NYC was 19.9%

That extrapolates to 2.4 million New Yorkers.

Given the current number of cases (327,374), 2.1 million (86.3%) of the already infected people were, by definition asymptomatic — having no or mild symptoms.

The implied deaths to infections rate is (1.0%).


The implications…


Senate may be starting to ask the right questions re: testing.

May 3, 2020

“Why do we have to have symptoms to get tested?”

The Senate (but not the Congress) is scheduled to get back on the job this week.

Roughly half the senators are 65 or older … and, thus, officially in the coronavirus’ “vulnerable” group.

So, it’s understandable that they’re eager that all colleagues have a clean bill of health before returning to the Senate chambers.


Here’s the rub…


Help Wanted: Vice President of Contact Tracing & Testing

May 1, 2020

Warning: Read this before you apply for the job.

Contact tracing & testing is front and center as a fundamental component of the Coronavirus Task Force’s plan to go forward.

Dr. Fauci has said (over & over again) that the process worked fine 30 years ago when he was fighting AIDs … and the media says that the test & trace model has been South Korea’s secret sauce fighting the coronavirus.

The essence of the process: Do diagnostic surveillance testing to ID people currently infected with the coronavirus, then trace back to ID the people with whom they’ve been in contact … then notify those people and test them … if they test positive, repeat the process … then again and again.

Sounds easy enough, doesn’t it?


But, it might not be as easy as it sounds.

Let’s run some numbers…


Still more about the NY antibody test results…

April 29, 2020

What about the 3% of New Yorkers floating around while infected but asymptomatic?

In a prior post, we squeezed the NY antibody test results pretty hard and estimated that about 600,000 New Yorkers are walking around at any one time infected with the coronavirus but exhibiting no or very mild symptoms. That means that about 3% of NY’s population are asymptomatic “hidden carriers” who may be unknowingly spreading the disease.

To understand their significance …

Most infectious disease epidemiology models are built on the “SEIR” construct: how many people are susceptible to a virus … of them, how many are likely to get exposed to it … of them, how many are likely become infected … and of them, how many are likely to recover, perhaps with some degree of immunity. The modelers then calibrate a virus’s behavior, estimating how long it takes people to move from susceptible to exposed to infected to final resolution (recovery or death).


My former strategy students should recognize the SEIR construct as a basic hierarchy-of-effects model, similar in design to, say, the classic marketing awareness – trial – repurchase model.

And, the spread effects are a classic Bass Diffusion Model application with infected people playing the role of “innovators” and susceptible people playing the role of “imitators”.

Let’s dive a little deeper…


NY Antibody Test < UPDATE>

April 27, 2020

Phase 2 Results

Gov. Cuomo reported Phase 2 results today.

Bottom line: Results are virtually identical to the Phase 1 results.

14.9% of the sample tested positive for coronavirus antibodies …  that’s up slightly from the Phase 1 results.

That projects up to 2.80 million New Yorkers.

Given the current number of cases (293,381), 2.6 million (89.8%) of the already infected people were, by definition asymptomatic — having no or mild symptoms.

The deaths to infections rate (.8%) was the same in both phases.


Squeezing the NY antibody test results…

April 24, 2020

Estimate: 3% of the NY state population are infected asymptomatics .. in circulation and potentially infecting others.

In my business analytics course, I used to nudge students to “squeeze the rock” .. to get as much possible information out of each test or piece of data.

OK, let’s apply that principle today …

In a prior post, I opined that NY antibody tests were missing an information opportunity.  If they also swabbed the random sample for C-19 diagnostic tests, they’d also have an estimate of the number of infected asymptomatics who are currently in circulation in NY.

OK, it was a missed opportunity.

But, let’s not fret.

We can squeeze the data to get a rough-cut estimate of  the number of infected asymptomatics who are currently in circulation in NY.

Let’s do some arithmetic …


NY’s missed testing opportunity

April 24, 2020

NY’s antibody testing program is highly commendable … and, it’’s already generating some very useful data.

But, I wonder..

Why aren’t they swabbing the same people for coronavirus diagnostic tests?

Doing so would tell us how many asymptomatic “hidden carriers” are currently in circulation.

My view: that’s one of the most important pieces of missing information … especially if it’s done on a periodic basis, say, weekly.

Would give us a good sense of how infectious the population is right now.